A Model of Mandatory Auditor Rotation Bryan K. Church College of Management Georgia Tech Atlanta, GA 30332-0520 404.894.3907 bryan.church@mgt.gatech.edu Ping Zhang Rotman School of Management University of Toronto Toronto, ON M5S 3E6 416.946.5655 pzhang@mgmt.utoronto.ca January 2006 We gratefully acknowledge the helpful comments of Shawn Davis. A Model of Mandatory Auditor Rotation Abstract We develop a theoretical model to compare the relative merits of a system that requires auditor rotation to one that does not. We show that auditor independence is improved with mandatory rotation, but that the net benefit is sensitive to the rotation period, startup cost, the cost associated with biased reports, auditors’ learning, and the time span of managers’ incentives. Mandatory auditor rotation is preferable if the rotation period is long, startup costs are high, the cost of biased reports is high, auditor learning is dramatic in terms of improving audit efficiencies over time, and the manager is myopic (focused on short-term payoffs). A Model of Mandatory Auditor Rotation I. Introduction The need for mandatory auditor rotation has long been debated. A primary concern is that if the incumbent is retained over time, the auditor may become complacent and objectivity may be impaired (e.g., Mautz and Sharaf, 1961). Regulators have periodically called for mandatory rotation in order to preserve auditor independence, including the Metcalf Report (United States Senate, 1976), the Securities and Exchange Commission’s Staff Report on Auditor Independence (SEC, 1994), and Congressional testimony (see e.g., Zeff, 2003). In the United States, the accounting profession has staunchly opposed mandatory rotation, arguing that such a requirement would increase the cost and risk of conducting an audit (e.g., American Institute of Certified Public Accountants, 1978; General Accounting Office, 1996). The profession contends that problems are more likely to occur in the initial years of the auditor-firm relationship because the auditor has less knowledge of the firm. In support of the argument, the American Institute of Certified Public Accountants (1992) documents that cases of alleged audit failure (between 1979 and 1991) occur three times more frequently in the first two years of the auditor-firm relationship (see also Palmrose, 1986; 1991). Carcello and Nagy (2004) show that fraudulent financial reporting, as evinced by the SEC’s Accounting and Auditing Enforcement Releases between 1990 and 2001, is more likely to occur in the first three years of the auditor-firm relationship. 2 Yet the issue of mandatory rotation continues to be discussed and has resurfaced as a fall out of the accounting scandals that occurred at the beginning of the 21 st century. Section 207 of the Sarbanes-Oxley Act of 2002 requires the U.S. Comptroller General to study the potential effects of mandatory auditor rotation. The inherent difficulty of such a study, though, is that the auditor-firm relationship can only be observed under the current system. Hence, it is problematic to assess the relative merits of a system that mandates auditor rotation compared to one that does not require rotation. 1 To shed insight into the auditor rotation issue, archival research has examined the relationship between auditor tenure and proxies for audit quality and/or the credibility of financial data. Geiger and Raghunandan (2002) find a negative association between auditor tenure and auditor-reporting failure. 2 Mansi, Maxwell, and Miller (2004) report a negative association between auditor tenure and the cost of debt financing. Others document a positive association between auditor tenure and earnings quality – real and perceived (Johnson, Khurana, and Reynolds, 2002; Myers, Myers, and Omer, 2003; Ghosh and Moon, 2005). Accordingly, archival findings suggest that benefits arise as auditor tenure is extended. But as mentioned earlier, archival studies are unable to determine the effect of mandatory auditor rotation because data are available only under the current system (i.e., one that does not require rotation). Dopuch, King, and Schwartz (2001) conduct a controlled, laboratory study, which allows them to assess behavior with and without 1 Some countries mandate auditor rotation, including Austria (every six years) and Italy (every nine years). Greece requires rotation every six years for public sector companies, and banks incorporated in Singapore must rotate every five years. Spain previously required rotation (every nine years), but repealed the law one year before it would have had a practical effect. 