1 Audit Quality and Auditor Switching: Some Lessons for Policy Makers CLIVE S. LENNOX Economics Department, Bristol University. Address for Correspondence University of Bristol Department of Economics 8 Woodland Road Bristol BS8 1TN England 2 Audit Quality and Auditor Switching: Some Lessons for Policy Makers 3 1. Introduction. This paper reviews the literature on audit quality and auditor switching to assess different countries’ policy regimes. It argues that policy-makers should limit managerial influence over auditor switching rather than reduce auditors’ economic dependency on clients. In particular, the paper advocates proper communication between shareholders and auditors, and a policy of mandatory auditor retention. In contrast, some countries have adopted policies of mandatory rotation, and have banned non-audit services and introductory fee discounts. It is argued that such policies are less desirable on both theoretical and empirical grounds. A lack of audit quality is most often alleged when it is believed that auditors should have warned investors about impending corporate failures (these are often termed “audit failures”). 1 The next section argues that audit failure is a significant cause for concern. Section 3 explains how auditor switching can affect audit quality. Section 4 describes regulatory regimes aimed at increasing audit quality. Sections 5 and 6 discuss the theoretical and empirical literature on auditor switching and audit quality. Section 7 concludes with policy recommendations. 2. Audit Failure - Is it a Problem? In the UK, an auditor who believes that a company is about to fail is required to signal this to investors by giving a going concern qualification in the audit report. The Guideline on Going Concern (1985) states, “The going concern concept identified in Statement of Standard Accounting Practice No. 2 is ‘that the enterprise will continue in operational existence for the foreseeable future’. This means in particular that the profit and loss account and the balance sheet assume no intention or necessity to liquidate or curtail significantly the scale of operation.” The rationale for the existence of going concern qualifications is to warn investors that the company’s financial statements are reported under the assumption that the company 1 Congressman Wyden (1986) has stated before Congress, “In one financial disaster after another . . . the disaster struck virtually on the heels of clean audit certificates issued by audit firms.” 4 will remain a going concern. In the event of liquidation, the company’s value may be very different to that reported in the accounts. However, the auditor is required to ensure that shareholders are warned, even if the reported value of the company corresponds to its liquidation value. 2 The Guideline states, “Where there is significant uncertainty about the enterprise’s ability to continue in business, this fact should be stated in the financial statements even where there is no likely impact on the carrying value and classification of assets and liabilities”. UK evidence indicates that audit reports are not accurate indicators of financial distress. Only 20-27% of failing companies receive qualified reports (Taffler and Tissaw, 1977; Taffler and Tseung, 1984; Citron and Taffler, 1992). Moreover, in a sample of 40 companies that received qualified reports, Taffler and Tseung (1984) found that only 10 failed. To illustrate this, Table 1 shows the correlation between audit reporting and company failure in the UK (1987-94). 3 Table 1 The correlation between financial health and audit reporting. Q it = 0 Q it = 1 GQ it = 0 GQ it = 1 FAILS it = 0 6804 177 6878 103 FAILS it = 1 97 26 102 21 Type I error (%) 78.86 82.93 Type II error (%) 2.54 1.48 FAILS it = 1 if company i received its final audit report in year t prior to entering bankruptcy; = 0 otherwise. Q it = 1 if company i received a qualified report in year t; = 0 otherwise. In addition, to going concern qualifications, qualified reports were given for non-compliance with Statements of Standard Accounting Practice, and due to uncertainty regarding provisions for bad debts, slow-moving stocks and litigation. GQ it = 1 if company i received a going concern qualification in year t; = 0 otherwise. 2 Altman (1982) has argued that the correlation between audit qualifications and corporate failure may not give an accurate measure of audit quality. This is because an auditor might choose to give an unqualified report to a company which is likely to fail, if the accounts give a true and fair view of the company’s liquidation value. Whilst this argument holds for the US, it does not apply to the UK. 3 More details about this sample are given in Lennox (1998a). 