Market reactions to audit firm rotation : does tenure matter?

(1)

MSc Accountancy & Control, variant Accountancy

Faculty of Economic and Business, University of Amsterdam

Author: Paola Monster Student number: 10399720

Date: June 23 2014

Program: Master Accountancy & Control Track: Accountancy

First Supervisor: Dr. Jeroen van Raak Second Supervisor: Dr. Sanjay Bissessur

Market reactions to audit firm rotation

(2)

1

Table of Content

Abstract ... 2

1 Introduction ... 3

2 Literature review and hypotheses ... 5

2.1 Demand for audit and auditor choice ... 5

2.3 Theory behind relationship between audit quality and industry specialization ... 8

2.4 Industry specialization, audit firm tenure and the perceived audit quality ... 10

2.5 Hypotheses ... 11

3 Research Design ... 13

3.1 Measurement of Industry specialization ... 13

3.2 Measurement of Perceived audit quality ... 16

3.3 Measurement of Audit firm tenure ... 17

3.4 Control Variables ... 18 3.5 Research Design ... 19 3.6 Sample Construction ... 21 4 Descriptive statistics ... 22 4.1 Independent variables ... 22 4.2 Dependent variable ... 24 4.3 Correlation Matrix ... 26 4.4 Empirical Results ... 28 4.5 Sensitivity analysis ... 30 5 Conclusion ... 32 References ... 35 Appendix ... 39

(3)

2

Abstract

A number of studies have investigated the stock market’s reaction to firms switching to and from Big4 auditors. However, a Big4 auditor is not the only possible indication of audit quality. Tenure and industry specialization are considered valuable aspects by investors as well. This study contributes to the existing literature by examining how the market reacts to audit firm switches after a short and long tenure and how the market reacts to firms switching from and to industry specialist audit firms. Additionally, this study contributes to the ongoing mandatory audit firm debate. Consistent with my hypotheses, I find that firms experience significant negative abnormal returns when the switch is made after a long tenure. This suggests that investors perceive a decrease in audit quality when a switch is made after a long audit firm tenure. Additionally, I find that firms switching between Big4 audit firms experience significant positive abnormal returns when the successor auditor is an industry specialist. This result suggest that the market does perceive audit quality differences based on industry specialization.

Keywords: Mandatory audit firm rotation, audit quality, market reaction, tenure, specialist

(4)

3

1 Introduction

The financial crisis has brought up several questions about the quality of the audit and the independence of the external auditor. One of the issues often discussed, is the reform of the audit market in Europe to create financial stability. Mandatory audit firm rotation is often suggested as one way to increase auditor independence, since it is important for stakeholders to be able to rely on the audited financial statements. The European Commission presented a legislative proposal indicating that the controlling audit firms should rotate every six to nine years (European commission, 2011). In December 2012 the Dutch Senate announced that they agreed on this proposal (Eerste Kamer der Staten Generaal, 2012) and mandatory firm rotation on which audit firms must rotate every eight years is currently implemented in the Dutch legislation which will go into effect on the first of January 2016.

Mandatory audit firm rotation refers to the imposition of a limitation on the period of which a particular registered public accounting firm is allowed to control a particular company (Sarbones-Oxley Act, 2002). Proponents of mandatory firm rotation argue that it can help eliminate the unconscious desire that auditors may have to please the client and hereby increasing independence (Arel et al., 2006). Arel et al. (2006) state that auditors might have an unintentionally self-serving bias and might cause them to agree with the client on certain aspects in order to retain the client. Thus with mandatory rotation they do not have to align their interest with the interest of the client. The study of DeAngelo (1981) supports this argument. It states that an existing client provides quasi rents to the audit firm which they expect to receive the entire life of the auditor-client relationship. This life is ‘eternal’ and this makes the customer an important source of income. According to DeAngelo (1981) is this 'eternal' income a factor which makes the audit firm depending on the customer and therefore harms the quality of the audit.

Conversely, Jackson et al. (2008) state that during rotation client-specific knowledge gets lost at the expense of audit quality. An important finding of their study is that audit quality increases as the tenure of the firm increases, because less errors are made as the firm gains more client-specific knowledge during the years. Auditor tenure is defined as the number of years an auditor is retained by the firm (Myers et al., 2003). In addition, Johnson et al. (2002) state that the duration of the term has an effect on the quality of an audit. They find that a short-term relationship of two to three years, is associated with lower audit quality

(5)

4

relative to medium audit-firm tenures of four to eight years. They find no evidence of reduced audit quality for longer audit-firm tenures of nine or more years.

It is obvious that different studies have contradictory results and therefore form different opinions of what mandatory audit firm rotation does to the quality of an audit. Instead of only looking at the quality of an audit due to audit firm rotations, many researchers have focused on the shareholders perception of audit firm switches. Chang et al. (2010) for example examined market responses to auditor switching from Big 4 to smaller accounting firms and found positive market response. Additionally, Ghosh and Moon (2005) provide evidence that longer auditor tenure is associated with stronger earnings response coefficients, which suggests that investors perceive earnings quality to be better of firms with longer auditor tenure compared to the earnings quality of firms with shorter auditor tenure. Knechel et al. (2007) also examines how the market reacts to auditor switches but examines firms that are considered to be industry specialists. They found that the market reacts positively to news of companies that switch from a non-specialist Big 4 auditor to a specialist Big 4 auditor. Their results suggest that the market finds industry specialization to be valuable in order to be able to get relevant valuations of a company’s market value and note that the market perceive this to be relevant due to audit quality differences.

While the prior market study of Gosh and Moon (2005) found a positive market response to long tenure, I have seen no studies that document the market reaction to audit firm switches after a long tenure compared to a short tenure. Additionally, Knechel et al. (2007) document a positive reaction for firms switching to an industry specialist, so this study examines if this is still the case if the switch is made after a short tenure or long tenure. The research question addressed in this study is therefore: Does audit firm tenure influence the stock market reaction when companies switch from a non-specialist Big4 audit firm to a specialist Big4 audit firm?

Following this question I test three hypotheses. I predict that abnormal returns turn out to be negative when a switch is made after a long audit firm tenure compared to switches after a short tenure. I expect this because prior literature (e.g. Gosh and Moon, 2005) indicates that long tenure is appreciated by investors and therefore depreciate that the long tenure is diminished with an audit firm switch. I find evidence for this hypothesis when testing the independent variable with a 5day window. This implies that that investors perceive a decrease in audit quality when a switch is made after a long audit firm tenure. I also hypothesize that positive abnormal returns are found when the switch is made to an industry audit firm. This hypothesis is also supported which suggest that investors appreciate companies switching to

(6)

5

an industry specialist which reflects higher audit quality. Finally, my third hypothesis states that the negative abnormal returns obtained when the switch is made after a long tenure is smaller when the switch is made to an industry specialist. I find no evidence for this hypothesis.

Investors are important stakeholders of a company and one of the main objectives of mandatory rotation is to provide more credible information to investors. The aim is to protect investors due to a gain in auditor independence and thus higher audit quality. Mandatory audit firm rotation has been subject to an ongoing debate between many stakeholders, including investors. This makes the study socially relevant due to the fact that mandatory audit firm rotation is being implemented in Dutch legislation and it is therefore good to know if shareholders perceive this rotation as an improvement in audit quality. This study find evidence that investors perceive a decrease in audit quality when the switch is made after a long tenure. Indicating that if mandatory audit firm rotation is implemented and firms have to switch every 8 years, investors perceive this negatively resulting in negative abnormal stock returns. Therefore this study can contribute to the ongoing debate. Additionally, from an academic point of view this study contributes to prior literature by examining if there is a difference in the stock market reaction after a short tenure compared to such a switch after a long tenure. This has not, to my knowledge, been examined before.

