Improving Experienced Auditors’ Detection of Deception in CEO Narratives

Abstract

We experimentally study the deception detection capabilities of experienced auditors, using CEO narratives from earnings conference calls as case materials. We randomly assign narratives of fraud and non-fraud companies to auditors as well as the presence versus absence of an instruction explaining that cognitive dissonance in speech is helpful for detecting deception. We predict this instruction will weaken auditors’ learned tendency to overlook fraud cues. We find that auditors’ deception judgments are less accurate for fraud companies than for non-fraud companies, unless they receive this instruction. We also find that instructed auditors more extensively describe red flags for fraud companies and more accurately identify specific sentences in narratives that pertain to underlying frauds. These findings indicate that instructing experienced auditors to be alert for cognitive dissonance in CEO narratives can activate deception detection capabilities.

Keywords: deception detection; earnings conference call; fraud; audit

JEL Classification: M41; M42; M48