Auditor tenure and financial restatements:
The moderating effects of complexity, low
balling, and professionalism
Nathalie Michelle Hew
Student number: 10004229/ 6268021Academic year 2014-2015 Master thesis
Master Accountancy and Control 22th of June 2015
Supervisor: Dr. Bo Qin Amsterdam Business School Research field: Auditing Final version
Statement of Originality
This document is written by student Nathalie Michelle Hew, 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 others than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business at University of Amsterdam is responsible solely for the supervision of completion of the work, not for the contents.
Abstract
This study investigates the relation between auditor tenure and financial restatements by examining three moderating factors, complexity, low balling and professionalism. In this paper financial restatements are used as a proxy for audit quality. To interpret the moderating variables proxies are also used. Client firm size is used as the proxy for complexity, audit fees as the proxy for low balling and Big 4 classification as the proxy for professionalism. Through use of logistic regressions U.S. based publicly listed firms in the period from 2003 to 2013, with or without restatements, are
analyzed. While overall tenure results indicated to be insignificant, the categorization of tenure into three terms provided significant evidence. By categorizing auditor tenure into short, medium and long term, significant evidence is provided for long auditor tenure which suggests that there is a positive association between tenure and restatement likelihood. Further analyses of the influences of moderating variables provides more insight into the relation between auditor tenure and the likelihood of restatement. Specifically, the results suggest that complexity and professionalism significantly influence the relation between auditor tenure and the likelihood of restatements. The findings suggest that complexity attenuates the positive association between the auditor tenure and restatement likelihood. This is consistent with the concerns about the complexity negatively affecting the audit quality. The findings also suggest that professionalism strengthens the association between the two variables. However, this finding is inconsistent with what is initially hypothesized based on prior literature. Furthermore, no evidence was found to support the
association of low balling with the relation. Ultimately, this paper provides insight to the relation between auditor tenure and restatement likelihood by investigating the moderating variables, complexity, low balling and professionalism.
Acknowledgements
My gratitude goes out to Dr. Bo Qin, from the University of Amsterdam, for guiding me through this process. I am also grateful for my family and friends, without your support I would not be here.
Content
1. Introduction ………..……….……... p. 6 2. Literature review and hypotheses ………..…………. p. 9
2.1. Audit quality……….……..………... p. 9 2.2. Rotation and audit quality………..……….………... p. 9 2.3. Client complexity and audit quality.……….………….……...………. p. 11 2.4. Low balling and audit quality………...….…………....…...… p. 13 2.5. Professionalism and audit quality…….….……….….……….. p. 14 3. Research methodology ……….….……….………....… p. 15
3.1. Research design ………..………….………...…….. p. 15 3.2. Sample selection …………...……..……….………. p. 21 4. Results ………..…...……….………...… p. 22 4.1. Overall tenure results………..……….………..… p. 23 4.2. Results of the moderating variables …..……….………. p. 25 4.2.1. Results for client complexity…..…………..….……….. p. 26 4.2.2. Results for low balling ………..…….………. p. 27 4.2.3. Results for professionalism …………..……….……….. p. 27 4.3. Additional sensitivity tests ……….………….……….……….. p. 27 4.3.1. Short, medium and long term ………....………... p. 27 4.3.2. Classification of restatements ………….………...….…...…….. p. 28 4.3.3. Industry classification ………..………….…...….…... p. 29 4.3.4. Year classification ………...……....….….... p. 30 5. Conclusion ………...…………...….. p. 31 6. References ………...……….….… p. 33
1. Introduction
In 2006 the 8th European Union Company Law was adopted, this directive was aimed at the responsibilities, independence and ethics of statutory auditors. Audit partner rotation is one of the requirements of the 8th EU Company Law directives
implemented to maintain audit partner independence and objectivity. This directive requires audit partners to rotate every seven years. This policy has been mandated because regulators believed that a continuous audit partner – client relationship might cause a familiarity threat, which could harm the auditors’ independence and in turn the audit quality. Additionally, to ensure even higher audit quality, the EU has recently mandated auditor rotation for Public Interest Entities (PIEs) every ten years (PwC, 2014).This rule requires European-listed companies, banks and financial institutions to appoint a new auditor every 10 years, though this can be extended if companies put their audit contract up for bid at the decade mark or appoint another audit firm to do a joint-audit. Prior to these European initiatives to enhance the
reporting quality, the Sarbanes-Oxley Act (SOX) was introduced by the United States in 2002. SOX was enacted as a response on the scandals of Enron and Worldcom, to help enhance the quality of information. Section 203 of SOX required of Securities and Exchange Commission (SEC) registrants to comply with the audit partner rotation, which requires the audit partners to rotate every five years. Yet, even after the implementation of SOX, the question remained whether audit firm rotation should also be mandated to further enhance the quality of information. The U.S. General Accounting Office (2003) found that mandatory auditor rotation could strengthen the perception of independence, but that this would be costly.
Mandatory auditor rotation implies a limit on audit firm tenures. There are two views on this requirement, those of the opponents and those of the proponents. The opponents of rotation believe that the discontinuation of engagement and the loss of institutional knowledge cause greater risk to audit quality than the objectivity gained. The literature provides evidence to support those claims. Like Johnson, Khurana and Reynolds (2002) who find that short audit tenure had a negative influence on audit quality, which provides support for the opponents. Similarly, Meyer, Myer and Omer (2003) find that extended audit firm tenure is associated with higher quality of audits. Yet, the proponents of mandatory rotation believe that extended tenure will impair the auditors’ independence. Carey and Simnet (2006) provide evidence consistent with the claims that extended tenure is negatively associated with audit quality. Besides the
main discussion around auditor rotation, there are also various factors that could influence the relationship between auditor tenure and audit quality. Studies that investigate whether fee dependence has an influence on the relationship, yet again result in conflicting evidence (Gul, Jaggi, Krishan, 2007; Stanley & DeZoort, 2007). The same holds for studies investigating the effects of industry specialization, which either find no evidence of a moderating effect (Myers et al, 2003a) or evidence suggesting that for short tenure relation, industry specialization and audit fees are negatively associated with the likelihood of financial restatements (Stanley & DeZoort, 2007).
As discussed above there is quite some research done on the relationship between auditor tenure and audit quality. Yet, there remains a discussion around the relationship and therefore research on the relationship between auditor tenure and audit quality remains interesting and necessary. In order to contribute to the literature I will further examine the effect of moderating variables on the relationship between audit firm tenure and audit quality. I have chosen to investigate the moderating effects of complexity, low balling and professionalism. In order to provide adequate
assurance, auditors’ competence and independence has often been associated with the length of auditor engagement (AICPA, 1978). The moderators discussed in this paper all have a direct effect on independence and competence, which are the two main aspects of audit quality. Therefore it is important to further investigate these variables in order to get a better understanding of the relation between audit quality and auditor tenure is. Firstly, it is important to investigate the moderating effect of complexity in relation to the length of auditor tenure as prior literature already recognizes that it can significantly influence the audit quality (Carcello & Nagy, 2004; Litt, Sharma,
Simpson & Tanyi, 2014). By including low balling as a moderating factor, insight is gained through fee incentives to whether the audit quality is affected by the length of auditor engagement. Lastly, professionalism is included to examine whether this factor causes moderating effects on the audit quality when the auditor tenure increases or decreases.
