VOCAL CUES AND MANAGER REPORTING GOALS A Dissertation Presented to the Faculty of the Graduate School of Cornell University In Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy by Blake Allen Steenhoven May 2021 © 2021 Blake Allen Steenhoven VOCAL CUES AND MANAGER REPORTING GOALS Blake Allen Steenhoven, Ph. D. Cornell University 2021 This dissertation presents experimental and qualitative evidence on the acoustic properties of financial communications and how they relate to managers’ reporting objectives. Managers frequently attempt to persuade investors and other market participants to interpret information in a particular way. While research in accounting has examined characteristics that make a disclosure persuasive to investors, few papers examine whether and how managers attempt to persuade. Drawing on research in psychology, I show that managers increase their vocal pitch, pitch variation, volume, volume variation, and speech rate when attempting to persuade prospective investors to invest. Relative to this goal, I find that managers lower their pitch and pitch variation when attempting to persuade current investors to maintain their investment. To examine key assumptions of my experiment and provide rich descriptive evidence to contextualize my findings, I also conduct semi-structured interviews with Investor Relations professionals. This study has implications for preparers of financial communications, for investors using these disclosures, and for researchers attempting to identify manager reporting goals and understand the determinants of nonverbal behaviors in financial communications. BIOGRAPHICAL SKETCH Blake Steenhoven received his Bachelor of Business Administration from the University of Alaska Anchorage in 2014, majoring in accounting and economics. Before beginning his graduate studies, he worked as an audit associate for KPMG in Seattle, Washington, where he is licensed as a Certified Public Accountant. He received a Master of Science in 2020 and a Doctor of Philosophy in 2021 from Cornell University. Following his doctoral studies, Blake will join the faculty of Queen’s University as an Assistant Professor of Accounting. His research focuses on the judgment and decision making of investors, managers, financial analysts, and other capital market participants. iv TO MY PARENTS, THALE AND JEFF, AND GRANDPARENTS, PAULA AND JIM for being a constant source of love and support TO MY ADVISER, ROB BLOOMFIELD for encouraging me to think bigger TO MY ADVISER, KRISTI RENNEKAMP for being a mentor, role model, and friend v ACKNOWLEDGMENTS I am extremely grateful for the guidance and feedback from the members of my dissertation committee: Rob Bloomfield (co-chair), Kristi Rennekamp (co-chair), Bob Libby, and Melissa Ferguson. I would also like to thank Natasha Bernhardt, Shana Clor-Proell, Mike Durney, Brian Gale, Shannon Garavaglia, Ryan Guggenmos, Erik Harvey, Christoph Hörner, Eunjee Kim, Mani Sethuraman, Rosh Sinha, Dominique Wasna, Donnie Young, Xinyu Zhang, and workshop participants at Cornell University, Harvard Business School, Indiana University, Queen’s University, and Tilburg University for their helpful comments on the paper, as well as Sam Tilsen, Bruce McKee, and members of the Cornell Phonetics Lab for their guidance on the acoustic analyses. Additional thanks to the Deloitte Foundation and Cornell University for funding support. A special thank you to Kyle Hampton and Soren Orley for encouraging me to pursue a career in academia. vi TABLE OF CONTENTS Biographical Sketch.................................................................................................................................. iv Dedications ................................................................................................................................................ v Acknowledgments .................................................................................................................................... vi Table of Contents .................................................................................................................................... vii List of Tables ......................................................................................................................................... viii Section 1: Introduction .............................................................................................................................. 1 Section 2: Theory and Hypothesis Development ...................................................................................... 7 Nonverbal Behavior in Financial Communications ......................................................................... 7 Vocal Cues and Persuasion .............................................................................................................. 9 Persuasion Goals in Financial Communications............................................................................ 10 Section 3: Experiment ............................................................................................................................. 16 Participants .................................................................................................................................... 16 Design ............................................................................................................................................ 16 Procedure ....................................................................................................................................... 19 Dependent Measures ..................................................................................................................... 22 Section 4: Results .................................................................................................................................... 24 Section 5: Semi-Structured Interviews .................................................................................................... 30 Action-Focused and Inaction-Focused Reporting Goals in Communications with Investors ....... 31 Vocal Cues in Financial Communications ..................................................................................... 33 Response to Experimental Scenario .............................................................................................. 35 Management of Nonverbal Behaviors in Financial Communications ........................................... 37 Section 6: Conclusion .............................................................................................................................. 40 Appendix A: Audio Prompt and Response Scripts .................................................................................. 45 Appendix B: Experimental Manipulations .............................................................................................. 46 Appendix C: Variable Descriptions ......................................................................................................... 47 Appendix D: Prepared Questions for Semi-Structured Interviews .......................................................... 48 References ............................................................................................................................................... 49 vii LIST OF TABLES Table 1: Descriptive Statistics ................................................................................................................. 25 Table 2: Correlation Table ....................................................................................................................... 26 Table 3: Effects of Reporting Goals on Manager Vocal Cues ................................................................. 27 viii SECTION 1 INTRODUCTION Recent literature examines managers’ vocal cues and other nonverbal behaviors in financial communications (Mayew and Venkatachalam 2013), with early research suggesting that investors react to these cues in earnings calls (Mayew and Venkatachalam 2012) and video disclosures (Elliott, Hodge, and Sedor 2012; Cade, Koonce, and Mendoza 2020). While archival research suggests that managers’ vocal cues may convey value-relevant information about their emotional states, a growing literature in psychology and sociolinguistics suggests that vocal cues are also used strategically (Weinstein, Zougkou, and Paulmann 2018; van Zant and Berger 2020). In this study, I examine how managers vary their vocal cues when trying to persuade investors, and how these cues differ based on managers’ reporting goals. Persuasion is an important goal in communications with investors. Much of the research in financial accounting can be viewed as examining how investor, disclosure, and manager characteristics affect the persuasiveness of financial communications (Mercer 2004; Hamilton and Winchel 2019). Given the considerable evidence that discretionary disclosure choices affect investor judgments (Libby and Emett 2014) and that firms devote significant effort to these choices (Brown, Call, Clement, and Sharp 2019; Bamber and Abraham 2019), it is important to understand how managers actually attempt to persuade investors. Research in psychology has shown that speakers modulate their voices to influence others (Weinstein et al. 2018), suggesting that managers may do the same 1 when trying to persuade investors. Prior research on persuasion focuses on contexts where the goal is to persuade others to take an action, with a recent study finding that speakers raise their vocal pitch, volume, and speech rate when trying to persuade others to purchase a product (van Zant and Berger 2020). Similar goals are common in financial reporting, such as when managers try to persuade investors to invest in the firm. However, managers often try to persuade others to inaction. For example, managers may try to persuade current investors not to react to poor performance (Li 2008), restatements (Elliott et al. 2012), or criticism of firms’ accounting practices (Cade 2018). Managers may also prefer a more muted response to good news, due to equity incentives (Aboody and Kasznik 2000) or competitive and regulatory pressures (Bhojraj, Blacconiere, and D’Souza 2004). Given the prevalence of these goals and the importance of disentangling the effects of different reporting incentives, I also examine whether distinct patterns of vocal