Master Thesis Accounting. To what extent do firms switch auditors to survive an economic crisis?

Transcription

1 Master Thesis Accounting To what extent do firms switch auditors to survive an economic crisis? Dennis P.H. van Ginneken Date of completion: 23 June 2012

2 Master Thesis Accounting To what extent do firms switch auditors to survive an economic crisis? Tilburg University Faculty of Economics and Business Studies Department of Accountancy Dennis P.H. Van Ginneken ANR: Date of completion: 23 June 2012 Date of graduation: 5 July 2012 Supervisor: S.N.M. Vandenbogaerde

3 iii Preface This thesis completes the master in Accounting, which I followed at Tilburg University. Together with my bachelor, I have studied in Tilburg for the last four years. During these years I received support and encouragement from friends and family, for which I want to thank them. I particularly want to thank my parents who were always there for me and without whom I would have not been able to successfully complete this study. Furthermore, I would especially like to thank my sister Esther for her encouragement, support and helpfulness during my study. Finally, I would like to thank my supervisor, Sofie Vandenbogaerde, for her suggestions and feedback which substantially improved my thesis. Dennis van Ginneken Tilburg, 23 June 2012

4 iv Abstract The current study compares the auditor switching behavior of U.S. listed firms during an economic crisis to the switching behavior in a non-crisis period. The study can indicate whether firms use auditor switches to save costs or mask their real financial performance during a crisis. Results show that non-big 4 clients are more likely to switch auditors during a crisis. On the contrary, Big 4 clients are less likely to switch auditors in the crisis period and also less likely to subsequently choose a non-big 4 successor in this period. Furthermore, firms have a significantly higher likelihood of switching auditors when financial performance is lower during a crisis period than when this performance deterioration occurs in a non-crisis period. By focusing on the abnormal part of the audit fee charged, this study also attempted to document that firms with relatively high audit fees are more likely to switch during a crisis compared to when this fee level is charged in a non-crisis period. Although the coefficients often have the predicted sign, results are only marginally significant for this test.

5 Table of contents Preface... iii Abstract... iv Table of contents... v 1 Introduction Literature review Auditor switches Big 4 and non-big 4 auditors Likelihood of switching auditors to save costs Methodology Sample selection The auditor switches model The audit fee model Results Descriptive statistics Auditor switches Audit fees Likelihood of switching auditors to save costs Additional tests Robustness checks Discussion References Appendices Index of tables v

6 Introduction 1 1 Introduction In 2007 the United States of America witnessed a sub-prime crisis. This crisis was caused by the falling house prices in America, leading to the fact that less credit-worthy borrowers could not pay for their mortgages, which in turn increased the default rates. Due to the complexity and illiquidity of the investments in these mortgages, the impact of this increased amount of defaults was greatly multiplied (Reinhart & Rogoff, 2008). Usually banks would act as intermediaries, but in this case they were primarily the investors themselves. As a result, they were the ones bearing the risk on these mortgages. This is seen as a risky strategy, since banks have a high amount of leverage. As a consequence, the reduced house prices led to the collapse of the market for assetbacked commercial paper and ultimately to the global financial crisis in 2008 (Acharya & Richardson, 2009). Because the financial market became globally involved in 2008, the whole economy was pushed into a recession. This also affects firms; they have to cope with lower demand which consequently reduces revenues. This causes firms to become more constrained in their financial budget and may even lead to financial losses. Campello, Graham and Harvey (2010) showed that constrained firms planned to cut costs on technology, employment and capital to a greater extent. The audit service is another category on which U.S. listed firms can attempt to cut costs, since these firms financial statements have to be audited. These audit fees are usually larger for clients of Big 4 auditors than for clients of non-big 4 auditors (e.g., Craswell, Francis, & Taylor, 1995; Palmrose, 1986a; Simon & Francis, 1988), giving Big 4 clients an opportunity to save costs by switching downwards. Furthermore, lateral switches are also associated with lower fees due to the low-balling practice (e.g., Baber et al., 1987; Ettredge & Greenberg, 1990; Francis & Simon, 1987). In addition, some authors have argued that non-big 4 audit firms provide lower quality audits (e.g., Becker, DeFond, Jiambalvo, & Subramanyam, 1998; Caramanis & Lennox, 2008; DeAngelo, 1981b) and that shorter client-auditor relationships are associated with lower audit quality (e.g., DeFond & Subramanyam, 1998; Johnson, Khurana, & Reynolds, 2002; Myers, Myers, & Omer, 2003), enhancing the accounting flexibility and opportunity to manage earnings. During an economic crisis firms might be more attracted to these opportunities to save costs and mask their financial performance; firms are therefore predicted to be more likely to switch auditors during a crisis and Big 4 clients are furthermore predicted to be more likely to switch downwards during a crisis. Delving deeper into the cost savings motive for switching auditors during an economic crisis, it is expected that firms switching auditors attain a larger fee drop