To proxy audit quality I will use financial restatement. Restatements are unique and direct output-based measure of audit quality (Stanley & DeZoort, 2007). By using restatements as proxy or audit quality the request of Carcello and Nagy (2004) for a more objective and direct measure of audit quality is addressed. Additionally, financial restatements significantly reduce the confidence of investors in financial
reporting and market efficiency (SEC, 2002).
Ultimately, the purpose of this study is to investigate the moderating effects of complexity, low balling and professionalism on the relationship between auditor tenure and financial restatements. Thus, my research question states: To what extent complexity, low balling and professionalism influence the relationship between auditor tenure and financial restatements?
From an academic aspect my paper will give more insight into the association between audit firm tenure and audit quality through studying the moderating effects of various moderators, and thereby using a direct measure of audit quality. From a societal point of view my research may contribute to the discussion on mandatory audit firm rotation by investigating the moderating effects on the relationship between auditor tenure and audit quality.
By investigating U.S. based publicly listed companies over the period 2003-2013, this paper analyzes the main relation between auditor tenure and financial restatements and the effect of the moderators on this relation. The overall tenure analysis initially provided insignificant results. However, when auditor tenure is categorized in short, medium and long tenure, significant evidence was provided, suggesting that long auditor tenure is indeed positively associated with the likelihood of restatements. Furthermore, the results of the moderating variables suggest that complexity attenuates the positive relation between auditor tenure and restatement likelihood, while professionalism seems to strengthen this relation. While the finding of the complexity is consistent with the second hypothesis, the finding of
professionalism contradicts the theory of Big 4 audit firms delivering higher audit quality. Finally, no significant results were found regarding the influence of low balling on the main relation. Collectively, these results contribute to the auditor tenure literature by providing additional insight in the relation between auditor tenure and audit quality.
The study is conducted through quantitative analysis. Data used in this paper is retrieved from COMPUSTAT and Audit Analytics. The remainder of this paper is organized as follows. Chapter 2 discusses the relevant literature and provides the hypotheses. Chapter 3 describes the research design and method. In chapter 4 the results are presented, upon which the conclusion is drawn in chapter 5.
2. Literature review and hypothesis 2.1 Audit quality
In this paragraph prior studies on the relationship between audit firm/ partner tenure and audit quality are reviewed. The quality of audit is important because shareholders and other stakeholders use the audited financial statements to make decision about the company. The quality of an audit is assured via external audits being performed by auditors. External audits are an instrument to enable efficient technology for
organizing business enterprises (Watts & Zimmerman, 1986). Audit quality is defined by the Public Company Accounting Oversight Board (p.3, 2013) as “meeting
investors’ needs for independent and reliable audits and robust audit committee communications on: (1) financial statements including related disclosures, (2) assurance about internal control and (3) going-concern warnings. According to DeAngelo (1981b), audit quality is dependent on the probability of the auditors discovering breaches in the client’s statements and the auditor’s ability to withstand pressure from the client to selectively disclose. Thus, the quality of an audit is dependent on auditor competence and independence.
Ultimately, research points out that both industry expertise and independence contribute to higher audit quality. According to Brooks (2011), audit quality will increase with auditor tenure due to a learning effect and decrease with auditor tenure due to a bonding effect. From the auditor experience perspective, audit quality increases with auditor tenure over time as the auditor gains a better understanding of the clients system, business and industry environment, and internal controls (AICPA 1978) (Learning Effect). On the other hand, from the auditor independence viewpoint, audit quality decreases with auditor tenure over time as the auditor binds himself to the client due to either the economic bond or the social bond (Bonding Effect). The learning effect and bonding effect illustrate the trade-off between industry expertise and independence associated with the audit firm – client relationship. In the following paragraphs prior research regarding the trade-off is discussed.
2.2. Rotation and audit quality
To ensure the highest audit quality, regulatory bodies have also enforced various requirements over the years, like audit firm and audit partner rotation. Prior research has investigated the effects of these (potential) requirements on the relationship between tenure and audit quality. In 2002, Johnson et al. found that short audit-firm –
client relationships are associated with lower-quality financial reports compared to medium audit-firm – client relationships. Additionally, the authors could not provide sufficient evidence to support that long term engagements (nine plus years) are linked to declining financial reporting quality relative to medium length engagements. Myers, Myers and Omer (2003a) indicated that longer audit firm tenure constraints managerial discretion with accounting accruals, which in turn indicates high audit quality. Auditor tenure is perceived as increasing audit quality by investors and information intermediates (Ghosh and Moon, 2005). This perception is also confirmed by Jackson, Moldrich and Roebuck (2008), their research finds that mandating audit firm rotation provides minimal benefits for share- and stakeholders compared to audit partner rotation. According to Geiger and Raghunandan (2002), more reporting failures occur during the initial years of the audit firm – client engagement, compared to when this relationship was longer. Likewise, Carcello and Nagy (2004) conclude that fraudulent financial are more likely to occur during short auditor engagements, implying that mandatory auditor rotation will impair the quality of audit. When auditor tenure is extended, firms’ earnings become more conservative, however this result only holds for large audit firms (Li, 2010). Knechel et al. (2012) also state that prior research has indicated that higher audit quality is associated with lower likelihood of accounting restatements. Thus, restatements of financial
statements indicate that the audit opinion and the underlying audit process are most likely questionable because their audit opinion was not free from material
misstatements. Lastly, Stanley and DeZoort (2007) provide evidence to state that financial restatements are negatively associated with the length of the auditor – client relationship.
While all these papers indicate a positive relationship between auditor tenure and audit quality, some studies indicate otherwise. Some researches provide evidence to support mandatory auditor rotation, they find that a long-term auditor – client relationship could decrease auditor independence (Cameran et al., 2005). In contrast to the larger firms, Li (2010) suggests that smaller audit firms do face a bigger threat to the auditor’s independence during a long-term engagement, which will lead to a decrease of the audit quality. According to Brooks (2011), there are three aspects of auditor independence concerns: 1) overfamiliarity threat; 2) close personal
relationship and 3) reduced investor confidence. He also states that close auditor – client relationships causes the auditors to not only lose their independence, but also
their objectivity and professional skepticism. In contrast to the findings of the opponents, Davis, Soo and Trompeter (2002) actually find that longer auditor tenure actually increases the use of discretionary accruals to manage earnings. Research on a Korean firm sample indicated that mandatory auditor rotation improves the audit quality, due to the shorter-term relationship between auditors and the clients (Chung, 2004). When examining restatements as a proxy for audit quality, Myers et al. (2003b) conclude that there is no evidence to support the assumption that auditor rotation increases the likelihood of restatements. Lastly, by having shorter auditor tenures new perspectives are gained, which could lead to the identification of issues that may have been unidentified on previous audits, thus increasing the audit quality (Carey & Simnett, 2006). These authors conclude that long auditor tenure does indeed impair the auditors’ independence, and therefore with time auditors may possibly be less critical towards their clients which will reduce the audit quality.
