Amsterdam Business School
The effect of deficient PCAOB inspection reports on the likelihood of
auditor dismissals
Name: Jordy van de Vlag Student number: 10184139 Date: June 22, 2015 Word count: 16,711
Thesis supervisor: dhr. dr. J.J.F. van Raak
MSc Accountancy & Control, specialization Accountancy Faculty of Economics and Business, University of Amsterdam
Statement of originality
This document is written by student Jordy van de Vlag, 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 Economic and Business is responsible solely for the supervision of completion of the work, not for the contents.
Abstract
One of the elements of the Sarbanes-Oxley Act was the establishment of the Public Company Accounting Oversight Board (PCAOB). The PCAOB inspects auditors of public companies by reviewing some of the audits performed by them. The inspection process results in the preparation of an inspection report and part of the report is published by the PCAOB. The public portion of the report describes the audit deficiencies identified by the inspection team. The reports can be classified as GAAP-deficient, GAAS-deficient or non-deficient reports. This study focuses on the potential use of PCAOB inspection reports of triennially inspected auditors as a publicly available signal of audit quality. More specifically, this study focuses on the potential client reaction to the reports by examining whether clients of GAAP-deficient auditors are more likely to switch to a non-GAAP-deficient successor auditor. Furthermore, this study examines the moderating effect of second-round inspections on the association between GAAP-deficient inspection reports and the likelihood of client switching to a non-GAAP-deficient auditor. The results do not indicate that clients of non-GAAP-deficient, triennially inspected auditors are more likely to switch to a non-GAAP-deficient successor auditor. Furthermore, this study does not find a stronger association between the receipt of a GAAP-deficient inspection report and the likelihood of client switching to a non-GAAP-deficient auditor regarding second-round inspections.
Finally, this study examines potential client reaction to GAAS-deficient PCAOB inspection reports. The results do not indicate that clients of GAAS-deficient auditors are more likely to switch to a non-deficient, successor auditor. However, when distinction is made between successor auditors that are labelled as non-deficient because the auditor received a clean inspection report and successor auditors that are labelled as non-deficient because the auditor has not received an inspection report yet, this study finds that clients of GAAS-deficient auditors are more likely to switch to an auditor that has received a clean inspection report.
Table of contents
1.INTRODUCTION ... 5
2.LITERATURE REVIEW AND HYPOTHESES ... 8
2.1BACKGROUND ON THE PCAOB INSPECTION PROCESS ... 8
2.2PRIOR RESEARCH ON THE INFLUENCE OF THE PCAOB INSPECTION PROCESS ... 9
2.2.1CLIENT, MARKET AND AUDITOR RESPONSES TO THE PCAOB INSPECTION REPORTS ... 10
2.3HYPOTHESIS DEVELOPMENT ... 11
2.3.1THE RECEIPT OF A GAAP-DEFICIENT PCAOB REPORT AND CLIENT’S SWITCH TO A NON -GAAP DEFICIENT SUCCESSOR AUDITOR ... 12
2.3.2THE MODERATING EFFECT OF SECOND-ROUND PCAOB INSPECTIONS ... 13
3.RESEARCH METHODOLOGY ... 15
3.1REGRESSION MODEL ... 15
3.2VARIABLE DEFINITIONS ... 16
3.2.1DEPENDENT VARIABLE DEFINITION ... 16
3.2.2INDEPENDENT VARIABLE DEFINITIONS ... 16
3.3SAMPLE SELECTION ... 18
4.DESCRIPTIVE STATISTICS AND RESULTS ... 21
4.1DESCRIPTIVE STATISTICS ... 21
4.1.1GENERAL DESCRIPTIVE STATISTICS ... 22
4.1.2SWITCHING BEHAVIOUR OF GAAP-DEFICIENT AND NON-DEFICIENT CLIENTS ... 24
4.2RESULTS ... 28
4.2.1MULTIPLE REGRESSION RESULTS ... 28
4.2.2GAAS-DEFICIENT PCAOB INSPECTION REPORTS ... 32
4.2.2.1SAMPLE, DESCRIPTIVE STATISTICS AND RESULTS ... 32
4.2.3ADDITIONAL ANALYSES ... 38
4.2.3.1ADDITIONAL ANALYSIS ON THE SAMPLE PERIOD CHOSEN ... 38
4.2.3.2AUDITORS THAT AUDIT MORE THAN 20 PUBLIC CLIENTS VERSUS AUDITORS WITH 20 OR LESS PUBLIC CLIENTS ... 40
4.2.3.3THE INFLUENCE OF SWITCHES TO AUDITORS THAT HAVE NOT RECEIVED AN INSPECTION REPORT ... 42
4.2.3.4OTHER ELEMENTS IN THE INSPECTION REPORT THAT COULD INFLUENCE CLIENT REACTION ... 44
5.CONCLUSION ... 45
REFERENCES ... 48
1. Introduction
In response to extraordinary accounting and auditing scandals such as Enron and Worldcom, the U.S. Congress enacted the Sarbanes-Oxley Act in the United States in 2002 (Nagy, 2014). One of the elements of SOX was the establishment of the Public Company Accounting Oversight Board (PCAOB). The PCAOB is, among others, responsible for overseeing auditors of public companies and for performing inspections of the audits performed by these audit firms (Church & Shefchik, 2012). The PCAOB inspections are performed annually for Big 4 auditors and national auditors with more than 100 public clients. The inspection process is performed every three years by the PCAOB for auditors with less than 100 public clients, the so-called triennially inspected auditors (Daugherty, Dickins and Tervo, 2011). The PCAOB inspectors adopt a so-called ‘risk-based approach’, by which they do not inspect every audit engagement of an auditor (Church & Shefchik, 2012).More attention will be paid to this in section 2.1. The inspection process results in the preparation of the PCAOB inspection report. The public portion of the inspection report describes the audit deficiencies identified in the inspected audit engagements. Most inspection reports also include a non-public portion about the audit firm’s quality controls (Nagy, 2014). In this research, the focus is on the public portion of the inspection report.
According to Abbott, Gunny and Zhang (2013), inspection reports can be classified into three categories based on severity. In a clean report, the PCAOB finds no audit deficiencies (non-deficient report). Deficient reports issued by the PCAOB can be divided in GAAS-deficient and GAAP-deficient reports. In a GAAS-deficient report, the PCAOB states that the financial statements are free from material error and conform GAAP, but the audit process itself is not fully performed in accordance with GAAS-recommended procedures. For example, there was insufficient testing or inadequate documentation in the audit process (Church & Shefchik, 2012). In a GAAP-deficient report, the PCAOB states that the auditor failed to identify a material departure from GAAP. Among other factors, which will be elaborated in chapter 2, this variation of PCAOB inspection reports may lead to market response by investors (Offermanns & Peek, 2011) or client reaction in the form of auditor switching (Abbott et al., 2013; Acito, Hogan and Mergenthaler, 2014; Daugherty et al., 2011).
