Detecting Deceptive Discussions in Earnings Calls
In this study, researchers investigate the use of linguistic markers to identify deceptive behavior during quarterly earnings conference calls. By analyzing subsequent financial restatements, the researchers label each call as "truthful" or "deceptive". They develop models based on word categories that have been shown to be related to deception and find that CEO or CFO narratives significantly outperform models based on financial and accounting variables. The researchers also identify specific linguistic markers used by deceptive executives, such as more references to general knowledge and extreme positive emotion, and fewer references to shareholder value.
The study sheds light on the use of language in financial reporting and its implications for detecting deceptive behavior. The findings suggest that linguistic markers can be useful in identifying deceptive behavior, and that models based on CEO or CFO narratives may be more effective than traditional financial metrics. The results of this research have important implications for investors and regulatory agencies seeking to detect and prevent financial fraud.
Read the research: Detecting Deceptive Discussions in Conference Calls