The remainder of this paper is organized as follows. The second section describes the prior research and the hypotheses of this study. The third section starts with explaining how I measure auditor industry specialization, the research design and discusses my sample construction. The fourth section presents the descriptive statistics on my independent and dependent variables. The fifth section presents the results from my empirical tests followed by section 6 where I discuss the sensitivity test that I performed. Finally, a short summary, limitations, and conclusion is given in the seventh section.

2 Literature review and hypotheses

2.1 Demand for audit and auditor choice

The Agency Theory describes the relationship between the principle (i.e. the shareholder) and the agent (i.e. managers) and concerns aligning the interests of the principal and the agent (Jensen and Meckling, 1976). This theory explains the demand for auditing as arising from conflicting interests between managers and shareholders and other entities that contract with a

(7)

6

client (e.g. creditors). Separation of ownership and control can introduce agency problems such as moral hazard because of an informational asymmetry. Information asymmetry occurs in a situation in which one party in a transaction has more or superior information compared to another. In this setting, informational asymmetry occurs when the manager has access to superior information concerning the manager's performance, and it is assumed that the shareholder cannot observe the manager's behavior. The existing concern is that the manager will misuse his position and his information in order to maximize his self-interest at the expense of the shareholder (Jensen & Meckling, 1976). There are several options to solve this problem such as incentive contracts to align the shareholders interest with the interests of the shareholders. Another option is to alleviate moral hazard by providing complete public disclosures of the firm’s information to remove the asymmetric information between the shareholder and the manager.

Williams (1988) argues that both of these options require complete and reliable financial information which creates a demand for auditing services. The shareholders require independent monitors to assure the fairness of the financial statement disclosures. Under the stewardship hypothesis a manager (agent) would seek an auditor who satisfies the shareholder’s need for assurance. The underlying assumption here is that there are incentives for managers to provide financial statements to facilitate monitoring activities by principles. As shareholders might not trust the managers to keep their best interest in mind, shareholders might reduce the compensation of the managers in order to compensate for the risk of loss they perceive. Hence, the manager will agree to provide evidence that the financial statements that they produce are reliable and free of material misstatements. To provide this evidence the manager will seek to an independent auditor in order to avoid the reduction in compensation. The shareholders expect the managers to select an auditor who is competent in conducting high quality audits. Thus is expected that the manager keeps the interest of the shareholder in mind when selecting the auditor.

So from a theoretical perspective, the need for auditing can be explained by Stewardship Theory and Agency Theory. Stewardship Theory applies where managers are hired by the principal (owner) to act in the best interests of the owners (Williams, 1988), and that when one party is delegated decision power, this party has an incentive to agree to be monitored if the benefits from these monitoring activities exceed the related costs (Wallace, 1980). The Agency Theory proposed by Jensen and Meckling (1976) further explains that, for a company where there is a separation of ownership and control, the principal (owner) should be willing to incur a financial cost to monitor the activities of their managers.

(8)

7

2.2 Theory behind relationship between audit firm tenure and audit quality

Agents and principles seek for high quality audit to get the best assurance as possible as explained in the previous section, but audit quality is not a concept which has a definition that has been universally recognized. Duff (2004) asserts that many attempts have been made to define the audit quality but that generally the definition of DeAngelo is used. DeAngelo (1981) defines audit quality as the joint probability that an auditor will 1) detect a breach or error in the accounting system and 2) withstand client pressure to report that breach. The first probability reflects the technological ability of an auditor, such as technology, competence and effort, on which they are able to conduct the audits and identify the breaches. The conditional probability reflects the objectivity of the auditor, i.e. auditor’s independence.

The academic debate surrounding mandatory audit firm rotation (MAR) has mainly focused on how auditor tenure affects audit quality (Cameran et al., 2009). Opponents for MAR argue that shortening auditor-client relationship enhances auditor independence, and thus audit quality. A long tenure increases the familiarity risk which might adversely affect auditor independence and a longer tenure might lead to routine work which lead to auditors tending to rely on earlier reports and audits (Arel et al., 2005). Thus, losing their professional skepticism and their attention to detail. A new incoming auditor is expected to provide a “fresh look” at the company’s accounts and to be more objective. MAR has therefore been proposed as a potential solution because limiting auditor tenure reduces concerns about deteriorating independence and audit quality.

On the other hand, opponents argue that audit quality decreases when the audit firm is forced to resign. Both theoretical and empirical evidence have shown that the skills of an auditor increases as the auditor gets more experienced with the client. Auditors experience a significant learning curve with new clients and as the years of audit firm tenure increase they acquire more client-specific knowledge as the new auditor becomes familiar with the client’s operations (Arruñanda and Paz-Arez, 1997). Evidence have been provided about higher occurring audit failures in the first years of the auditor-client relationship because of the lack of this client specific knowledge. Nashwa (2004), for example, showed that errors are mainly made in the first three years of the audit firm tenure. He argues that this errors occur in the audit because of the lack of specific company information and experience and that the possibility of failure in the audit increases with each change of auditor. This indicates that the quality suffers as long as the auditor is not familiar with the activities of the company. Also, Carcello and Nagy (2004) find that accounting fraud is more likely to occur in the early years

(9)

8

of tenure (three years or less) compared with medium audit tenure (four to eight years) and long auditor tenure (nine years or more).

Additionally, Johnson, Khurana and Reynolds (2002) investigate the effect of audit firm tenure on earnings quality, which is a proxy for audit quality, for U.S.-listed firms. Johnson et al. (2002) find that short audit firm tenure is associated with higher unexpected accruals, indicating lower earnings quality. They also find less persistence of the accrual component with a short tenure, which indicates a reduction in earnings quality. On the other hand, they did not find a significant relation between long audit firm tenure and unexpected accruals. These results indicate that short audit tenure is correlated with lower-quality financial reports and that financial-reporting quality does not deteriorate with longer audit tenures. Johnson et al. (2002) state that a firm’s management has more discretion in the financial reporting process in the early years of the audit engagement relationship and that earnings quality is therefore lower. Consistent with Johnson et al. (2002), Myers et al. (2003) find that earnings quality is relatively lower when audit firm is short and that the magnitude of both discretionary and current accruals decline with longer auditor tenure. Although they use different proxies to measure earnings quality their results also suggest that the quality of the audit is lower with a short tenure and increases along with the length of the tenure. Both studies therefore indicate that audit firm tenure has a significant impact on the quality of the audit provided.

The findings of the mentioned studies show the importance of the skills of an auditor in relation to audit quality and that they suggest that it takes time for auditors to develop client specific knowledge to perform an effective audit. There is a reduction in the ability to detect distortions in the financial statements and therefore do not provide a true and fair view of the firm’s performance if the audit firm lacks client-specific knowledge and expertise (Arruñada & Paz-Ares, 1997). Considered collectively, prior literature on auditor tenure and audit quality assert that audit quality is lower in the early engagement years and there is no evidence of a deterioration of audit quality with longer tenure. This imposes that mandatory audit firm rotation may not lead to improved audit quality.