Low quality audits tend to neglect errors and fraud, and could therefore lead to more financial restatements. This would mean that during a period of long audit-firm – client relationship more restatements are issued. Thereby suggesting that short auditor tenure leaves room for professional skepticism, objectiveness and ultimately independence, and therefore audit quality will be higher and will result in fewer financial restatements. However, other researches indicate that during long auditor tenure more familiarity and expertise is gained, and this will lead to higher audit quality (Geiger and Raghunandan, 2002; Johnson et al., 2002; Myers et al., 2003a; Ghosh & Moon, 2005). Often higher audit quality is determined on the basis of whether certain negative outcomes are present, such as financial restatements (Knechel, Krishnan, Peyzner, Shefchik & Velury, 2012). Thus, high quality audit should prevent mistakes and fraud in the financial statements. With higher levels of audit quality it should be more likely to detect material misstatements (Francis, Michas & Yu, 2013). If during these long audit-firm – client relationships fewer restatements are issued, it would indicate higher audit quality than when the audit-firm – client relationship is shorter and more restatements issued. This could imply that short audit-firm tenure leads to lower audit quality partly due to lack of
knowledge. However, if more restatements are issued during long auditor tenure, this would imply possible lack of independence. Based on the above discussed
H1: Audit firm tenure is positively associated with the likelihood of financial restatements
2.3 Complexity and audit quality
The first moderating factor of interest is complexity. The complexity of an
organization is defined as geographical and/or product line diversification (Bushman, Chen, Engel & Smith, 2004). Complexity relates to the client complexity and is also referred to as audit complexity. Complexity is positively associated with client firm size, implying that the greater the client firm size, the higher the complexity, and due to this higher level of complexity it is far less likely an audit firm would be an expert in each line of business in which the company operates (Carcello and Nagy, 2004). Larger firms with greater complexity face more severe and consistent repercussion as a result of audit partner rotation, these repercussions are worsened if the audit firm has limited resources and technology (Litt, Sharma, Simpson and Tanyi, 2014). After the introduction of SOX, the profession argued that audit partner rotation could threaten the audit quality, especially for larger clients. These larger clients hold a higher degree of complexity due to the volume of transaction, the complexity of processes and controls, the widespread nature of operations, and the greater variety of business risks. Because these larger clients are usually more complex, an effective audit will require a higher degree of audit effort and organizational expertise (Litt et al., 2014).
Daugherty, Dickins, Hatfield and Higgs (2012) suggest that getting well acquainted with complex issues like management and governance characteristics, getting senior management and audit committee members their trust, and applying client-specific knowledge to deliver a successful audit, requires two to three years of client
engagement. Ultimately, they find that the quality of audit is lower during the initial two to three years, especially for larger clients. Chin, Chen and Hsiesh (2009) found that higher degrees of complexity are positively associated with discretionary accruals, indicating lower audit quality when an organization is more complex.
Thus, client size can have significant impact on results. According to business professionals (e.g. PwC, 2011), the greater the client size, the more complex the audit due to the greater volume of transactions, the complexity of processes and controls, broad range of operations, and the diverse business risks. High complexity of a client
demands more effort from the auditor (Litt et al., 2014). For the auditor to get well acquainted with such complex issues, sufficient time is required. However, when such time is not available due to mandatory rotation, the quality of the audits will be
impaired. Thus, high levels of client complexity is expected to negatively influence the relationship between auditor tenure and audit quality, making it more likely restatements will occur. Therefore the second hypothesis states:
H2: Client firm complexity will attenuate the positive association between auditor tenure and restatement likelihood.
2.4 Low balling and audit quality
The second moderator in question is low balling. Low balling is when the audit firms offer an audit fee that is lower than the total current costs, this is offered upon the initial engagement to gain customers. Regulators and the profession claimed that low balling negatively influenced the independence of the auditor by establishing future economic interest in clients and thereby becoming financially bonded. However, Deangelo (1981a) provided evidence suggesting that low balling is a competitive reaction to the prospects of future economic quasi-rents, and does in fact not reduce the auditor independence. Although this implications indicates that low balling does not necessary cause low audit quality on the initial engagement, the prospect of future quasi rents does indeed create some incentive to lower audit quality in order to retain clients (DeAngelo, 1981a). Consistent with the literature, when a client is a significant part of the auditors’ portfolio, the independence of the auditor is likely to be impaired. The auditor will be more likely to act in favor of the client, thus decreasing the audit quality (Blay, 2005). When a client firm is more complex, the associated costs should raise due to the more complex procedures that need to be performed (Hay, Knechel & Wong, 2006). However, if the audit firm engages in low balling it could result in reduced audit effort (especially for the more complex firms), which subsequently lowers the quality of audits, in order to keep the associated costs low (AICPA, 1978). Although it was claimed that low balling of audit fees impaired the auditors’ independence, DeAngelo (1981a) provided evidence that low balling is actually a competitive response of prospects of future economics quasi-rents. However, this does not take away that the quality of audits could suffer as a consequence. When low balling occurs the initial start-up costs in the initial engagement should be recognized
as sunk costs and should therefore not be taken into consideration during decision making. Yet, Geiger and Raghunandan (2002) state that on the contrary to what the literature has documented, these costs are in fact taken into consideration and therefore could impair the auditors' independence. When taking auditor tenure into account, especially for shorter tenures, the threat of dismissal by the client in the initial years of the engagement could affect the auditor independence. More
specifically, during short auditor tenures the quality of audit could be impaired due to the incentive of the auditors to retain clients and thereby possibly reducing the audit effort in order to recover the costs of the initial engagements in the later periods. However, it can also be argued that audit firms with shorter tenures, have less incentive to protect their reputation and will be more likely to place emphasis on profits (Gul et al., 2007). Thus, Gul et al. (2007) mention that the competitive practice of low balling also reduces audit quality due auditor wanting to be more
accommodating towards the client, which could lead to lack of independence and auditor effort. Therefore ultimately, it can be argued that when auditors engage in low balling, they have incentives not to tightly monitor over the client’s financial reporting in the initials years of their tenure as they need long tenure to recover the initial price discount, thereby weakening the positive relationship between auditor tenure and restatements. Hence, short tenure leads to lower audit quality. Thus, I expect that low balling will have a negative effect on the relationship between auditor and financial restatements. More specifically the hypothesis reads:
H3: Low balling will attenuate the positive association between auditor tenure and restatement likelihood.
