British Accounting Review (1999) 31, 151–183 Article No. bare.1999.0096, available online at http://www.idealibrary.com on AUDITOR’S SKILL, AUDITING STANDARDS, LITIGATION, AND AUDIT QUALITY PASCAL FRANTZ London School of Economics and Political Science This paper introduces a model of the market for audit services in which auditors differ in their levels of skill, which may or may not be observable and capture differences in ability. The model captures the interplay amongst auditing standards, litigation, and auditors’ levels of skill, which determines auditors’ responses to auditing standards. The paper shows that the quality of audit supplied by any auditor is increasing in the auditor’s level of skill regardless of whether or not auditors’ levels of skill are observable. An increase in the quality of audit prescribed by auditing standards is shown to induce some auditors endowed with low levels of skill to decrease the quality of their audits so that the average quality of audit and economic welfare may actually decline as auditing standards are raised. Auditors’ choices of audit quality are furthermore shown to be increasing in trial awards. Incentives for trials and out-of-court settlements are shown to depend crucially on whether or not auditors’ levels of skill are observable. Only when auditors’ levels of skill are unobservable do trials obtain with some probability. When auditors’ levels of skill are unobservable, the introduction of either restrictions on costs awarded by the courts or an imperfection in the courts’ technology is shown to lead the most skilled auditors to supply audits of a quality strictly exceeding the quality prescribed by the prevailing auditing standards. When the courts err often enough, the most skilled auditors having exercised due care furthermore make offers to settle when sued.  1999 Academic Press INTRODUCTION The last decades have seen a dramatic increase in litigation against auditors. Although the litigation crisis is most profound 1 in the USA, it has not spared the UK, where Price Waterhouse and Ernst & Young face a multi-billion This paper is based on Discussion Paper No. 235, Financial Markets Group, London School of Economics and Political Science. The author would like to thank Sudipto Bhattacharya, Michael Bromwich, Ronald Dye, Christopher Napier, Bharat Sarath, Marleen Willekens, and other participants at various workshops, the 1996 Annual Congress of the European Accounting Association, and the 1996 International Symposium on Audit Research. Please address all correspondence to: Pascal Frantz, Department of Accounting and Finance, London School of Economics and Political Science, Houghton Street, London WC2A 2AE. Received February 1998; revised September 1998; accepted November 1998. 0890–8389/99/020151+33 $30.00/0  1999 Academic Press 152 P. FRANTZ dollar writ in connection with audits of the collapsed Bank of Credit and Commerce International. This paper introduces a model of the market for auditing services designed to capture the interplay amongst auditing standards, litigation, and auditors’ levels of skill, which determines auditors’ responses to auditing standards and other legal parameters. The model is a vehicle for asking the following questions: How is the choice of audit quality related to an auditor’s level of skill? Why do some auditors settle out-of-court while others go to trial? What effect does the increasing number of auditing standards have on both the equilibrium schedule of audit quality chosen by auditors and the resulting frequency of trial? Do limits on trial awards reduce the average quality of audit? How do restrictions on litigation costs awarded by the courts or a move from the English to the American rule on litigation costs affect the equilibrium schedule of audit quality and the resulting frequency of trial? How does an imperfection in the courts’ technology affect the average quality of audit supplied or the frequency of trial? The accounting literature suggests that audit firms differ in the quality of the audits supplied (Palmrose, 1988) 2 . Dye (1993) explains systematic differences in the quality of audit supplied by various auditors through differences in auditors’ levels of wealth. Nelson et al. (1988) show how the quality of an audit is related to the quality of the client’s internal control system. This paper explains differences in audit quality through differences in levels of skill, which may or may not be observable and capture differences in ability. Any auditor, whatever his level of skill, can supply audits of any quality. The direct cost of an audit of any quality is however strictly decreasing in the auditor’s level of skill. Differences in auditors’ levels of skill may be explained by differences in knowledge 3 or audit technology 