2.3 Theory behind relationship between audit quality and industry specialization

Audit quality and tenure are significantly correlated as showed in the previous paragraph but industry specialization also play an increasingly important role in audit quality. An industry specialist is defined as auditors whose training and practice experience particular lay in a

(10)

9

particular industry and this focused training and concentrated experience result in more knowledge of financial statements errors (Solomon et al., 1999). Standards setters, such as the AICPA 1997, have suggested that industry expertise results in higher audit quality. Audit firms are placing more emphasis on industry specialization since industry specific expertise provide audit firms to assure higher quality audits (Carcello and Nagy, 2002).

Carcello and Nagy (2002) investigated this association and found a negative relation between auditor industry specialization and client financial fraud indicating that fraud is less likely with industry specialization. This result suggest that when audit firms are specialized in the specific industry in which the client is operating, it is less likely that fraud occurs and thus presents a higher audit quality. Furthermore, Maletta and Wright (1996) investigated differences in the incidence, magnitude, income effect, cause and method of detection of large financials statement errors. They looked at a broad case of industries utilizing actual audit work paper information and found that if an audit firm has specialized knowledge in an industry the ability increases to asses audit risks, detect errors and misstatements, and improve earnings quality. Both studies revealed that auditor having industry specialism are more likely to provide an audit of higher audit quality.

Thus, industry-specialists have more industry expertise that enables them to identify misstatements more effectively. They get this industry expertise from serving other clients in the same industry and learning and sharing best practices across the industry. Industry-specialists are incentivized to correct or report identified misstatements in order to protect their market shares and an increase in audit quality should impact disclosure quality by enhancing financial statement credibility (Dunn and Mayhew, 2004). Dunn and Mayhew (2004) tested the relationship between the use of an industry specialist audit firm and the quality of the firm’s disclosure. They provided evidence on the effects of hiring an industry specialist auditor and what impact the choice of an auditor has on financial reporting quality. Their results suggest that industry-specialist audit firms improved disclosure quality, and that the choice of an industry-specialist auditor is a signal of enhanced disclosure quality.

Other research showed that industry specialist are associated with higher earnings response coefficients (Balsam et al. 2003). Furthermore, evidence is provided that industry specialists are associated with a stronger association between current earnings and subsequent cash flows, more specifically a positive relation between auditor industry experience and the ability of client earnings to predict future cash flows was founded (Gramling et al., 2001), and a negative relation between auditor industry specialization and absolute discretionary accruals was obtained (Krishnan, 2003).

(11)

10

In conclusion, all these studies suggest that higher audit quality is obtained if the auditing firm is an industry specialist. This allows me to suggest that if an audit firm rotation takes place between a non-industry specialist to an industry specialist the client‘s audit will improve and thus an enhanced audit quality will be obtained.

2.4 Industry specialization, audit firm tenure and the perceived audit quality

Previous paragraphs showed that auditor’s industry expertise and increasing tenure has a significant positively impact on the quality of the audit. Audit quality comprises actual and perceived quality accounts (Jackson et al., 2008). Actual audit quality is the degree to which material misstatement in the financial accountants are discovered and reported (DeAngelo 1981) while perceived quality is how effective users of financial statements believe the auditor is at reducing material misstatements in the financial accounts (Jackson et al., 2008). Shareholders are the principal users of financial statements and how they perceive the quality of the financial statements is an important factor as it might have an impact on the overall business. If shareholders have the feeling that the financial statements do not present a true and fair view of the business operations it might have a negative impact on the business as it might cause the share prices to drop. Therefore it is good to know how shareholders perceive a rotation of an audit firm.

Prior literature on the perceived audit quality due to auditor switches has produced contradictory findings. For example, Knechel et al. (2007) argue that auditing quality is perceived by investors to be different as some auditors are perceived to be more credible than other auditors. They state that previous market studies propose that a switch from or to a Big4 auditor will induce a positive or negative market reaction, due to a change in the perception of audit quality, at the time of change. However, evidence showed inconsistency in these proposition as some have find a positive or negative reaction while others find no reaction.

The study of Knechel et al. (2007) was drawn on these findings as they state that there have been a numerous studies that have examined the market reaction to auditor switches by focusing on switches to and from the Big4, but that no study has investigated investor’s reaction to auditor switches by focusing on industry specialist auditors. Hereby they were able to evaluate if the market considers industry expertise to be valuable. The results provided evidence that the market reacts positive to new of a company switching from a Big4 non-specialist to a Big4 non-specialist suggesting a perceived increase in audit quality. They were also able to provide evidence that market reacts negatively to news of companies switching from a

(12)

11

Big4 specialist to a Big4 non-specialist. This in its turn suggest that investors perceive that audit quality decrease when the company switches to a non-specialist audit firm.

Additionally, Balsam et al. (2003) examined whether an auditor’s industry specialist status is associated with earnings response coefficients (ERC). They examined this relation because ERC measures the extent of stock market responsiveness to earnings surprises. When the perceived uncertainty and noise in reported earnings are reduced, as obtained through higher audit quality, the ERC will be higher. Balsam et al. (2003) showed a positive association between industry specialization and ERC. Their results suggest that if an auditor is a specialist in the client’s industry, specialist auditors increase the markets’ perception of earnings quality.

Besides the market reaction to industry specialization, Gosh and Moon (2005) examined how investors perceive auditor tenure. They analyzed whether investors perceive earnings quality as being affected by tenure and found that investors perceive auditor tenure as improving audit quality. They stated that therefore, to the extent that capital market participants view tenure as improving independence and audit quality, financial statements are perceived as more reliable for financial decisions as tenure lengthens. In addition, Mansi et al. (2004) argue that the auditors have a dual characterization, namely as insurance provider and information intermediary which suggests that audits prove value to the capital market. Their results suggest that investors value the insurance role of auditors in addition to their information role. Furthermore, they provide evidence that investors value audit tenure as they require lower rates of return when the length of the audit firm tenure increases. This suggests that mandatory audit firm rotation may not be beneficial and could be viewed negatively by the capital market for riskier firm.

So although prior literature produced contradictory findings on the market reaction to auditor switches, the mentioned studies showed that investors perceive tenure and industry expertise to be valuable. Financial statements are perceived as more reliable when the tenure lengthens and when the auditor has industry expertise. This allows me to suggest that 1) the market reacts positively to the length of the audit firm tenure and 2) the market reacts positive to companies switching from a Big4 non- industry specialist to a Big4 industry specialist.

2.5 Hypotheses

As mentioned in the previous sections, prior studies on audit firm tenure suggest that audit quality increases with firm tenure, consistent with a “learning curve” for new auditors and the

(13)

12

critical importance of client-specific knowledge and experience, which can only be acquired over time with the client, for producing a high-quality audit. (Arruñada & Paz-Ares, 1997; Nashwa, 2004; Johnson et al., 2002). In addition, short tenure may involve higher risk for audit failures and fraudulent financial reporting (Carcello and Nagy, 2004).

