Does Auditing Deter Potentially Fraudulent Earnings Management? An Experimental Investigation

Abstract

Effective audits are claimed to not only enhance the detection of fraud but also the deterrence of fraud in the first place. To focus on whether auditing has a deterrence effect separate from the detection effect, we design an experiment that isolates the deterrence effect from the detection effect. The experiment with 96 senior corporate managers features changes in audit procedures that would increase the likelihood of fraud being detected in the other divisions of the company but not in the focal manager’s division. Indeed, audit attention is reduced in the focal manager’s division due to a planned rotation in audit procedures that results in less substantive tests of details and more substantive analytical procedures in the focal manager’s division. We find that focal managers commit less potentially fraudulent earnings management in their own division when the nature of evidence collected changes towards increased probative value in the other managers’ divisions, but not when there is an increase in the extent of evidence collected employing the same audit tests. We also find that changing the nature of audit evidence collected results in more of a reduction in potentially fraudulent earnings management than the increased extent of audit evidence does, due to the auditor being perceived as more unpredictable under evidence nature changes than evidence extent changes. Overall our findings demonstrate that auditing has deterrence effects separate from an audit’s detection effects and that enhancements in deterrence are associated with the increased unpredictability of the audit procedures.