4 . This paper also differs from the above papers by explicitly modelling the litigation phase. Related papers in the legal literature include Hylton (1993), Hay (1995), and Hughes & Snyder (1995). As in the contemporaneous literature, the quality of an audit is assumed not to be observable either at the time the audit is purchased or performed, incentives to supply high quality audits are provided by the threat of litigation (Moore & Scott, 1988; Sarath & Wolfson, 1988; Balachandran & Nagarajan, 1991; Narayanan, 1994; Thoman, 1996; Schwartz, 1997). The model captures the following features of litigation against auditors in the UK. Auditors owe a duty of care to the company 5 . Litigation against auditors is initiated by insolvency practitioners such as receivers and liquidators in financially distressed firms 6 . Auditors are liable under a rule of joint and several liability. Courts’ decisions are based on a rule of negligence. The relevant law on litigation costs is defined by the English rule 7 . The paper considers exogenous differences in levels of skill amongst auditors and examines the dependence of the quality of audit supplied by an auditor and the subsequent litigation choices made by this auditor if sued on his level of skill, trial awards, litigation costs, and auditing standards. Regardless of whether or not auditors’ levels of skill are observable, the AUDITOR’S SKILL, AUDITING STANDARDS, LITIGATION AND AUDIT QUALITY 153 quality of audit supplied by any auditor is shown to be increasing in the auditor’s level of skill. An increase in the quality of audit prescribed by auditing standards is shown to induce some auditors endowed with low levels of skill to decrease the quality of their audits so that the average quality of audit and economic welfare may actually decline as auditing standards are raised. Auditors’ choices of audit quality are furthermore shown to be increasing in trial awards. Incentives for trials and out-of-court settlements are shown to depend crucially on whether or not auditors’ levels of skill are observable. Only when auditors’ levels of skill are unobservable do trials obtain with some probability. When auditors’ levels of skill are unobservable, the introduction of either restrictions on costs awarded by the courts or an imperfection in the courts’ technology is shown to lead the most skilled auditors to supply audits of a quality strictly exceeding the quality prescribed by the prevailing auditing standards. When the courts err often enough, the most skilled auditors having exercised due care furthermore make offers to settle if sued. The paper proceeds as follows. Section 2 introduces the basic model and derives an auditor’s equilibrium litigation and audit quality choices as a function of his level of skill, trial awards, litigation costs, and auditing standards. Section 3 analyses in typical legal environments the effect of changes in the quality of audit prescribed by auditing standards on the quality of audit adopted by auditors, economic welfare, and the resulting frequency of trial. Section 4 analyses the effect of changes in legal parameters on the quality of audit adopted by auditors and the resulting frequency of trial. Section 5 considers the effect of restrictions on costs awarded by the courts. Section 6 introduces an imperfection in the courts’ technology. Section 7 presents a brief summary of the paper. AUDITOR’S SKILL AND AUDIT QUALITY This section first introduces a model of the market for audit services in which auditors differ in their levels of skill and audits may or may not be discretionary. It then analyses an auditor’s litigation and audit quality choices as a function of his level of skill, trial awards, litigation costs, and auditing standards. The basic model Demand for audit services is generated by the sale of firms by one generation to another generation of risk-neutral 8 investors for life-cycle reasons. Each firm sold may ultimately turn out to be either a success (denoted by S), generating cash-flow X S , or a failure (denoted by F), generating cash-flow X F ,with0<X F <X L <X S and X L representing the cash-flow that can be obtained from an early liquidation of the firm. If investors could distinguish between successful and failing firms, they would thus liquidate 9 the latter 154 P. FRANTZ firms. Investors however cannot 10 distinguish between successful and failing firms. Only the fraction of successful firms 11 in the population, p20, 1, is known. Audit technology. Auditors own a technology that aids in distinguishing between successful and failing firms. The audit process may be summarized as follows. Every firm that hires an auditor requests him to attest to the report that it is ultimately successful. In his duties, the auditor acquires additional