So if this is the case, audit quality will decrease after an auditor switch in the first term (1 – 3 years) of the audit firm and client engagement. Assuming that the financial markets are efficient, I expect the investors to be able to appreciate the level of audit quality obtained over the auditor engagement term and thus depreciate the fact that a rotation will abolish all the obtained client specific knowledge and experience. Gosh and Moon (2005) showed that investors perceive an increase in audit quality as the tenure lengthens. This means that if the firm-client relationship is existing for 8 years or more, the investors appreciate this and will dislike the fact that the switch is made after such a long time as all the knowledge is getting lost. On the other hand, if the relationship only exist for a couple of years (2-3 years) when the switch is made, this would not have a negative effect on the market’s reaction. This because the investors have not seen any increase in audit quality during this short tenure. Based on the aforementioned I predict that:

H1: Investors perceive a decrease in audit quality when a switch is made after a long audit

firm tenure resulting in negative abnormal returns.

Furthermore, the Stewardship Theory implies that a manager would seek an auditor who satisfies the shareholder’s need for assurance. I expect that investors will perceive the switch as a positive aspect based on this assumption. This because the shareholders assume that the managers choose an audit firm which gives the best assurance to the financial information provided. Additionally, prior literature has shown that if an audit firm has industry specialization of the audited company, higher audit quality is obtained (Carcello & Nagy, 2004; Dunn & Mayhew, 2004; Gramling et al., 2000; Krishnan, 2003; Maletta & Wright, 1996). Given that, an increase in the industry knowledge of a firm’s external auditor is positively received by investors (Knechel et al., 2007; Gosh and Moon, 2005; Balsam et al., 2003).

This leads to the following hypotheses:

(14)

non-13

specialist audit firm to a specialist audit firm, resulting in positive abnormal stock returns.

Based on the argumentation of Gul et al. (2009) who found that the association between shorter auditor tenure and lower earnings quality is weaker for firms audited by industry specialists compared to non-specialists, I predict that:

H3: Investors perceive a smaller decrease in audit quality when a switch is made after a long

tenure if the switch is made from a non-specialist audit firm to a specialist audit firm.

3 Research Design

This archival study tests the relation between an audit firm switch to an industry specialist, and the abnormal returns. Prior research showed that audit firm rotations have an effect on the stock market returns of a firm. Therefore this study is focused on how the moderating variable, namely audit firm tenure and industry specialization, has an effect on the showed association. Figure 1 provides a schematic overview of the relationship that is tested.

Figure 1 Schematic Overview of tested association.

3.1 Measurement of Industry specialization

To test the second and third hypothesis a proxy for industry specialization is needed. Auditor industry specialization can be measured in terms of the audit firm’s market share and is considered a specialist in an industry if the audit firm has 30 percent market share of the industry’s total assets. A minimum of 30 companies per industry is needed to obtain a stable indicator of audit specialization (Craswell and Taylor, 1991, Craswell et al., 1995). I consider

(15)

14

a specialist in an industry if the audit firm has the largest share of the industry’s total assets. I calculate this by taking the audit firm’s total assets in the specific industry and divide this by the industry’s total asset. Additionally, to state that the auditor switch is made from a non-industry specialist to an non-industry specialist I require that the successor auditor has at least 30 percent market share in a given industry in the year that the switch is made at the U.S. national level. An industry is defined as all companies within each two-digit primary Standard Industry Classification (SIC) code.

TABLE 1

Auditors considered as Industry Specialists in Auditor Switches for Each Year during the Sample Period

(16)

15

TABLE 1 (continued)

SIC INDUSTRY 2005 2006 2007 2008 2009 2010 2011 2012

Agricultural DEL EY EY EY DEL EY NONE NONE

Business Services PWC NONE PWC PWC PWC PWC NONE PWC

Chemicals & Allied Products PWC PWC PWC,

KPMG

PWC PWC PWC PWC PWC

Coal Mining, Oil & Gas Drilling EY EY EY EY EY EY EY EY Communications EY EY EY EY EY EY NONE KPMG Detection, Laboratory, Surgical products PWC EY, PWC EY, PWC EY, PWC EY, PWC EY, PWC EY, PWC EY, PWC

Durable goods - Wholesale DEL DEL DEL DEL DEL DEL DEL EY

Eating and Drinking Places EY EY EY,

KPMG EY, KPMG EY EY, KPMG EY, KPMG EY, KPMG

Educational Services EY,

PWC EY, PWC PWC EY, PWC EY, PWC PWC EY PWC

Electric, Gas, Sanitary Services

DEL, PWC

DEL DEL DEL DEL DEL DEL DEL

Electronic & Other Electrical Equipment

PWC PWC PWC PWC PWC PWC PWC EY,

PWC

Engineering, Commercial & Mgt Services EY EY EY, KPMG EY, KPMG EY, KPMG EY, KPMG EY, KPMG EY, KPMG

Food & Beverages KPMG,

PWC KPMG, PWC KPMG KPMG KPMG KPMG, PWC KPMG, PWC KPMG, PWC

Health Services EY EY,

KPMG

EY EY EY EY EY EY

Hotels, Other Lodging Places EY EY EY EY EY EY EY EY

Machinery & Equipment EY EY EY PWC,

EY

EY EY EY EY

Mining - Metal, Gold & Silver PWC,

KPMG PWC, DEL PWC PWC PWC PWC, KPMG PWC, KPMG PWC, KPMG Misc Manufacturing Industries

PWC NONE NONE NONE NONE NONE None EY

SIC INDUSTRY 2005 2006 2007 2008 2009 2010 2011 2012

Mobile Homes & House Hold Furniture PWC, KPMG PWC, KPMG PWC, KPMG PWC PWC PWC PWC PWC

Motion Pictures DEL,

PWC DEL, PWC PWC DEL, PWC DEL, PWC DEL, PWC PWC DEL, PWC

Motor, Aircraft, Railroad eq DEL,

PWC

PWC PWC PWC PWC PWC PWC PWC

Nondurable goods - Wholesale EY EY,

DEL

(17)

16

Variable Definitions:

DEL = Deloitte Touche Tohmatsu; PWC = PricewaterhouseCoopers; and EY = Ernst & Young.

To be considered as an industry specialist in year t, the audit firms were required to either audit more than 30 percent of firms in an industry measured by the industry’s total asset.

I construct my measure of auditor industry specialization using information for 6,023 firms in 2005; 5797 firms in 2006; 5556 firms in 2007; 5384 firms in 2008; 5349 firms in 2009; 5266 firms in 2010; 5230 firms in 2011; and 4525 in 2012. As in prior studies I found that only Big 4 firms qualify as industry specialist. I only used the industries that are considered in my sample and identified 30 industries over the entire sample period 2005 – 2012. Table 1 gives an overview of the industry specialists in each year during my sample period per industry. This table shows that Ernst and Young is classified as a specialist the most often, followed by PricewaterhouseCoopers, Deloitte, and finally KPMG. The identification of industry specialists is not stable. Only four of the 30 industries have the same firm classified as the lone specialist in all eight years. There are also 17 industries that have two specialists during the eight year and another 4 industries have no specialist during several years.

3.2 Measurement of Perceived audit quality

Perceived audit quality is measured using abnormal returns which is the difference between actual and expected returns. In order to explore investors’ reaction to auditor switches, I examine three-day cumulative abnormal returns (CAR) around the date of the auditor switch. This narrow window is used because the association can be measured more reliably as other events that might have caused positive or negative abnormal returns are excluded.

Furthermore prior studies showed that the actual event date rather the 8-K filing date is the relevant event date to study auditor switching (Carter and Soo, 1999; Knechel, 2007).

