Corporate disclosures increasingly include multimedia content such as audio, images, and videos. In fact, as of 2022, 83 percent of U.S. public firms used graphics in annual reports and various other communications. Additionally, companies commonly provide their earnings calls, shareholder calls, M&A calls, or earnings guidance calls in audio format. However, while such multimedia disclosures aim to help investors process information, whether and how nonverbal disclosures influence investor trading is not well understood.
The challenge is twofold. First, nonverbal information is difficult to quantify. Second, the decision to incorporate multimedia into corporate communication is often endogenous and lacks standardization, which complicates cross-company comparisons. In a new paper, we use a deep-learning algorithm to assess the quality of managers’ real-time vocal delivery during earnings conference calls. Concentrating on earnings calls is advantageous as they are a common platform for multimedia disclosures and provide an opportunity to observe granular real-time market reactions.
Measuring Vocal Delivery Quality and Its Impact on the Stock Market
Our measure of vocal delivery quality evaluates how clearly someone’s speech can be comprehended by average listeners. It captures the pronunciation, fluency, and diction of spoken words. Importantly, it focuses solely on the comprehensibility of audio and differs from previously studied dimensions of vocal information such as emotions or other nonverbal communications.
We aim to evaluate whether managers’ vocal delivery quality is associated with investors’ reactions in the stock market. Existing empirical evidence suggests that disclosure-processing costs are associated with capital market outcomes. Extending this line of literature, we hypothesize that if investors incur higher information processing costs to comprehend managers who speak with low-quality vocal delivery, the contemporaneous market reactions could be subdued. In contrast, if what managers say is easy to understand, investors may incur lower processing costs, leading to accelerated real-time market reactions.
We indeed find that investors react instantaneously to the quality of managers’ vocal delivery during earnings calls. Specifically, the managers’ vocal delivery quality is positively associated with real-time abnormal trading volume and abnormal absolute returns. This result holds true across a variety of speakers and after controlling for call-level fixed effects.
This positive association is more pronounced in retail trades than institutional trades. Additionally, calls with low-quality vocal delivery are slower to influence stock prices than those with high-quality vocal delivery. In fact, we observe that the market needs a few days to fully correct the underreactions caused by low quality vocal delivery.
Our results are robust to various sensitivity checks. We introduce an alternative measure of vocal delivery quality, which uses a different deep learning algorithm, and confirm that our findings are not specific to one measurement algorithm. Also, we repeat our tests using the discussion sessions of earnings calls and find similar results. Even though our measure could be noisy during the discussion sessions due to short conversation fragments and managers being spontaneous, our main findings remain qualitatively similar.
Can Managers Adjust Their Vocal Delivery Quality?
Voice quality can change, prompting the question of whether economic incentives can motivate managers to deliberately control the quality of their vocal delivery. Prior work suggests that managers could use verbal disclosures to influence investor reactions. On the other hand, it could also be that managers’ vocal delivery quality is unconsciously affected by the content they deliver. Though it is empirically challenging to identify managerial intent, we conduct several analyses to provide some insight into this issue.
We find that vocal delivery quality worsens when a firm reports a decrease in earnings, has high earnings volatility, and presents a negative-tone transcript. We also find evidence suggesting that managers were less clear in vocal delivery when trying to obfuscate. Moreover, firms whose managers have worse vocal-delivery quality have less persistent positive earnings, suggesting economic motivations of managers at least partially explain variations in vocal-delivery quality. These results are also robust to using a different deep learning algorithm to measure vocal delivery quality.
We contribute to the literature in several ways. First, we advance the understanding of multimedia disclosures by demonstrating that nonverbal disclosure can influence investor reactions through information processing costs. We construct a measure of an unexplored dimension of voice, vocal-delivery quality, which complements other measures of vocal information.
Furthermore, our study contributes to the literature on strategic disclosure, providing evidence suggesting that managers could alter their nonverbal disclosures as a strategic tool to influence investor perception. This is a new dimension of strategic behavior that hasn’t been studied in depth before, extending our understanding of the dynamics of corporate disclosure strategies.
Finally, we highlight some important implications for practitioners. Investors should be aware that nonverbal information could be influencing their decisions. Our results imply that retail investors, in particular, are more influenced by vocal delivery quality than institutional investors, potentially due to differences in levels of sophistication. We also find that low-quality vocal delivery can slow down the price discovery process, which could be particularly relevant for traders who rely on rapid price adjustments. Moreover, regulatory bodies might consider whether guidelines or requirements are needed to ensure consistency in the quality of vocal delivery across firms.
This post comes to us from Bok Baik at Seoul National University, Alex G. Kim at the University of Chicago’s Booth School of Business, David Sunghyo Kim at MIT’s Sloan School of Management, and Sangwon Yoon at Seoul National University. It is based on their recent article, “Managers’ Vocal Delivery and Real-Time Market Reactions in Earnings Calls,” available here.