Audit quality is generally not directly observable, by which investors, clients and other stakeholders have to use proxies for audit quality that are publicly available (Barton, 2005). Therefore, consistent with Abbott et al. (2013), this study focuses on the potential use of the PCAOB inspection reports as a publicly available signal of audit quality. More specifically,
this study focuses on the potential client reaction to the PCAOB inspection reports in the form of auditor switching by examining whether clients are more likely to switch to a non-GAAP-deficient, triennially inspected auditor when their incumbent triennially inspected auditor received a GAAP-deficient PCAOB inspection report. The focus of this research is on triennially inspected auditors and their clients, as prior research indicates that there is limited variation in the PCAOB inspection reports for the Big-4 and national U.S. auditors (Church & Shefchik, 2012; Gunny & Zhang, 2013). They found that none of these auditors receive a clean PCAOB inspection report and all Big-4 auditors receive a GAAP-deficient PCAOB inspection report. Due to this lack of variation in the inspection reports, it is not of particular interest to focus on Big-4 auditors and national U.S. auditors. Furthermore, switching costs may be lower for clients of triennially inspected auditors, because these clients are smaller, the clients are able to amortize the switching costs over a three-year period as the auditors are inspected triennially and the auditor market for smaller auditors is more competitive (Ghosh & Lustgarten, 2006). This makes it more interesting to examine the reaction of these clients to GAAP-deficient PCAOB inspection reports as their incentive to switch to a non-GAAP-deficient auditor may increase due to the lower switching costs.
According to Acito, Hogan & Mergenthaler (2014), prior research on auditor and client reaction to PCAOB inspection reports and how these reports affect auditor-client relationships is limited. They examine, among others, whether audit deficiencies identified in the PCAOB inspections lead to changes in auditor turnover. However, the focus of their research is on PCAOB inspection reports of Big-4 audit firms. Prior research of Daugherty et al. (2011) and Abbott et al. (2013) already address the issue of auditor switching of clients from (GAAP-)deficient, triennially inspected auditors to non-(GAAP-)deficient, triennially inspected auditors. However, Abbott et al. (2013) focus in their research on the period 2005-2007 in which the PCAOB only performed some initial (first-round) inspections. Daugherty et al. (2011) also focus almost entirely on first-round PCAOB inspections. Daugherty et al. (2011) call for further research on the influence of second-round and later PCAOB inspection reports on the public audit client base of triennially inspected auditors. Hermanson and Houston (2009) and Daugherty et al. (2011) argue that the existence of audit deficiencies for triennially inspected auditors may be less in second-round inspections. However, there is not yet a clear indication about the association between second-round PCAOB inspection reports and client’s auditor changes. This study addresses this issue by examining the moderating effect of second-round PCAOB inspections on the association between GAAP-deficient
PCAOB inspection reports and the likelihood of client’s switching to a non-GAAP-deficient, triennially inspected auditor.
This study predicts a positive association between the receipt of a GAAP-deficient PCAOB inspection report and the likelihood of client’s switching to a non-GAAP-deficient, triennially inspected auditor and predicts a stronger positive association with respect to second-round PCAOB inspections. However, the findings do not correspond to these predictions. Firstly, this study does not find that clients of GAAP-deficient, triennially inspected auditors are more likely to switch to a non-GAAP-deficient, triennially inspected successor auditor, irrespective of whether the successor auditor is labelled as non-GAAP-deficient because he received a non-GAAP-non-GAAP-deficient inspection report or because he has not received an inspection report yet. In addition, this study does not find a stronger positive association between receiving a GAAP-deficient inspection report and the likelihood of client switching to a non-GAAP-deficient successor auditor with respect to second-round inspections. Finally, this study provides some evidence that clients have an incentive to switch to another auditor in order to obtain a more favourable going-concern opinion.
This study contributes to the literature by providing initial evidence on the effect of second-round PCAOB inspections on the association between the receipt of a GAAP-deficient inspection report and the likelihood of client’s switching to a non-GAAP-GAAP-deficient, triennially inspected auditor. In addition, this study is the first that makes a distinction between successor auditors that are labelled as non-GAAP-deficient (or non-deficient) because the auditor received a non-GAAP-deficient (or non-deficient) PCAOB inspection report and successor auditors that are labelled as non-GAAP-deficient (or non-deficient) because the auditor has not received an inspection report yet. The findings of this study regarding the GAAS-deficient sample indicate that this distinction does matter. More specifically, this study does not find that clients of GAAS-deficient auditors are more likely to switch to a non-deficient successor auditor. However, when the above-mentioned distinction is made, this study finds that clients of GAAS-deficient auditors are more likely to switch to an auditor that has received a non-deficient PCAOB inspection report. Making this distinction may also have consequences for the results of Abbott et al. (2013), as their sample is focused on the initial years of the PCAOB inspection program (2005-2007), by which it is likely that many switches in their research are related to successor auditors that have not received an inspection report at the time of the switch.
The remainder of this study is organized as follows. Chapter 2 provides a review of prior literature and defines the hypotheses. Chapter 3 describes the study’s research design
and sample selection. Chapter 4 provides descriptive statistics and presents the results and additional analyses and chapter 5 provides the study’s conclusion.
2. Literature review and hypotheses
In this chapter, the background with respect to the PCAOB inspection process will be presented and insights from prior literature on the influence of the PCAOB inspection process will be provided. This will ultimately lead to the development of the hypotheses in the final paragraph.
2.1 Background on the PCAOB inspection process
In 2002, the Sarbanes-Oxley Act was enacted in the United States (Nagy, 2014). Section 101 of SOX established the Public Company Accounting Oversight Board. The PCAOB has four major tasks to focus on: registration, enforcement, inspections and standard setting (Abbott et al., 2013). The most relevant task of the PCAOB with respect to this study is the performance of inspections. According to section 104 of SOX, the PCAOB is responsible for conducting inspections of each public audit firm of public companies. These inspections of audit firms by the PCAOB replaced the self-regulated peer review program (Nagy, 2014). As stated in chapter 1, these PCAOB inspections are performed annually for audit firms with more than 100 publicly held clients and triennially for audit firms with 100 or fewer public clients (Daugherty et al., 2011).
The PCAOB inspectors select audit engagements for inspection based on an internally developed risk model, by which only a certain number of audit engagements of a specific audit firm are inspected (Nagy, 2014). With respect to these selected audit engagements, inspectors examine the quality of the audit work performed by the auditor and evaluate the quality control system of the auditor (Church & Shefchik, 2012; Abbott et al., 2013). The inspectors inform the auditor during the inspection process about deficiencies that were identified by the inspection team. Stakeholders are not informed during the inspection process, by which they know the inspection results only after public disclosure of the report.
At the end of the inspection process, the PCAOB inspectors prepare and issue an inspection report. The deficiencies identified in the auditor’s audit work are described in the public portion of the PCAOB inspection report (Church & Shefchik, 2012). As mentioned in chapter 1, there are two types of deficient reports issued: GAAP-deficient and
GAAS-deficient inspection reports. Deficiencies identified during the evaluation of the quality control system of the auditor are described in the non-public portion of the inspection report (Nagy, 2014). This section of the report only becomes public when the auditor fails to make reasonable progress to address and remediate the respective deficiencies to the satisfaction of the PCAOB within 12 months after the issuance of the PCAOB report (Nagy, 2014). From now on, only the public portion of the PCAOB inspection report is of particular interest in this study.
2.2 Prior research on the influence of the PCAOB inspection process
In this paragraph, distinction is made between research that focuses on the association between the PCAOB inspection process and actual audit quality and research that focuses on the PCAOB inspection reports as a signal of audit quality.