Operative Builders & Constuctions

EY EY EY EY EY EY EY,

PWC

EY

Publishing & Printing PWC PWC PWC PWC PWC PWC PWC PWC

Retail – Other NONE DEL DEL,

EY

EY EY EY EY EY

Retail Stores EY EY EY EY EY EY EY DEL

Rubber & Misc Plastics & Glass,Concrete Pds

PWC PWC PWC PWC PWC PWC PWC PWC

Steel & Fabricated Metal Products DEL, PWC PWC PWC DEL, PWC DEL, PWC DEL, PWC DEL, PWC PWC

Textile Mill Products & Apparel

DEL DEL DEL EY EY NONE NONE NONE

(18)

17

Carter and Soo (1999) show that due to late frequent filing of the 8K filing prior research failed to find significant market reactions and that the informativeness of these dates is then severely affected. Therefore, I use the actual event date, the date on which the relationship with the predecessor auditor is officially terminated and a successor is appointed, to observe the market effect.

Additionally, Mackinlay (1997) states that there are two common choices for modeling the normal return, namely the constant mean return model and the market model. For this study I use the market model as prior studies have typically employed the market model to compute abnormal returns around auditor switches (Chaney et al. 2007, Knechel et al., 2007). 1

3.3 Measurement of Audit firm tenure

Prior literature has identified different ways to measure audit firm tenure such as using a continuous variable (Chen et al., 2008; Myers et al., 2003) or dichotomous variable (Johnson et al., 2002; Ghosh & Moon, 2005). Johnson et al. (2002) split up their sample into a short tenure (≤ 3 years), medium tenure (4 – 8 year), and long tenure (≥ 9 years). Their results document lower audit quality for short audit tenures but not for long tenures. In this study I use the dichotomous approach whereby I split up audit tenure into the same time spans as Johnson et al (2002). I look at the tenure of the predecessor auditor since 1980 and determine if the audit firm has been working for the company for a short, medium or long period until the switch. Within this approach I am able to test my hypotheses in which I am going to investigate whether audit firm rotation affects the abnormal returns if the rotation takes place

Mackinlay (1997) states that the abnormal return is the following for a firm i and event date t in which is controlled for the effects of market wide fluctuations: 𝐴𝐴𝐴𝐴_{𝑖𝑖𝑖𝑖}= 𝐴𝐴_{𝑖𝑖𝑖𝑖}− 𝐸𝐸 (𝐴𝐴_{𝑖𝑖𝑖𝑖} | 𝑋𝑋_𝑖𝑖 ) where,

𝐴𝐴𝐴𝐴𝑖𝑖𝑖𝑖 = abnormal return respectively for time period t, 𝐴𝐴𝑖𝑖𝑖𝑖 = actual return respectively for time period t, 𝐸𝐸 (𝐴𝐴𝑖𝑖𝑖𝑖 | 𝑋𝑋𝑖𝑖 ) = normal return respectively for time period t, 𝑋𝑋𝑖𝑖 = the conditioning information for the normal return model.

The concept of a cumulative abnormal return is necessary to accommodate the three day period event window and therefore the following model is used: 𝐶𝐶𝐴𝐴𝐴𝐴_(−1,+1)= ∑+1_{𝑖𝑖=−1}𝐴𝐴𝐴𝐴_{𝑖𝑖𝑖𝑖} where,

𝐶𝐶𝐴𝐴𝐴𝐴(−1,+1)= cumulative abnormal returns firm I , 𝐴𝐴𝐴𝐴𝑖𝑖𝑖𝑖= daily abnormal return for firm i on day t ,and 𝐴𝐴𝐴𝐴𝑖𝑖𝑖𝑖=_{𝑆𝑆𝑆𝑆}𝑆𝑆𝑆𝑆_{𝑖𝑖𝑖𝑖−1}𝑖𝑖𝑖𝑖 −_{𝑆𝑆𝑆𝑆}𝑆𝑆𝑆𝑆_𝑖𝑖−1𝑖𝑖 where,

𝐴𝐴𝐴𝐴𝑖𝑖𝑖𝑖 = Daily abnormal return for firm i on day t ,𝑆𝑆𝑆𝑆𝑖𝑖𝑖𝑖 = Stock price for firm i on date t , 𝑆𝑆𝑆𝑆𝑖𝑖𝑖𝑖−1 = Stock price for firm i on date t minus one day ,𝑆𝑆𝑆𝑆_𝑖𝑖 = Stock Index on date t ,𝑆𝑆𝑆𝑆_𝑖𝑖−1 = Stock Index on date t minus one day

(19)

18

after long tenure.

3.4 Control Variables

In order to test the relation I control for several factors. This section contains the motivation for the control variables used in this study.

Disagreement (DISAGREEi)

Davidson and Gribbin (1995) find evidence of negative market reactions to auditor changes when they are accompanied by auditor disagreements. Therefore, this study will examine only SEC registered companies as SEC registered companies are required to disclose the reason for auditor change in an 8-K report.

Hackenbrack and Hogan (2002) distinct four reasons for an audit firm switch namely, Service-, Disagreements-, Fee-, Uninformative (such as mergers and acquisitions) related reasons. I use a dummy variable with a value of 1 if the 8-K reveals that the auditor and company experienced a disagreement about accounting issues or service related issues at the time of switch, 0 otherwise. Fee and Uninformative reasons will be excluded from the sample.

Resignation (RESIGNi)

Although some of the resignations will be due to a disagreement there are still other reasons for resignations and therefore separately controlled for. Beneish et al. (2005) state that resignation announcements without a reason have an insignificant market response. Dunn et al. (1999) considered the effectiveness of the legislation in provision information to the market on the circumstances surrounding the resignation. They found a negative market response on the day of the resignation. Additionally, Zhang Shu (2000) argue that auditors have private information about their clients and therefore auditor resignations potentially signal the client’s high litigation risk or lack of growth causing the stock prices to drop. So in order to control for this I use a dummy variable with a value of 1 if the predecessor auditor resigns, 0 otherwise.

Timing (TIMINGi)

Schwartz and Soo (1996) demonstrate that auditor changes is one of the determinants for reporting delays. They argue that the optimal timing for an auditor change is early in the year

(20)

19

as it improves reporting timeliness which is perceived more positively by investors than switches late in the year. Therefore I use 1 if the auditor switch occurred in the fourth quarter of the fiscal year, 0 otherwise.

Size (SIZEi)

The inclusion of SIZE is motivated by the study of Diamond & Verrecchia (1991) which suggest that it is more likely that larger firms have better controls in place and have established well working financial reporting and disclosure practices. This results in less information asymmetry and the reported earnings of those firms are, ceteris paribus, perceived to be more reliable and relevant. Firm size therefore might explain different market reactions to auditor changes and thus affecting the stock returns (Eichenseher et al., 1989). Additionally, stock markets incorporate information of larger firms earlier in the share price than smaller firms and evidence is provided that the abnormal returns relation varies with firm size (Collins & Kothari, 1989).

Firm age (AGEi)

The inclusion of AGE is motivated by two reasons. First, previous studies argued that older firms are more likely to be stable and face less information asymmetry problems (Diamond & Verrecchia, 1991, Ghosh & Moon, 2005). Second, the probability exist that firm age and audit firm tenure are positively correlated, and by including this control variable the probability decrease that it leads to omitted correlated variable bias (Ghosh & Moon, 2005).

