Welfare-enhancing fraudulent behavior

The Statement of Auditing Standard (SAS) No. 99, Consideration of Fraud in a Financial Statement Audit, issued in November 2002, defines fraud to include intentional material misstatement of financial accounts and misappropriation of assets. The Committee of Sponsoring Organizations of the Treadway Commission (1999) notes the type of managerial manipulation that can ultimately result in fraud. The commission observes that “ many frauds allegedly were initiated in a quarterly Form 10-Q, with the first manipulation sometimes at relatively small amounts. After observing that the fraud was undetected in initial attempts, the fraud scheme was repeated in subsequently issued quarterly or annual financial staements, with the fraud amount often increasing over time.” Perols and Lougee (2011) report that firms in which fraud has been committed are more likely than other firms to have experienced managed earnings in prior years.

Of course, if the corporation could costlessly eliminate all potential for fraud, it would optimally do so.

Before it was dissolved in 2002, the Public Oversight Board (POB) served as an independent private-sector entity that oversaw the peer review program of the SEC Practice Section of the AICPA Division of Firms. The Sarbanes-Oxley Act of 2002 established the Public Company Accounting Oversight Board, a government regulatory agency, to oversee the work of public accountants.

Nelson et al. (2003) identify the most common forms of earnings management in practice. Krishnan (2003) finds that auditors with specialized industry knowledge are better able than auditors without this knowledge to mitigate accrual-based earnings management. Dyck et al. (2013, p. 4) estimate that “14.5 % of large publicly traded corporations engage in fraud.”

Subramanyam (1996) and Jiraporn et al. (2008) provide empirical support for this view.

Hillegeist (1999) analyzes the impact of three distinct damage apportionment rules on audit quality and audit failure rates. Patterson and Wright (2003) demonstrate how a proportionate liability rule can reduce audit failure rates when audit evidence about fraud is inconclusive.

Schwartz (1997)’s work is more similar to ours in that it examines the impact of different liability rules on a client’s investment decision, an auditor’s effort, and their joint welfare. In Schwartz’s model, changes in the legal liability rule do not change joint welfare.

In a related analysis, Laux and Newman (2010) investigate the effects of pronounced auditor liability on the ability of risky clients to secure auditing services.

Goldman and Slezak (2006) also analyze a setting in which increased earnings manipulation can give rise to widespread gains. The gains arise when increases in stock-based compensation induce managers to increase productive effort as well as manipulative effort.

We model the manager’s compensation as a fixed wage for expositional simplicity. The primary qualitative conclusions drawn below persist if the manager’s compensation includes payments that vary with the client’s operating profit.

\(\overline{\Pi }(M)\) might decrease with \(M\) if, for example, the manager’s productive effort (which increases the client’s profit) declines as his manipulative effort (\(M\)) increases. Alternatively, the client’s operating profit might not vary with \(M\) when no fraud arises. The key qualitative conclusions reported below hold in both cases.

The ensuing analysis can be extended to model formally competition among potential auditors for the right to conduct an audit for the client. See Lin and Sappington, “ Welfare Enhancing Fraud with Binary Auditor Effort,” working paper, University of Houston, January 2013.

The relevant range for \(M\) is the range in which probabilities \(\phi (M)\) and \(\theta (E(M))\) are well defined.

The conclusions drawn in this section do not rely on conditions (vi), (vii), and (viii) in the definition of the LQ Example. These assumptions are only employed to determine the manager’s choice of \(M\), which is considered in Sect. 5.

In practice, the auditor’s loss from undetected fraud and the manager’s loss from detected fraud might increase with \(M\). These possibilities complicate the ensuing analysis considerably. However, conclusions similar to those drawn below hold when \(L_{a}(\cdot )\) and \(L_{m}(\cdot )\) are both strictly increasing, linear functions of \(M\). See Lin and Sappington (2014) for details.

To ensure condition (C1) is not vacuous, we assume \(\lambda <\frac{1}{3}\) throughout the ensuing analysis.

The proof of Proposition 1 is presented in “Appendix 1”, as are the proofs of all other propositions in Sects. 4 and 5.

The literature provides mixed evidence regarding the impact of corporate governance and regulatory reform on financial misreporting. Some studies (e.g., Beasley 1996; Klein 2002; Farber 2005) suggest that board independence and auditor independence are instrumental in enhancing oversight and deterring such misreporting. Other studies (e.g., Cohen et al. 2008; Larcker et al. 2007) report that the regulatory reforms surrounding SOX have little impact on oversight. Our study suggests the importance of considering the impact of corporate goverance on incentives for audit effort to detect financial misreporting.

This conclusion is established in the proof of Proposition 4.

Condition (C2) and the first inequality in Condition (C3) ensure \(\theta (E(\widehat{M} ))\in ( 0,1)\). The second inequality in Condition (C3) ensures \(U_{m}(w,\widetilde{M} )>U_{m}\left( w,0\right)\) and \(\widetilde{M}>\widehat{M}\).

Thus a reduction in \(k\) and an increase in \(b\) can reflect corresponding changes within the client’s organization. (Recall the discussion in Sect. 4.)

Even in the relatively simple LQ Example, parameter changes typically do not have unambiguous effects on \(\widetilde{M}\). Proposition 5 in “Appendix 2” identifies conditions under which these effects can be signed.

The client experiences only a small percentage increase in her expected net return in this example. Notice, though, that this gain for the client does not include any cost saving she might enjoy as \(k\) declines (perhaps due to less stringent internal controls, for instance). Furthermore, the client’s net payoff increases here even though the manager’s wage does not vary as his cost of undertaking manipulative effort declines. In principle, a manager might accept a lower wage as his expected net return from engaging in fraud increases. (Denis and Xu (2013) observe that executives tend to receive lower pay in countries that implement weaker insider trading restrictions.) The wage reduction could provide an additional source of gain for the client in our setting.

The same will be true, for example, when the auditor commands most of the gains from his expanded auditing effort (so \(\lambda\) is close to \(1\)).

This conclusion is consistent with Bedard and Johnstone (2004) finding of a positive association between an auditor’s effort and billing rate and the client’s earnings manipulation risk, which is more pronounced when the client has weaker corporate governance.

An auditor also would be more inclined to deliver a complex audit if he experienced a direct reward when he uncovered fraud. In practice, the reward might reflect the benefit of an enhanced reputation. In principle, the reward could take the form of a contingent payment (e.g., Matsumura and Tucker 1992; Dye et al. 1990; De and Sen 1997).

See Schwartz (1997), Chan and Pae (1998), Hillegeist (1999), Radhakrishnan (1999), and Patterson and Wright (2003), for example.