Several studies focus on the association between the PCAOB inspection process/reports and actual (improvements in) audit quality. For example, Carcello, Hollingsworth and Mastrolia (2011) examine whether the PCAOB inspection process results in an improvement in audit quality by focusing on the Big-4 audit firms. Their results provide evidence that the inspection process has led to an improvement in audit quality, measured by a significant reduction in client’s use of abnormal accruals following a PCAOB inspection of the accompanying audit firm. Furthermore, Gunny & Zhang (2013) examine whether the PCAOB inspections are able to distinguish actual audit quality. With respect to triennially inspected auditors, they find that the PCAOB inspection reports are associated with lower actual audit quality (measured by restatement frequency and the use of abnormal accruals by clients) when the inspection reports are ‘seriously deficient’ (GAAP-deficient PCAOB inspection reports).
On the other hand, there is some prior research that focuses on the PCAOB inspection process/report as a potential publicly available signal of audit quality. According to Barton (2005), audit quality is not directly observable, by which publicly available proxies are necessary. Lennox and Pittman (2010) argue that the PCAOB inspection reports are not informative to clients. If clients perceive the PCAOB inspection reports as a valuable signal of audit quality, they would switch to an auditor that receives a clean (favourable) report (Lennox & Pittman, 2010). In their research, no association is found between deficient PCAOB inspection reports and client’s audit firm choice, by which they argue that clients do not perceive the PCAOB inspection reports as being valuable in signalling audit quality.
However, Abbott et al. (2013) argue that the inspection reports of the PCAOB may potentially serve as a publicly available proxy of perceived audit quality, thereby contrasting the results of Lennox and Pittman (2010). More specifically, they focus on the potential use of GAAP-deficient PCAOB inspection reports of triennially inspected auditors as a perceived audit quality signal for the clients of these auditors. Their results suggest that the clients of these GAAP-deficient, triennially inspected audit firms are using the inspection reports as a signal of audit quality as these clients react to these reports in the form of switching to a non-GAAP-deficient, triennially inspected auditor (The topic of stakeholder reaction will be elaborated further in paragraph 2.2.1). The contrasting results of both researches may be explained by the inclusion of annually inspected audit firms in the research of Lennox and Pittman (2010).
2.2.1 Client, market and auditor responses to the PCAOB inspection reports
In this paragraph, the focus is on market, client and auditor responses to the inspection reports published by the PCAOB. As stated earlier, the PCAOB does not communicate their inspection results to clients of the audit firm or other stakeholders during the inspection process (Abbott et al., 2013). Therefore, the PCAOB inspection report is (after publication) the first source of information for these stakeholders regarding the inspection process and the audit deficiencies identified and may lead to stakeholder reaction after publication. Prior research has focused on the reaction of different stakeholders (e.g. audit clients, investors) to these PCAOB inspection reports (Offermanns & Peek, 2011; Daugherty et al., 2011; Acito et al., 2014; Abbott et al., 2013). The auditor itself may also react to the deficiencies disclosed in the inspection report (Gramling, Krishnan and Zhang, 2011; Acito et al., 2014).
First of all, Offermanns and Peek (2011) focus on the investor (market) response to the issuance of the PCAOB inspection reports. They observed significant investor reaction to the PCAOB inspection reports in their study. More specifically, they provide evidence that the stock returns of the clients audited by the inspected audit firm exhibit abnormal variance around the publication date of the PCAOB inspection report, which may indicate that the inspection reports are informative to investors.
Secondly, Daugherty et al. (2011) focus on the client responses to the PCAOB inspection reports in the form of auditor switching. They provide evidence that the clients of deficient, triennially inspected auditors dismiss their auditor and are more likely to switch to a triennially inspected auditor without a deficient PCAOB inspection report. Consistent with Daugherty et al. (2011), Abbott et al. (2013) also focus, as mentioned earlier, on client
responses to the PCAOB inspection report and find that clients of GAAP-deficient, triennially inspected auditors are more likely to switch to a non-GAAP-deficient triennially inspected auditor within one year of the public disclosure of the PCAOB inspection report. In addition, Acito et al. (2014) examined whether clients react to the deficiencies identified in the PCAOB inspection reports in the form of Big-4 auditor turnover. They provide evidence that clients are more likely to switch to another Big-4 auditor when their incumbent Big-4 auditor has significant deficiencies (relative to the other Big 4 auditors) in areas that are important to the client’s accounting.
Finally, as mentioned before, the audit firm itself may also react to the PCAOB inspection reports. Gramling et al. (2011) look at auditor reaction by examining whether audit deficiencies identified by the PCAOB are associated with a change in going concern decisions made by triennially inspected auditors for their clients. Their results indicate that triennially inspected auditors with a deficient PCAOB inspection report are more likely to issue a going-concern opinion for their financially distressed clients subsequent to the publication of their PCAOB inspection report. In addition, Acito et al. (2014) examined whether auditors react to the audit deficiencies identified in the PCAOB inspection reports in the form of changing the audit fees charged. Their results do not indicate that the audit fees charged significantly changed for clients with significant exposure to the areas in which audit deficiencies were identified.
2.3 Hypothesis development
In this paragraph, the hypotheses of this study are developed and described. Consistent with Daugherty et al. (2011) and Abbott et al. (2013), the hypotheses are build on the basis of the potential use of the GAAP-deficient PCAOB inspection report as a publicly available signal of audit quality. Section 2.3.1 presents the development of the first hypothesis by describing the main factors (e.g. variation, accessibility and objectivity) that suggest why audit clients could view a GAAP-deficient PCAOB inspection report as a signal of audit quality and why clients may react (in this study in the form of switching to a non-GAAP-deficient, triennially inspected auditor). Section 2.3.2 presents the development of the second hypothesis by introducing the potential moderating effect of second-round PCAOB inspections.
2.3.1 The receipt of a GAAP-deficient PCAOB report and client’s switch to a non-GAAP deficient successor auditor
Abbott et al. (2013) present five factors that clarify why stakeholders (in this case audit clients) could use a GAAP-deficient PCAOB inspection report as a publicly available signal of audit quality and may react accordingly. First of all, the PCAOB inspectors that conduct the inspections can be characterized as independent and experienced inspectors. These PCAOB inspectors are not allowed to be active auditing practitioners (Church & Shefchik, 2012). Secondly, consistent with the concept of signalling, a signal’s value and informativeness is likely to be dependent on its variation. So, in order to be a potential signal of audit quality, there has to be variation in the various published PCAOB inspection reports. This also clarifies why this study is focused on the use of PCAOB inspection reports of triennially inspected auditors as a signal of audit quality. With respect to these auditors, there is variation in the severity of their PCAOB inspection reports. As mentioned in chapter 1, there is limited variation in the severity of the PCAOB inspection reports of Big-4 and national auditors. Thirdly, DeAngelo (1981) states that audit quality is defined as the joint probability that a given auditor will discover a breach in the client’s accounting system and will report this breach. As discussed earlier, a GAAP-deficient PCAOB inspection report suggests that the respective auditor has failed to prevent or discover a material departure from GAAP by their audit client, which is consistent with the above concept of audit quality of DeAngelo (1981). Fourthly, the PCAOB inspection reports are publicly available at the website of the PCAOB. In order to be a potentially valuable signal of audit quality, the PCAOB inspection reports need to be publicly available for the clients, because otherwise it is not possible for the client to consider the results of the inspection process and to use the inspection reports in evaluating audit quality and making auditor choice decisions. The final factor that clarifies why clients could view a GAAP-deficient PCAOB inspection report as a signal of audit quality is about the specificity of the inspection reports. The scope of the PCAOB inspections is wide due to the fact that the PCAOB has access to many confidential documents and clients, by which the inspection report is very specific about the particular audit deficiencies identified in the inspection process.