2 According to Geiger and Raghunandan (2002), auditor-reporting failure occurs when the auditor issues a clean audit report to a firm that files for bankruptcy in the following year. 3 auditor rotation. Their findings indicate that mandatory rotation can improve auditor independence, but that welfare may be higher (for the auditor and the firm) in the absence of rotation. While the findings are intriguing, the experimental design does not provide for a comparison of the total costs and benefits of the alternative systems. Such a comparison is necessary, however, to determine whether mandatory auditor rotation is optimal. Because prior research does not provide a basis to perform a complete analysis of the relative merits of mandatory rotation, we develop a theoretical model to formally consider the issue. We find that auditor independence is improved with mandatory rotation, but that the net benefit is sensitive to the rotation period, startup cost, the cost associated with biased reports, auditors’ learning, and the time span of managers’ incentives. Mandatory auditor rotation is preferable if the rotation period is long, startup costs are high, the cost of biased reports is high, auditor learning is dramatic in terms of improving audit efficiencies over time, and the manager is myopic (focused on short-term payoffs). The remainder of the paper is organized as follows. Section II presents the basic model, and Section III introduces the firm’s incentive to report biased accounting values. Sections IV and V examine the effect of auditor experience and managerial payoffs, respectively, on the results. Section VI provides concluding remarks. II. The Model We start with a model similar to that of DeAngelo (1981) and Magee and Tseng (1990). Each operating period a firm hires an auditor to verify the reported accounting 4 value (financial data). The engagement is for one period and contingent fees are not allowed. At the beginning of the period, auditors submit bids to provide services. A firm hires the auditor whose bid provides the lowest cost. If the incumbent auditor submits the lowest cost bid, the incumbent is retained. Otherwise, a new auditor is hired. 3 When a new auditor is hired, startup costs (i.e., one-time fixed costs) are incurred by each party, denoted M for the firm and A for the auditor. We assume that the audit market is competitive and all auditors possess the same audit technology. 4 To compete for an engagement, auditors submit bids that cover their minimized costs. The auditor’s cost per period includes two components: the cost of conducting the audit and the expected cost of liability from audit failures. Let ε be audit effort, C be the examination cost, u be misstatement in the reported accounting value, and L be the expected liability. Then C is an increasing function of ε, and L is an increasing function of the absolute value of u. If there is no known bias in the reported accounting value, the size of misstatement equals the size of audit error. The audit error, denoted x, is a random number with an expected value of zero and its absolute value is a decreasing function of ε. Let L x ≡L(u=x), the auditor’s expected liability from audit errors alone. Auditor effort is chosen to minimize the cost C+L x each period. 5 We first consider the auditor’s pricing strategy when the firm’s manager does not have an incentive to report any specific (biased) accounting value, referred to as the benchmark case. We start by introducing three parameter variables. The first variable is the discount factor, denoted γ. The second is the expected auditor turnover rate per period, 3 If more than one new auditor submits the lowest cost bid, the firm randomly chooses a new auditor. 4 These assumptions are not essential for the conclusions obtained in the study, but allow us to streamline the analysis and presentation. 5 The bias incorporated in the accounting value is negotiated between the auditor and manager and is not directly affected by the auditor’s effort. 5 denoted δ. The value of δ reflects the possibility that (1) the firm ceases operations at period end and (2) the auditor severs the relationship with the firm due to cost and/or specialization considerations. Thus, δ reflects a normal turnover rate before considering disagreements. 6 At period end, the incumbent knows whether the firm requires services in the next period and whether the incumbent will be available to submit a bid to provide such services. The third parameter variable is the maximum number of consecutive periods that the incumbent can be retained by a specific firm, denoted N. The incumbent is retained (prior to period N+1) as long as the incumbent’s bid equals the lowest bid from competing new auditors plus M (the startup cost incurred by the firm to engage a new auditor). If P is the lowest price that a new auditor bids, then in the first period all auditors bid P and one is hired. In all subsequent periods the incumbent bids P+M and is retained. At time 1, the incumbent is available to bid with a probability of δ in the second period, a probability of δ i-1 for the i th period, and a probability of δ N-1 for the final period of service. At the beginning of the first period, the auditor prices the audit such that the expected discounted total cost equals the expected discounted total revenue: that is, economic profit is zero in a competitive audit market. The quasi rent generated by the auditor in period i, denoted q i , is revenue minus cost for the period and the total quasi rents after period i are the expected present value of all future quasi rents, which has a value of Q i+1 = q i+1 + q i+2 δγ+q i+3 (δγ) 2 +…+ q N-i (δγ) N-i-1 . 