information about the audited firm’s type. Based on this information, the auditor either agrees to attest to the firm’s original favourable report and issues an unqualified opinion (denoted by ‘S’) or, instead, disagrees with the proposed report and issues a going concern qualification 12 (denoted by ‘F’). The relation between the audit report, the audited firm’s actual condition, and the quality q of the audit is summarized by the following conditional probabilities:  Pr‘F’jF, q Dq Pr‘S’jS, q D1 1 That is, if the auditor adopts an audit of quality q, the likelihood that he correctly issues a qualified report, given that the audited firm ultimately fails, is q. Conversely, if the audited firm is ultimately successful, the auditor is assumed not to obtain any information indicating that the firm might fail and thus always issues an unqualified opinion for such firms 13,14 . Auditor’s level of skill and audit cost. Each auditor is a utility-maximizing, risk-neutral agent, who has the ability to audit at most one firm in a period. He strategically chooses how hard to work, that is his level of effort e,and thus the quality q of his audit, which is strictly increasing in the level of effort supplied. Both the level of effort e and the quality of audit q are assumed not to be observable either at the time the audit is purchased or performed. Auditors differ in a factor n, which may or may not be publicly observable and will be referred to as skill. Any auditor, whatever his level of skill, can supply audits of any quality in the interval  q , 1  ,whereq is a lower bound on audit quality. The direct cost of an audit of quality q supplied by an auditor of skill n, Cq, n, is however strictly increasing in q on  q , 1  , C q >0, and strictly decreasing in n, C n <0. The marginal cost of producing an audit of quality q strictly increases in q , C qq >0, and strictly decreases in n, C qn <0 15 . Cq, n furthermore satisfies the following boundary conditions: 8n, Cq , nD0, C q q, nD0, and lim q!1 Cq, nDC1. The distribution of auditors’ skill has a density wn with support [ 0, n max ] . Its associated distribution function is denoted by W n. Auditor’s liability. Unless disciplined, an auditor would always choose the lowest level of audit quality q . The incentive to supply higher quality audits is provided by the threat of litigation: conditional on a firm’s failure, the firm’s auditor may be sued on behalf of the ‘company’ by the firm’s AUDITOR’S SKILL, AUDITING STANDARDS, LITIGATION AND AUDIT QUALITY 155 insolvency practitioner. The relevant law is defined by a rule of negligence and the courts own a technology allowing them to discover the defendant’s behaviour without error 16 . There is some level of audit quality q Ł prescribed by the prevailing auditing standards such that, conditional on the failure of the firm and issuance of an unqualified opinion by the firm’s auditor, the auditor is only liable if the quality q of his audit is strictly lower than q Ł17,18 . The relevant law on litigation costs is defined by the English rule: the loser at trial pays the litigation costs of both parties. Suits need not end up in trials: the defendant, that is, the auditor, can make a settlement offer and the plaintiff, that is, the insolvency practitioner, can accept the offer or go to trial. More formally, the litigation process is as follows. The plaintiff, who believes that the defendant is negligent with some probability p, must decide whether or not to file suit on behalf of the company. The defendant, who knows whether or not he is negligent, then responds with a settlement offer (a settlement offer of zero is equivalent to no settlement offer). The plaintiff finally chooses between accepting the settlement offer and going to trial (acceptance of a zero settlement offer is equivalent to dropping the action). Both parties, who are assumed to be risk-neutral, take their decisions knowing that, if the defendant is found negligent by the courts, he must pay trial awards A to the plaintiff in addition to the plaintiff’s litigation costs P and his own litigation costs D 19 . Neither party incurs any costs 20 if the plaintiff drops the case before trial or if the case is settled. Demand for audit services. Audits may be discretionary or mandatory. As in Dye (1995), the audit market operates as follows. Each potential auditor chooses which fee to charge. Firms’ owners observe the posted audit fees (and levels of skill when observable) and conjecture about both the quality of audit supplied by each potential auditor and the expected value derived from hiring each potential auditor. The owners of each firm then either hire an auditor from amongst those offering the highest audit value or hire none, depending on which decision makes them better off (the