7 Introduction 2 during the crisis than in a non-crisis period. This due to the anticipated incremental effort of the firm to find the cheapest auditor and the potential additional price competition between auditors. Focusing on the relative level of the audit fee, clients which have relatively high audit fees are predicted to be more likely to switch during an economic crisis due to the additional cost savings potential. Furthermore, clients which are trying to cope with low financial performance are expected to have a higher likelihood of switching auditors when this occurs during a crisis as opposed to a non-crisis period. Reasons for this expectation are that firms might be more aware of the opportunity to save costs through switching auditors during a crisis and that it is more interesting to switch during a crisis when auditors indeed engage in fiercer price competition. In summary, the current study examines the extent to which firms switch auditors during an economic crisis to deal with the downward tendency in performance. In this research, data from U.S. listed firms in the non-crisis period (2005 & 2006) and the crisis period (2009 & 2010) is used to test the hypotheses. The auditor switches used in this research are those initiated by the client, which means that the focus is on the demand side of the audit market. To test whether firms are more likely to switch auditors during an economic crisis and whether Big 4 clients are more likely to switch downwards a logistic regression model, similar to the model estimated by Ettredge, Li and Scholz (2007), is employed. Contrary to expectations, the results imply that Big 4 clients are less likely to switch auditors during the crisis. However, when focusing on clients of smaller audit firms, findings do support the predictions. A potential explanation for these results relates to the reputation and credibility that clients of large accounting firms might want to signal to the public, where switching auditors could signal an impairment of independence and quality. Also in contrast with expectations, but consistent with the previous argumentation, is that the results indicate that Big 4 clients are less likely to switch to smaller accounting firms during an economic crisis. Next, an audit fee model comparable to the one estimated by Ettredge, Heintz, Li and Scholz (2011) is employed to investigate whether fee drops are larger for switchers during the crisis compared to switchers before the crisis. The findings confirm that switching during a crisis is associated with an additional fee cut when compared to switching in a non-crisis period. This implies that firms might indeed be putting additional effort into finding the cheapest auditor that can supply the services needed. It further supports the notion that auditors are engaging in additional price competition during an economic crisis to attract these clients. In order to test whether firms with relatively high audit fees or low financial performance are more likely to switch when this occurs during a crisis as opposed to a non-crisis period, the logistic regression model mentioned before is estimated again.

8 Introduction 3 Using the residual from the audit fee model as a proxy for the relative level of audit fees, results only weakly imply a more positive association between the abnormal audit fee and the likelihood of switching auditors during an economic crisis. Thus, clients do not seem to be influenced extensively by the relative level of audit fees as a determinant for dismissing their current auditor. Finally, firms with lower performance do seem to be more likely to switch auditors during an economic crisis. This suggests that clients are indeed more aware of the opportunity to save costs by switching auditors during a crisis and also more interested in this practice, potentially due to the additional price competition between auditors. This research is scientifically relevant because it is one of the first to provide evidence regarding the switching behavior of firms during an economic crisis. It also adds to the existing literature on auditor switches and the determinants of these switches. From a societal point of view this research is relevant because it implies that firms see audit fees as a cost on which they can save, especially during an economic crisis. Of course, auditors might also be interested in the motives of clients for switching auditors during a crisis. Knowing these motives can help them in their attempts to retain their current clients and attract new clients by adapting to these interests of firms. The results confirm this interest of auditors since audit fees are indeed used as a tool to attract and retain clients during a crisis. Furthermore, the fact that Big 4 clients are less likely and non-big 4 clients are more likely to switch during an economic crisis contributes to societal knowledge on the auditing practice. This because it suggests that audit quality may on average be lower with non-big 4 clients and higher with Big 4 clients during a crisis compared to a non-crisis period. For investors and regulators, this means that financial statements of Big 4 clients might actually become more reliable and credible overall during an economic crisis. It further suggests that Big 4 clients, in contrast to non-big 4 clients, attach substantial value to the (perceived) quality of their financial statements. Finally, contributing to the debate on mandatory auditor rotation, the results imply that firms perceive retaining an auditor as a signal for credibility. If this is indeed the case, mandatory auditor rotation might actually reduce the overall quality of financial statements provided to the public. Therefore, the proposed rule of mandatory rotation could have the opposite effect of what is desired by regulators. The remainder of this paper is structured as follows: section 2 discusses the prior literature on auditor type, auditor switches and motives for switching auditors. In section 3, I outline the method and data used in this research. The analyses of the data and hypotheses are presented in section 4. Finally, Section 5 contains the discussion.