cues are used when persuading investors to inaction. While no research has examined how vocal cues are used to persuade others to inaction, regulatory fit theory (Cesario and Higgins 2008) suggests that managers will use vocal cues associated with a less eager (more vigilant) delivery style than when persuading others to action. I use a controlled experiment to examine whether managers vary their vocal cues when trying to persuade investors, and whether these cues differ for action- focused (persuading prospective investors to invest) and inaction-focused (persuading current investors not to sell) reporting goals. In my experiment, MBA students act as managers of a hypothetical company that is reporting below-expectations earnings for 2 the quarter, but has raised its full-year guidance. Participants are seated in a sound- attenuated room and record themselves giving scripted responses to questions from buy-side analysts. I use a within-participants design and manipulate reporting goals at three levels. In a practice condition, managers practice reading the scripted response to their Investor Relations Officer. I compare this practice condition to an action-focused condition where the analyst is described as a prospective investor and participants are given the goal of persuading him to invest in the company. I then compare the action- focused condition to an inaction-focused condition, where the analyst is described as a current investor and participants are given the goal of persuading him not to sell his investment. To measure managers’ vocal cues, I use Praat (Boersma and Weenink 2019), a speech analysis software used in phonetics research. Praat allows for automated extraction of vocal cues from audio recordings, which provides more objective measurements than the subjective ratings used in early persuasion research (Mehrabian and Willams 1969; Hall 1980; Burgoon, Birk, and Pfau 1990). Further, as a free, open-source software, Praat is more accessible to researchers and capital market participants than the proprietary speech analysis software used in prior accounting research (Mayew and Venkatachalam 2012; Hobson, Mayew, and Venkatachalam 2012). I find that, compared to their normal conversational speaking style, managers speak faster and use higher and more variable pitch and volume when persuading prospective investors to invest in the company. These results are consistent with prior research on action-focused persuasion goals in other contexts and suggest that prior 3 findings in the psychology literature replicate in an accounting context. I also find that managers use different vocal cues when persuading investors to inaction. Relative to action-focused persuasion attempts, managers use lower and less variable pitch when persuading current investors not to sell their investment. I also find that female managers (but not male managers) use lower volume for inaction-focused persuasion. These findings are generally consistent with the predictions of regulatory fit theory (Cesario and Higgins 2008) and suggest that managers use different patterns of vocal cues for different reporting goals. To examine key assumptions in my experiment and provide descriptive evidence to contextualize my findings, I conduct semi-structured interviews with twelve investor relations professionals. In these interviews, respondents shed light on managers’ reporting goals in communications with investors, the importance of vocal cues and other nonverbal behaviors to capital market participants, and how firms attempt to manage nonverbal communications. Investor relations professionals confirm that managers are sensitive to the differences between the action-focused and inaction-focused reporting goals I study. Respondents also indicate that investors are attentive to managers’ vocal cues, and suggest that managers use nonverbal behaviors to influence investors. However, they also express concerns about the inferences investors may draw from these cues, and suggest that these cues can communicate material nonpublic information and therefore present legal risks. To shed light on how executive presentation coaching could affect the generalizability of my results, I ask IROs about how they would advise managers to speak in the two scenarios presented in the experiment. Although there is some variation in recommendations for 4 persuading a prospective investor to invest, IROs generally recommend a calmer, more professional delivery style when persuading current investors not to sell, consistent with the effects observed in my experiment. Finally, respondents describe a number of methods used to manage vocal cues and other nonverbal behaviors, ranging from informal coaching during disclosure rehearsals to formal presentation training. These interviews provide rich, descriptive evidence that helps contextualize my findings and suggests interesting areas for future research. My study contributes to our understanding of managers’ nonverbal behavior in financial communications. Given channels like earnings calls, road shows, and private phone calls are viewed as among the most important for conveying a company’s message (Brown et al. 2019), a better understanding of the information unique to these channels is warranted. Whereas prior research has largely focused on information unintentionally revealed by managers’ vocal cues (Mayew and Venkatachalam 2012; Hobson et al. 2012), my findings suggest that these signals may reflect strategic efforts by managers to advance their reporting goals. I also contribute to the literature on strategic reporting. While extensive analytical research in accounting has examined strategic disclosure (Stocken 2013), archival research in this area is difficult because managers’ disclosure incentives are not uniform and actual intent is unobservable. It is also difficult to link specific disclosure choices to the manager, given the numerous other parties involved in crafting disclosures (Amel-Zadeh, Scherf, and Soltes 2019) and the difficulty of separating discretionary from non-discretionary choices. Experiments are well-suited to examining managers’ strategic disclosures, because they can manipulate or hold 5 constant factors confounded in archival settings (Libby, Bloomfield, and Nelson 2002). Rather than trying to infer managers’ goals from firm disclosures, I directly manipulate managers’ reporting goals and observe the effect on their behavior. In providing causal evidence on the effects of reporting goals on disclosure characteristics, my findings suggest that distinct patterns of vocal cues are used for different reporting goals. As such, managers’ vocal cues may contain useful signals for researchers and capital market participants attempting to identify managers’ reporting goals. This is an important finding, as several analytical papers suggest that uncertainty about managers’ reporting objectives can facilitate reporting bias (Fischer and Verrecchia 2000) or nondisclosure (Einhorn 2007) because investors are unable to anticipate and back out the effects of these choices. My study also makes a methodological contribution to the accounting literature. Audio from earnings calls, road show presentations, and other management disclosures is already being used by practitioners (Cohen and Malloy 2012) and is increasingly available to researchers (Mayew and Venkatachalam 2013; Teoh 2018). I introduce techniques and best practices from the phonetics literature for the processing and analysis of audio data and demonstrate how this data can be generated in a controlled lab setting to provide strong causal evidence. The remainder of this dissertation is organized as follows. In Section II, I provide my theory and develop my hypotheses. I describe my experimental design and present my results in Sections III and IV, respectively. In Section V, I summarize themes in responses from semi-structured interviews with investor relations professionals and provide representative quotes. Section VI concludes. 6 SECTION 2 THEORY AND HYPOTHESIS DEVELOPMENT Nonverbal Behavior in Financial Communications In addition to the quantitative information in disclosures, presentation features of financial communications are also informative to capital markets (Libby and Emett 2014). Drawing on advances in textual analysis (Loughran and McDonald 2016), substantial research in accounting and finance has examined language choices in firm disclosures (Li 2010). However, managers also communicate with market participants in richer, more interactive settings, such as earnings conference calls, road show presentations, and private phone calls, and investor relations officers rank these among the most important disclosure channels (Brown et al. 2019). While research on audio and visual cues in financial communications is in earlier stages, there is growing evidence that these cues are informative to capital markets (Mayew and Venkatachalam 2013; Blankespoor 2018). Although experimental and archival studies have largely focused on nonverbal behaviors in video disclosures (Elliott et al. 2012; Cade et al. 2020; Blankespoor, Hendricks, and Miller 2017), an emerging literature in accounting also examines managers’ vocal cues, driven in part by the increasing availability of audio data from earnings conference calls and tools for speech analysis (Mayew and Venkatachalam 2013; Teoh 2018).1 Several studies use proprietary software to analyze managers’ 1 Depending on the field and research paradigm, vocal cues are also referred to as elements of prosody, vocalics, acoustics, or paralanguage. For simplicity, I use “vocal cues” throughout the paper to refer to the non-lexical elements of speech, such as pitch, loudness, and speech rate. 7 speech in earnings calls, providing evidence that the acoustic signal contains value- relevant information (Mayew and Venkatachalam 2012; Price, Seiler, and Shen 2017) and may help detect financial misreporting (Hobson et al. 2012). Other research has examined specific vocal cues exhibited by managers in earnings calls, finding that managers’ voice pitch is negatively associated with CEO career outcomes (Mayew, Parsons, and Venkatachalam 2013) and with stock price movement during earnings calls (Mayew, Sethuraman, and Venkatachalam 2020). Despite growing evidence that managers’ vocal cues contain relevant information and affect investor judgments, there is little research in the accounting literature on the determinants of these cues. Existing experimental research generally focuses on investor reactions to experimentally manipulated nonverbal behaviors, whereas archival research largely draws on the “leakage hypothesis” (Ekman and Friesen 1969) to argue that observed cues are produced unintentionally due to managers’ affective and physiological states. Although substantial research has examined associations between nonverbal behaviors and emotional states (Scherer 2003; Bachorowski and Owren 2008) or the use of deception (Vrij, Hartwig, and Granhag 2019), speakers also use nonverbal cues for communicative purposes (Burgoon, Guerrero, and Floyd 2016).2 As such, vocal cues and other nonverbal 2 The difference between nonverbal cues produced intentionally and unintentionally is consistent with the distinction between nonverbal communication and nonverbal behavior (Hall, Horgan, and Murphy 2019). While the terms are often used interchangeably, nonverbal communication describes cues which are intentionally produced to inform or influence others. In contrast, nonverbal behavior represents cues that are unintentionally produced. While nonverbal behavior may be informative (e.g., a person can infer that someone who yawns is tired or bored), its function is not to communicate that information. 