3.5 Research Design

The regression model I use to investigate the effect of audit firm rotation on the perceived audit quality, is derived from the model used in Knechel et al. (2007). I adjusted their model to fit my research:

CAR =

β

0

+

β

1INDUSTRY +

β

2NONINDUSTRY +

β

3SHORT-TENURE +

β

4

LONG-TENURE

+ β

5DISAGREE

+ β

6RESIGN

+ β

7TIMING

+

β

8SIZE

+ β

9AGE

+

ε

(1) This cross-sectional regression model specifies that a company’s three day cumulative abnormal return depends on the audit firm tenure, industry specialism and the specified control variables. Model (1) is used to test the first hypothesis in which I explore if abnormal

(21)

20

returns decrease when audit firms rotate after a long tenure. A negative coefficient for LONG- TENURE (β4) is consistent with my hypothesis that investors perceive a decrease in audit

quality when a switch is made after a long tenure.

This model is also used to explore if the switch from a non-industry to an industry specialist audit firms affects CAR positively or not. I expect a positive coefficient for INDUSTRY and a negative coefficient for NONINDUSTRY. In other words, I will test whether β1 is significantly greater than 0 and β2 is significantly less than 0. All other forms of auditor switch are included in the intercept of the regression model.

In order to test my third hypothesis I include two moderating variables to Model (1), namely SHORT-TENURE*INDUSTRY and LONG-TENURE*INDUSTRY. This cross-sectional regression model specifies that a company’s three-day cumulative abnormal return depends on the type of auditor switch that occurs moderated by the audit firm tenure and the specified control variables. I expect the coefficient β6 to be positive and β4 to be negative as

LONG TENURE is not moderated by INDUSTRY.

CAR =

β

0

+ β

1INDUSTRY +

β

2NON-INDUSTRY +

β

3SHORT-TENURE +

β

4

LONG-TENURE

+

β

5SHORT-TENURE * INDUSTRY +

β

6LONG-TENURE * INDUSTRY

+

β

7DISAGREE

+ β

8RESIGN +

β

9TIMING

+ β

10SIZE

+

β

11AGE+

ε

(2)

The variables used in the models are explained as follows:

Dependent variable:

CAR = Three day Cumulative Abnormal Return

Independent variables:

SHORT-TENURE = 1 if consecutive audit firm tenure in year t is 1- 3 years, 0 otherwise LONG-TENURE = 1 if consecutive audit firm tenure in year t is ≥ 9 years, 0 otherwise INDUSTRY = 1 if the auditor switch is from to a specialist Big 4 firm, 0 otherwise. NON – INDUSTRY = 1 if the auditor switch is from a specialist Big 4 firm to a

non-specialist Big 4 firm and 0 otherwise.

(22)

21

DISAGREE = 1 if company experienced a disagreement about accounting issues or service related issues at time of switch and 0 otherwise.

RESIGN = 1 if the audit firm switch was due to a resignation and 0 otherwise. TIMING = 1 if the auditor switch occurred in the fourth quarter of the fiscal year,

0 otherwise.

SIZE = the natural log of total assets in the year prior to the auditor switch. AGE = number of years the firm is a listed company.

3.6 Sample Construction

The data retrieved is based on the years 2005 till 2012. My choice for this period is motivated by two reasons. First, the fall of Arthur Anderson in 2002 influenced the abnormal returns of the companies that were audited by this firm (Chaney and Philipich, 2002) and might have influenced many other companies. Thus by selecting auditor switched after 2003 I eliminate the auditor switches due to the fall of this audit firm which took place in 2002 and 2003. Second, the implementation of the SOX Section 404 requirements on auditing of internal controls over financial reporting (ICOFR) in 2004, where companies that have an excess of 75 million of market capitalization are required to engage an auditor to audit their ICOFR (Chang et al., 2010). Due to this fact clients may have switch auditors because the audit of an ICOFR is very costly and the reason for switching may differ when firms consider the Section 404 requirements (Chang et al., 2010). So to eliminate these types of audit firm rotations I constructed my sample after 2005.

The sample consists of only SEC registered companies as these companies are required to disclose the reason for auditor change in an 8-K report. I first retrieved all the information from audit analytics and found 1295 auditor changes for the year 2005 – 2012 that involved only BIG4 audit firms and only US firms. Next, I excluded 46 observations that had confounding events accompanying the auditor switch such as mergers. I also excluded 97 firms operating in four financial service industries, namely SIC Codes 6000-6999. I did this because some explanatory variables employed in this studies such as this one may be inappropriate for these firms (Knechel, 2007; Krishnan, 1994). Finally, I dropped all the firms all changes that did not include information regarding tenure (auditor engagement date in Audit Analytics) and dates that are not considered accurate in Audit Analytics (Since event types” with the letter E, L and O).

(23)

22

The other variables used in this study were obtained from several databases; GROWTH and AGE from Thomson One, SIZE from DataStream and Compustat, and CAR from Eventus. There were about 40 companies who did not have the date of listing in Thomson One. In order to complete my dataset as much as possible I looked this up in Compustat and used the numbers of years the firm appeared in Compustat since 1950. After deletion of outliers and observations with incomplete data, the final sample consisted of 286 firm-year observations.

TABLE 2

No of Firms No of Industries

Initial sample: Auditor Switches during the period 2005 - 2012 1295 43

Less: Confounding events around the switch 46 1

1249 42

Less: Firms operating in financial services industries 97 4

1152 38

Less: Firms with unavailable tenure information 144 2

1154 36

Less: Firms with missing information CAR, SIZE & Outliers 868 2

286 34

* Less: Industries merged 4

Final Sample 286 30

4 Descriptive statistics

4.1 Independent variables

Table 3 presents the descriptive statistics of the full sample divided into the switches that occurred from a non-industry specialist to an industry specialist, and vice versa, and switches that had no specialists involved. The total sample consists of 286 observations and from this full sample 33% involved switches to an Industry specialist, 26 % to a non-specialist, and 41% involved no industry specialist switches at all. I find only 6 reported disagreements with the predecessor auditor in my full sample, whereby 2 disagreements in each type of switch. In total

* Four industries did not contain enough companies to conduct a stable indicator (>30). Therefore I merged these industries with common industries.

(24)

23

TABLE 3

Independent Variable Descriptive Statistics—Sample Consists of 286 Firms

TO

INDUSTRY

TO NON-

INDUSTRY OTHER Full Sample

No. of Observations in Sample 94 75 117 286

(33 %) (26 %) (41%) (100%)

Reported Auditor Disagreements 2 2 2 6

(2%) (2,5%) (1,7%) (2%)

Switch made in Fourth Quarter 17 11 17 45

(18%) (15%) (15%) (16%)

Predecessor Auditor Resigned 5 6 10 21

(5,3%) (8%) (8,5%) (7%)

Auditor Short Tenure 14 9 11 34

(15%) (12%) (9%) (12%)

Auditor Medium Tenure 30 24 40 94

(32%) (32%) (34%) (33%)

Auditor Long Tenure 50 42 66 158

(53%) (56%) (56,4%) (56%) Size (mln) Minimum 10,4775 9,6930 9,6763 9,6763 Maximum 17,2945 17,4901 16,5498 17,4901 Mean 13,9700 13,5739 13,6695 13,7432 Std. Devation 1,5484 1,7222 1,5588 1,6025