2.5. Professionalism and audit quality
The last moderating variable tested for is professionalism. As DeAngelo (1981b) professionalism will be linked to audit firm size. DeAngelo (1981b) suggests that audit firm size is a measure for audit quality because large audit firms are indifferent to one client, as these audit firms face greater reputation loss to their entire clientele in case of misjudgments. This will be referred to as professionalism and thus the size of the audit firm will determine the degree of professionalism. The higher degree of professionalism is mainly attributable to larger (perceived) independence. Large audit firms can be seen as more professional because they often have more incentives to
retain their reputation rather than clients, due to the large client base (Deis & Giroux, 1991). Their incentive to issue an accurate report is higher, because misstatements could lead to loss of rents (Lennox, 1999). The implication of Deis & Giroux’s (1991) study is that large audit firms will not impair the quality of audit in return for
retaining or obtaining clients, therefore these audit firms will remain more
independent. When an auditor is indifferent to the termination of its relationship with the client, he no longer has economic incentive to cover a discovered breach and is thus completely independent (DeAngelo, 1981a). Accordingly Defond, Erkend & Zhang (2014) find that Big N auditors are commonly associated with higher audit quality.
When examining the association between professionalism and auditor tenure, Lim and Tan (2010) indicate that longer periods of auditor tenure are associated with greater audit quality when auditors are less fee-dependent on clients, ergo more professional. As mentioned above, audit firms with shorter tenures will probably have less incentive to protect their reputation and will be more likely to place emphasis on profits and therefore diminish their auditor professionalism (Gul et al., 2007). Auditor professionalism is often found in larger audit firms because of their increased
incentive to remain independent in order to prevent reputation loss to the audit firm. If the audit firm is more professional, they are more independent and focused on
retaining their reputation and therefore it will be less likely that financial restatements will occur. I expect that when auditor tenure is long, an audit firm has more incentive to be professional in order to protect their reputation, and thus they are more likely to find and report breaches and errors in the financial statements. Thus, the positive association between auditor tenure and restatement likelihood will weaken as a result of auditor professionalism. The fourth hypothesis stated formally:
H4: Auditor professionalism will attenuate the positive association between auditor tenure and restatement likelihood.
3. Research methodology 3.1. Research design
In this paper an empirical evaluation is undertaken to investigate the influence of moderators on the relationship between auditor tenure and audit quality. This is done
through a database analysis. Data is collected from COMPUSTAT and Audit
Analytics from U.S. based-firms over the period 2003-2013. By using a sample over such a large period, it is assured that sufficient data will be available to conduct this research. Furthermore, for this paper I chose to work with U.S. based firms because there is more information available compared to European samples, which will hopefully lead to more robust results.
As stated it earlier financial restatements are used as a proxy for audit quality. Financial restatements will be the dependent variable in this paper. In a report of the Center of Audit Quality (Scholz, 2014) restatements are defined as “corrections of errors in public company GAAP financial statements filed with the SEC”. These financial restatements are a used indicator as proxy for negative audit quality (Knechel et al., 2012).
When interpreting our independent variable, auditor tenure, I will use the definition as used in Myers et al. (2003a), auditor tenure will equal the length of the auditor – client relationship during a fiscal year covered by audited financial
statements. Auditor – client relationships can have different lengths. These lengths will be defined as short, medium and long auditor tenure. To further clarify these terms of auditor tenure, Johnson et al.’s (2002) term definitions are used. Short auditor tenure is audit firm – client relationship with a duration of two to three years, medium length auditor tenure is associated with a four to eight year audit-firm – client relationship and long auditor tenure is defined as nine plus years.
Besides the dependent and independent variable, there are several factors which could influence the quality of an audit. To refine the results researchers often include some of these control variables to their research. These control variables could include various client characteristics such as client size, return on assets, leverage, current ration and asset turnover. As stated in the hypotheses development my main moderating variables in this study are complexity, low balling and
professionalism, because these variables are essential for gaining insight in the relationship between auditor tenure and audit quality. As previously explained it is expected that these variables will have moderating effects on the relationship between audit firm tenure and audit quality.
The first moderating factor complexity is measured by SIZE. The variable SIZE refers to client firm size. As stated earlier the size of a client firm is expected to negatively influence the relationship between auditor tenure and audit quality. Another
moderator in this paper is LOWBALLING. I added low balling to the regression model to control for fee dependence. Fee dependence will illustrate the auditors’ incentives; whether or not independence is compromised with respect to auditor tenure. In
accordance with DeAngelo (1981a) low balling occurs when the audit fee in the initial period is lower than the cost occurred for the initial audit. Through using an audit fee regression model, low balling is measured by comparing the predicted level of audit fees to the residual value. If the residual value results in a significant negative value this will indicate that the audit firm did engage in low balling. To estimate the professionalism of an audit firm, the variable BIG 4 is introduced. This variable will make a distinction between Big 4 audit firms and non- Big 4 audit firms.
Additionally, various control variables are included because these variables are previously known to have an influence on the relationship between auditor tenure and financial restatements. One of which is company GROWTH which indicates the firms’ growth opportunities. Furthermore, I included the control variables
LEVERAGE and LOSS because previous literature presumes that these variables could possibly cause greater incentive to misrepresent their financial reports if leverage is high and the company is less profitable (Myers et al., 2003b; Woodland and
Reynolds, 2003). Myer et al. (2003b) and Stanley and DeZoort (2007) also suggest that merger activity provides management to manage earning prior to the merger, thus lastly the control variable MERGER was included. Ultimately, previous literature concludes that firms that restate their financials usually have higher leverage and are less profitable and are more frequently involved in a merger than the companies that do not restate their financial statements.
Unlike papers using e.g. discretionary accruals (Jones-model), I could not find a standard logistic model for financial statement restatements. Based on the model used in Stanley and DeZoort (2007) and two working papers from Myers et al.
(2003b) and Woodland and Reynolds (2002), I estimated the following main model to test the first hypothesis:
RSTMT= β0 + β1Tenure (LT/ST) + β2 SIZE+ β3 LOWBALLING + β4 BIG_4+ β5 GROWTH + β6 LEVERAGE + β7 LOSS + β8 MERGER + ε
The dependent variable is RSTMT, which is equal to 1 if the financial statements are restated, and otherwise 0. To estimate the above mentioned model I use a logistic
regression. Tenure equals the length of the audit firm – client engagement in years; SIZE equals the natural log of the clients firms’ sales. LOWBALLING is a dummy variable and equals 1 if audit fees are considered abnormally low on the initial
engagement, otherwise 0. BIG 4 is a dummy variable and equals 1 if audit firm is one of the Big 4, otherwise 0. GROWTH is a dummy variable and is measured by
calculating the book value of the client companies’ equity. If market-to-book ratio is higher than 1 than growth equals 1, otherwise 0. LEVERAGE equals the total liabilities divided by the total assets. LOSS is a dummy variable, to identify whether or not the client firm has suffered a loss, if so than 1, otherwise 0. MERGER is the final dummy variable and equals 1 if the client was involved in a merger, otherwise 0.