8 behaviors may be informative about managers’ intentions and reporting goals in financial communications. Vocal Cues and Persuasion Although it is commonly accepted that a substantial amount of meaning is conveyed nonverbally, research in psychology and sociolinguistics examines how speakers use vocal cues for various communicative functions. Research has shown that speakers use vocal cues to inform others about their attitudes (Mitchell and Ross 2013; Freeman 2014) and intentions (Hellbernd and Sammler 2016), and to influence the perceptions and behaviors of others (Fraccaro et al. 2013; Hughes, Mogilski, and Harrison 2014; Weinstein et al. 2018). Following Mehrabian and Williams (1969), which finds that speakers increase speech rate, volume, and intonation when attempting to persuade, several studies have examined the relationship between vocal cues and persuasion. Hall (1980) similarly finds that speakers modulate their vocal cues when attempting to persuade, and provides some evidence that these cues are effective. Focusing on receivers of persuasive communications, Burgoon et al. (1990) provide evidence that vocal cues can affect persuasiveness through perceptions of speaker credibility. Of the specific vocal cues associated with persuasion, early research focused primarily on speech rate, suggesting that faster speech may increase persuasiveness by signaling source credibility (Miller, Maruyama, Beaber, and Valone 1976) or imposing processing costs on listeners (Woodall and Burgoon 1983, Smith and Shaffer 1991). More recent research has examined vocal characteristics like vocal pitch, intonation, and volume. Jiang and Pell (2017) find that speakers vary their voices 9 when projecting different levels of confidence, and that listeners’ ratings of confidence are associated with louder speech and more variable pitch and volume. Guyer, Fabrigar, and Vaughan-Johnston (2019) manipulate audio files to vary the pitch of speakers, and find that lowered pitch and falling intonation increase persuasiveness through perceptions of speaker confidence. A recent study by van Zant and Berger (2020) examines both the cues that speakers produce when attempting to persuade others and the cues associated with attitude change. They find that speakers use higher pitch, volume, and speech rate, and more variable pitch and volume when attempting to persuade. For receivers of these communications, higher volume and more variable volume are associated with increased persuasiveness. These studies suggest that vocal cues may be an important tool for managers to increase persuasiveness, or for market participants to identify managers’ persuasive intent, in financial communications. Persuasion Goals in Financial Communications Persuasion is an important goal in communications with investors. Much of the research in financial accounting can be viewed as examining how investor, disclosure, and manager characteristics affect the persuasiveness of financial communications (Mercer 2004; Hamilton and Winchel 2019). Despite considerable evidence that discretionary disclosure choices affect investor judgments (Libby and Emett 2014) and that firms devote significant effort to these choices (Brown et al. 2019; Bamber and Abraham 2019), less focus has been given to how managers actually attempt to persuade investors. Understanding the effects of managers’ persuasive intent, or reporting goals, is important to researchers, as well as consumers of firm disclosures, attempting to identify strategic disclosure choices by managers (Fischer and 10 Verrecchia 2000; Einhorn 2007). However, because manager intent is unobservable, reporting goals and their consequences are difficult to study using archival data. Experimental methods are well-suited for examining these effects, because manager reporting goals can be manipulated and manager decisions can be observed, controlling for other determinants of disclosure which are confounded in natural settings (Libby et al. 2002). For example, Asay, Libby, and Rennekamp (2018) examine the effects of managers’ reporting goals on linguistic features of disclosure by directly manipulating reporting goals, while holding constant other drivers of linguistic characteristics, like firm or investor characteristics. Other studies examine how characteristics of the intended audience affect discretionary disclosure choices. Clor-Proell and Maines (2014) find that the requirement to recognize or disclose an estimate affects the reporting decisions of managers of public companies, but not managers of private companies. The authors argue that the effects for public company managers reflect anticipated scrutiny from capital market participants. Several recent working papers also show that managers’ disclosure decisions are affected by audience characteristics (Durney 2019; Witz 2020). These studies collectively demonstrate that managers disclose differently to different audiences and to achieve different reporting goals. While research in psychology finds consistent associations between vocal cues and persuasion, these studies focus exclusively on scenarios where the persuasion goal is to induce action in others (e.g., persuading others to buy a product). This type of persuasion goal is pervasive in financial communications, such as when managers attempt to persuade prospective investors to invest in the company. To the extent that 11 patterns of vocal cues identified in the persuasion literature reflect action-focused persuasion, I predict that managers will use similar cues when attempting to persuade investors to take an action. My first hypothesis is therefore: H1: When attempting to persuade investors to action, managers will use higher pitch, pitch variation, volume, volume variation, and speech rate than when they have no reporting goal. While the predictions of H1 are consistent with the findings of prior research in psychology, it is important to establish that these effects extend to persuasion goals in accounting settings. Given that contextual and interpersonal factors can affect nonverbal communication (Burgoon et al. 2016), it is unclear whether cues identified in prior research will be the same as those used when persuading investors to action.3 Although action-focused goals are common in communications with investors, extensive research in accounting has also examined settings where managers may attempt to persuade others to inaction. Much of this research focuses on how content and formatting choices of disclosures can limit reactions to poor performance (Li 2008), restatements (Elliott et al. 2012), criticism of firms’ accounting practices (Cade 2018), earnings management (Nelson, Elliott, and Tarpley 2002), and other negative information.4 Managers may also attempt to limit reactions to positive information due to equity-based compensation incentives (Aboody and Kasznik 2000), or when 3 My semi-structured interviews with Investor Relations Officers in Section V highlight a number of differences between communications with investors and the marketing settings which are largely the focus of prior research. 4 Blankespoor, deHaan, and Marinovic (2020) note that managers might prefer a slower response to disclosed information “in order to reduce frivolous litigation from abrupt stock drops, allow time for good news to offset bad news, […] to allow time to place insider trades, or to reduce negative media attention that can damage career prospects.” 12 disclosures may be used by competitors or regulators (Bhojraj et al. 2004). Audience characteristics may also induce different persuasion goals, such as when managers persuade current investors to maintain (not to sell) their position. While no prior research has examined how speakers vary their vocal cues when persuading others to inaction, research on regulatory fit (Higgins 2000; 2005) suggests that these cues are likely to differ from those used in action-oriented persuasion attempts. Regulatory fit theory suggests that individuals are more receptive to messages that are consistent with their motivational focus (Cesario, Higgins, and Scholer 2008). Specifically, individuals with a “promotion” focus visualize goals as desirable end-states, and are more receptive to “eager,” advancement-focused messages. In contrast, individuals with a “prevention” focus visualize goals as the avoidance of undesirable end-states, and are more receptive to “vigilant,” safety- oriented messages. Prior research has found that persuasion attempts are more effective when the content or framing of the message is consistent with the regulatory focus of the recipient (Cesario, Grant, and Higgins 2004; Lee and Aaker 2004). Studies focusing on physical nonverbal behaviors have similarly shown that messages communicated with an eager (vigilant) style of delivery are more effective for promotion-focused (prevention-focused) receivers (Cesario and Higgins 2008). Although Cesario and Higgins (2008) manipulate speech rate, the literature has otherwise been silent on the vocal