In addition, Acito et al. (2014) argue that clients may react to a GAAP-deficient inspection report (by switching to another auditor) if the report shows that the auditor has specific deficiencies in an area that is important to the client’s accounting (e.g. due to a lack of expertise). More specifically, if the inspection report shows that an auditor has audit
deficiencies regarding a particular accounting standard and if the client frequently uses this accounting standard in their financial reporting, the client may decide to switch to another auditor (that is not deficient on this standard) to ensure that their accounting information related to that standard is audited well.
Finally, according to Barton (2005), clients may also react because of their concerns about the reputation of their auditor and their own reputation. If clients are associated with an auditor that receives a deficient PCAOB inspection report, they may have an incentive to switch to an auditor that receives a clean report in order to protect their own reputation for credible financial reporting. All the above-mentioned factors suggest the potential for client reaction to the PCAOB inspection reports in the form of switching from a triennially inspected auditor that has received a GAAP-deficient report to a non-GAAP-deficient, triennially inspected auditor. This results in the first hypothesis of this study:
H1: There is a positive association between the receipt of a GAAP-deficient PCAOB
inspection report and the likelihood of client’s switching to a non-GAAP-deficient, triennially inspected auditor.
2.3.2 The moderating effect of second-round PCAOB inspections
With respect to the second hypothesis of this study, second-round PCAOB inspections are introduced as a moderator variable. More specifically, this hypothesis is focused on the moderating effect of second-round PCAOB inspections on the association between GAAP-deficient PCAOB inspection reports and client’s switching behaviour.
On the one hand, audit firms may resolve various audit deficiencies identified in their initial inspections (including GAAP-deficiencies) in order to show in second-round PCAOB inspections that they are improving their audit work. With respect to this, Acito et al. (2014) argue that auditors want to address their deficiencies, because if they fail to do this, the PCAOB may judge this failure as a lack of quality control. With regard to this incentive to improve, Hermanson & Houston (2009) and Daugherty et al. (2011) argue that the rate of deficiencies for triennially inspected auditors may be significantly less in second-round inspections. If auditors demonstrate in second-round inspections that they are resolving various audit deficiencies, clients may become more confident that these auditors will do everything to improve their audit work even further. As a result of this, clients may be less likely to switch to another auditor following these second-round PCAOB inspections. It may
be even the case that auditors have solved all of their deficiencies, by which they receive clean PCAOB inspection reports following the second-round inspections.
On the other hand, even though auditors may demonstrate that they are improving their audit work, clients may have a stronger incentive to switch to a non-GAAP-deficient auditor after second-round inspections, because they may not be willing to be associated with an auditor that receives a GAAP-deficient inspection report for the second time. Auditors may again receive a deficient PCAOB inspection report after the second-round inspection when some of their audit deficiencies are still present or when other new audit deficiencies are identified. Clients may give their auditor the benefit of the doubt after the auditor’s initial deficient report, by which they remain loyal to their auditor. For example, Abbott et al. (2013) find that less than half (44.3 percent) of the clients dismissed their auditor following the public disclosure of the initial GAAP-deficient inspection report. However, capital market participants (such as investors) get even less assurance that client firms are reporting in accordance with GAAP due to the publication of the auditor’s second GAAP-deficient inspection report. In order to avoid growing concerns among these capital market participants about the credibility of their financial reporting and to protect their own reputation, clients may not be willing to remain loyal to their (GAAP-)deficient auditor and may be more likely to switch (Barton, 2005).
Furthermore, it may be the case that an auditor receives a non-GAAP-deficient PCAOB inspection report in the first-round and a GAAP-deficient report in the second-round, which indicates a decrease in the quality provided by the incumbent auditor (Abbott et al., 2013). Clients may be unwilling to be associated with such an auditor (e.g. in order to protect their reputation) and therefore they may have an increased incentive to switch to a non-GAAP-deficient auditor following second-round inspections.
Overall, the expectation of this study is that clients have a stronger incentive to switch to a non-GAAP-deficient, triennially inspected auditor following second-round PCAOB inspections, because being associated with an auditor that receives a GAAP-deficient report for the second time or with an auditor that moves from non-deficient to GAAP-deficient will lead to increased concerns about their financial reporting credibility and growing harm to their reputation. This leads to the second hypothesis of this study:
H2: With respect to second-round inspections, there is a stronger positive association between
the receipt of a GAAP-deficient PCAOB inspection report and the likelihood of client’s switching to a non-GAAP-deficient, triennially inspected auditor.
3. Research methodology
Paragraph 3.1 presents the regression model that will be used to test the hypotheses. Paragraph 3.2 provides a more detailed description of the variables that are included in this model. Finally, paragraph 3.3 provides an overview of the sample selection process.
3.1 Regression model
Consistent with Abbott et al. (2013) and Barton (2005), a logistic regression will be used to model the relationship between the study’s dependent variable (SWITCH) and the independent variables. The model is presented below:
SWITCH = β0 + β1DEFRPT + β2SECOND + β3SECOND×DEFRPT + β4AUDITOR_SIZE +
β5AUDITOR_SIZE×DEFRPT + β6GOING_CONCERN +
β7GOING_CONCERN×DEFRPT + β8CLIENT_SIZE +
β9CLIENT_SIZE×DEFRPT + β10LOSS + β11LOSS×DEFRPT + ε
Where:
SWITCH = dummy variable coded ‘1’ for clients that dismissed their triennially inspected auditor and switched to a non-GAAP-deficient, triennially inspected auditor within one year after the public disclosure of the PCAOB inspection report and ‘0’ otherwise (from AuditAnalytics).
DEFRPT = dummy variable coded ‘1’ for clients whose triennially inspected auditor received a GAAP-deficient PCAOB inspection report and ‘0’ for clients whose triennially inspected auditor received a clean PCAOB inspection report (from PCAOB inspection reports, hand-collected).
SECOND = dummy variable coded ‘1’ if it is a second-round PCAOB inspection for the triennially inspected auditor and ‘0’ if it is a first-round inspection (from PCAOB inspection reports, hand-collected).
AUDITOR_SIZE = dummy variable coded ‘1’ when the current triennially inspected auditor audits fewer than five public clients and ‘0’ otherwise (from PCAOB inspection reports).
GOING_CONCERN = dummy variable coded ‘1’ in instances where the incumbent auditor’s opinion contains an explanatory paragraph regarding the going-concern assumption and ‘0’ otherwise (from AuditAnalytics).
CLIENT_SIZE = the natural log of total assets (from Compustat or AuditAnalytics)
LOSS = dummy variable coded ‘1’ when the client reports a loss (when net income is negative) and ‘0’ when the client reports a profit (when net income is positive) (from Compustat or AuditAnalytics).