9 Literature review 4 2 Literature review An extensive part of the academic literature on auditor switches investigates the type of switch and the effect of auditor switches on audit quality and audit fees. Overall, the results of these studies are mixed. However, there is limited literature on the impact of an economic crisis on switching behavior. The following sections address each of these previous literature topics and develop the hypotheses. I start with a review of the general literature on auditor switching, after which I discuss papers investigating possible differences between Big 4 and non-big 4 auditors. The final part of the literature review goes into more detail on some factors that may influence the likelihood of switching auditors. 2.1 Auditor switches Two topics which are often discussed in the literature on auditor switches are low-balling and audit quality. The questions addressed are whether low-balling exists for initial engagements and whether audit quality is lower in the earlier years of an engagement. Each of these topics will be discussed in the next sections Low-balling Low-balling refers to the practice of reduced audit fees for initial engagements, which is a competitive response to the expected future quasi-rents (benefits) that the incumbent auditor will receive from the future engagements. Features such as the client s transaction costs associated with switching auditors, the auditor s start-up costs for new engagements, and advantages that the auditor derives from offering services to a certain type of customers (specialization) may enable the incumbent auditor to derive benefits from the future engagements. These future benefits give auditors an incentive to lower the initial price to attract the new client (Chan, 1999; DeAngelo, 1981a; Kanodia & Mukherji, 1994). The initial fee cut can also be explained as being part of an efficient contracting mechanism. In this case, the fee cut acts as collateral which the auditor puts at risk when she colludes with the manager (Lee & Gu, 1998). The existence of low-balling is thoroughly verified by empirical research, although the magnitude of the fee cut depends on the differences between the new and old auditors and on how many auditors are trying to obtain the client (Baber et al., 1987; Ettredge & Greenberg, 1990; Francis & Simon, 1987). Consistent with low-balling being temporary to attract the client, these fee reductions are typically present from the first until the third year, while they vanish in the fourth year of the engagement (Gregory & Collier, 1996; Simon & Francis, 1988; Walker & Casterella, 2000). Furthermore, the practice of

10 Literature review 5 low-balling is still documented in the Sarbanes-Oxley era when prices are generally increasing; the increases are smaller for clients that switch auditors (Ettredge, Li, & Scholz, 2007; Griffin & Lont, 2005). However, outsiders may perceive the financial statements as more reliable when the auditor is not likely to earn future benefits from the client. This may be due to the fact that the auditor appears to be independent of the client in this case. One requirement for this argument is that the quasirents for the auditor are observable by the market (Dye, 1991). Using Australian data, some researchers documented no significant fee cuts for initial engagements (Butterworth & Houghton, 1995; Francis, 1984). This suggests that auditors might refrain themselves from lowballing in order to be (perceived) independent. Firms plan to cut costs in almost all expenditure categories in bad financial times, where more constrained firms plan even deeper cuts (Campello, Graham, & Harvey, 2010). Summarizing the previous discussion, there is a considerable amount of literature that documented the existence of low-balling. While a few studies show no significant fee reductions for initial engagements, these results are found in audit markets outside the United States and may therefore not apply to firms in the United States. Due to the historical pattern, firms may have a reasonable expectation of receiving price cuts when switching auditors (Ettredge et al., 2007). Firms may therefore be increasingly interested in the practice of low-balling during a crisis, since this is one way in which they can temporarily cut costs Audit quality Switching auditors is also likely to influence the financial statement quality. This can occur due to a reduction in the audit quality, which refers to the probability that a misstatement will be detected and reported. Literature indicates that audit quality can be lower after switching auditors, due to both auditor independence impairment and the short tenure of the auditor-client relationship. Auditor independence refers to the likelihood that an auditor will report a detected misstatement in the financial report of the client. Until the low-balling cost is recovered, the auditor might attach substantial importance to future fees and put too little weight on the loss associated with reduced independence (Simon & Francis, 1988). In addition, when the future benefits to the auditor are observable, the existence of low-balling reduces the perceived level of auditor independence (Dye, 1991). Apart from the effect of independence impairment on audit quality, the short tenure can also affect audit quality. The short tenure affects audit quality

11 Literature review 6 through the limited knowledge that the new auditor has on the firm in the initial years of the engagement. This relates directly to the learning curve argument which suggests that the auditor will gain knowledge on the client over time, allowing better audits in later years. Consistent with the argument of reduced audit quality after switching auditors, accruals are usually of lower quality in the first years of a new audit engagement (DeFond & Subramanyam, 1998; Gul et al., 2007; Gul, Yu Kit Fung, & Jaggi, 2009; Johnson et al., 2002; Myers et al., 2003). When focusing on earnings response coefficients as a proxy for perceived earnings quality, shorter auditor tenure is associated with lower perceived quality (Ghosh & Moon, 2005). Two alternative explanations for the association between earnings quality and auditor tenure are that firms with lower earnings quality may just be more likely to switch auditors or that auditors are more likely to resign from engagements with clients that have lower earnings quality (Gul et al., 2009). There are also significantly more audit failures, referred to as the failure of the auditor to issue a going-concern modified audit report before the firm entered into bankruptcy, in the early years of the auditorclient relationship (Geiger & Raghunandan, 2002). However, since fee discounts are a sunk cost in the future, they should not affect auditor behavior according to economic theory (DeAngelo, 1981a). Others suggest that there must be a disagreement on the appropriate reporting practice amongst auditors themselves, the issue must affect the firm for multiple periods, and the issue is neither by the client nor by the auditor seen as very important before the threat of reduced auditor independence occurs (Magee & Tseng, 1990). Low-balling can also be perceived as collateral, serving as an efficient contracting mechanism. Through this collateral the owner can enforce that the auditor acts honestly, since the auditor risks losing it when the owner finds out that she colluded with the managers of the firm (Lee & Gu, 1998). Moreover, the huge fees which the auditor requires to approve misstated reports ruin the credibility of the report in the capital market s view making it unattractive to the firm. When fees are relatively small compared to the legal liabilities of the auditor, the auditor has incentives to be conservative due to the risk of litigation. In addition, low-profitability clients with higher risk are more likely to switch auditors. This enhances the auditor s concern for potential litigation and increases the level of conservatism (Lu, 2006). These arguments are supported by the fact that while clients with a more conservative auditor are more likely to switch, they are approached with the same amount of conservatism by the new auditor (Chow & Rice, 1982; Krishnan, 1994; Krishnan & Stephens, 1995). An experiment among CPAs, financial analysts and commercial loan officers indicated that these groups do not perceive shorter clientauditor relationships as having a higher risk of impaired auditor independence (Shockley, 1981).