latter alternative being only available when audits are discretionary). The benefit derived by a firm’s owners from an audit gross of the audit fee that is, the gross value of an audit, lies in the rise in the firm’s expected selling price. An audit increases the firm’s expected selling price for the following reasons. First, it assists in identifying and liquidating firms whose liquidating values exceed their going-concern values (information value of an audit). Second, it provides gains from litigation if the firm subsequently fails, the auditor did not issue a going concern qualification, and the audit was negligent (litigation value of an audit). Equilibrium audit fees are determined as follows. In any market equilibrium, as all firms are ex-ante identical, the net value of an audit, that is, the gross value of an audit less the audit fee, must be the same across all firms that hire auditors and across all potential auditors seeking to be employed. 156 P. FRANTZ Strategic analysis This subsection analyses the schedule of audit quality supplied by auditors and litigation choices made by insolvency practitioners and auditors. The analysis is based on the concept of a perfect Bayesian equilibrium (PBE). A set of strategies and beliefs constitutes a PBE if it is sequentially rational, that is, if at any stage of the game, the strategies are optimal given beliefs and beliefs are obtained from equilibrium strategies and observed actions using Bayes’ rule. In addition, we impose the condition that weakly dominated strategies be excluded 21 . Equilibrium litigation choices. The equilibrium litigation choices made by auditors and insolvency practitioners depend on whether or not auditors’ levels of skill are observable. When auditors’ levels of skill are unobservable, there are two possible sequences of equilibrium litigation strategies: no suit and semi-separating suit equilibrium strategies. If the insolvency practitioners’ expected return from trial is non-positive, that is, pA1pP C DÄ 0, the insolvency practitioners do not sue and the no suit sequence of equilibrium litigation strategies obtains. If the insolvency practitioners’ expected return from trial is strictly positive, that is, pA1pP CD>0, the semi-separating suit sequence of equilibrium litigation strategies described in Lemma 1 obtains. L EMMA 1: Assuming that auditors’ levels of skill are unobservable and that insolvency practitioners’ expected return from trial is strictly positive, that is, A1PCD>0, the following sequence of strategies are equilibrium strategies 22 : Insolvency practitioners file suit; Negligent auditors make offers of A with probability A1PCD A , and make no offer to settle with the complementary probability; Careful auditors make no offer to settle; Insolvency practitioners accept offers to settle of A, but in response to no offer to settle, drop their actions with probability PCD ACPCD , and go to trial with the complementary probability. P ROOF: A proof of Lemma 1 can be found in the Appendix. In the semi-separating suit sequence of equilibrium strategies, a pro- portion of negligent auditors thus reveal their types through offers to settle. The complementary proportion of negligent auditors mimic the behaviour of the careful auditors by making no settlement offer. The information content of a refusal to settle is thus clouded. Some insolvency practitioners respond by bringing their cases to trial while others drop their actions. C OROLLARY 1: Conditional on the failure of a firm and the firm’s auditor having filed an unqualified report, the probabilities of the various litigation outcomes are: AUDITOR’S SKILL, AUDITING STANDARDS, LITIGATION AND AUDIT QUALITY 157 1 for trial;   1 1  PCD A  for settlement; 1 PCD A for the action being dropped. C OROLLARY 2: Conditional on the failure of a firm and the firm’s auditor having filed an unqualified report, the expected payoffs to the various parties arising from litigation are as follows: AgainofA1PCD for insolvency practitioners; A loss of A for negligent auditors; No gain or loss for careful auditors. When auditors’ levels of skill are observable, there are two possible sequences of equilibrium litigation strategies: no suit and separating suit equilibrium strategies. If no auditor is negligent, insolvency practitioners never sue and the no suit sequence of equilibrium litigation strategies obtains. Conversely, if some auditors are negligent, the separating suit sequence of equilibrium litigation strategies described in Lemma 2 obtains. L EMMA 2: Assuming that auditors’ levels of skill are observable and that some auditors are negligent, the following sequence of strategies are equilibrium litigation strategies 23 : Insolvency practitioners file suit; Negligent auditors make offers of A; Careful auditors