A second model is estimated containing interaction terms. These interactions terms capture the interaction between with the independent variable Tenure and the moderators. This model is used to test the second through fourth hypothesis, which relate to the moderating effects of complexity (which will be measured by size), low balling and professionalism (Big 4/ non-Big 4) on the relationship between auditor tenure and audit quality. The model is estimated as follows:
RSTMT= β0 + β1Tenure (LT/ST) + β2 SIZE+ β3SIZE x Tenure + β4 LOWBALLING+ β5 LOWBALLING x Tenure+ β6 BIG_4+ β7 BIG_4 x Tenure + β8 GROWTH + β9
LEVERAGE + β10LOSS+ β11 MERGER + ε
The table below will demonstrate the various variables that will be used for my research.
Table 1 The variables
Variable Measure Proxy Description
Dependent Audit Quality Financial statement restatements
1 if the client restates its financial statements and otherwise 0
Independent Audit firm tenure Length of audit-firm – client relationship
The number of years the same auditor has been auditing the same client.
Moderating Control Low balling Complexity Professionalism Audit fees Size Big 4 firms Growth Leverage Merger Loss
The difference between the initial audit fee and the actual cost of an initial audit. Identified as dummy variable, 1 if auditor engaged in low balling; otherwise 0.
Client size is interpreted by taking the natural logarithm of client sales in year t.
Dummy variable, if Big 4 audit firm then 1, otherwise 0
Dummy variable, measured by the market-to-book value. Market value of equity divided by the book value of equity. If market-to-book ratio is higher than 1 than growth equals 1, otherwise 0.
Total liabilities/ total assets
Dummy variable, if company is involved in a merger activity then equal to 1, otherwise 0
Dummy variable, measured by net income. If net income is negative, loss equals 1, otherwise 0.
Tabel 2
descriptive statistics regression LAF
In order to determine whether audit firms engaged in low balling, an additional regression is carried out. A sample of audit firms is identified that are more likely to be associated with low balling (N=19844). Low balling will occur when the audit fees (publicly available) are below one standard deviation from the predicted level
(Lowballing=1), if this is not the case the firm supposedly did not engage in low balling (Lowballing=0). The predicted level of audit fees is estimated through a cross-sectional OLS audit fee model, which is similar to that used in Gul, Fung and Jaggi (2002). The following regression model is measure over a period from 2003 to 2013 as follows:
LAF = β0 + β1BIG 4 + β2 SIZE+ β3 GROWTH + β4 LOSS+ β5 LEVERAGE + fixed effects +ε
LAF is the natural logarithm of audit fees. BIG_4 indicated whether client is audited by a Big 4 company, if so BIG 4=1, otherwise 0. The variable SIZE relates to the client size, this is calculated by taking the natural logarithm of the total client sales. GROWTH is measured by the market-to-book ratio, if the market-to-book ratio is bigger than 0 it indicates whether the client experienced growth during the fiscal year, if so 1, otherwise 0. LOSS indicates whether client firm experienced losses, if so LOSS=1, otherwise 0. LEVERAGE is calculated by dividing the total liabilities of the client firm by the total assets. Fixed effects include the year and industry dummy variables. And lastly ε is the random error term.
Furthermore, the audit fee data is extracted from Audit Analytics and observations from tenures shorter than 3 years are excluded, as with low balling the focus is on short tenures.
LAF Coeff. T P>|t| BIG 4 0.6160373 37.16 0.000 LEVERAGE -.1883391 -7.02 0.000 SIZE .4455625 101.01 0.000 LOSS .3415445 22.44 0.000 GROWTH -.0585726 -2.29 0.022 _cons 6.245.551 105.38 0.000 (3)
3.2 Sample selection
As mentioned previously to test the hypothesis two datasets are used. Before combining the datasets retrieved from COMPUSTAT and Audit Analytics, the
datasets are filtered for relevant information. Consistent with the literature, companies in the financial services industry are excluded, as companies in such industries have special auditing regulation. The initial COMPUSTAT sample contained 102.974 observations. The initial Audit Analytics sample contained 155.947. When combining the dataset from COMPUSTAT with that from Audit Analytics, 85.884 observations are matched. Afterwards, the sample is corrected for specifically U.S. based
companies. Subsequently, when eliminating the observations with missing values the final sample contains 61,828 observations. The observations that did not include the necessary data for the control variables were excluded during the regression testing. Furthermore, the observations for auditor tenure and client sales are winsorized in order to eliminate the extreme value or outliners. These extreme values could have wrongly influenced the results, therefore these values are replaced with the acceptable extreme values.
The financial restatements are identified using Audit Analytics: Audit Fees with Restatements database as of the 7th of May 2015. The sample is cut off at 2013, which allows somewhat less of time for subsequent restatements to occur for the two latter years. According to Francis et al. (2013), there is an average delay of
approximately 700 days (or 2 years) between the issuance of the financial statements and the subsequent restatement. In table 3 the number of restatements per year are identified. Especially in the first sample years, the number of restatements is relatively high. We see that over time the number of restatements significantly decreases. In total there were 7,061 restatements issued in the period from 2003 till 2013. Furthermore, in table 3 the second column, the number of restatements retrieved from Audit Analytics: Audit Fees with Restatements is combined with the restatements retrieved from Audit Analytics: Non-reliance Restatements, thereby amounting to a total of 10,534 restatements in the period from 2003-2013.
Table 3
Sample characteristics for restatements per year
Audit fees with restatements
Non-reliance restatements1
Fiscal Year Restatement Restatement
0 1 0 1 2003 4,854 1,113 4,404 1,563 2004 5,095 862 4,630 1,327 2005 5,559 886 5,167 1,278 2006 5,419 785 5,029 1,175 2007 5,216 646 4,905 957 2008 4,930 529 4,639 820 2009 4,833 451 4,508 776 2010 4,777 402 4,442 737 2011 4,562 551 4,291 822 2012 4,685 517 4,480 722 2013 4,837 319 4,799 357 Total 54,767 7,061 51,294 10,534 4. Results
In this paragraph the results from the regressions will be discussed. The main regression type used is a logistic regression (table7). To test the robustness of the results probit and OLS-regressions are also run. Yet, there is no significant difference in results between the 3 types of regressions. The regressions are run on the two model discussed earlier, model I to test hypothesis one and model II to test hypothesis two through four.
Furthermore, the descriptive statistics of our sample variables can be found in table 4. In table 5 the results of the Spearman correlations test are displayed. The results of the Spearman correlation do indicate positive correlation between the independent variable and the moderating variables, TENURE and SIZE (correlation of 0.4846), TENURE and BIG 4 (correlation of 0.4722) and a negative correlation between TENURE and LOWBALLING (correlation of -0.2081). The dependent variable is also significantly correlated with LOWBALLING (correlation of -0.0108), BIG 4 (correlation of 0.0093), GROWTH (correlation of 0.0140) and LEVERAGE (correlation of -0.0194). The control variables and moderating variable are also all significantly correlated with each other (see table 5). However, inconsistent with what is predicted, no significant correlation can be observed between the independent and
1 With Non-reliance restatements column represents the combined number of restatements retrieved from Audit Analytics: Audit fees with restatements and Audit Analytics: Non-reliance restatements.