12 Literature review 7 A familiarity threat, for which evidence is reported in the United Kingdom (Hudaib & Cooke, 2005), even implies a higher probability of audit quality impairment in later engagement years instead of earlier. This because it is more likely that familiarity becomes a threat in later years, when the client and auditor have a longer connection. This overview indicates that there is mixed evidence with regard to whether audit quality is lower subsequent to an auditor switch. When a client does believe that audit quality can be temporarily reduced by switching, through impaired auditor independence or less client-specific knowledge, the firm may see some benefits to this opportunity in the short term. Earnings management will be less likely to be detected or constrained when audit quality is lower, which increases the opportunities for management to hide actual corporate performance. Furthermore, the client can try to prevent the issuance of a qualified opinion by switching auditors or switch to an auditor that is willing to approve the firm s preferred reporting practice. These attempts are referred to as opinion shopping. In summary, the firm must first of all have some interest in reducing audit quality and the firm must believe that a temporary reduction in audit quality is achievable through switching auditors. During the economic crisis more firms are struggling with lower performance and may have an incentive to hide these poor performances. Furthermore, the firms that are not yet facing a deterioration in performance may be more interested in this additional accounting flexibility due to increased uncertainty about the future during an economic crisis. Therefore, during a crisis the importance of these motivations for auditor switching may be magnified. Overall, the potential for low-balling and reduced audit quality may provide firms with incentives to switch auditors. These motives may be present to a greater extent during an economic crisis, when a downward tendency in financial performance occurs at more firms and is more substantial. In addition, firms whose performance is not yet declining are facing an increased risk of deterioration during a crisis. This may also make these firms more sensitive to the previously mentioned motives for switching in an attempt to stay ahead of the potential downward tendency. The fact that more firms have to deal with a downward tendency in corporate performance suggests that a larger number of firms will be interested in the practice of lowballing to save costs or the reduced audit quality in the early years of an audit engagement to be able to report more favorable financial statements. The notion that during a crisis the downward tendency is more substantial than in better financial times implies that the previously mentioned motives will become more relevant. Furthermore, firms will probably be more active in searching for ways to reduce costs or mask their financial performance when the downward tendency is

13 Literature review 8 more substantial. This may suggest a higher probability of switching auditors during an economic crisis, which brings me to the following hypothesis: H 1a : Firms are more likely to switch auditors during an economic crisis as opposed to a non-crisis period. 2.2 Big 4 and non-big 4 auditors The market for audit services is often separated into two distinct types of firms: the Big 4 (large) and the non-big 4 (smaller). Although the audit market has changed from the presence of a Big 8, to a Big 6, to a Big 5, and currently a Big 4, the distinction has always been the same; large accounting firms versus smaller accounting firms. In the past, researchers have frequently focused on possible differences in the audit fees that these firms charge. Others have investigated whether there is a systematic difference in audit quality between the large and smaller accounting firms. I will address each of these topics in the next two sections Audit fees Linking firm size to the pricing of audit services indicates that the large accounting firms charge a price premium (Craswell et al., 1995; Palmrose, 1986a; Simon & Francis, 1988) and that clients could gain additional fee cuts when switching from a large to a small accounting firm (Ettredge & Greenberg, 1990; Sankaraguruswamy & Whisenant, 2004; Walker & Casterella, 2000). There is also a price premium relative to smaller auditors when second-tier auditors are combined with the Big 4 (Carson, 2009). After the Sarbanes-Oxley act a price premium still exists, although this premium appears to be mainly driven by smaller clients of the Big 4 accounting firms. This seems plausible since smaller clients often do not need the large scope of services provided by the Big 4; the services that the smaller audit firms offer are often sufficient. Diseconomies of scale, which refer to the case where small auditors have to make disproportionately large costs to properly audit large clients due to a lack of the right tools, provide further support for this result. This because smaller auditors experience fewer diseconomies of scale for the smaller Big 4 clients than for the larger clients (Ettredge et al., 2007). Apart from studies focusing on the United States, evidence for the existence of a price premium is also documented in, for instance, the United Kingdom (Chan, Ezzamel, & Gwilliam, 1993; Ireland & Lennox, 2002), Hong Kong (DeFond, Francis, & Wong, 2000; Gul, 1999), Australia (Francis, 1984; Francis & Stokes, 1986) and in an analysis including 15 countries all over the world (Choi, Kim, Liu, & Simunic, 2008).