make no offers to settle; Insolvency practitioners accept offers to settle of A, and in response to no offer to settle, drop their actions. P ROOF: In the event of failure, the insolvency practitioner infers from the equilibrium schedule of audit quality qn and from knowledge of the auditor level of skill n the quality of audit supplied. There is no asymmetry of information between the auditor and the insolvency practitioner and no trial ever takes place. Litigation outcomes depend on the level of care taken by the auditor in only two sequences of equilibrium litigation strategies: the semi-separating and separating suit sequences of equilibrium strategies. In the no suit sequence of equilibrium strategies, no suits are filed and there are thus no incentives for auditors to perform audits whose quality exceeds the lower bound on audit quality q . The lack of filing of any suits can thus only arise in equilibrium if the quality level q Ł prescribed by the prevailing auditing standards does not exceed the lower bound on audit quality q , which is not very interesting. As equilibrium strategies arising in the litigation phase, we shall therefore adopt those described in Lemma 1 when auditors’ levels of skill are unobservable and those described in Lemma 2 when auditors’ levels of skill are observable. Whether or not auditors’ levels of skill are observable, 158 P. FRANTZ there is thus no gain or loss in the litigation stage for careful auditors and the expected loss of a negligent auditor is equal to trial awards A. Equilibrium choices of audit quality. As the quality of an audit is not publicly observable when the auditor’s report is issued, the equilibrium choice of audit quality made by an auditor is the one minimizing the total cost of the audit, with the total cost of an audit being the sum of the direct cost Cq, n and the expected loss arising from any litigation by the insolvency practitioner working on behalf of the auditee in the event of the auditee’s failure. An auditor can only be sued if the audited firm fails and the auditor did not issue a going concern qualification. Conditional on performing an audit of quality q, an auditor’s expected cost arising from litigation can thus be written as lqdq, q Ł A,wherelqÁ1p1q represents the auditor’s joint probability of auditing a failing firm and issuing an unqualified report and dq, q Ł ,withdq, q Ł Á1ifq<q Ł and dq, q Ł Á0ifq½q Ł , is an indicator of auditor’s negligence (dq, q Ł D1 indicates that the auditor supplied a negligent audit while dq, q Ł D0 indicates that the auditor exercised due care). An auditor of skill n thus chooses the level of audit quality (and hence the level of effort) minimizing the function f Cq, nClqdq, q Ł A g over  q , 1  . As shown in Figure 1, this function presents a discontinuity: it follows the function fCq, nClqAg up to the quality qDq Ł , whereupon it follows the function fCq, ng. Figure 1. Total expected cost of an audit. AUDITOR’S SKILL, AUDITING STANDARDS, LITIGATION AND AUDIT QUALITY 159 This analysis confines attention to ‘typical legal environments’, with typical legal environments satisfying the following definitions: D EFINITION 1: A legal environment is defined as the set f q Ł , A, P, D g . D EFINITION 2: A legal environment is said to be typical if the following conditions are satisfied: At least one employed auditor elects to be negligent; At least one employed auditor elects to exercise due care. Restricting the analysis to legal environments in which at least one employed auditor elects to supply a negligent audit does not entail much loss of generality. If all employed auditors elect to exercise due care and the quality level q Ł prescribed by the prevailing auditing standards strictly exceeds the lower bound on audit quality q , no equilibrium can obtain 24 .The intuition is as follows. If all employed auditors exercise due care, there is no incentive for insolvency practitioners to file suit. But, given that insolvency practitioners do not file any suits, all auditors supply negligent audits. Audit cost functions corresponding to different levels of skill n in a typical legal environment are shown in Figure 2. L EMMA 3: The level of effort e supplied by an auditor of skill n if employed, en, and hence the quality q of the audit supplied, qn, are increasing in n. Figure 2. Total expected cost for auditors with various levels of skill. 