5 Following Koh (1991), the accuracy of audit reports is measured in terms of type I and type II errors. The type I error rate is equal to the proportion of failing companies that are given unqualified reports; the type II error rate is equal to the proportion of non-failing companies that are given qualified reports. The type I error rate is very large (approximately 80%) compared to the type II error rate (approximately 2%) - it is therefore unsurprising that audit failure (a type I error) is perceived to be a significant problem. Altman (1977) has estimated that type I errors are seven times more costly to accounts users than type II errors. This implies that, as well as reducing the incidence of both type I and type II errors, policy should ensure that there are approximately seven times as many type II errors as type I errors. For most countries, this means that the rate of audit qualifications needs to be substantially higher. 4 Although most failing companies are not given qualified reports and most qualified reports are not followed by bankruptcy, it does not necessarily follow that auditors are failing in their responsibilities. The low predictive power of most bankruptcy models suggests that auditors may not accurately warn about impending failure when bankruptcy is an unforeseeable event. However, there are three reasons for arguing that audit reports could be more accurate. First, reports are poor indicators of financial distress compared to the predictions of bankruptcy models (Altman and McGough, 1974; Koh, 1991; Lennox, 1998a). Secondly, the number and scale of successful litigation cases suggest that auditors sometimes fail in their responsibilities towards shareholders (Palmrose, 1988). Finally, regulatory investigations (by the Department of Trade and Industry (DTI) in the UK, and the Stock Exchange Commission (SEC) in the US) have often been critical of auditors (Firth, 1990; Wilson and Grimlund, 1990; Davis and Simon, 1992). Therefore, audit failure is a significant and important problem - the following section discusses how auditor switching affects the quality of audit reporting. 4 In the US, Carcello et al. (1995) report that type I error rates were approximately 48% between 1972-92 (evidence for type II error rates was not given). 6 3. The relationship between auditor switching and audit quality It has been argued that companies use auditor switching to avoid receiving qualified reports. This argument assumes that managers dislike qualified reports and that managers influence the auditor appointment decision. The first assumption is relatively uncontroversial - a qualified report may signal to investors that managers are poor stewards of the company’s affairs, or that managers have attempted to present an over-favourable view of the company’s performance. In addition, qualified reports cause share prices to fall - this reduces managerial utility if managers own shares or if their compensation is related to market value (Firth, 1978; Banks and Kinney, 1982; Fleak and Wilson, 1994; Chen and Church, 1996; Jones, 1996). Therefore, there are strong grounds for believing that managers dislike receiving qualified reports. The second assumption is more controversial because, de jure, auditors are appointed by shareholders. However, de facto, managers exert considerable influence over auditor appointments. For example, managers often dismiss incumbent auditors without consulting shareholders - shareholders merely vote on whether to accept their recommendation regarding the appointment of a new auditor or the re-appointment of the incumbent auditor. Thus, the right to dismiss an auditor lies mainly with managers. Secondly, managers have some influence over the appointment of a new auditor or the re-appointment of the incumbent auditor. They set meeting agendas when auditor appointments are proposed, and they typically have the proxy votes of a large number of shareholders. Thus, managers have considerable influence over auditor hiring and firing. If managers dislike qualified reports and have some influence over auditor appointment, they may try to use auditor switching to avoid receiving qualified reports. Teoh (1992) usefully identified two ways in which this could occur. First, a manager may actively use the auditor switch decision to avoid receiving a qualified report. If a new auditor is less likely to give a qualified report compared to the incumbent auditor, the manager may choose to 7 switch; similarly, if a new auditor is more likely to give a qualified report compared to the incumbent, the manager may choose not to switch. This active use of auditor switching is called ‘opinion-shopping’ in this paper. Secondly, if auditors earn client-specific rents, a manager may obtain a more favourable report from an incumbent auditor by threatening to switch to a new auditor. The next section describes regulatory safeguards that reduce the scope for opinion-shopping and the switch threat. 4. Regulatory Safeguards Policies aimed at curbing opinion-shopping focus on reducing managerial influence over auditor switching. On the other hand, policies aimed at reducing the potency of the switch threat deal with both managerial influence and auditors’ economic dependency on clients. First, the discussion focuses on managerial influence over auditor switching. 4.1 Managerial influence over auditor switching 4.1.1 Communication between an outgoing auditor and shareholders Most countries require outgoing auditors to communicate with shareholders and/or regulatory bodies. 