cues associated with an eager or vigilant style of delivery. However, the vocal cues previously associated with persuasion, such as higher and more variable pitch and volume, are also associated with an activated, excited emotional state (Goudbeek and 13 Scherer 2010; Scherer 2013). Given that these studies have focused exclusively on action-oriented persuasion goals, I predict that these cues are associated with a more eager delivery style. As regulatory fit theory suggests a vigilant style is more effective than an eager style for inaction-focused persuasion (Cesario and Higgins 2008), I predict that managers will use vocal cues associated with a more vigilant, cautious style when persuading investors to inaction. Underlying my prediction is that this delivery style has been effective for inaction-focused persuasion by managers in other contexts; however, it is worth noting that this prediction does not require that a vigilant style is actually more effective for inaction-focused goals in an accounting context.5 To the extent that a more vigilant delivery style reflects a lower level of activation or emotional excitement, I expect that managers will use lower and less variable pitch and volume and a lower speech rate when persuading investors to inaction than when persuading investors to action. My second hypothesis is as follows: H2: When attempting to persuade investors to inaction, managers will use lower pitch, pitch variation, volume, volume variation, and speech rate than when persuading investors to action. For completeness, I also compare vocal cues used when persuading investors to inaction to those used with no persuasion goal. As noted previously, prior research has shown that individuals use higher levels of these vocal cues when persuading other 5 While not the primary focus of this study, prior research argues that current and prospective investors differ in regulatory focus (Cianci and Falsetta 2008; Harris, Hobson, and Jackson 2016). As such, regulatory fit theory suggests that the changes in delivery I predict may increase persuasiveness. I shed additional light on the effectiveness of different delivery styles in an accounting context in my semi- structured interviews with Investor Relations Officers in Section V. 14 to take an action than when they have no persuasion goal (van Zant and Berger 2020); however, no studies have compared vocal cues used when persuading others to inaction to those used with no persuasion goal. Further, regulatory fit theory predicts only that different delivery styles will be used for action-focused and inaction-focused persuasion; it does not provide a basis for predicting how vocal cues used when persuading investors to inaction will differ from those used with no persuasion goal. As theory and prior literature do not provide a basis for prediction, I treat this comparison as a research question and state the following hypothesis in the null form: RQ: When attempting to persuade current investors not to sell their investment in the company, managers’ pitch, pitch variation, volume, volume variation, or speech rate will not differ from when they have no reporting goal. 15 SECTION 3 EXPERIMENT Participants Participants are 108 MBA students (76 male and 32 female) enrolled in a top- ranked MBA program in the Northeastern United States. I exclude one participant who did not provide all of the recordings and two who deviated from the script, resulting in a total sample of 105 participants.6 MBA students enrolled in this program are well- suited to the goals of the experiment, as over one-third take jobs in financial services after graduation, and participants in my experiment have approximately 6 years of work experience, with 67 percent indicating experience in a managerial role. Further, I examine linguistic cues, which are naturally acquired and reinforced components of speech (Xu 2010).7 All participants identify as native English speakers. Design Participants assume the role of CEO of Galaxy Brewing, a fictitious craft beer company, and are tasked with responding to questions asked by investors in private phone calls. I use a 1 x 3 within-participants design in my experiment, where participants respond once without a persuasion goal (Practice condition), once with the goal of persuading a prospective investor to invest in the company (Action Goal 6 I do not remove recordings that contain speech disfluencies (e.g., “uh” or “um”), as the length of the recordings makes it unlikely that they will have a significant impact on measurements of vocal cues. Similar research by van Zant and Berger (2020) finds no differences in vocal cues for recordings containing disfluencies. 7 The generalizability of my results may be limited to the extent that managers receive vocal coaching for executive communications. I shed light on the nature of this training in my semi-structured interviews with Investor Relations Officers in Section V. 16 condition), and once with the goal of persuading a current investor not to sell their investment in the company (Inaction Goal condition). I use the setting of private phone calls with investors for several reasons. First, it allows a more direct manipulation of reporting goals. In public disclosures, managers are simultaneously communicating with multiple audiences, including current and prospective investors (Amel-Zadeh et al. 2019). Because managers’ reporting goals can be audience-specific, strategic disclosure choices in these settings will be a function of the salience, importance, and characteristics of these audiences (Bhojraj et al. 2004; Durney 2019). By using a private disclosure setting with individual investors, I abstract away from the influence of different audiences to provide a cleaner manipulation of reporting goals. Second, private communications with management are common, and buy-side and sell-side analysts view these communications as important (Brown, Call, Clement, and Sharp 2015, 2016). While these meetings are difficult to study using archival methods, an experiment allows me to shed light on communication in this important disclosure setting. All participants complete the Practice condition recording first, while the ordering of the Action Goal and Inaction Goal recordings is randomized across participants. This design choice serves at least two purposes. First, the primary focus of this study is on how vocal cues differ between reporting goals. For this reason, to the extent that a lack of familiarity with the script creates noise in the vocal cues produced (e.g., stuttering, speech disfluencies, or misreading), these effects will be less pronounced in the Action Goal and Inaction Goal conditions if those recordings come after participants have had an opportunity to practice. Second, because I model 17 the first recording as “practice” before taking calls from investors, the ordering is consistent with preparation and rehearsal by practitioners prior to verbal disclosures (Brown et al. 2019). In addition to increasing the ecological validity of the study, this choice increases the likelihood that participants will use vocal cues appropriate for the setting (Xu 2010). Participants are provided a scripted response to use in all three conditions. The use of a scripted response is consistent with prior phonetics research. As noted by Wagner, Trouvain, and Zimmerer (2015), the “dominant form of speech material used in phonetics research remains read speech recorded in a highly controlled laboratory situation.” The provision of a script also allows me to control for the effects of content or language differences on vocal cues (Xu 2010), which increases internal validity for the effects of reporting goals on managers’ vocal cues in my experiment. Although I use a script in the experimental task, I make certain design choices to increase the chances that participants’ recordings reflect more natural speech. First, I hire voice actors to record audio prompts for each recording. This increases engagement with the task, encourages participants to adopt a speaking style appropriate for the context, and approximates an interactive setting. Scripts provided to voice actors for these audio prompts are presented in Panels A, B, and C of Appendix A.8 Second, to address concerns that lab speech is “overplanned” relative to 8 An alternative approach to a recorded prompt would be to assign a separate group of participants to the role of current and prospective investors. It is likely that a more interactive design would better capture the richness of communication between managers and investors at the expense of some experimental control (see Kachelmeier (2018) for a related discussion). While it is possible that greater interactivity would affect the vocal cues used for various reporting goals, sociophonetics research has found that speakers use similar cues when prompts are prerecorded or spoken live (Xu 2010). Given that research on managers’ vocal cues is in early stages, I use a less interactive design to establish clean causal relationships that future research can explore in less controlled interactive environments. 