3.2 Variable definitions
3.2.1 Dependent variable definition
As stated above, the dependent variable, SWITCH, is a dummy variable coded ‘1’ for clients that dismissed their triennially inspected auditor and hired a non-GAAP-deficient successor auditor within one year after the inspection report was publicly disclosed on the PCAOB website and ‘0’ otherwise. If a client does not switch, the dependent variable is coded ‘0’. Furthermore, if a client switches to another GAAP-deficient triennially inspected auditor (lateral switch), the dependent variable will also be coded as ‘0’. Finally, the dependent variable will also be coded as ‘0’ if the client switches to a Big-4 or national auditor1.
This study uses a period of one year subsequent to the public disclosure of the PCAOB inspection report, because auditor appointment decisions are typically made within one year (Lennox & Pittman, 2010). Finally, consistent with Hilary and Lennox (2005) and Abbott et al. (2013), auditor resignations are not included in this research, because these resignations are switches that are auditor-initiated. In these cases, the audit firm itself decided to stop being auditor of the client firm.
3.2.2 Independent variable definitions
As stated earlier, DEFRPT covers whether a triennially inspected auditor of a certain company received a GAAP-deficient PCAOB inspection report. A triennially inspected auditor is labelled as GAAP-deficient if he failed to identify, or to address appropriately, a
1 As stated earlier, prior research indicates that Big 4 and national auditors do not receive clean PCAOB
departure from GAAP that related to potentially material misstatements in the audited financial statements of the client. An auditor is labelled as non-deficient (‘clean’) if the PCAOB inspection team did not identify any audit performance issues that resulted in the auditor failing to obtain sufficient appropriate evidence to support its opinion on the client’s financial statements.
Furthermore, the second hypothesis focuses on second-round PCAOB inspections. Therefore, a dummy variable is introduced in the model to show whether it is a first-round or a second-round PCAOB inspection (SECOND). As stated in paragraph 2.3.2, clients do not want to be associated with an auditor that is (GAAP)-deficient for the second time. Moreover, if an auditor is labelled as non-GAAP-deficient in the first-round and GAAP-deficient in the second-round, this may indicate a decrease in quality the auditor delivers, which may increase the client’s incentive to dismiss the auditor (Abbott et al., 2013). Therefore, this study expects that receiving a GAAP-deficient PCAOB inspection report in the second-round (SECOND×DEFRPT) increases the likelihood of client’s switching to a non-GAAP-deficient, triennially inspected successor auditor.
In addition, there may be other factors that influence the client’s decision to switch to another audit firm. These factors do not necessarily relate to whether the audit firm is labelled as GAAP-, GAAS- or non-deficient. For example, according to Ettredge et al. (2007), opinion shopping may play a role in the client’s decision to switch to another auditor. Clients may switch to another auditor in order to obtain a more favourable going-concern opinion. Therefore, this study includes the variable GOING_CONCERN. This study expects a positive association between GOING_CONCERN and the likelihood of switching to another auditor in general. However, consistent with Abbot et al. (2013), this study does not provide an expectation on the influence of GOING_CONCERN on the likelihood of dismissing a GAAP-deficient, triennially inspected auditor. Moreover, this study includes variables to control for the size of the auditor (AUDITOR_SIZE) and the size of the client (CLIENT_SIZE). In case of the size of the auditor, if the auditor receives a GAAP-deficient PCAOB inspection report and if the auditor has only a few public clients, there is a greater likelihood for each client that the audit deficiencies found by the inspection team relate to the audit of their financial statements (the client does not know this for sure as client identities are not published in the inspection report). Therefore, consistent with Abbott et al. (2013), this study expects that the likelihood of switching to a non-GAAP-deficient, triennially inspected auditor increases when the incumbent auditor audits a few public clients. With respect to CLIENT_SIZE, as the size of the client increases, management’s actions become more difficult to observe, thereby
increasing the need for a firm to signal higher audit quality (DeFond, 1992). From this perspective, consistent with DeFond (1992) and Abbott et al. (2013), this study expects that CLIENT_SIZE increases the likelihood of switching to a non-GAAP-deficient, triennially inspected auditor. However, switching costs may be higher as the size of the client increases (Ghosh & Lustgarten, 2006). Due to the higher switching costs, it may be less likely that clients switch to a non-GAAP-deficient, triennially inspected auditor. Finally, this study controls for the profitability of the client by including the dummy variable LOSS. A client may decide to switch to another auditor if it has reported a loss in their financial statements and/or if it expects to face difficulties in becoming profitable in the coming year(s). If a firm switches to a new auditor, it takes time for the new auditor to gather the same level of client-specific knowledge as the incumbent auditor (Bedard & Johnstone, 2010). Management may take advantage of the information asymmetry and may engage in earnings management in order to avoid reporting losses again. Based on this, this study expects a positive association between LOSS and the likelihood of switching to another auditor in general. However, the study does not provide an expectation on the influence of LOSS on the likelihood of dismissing a GAAP-deficient, triennially inspected auditor.
3.3 Sample selection
First of all, this research starts with obtaining PCAOB inspection reports from triennially inspected auditors in the United States. The triennially inspected auditors that received a first-round2 and a second-round PCAOB inspection report are included in this research and their reports are analysed. The first-round inspection reports, included in this research, are published by the PCAOB from January 21, 2005 to December 22, 2008 and the second-round inspection reports of the triennially inspected auditors are published from September 24, 2007 to December 22, 2011. The total sample period will be 2005-2012 in order to cover the first-round and second-first-round PCAOB inspections as well as to cover the potential for subsequent client reaction in the form of auditor switching after publication of the PCAOB inspection reports. With respect to the PCAOB inspection reports, the following data are captured: name of the audit firm, publication date of the inspection report, number of public clients, whether the triennially inspected audit firm is GAAP-, GAAS- or non-deficient, the number of audit
2 First-round inspections are mainly conducted in the period 2005-2008, starting from January 21, 2005
deficiencies identified, the number of audit engagements reviewed by the inspection team and the number of audit engagements reviewed in which deficiencies are identified.
An overview of the distribution of the triennially inspected auditors is presented in Table 1. A total of 3863 triennially inspected auditors received a first-round and a second-round inspection report between January 21, 2005 and December 22, 2011. In the first-second-round, 49 audit firms were labelled as GAAP-deficient, 163 audit firms as GAAS-deficient and 174 audit firms as clean auditors. In the second-round, 44 audit firms received a GAAP-deficient inspection report, which is not significantly different from the number of GAAP-deficient auditors in the first-round (p-value: 0.580 > 0.05). Furthermore, the number of audit firms that received a GAAS-deficient inspection report decreased significantly in the second-round (p-value: 0.000 < 0.01). Finally, the number of audit firms that received a clean inspection report increased significantly in the second-round (p-value: 0.000 < 0.01).