14 Literature review 9 Nevertheless, some studies only found a price premium for smaller clients of the Big 4 accounting firms. The diseconomies of scale experienced by smaller auditors when auditing large clients could explain this result, since the higher costs for these auditors can reduce the fee spread between large and small auditors (Che-Ahmad & Houghton, 1996; Francis & Stokes, 1986). Similarly, Simunic (1980) argued that large auditors pass their cost savings due to economies of scale on to their large clients. Previous literature seems to agree to a large extent that a price premium exists for the large accounting firms. There are only a few studies that did not document this premium and these studies only provide evidence for large and medium-sized clients. In a period of economic downturn firms plan significant reductions in almost all categories of expenses, where the more constrained firms plan even deeper cuts (Campello et al., 2010). This suggests that the price premium that clients will probably have to pay for the audit engagements with a Big 4 accounting firm may increase in relevance during a crisis. This because switching to a non-big 4 instead of a Big 4 auditor saves the company costs, in the form of the price premium. Therefore, during a crisis in which firms want to cut costs there may be an increase in clients switching from Big 4 to non-big 4 auditors relative to switches to another Big 4 auditor Accounting flexibility If auditors earn quasi-rents on engagements with clients, auditors with more clients have more at stake leading to increased motives to provide high audit quality (referred to as the deep pockets hypothesis). Since it is costly for clients to assess actual audit quality, auditor size might also be used as a reasonable signal for audit quality. Additionally, the potential threat of termination that might motivate the auditor to impair independence is less relevant for large auditors. This because the incentive of colluding with one client is tempered by the amount of quasi-rents of the remaining clients that is put at risk with reducing audit quality (DeAngelo, 1981b). Confirming these arguments, large auditors are more accurate in providing both unqualified opinions to successful firms and qualified opinions to firms that go bankrupt subsequently (Lennox, 1999a). An alternative hypothesis is the reputation hypothesis, which states that large auditors are more likely to provide high quality audits because of the increased motives to avoid criticism that might damage reputation. This hypothesis predicts that large accounting firms provide higher quality since they are more visible if something goes wrong, and therefore have a lower probability of being sued or criticized. Furthermore, the reputation hypothesis predicts that criticism should reduce demand due to reputation damage. However, the fact that large

15 Literature review 10 accounting firms actually have a higher probability of being sued or criticized and that this criticism does not significantly affect the demand for audit services is inconsistent with these predictions. These findings do, however, provide support for the deep pockets hypothesis. This theory can explain the increase in litigation and also notes that criticism should not affect demand because it is not a reliable indicator of auditor accuracy (Lennox, 1999b). Evidence on the level of discretionary accruals (Becker, DeFond, Jiambalvo, & Subramanyam, 1998; Francis, Maydew, & Sparks, 1999; Krishnan, 2003), the level of conservatism (Caramanis & Lennox, 2008; DeFond & Subramanyam, 1998; Francis & Krishnan, 1999; Jenkins & Velury, 2011; Reynolds & Francis, 2001), and compliance with GAAP requirements (Krishnan & Schauer, 2000) provides further support for the argument that audit quality is higher with large accounting firms. Since audits are performed to add credibility to the financial statements used by outsiders, the perception of credibility and reputation by outsiders is also important. The fact that clients of large accounting firms have lower equity risk premiums (Boone, Khurana, & Raman, 2010) and higher perceived earnings quality (Teoh & Wong, 1993) suggests that large auditors provide higher quality. Furthermore, perceived auditor independence is higher with large accounting firms (Al-Ajmi, 2009; Shockley, 1981) and outsiders react most positively to auditor switches to these auditors (Knechel, Naiker, & Pacheco, 2007). Clients of large accounting firms also have more accurate earnings forecasts and lower levels of forecast dispersion (Behn, Choi, & Kang, 2008). Additionally, when firms are issuing new shares and wish to provide more certainty to the market, they are more likely to hire a large auditor (Balvers, McDonald, & Miller, 1988; Beatty, 1989; Firth & Smith, 1992). Despite this compelling amount of literature that supports the quality differentiation between Big 4 and non-big 4 auditors, the level of audit market concentration and differences between auditors can also be explained by determinants other than quality differences in the services provided by auditors. Client size is one of these other determinants that significantly affects quality differences between auditors (Doogar & Easley, 1998; Lawrence, Minutti-Meza, & Zhang, 2011). Additionally, the auditor s dependence on one client s fees does not impair auditor independence according to Australian data (Craswell, Stokes, & Laughton, 2002). This indicates that smaller auditors, where dependence on one client s fees is often greater, do not have a greater risk of losing independence. The comment to this argument that could be made is that results from Australia may not apply to the United States. Focusing on the United States again, the market reaction to auditor switches can also give an indication of audit quality, or at least how it is perceived by the market. The fact that the market does not react more positively, sometimes