18 spontaneous speech, participants use a relatively long script (presented in Panel D of Appendix A), reducing participants’ ability to fully plan their vocal cues before speaking (Xu 2010). Third, to increase the “naturalness” of the response, I adapt the script from actual earnings call transcripts, retaining elements associated with conversational speech (e.g., beginning sentences with conjunctions and using colloquial expressions). Participants are seated in a private, sound-attenuated room for the duration of the experiment in order to reduce environmental noise in audio recordings. Voice recordings are made using a table-mounted Shure SM57 dynamic cardioid microphone with a windscreen, positioned approximately 30cm in front of the participant (Švec and Graqvist 2010). Participants are asked to maintain a constant distance from the microphone when speaking. Audio is digitized using a Zoom H4n Pro digital recorder (24-bit, mono, 44.1kHz sampling rate) and recorded as WAV files. Procedure Participants receive background information about the firm and record a statement describing the company and their role as CEO, which reinforces the setting. Participants are then informed that the company has just released its earnings for the quarter and has updated its full-year guidance. In all conditions, the company missed its previously issued guidance for the current quarter, but is raising its full year guidance. Management forecasts provide a useful setting for examining persuasion goals, due to the uncertainty about estimation error and bias in forward-looking disclosures (Rogers and Stocken 2005), and they are most commonly bundled with earnings announcements (Anilowski, Feng, and Skinner 2007; Rogers and Van 19 Buskirk 2013). Further, the bundling of positive and negative news creates ambiguity about earnings persistence, making it plausible that prospective investors are considering investing in the company and current investors are considering selling their investment. After reviewing the company’s press release, participants are informed that they will be taking private phone calls from several investors, and that their Investor Relations Officer has scripted a response to anticipated questions about the recent quarter performance and full year outlook. The response explains that the earnings miss is due to poor performance by a product line the company introduced at the beginning of the quarter. However, the company is increasing its full-year guidance, asserting that investments in the brand during the quarter will drive growth in the remainder of the year. All participants are instructed by their Investor Relations Officer to practice reading the response once out loud “in a conversational way” (Practice condition). This recording allows an assessment of the participant’s vocal cues in the absence of explicit reporting goals or investor characteristics, and is consistent with the instructions given in the no-goal conditions in prior research (van Zant and Berger 2020). Next, participants are informed that they will be taking calls from important investors. In the Action Goal condition, the investor is described as a prospective investor “known for pursuing investment opportunities to maximize his portfolio value” and that he is considering investing “because of the strong outlook for the year.” In the Inaction Goal condition, the investor is described as a current investor “known for monitoring threats to his investments to protect his portfolio value” and is 20 considering selling his investment “because of the weak performance this quarter.” Descriptions provided to investors are provided in Appendix B. While the investor descriptions may be sufficient to activate persuasion goals, I increase the strength of my manipulation in two ways. First, I include a comprehension check asking whether their goal is to persuade the investor “to make an investment in Galaxy” or “not to sell his investment in Galaxy.” Participants are then directed to a page that indicates whether they responded correctly or incorrectly and repeats their goal. Second, participants are instructed to try to respond in a way that “plays up the positive full-year outlook” when responding to the prospective investor, and in a way that “downplays the weak performance in the first quarter” when responding to the current investor. Participants then take both calls and respond to the prospective and current investors, with the order randomized across participants.9 To increase realism and engagement with the task, participants are given audio prompts to begin each recording. I hired three professional voice actors to read prepared scripts for the prompts, presented in Appendix B. One actor reads as the firm’s investor relations officer and introduces the investor (Panel A), while the other two actors read as investors asking about the company’s current quarter earnings and future guidance (Panels B and C). Because participants take calls with two investors, each actor reads 9 One potential concern with a within-participants design is that differences in responses may reflect demand effects. However, participants are not informed that their vocal cues will be analyzed. Further, although participants may infer that their delivery should vary based on their reporting goals, they are given no direct or implied indication of how they should vary their vocal cues, either in the variables they should adjust or the direction in which they should adjust them. Thus, it seems unlikely that participants are drawing inferences about whether and how they should alter their delivery from seeing a second persuasion goal. Consistent with this, all analyses control for the order in which participants are assigned reporting goals, and I find no evidence of order effects in any analysis. 21 two variants of the question. I randomly assign each voice actor and question variant to the current and prospective investor to control for differences in vocal delivery and minor wording differences between the two questions. After recording both responses, participants provide demographic information and are debriefed. Dependent Measures To measure the effect of persuasion goals on manager vocal cues, I extract the acoustic properties of each recording using Praat software (Boersma and Weenink 2019). Praat is a free, open-source software commonly used in phonetics research to analyze and manipulate acoustic properties of recorded speech. Whereas early research on vocal cues and persuasion relied on subjective ratings of vocal cues, Praat allows for a more objective and standardized assessments of these cues and can easily analyze large samples of audio files that would be cost-prohibitive using subjective raters. To ensure that measurements are consistent across recordings, I manually segment each file to begin with the first voice onset (i.e., the first moment where vocal pitch is identified) and end with the last voice offset (i.e., the last moment where vocal pitch is identified). I measure speech rate in words-per-second, calculated by dividing the number of words in the script (held constant at 164) by the duration of the recording. I automate the extraction of all other vocal cues using a script and use default Praat settings except as noted below. To measure pitch and pitch variation, I use the mean and standard deviation of fundamental frequency (F0), respectively. Fundamental frequency is the acoustic correlate of pitch,10 and captures the rate at which peaks in a sound wave occur; in 10 As noted by Ladefoged (2003), the term “pitch” refers to an auditory property, reflecting the way a 22 recorded speech, these peaks are caused by vibrations in a speaker’s vocal folds (Ladefoged and Johnson 2014). The standard measurement of F0 is hertz (Hz), which represents the number of complete vibrations of the vocal folds per second. Consistent with recommendations for vocal acoustics research (Vogel, Maruff, Snyder, and Mundt 2009) and with prior research (van Zant and Berger 2020), I apply pitch-range settings of 70-250Hz for male participants and 100-300Hz for female participants. To reduce measurement error, I extract the pitch listing for each recording and remove observations using the median absolute deviation following the recommendations of Leys et al. (2013).11 To measure volume and volume variation, I use the mean and standard deviation of intensity, respectively. Intensity is the acoustic correlate of perceived loudness, and measures the acoustic energy in the sound wave reflecting a speaker’s respiratory effort (Ladefoged 2003). Intensity is measured as the root mean square amplitude of the sound wave and is most commonly expressed in decibels (dB). Consistent with prior research, I measure intensity over only the voiced portions of each recording (van Zant and Berger 2020). listener perceives a sound. As such, pitch is not an objective property that can be measured from audio data. In contrast, fundamental frequency is an acoustic property which can be measured from the sound wave. Pitch and fundamental frequency are strongly correlated, such that many texts use the terms interchangeably. 11 The automated process of extracting vocal cues can result in extreme observations which are simply measurement errors, rather than extreme deviations in vocal pitch. Ideally, this would be addressed by manually inspecting each pitch listing to identify these non-pitch artifacts. However, this is infeasible, as pitch measurements are generated every ten milliseconds, resulting in approximately 1.7 million observations in my dataset. As such, I use the median absolute deviation to identify and remove outliers following the standards outlined in Leys et al. (2013). All significant results remain significant if I alternatively exclude observations greater than 2 standard deviations above or below the mean. 23 SECTION 4 RESULTS Descriptive statistics for vocal cues by reporting goal condition are presented in Table 1, with raw means and standard deviations for female (male) participants presented in Panel A (Panel B). Vocal cues are as defined in the previous section and described in Appendix C. Correlations between vocal cues are presented in Table 2. I test differences across reporting goal conditions using mixed-effects models with participant random effects (Baayen, Davidson, and Bates 2008). I model participant gender, its interaction with reporting goals, and the order in which participants responded to the current and prospective investor as fixed effects, consistent with prior research (van Zant and Berger 2020). I find no evidence of order effects or, except where noted below, gender by reporting goal interactions (all p > 0.10). As such, I present coefficient estimates for the effects of reporting goals and, for brevity, discuss other effects only where significant. Table 3 presents the coefficients, standard errors, and tests of the effects of reporting goals on vocal cues. Hypothesis 1 predicts that managers attempting to persuade prospective investors to take a position in the company will use vocal cues associated with an eager speaking style. Consistent with prior research on persuasion, H1 predicts that the Action Goal condition will have higher pitch, pitch variation, volume, volume variation, and speech rate, relative to the Practice condition. Results of my test of H1 are presented in Panel A of Table 3. Consistent with my predictions, participants with an Action Goal use higher pitch (p < 0.001), more variable pitch (p < 0.001), higher 24 TABLE 1. Descriptive Statistics Panel A. Female Participants Practice Action Goal