TABLE 1
Distribution of the triennially inspected audit firms First-round PCAOB inspections Second-round PCAOB inspections Z-score P-value (P > |z|) GAAP-deficient 49 (12.7 %) 44 (11.4%) 0.553 0.580 GAAS-deficient 163 (42.2%) 85 (22.0%) 6.012 0.000 Non-deficient (clean) 174 (45.1%) 257 (66.6%) -6.016 0.000 Total 386 (100%) 386 (100%) - -
Knowing the respective GAAP-, GAAS- and non-deficient triennially inspected auditors, the public clients of these auditors are identified by looking at the audit opinions provided by these auditors (using AuditAnalytics) before the publication date of their PCAOB inspection report. These client portfolios are used to look at the client’s switching behaviour in the year after the publication of the auditor’s PCAOB inspection report. Table 2 provides an overview of the client firm observations used in this research. First of all, a full sample and a reduced sample (including some financial statement information of the client) are used to test the hypotheses, because for a significant amount of companies, financial statement data for the required year(s) is not available in either Compustat or AuditAnalytics. Furthermore,
3 In some cases, a company ceased to be audit client before the inspection, by which the triennially inspected
auditor (with only one issuer audit client) had no issuer audit client at time of inspection. These auditors are not included in this research as it is not possible to look at client switching behaviour for these auditors with no issuer audit client.
consistent with Abbott et al. (2013), clients in the financial industry are not included in this research. After deducting these client observations, the full sample consists of 895 GAAP-deficient, 2103 GAAS-deficient and 1473 non-deficient client observations. The reduced sample consists of 462 GAAP-deficient, 1117 GAAS-deficient and 943 non-deficient client observations.
TABLE 2
Overview of the client observations GAAP-deficient client observations GAAS-deficient client observations Non-deficient client Observations Initial sample 1167 2832 2367 Less: financial companies (or SIC-code not available) (272) (729) (894) Full sample 895 2103 1473
Less: total assets not available
(374) (888) (491)
521 1215 982
Less: net income not available
(59) (98) (39)
Reduced sample 462 1117 943
Table 3 provides an overview of the client observations per industry (for the full sample). The industry distribution largely corresponds to the industry distribution of Abbott et al. (2013). For example, the four largest industries (consumer product and food, energy, information and communication and manufacturing) in Abbott et al. (2013) are also the largest industries represented in this research.
Finally, important to note is that the non-deficient (clean) client observations are used as a control sample in order to determine if client firms react differently to the different types of PCAOB inspection reports (GAAP, GAAS or clean). A client whose triennially inspected auditor receives a clean PCAOB inspection report does not have an a priori incentive to dismiss their auditor (Abbott et al., 2013), while a client whose auditor receives a deficient PCAOB inspection report may have the incentive to dismiss their auditor.
TABLE 3
Overview of the client observations per industry (full sample)
Industry Two-digit SIC-codes GAAP-deficient client observations GAAS-deficient client observations Non-deficient client observations Construction 15-17 10 27 9 Consumer product and food 20-33 157 393 266 Energy 10-14, 46, 49 132 292 181 Information and communication 48, 73, 78, 79, 84 210 480 302 Manufacturing 34-39 179 479 338 Personal services and healthcare 72, 80, 83 25 57 30 Professional and educational services 75, 76, 82, 87, 89 33 64 49 Real estate 65, 70 22 70 137 Retail and wholesale 50-59 85 169 115 Transportation 40-42, 44, 45, 47 16 33 23 All other 1, 2, 7, 8, 9, 99 26 39 23 Totals 895 2103 1473
4. Descriptive statistics and results
Paragraph 4.1 presents descriptive statistics for the clients of GAAP-deficient and non-deficient, triennially inspected auditors, including an overview of the client switching behaviour. Paragraph 4.2 presents the results of this study.
4.1 Descriptive statistics
Paragraph 4.1.1 provides some general descriptive statistics and paragraph 4.1.2 provides an overview of the switching behaviour of clients from GAAP-deficient and non-deficient, triennially inspected auditors within one year after the public disclosure of the PCAOB inspection report.
4.1.1 General descriptive statistics
Table 4 provides descriptive statistics for the client observations regarding GAAP-deficient, triennially inspected auditors and non-deficient, triennially inspected auditors.
First of all, regarding the GAAP-deficient client observations, the mean value for SWITCH is 0.097. This indicates that 9.7 percent of the clients of GAAP-deficient, triennially inspected auditors switched to a non-GAAP-deficient successor auditor within one year after the public disclosure of the inspection report. Client switching behaviour will be more extensively discussed in paragraph 4.1.2. Secondly, 51.4 percent of the GAAP-deficient client observations relates to second-round PCAOB inspections. Thirdly, 10.1 percent of the GAAP-deficient clients were audited by a triennially inspected auditor that audits fewer than five public clients. This differs significantly from the percentage found by Abbott et al. (2013), as they found that 14.5 percent of the GAAP-deficient clients were audited by an auditor with less than five public clients (t = -2.265, p = 0.024 < 0.05). Furthermore, a majority of the GAAP-deficient clients (60.3 percent) received an explanatory paragraph regarding the going-concern assumption, which is consistent with Abbott et al. (2013). In addition, the financial performance of the GAAP-deficient clients was remarkably poor, as 67.1 percent of these clients reported a loss in their financial statements. Finally, interesting to note is that the mean value of total assets, for the 462 GAAP-deficient client observations in the reduced sample, is 41.4 million dollar. This differs significantly from the results of Abbott et al. (2013), as they found a mean value of total assets of 10.28 million dollar for their GAAP-deficient clients (t = 6.676, p = 0.000 < 0.01). This difference may be explained by the different samples used. Abbott et al. (2013) take into account triennially inspected auditors that received a GAAP-deficient inspection report between 2005 and 2007. However, most of the larger triennially inspected auditors (i.e. with more than 20 public clients) in this study’s sample received a GAAP-deficient inspection report after 2007, by which the auditors and client firms included in this study’s sample are quite different. These larger triennially inspected auditors audit larger client firms (mean value of total assets for the clients of GAAP-deficient auditors with more than 20 public clients is 49.2 million dollar, whereas mean value of total assets for the clients of GAAP-deficient auditors with less than 20 public clients is 24.1 million). The difference may also be explained by the different ways of data collection. Abbott et al. (2013) collect financial statement data largely by hand, by which it is likely that they were able to obtain data for smaller companies that is not available in AuditAnalytics or Compustat.
TABLE 4
Descriptive statistics for the GAAP-deficient and non-deficient client observations
N GAAP Mean N CLEAN Mean GAAP vs. CLEAN (t-value) P-value (P > |t|) SWITCH 895 0.097 1473 0.063 3.038 0.002 SECOND 895 0.514 1473 0.722 -10.479 0.000 AUDITOR_SIZE 895 0.101 1473 0.158 -3.930 0.000 GOING_CONCERN 895 0.603 1473 0.413 9.133 0.000 ASSETS 462 41,400,000 943 52,300,000 -1.715 0.087 CLIENT_SIZE (based on ASSETS) 462 15.867 943 16.342 -4.125 0.000 LOSS 462 0.671 943 0.661 0.373 0.710
As stated earlier, Table 4 also provides descriptive statistics for the client observations regarding non-deficient, triennially inspected auditors.