Firm age (years) Minimum 2,0 1,0 1,0 1,0

Maximum 59,0 56,0 56,0 59,0

Mean 21,394 24,507 19,162 21,297

Std. Devation 13,6423 14,9118 13,1935 13,9248

Variable Definitions:

− TO INDUSTRY = firms that switched from a non-industry specialist auditor to an industry specialist auditor; − TO NON- INDUSTRY = firms that switched from an industry specialist auditor to a non-industry specialist

auditor; and

(25)

24

21 auditors resigned whereby 5,3 % in the switch to specialist group, 8 % in the switch to non-specialist group, and 8,5% in the group that had no specialists involved. Also, from the total sample, 45 switches occurred in the last quarter of the year and the INDUSTRY has the highest percentage (18%). Additionally, 56% of the audit firm switches had a predecessor auditor for more than nine years, 33% for a medium period of 3 -8 years and 12 % for a short period of 1 -3 years. All the percentages between the different groups and different variables are comparable. There are no extremes found. This is also the case for the variables SIZE and FIRMAGE. The average SIZE mean for the full sample is 13.7 million, whereby 13.9 million of the group moving to industry specialist (IND), 13.5 million for the group moving to non-specialist (NON-IND), and 13.6 million for the group that involve other type of switches (OTHER). It seems that bigger firms, measured by total assets, tend to switch from an industry specialist to an industry specialist or from a non-industry to an industry specialist. Finally, I observe an average age of 21.3 years in the full sample, whereby 24.5 years in the IND group, 21.4 years in the NON-IND group, and 19.2 years in the OTHER group. It seems that firms audited by specialists are in general older and larger in size which is consistent with the suggestions of Gul et al. (2009).

4.2 Dependent variable

Table 4 presents the descriptive statistics of the dependent variable CAR. Again I divide the sample in BIG 4 firms that switch to specialist auditors (IND), that move away from specialist auditors (NON-IND), and switches that do not involve specialist auditors (OTHER). I perform an Independent Sample T-Test to test whether there is a difference between these groups. I am interested in this comparison as one of my predictions of this study is that abnormal stock returns are more positive when companies switch to a specialist auditors compared to companies that switch to a non-specialist. Consistent with my expectation, I find that companies switching to an industry specialists compared to companies that switch to non-specialist or have no non-specialist involved experience a positive mean CAR of .00480, while the other kind of switches experience a negative CAR of -.00310. Panel B compares firms that move away from specialist auditors (NON -IND) to firms that switch to specialist auditors and to firms and switches that do not involve specialist auditors (OTHER). At a significance level of 10 percent is showed that firms switching to non- specialist experience a more negative CAR (-.00414) compared to other switches (-.00078). Finally, the comparison between firms switching to industry specialist and firms switching to non-industry specialist in Panel C

(26)

25

shows at a significance level of 5 percent that in average switches to specialist auditor is more positive compared to switches to non-specialist. This is consistent with my second hypothesis and prior studies (Knechel et al., 2007; Gosh and Moon, 2005; Balsam et al., 2003). It is also of interest to note that all other switches, on average, reveal a slightly negative CAR.

TABLE 4

Three-Day Cumulative Abnormal Returns of Sample Firms around the Auditor Switch Event Dates—Sample Consists of 286 Firms

Panel A: Mean Comparison of firms that switch to specialist auditors (IND) to firms that move away from

specialist auditors and switches that do not involve specialist auditors (OTHER).

IND OTHER

Mean ,00480** -,00310**

N 94 192

Sig. (2 tailed) 0,002 (Equal variances assumed)

Panel B:Mean Comparison of firms that move away from specialist auditors (NON -IND) to firms that

switch to specialist auditors and to firms and switches that do not involve specialist auditors (OTHER).

NON -IND OTHER

Mean -,00414* -,00078*

N 75 211

Panel C: Mean Comparison of firms that switch to specialist auditors (IND) to firms that move away from

specialist auditors (NON- IND).

IND NON IND

Mean ,00480** -,00414**

N 94 75

(27)

26

4.3 Correlation Matrix

The correlations among all the variables are presented in table 5. This table shows that INDUSTRY is positively correlated with CAR and is significant at the .01 level. I expected this positive correlation, as I hypothesize that positive abnormal returns are associated with switches to industry specialists, thus it provides initial support for the study’s hypothesis. CAR is negatively correlated with RESIGN at a .05 significance level and is positively correlated with

SHORT TENURE at a .01 significance level. The negative correlation between CAR and RESIGN is also predicted as prior literature showed that resignations are perceived negatively by investors and negative market responses were found (Zhang Shu, 2000; Dunn et al, 1999).

Firm AGE is negatively correlated with SHORT TENURE (-.119, p<.05) and positively correlated with LONG TENURE (.224, p<0.01). Notable is that SIZE is negatively correlated with SHORT TENURE (-.173, p< .01) and positively correlated with LONG TENURE (.146, p<.05). Taking these two correlations together suggests that bigger firms tend to maintain a longer relationship with the audit firm. This is consistent with prior literature because bigger firms are more complicated which takes more time for an audit firm to acquire client specific knowledge. A switch after a short tenure would be therefore harmful for a big company as the successor auditor would again need a lot of time to get the same knowledge as the predecessor auditor. This implies that big firms understand the advantage of staying with the same auditor for a longer period when it comes to client specific knowledge.

Another notable correlation is the positive significant correlation between disagreement and short tenure (.172, p<0.01). This correlation can suggest two different things. First, this could mean that disagreements in for example accounting issues occur more often with a short tenure as the client firm relation is not that long established and therefore client specific knowledge is missing which might lead to disagreements. On the other hand, this could also mean that due to disagreements the tenure is short. This is an aspect for further research.

Furthermore, the correlations between all the other independent and control variables are generally between +/- 0.3 and multicollinearity does not appear to be a problem as the VIF values are less than 2 and the tolerance statistics are well above 0.2.2

2

(28)

27

TABLE 5

Correlation matrix of all the variables

**. Correlation is significant at the 0.01 level (2-tailed). *. Correlation is significant at the 0.05 level (2-tailed).

Variable definition

CAR = Cumulative Abnormal Return

SHORT-TENURE = 1 if consecutive audit firm tenure in year t is 1- 3 years, 0 otherwise LONG-TENURE = 1 if consecutive audit firm tenure in year t is ≥ 9 years, 0 otherwise INDUSTRY = 1 if the auditor switch is to a specialist Big 4 firm and 0 otherwise.

NON – INDUSTRY= 1 if the auditor switch is from a specialist Big 4 firm to a non-specialist Big 4 firm and 0 otherwise. DISAGREE = 1 if company experienced a disagreement about accounting issues, 0 otherwise

RESIGN = 1 if the audit firm switch was due to a resignation and 0 otherwise.

TIMING = 1 if the auditor switch occurred in the fourth quarter of the fiscal year, 0 otherwise. SIZE = the natural log of total assets in the year prior to the auditor switch.

AGE = number of years the firm is a listed company.

