ABSTRACT We estimate linguistic‐based classification models of deceptive discussions during quarterly earnings conference calls. Using data on subsequent financial restatements and a set of criteria to identify severity of accounting problems, we label each call as “truthful” or “deceptive.” Predict...