Audio signals reveal evasive managers during earnings calls
Lie to me: Detecting Managerial Evasiveness in Earnings Calls via Conversational Audio Encoders
Machine LearningComputational Engineering, Finance, and Science
Summary
When company leaders talk to analysts on earnings calls, sometimes they try to avoid answering questions directly. The authors found that by combining what is said with how it is said—using both text and audio analysis—they can better spot when managers are being evasive. This helps predict when companies might face serious regulatory problems, like late financial filings. Their method works better than just looking at the words or call length alone.
What this means in practice
- •For financial compliance teams: Detect early signs of regulatory risks from earnings calls by analyzing combined audio and text cues for evasiveness indicators.
- •For financial analytics providers: Enhance risk scoring products by integrating conversational audio analysis to better flag companies with potential reporting issues.$Commercial implications: Enables creation of advanced financial risk detection services sold to investors and compliance clients using novel audio-text fusion.
Authors
Huizhong Chen, Huan Zhang
Abstract
Earnings conference calls are a primary channel through which managers disclose information under analyst scrutiny. Prior work has linked vocal and lexical cues to future adverse outcomes, but often pools features over an entire call and underuses the interactive structure of Q&A. We propose a two-branch late-fusion framework for detecting managerial evasiveness as a predictor of extrinsic SEC events (primarily late filings): (i) an LLM-as-a-judge that maps Q&A text to an interpretable call-level vector X_text via a structured binary rubric, and (ii) a frozen conversational encoder whose temporal hidden states are read by a DeepVoice-style sequential reader to produce an audio representation h. Late fusion of (X_text, h) yields a call-level risk score p. On n=1,039 calls (212 late filings) with firm-grouped 5-fold CV, fusion reaches AUROC approx. 0.89, versus 0.55 for the text judge and 0.71 for duration alone. These results show that conversational audio dynamics encode managerial evasiveness beyond lexical content and call length, yielding a stronger early-warning signal of adverse SEC outcomes.