THE EFFECTS OF AUDITOR REPUTATION IN MORAL HAZARD AND ADVERSE SELECTION SETTINGS.

摘要

The dissertation analyzes two models of auditing. The first is a multi-period model of auditing with incomplete information. An owner contracts with a manager to take actions on the owner's behalf. The outcomes of the manager's actions are unobservable to the owner. The auditor's role is to investigate and provide information to the owner about the actual outcomes. The auditor is modeled as a rational economic agent taking investigative acts and making reports under moral hazard. Issues of reputation formation and auditor independence are addressed as a means of reducing auditor moral hazard. In the second model, an informed entrepreneur wishes to issue shares in a project to uninformed investors. The auditor's role is to reduce information asymmetry by reporting to investors about the quality of the entrepreneur's project. Necessary and sufficient conditions for auditing to be useful are examined.;The second model is an adverse selection model in which an entrepreneur issues shares in a project to outside investors. The entrepreneur can perfectly signal the value of the project via the fraction of equity he retains in the project. However, such signaling is costly. Our main result is that a necessary and sufficient condition for auditing to be useful is that the audit report alter the support of the investors' probability distribution about the project's quality. We show conditions under which heterogeneous entrepreneurs choose the 'best' available auditor. In general, the choice of auditor depends on the audit cost and the 'productive effect' of audit quality as captured by the precision of the audit report.;A key result of the first model is that even though the auditor prefers to shirk, reputation effects ensure that the auditor works hard and reports truthfully over all but a few periods of the game. Creating incentives to build a reputation reduces the investigation costs the owner must incur to monitor the auditor. Reputation formation serves as a partial substitute for costly monitoring by the owner. The model is useful in understanding auditor independence. The main insight is that multi-period reputation effects preclude cooperative and collusive play between the auditor and manager over many periods of the game.