5 The aim of such communication is to prevent managers using switching to conceal unfavourable information. In the US, the SEC must be informed of the reasons for a change in auditor, and the auditor is entitled to be heard by shareholders. The SEC have introduced a number of disclosure requirements over the years seeking to increase communication between shareholders and auditors (for example SEC, 1988 and 1989). More recently, the US Public Oversight Board (1994) recommended that auditors express judgements to boards of directors and audit 5 In the UK, a dismissed auditor has the right to make written representations to shareholders or to speak against the resolution for his removal at the company’s Annual General Meeting (AGM). In the UK, an auditor’s notice of resignation is not effective unless it contains either: (a) a statement to the effect that there are no circumstances connected with the resignation which should be brought to the notice of the members or creditors of the company, or (b) a statement of any such circumstances. In Denmark, Ireland, the Netherlands, Spain and Sweden, auditors are required to disclose the reason for a change of auditor. In all EU countries with the exception of Spain, auditors have the right to defend their positions if threatened with dismissal. 8 committees about the appropriateness of the financial statements, and that auditors meet with boards of directors and audit committees at least once a year. However, in many countries outgoing auditors rarely communicate with shareholders even when there is a genuine cause for concern. 6 For example, the DTI’s investigation of Ramor Investments (1983) concluded, “We are critical of the auditors’ work . . . and of the fact that when Price Waterhouse did resign they went too quietly.” 7 One problem may be that auditors do not wish to gain reputations as trouble-makers amongst the managers of other companies. 8 In addition, auditors may be concerned that communication would reveal previous audit errors and increase the likelihood of litigation. 4.1.2 Communication between outgoing and incoming auditors 6 A problem in the UK is that the auditor’s statement must first be sent to the company’s managers, who are then required to send copies to all shareholders (Dunn, 1991). If managers fail to do this, the auditor has the right to read the statement at the AGM. However, such meetings are often attended by few shareholders. Therefore, scope for direct communication between auditors and shareholders is limited. 7 The inspectors found that Price Waterhouse had resigned because of: the intermingling of a director’s (Mr. S) personal accounts with those of the company; a lack of co-operation from the same director; an instance of outright deception; and errors in a previous audit. Rather than revealing these facts to shareholders at the company’s AGM, the partner in charge of the audit (Mr. A) had written to Mr. S the following letter: “Dear Mr. S, As arranged I am writing to let you know in advance of the Annual General Meeting the replies I will give if I am asked by a shareholder for the reasons why my firm is not seeking re-election as auditors. If no questions are asked then, of course, no further information in addition to that contained in the Annual Report need be provided. However, if a shareholder asks for further information I propose to reply as follows: “In recent years we have experienced certain difficulties in obtaining the necessary information for our audit and being sure that all relevant explanations have been provided to us. In the final outcome we have been satisfied that we have received all such information and explanations; otherwise this would have been reflected in our audit report. However, the situation created by these difficulties caused us to agree with the directors that we would not seek re-election at this meeting, a step we are permitted to take under the provisions of the Companies Act.” If there should be a follow-up question asking for more information about the difficulties referred to in the foregoing statement I would propose to reply as follows: “There was no one matter which in itself caused us to reach this agreement with the directors. In view of this, there is nothing more that can be added to the answer that has already been given.” I would not intend to give any more information nor to respond to any other questions. Yours sincerely, Mr. A.” In fact Mr. A was not asked any questions at the AGM. As he had pointed out, shareholders could only have known what was in the annual report. 8 In theory, an auditor should wish to gain a reputation as a ‘whistle-blower’ so as to improve his reputation amongst shareholders. In practice, this argument does not seem to be important. For example, the Financial Times (23:1:97) has reported that an anonymous hotline has been set up for UK auditors to act as whistle-blowers to the pensions regulator. The fact that the regulator believed the hotline needed to be anonymous suggests that auditors do not have an incentive to gain a reputation for being whistle-blowers. This is consistent with the argument that managers rather than shareholders have most influence over auditor appointment. 