Inaction Goal Vocal Cue Condition Condition Condition Mean Pitch 182.19 183.91 182.47 (21.89) (18.78) (19.56) St. Dev. Pitch 25.46 27.46 26.37 (8.01) (7.40) (7.33) Mean Volume 48.27 49.08 48.68 (2.63) (2.56) (2.61) St. Dev. Volume 3.63 3.70 3.69 (0.57) (0.51) (0.52) Speech Rate 3.06 3.17 3.17 (0.35) (0.33) (0.29) Panel B. Male Participants Practice Action Goal Inaction Goal Vocal Cue Condition Condition Condition Mean Pitch 113.44 115.56 115.37 (14.32) (14.51) (14.18) St. Dev. Pitch 15.55 16.83 16.47 (5.36) (5.77) (4.95) Mean Volume 50.04 50.54 50.70 (3.60) (3.28) (3.32) St. Dev. Volume 3.68 3.74 3.75 (0.58) (0.59) (0.54) Speech Rate 3.13 3.26 3.24 (0.37) (0.34) (0.35) This table reports descriptive statistics for vocal cues examined in the main experiment. Panel A (Panel B) presents descriptive statistics for female (male) participants. Mean values are presented for each reporting goal condition, with standard deviations presented in parentheses below. All variables are described in Appendix C. 25 TABLE 2. Correlation Table Mean St. Dev. Mean St. Dev. Speech Pitch Pitch Volume Volume Rate Mean Pitch 1.000 0.770 -0.073 0.174 -0.124 <0.001 0.196 0.002 0.027 St. Dev. Pitch 1.000 0.179 0.303 -0.081 0.002 <0.001 0.149 Mean Volume 1.000 0.570 0.064 <0.001 0.255 St. Dev. Volume 1.000 -0.089 0.116 Speech Rate 1.000 This table reports correlations between vocal cues examined in the experiment. Pairwise correlations are presented, with two-tailed p-values below. All variables are described in Appendix C. volume (p < 0.001), more variable volume (p = 0.030), and speak faster (p < 0.001) than in the Practice condition.12 These results provide strong support for H1. Hypothesis 2 predicts that managers attempting to persuade current investors not to sell their position in the company will use vocal cues associated with a more vigilant speaking style. As such, compared to the Action Goal condition, H2 predicts that the Inaction Goal condition will have lower pitch, pitch variation, volume, volume variation, and speech rate. Results of my test of H2 are presented in Panel B of Table 3. Consistent with my predictions, participants with an inaction goal use lower pitch 12 Tests of directional predictions are one-tailed. All other p-values are based on two-tailed tests. 26 TABLE 3. Effect of Reporting Goals on Manager Vocal Cues Panel A. Action Goal vs. Practice (Hypothesis 1) Vocal Cue Estimate Std. Error t-stat p-value Mean Pitch 0.96 0.26 3.63 <0.001 St. Dev. Pitch 0.82 0.17 4.96 <0.001 Mean Volume 0.33 0.06 5.11 <0.001 St. Dev. Volume 0.03 0.02 1.90 0.030 Speech Rate 3.62 0.67 5.41 <0.001 Panel B. Inaction Goal vs. Action Goal (Hypothesis 2) Vocal Cue Estimate Std. Error t-stat p-value Mean Pitch -0.41 0.22 -1.89 0.031 St. Dev. Pitch -0.36 0.15 -2.45 0.008 Mean Volume -0.06 0.05 -1.33 0.093 St. Dev. Volume 0.00 0.01 0.11 0.544 Speech Rate -0.25 0.51 -0.49 0.312 Panel C. Inaction Goal vs. Practice (Research Question) Vocal Cue Estimate Std. Error t-stat p-value Mean Pitch 0.55 0.25 2.22 0.029 St. Dev. Pitch 0.46 0.17 2.67 0.009 Mean Volume 0.27 0.34 4.63 <0.001 St. Dev. Volume 0.03 0.01 2.33 0.022 Speech Rate 3.37 0.61 5.54 <0.001 This table summarizes the effects of reporting goals on manager vocal cues. Panel A compares vocal cues used with an action goal to those used in a practice condition (Hypothesis 1). Panel B compares vocal cues used with an inaction goal to those used with an action goal (Hypothesis 2). Panel C compares vocal cues used with an inaction goal to those used in a practice condition (Research Question). All variables are described in Appendix C. I present coefficients and standard errors for the effects of persuasion goal condition on vocal cues from a linear mixed model controlling for gender, a gender by reporting goal interaction, and order as fixed effects, and participant random effects. Reported p-values in Panel A and Panel B are one-tailed, given my directional predictions. 27 (p = 0.031) and less variable pitch (p = 0.008) than with an action goal. I also find that participants with an inaction goal use lower volume, although this effect is only marginally significant (p = 0.093) and is qualified by a significant gender by reporting goal interaction (p = 0.003, untabulated). Probing the simple effects, I find that volume is significantly lower in the Inaction Goal condition than the Action Goal condition for female participants (p = 0.005, untabulated), but not for male participants (p = 0.939, untabulated). I find no differences between reporting goals for volume variability (p = 0.544) or speech rate (p = 0.312). Collectively, these results support the predictions of H2 for pitch and pitch variation, but support the predictions for volume only among female participants. Given the lack of prior research on how vocal cues used for inaction-focused persuasion goals differ from those used without a persuasion goal, I make no formal predictions regarding how vocal cues in the Inaction Goal condition differ from those in the Practice condition. The comparison of these two conditions is presented in Panel C of Table 3. Relative to the Practice condition, participants with an inaction goal use higher pitch (p = 0.029), more variable pitch (p = 0.009), higher volume (p = 0.001), more variable volume (p = 0.022), and speak faster (p < 0.001). While I make no ex ante predictions regarding these effects, these results largely mirror the effects of an action goal relative to the Practice condition in Panel A of Table 3 suggesting several possible explanations. First, participants in the Practice condition were instructed to try to sound unscripted in reading their response. It is possible that, relative to a persuasion goal, participants try to sound unscripted by speaking in a slower, more relaxed manner. Second, given that participants always 28 complete the Practice condition first, differences could reflect a lack of familiarity with the script. Third, prior research suggests that questioning from analysts may affect managers’ emotional states (Mayew and Venkatachalam 2012). As such, relative to the Practice condition delivered to the company’s Investor Relations Officer, participants may have felt greater stress when responding to analysts. Results are consistent with research showing that increases in pitch, volume, and speech rate are associated with emotional arousal (Goudbeek and Scherer 2010; Scherer 2013). 29 SECTION 5 SEMI-STRUCTURED INTERVIEWS In designing my laboratory experiment, I relied on several assumptions about persuasion in real-world investment settings. To test these assumptions and to provide additional context for my findings, I conduct semi-structured interviews with twelve Investor Relations Officers (IROs) of publicly traded companies. Survey evidence suggests that IROs are heavily involved with preparation and rehearsal for communications with investors (Brown et al. 2019). As such, they have relevant experience and are well-positioned to speak to these issues. Each interview was approximately 30 minutes long and was conducted using video conferencing software. Interviews were recorded and transcribed to exclude identifying information. In this section, I summarize themes in responses, with representative quotes included in the text.13 My prepared interview questions are shown in Appendix D. First, to confirm that managers are sensitive to the differences between the action-focused and inaction- focused reporting goals I study, I ask about how managers prepare for communications with investors and how characteristics of the disclosure audience affect this process (Questions 1 and 2). Second, I ask about how firms consider delivery style in communications with investors to confirm that managers believe investors are attentive to vocal cues, such that managers are likely to vary them 13 For each quote, I include an identifier for the respondent number and the question to which the participant is responding. For example, a response to Question 2 (see Appendix D for the list of questions) from the third respondent I interviewed would be labeled [R3-2]. 30 (Question 3). Finally, to shed light on how executive presentation coaching could affect the generalizability of my results, I ask IROs about how they would advise managers to speak in the two scenarios presented in the experiment (Question 4). Because I adopt a semi-structured approach, I deviate from the list of prepared questions to follow up on participant responses to provide additional context for their experiences. Respondents note a variety of ways that IROs attempt to manage vocal cues and other nonverbal behaviors in communications with investors. While outside the scope of my study, I summarize these responses to provide additional depth of understanding for the setting and to motivate future research in this area. Action- and Inaction-Focused Reporting Goals in Communications with Investors My laboratory experiment assumes that managers are sensitive to the differences between action-focused and inaction-focused reporting goals in communications with investors. Respondents note that they explicitly consider reporting goals when preparing for communications with investors. Consistent with the assumptions in my experiment, respondents note that persuading prospective investors to invest in the company and persuading current investors not to sell their position are both important goals. What are we trying to accomplish in this meeting, right? Do we want them to buy shares? Do we want them to be a shareholder of ours? Along those lines. […] In some cases they might be an existing shareholder. So, not wanting them to sell their position is equally as important. So, it’s not necessarily always about increasing, it’s also about maintaining. [R9-2] I always think about being positive. I will acknowledge when, yes, we had a tough month, there is no doubt. And I have the empathy, because I know they’re going to want to hear that, too. Because they don’t want you to just come out and say like, “Oh sure, whatever, bad month. But look at the longer track, of course we’re great. We’re great.” Nobody wants to hear that. [R4-3] 31 Whatever you put out for guidance or have put out for guidance is like a promise to the street, and if you don’t deliver, you’ve broken the promise. So, the street doesn’t want to hear that you’re going to get better. They’re still