6 Turnover also may arise due to disagreements over the accounting value, which can result in the auditor resigning or being fired. In the benchmark case, the firm does not have an incentive to report a biased accounting value and, as such, disagreements over the accounting value do not arise. We relax this assumption later. 6 Lemma 1: When the maximum number of consecutive periods that the auditor can be retained is N, the expected present value of all future quasi rents after period i, Q i+1 , is (M+A)(1-(δγ) N-i )/(1-(δγ) N ), which is a decreasing function of i. If N is infinitely large, the future quasi rent after period i is a constant of (M+A). Proof: See Appendix. III. Auditor Independence in a Non-Cooperative Game We introduce the assumption that the firm’s manager has an incentive to report a specific accounting value in some periods. The incentive arises with a probability of α (≤1) at the end of each period. Let d be the difference between the reported accounting value and the auditor’s unbiased estimate of value. 7 The size of misstatement (u) is the sum of the deviation (d) and the audit error (x), where the deviation (d) equals the difference between the manager’s desired reporting value and the auditor’s unbiased estimate. 8 The auditor’s expected liability is L(u). The auditor chooses effort, ε, to minimize the cost C(ε)+L x , where L x ≡L(u=x) is the auditor’s expected liability from audit errors alone. Let D be the auditor’s expected incremental liability of endorsing the firm’s desired accounting value, then D=L(u)-L x for a given audit effort. The value of D is an increasing function of d and D>Q i if d>d i * ). When the firm’s manager has an incentive to report a specific, desired accounting value, the manager presents the value to the auditor for endorsement. If the auditor 7 In this paper we assume that the manager knows the auditor’s unbiased estimate of the firm value. This assumption simplifies the analysis, but does not alter the paper’s main results. Interested readers are referred to Zhang (1999) in which the auditor’s unbiased estimate is private information and the manager is unable to observe it. 8 The manager’s desired reporting value arises for various reasons, including the market’s expectation of the reported accounting value and the effect of the reported value on the manager’s compensation. 7 endorses the desired value, the manager receives a benefit of B and the auditor incurs an incremental liability with expected value of D. If the auditor does not endorse the desired accounting value, the manager does not receive the benefit and the auditor does not incur the incremental liability. However, the auditor may not be retained in future periods (i.e., the auditor may not be permitted to submit a bid for the next period’s audit service). Proposition 1: The auditor does not endorse the manager’s desired accounting value in period i if a. the auditor is not available to provide audit services in the next period, even if the reported accounting value is endorsed, or b. the required deviation to report the desired accounting value is greater than d i * , or c. the maximum number of consecutive periods that the auditor can be retained is definite. Otherwise, the auditor compromises independence by endorsing the manager’s desired accounting value. Proof: See Appendix. Proposition 1 implies that the auditor compromises independence as long as each of the following conditions is satisfied: (1) the auditor derives benefits from repeat engagements with the firm, (2) the expected future benefits exceed the incremental expected liability that arises from endorsing the firm’s desired accounting value, and (3) the auditor may be retained indefinitely. 8 The benefit from endorsing the desired accounting value is zero if the auditor is not available to provide audit services in the next period (B=0). In this situation, the auditor does not endorse a biased report because such behavior increases the auditor’s expected liability without a consequent gain (D>0). If the auditor is available to provide audit services in the next period, endorsing the desired accounting value produces expected benefits from potential future engagements. However, the auditor does not endorse a biased report if the expected benefit is less than the incremental expected liability that accompanies such behavior (B<D). Hence, the auditor does not endorse a biased report that exceeds an upper limit. If the maximum number of consecutive periods that the auditor can be retained is definite (i.e., auditor rotation is mandated), the auditor does not endorse a biased report even if the expected benefit from potential future engagements exceeds the incremental expected liability. When the number of consecutive engagements with a specific auditor is fixed, backward induction leads to equilibrium in which the auditor does not endorse biased reports. By comparison, if the number of consecutive engagements is indefinite (i.e., auditor rotation is not required), backward reduction cannot be used to derive equilibrium behavior. In this case, the forward reductions and trembling hand refinement lead to equilibrium in which auditors endorse a biased report if gains from future engagements exceed the incremental expected liability (B>D). Magee and Tseng (1990) find that the auditor does not compromise independence unless specific conditions are satisfied. In their model, the firm has a definite life. As a result, the maximum number of consecutive engagements cannot exceed the life of the firm (N periods). When the firm’s life is not definite and auditor rotation is not required, [...]