Table 5 Correlations Table 4 Descriptive statistics
dependent variables, restatements and tenure (correlation of -0.0010). Therefore to gain further insight into the relationship various regression are run, the results will be discussed in the following paragraphs.
***, **. * indicate significance at 1%, 5% and 10% levels (two-tailed), respectively.
4.1.1 Overall tenure results
The first hypothesis aims to test the main relationship between auditor tenure and financial restatements. In table 4, the descriptive statistics can be found that were
2 Restatements are the restatements solely retrieved from Audit Analytics: Audit Fees with Restatements. 3 Restatements combined are the restatements retrieved from Audit Analytics: Audit Fees with Restatements,
combined with the restatements retrieved from Audit Analytics: Non-reliance Restatements.
Variable Obs. Mean Std. Dev. Min Q1 Median Q3 Max
RESTATEMENT2 61,828 .1142039 .3180163 0 0 0 0 1 RESTATEMENT combined3 61,828 .1703759 .3759657 0 0 0 0 1 TENURE 61,795 11.34025 14.27914 0 2.46 6.98 13.34 80.52 SIZE 58,120 14.58856 2.29395 9.21 12.96 14.60 16.20 22.28 LOWBALLING 61,828 .3505693 .4771521 0 0 0 1 1 BIG 4 61,828 .6176975 .4859538 0 0 1 1 1 GROWTH 61,828 .9067575 .2907744 0 1 1 1 1 LEVERAGE 61,828 .5750174 .2878319 0 0.34 0.58 0.84 1 LOSS 61,828 .3750243 .4841331 0 0 0 1 1 MERGER 61,828 .1430258 .350102 0 0 0 0 1 1 2 3 4 5 6 7 8 9 1.RESTATEMENT 1.0000 2. TENURE -0.0010 1.0000 3. SIZE -0.0074* 0,4846*** 1.0000 4. LOWBALLING -0.0108*** -0.0698*** -0.0791*** 1.0000 5. BIG 4 0.0093** 0.4722*** 0.5981*** -0.1663*** 1.0000 6. GROWTH 0.0140*** 0.1440*** 0.1999*** -0.0617*** 0.1649*** 1.0000 7. LEVERAGE -0,0194*** -0.0403*** 0.1094*** -0.0492*** -0.0584*** -0,3960*** 1.0000 8. LOSS 0,0034 -0.2428*** -0.3690*** -0.0984*** -0.1796*** -0.2646*** 0.0913*** 1.0000 9. MERGER 0.0081 -0.0289*** 0.1217*** -0.1149*** 0.0740*** 0.0438*** -0,0464*** -0,0285** 1.0000
used to test H1. When the number restatements (7,061 restatements) retrieved from Audit Analytics is used in the regression, we find that despite prior literature no significant relationship is found between auditor tenure and financial restatements (P > |z|= 0.431). Meaning that no association can be confirmed between likelihood of financial restatements and the length of auditor tenure. Thus, with these results we reject our first hypothesis, as the results indicate no association between financial restatements and auditor tenure.
Presumably, the restatements retrieved from Audit Analytics Audit fees with Restatements are more reliable and therefore this sample will be used for further analysis. However, the results for the combined restatements groups are also analyzed for the robustness. When the number of restatements retrieved from Audit Analytics Audit Fees with Restatements is merged with those retrieved from Non-reliance Restatements, the number of restatement significantly increases (10,534 restatements) (table 6). When proceeding to run the regression with those numbers, the result then show a significant relationship between the independent and dependent variable, financial restatements and auditor tenure (P>|z| = 0.086) (table 7, panel A). The result indicate that there is a negative relationship between financial restatements and auditor tenure, even though is relatively small, -0.001543. Thereby, again rejecting the initial hypothesis as the hypothesis stated that auditor tenure is positively associated with the likelihood of financial statements.
The negative association between auditor tenure and financial restatements, as mentioned earlier, can be explained by the learning effect (Li, 2010). The learning effect suggests that as the auditor tenure increases, the audit quality will likewise increase due to auditors gaining more experience and expertise over time. Yet, the coefficient of tenure is relatively small, indicating that with every year of tenure increases, the likelihood decreases by 0.001543.
In addition, the control variable results in model I indicate that the size of the client company, size of the auditor (Big 4 or non-Big 4), growth and merger are significant factors that could influence the relationship between the independent and dependent variable. However, for the second moderating variable, low balling, no significant relation was found, in the initial regression run for hypothesis 1.
Table 7 Logistic Regression Results for both samples
Panel A
Table 6 Number of restatements
This table contains the logit regression of auditor tenure and financial restatements. Column 1 includes the main sample with restatements retrieved solely from Audit Analytics: Audit Fees with
Restatements, whereas column 2 contains the sample of combined restatements retrieved from Audit Fees with Restatements and Audit Analytics: Non-reliance Restatements.
4.2. Results of the moderating variables
As mentioned before in the paper investigates the three moderating variables, complexity, low balling and professionalism, and whether these variables have an influence on the relationship between financial restatements and auditor tenure. To measure each of the moderating factor, the proxy size, audit fees and audit firm size (Big 4 vs. non-Big 4) are used. In table 7, panel B the results of logistic regression of the moderating variables are shown.
Restatements Audit fees with Restatements 7,061 Non-reliance Restatements 3,473
Total 10,534
Audit fees with Restatements Non-reliance Restatements
Independent Dependent variable
Restatements (1) Restatements (2) Coeff. Z-value P> | z | Coeff. Z-value P> | z | TENURE 0.000801 0.79 0.431 -.001543 -1.711 0.086 SIZE -.0315103 -3.89 0.000 -.0243245 -3.54 0.000 LOW BALLING -.0715971 -2.69 0.008 -.0418495 -1.84 0.066 BIG 4 .1191175 3.51 0.000 .0701939 2.44 0.015 GROWTH .1595381 2.66 0.007 .0118801 0.25 0.803 LEVERAGE -.1061507 -2.01 0.045 .0033446 0.07 0.941 LOSS .0212344 0.70 0.482 .1995504 7.87 0.000 MERGER .0663407 1.83 0.067 .2295252 7.64 0.000 _cons -1.743508 -14.95 0.000 -1.36297 -13.99 0.000 Number of observations 58,087 58,087 R2 0.0013 0.0034 LR chi2 53.72 182.22
Table 7 Logistic Regression Results moderating factors
Panel B
This table contains the logit regression of auditor tenure and financial restatements with the moderating factors. Column 3 includes the main sample with restatements retrieved solely from Audit Analytics: Audit Fees with Restatements, whereas column 4 contains the sample of combined restatements retrieved from Audit Fees with Restatements and Audit Analytics: Non-reliance Restatements.