in the moment, mildly pissed off that you broke the promise. And it may be – you may be promising blue skies and rainbows going forward, but they don’t want to focus on that on the day. [R7-2] Respondents also confirm that reporting goals vary based on the disclosure audience, consistent with the manipulation of reporting goals in the experiment, and that they use information about investors to tailor their communications. I think it’s important to know your investors, know what they’re looking for. What’s their profile? What I mean by that is, what are they looking for as an investor? What matters most to them? Is it growth, profitability? What are they looking to hear? [R9-2] I have to be consistent with what I present from a facts and figures perspective on the financial disclosures and strategy, and things like that. But the nuance and the art of what you’re doing as an IR person is figuring out how to tell the story to the person sitting across from you in a way that’s meaningful and creates an actionable process for them to come into the stock. [R6-2] Respondents also express limitations to action-focused goals in financial communications, noting that trying to play up positive performance can create unrealistic expectations. There were times that I felt like he was coming at it from the angle of, “I really want to put a good face on this.” And those were what I tried to guard against. If he was excited about a new product, I’d try to rein him back a little bit, like don’t overpromise on this, because in six months you’re going to have to talk about it again, and it’s not going to be good if it didn’t work. [R1-3] If we had a really good performance, we’re going to say, “Look, we hope to see more gains like this again, but we don’t want people to draw a hockey stick of improvement,” right? We’re trying to be…internally sensible about what that looks like. [R7-2] These responses collectively confirm that managers explicitly consider action- focused and inaction-focused reporting goals when communicating with investors, that 32 these goals are shaped by audience characteristics, and that action-focused persuasion goals may be moderated in financial communications, relative to the other settings where persuasion has been examined. Vocal Cues in Financial Communications Respondents confirm that, in addition to the information content, investors analyze managers’ vocal cues and other nonverbal behaviors. We even at one of the places I worked, publicly traded company, had to analyze our earnings call transcript because we wanted to get feedback. Because different hedge funds are using these services these days that interpret the tone, the voice, the words, how they were communicated. These are people from the military, and from government that do this now for different hedge funds. [R5-3] I mean, in reality now, when our calls are being recorded, there are now AI algorithms running in the background listening to our calls with some of these hedge funds, that are analyzing not only the voice inflection of the CEO as he’s reading the script at the beginning of the call and answering questions, but also looking for patterns in speech […] it’s getting to the point where it’s crazy how much is watched and recorded and analyzed. [R6-3] Consistent with managers using vocal cues and other nonverbal behaviors to influence investors, respondents note that managers generally try to project optimism in their delivery. My personal view is you’re also not trying to sell the stock, you’re just trying to present the facts and let them make the decision according to their investment portfolio and thesis. I know that…I would typically find that with new management that hadn’t dealt with investors, they tried to sell the stock a little bit more. [R1-3] Even if it was a crappy quarter and we missed something, our executives are still going to sound as positive and upbeat as they can. So, you don’t have to tell them that. [R8-3] Although respondents largely agree that managers’ nonverbal behaviors should convey positivity, most indicate that managers should generally use a more moderate tone when communicating with investors. 33 So, tone of voice matters very much because the CFO is talking about something that may have been difficult in the quarter – you know, you don’t want him to come across as defensive or overly excited about an opportunity, or overly mopey about something that may not have gone well. So, the tone needs to be positive, but not exaggerated in any way on any front. [R3-1] You don’t want the ShamWow guy selling your stock. […] I guess it’s – if you had to take a survey of “How did you find this person’s presentation?” You want them – you want every investor to come out and be like, “Slightly positive”, “Slightly positive.” So, it gives – legitimately, because you don’t want to be like, “Okay, this guy’s clearly in love with this company and is on board.” At the same time, you don’t want this person to be like Ben Stein in Ferris Bueller, where he’s just like, “Bueller. Bueller. Bueller.” [R7-3] Respondents’ recommendations for delivery style are consistent with their discussion of reporting goals, suggesting that managers should generally communicate positivity, but encouraging moderation in action-focused reporting goals and in vocal cues. However, respondents also indicate that nonverbal cues can present legal risks in private communications, as regulators believe they can communicate material nonpublic information to investors. Under RegFD, it’s not just what you say, it’s how you say it. So, if you’re sitting across the table from somebody and…you might tell everybody that it’s going to be a good quarter on the call, but you roll your eyes at an investor meeting as you say it, you just communicated something to that person that could be material that is now…[laughs] very…it could get you in a lot of trouble. Because you’ve now communicated something to them that’s different from what you’ve communicated to everybody else who couldn’t see you in person. [R6-3] We tried to be as careful on that as we could. There was an FD case on nonverbal cues that our general counsel liked to bring up where the investors met with the CEO, and the CEO was – I don’t remember if he was really positive or really negative that day and they went out and did a transaction on the stock and caused all kinds of problems. [R1-3] So, quiet period, you try to stay away from that because everything is being analyzed, and sometimes, in certain places I’ve worked you would only answer via email so there couldn’t be any interpretation of tone of voice of body language or anything. [R5-3] 34 Responses collectively confirm that managers believe investors are attentive to vocal cues, and provide additional support for the notion that managers vary their vocal cues based on reporting goals. Response to Experimental Scenario While respondents largely agree about the general tone that managers should convey when communicating with investors, responses about the scenarios in the main experiment are more varied. Some participants recommend that managers use the same tone when persuading prospective investors to invest and when persuading current investors to maintain their position. You can’t – one can’t be realistic, and the other one can’t be a Hulk, right, whatever it is. You’re doing yourself a disservice, again, because you have to assume they both have access to the same amount of information. [R7-4] I think I’d want them to be just upbeat in both calls. Obviously, you’re exactly right that what they would say would be different between those two calls. But you’d want the same sort of positive, upbeat tone and trying to sell the business as much as you can in either situation. [R8-4] When persuading prospective investors to invest in the firm, some participants recommend that the manager use an upbeat, excited delivery style. In contrast, others advise against trying to “sell” the stock, arguing that persuading investors to invest in the firm is different than marketing a product. When you’re talking about the future things and the upside and you’re being more boisterous – I mean, I think that can definitely…I think we do that as humans naturally because that does create more positivity. [R1-4] I’ll walk into a meeting – not a lot, but occasionally, and someone will be like, “Alright, give me your ninety second pitch on [Company Name],” and I’ll just reframe the whole conversation. I’m like, “Look, I’m not here to sell you anything. I’m here to answer your questions, but I’m not here to pitch you on anything. I’m happy to talk to you about our strategy, our results, where we’re taking the company, where we’ve come from. But I’m not here to ‘pitch’ you on our stock. I’m 35 here to educate you and let you make the decision that you think is best as an investment professional.” And clearly, based on the number of times I get that, there are companies that walk in and they’re pitching their story to an investor. [R10-4] Consistent with the findings in my experiment, when persuading current investors to maintain their position in the firm, respondents generally recommend using a more calm, professional style. I would remind them, like, “Listen, they are giving us a chance. They’re not redeeming like most investors do without even the opportunity. So, we should be very respectful of that, and answer very calm, and just try to keep it very professional, because that’s what they’re going to want to hear. If there’s any chance of keeping them it’s going to be this opportunity. This is the opportunity.” [R4-4] The reality is that, sometimes, it’s like, some twenty-five-year-old kid who’s never run a business or really done anything, and you know that, […] and you kind of want to step back and say, “Dude, this is a lot harder than you realize. You don’t know X, Y, and Z.” But you can’t – [laughs] you can’t really come out and say that. And it’s sort of hard to handle some of that stuff sometimes. But like I said, just do your best to maintain your composure and not… just stay as professional as you can. [R8-4] A common theme in these responses is the need to express empathy when persuading current shareholders not to sell, suggesting empathy as a potential mechanism for the effects I observe. At the end of the day, behind every algorithm is still a human being. So, there’s still