160 P. FRANTZ PROOF: A proof of Lemma 3 can be found in the Appendix. The intuition is as follows. Any auditor, whatever his level of skill, can supply audits of any quality. The direct cost of an audit of any quality is however strictly decreasing in the auditor’s level of skill. The level of effort supplied by an auditor if employed, and hence the quality of audit supplied, are thus increasing in the auditor’s level of skill. From Figure 2, it can be inferred that, for any level of audit quality q Ł prescribed by the prevailing auditing standards, there exists a critical level of skill n B such that an auditor of this level of skill would be indifferent between being negligent and exactly meeting the level of quality prescribed by the prevailing auditing standards. More formally, this condition may be rewritten as: min q fCq, n B ClqAgDCq Ł , n B 2 The solution to the left-hand side of this equality may be characterized by the first order condition: C q q, n B Cl q qAD0 3 P ROPOSITION 1: Given a typical legal environment, there is a strictly positive constant n B such that: 1. Employed auditors with levels of skill n,with0Än<n B , elect to supply negligent audits of quality q minimizing fCq, nCqAg; 2. Employed auditors with higher levels of skill n,withn B ÄnÄn max ,electto supply audits of quality q Ł . P ROOF: A proof of Proposition 1 can be found in the Appendix. As illustrated in Figure 3, auditors with lower levels of skill thus choose to supply negligent audits. Auditors with higher levels of skill elect to provide audits of the quality prescribed by the prevailing auditing standards. The equilibrium schedule of audit quality qn is furthermore discontinuous at n B . P ROPOSITION 2: The expected cost of an audit supplied by an auditor of skill n strictly decreases in n over [ 0, n max ] . When auditors’ levels of skill are unobservable, there thus exists  n B ½0 such that the set of levels of skill employed auditors are endowed with is the interval   n B , n max  . P ROOF: A proof of Proposition 2 can be found in the Appendix. When auditors’ levels of skill are unobservable, the set of employed auditors thus consists of the most skilled auditors. The intuition is as follows. When auditors’ levels of skill are unobservable, the fee related to [...]... an increase in the audit quality prescribed by auditing standards on the quality of audit supplied by various types of auditors, if employed, is illustrated in Figure 4 COROLLARY 4: When auditors’ levels of skill are unobservable27 , an increase in the quality of audit prescribed by auditing standards leads to a decrease in AUDITOR’S SKILL, AUDITING STANDARDS, LITIGATION AND AUDIT QUALITY 163 Figure... in their levels of skill, which may or may not be observable, and examines the dependence of the quality of audit supplied by an AUDITOR’S SKILL, AUDITING STANDARDS, LITIGATION AND AUDIT QUALITY 175 auditor and the subsequent litigation choices made by this auditor if sued on his level of skill, trial awards, litigation costs, and auditing standards Regardless of whether or not auditors’ levels of... increase in the average quality of audit supplied The effect of an increase in the quality of audit prescribed by auditing standards on the quality of audit supplied by various types of auditors, if employed, is summarized in Proposition 3: PROPOSITION 3: For any move from auditing standards prescribing 0audit quality qŁ to auditing standards prescribing a strictly higher audit quality qŁ , there is... prescribed by auditing standards causes some auditors endowed with lower levels of skill who would otherwise have exercised due care to supply negligent audits and auditors endowed with higher levels of skill to raise the quality of their audits to implement the new standards As qBest n is strictly increasing in n, the auditors endowed with AUDITOR’S SKILL, AUDITING STANDARDS, LITIGATION AND AUDIT QUALITY. .. awarded by the courts leads to a decrease in the quality of audit supplied by employed auditors with low levels of skill and an increase in the quality Figure 6 Effect of restrictions on costs awarded on audit quality AUDITOR’S SKILL, AUDITING STANDARDS, LITIGATION AND AUDIT QUALITY 171 of audit supplied by auditors with high levels of skill When audits are mandatory, the introduction of restrictions on... independent of the quality of audit prescribed by the auditing standards If the quality of audit prescribed by the auditing standards were 0 to increase to qŁ , this auditor would thus strictly prefer to supply the 0 negligent audit of quality qNeg nB All auditors with levels of skill nB . costs, and auditing standards. Regardless of whether or not auditors’ levels of skill are observable, the AUDITOR’S SKILL, AUDITING STANDARDS, LITIGATION AND AUDIT QUALITY 153 quality of audit. amongst auditing standards, litigation, and auditors’ levels of skill, which determines auditors’ responses to auditing standards. The paper shows that the quality of audit supplied by any auditor. the quality of audit prescribed by auditing standards leads to a decrease in AUDITOR’S SKILL, AUDITING STANDARDS, LITIGATION AND AUDIT QUALITY 163 Figure 4. Effect of an increase in the quality