Accounting Accruals, Audit Quality, and Audit Pricing

Abstract

AbstractThis study examines the impact of the interaction between firms' accounting choices and auditors' audit effort choices on audit pricing. In a multi-period model, I focus on a central feature of accounting: reversing accruals. I find that initial low-balling, defined as lower audit fees than audit costs, arises independently from the direction of accruals, but increases with more conservatism in the first engagement period. The reason is that a conservative accruals strategy implies understated earnings today and overstated earnings in the future, which creates a higher investor demand for auditing services in future periods. The incumbent auditor can satisfy this future demand better than any competitor, because the first-period learning cost is sunk. Thus, conservatism enhances the competitive advantage of the incumbent and increases the future quasi-rent, which is used for a higher initial low-balling discount. I further show implications of the auditor-firm interaction for fee-cutting, a common proxy for low-balling, and the frequency of modified audit opinions, a common proxy for audit quality. Thus, my model provides a theoretical foundation for the connection between high low-balling discounts, accounting accruals, and audit quality.Keywords: Low-ballingFee-cuttingAudit qualityAccounting accruals AcknowledgmentsAccepted by Robert F. Göx. I thank Christopher Bleibtreu, Ralf Ewert, Qi Gao, Yasmin Hoffmann, Bill Kinney, Thomas Kourouxous, Sandra K. Kronenberger, Peter Krenn, Volker Laux, Paul Newman, Elisabeth Plietzsch, Stefan Schantl, Thomas Simon, Dirk Simons, Alfred Wagenhofer, Anna Waldner, and participants of the 12th Workshop on Accounting and Economics in Tilburg, the 39th EAA Annual Congress in Maastricht, the 83th VHB Annual Congress, the 12th EARNet Symposium Komma and seminar participants at the University of Magdeburg for helpful comments. I acknowledge funding from the Austrian Science Fond (FWF), Project W 1229.Disclosure statementNo potential conflict of interest was reported by the author(s).Notes1 Over the total time period, the auditor never earns rent because it is used to attract the engagement in the first period. Therefore, it is called quasi-rent.2 See also Kanodia and Mukherji (Citation1994), Lee and Gu (Citation1998), Chan (Citation1999), Zhang (Citation1999), and Simons (Citation2011).3 See also Göx and Wagenhofer (Citation2009, Citation2010), Nan and Wen (Citation2014), and Caskey and Laux (Citation2017).4 Further studies include Gul et al. (Citation2009), Ghosh and Lustgarten (Citation2006) and Ettredge and Greenberg (Citation1990).5 In addition, Barua et al. (Citation2020) provide evidence that audit fee studies are often subject to a systematic bias because only the succeeding auditor's fee must be mandatorily disclosed. The study shows that the fee discount often disappears after adjusting for the bias.6 To economize on notation, I drop the time index on the economic state, the accounting signal, and the audit opinion. Thus, θt,h=θh and θt,l=θl, St,h=Sh and St,l=Sl, Rt,h=Rh and Rt,l=Rl.7 The owner-manager already designed the distributions in the most favorable manner and they cannot be improved further.8 As I discuss in Section 5, as long as the lenders and auditors rationally anticipate the reporting incentives, and there is a variation between periods, the results of the paper hold. This includes settings, in which the owner-manager chooses accruals in each period.9 Figure 2 uses symmetric distributions along c = 0.5 for the interval Et∈[0,1], fθh(Et)=1−(1−Et)2 and fθl(Et)=1−Et2. A neutral accounting system aligns economic and accounting income. In an asymmetric distribution, where probability mass of fθh(Et) shifts to the left but still fulfills all assumptions stated above, a neutral accounting system would imply overstated accounting income relative to the economic income.10 This aspect of the assumption is not critical to the model results. Variable start-up cost could also decrease with more audit effort; for example, when an audit team member becomes faster every day, with more repetitions of the same task. As demonstrated in Section 3.3, the results of the study hold as long as the incumbent has a competitive advantage over the audit competitor, which is connected to the audit quality in period two.11 P1(θh|Rl)=p∫0cfθh(E1)dE1p∫0cfθh(E1)dE1+(1−p)(∫0cfθl(E1)dE1+∫c1fθl(E1)qE1dE1)andP2(θh|Rl)=p∫01−cfθh(E2)dE2p∫01−cfθh(E2)dE2+(1−p)(∫01−cfθl(E1)dE1+∫1−c1fθl(E2)qE2dE2). 12 I assume that the learning cost in the second period is sunk also with a minimum audit effort, qt=0, in period one. For example, the parties are already familiar with each other because of the fee negotiation process. The results do not depend on the extent of learning (see Section 3.3 for details).13 An implicit assumption in this section is that the incumbent auditor is rehired in period two. In Section 3.3, I show that this is optimal in equilibrium.14 The expected audit quality expressed as the frequency of modified audit opinions is discussed in Section 4.15 The firm re-hires the incumbent regardless of the occurrence of accounting mistakes. The realization of payoffs and uncovering of mistakes occur after period two. However, even when considering the realization, the incumbent still provides a higher net value to the firm because all auditors act to the best of their knowledge (they report truthfully) and differences exist only because of learning cost. Supporting these simplifications, Hennes et al. (Citation2014) find that clients retain auditors after restatements, especially when switching cost is high. In their sample, auditor turnover after a restatement is only 16.2% and the dismissals initiated by the client account for only 10.7% of all restatement cases. Thus, it is a common practice to retain the auditor, even after restatements.16 I assume that the firm remains with the incumbent auditor in case of indifference.17 Full expressions are in the Appendix, Proof of Proposition 1.18 M. DeFond and Zhang (Citation2014) provide an overview of studies that use FMAO or Going-Concern Opinions, which is the only type of modified opinion in SEC-filings, as a proxy for audit quality in Table 3. This measure is particularly useful in my study because the literature indicates that a higher FMAO captures auditor independence (M. DeFond & Zhang, Citation2014, p. 2879). A direct measure of audit quality is the conditional probability P1(Rl|Sh), which requires information on the audit process itself, for example, the audit hours.Additional informationFundingThis work was supported by Austrian Science Fund [W 1229].