Firstly, the mean value for SWITCH is 0.063. This indicates that 6.3 percent of the clients of clean, triennially inspected auditors switched to a non-GAAP-deficient successor auditor, which is significantly different from the GAAP-deficient sample (t = 3.038, p = 0.002 < 0.01). As stated earlier, the switching behaviour of client firms will be more extensively discussed in paragraph 4.1.2. Secondly, about three-quarters (72.2 percent) of the non-deficient client observations relate to second-round PCAOB inspections and a quarter of the non-deficient client observations relate to first-round inspections. This differs from the GAAP-deficient sample in which about half of the client observations relate to first-round inspections and the other half to second-round inspections (t = -10.479, p = 0.000 < 0.01). As stated in Table 1, the percentage of audit firms that were labelled as GAAP-deficient is largely the same in the first-round (12.7%) and in the second-round (11.4%), by which it makes sense that the number of GAAP-deficient client observations is also largely the same in the first- and second-round. However, as stated in Table 1, the number of audit firms that received a clean inspection report increased significantly in the second-round, by which it makes sense that a major part of the non-deficient client observations relate to second-round inspections. Thirdly, 15.8 percent of the non-deficient clients were audited by a triennially inspected auditor that audits fewer than five public clients, which differs significantly from the GAAP-deficient sample (t = -3.930, p = 0.000 < 0.01). This implies that the clients of clean auditors were more likely to be audited by a triennially inspected auditor with fewer than five public clients, which is consistent with Abbott et al. (2013). Furthermore, 41.3
percent of the non-deficient clients received an explanatory paragraph in the audit report regarding the going-concern assumption, which is significantly different from the GAAP-deficient sample (t = 9.133, p = 0.000 < 0.01). The rate is also significantly different from the rate found by Abbott et al. (2013), as they found that 45.9 percent of the non-deficient clients received an explanatory paragraph regarding the going-concern assumption (t = -1.760, p = 0.079 < 0.10). In addition, the mean value of total assets, for the 943 non-deficient client observations, is 52.3 million dollar, which is significantly different from the GAAP-deficient sample (t = -1.715, p = 0.087 < 0.10). This indicates that the clients of GAAP-deficient auditors are not as large as the clients of non-deficient auditors. Furthermore, the mean value of total assets differs again significantly from the results of Abbott et al. (2013), where the mean value of total assets is 12.6 million dollar for their non-deficient clients (t = 7.117, p = 0.000 < 0.01). Finally, 66.1 percent of the non-deficient clients reported a loss in their financial statements, which is not significantly different from the GAAP-deficient clients (t = 0.373, p = 0.710 > 0.05). This indicates that the financial performance of the non-deficient clients corresponds to the financial performance of the GAAP-deficient clients.
4.1.2 Switching behaviour of GAAP-deficient and non-deficient clients
Table 5a provides an overview of the switching behaviour of clients from GAAP-deficient, triennially inspected auditors. Firstly, about 9.7 percent of the clients of GAAP-deficient, triennially inspected auditors (87 clients) switched to a non-GAAP-deficient, triennially inspected auditor within one year after the public disclosure of the PCAOB inspection report (6.9 percent in case of the reduced sample). This percentage is significantly lower than the dismissal rate shown by Abbott et al. (2013), as they find that 44.3 percent (168 out of 379 clients) of the GAAP-deficient clients switched to a non-GAAP-deficient, triennially inspected auditor within one year after public disclosure of the PCAOB inspection report (t = -15.369, p = 0.000 < 0.01). This implies that the clients of GAAP-deficient auditors included in the study of Abbott et al. (2013) were more likely to switch to a non-GAAP-deficient successor auditor than the clients of GAAP-deficient auditors included in this study’s sample. This may be explained by several factors (e.g. different sample periods used, larger client firms included in this study, different ways of data collection), which are discussed in great detail in paragraph 4.2.1 (based on the multiple regression results). Important to note is that 52 of the 87 clients switched to a triennially inspected auditor that is labelled as non-GAAP-deficient, because the auditor has not received a PCAOB inspection report yet. This implies
that these clients have not consciously looked for a triennially inspected successor auditor that received a non-GAAP-deficient inspection report. Abbott et al. (2013) do not take into account whether the successor auditor has received a PCAOB inspection report at the time of the client’s switch. In order to address this issue, this study conducts an additional analysis (paragraph 4.2.3.3) in which it distinguishes between successor auditors that are labelled as non-GAAP-deficient because the auditor has received a non-GAAP-deficient PCAOB inspection report and successor auditors that are labelled as non-GAAP-deficient because the auditor has not received an inspection report yet.
TABLE 5A
Switching behaviour of clients from GAAP-deficient, triennially inspected auditors
Full sample Reduced sample
First-round Second-round Total First-round Second-round Total Number of GAAP-deficient client observations 435 460 895 200 262 462 Switching: To non-GAAP-deficient, triennially inspected auditor 44 (10.1%) 43 (9.4%) 87 (9.7%) 17 (8.5%) 15 (5.7%) 32 (6.9%) To non-GAAP-deficient, triennially inspected auditor that has not received an inspection report yet 38 out of 44 14 out of 43 52 out of 87 15 out of 17 2 out of 15 17 out of 32 To GAAP-deficient, triennially inspected auditor (lateral) 2 (0.5%) 6 (1.3%) 8 (0.9%) 0 (0.0%) 2 (0.8%) 2 (0.4%) To Big-4 or national auditor 10 (2.3%) 7 (1.5%) 17 (1.9%) 6 (3.0%) 5 (1.9%) 11 (2.4%) No switch (within
one year of the public disclosure of the PCAOB report)
379 (87.1%) 404 (87.8%) 783 (87.5%) 177 (88.5%) 240 (91.6%) 417 (90.3%) Total 435 460 895 200 262 462
In addition, 8 lateral switches (from GAAP-deficient to GAAP-deficient) and 17 switches to Big-4 or national auditors are identified in this research. Abbott et al. (2013) report one lateral switch and 8 switches to Big-4 or national auditors, which is not significantly different (lateral switch: t = 1.227, p = 0.220 > 0.05; switch to Big-4 or national auditor: t = -0.249, p = 0.803 > 0.05). It is likely that the client firms related to these types of switches consider other factors, besides the inspection results of the PCAOB, when they decide to dismiss their auditor and choose a successor auditor or they may even not consider the PCAOB inspection reports when making switching decisions.
With respect to the difference between the first-round and second-round, it is interesting to observe that 10.1 percent of the clients of GAAP-deficient, triennially inspected auditors switched to a non-GAAP deficient, triennially inspected auditor after the first-round inspections and 9.4 percent of the clients made this switch after the second-round PCAOB inspections, which is not significantly different (t = 0.387, p = 0.699 > 0.05). This may indicate that the clients of GAAP-deficient auditors do not have a stronger incentive to switch to a non-GAAP-deficient, triennially inspected auditor following second-round inspections. Paragraph 4.2.1 will elaborate further on this issue.