9 In most countries, outgoing auditors are required to inform incoming auditors if there are worrying circumstances that prompted the change in auditor. 9 The aim of this requirement is to prevent a manager concealing unfavourable information by switching to a less well-informed new auditor. However, communication between incoming and outgoing auditors does not occur very frequently. For example, the DTI’s investigation of Ramor Investments found that Price Waterhouse had resigned because of matters relating to fraud. However, Price Waterhouse did not reveal this to the incoming auditors Norton Keen - rather they were told that the resignation decision was based on commercial grounds. These facts led the inspectors to conclude that Price Waterhouse’s conduct with regard to the incoming auditors was “indefensible”. More systematic evidence has been provided by Dunn et al. (1994) who found that UK auditors only disclosed why resignations occurred in 19 out of 793 (2.4%) cases - this was despite a high incidence of audit qualifications before and after the resignations. 10 Scope for communication between incoming and outgoing auditors appears to be greater in the US. In 1975, the SEC introduced Accounting Series Release No. 165 requiring auditor changes to be disclosed when the incumbent auditor’s appointment was terminated rather than when a new auditor was appointed. Smith (1988) found that the increased timeliness of information about auditor changes provided more useful information to investors. This suggests that policies aimed at improving disclosure about auditor changes can be effective. More recently, the American Institute of Certified Public Accountants (AICPA) issued SAS 84 (1997) which increased incoming auditors’ rights of access to outgoing auditors’ working papers. 11 4.1.3 Mandatory rotation and retention of auditors 9 In the EU, only Austria, Finland, Greece and Italy do not require communication between incoming and outgoing auditors (Buijink et al., 1996). 10 In the UK, one problem is that legislation does not specify what an outgoing auditor should reveal. 11 In its modified audit report on Alexanders Holdings (a UK quoted company), Clark Whitehill stated that it was unable to check the opening financial position because the previous auditors did not allow access to the prior year’s working papers. 10 Mandatory rotation or retention of audit firms reduces managerial influence over auditor switching, and therefore limits the use of opinion-shopping and the switch threat. In addition, mandatory rotation reduces the potency of the manager’s switch threat by lowering the expected period of incumbency and reducing expected rents. 12 Voluntary rotation occurs very infrequently - in the US, the rate of auditor switching is only 1-6% per year with 75% of clients hiring the same auditor for 17 or more years (DeAngelo, 1981; Beck et al., 1988). 13 Whilst Italian companies regulated by the Stock Exchange Regulatory Authority are required to rotate audit firms every nine years, no such provisions exist in other EU countries, the US or Australia. 14 In the US, the Metcalf report (1976) recommended mandatory rotation of audit firms but this was later rejected by AICPA (1978) and the SEC Practice Section (AICPA, 1992). Mandatory retention of audit firms has been adopted in Belgium, Spain and Italy (three years), Portugal (four years) and France (six years), but not in other EU countries, the US or Australia. Some policy-makers have argued that mandatory rotation could reduce audit quality because incumbent auditors have greater knowledge of clients (AICPA, 1978; Ryan Commission, 1992). Consistent with this, studies have shown that it takes time for newly appointed auditors to become familiar with clients and shorter periods of incumbency reduce auditors’ incentives to invest in learning-by-doing (St. Pierre and Andersen, 1994; Arrunada and Paz-Ares, 1997). In a survey of European auditors, Ridyard and Bolle (1991) found that 1- 2 years were needed to gain client familiarity in industries where the auditor had no previous experience; 2 years were needed for industries in which the auditor had experience. 12 There is little systematic evidence for the view that long auditor-client relationships reduce auditors’ incentives to maintain independence, although there is some anecdotal evidence. In the DTI’s investigation of Rotaprint (1991), the fact that Joselyne Layton-Bennet had been Rotaprint’s auditor for thirty years was identified as being a potential problem (although the reason why this was perceived to be a problem was not revealed). 13 In the UK, the rate of auditor switching is 4% per year with more than half of companies having the same auditor for more than 20 years (Ridyard and De Bolle, 1992; Lennox, 1998b). 