4.2.1. Results for complexity
When testing hypothesis two, the influence of client complexity on association between the financial restatements and auditor tenure, the result indicate a significant relationship (P > |z| < 0.004). The coefficient of -0.0015361 suggest that client complexity does indeed attenuate the relationship between auditor tenure and
likelihood of financial restatements. Thereby providing sufficient evidence to confirm hypothesis two. As complexity of the client is determined by the size of the client, the results suggest that the larger the client firm, the higher the level of complexity and
Audit fees with
Restatements Non-reliance Restatements
Independent Dependent variable
Restatement (3) Restatement (4) Coeff. Z-value P> | z | Coeff. Z-value P> | z | TENURE .0094425 1.11 0.268 -.0120533 -1.60 0.109 SIZE -.010602 -1.07 0.284 -.0121916 -1.46 0.143 LOW BALLING -.088923 -3.03 0.002 -.0543059 -2.18 0.029 BIG 4 -.010134 -0.22 0.824 -.0885245 -2.27 0.023 SIZE X TENURE -.0015361 -2.91 0.004 -.000921 -2.02 0.044 LOW BALLING X TENURE .0029091 1.33 0.184 .0025797 1.32 0.186 BIG 4 X TENURE .017325 3.57 0.000 .0268184 5.98 0.000 GROWTH .1540214 2.56 0.010 .0181292 0.38 0.704 LEVERAGE -1071959 -2.02 0.043 .0006575 0.01 0.988 LOSS .027156 0.89 0.371 .1980658 7.77 0.000 MERGER .0650931 1.79 0.073 .2260699 7.52 0.000 _cons -1.95468 -14.17 0.000 -1.421924 -12.37 0.000 Number of observations 58,087 58,087 R2 0.0018 0.0042 LR chi2 74.63 222.54
thus the more likely financial restatements will occur as tenure lengthens.
4.2.2. Results for low balling
The second moderating variable, low balling, is tested whether it will indeed attenuate the association between the independent and dependent variable. Prior research indicated no relationship between audit quality and low balling, yet due to the prospects of future quasi rents the quality could still be somewhat impaired
(Deangelo, 1981a). However, the results of low balling as a moderating factor on the relationship between auditor tenure and the likelihood of financial restatements provided no such evidence (P >| z | = 0.184). Thus, no significant evidence found is to support the third hypothesis, and therefore hypothesis 3 is rejected. The insignificant result of low balling is on the other hand somewhat consistent with what prior literature already indicated, low balling does not impair audit quality over time.
4.2.3. Results for professionalism
Professionalism is identified as auditor being either Big 4 or non-Big 4 audit firms. Upon testing the fourth and last hypothesis, results indicate that professionalism is indeed a significant variable (P >| z | = 0.000) and positively influence the association between auditor tenure and the likelihood of financial restatements (coefficient of .017325). This evidence is inconsistent with what was initially hypothesized, as the hypothesis stated that professionalism will attenuate the association between auditor tenure and likelihood of financial restatements. Thus the positive influence, indicates that professionalism does not attenuate but rather strengthens the association between the two variables.
4.3. Additional sensitivity analysis
To test for robustness additional logistic regressions are run. First, a distinction is made between three different types of auditor tenure. To see whether results differ dependent on short, medium or long term. Secondly, restatements are classified by characteristics. And finally additional sensitivity tests are done for classifications per industry and per year.
4.3.1. Short, medium, long tenure
and long term. For this test the same term definition is used as Johnson et al. (2002). Short auditor tenure is defined as a relationship of two to three years, a medium auditor tenure is defined as a relationship of four to eight years, while long auditor tenure is defined as a relationship of more than nine years.
A regression is run for each of the three terms. The result indicate that for short auditor tenure there is no significant relationship between tenure and the likelihood of financial restatements (P >| z |= 0.451). When proceeding to medium auditor tenure, again results point out no significant relationship between the two main variables (P >| z |= 0.144). Lastly, looking at the results for auditor tenure of nine plus years, results then indicate a significant association between the independent and dependent variable (P>| z | =0.075 and coefficient of .0022291). This result is in line with the first hypothesis which states a positive association between auditor tenure and the likelihood of financial restatements.
The results gained from this sensitivity analysis can somewhat explain the overall tenure results, as those results indicated no significant relation between auditor tenure and restatement likelihood. By categorizing auditor tenure into three
categories, it becomes clear how the overall tenure result is reached. The results indicate that although for the short and medium auditor tenure terms no significant results were found, the results for the long term auditor tenure suggest a positive association between auditor tenure and the likelihood of restatements. The longer the tenure, the higher the likelihood of financial restatements.
4.3.2. Classification of type of restatements
To get an even greater insight into the association between financial restatement and auditor tenure, an additional distinction can be made between various types of restatements. In this sample we made five distinctions, namely restatements due to fraud, restatement due to clerical errors, restatements that influence the financial statements negatively, restatements that influence the financial statements positively and restatements under investigation by SEC. Noticeably, the majority of restatements negatively influence the financial statements (table 8).
The regression is run for each of the five types of restatements. Untabulated results then indicate that for restatement caused by fraud there is significant negative association between tenure and likelihood of restatements (P>| z |= 0.018) with a coefficient of -.0427334. The other four types of restatements do not suggest any
Table 8
Restatement classifications
significant association with the independent variable auditor tenure.
The regression is also run to test the moderating effects by the characteristics of restatements. For restatements characterized by fraudulent activities and those characterized by clerical errors significant evidence was found consistent with the initial findings. The results for restatements due to fraud, indicate a significant evidence that professionalism strengthens the relation between auditor tenure and restatement likelihood (P>| z |= 0.060 and coefficient of .2102591). This result is consistent with the moderating effect of professionalism found in the main sample. Furthermore, the results indicate that for restatements characterized by clerical errors, the moderating variable size has significant influence to attenuate the relation between auditor tenure and the likelihood of restatements (P>| z |= 0.010 and coefficient of -.0109959). This result is also consistent with the moderating effect of client
complexity found in the main sample.
4.3.3. Industry classification
As part of the sensitivity tests, the standard industrial classification code (SIC) is used to identify the various industries. In order to classify the various industries, dummy variables are added. In this case a distinction is made between 12 major industries (table 9). In the sample it can be observed that some industries are rather large while others are relatively small. The largest observations are gained from the Durable material industry, with 23.42% of the total observations, while the smallest industry amounts to 2.29% of the total observations.
For each of the industries a logistic regression is run. The results suggest that mainly in the Utilities, Retail, Textiles and printing, and lastly Extractive industry, auditor tenure is significantly associated with the likelihood of financial restatements. Type of restatement Number of restatements Fraud 59 Clerical error 239 Adverse 3,386 Improve 689 SEC Investigation 257
Table 9
Industry classification categories
However, not all industry have the same association between the two variables. For instance in the Utilities, Retail and Extractive industry, auditor tenure is positively associated with restatement likelihood, while in the Textiles and printing industry, auditor tenure is negatively associated with restatement likelihood. Ultimately, indicating that in the positively associated industries the longer the tenure is the more likely restatements will occur, while the negatively associated industries suggest that shorter tenure will lead to a higher likelihood of restatements.