Table 5b provides an overview of the switching behaviour of clients from clean, triennially inspected auditors. Firstly, 4.5 percent of the clients of clean, triennially inspected auditors switched to another clean, triennially inspected auditor and 1.8 percent of the clients switched to a GAAS-deficient, triennially inspected auditor, by which in total 6.3 percent of the clients of clean, triennially inspected auditors switched to a non-GAAP-deficient, triennially inspected successor auditor. Abbott et al. (2013) find that 17.9 percent of the clients of clean, triennially inspected auditors switched to a non-GAAP-deficient successor auditor. The rate found in this research (6.3 percent) is significantly different from the rate found by Abbott et al. (2013) (t = -7.716, p = 0.000 < 0.01). This implies that the clients of clean auditors included in the study of Abbott et al. (2013) were more likely to switch to a non-GAAP-deficient successor auditor than the clients of clean auditors included in this study’s sample. Furthermore, when looking at the difference between the switching rate of clients of GAAP-deficient auditors (to GAAP-GAAP-deficient auditors) and clients of clean auditors (to non-GAAP-deficient auditors), this research shows a difference of 3.4 percent (9.7%-6.3%), which is statistically significant (t = 3.038, p = 0.002). Abbott et al. (2013) also found a significant difference in their study (44.3%-17.9% = 26.4%). So, consistent with Abbott et al. (2013), clients of GAAP-deficient, triennially inspected auditors are more likely to switch to a
non-GAAP-deficient, triennially inspected successor auditor than clients of clean, triennially inspected auditors.
TABLE 5B
Switching behaviour of clients from clean, triennially inspected auditors
Full sample Reduced sample
First-round Second-round Total First-round Second-round Total Number of non-deficient client observations 410 1063 1473 253 690 943 Switching: To non-deficient, triennially inspected auditor (lateral) 17 (4.2%) 49 (4.6%) 66 (4.5%) 8 (3.2%) 24 (3.5%) 32 (3.4%) To non-deficient, triennially inspected auditor that has not received an
inspection report yet
14 out of 17 26 out of 49 40 out of 66 6 out of 8 13 out of 24 19 out of 32 To GAAS-deficient, triennially inspected auditor 7 (1.7%) 20 (1.9%) 27 (1.8%) 3 (1.2%) 12 (1.7%) 15 (1.6%) To GAAP-deficient, triennially inspected auditor 1 (0.2%) 3 (0.3%) 4 (0.3%) 1 (0.4%) 1 (0.2%) 2 (0.2%) To Big-4 or national auditor 3 (0.7%) 21 (2.0%) 24 (1.6%) 3 (1.2%) 16 (2.3%) 19 (2.0%) No switch (within
one year of the public disclosure of the PCAOB report)
382 (93.2%) 970 (91.2%) 1352 (91.8%) 238 (94.0%) 637 (92.3%) 875 (92.8%) Total 410 1063 1473 253 690 943
In addition, 4 switches (0.3 percent) from a clean, triennially inspected auditor to a GAAP-deficient, triennially inspected auditor and 24 switches (1.6 percent) to a Big-4 or national auditor are identified in this research. Abbott et al. (2013) identified two switches (0.4 percent) from a clean, triennially inspected auditor to a GAAP-deficient auditor and 6 switches (1.3 percent) to Big-4 or national auditors, which is not significantly different from
this study (lateral switch: t = -0.524, p = 0.601 > 0.05; switch to Big-4 or national auditor: t = 0.539, p = 0.590 > 0.05).
Finally, regarding the difference between the first- and second-round, it is interesting to note that 5.9 percent (4.2% + 1.7%) of the clients of clean, triennially inspected auditors switched to a non-GAAP-deficient successor auditor after the first-round inspections, whereas 6.5 percent (4.6% + 1.9%) of the clients made this switch after the second-round inspections, which is not significantly different (t = -0.450, p = 0.653 > 0.05). This may imply that the clients of clean, triennially inspected auditors do not have a stronger or weaker incentive to switch to a non-GAAP-deficient, triennially inspected successor auditor following second-round inspections.
4.2 Results
In paragraph 4.2.1, the multiple regression results are discussed. Paragraph 4.2.2 describes the GAAS-deficient sample and provides the accompanying results. Paragraph 4.2.3 describes the additional analyses that are performed.
4.2.1 Multiple regression results
The regression model used is presented in paragraph 3.1 and an overview of the results is given in Table 6. The regression model used to test the hypotheses is statistically significant (Chi2 = 32.33, p = 0.000 < 0.01; Chi2 = 21.09, p = 0.021 < 0.05).
First of all, regarding the first hypothesis, this study does not find a significant positive association between the receipt of a GAAP-deficient PCAOB inspection report and the likelihood of client’s switching to a non-GAAP-deficient, triennially inspected successor auditor (full sample: DEFRPT: β = 0.458, p = 0.192 > 0.05). Therefore, the first hypothesis (H1) cannot be supported. This finding suggests that clients of GAAP-deficient, triennially inspected auditors are not more likely to switch to a non-GAAP-deficient auditor than clients of non-deficient, triennially inspected auditors. This may indicate that the clients of GAAP-deficient, triennially inspected auditors do not use the PCAOB inspection reports as a publicly available signal of audit quality. This result is not consistent with Abbott et al. (2013). Their finding imply that the clients of GAAP-deficient, triennially inspected auditors are using the inspection reports as an audit quality signal. Firstly, the opposite results of this study may be
due to the difference in sample (period) used4. Abbott et al. (2013) only take into account triennially inspected auditors that received a first-round inspection report between January 2005 and December 20075. This study takes into account triennially inspected auditors that received a first-round and second-round inspection report between January 2005 and December 2011. Most of the triennially inspected auditors, of which a large part audit more than 20 public clients, received a (first-round or second-round) GAAP-deficient PCAOB inspection report after December 2007, by which the auditors and accompanying clients included in this study’s GAAP-deficient sample are significantly different. Clients of large triennially inspected auditors (with more than 20 public clients) may react differently to the GAAP-deficient PCAOB inspection reports. For example, these clients may be less likely to dismiss their auditor as it is less likely that the audit deficiencies identified during the inspection are related to the audit of their financial statements.6 Furthermore, clients included in this study’s sample are larger (on average) than the clients included in Abbott et al. (2013). As stated in paragraph 4.2.1, the mean value of total assets is 10.28 million for their GAAP-deficient clients, while the mean value of total assets (for the 462 GAAP-GAAP-deficient clients) is 41.4 million dollar in this study’s reduced sample. According to Ghosh & Lustgarten (2006), switching costs may be higher as, among others, client size increases. These potentially higher switching costs may suggest why clients of GAAP-deficient, triennially inspected auditors are not more likely to switch to a non-GAAP-deficient auditor in this research. Finally, the different results may be due to different ways of data collection. In this study, auditor change data is obtained from AuditAnalytics. However, Abbott et al. (2013) have used the SEC EDGAR website to obtain auditor change data, which may have provided them with a richer set of data.
On the other hand, the result found regarding the first hypothesis is consistent with Lennox & Pittman (2010), as they find that audit clients do not use the PCAOB inspection reports as a valuable signal of audit quality. One reason why clients may not react to the PCAOB inspection reports (by switching to another auditor) is the lack of information about differences in the overall quality of the audit firms. The PCAOB inspection report does not
4 Abbott et al. (2013) exclude, without any reason, some clean triennially inspected auditors from their sample
(e.g. Porter, Keadle & Moore LLP, Payne Falkner Smith & Jones PC).
5 This study provides an additional analysis in paragraph 5.3 in which only the sample period January 2005 to
December 2007 is considered.
6 This study provides an additional analysis in paragraph 5.3 in which the GAAP-deficient sample and control
sample are split into a sample comprising triennially inspected auditors that audit more than 20 public clients and a sample comprising triennially inspected auditors that audit 20 or less public clients (based on the public client number in the PCAOB inspection report).