14 Mandatory rotation of audit firms every 12 years was proposed by the Fifth EU Directive (1989). In the UK, the Cadbury Committee (1992) recommended rotation of individual audit partners but not the rotation of audit firms. The Australian Society of Certified Public Accountants (1993) recommended partner rotation every seven years, but this was subsequently rejected by the Australian Joint Standing Committee (1996) because of the belief that second partner review would overcome problems of lengthy partner tenure. In the US, audit members of the SEC Practice Section are required to rotate partners every seven years. [...]... Incremental Information Content Of Audit Reports, WORKING PAPER (1998a) C Lennox, Bankruptcy, Auditor Switching And Audit Failure: Evidence From The UK 198 7-9 4, WORKING PAPER (1998b) C Lennox, Non -Audit Fees, Disclosure And Auditor Independence, WORKING PAPER (1998c) R Magee and M Tseng, Audit Pricing And Independence, THE ACCOUNTING REVIEW, 65 (1990) E Matsumura, K Subramanyam and R Tucker, Strategic Auditor. .. undesirable consequences for auditor effort and audit quality Even if it were desirable to reduce the level of auditor dependence, it is unclear how this could be achieved Theory predicts an ambiguous relationship between the provision of non -audit services and auditor dependence, and the evidence does not indicate that non -audit fees significantly affect dependence Therefore, a ban on non -audit services does... been taken - the first favoured by the US, UK and Australia relies on communication between shareholders, incoming and outgoing auditors; the second, favoured by some European countries includes mandatory rotation and/ or retention of auditors Policy- makers have also limited the potency of the switch threat by reducing auditors’ client-specific rents With respect to non -audit services, the UK and Australia... audit quality is also ambiguous On the one hand, when auditors low ball the higher future rents needed to offset the initial loss could increase auditor dependence (DeAngelo, 1981; Simunic, 1984).21 On the other hand, auditors may use low balling as a signal of quality (Craswell et al., 1996) If high quality auditors are more likely to be retained than low quality auditors, then high quality auditors... controlled for financial health, and so their results may be more reliable 18 This paper drew a distinction between policies that reduce managerial influence over auditor switching (improved communication and mandatory auditor rotation/retention) and policies that reduce auditors’ client-specific rents (bans on low balling and non -audit services) Policies that limit managerial influence over auditor switching. .. problem, Lennox (1998b) controlled for the effects of opinion-shopping and financial health on auditor switching and showed that short tenure increases the likelihood of audit failure This supports the argument that mandatory rotation is not a desirable policy 4.2 Reducing auditors’ client-specific rents Concern over the switch threat has led some countries to maintain auditor independence by reducing auditors’... switch threat, and that this could compromise auditor independence In addition, there have been concerns that audit firms are unwilling to criticise work carried out by non -audit departments On the other hand, non -audit services could improve audit quality by increasing auditors’ knowledge of clients Policies on non -audit services include laissez-faire, disclosure of non -audit fees, and prohibition... ON THE ROLE, THE POSITION AND THE LIABILITY OF THE STATUTORY AUDITOR WITHIN THE EUROPEAN UNION (1996) Z Palmrose, An Analysis Of Auditor Litigation And Audit Service Quality, THE ACCOUNTING REVIEW 63 (1988) K Pany and P Rekers, Auditor Independence And Non -Audit Services: Director Views And Their Policy Recommendations, JOURNAL OF ACCOUNTING AND PUBLIC POLICY (1983) K Pany and P Rekers, Within Versus... outgoing and incoming auditors Whilst mandatory rotation reduces managerial influence, it also tends to replace well-informed incumbent auditors with less well-informed newly appointed auditors In contrast, mandatory retention does not suffer from this problem, but does reduce the ability of managers to engage in opinion-shopping Therefore, mandatory retention may be preferable to a policy of mandatory... rotation - this conclusion is strengthened if one takes into account the switching costs that are incurred when companies change auditor Unfortunately, theory and evidence is much less clear regarding auditors’ rents and dependency on clients Theory implies that the optimal level of auditor dependence ensures that auditors have proper incentives to exert effort Therefore, policies aimed at reducing auditor . Woodland Road Bristol BS8 1TN England 2 Audit Quality and Auditor Switching: Some Lessons for Policy Makers 3 1. Introduction. This paper reviews the literature on audit quality and auditor switching. shareholders and auditors, and a policy of mandatory auditor retention. In contrast, some countries have adopted policies of mandatory rotation, and have banned non -audit services and introductory. influence over auditor switching (improved communication and mandatory auditor rotation/retention) and policies that reduce auditors’ client-specific rents (bans on low balling and non -audit services).