4.3.4. Year classification
The last sensitivity analysis is to run regressions for each year. The period used in the sample is from 2003 till 2013, thus containing 11 years. The number differs per year, this was already observed in table 3. Due to the rather significant fluctuations in restatements per year, it is interesting to further analysis the association between the independent and dependent variable for each year.
Although, in the initial years of the sample the number of restatements are rather large, the results indicate that solely in the last two year, 2012 and 2013, a positive association is found between auditor tenure and restatement likelihood (P>| z |= 0.025& P>| z |= 0.050), the results of the prior years are found to be not
4 Industry composition is defined similar to Litt et al. (2014). The SIC codes are defined as follows: Chemicals (2800-2824, 2840-2899); Computers (3570-3579, 3670-3679, 7370-7379); Durable Materials (3000-399, excluding 3570-3579 and 3670-3679); Extractive (1300-1399, excluding 2900-2999); Food (2000-2111); Mining and Construction (1000-1999, excluding 1300-1399);
Pharmaceuticals (2830-2836); Retail (5000-5999); Services (7000-8999, excluding 7370-7379)); Textiles and Printing (2200-2799); Transportation (4000-4899) and Utilities (4900-4999).
Industry4 Observations 1 Chemicals 1,265 2 Computers 8,150 3 Durable Materials 10,465 4 Extractive 1,996 5 Food 1,105
6 Mining and Construction 1,024
7 Pharmaceuticals 3,160
8 Retail 5,051
9 Services 4,984
10 Textiles and Printing 1,922
11 Transportation 2,946
significant. Thus, the results suggest that while the number of restatements decreases over time, in the latter two years (the years with the least number of restatements) it is more likely restatements will occur while tenure lengthens. Explanations for this could possibly be found in the length of auditor tenure in the latter two years, maybe in those years auditor tenures were not as long as the prior years, however further analysis would be needed to fully understand why this result is observed.
5. Conclusion
This paper examines the moderating effects of complexity, low balling and
professionalism on the relationship between auditor tenure and financial restatements. Prior research extensively researched the main relationship between auditor tenure and audit quality, yet the discussion remained whether or not, and more importantly how auditor tenure influences the audit quality. To gain further insight into the relationship this paper examines three moderating variables. Thus the research question states: “To what extent do complexity, low balling and professionalism influence the relationship between auditor tenure and financial restatements?” In order to answer the research question this study investigates U.S. based publicly listed firms in the period 2003-2013, through a quantitative analysis. Although prior literature had often used discretionary accruals as proxy for audit quality, this paper uses financial restatement as a measure of audit quality. Financial restatements are a unique and direct measure of audit quality, which can provide different insights into the relationship compared to the frequently used proxies like discretionary accruals or propensity of going-concern issuance.
The overall tenure findings suggest that there is no significant evidence to support the association between auditor tenure and the likelihood of financial restatements. However, auditor tenure can be divided in three categories, short, medium and long term (Johnson et al., 2002). Short auditor tenure is defined as a relationship of two to three years, a medium auditor tenure is defined as a relationship of four to eight years, while long auditor tenure is defined as a relationship of more than nine years. Thus, when auditor tenure is categorized in short, medium and long term, the result in fact indicate that long auditor tenure is positively significant associated with restatement likelihood.
between auditor tenure and financial restatements, this paper investigates the
moderating variables complexity, low balling and professionalism. Through analysis I can conclude that complexity and professionalism both have a significant impact on the relationship between auditor tenure and the likelihood of restatements. The findings suggest that complexity attenuates the positive association between auditor tenure and the likelihood of restatement, while professionalism seems to strengthen the positive association. The attenuating effect of client complexity is consistent with the second hypothesis, thus it can be concluded that for clients with higher levels of complexity (greater company size) the likelihood of restatements will increase as tenure lengthens. However, the strengthening effect, rather than the expected attenuating effect of professionalism, is inconsistent with what initially was
hypothesized. As stated earlier the assumption that Big 4 audit firms deliver higher audit quality and are more professional is based on the perceived auditor
independence, however this does not represent the actual auditor independence. It could be that although Big 4 audit firms are perceived as independent, in reality this perceived independence could be a mere illusion, as results suggest that in reality the audit quality diminishes over time. Some research pointed out that Big 4 audit firms offer higher audit quality and accordingly are sued less frequently and are even sanctioned less frequently. However, this does not suggest that they actually offer higher audit quality, it could be that these Big 4 audit firm just have better resources to fight lawsuits and regulators (Francis, 2004). In addition, Francis et al. (2013) find evidence to state that it is likely that the prior findings of quality differences between Big 4 and non-Big 4 are driven by a subset of large Big 4 firms, as smaller Big 4 offices are found to deliver lower audit quality at times. It could be that in the sample used there are mostly smaller Big 4 offices, rather than a large number of larger Big 4 offices, and therefore explaining the strengthening rather than attenuating influence on the association between auditor tenure and restatement likelihood. Lastly, regarding the final moderating variable, low balling, no significant evidence was found to support the assumptions made. Even the sensitivity test to further investigate the underlying results were found insignificant.
Furthermore, some sensitivity analyses were run to test the robustness of the results. Thereby restatements were categorized into five categories. Significant evidence was found to conclude that auditor tenure and restatement likelihood are negatively associated for restatements caused by fraudulent activity. For the same
type of restatements results also indicated that complexity strengthens the association between auditor tenure and restatement likelihood. Professionalism also seems to attenuate the positive association between auditor tenure and restatement likelihood for restatements characterized by clerical errors.
Additionally, sensitivity analyses were run for industry and year
classifications. Results indicated that in the Utilities, Retail and Extractive industries tenure is positively associated with restatement likelihood, while in the Textile and printing industry tenure is negatively associated with restatement likelihood. To understand this difference I would advise to further investigate the factors that could influence each individual industry. Lastly, the sensitivity test of year classifications implied that in the latter two years of the sample tenure is positively associated, while prior years no significant association could be observed.
Ultimately, this paper finds that complexity and professionalism are factors which have a significant influence on the relationship between auditor tenure and the likelihood of financial restatements. Yet, limitations in this paper might exist due to the sample cut off in 2013 the financial statements of 2013 have less time, than the estimated two years needed, for the restatement to be discovered. Furthermore, the number of restatements could be biased as the audit failures occurred in the financial crisis, causing lower audit quality due to external factors rather than tenure. Lastly, I would suggest further investigation into the audit quality of Big 4 audit firms, as results were inconsistent with what prior literature stated.
6. References
American Institute of Certified Public Accountants (AICPA, 1978). The Commission on Auditors' Responsibilities: Report, Conclusions and Recommendations. The Cohen Commission. New York: AICPA.
Basu, S. (1997). The conservatism principle and the asymmetric timeliness of earnings, Journal of Accounting and Economics, 24, (1), pp. 3-37.
Benh, B., J. Choi & T. Kang. (2008). Audit quality and properties of analyst forecasts. The Accounting Review, 83, (3), pp. 327-349.
Blay, A. D. 2005. Independence threats, litigation risk, and the auditor’s decision process. Contemporary Accounting Research 22 (4): 759-789.
