Reported earnings, auditor's opinion, and compensation: theory and evidence

Abstract

We study the effect of the auditor's independence on executive compensation and executive effort allocation. Using principal-agent theory, we examine a compensation contract involving two signals, one for incentives and the other for control. The incentive signal is the earnings reported by the executive and the control signal is the auditor's opinion. The optimal weights on earnings and audit opinion in the agent's compensation contract are obtained in a LEN (linear compensation plan, exponential utility, normally distributed outcome) framework. The pay-performance sensitivity (incentive weight on earnings) increases monotonically as the auditor becomes more independent. However, the pay-opinion sensitivity (incentive weight on audit opinion) first increases and then decreases as the auditor becomes more independent.

We test some of these results empirically with publicly available data and find that the executive is rewarded for higher reported earnings and penalised for audit qualification. Evidence also shows that the pay-performance sensitivity increases as the auditor becomes more independent.

Keywords:

• auditor's opinion
• earnings management
• LEN framework
• principal-agent theory

Acknowledgement

The authors would like to thank two anonymous reviewers and the editor for their helpful comments. This paper is based on Atasi Basu's dissertation research. She is grateful to Professors Amiya K. Basu, Ravi Dharwadkar, Susan Gensemer, Badr Ismail, Gerald Lobo, Jan Ondrich, and Alex Thevaranjan for their help. She would also like to thank the NERAAA 2005 and the AAA 2006 conference participants, and the Ph.D. students of the Lubin School of Accounting, Whitman School of Management at Syracuse University.

Notes

Linear compensation plan (the agent's compensation is a linear function of outcomes), exponential utility (the agent has a negative exponential, that is, constant absolute risk aversion utility function), and the outcomes (e.g., reported earnings) are normally distributed.

We thank an anonymous reviewer for pointing this out. A discussion of implicit effect of audit signal on compensation is included in the appendix.

DeAngelo (1981, p. 117, footnote 8) mentions that ‘It is important to note that perfect independence is a Nirvana-type construct useful only as a benchmark'. Watts and Zimmerman (1986, p. 315, footnote 3) say ‘Note that we do not expect auditors to be totally independent (i.e., report discovered breaches with probability one)'.

This assumption can be linked to DeAngelo (1981, p. 116). She writes ‘If the capital market expected the auditor never to deviate from management's position, then it would assess the value of the auditor's opinion as zero'. This situation is discussed again in Remark 3.

We thank an anonymous reviewer for pointing this out.

Data178 and data6 represent operating income after depreciation and total assets respectively in the Compustat data file.

The results for auditor opinion do not control for the reputation of the auditor; 97% of the sample companies are audited by the Big 4 auditors. However, in the analytical model, the effect of the auditor's opinion on compensation depends on the independence of the auditor, which is one dimension of auditor reputation.