16 Literature review 11 even more negatively, to switches from non-big 4 auditors to Big 4 auditors than vice versa implies no quality differences in favor of the Big 4 (Chang, Cheng, & Reichelt, 2010; Nichols & Smith, 1983). Furthermore, acquiring firms in merger activities perform even better when the audit services are provided by non-big 4 auditors (Louis, 2005). Relating the audit quality to earnings management and conservatism, clients of smaller accounting firms should enjoy greater levels of accounting flexibility when smaller auditors indeed provide lower quality. Nevertheless, smaller auditors seem to handle their large clients with a similar level of conservatism as their small clients, suggesting no effect of fee dependence on audit quality (Hunt & Lulseged, 2007). Since accruals are often used to examine earnings management, the insignificant differences between abnormal accruals for clients of Big 4 versus second-tier auditors (Boone et al., 2010) and the insignificant changes in discretionary accruals when clients switch from large to smaller auditors and vice versa also imply that audit quality provided by both groups is equivalent (Jeong & Rho, 2004). One note with regard to the last study is that it is based on the Korean environment, which provides little incentive for auditors to provide high quality. This makes it questionable whether these results can be generalized to the United States. Taken together, the literature that confirms the existence of higher audit quality provided by the Big 4 auditors is substantial. Although there are some studies that do not document this difference, firms may have a reasonable expectation that non-big 4 auditors provide lower quality audits on average. This lower audit quality can be associated with a larger ability of clients to manage earnings upward. Because of the economic crisis, more firms are having difficult times in terms of their performance. This may provide them with incentives to mask the poor performance, which can for instance be achieved through earnings management. Chia, Laplsey and Lee (2007) investigated the effects of Big 6 auditors on the level of earnings management during an economic crisis and found that clients of Big 6 auditors are constrained even more during a crisis. The authors argued this to be caused by the additional oversight of external stakeholders due to uncertainty and poorer performances during a crisis, which leads large accounting firms to be more conservative to protect their reputation. Consistent with greater accounting flexibility, non-big 6 auditors did not force their clients to report more conservative during a crisis. Therefore, during a crisis clients of Big 4 auditors may have an increased incentive to switch to a non-big 4 auditor instead of another Big 4 auditor, because of the potential for greater accounting flexibility.

17 Literature review 12 In sum, the plausible audit fee savings from switching to a non-big 4 instead of a Big 4 auditor and the potential for greater accounting flexibility may influence the choice of the client for a new accounting firm. During an economic crisis, when more firms have to deal with deteriorating performance and when there is an increased uncertainty about the future, the additional savings and future accounting flexibility may be of particular interest to clients. It may help them to reduce costs and mask poor financial performances. This suggests that Big 4 clients will be more likely to switch downwards during an economic crisis. I therefore hypothesize the following: H 1b : Big 4 clients are more likely to switch downwards during an economic crisis as opposed to a non-crisis period. 2.3 Likelihood of switching auditors to save costs I previously mentioned two possible reasons for switching auditors during an economic crisis; the lower audit fees and the increased accounting flexibility. In the next sections I will further investigate the reduced costs reason for switching auditors. I will elaborate on two factors that might influence the likelihood of switching in order to save costs: the level of audit fees and the financial performance of the firm Level of audit fees Audit fees are one of many expenditures of firms. As previously noted, Campello et al. (2010) showed that firms plan to reduce expenditures in almost all categories during an economic crisis. This suggests that firms may also try to reduce the expenses associated with audit services. Since clients have generally witnessed a history of fee cuts associated with auditor switches, they may have a reasonable belief that reducing audit fees through switching audit firms is achievable (Ettredge et al., 2007). If there is a larger proportion of auditor switches motivated by an attempt to reduce audit fees during a crisis, it can be expected that firms will more actively search for the auditor that provides the required services at the lowest price. Also, if there is indeed an increased interest of clients for low prices, auditors may respond to this by aggravating price competition. These arguments suggest that during an economic crisis, the low-balling practice of auditors may be enhanced. The hypothesis is therefore as follows: H 2 : Firms that switch auditors during an economic crisis experience larger drops in audit fees in the year subsequent to switching as opposed to firms that switch auditors in a non-crisis period.