1 2 3 4 5 6 7 8 1. CAR 1 2. INDUSTRY ,187** 1 3. NON-INDUSTRY -,109 -,417** 1 4. SHORT-TENURE ,178** ,065 ,002 1 5. LONG-TENURE ,018 -,029 ,009 -,408** 1 6. RESIGN -,139* -,054 ,015 ,021 ,011 1 7. AGE ,012 ,005 ,138* -,119* ,224** -,030 1 8. SIZE ,066 ,099 -,063 -,173** ,146* ,011 ,194** 1 9. DISAGREE ,082 ,001 ,024 ,172** -,065 -,041 ,083 ,04 ,4 10. TIMING ,103 ,045 -,017 ,079 ,022 -,011 ,041 ,02 ,70

(29)

28

4.4 Empirical Results

The results of the multivariate regression analysis are presented in Table 6. I performed two regressions; model 1 presents the first regression without SHORT TENURE *INDUSTRY and LONG TENURE * INDUSTRY, and model 2 includes these variables.

I predicted a negative coefficient for LONG-TENURE (β4) as I hypothesize that

investors perceive a decrease in audit quality when a switch is made after a long tenure, resulting in negative abnormal returns. My results indicate a positive coefficient for LONG-TENURE of .004 which is not consistent with my hypothesis. However there is no evidence that LONG TENURE is actually positive because the coefficient is not significant. Conversely, the results show a positive coefficient of .013 with a significance level of 1 percent for SHORT TENURE. This suggests that investors do not perceive a decrease in audit quality when the company decides to hire a new auditor after having their current auditor for a short period. This is in line with my prediction as I expect that investors appreciate the level of audit quality obtained over the auditor engagement. In other words, investors appreciate the obtained client specific knowledge and experience during the years resulting in higher audit quality (ArruNanda & Paz-Arez, 1997; Carcello & Nagy, 2004; Johnson, et al., 2002). Gosh and Moon (2005) find that investors perceive auditor tenure as enhancing earnings quality. Therefore I predict that investors depreciate the fact that a rotation, when the audit firm is retained for a long period, abolishes all the obtained client specific knowledge and experience, which results in negative abnormal returns. A possible explanation for the positive coefficient of SHORT TENURE is therefore that investors do not perceive a decrease in audit quality after a short tenure. This because the auditors did not get the chance to get client specific knowledge and experience in the short period. In other words, there is no waste of quality and therefore the investors do not react negatively.

Hypothesis 2 predicts that abnormal returns are positive when companies switch to an industry specialist audit firm. Consistent with this expectation, the results show a positive coefficient of .006 at a significance level of 5 percent. This means that there is support for the second hypothesis. These results are consistent with the studies of Balsam (2003) and Knechel et.al (2007) in which they find that specialist auditors increase the markets’ perception of audit quality. Switches to a non-industry specialist are negatively received by investors as the parameter estimate of NON-INDUSTRY is negative (-.002). However, there is no evidence to state that switches to industry specialist are considered more positively by investors compared to switches to non-industry because the p-value of NON-INDUSTRY is not significant (p=.485).

(30)

29

TABLE 6

Cross-Sectional Regression of Three-Day Cumulative Abnormal Returns around Auditor Switches on the Independent Variables—Sample Consists of 286 Firms

CAR = β0 + β1INDUSTRY + β2NON-INDUSTRY + β3SHORT-TENURE + β4

LONG-TENURE +β5SHORT-TENURE * INDUSTRY + β6LONG-TENURE * INDUSTRY +

β7DISAGREE + β8RESIGN + β9TIMING + β10SIZE + β11AGE + ε

Model 1 Model 2 β p-value β p-value INTERCEPT -,018 ,084 -,016 ,128 INDUSTRY ,006 ,028** ,001 ,767 NON INDUSTRY -,002 ,485 -,002 ,499 SHORT TENURE ,013 ,001*** ,006 ,209 LONG TENURE ,004 ,109 ,003 ,405 DISAGREE ,005 ,502 ,009 ,296 RESIGN -,010 ,019** -,009 ,031** TIMING ,004 ,216 ,004 ,260 SIZE ,001 ,236 ,001 ,269 AGE -6,45 ,941 5,69 ,947

SHORT TENURE * IND ,017 ,042**

LONG TENURE * IND ,005 ,382

*, **, *** Significant at the 10 percent, 5 percent, and 1 percent levels, respectively. Variable definition

(31)

30

I also hypothesize that investors perceive a smaller decrease in audit quality when a switch is made after a long tenure if the switch is made to a specialist audit firm. I find a higher parameter of LONG TENURE *INDUSTRY (.005) compared to LONG TENURE (.003) but no support is found for this hypothesis and therefore it is rejected. On the other hand, I find evidence that abnormal returns are better when the switch is made to an industry specialist after a short tenure compared to a switch made after a short tenure when no industry specialist is involved. This is showed by the variable SHORT TENURE *INDUSTRY (.017, p<.05) which is higher than the parameter of the switch to an industry specialist after a short tenure without industry specialization involved.

Finally, RESIGN is significant and negative in both models, which is consistent with my expectation and prior literature (Zhang Shu, 2000; Dunn et al. 1999). This finding might imply that auditor resignations potentially signal the client’s high litigation risk or lack of growth causing the stock prices to drop (Zhang Shu, 2000).

4.5 Sensitivity analysis

I also performed a sensitivity test to determine if my primary results are sensitive to some specifications I used in my research design. I especially examine the sensitivity of my results by investigating whether the results are still the same when enlarging the window. Prior research suggests that the market may anticipate the change in auditor prior to the date on which firms actually makes the switch (Knechel et al., 2007). So instead of using three day cumulative abnormal returns I use a five day window (-3, +1). The results of these analysis are presented in Table 7 in which both models are included.

Model 1 presents the result for my first and second hypothesis. LONG TENURE is negative (-.004) with a significant level of 5 percent and SHORT TENURE is positive (.048, p<.005) which is consistent with my prediction, providing support for the first hypothesis. This implies that when the switch is made after a long tenure the market reaction is negative while with a short tenure the market reaction is positive. Prior research suggests that the audit quality increases with auditor tenure (Carcello and Nagy, 2004; Ghosh and Moon, 2003; Johnson et al., 2002; Myers et al., 2003). A negative coefficient for LONG TENURE suggest that investors acknowledge that the obtained audit quality will decrease at the time a new auditor is hired. As mentioned in the previous section, a positive coefficient for SHORT TENURE, suggests that investors have not seen an increase in audit quality and therefore they do not perceive a decrease in audit quality when the switch is made.

(32)

31

TABLE 7

Cross-Sectional Regression of Five-Day Cumulative Abnormal Returns around Auditor Switches on the Independent Variables

CAR = β0 + β1INDUSTRY + β2NON-INDUSTRY + β3SHORT-TENURE + β4

LONG-TENURE +β5SHORT-TENURE * INDUSTRY + β6LONG-TENURE * INDUSTRY +

β7DISAGREE + β8RESIGN + β9TIMING + β10SIZE + β11AGE + ε

Model 1 Model 2 β p-value β p-value INTERCEPT ,003 ,610 ,002 ,770 INDUSTRY ,003 ,159 ,006 ,082 NON INDUSTRY -,007 ,001*** -,007 ,001*** SHORT TENURE ,005 ,048** ,014 ,000*** LONG TENURE -,004 ,036** -,004 ,127 DISAGREE -,003 ,615 -,001 ,848 RESIGN -,007 ,006*** -,007 ,011** TIMING -,004 ,088* -,004 ,057* SIZE -8,48 ,986 -5,16 ,991 AGE 5,67 ,355 6,18 ,308

SHORT TENURE * IND ,016 ,005**

LONG TENURE * IND -,001 ,753

*, **, *** Significant at the 10 percent, 5 percent, and 1 percent levels, respectively. Variable definition