... 2002 Audit-firm tenure and the quality of financial reports Contemporary Accounting Research 19: 63 7-6 60 Mansi, S .A. , W.F Maxwell, and D.P Miller 2004 Does auditor quality and tenure matter to investors? Evidence from the bond market Journal of Accounting Research 42: 75 5-7 93 Mautz, R.K., and H .A Sharaf 1961 The Philosophy of Auditing Sarasota, FL: American Accounting Association Magee; R P and M-C Tseng,... Auditing: A Journal of Practice and Theory 23: 5 5-6 9 DeAngelo, L 1981 Auditor independence, ‘low balling,’ and disclosure regulation Journal of Accounting and Economics 3: 11 3-1 27 Dopuch, N., R.R King and R Schwartz 2001 An experimental investigation of retention and rotation requirements Journal of Accounting Research 39: 9 3-1 17 Geiger, M., and K Raghunandan 2002 Auditor tenure and audit reporting failures... analysis indicates that it is not appropriate to interpret higher failure rates in the early years of the auditor- firm relationship as evidence that a system of no rotation is better than one of mandatory rotation In fact, the empirical evidence may well indicate that a system of auditor rotation is more beneficial if the rotation period is sufficiently large V Manager Incentives and Auditor Independence... 1990 Audit Pricing and Independence, the Accounting Review, April pp 31 5-3 36 Myers, J.N , L .A Myers, and T.C Omer 2003 Exploring the term of the auditor- client relationship and the quality of earnings: A case for mandatory auditor rotation? The Accounting Review 78: 77 9-7 99 29 Palmrose, Z.-V 1986 Litigation and independent auditors: The role of business failures and management fraud Auditing: A Journal... mandatory rotation is lower than that with no rotation In other words, mandatory auditor rotation is beneficial if rotation is not required too frequently Corollary 1: The threshold number of consecutive engagements that produces indifference between a system of mandatory rotation and one of no rotation is an increasing function of startup costs and a decreasing function of the size and frequency of the... auditor rotation Is mandatory auditor rotation beneficial? Based on the earlier discussion, auditor independence is preserved with a system of mandatory auditor rotation (i.e., N is definite) But the benefits of such a system relative to one in which rotation is not required are not entirely one sided The benefits of mandatory rotation as compared to no rotation are a function of the rotation requirements... higher reporting bias and, in turn, a higher cost of audit failure (i.e., higher expected cost of auditor liability) Mandatory auditor rotation can reduce the high cost of audit failure Furthermore, the benefit of a system of mandatory rotation, over one of no rotation, increases as startup costs increase given that the rotation period exceeds the threshold number of periods IV Auditor Experience with... rotation systems is as follows Tnr- Tr= ( C+ Lx +αD)/( 1- γ) +A+ M-[( C+ Lx)/( 1- γ)+ (A+ M)/( 1-( δγ)N)] =αD/( 1- γ) .- (A+ M)(δγ)N/( 1-( δγ)N) 23 For given A, M, α, D, δ and γ, larger values of N result in a larger difference TnrTr But as N decreases, the benefits of mandatory rotation decrease Ultimately, the benefits may disappear Proof of Corollary 1 Assume that Nt is the maximum number of consecutive engagements... auditor rotation is one sided and inappropriate If the rotation period is small, high startup costs can result in greater costs under a system of mandatory rotation than no rotation However, if the rotation period is sufficiently large, high startup costs can result 11 in greater benefit from mandatory rotation In the absence of auditor rotation, high startup costs lead to higher quasi rents, which lead to... hand, is higher under a system of no rotation because the auditor may endorse a biased report The net effect, as to which system is beneficial, depends critically on the rotation period (N) If N is small, the total cost under a system of mandatory rotation could be greater than that with no rotation because startup costs are high But if N is sufficiently large, the total cost under a system of mandatory . for audit quality and/ or the credibility of financial data. Geiger and Raghunandan (2002) find a negative association between auditor tenure and auditor- reporting failure. 2 Mansi, Maxwell, and. time, and the manager is myopic (focused on short-term payoffs). A Model of Mandatory Auditor Rotation I. Introduction The need for mandatory auditor rotation has long been debated. A primary. A Model of Mandatory Auditor Rotation Bryan K. Church College of Management Georgia Tech Atlanta, GA 3033 2-0 520 404.894.3907 bryan .church@ mgt.gatech.edu Ping Zhang Rotman