18 Literature review 13 Some researchers have tried to link high audit fees to auditor switches, by making use of the substantial audit fee increases right after the Sarbanes-Oxley Act (Ettredge et al., 2007; Raghunandan & Rama, 2006). This linkage is first of all expected because clients with relatively high fees observe comparable firms that pay lower audit fees, enhancing their belief that reduced audit fees are attainable. Secondly, these firms can potentially save the most on their audit costs. Results provided by Ettredge et al. (2007) confirm these expectations. I examine whether firms that experience relatively high audit fees are more likely to switch auditors during an economic crisis. If companies indeed see audit fees as a cost on which they can save, the magnitude of these potential savings will likely impact the probability of using auditor switching as a tool to reduce costs. Furthermore, due to the tendency to cut costs during an economic crisis, this motive of relatively high audit fees may become increasingly relevant during these periods. I therefore hypothesize the following: H 3 : Firms with relatively high audit fees are more likely to switch auditors during an economic crisis as opposed to a non-crisis period Financial performance During an economic crisis, in which more firms struggle with lower financial performance, most of the firms try to dampen this downward tendency by reducing costs. Additionally, the more struggling firms plan the most substantial cost reductions (Campello et al., 2010). As already discussed, firms may switch auditors because of an expectation to receive fee cuts for initial engagements, as witnessed in past practice. When financial performance of a firm is low, this low-balling practice may become increasingly relevant for companies. In an attempt to improve corporate performance by reducing costs, the firm may dismiss its current auditor and engage with a new accounting firm. That clients with lower profitability and clients that eventually entered into bankruptcy are more likely to switch auditors supports the previous argument (Ettredge et al., 2007; Johnson & Lys, 1990; Schwartz & Menon, 1985). There are, however, some potential reasons why low-balling may be less interesting for companies closer to financial distress. One of the reasons might be that an auditor s share of residual liability losses increases with substantial reductions in the financial performance of a client (Simunic, 1980). Another potential reason is that quasi-rents are earned in the future and that when a client is more likely to go bankrupt due to financial distress this will reduce the potential for future benefits. This may affect the amount of fee cutting that the auditor is willing to provide to new clients with low financial performance. Although empirical support for these arguments is not consistently

19 Literature review 14 documented (Ettredge & Greenberg, 1990), there is some evidence that discounts tend to be smaller for clients in worse financial conditions (Walker & Casterella, 2000). These smaller discounts may sometimes not outweigh the transaction costs which the firm faces when switching auditors, making an attempt to save costs by switching futile. It is probable that firms with lower financial performance will have a higher likelihood of switching auditors in an attempt to reduce auditing costs than comparable firms with better performance. This is due to the fact that the necessity for saving costs is more relevant for firms closer to financial distress. Because an economic crisis increases the financial problems of firms, it can be expected that firms will be more aware of the opportunities to save costs during these periods and auditor switches will occur more often. Also, if auditors indeed engage in incremental price competition during an economic crisis, it is more interesting for firms to switch during a crisis than in a non-crisis period. Furthermore, since switching (especially switching to smaller auditors) can be perceived as reducing audit quality by outsiders, clients of large audit firms that perform relatively well might be discouraged to switch during the crisis in an attempt to signal credibility (Chia et al., 2007). Consequently, it is expected that the relationship described above will be magnified during an economic crisis. This can be summarized in the following hypothesis: H 4 : Firms with lower financial performance are more likely to switch auditors during an economic crisis as opposed to a non-crisis period. In conclusion, one of the actions firms can undertake to survive an economic crisis is switching auditors. Incentives for this choice may be the reduced audit fees in the first years of an engagement or the reduction in audit quality. Furthermore, Big 4 clients may be more likely to choose a non-big 4 successor during a crisis due to the incremental fee reductions and the increased accounting flexibility. Auditors could adapt to the interests of clients by increasing price competition, which would increase the fee drops that switching clients experience. Finally, firms that pay higher audit fees than comparable firms and firms that experience lower financial performance are expected to be more likely to switch in order to save costs during a crisis.

20 Methodology 15 3 Methodology In the first section I will discuss how the sample is constructed. After that, the methodology for each of the two models estimated will be discussed. 3.1 Sample selection This study only contains publicly traded firms in North America. To test the hypotheses, auditor switches in two different periods are compared. Auditor switches in 2009 and 2010 are of primary interest, this is the period representing the economic crisis. The control sample period must be before the crisis. However, Sarbanes-Oxley (SOX) was enacted in 2002, which increased audit fees substantially (Asthana, Balsam, & Kim, 2004; Raghunandan & Rama, 2006). Also, Ettredge et al. (2007) documented a significant effect of audit fees on auditor switches in the immediate post-sox period. Since it can be assumed that this effect will only last for a few years, I use auditor switches in 2005 and 2006 as a control group. The database Audit Analytics is used to collect the information related to the auditor and Compustat is used to collect the other financial data. The initial dataset consists of firm-year observations concerning the economic crisis period and firm-year observations concerning the control period. Some of the variables contain values that are illogical or unusable. Therefore, multiple adjustments are made in order to make the results more reliable. First, I eliminate observations with zero values on variables for which these values are not reasonable, such as total audit fees and total assets. Second, observations with a more than three standard deviations divergence from the variable mean on one or more of the continuous variables needed are eliminated. Also, since the focus of this research is on firm initiated switches, auditor switches marked as auditor resignations are excluded from the sample. Finally, observations with missing values are excluded. The resulting sample sizes for all of the regression models in terms of firm-year observations per period are reported in Table 1. Table 1: Sample size for each of the regression tests Non-crisis period Crisis period Hypothesis Hypothesis Hypothesis Hypothesis

21 Methodology The auditor switches model To test whether the likelihood of switching auditors as well as the likelihood of switching from Big 4 to non-big 4 auditors is higher during an economic crisis (H1a and H1b, respectively), I estimate the following logistic regression model: Switched = b 0 + b 1 Crisis + b 2 LnSales + b 3 Loss + b 4 DebtRatio + b 5 ROA + b 6 Cash + b 7 Growth + b 8 Gocern + b 9 ModOp + b 10 InvRec + b 11 Tenure + b 12 IcWeak + b 13 Big4 + e (1) The dependent variable of this model is the dummy variable Switched that equals one if the firm switched auditors in the relevant fiscal year and equals zero otherwise. This variable thus indicates whether it is the first year of the new auditor. The independent variable of specific interest for this test is the dummy variable Crisis that equals one if the fiscal year is 2009 or 2010 and equals zero otherwise. As I expect that during the crisis firms are more likely to switch auditors, a positive sign is predicted. The control variables in this logistic regression model are selected based on prior research (Ettredge et al., 2007; Ettredge, Heintz, Li, & Scholz, 2011; Landsman, Nelson, & Rountree, 2009). A variable is included to control for size, which is the natural logarithm of the firm s sales 1 (LnSales). Client risk factors are controlled for by financial risk and audit risk proxies. Financial risk can be explained as the risk that a client will run into economic difficulties. Proxies to control for financial risk are the dummy variable Loss that equals one if the firm reports negative earnings and equals zero otherwise, total debt divided by total assets (DebtRatio), income before extraordinary items divided by total assets (ROA), and cash divided by total assets (Cash). Audit risk can be defined as the risk that an unqualified opinion is given by the auditor while the client s statements contain material errors. Proxies included for this type of risk are a measure of growth that equals the book value of equity divided by the market value of equity (Growth), the dummy variable Gocern that equals one if the company received a goingconcern opinion and equals zero otherwise, the dummy variable ModOp that equals one if the company received a modified audit opinion which is not a going-concern opinion and equals zero otherwise, the level of inventories and receivables divided by total assets (InvRec), the number of years (with a maximum of ten) that the current auditor audited the statements (Tenure), and the dummy variable IcWeak that equals one if the auditor assessed a material weakness in its internal controls and equals zero otherwise. Finally, the dummy variable Big4 that equals one if the auditor is a Big 4 accounting firm and equals zero otherwise is added. 1 Corrected for inflation using the consumer price index (U.S. Department Of Labor, 2012).

22 Methodology 17 H1b is tested using a comparable model. However, the sample will be restricted to clients that are engaged with a Big 4 auditor or dismissed a Big 4 auditor in the relevant fiscal year. Due to this sample restriction the Big 4 dummy variable can be eliminated from the model. The model is then estimated twice: once excluding downward switches and once excluding lateral switches. Downward switches refer to switches to a smaller auditor, whereas lateral switches refer to switches to a comparable auditor. When only downward switches are included, the expected sign for the crisis variable is positive. On the contrary, the coefficient is expected to be negative when only lateral switches are included. These expected signs are derived from the prediction that Big 4 clients are more likely to switch downwards during an economic crisis. Comparing these coefficients can indicate whether there is an increased likelihood of switching to non-big 4 auditors. As a further test, the sample is focused on firms that left a Big 4 accounting firm. This model includes the same variables as previously discussed, with some adjustments. The dependent variable is the dummy variable Downward that equals one if the new auditor is a non- Big 4 accounting firm and equals zero otherwise, and the Big 4 dummy variable is again excluded. The Crisis variable is expected to have a positive coefficient in this regression. In order to test whether firms are more likely to switch auditors when they are charged relatively high audit fees during an economic crisis (H3), I add a proxy for the level of abnormal audit fees charged (Afee). This variable is the residual from a fee model that will be discussed in the next section. Since higher relative audit fees are expected to be associated with a higher likelihood of switching auditors, a positive sign is predicted for the abnormal fee variable. An interaction term between the dummy variable for whether the fiscal year is during the economic crisis and the measure of abnormal fees is included to assess the impact of the economic crisis on the client s sensitivity to the relative level of audit fees. A positive sign is also expected for this variable due to the incremental sensitivity to the abnormal fee level predicted during an economic crisis. To test H4, whether firms experiencing lower financial performance are more likely to switch auditors during an economic crisis, two interaction variables are added to the model. These two variables are the return on assets (ROA) and negative earnings dummy (Loss) interacted with the dummy variable for whether the fiscal year is during the economic crisis (Crisis). The return on assets and negative earnings variables can give an indication of whether the motive of switching auditors due to the firm s financial performance is mainly related to reporting a profit or loss, or to the magnitude of the losses. The interaction terms are used to investigate the impact of the economic crisis on the sensitivity of the client to these motives for switching auditors. As I