someone pressing the button, making these sorts of decisions for a lot of these big, moving funds. So, you have to be…you have to acknowledge that individual. So, there is some sort of emotive feeling when things are going badly, no matter what the computer tells you to do. [R7-4] Well, look, let’s say for instance they’re down 15%. I think it’s important to be empathetic and take responsibility for what happened. And tell them how you believe things will improve. I think it’s a level of empathy that’s required. [R9-4] 36 While responses suggest that there is variation in how managers are advised to communicate with prospective investors, IROs generally recommend a calmer, more professional delivery style when persuading current investors not to sell, suggesting that preparation undertaken by managers for communications with investors is likely to reinforce the effects observed in my experiment. Management of Nonverbal Behaviors in Financial Communications Not surprisingly, in light of the importance of vocal cues to investors and regulators, respondents describe a variety of ways that firms manage nonverbal behaviors. Several indicate that executives receive coaching, ranging from informal feedback to formal presentation training. So, we manage every aspect of external communication. We sometimes laugh that we spend a lot of time on the script and on the press-release turning words from, like, “happy” to “glad.” So, a lot of it seems unimportant but the overall – but it is at the end of the day very important because it’s tone […] and it’s how you say it, it’s the tone of voice that you say it in, it’s the “glad” versus the “happy” word, and all of those pieces need to come together during that earnings release preparation process. [R3-1] We actually had somebody come in and spend an entire day with our executives – CEO, CFO, myself, and a couple other folks who were going to be deemed able to speak to investors, and have them do some communication stuff with us. For how to stand up in front of people, how to hold your hands, how to be consistent in your vocal inflection, and things like that. You know, basically what you do when you’re training someone to be a news reporter, or something like that. [R6-3] Other strategies for improving managers’ nonverbal behaviors include scheduling meetings with important investors at times when the manager has more energy and even manipulating audio files for prerecorded disclosures. We tend to be very, very strategic, in terms of how we move meetings. So, I want to make sure that if I have to deal with major shareholders, I want them first thing in the morning, or I want them right after lunch […] because I don’t want at three- 37 thirty in the afternoon, when they’ve got two more of these to go and they’ve been through ten, that they’re dealing with a major shareholder. [R7-3] It was either myself or one of my colleagues would sit with the executive while they were reading the script. And like I said, they would do it two times in case there was something that maybe just came across choppy in the audio […] and if something sounded weird, or there was a pause, or any sort of issue, we could just decide to swap out – we could use one of the two recordings and move things around a bit, and speed things up or slow things down. [R8-1] Respondents note that interpersonal factors can make managers’ nonverbal behaviors more difficult to manage in communications with investors. I will tell you this from my experience, it’s sometimes hard to keep your composure and stay as nice and professional as you can be when someone’s not giving you that – doing that, the other way. Like, if someone’s yelling at you or trying to get information that all parties know is nonpublic, it’s sort of hard to keep your patience. [R8-3] It’s really in those instances at conferences, at earnings releases, or any other event where there might be an investor Q&A – that’s where the bated breath comes into play, because you never know how the question is going to be asked. It could be the same question, depending on how it’s asked, could come across as confrontational or hopeful. And then you have to…the response to that needs to be appropriate, regardless of the way it was asked. [R3-1] Some respondents note that managers have individual styles, and that for some managers, trying to change these styles is either difficult or inadvisable. I think the answer to your question depends a lot on the personality that you are trying to manage. Because you want the personality to come across, but you want it to come across in a productive way. So, in some instances, that’s easy to do. And in some instances, it’s actually impossible. [R3-3] The job is 24/7, 365. There is no downtime. So, it would be tough, I think, to be like, “Alright, we’re going to hire a consultant to teach you how to talk to investors.” I’m sure companies do it, and I’m sure they get convinced it’s a good thing. But part of me is like, “Look, you have that skillset, or you don’t.” [R10-3] 38 Consistent with this view, several respondents note that they consider managers’ personalities and presentation styles when arranging meetings with investors. There are CFOs I can think of off the top of my head where I’m like, “I think they’re really skilled at being a CFO, but they actually were not skilled at all talking to investors.” So, they actually have somebody do it in their place. I have no idea if they tried to get that person more comfortable with it, or if the CEO, and the board, and the executive management team said, “Well, it’s just not important. We have somebody who can do that job fine. So, let this person focus on their strength.” [R10-3] We have our chief compliance officer who is a very dry and studied guy who goes in with pages of notes. He is not who I would put on every presentation, but we had a presentation two weeks ago with an international investor that is super focused on risk, and they loved him. […] I mean, he’s our chief compliance officer. He’s just…that’s what he does. And he’s not a flashy presentation guy. But that’s what they needed. That’s what they wanted. [R2-3] 39 SECTION 6 CONCLUSION Research in accounting has found that managers’ vocal cues contain value- relevant information. Using a controlled experiment, I provide evidence on the determinants of these cues. Specifically, I find that managers vary their vocal cues based on their reporting goals in communications with investors. Relative to their normal conversational style, I find that managers increase their pitch, pitch variation, volume, volume variation, and speech rate when persuading prospective investors to invest in the company. These results are consistent with prior research examining how people persuade others to take an action. However, I also find that managers use different vocal cues for different reporting goals. Relative to when persuading prospective investors to invest, I find that managers use lower pitch and pitch variation when persuading current investors not to sell their investment. These findings are consistent with the predictions of regulatory fit theory, with managers varying their vocal cues to convey an eager delivery style for action-focused persuasion goals and a vigilant delivery style for inaction-focused goals. Finally, in semi-structured interviews with investor relations professionals, I provide rich, descriptive evidence on managers’ reporting goals, how firms and capital market participants consider vocal cues and other nonverbal behaviors in financial communications, and the methods firms use to manage nonverbal communication. My study adds to a growing body of literature on nonverbal behavior and vocal cues in financial communications. Research in accounting has shown that investors 40 react to managers’ physical nonverbal behaviors (Elliott et al. Sedor 2012; Blankespoor et al. 2017; Cade et al. 2020) and vocal cues (Mayew and Venkatachalam 2012; Mayew et al. 2020). However, little is known about the determinants of these behaviors. While prior research has examined whether managers unintentionally produce vocal cues as a result of their emotional state (Mayew and Venkatachalam 2012; Hobson et al. 2012), I provide evidence that managers use these cues to advance their reporting goals. These results add to our understanding of the determinants of managers’ nonverbal behaviors, which may help understand previous findings and guide future research on nonverbal behavior in financial communications. My study also contributes to research on managers’ reporting objectives. As noted by Verrecchia (2001), early research on discretionary disclosure assumed that a manager’s objective is to maximize the current market capitalization of their firm. However, managers may have other objectives, due to equity incentives (Aboody and Kasznik 2000) or competitive and regulatory pressures (Bhojraj et al. 2004). Analytical research suggests that uncertainty about managers’ reporting objectives can facilitate reporting bias (Fischer and Verrecchia 2000) or nondisclosure (Einhorn 2007) because investors are unable to anticipate and back out these choices. Archival research on strategic reporting faces challenges in that managers’ intent is unobservable and disclosure choices may not be directly attributable to the manager, given the influence of other parties on disclosures (Amel-Zadeh et al. 2019). My results suggest that managers’ vocal cues may be useful for researchers and capital market participants attempting to infer managers’ reporting objectives. 41 My findings also have practical implications. For managers, my experiment and semi-structured interviews collectively suggest that vocal cues may be useful for achieving reporting goals, and that a better understanding of these cues could help managers communicate more effectively. Specifically, managers may consider varying their vocal cues based on the patterns I identify when trying to persuade others to action or inaction. Given evidence that analysts and investors scrutinize these cues (Cohen and Malloy 2012), my findings may also help managers understand how these signals will be interpreted. For analysts and investors, my study suggests that the ability to observe managers’ vocal cues may be a double-edged sword. On the one hand, these cues may provide useful signals of managers’ reporting goals. However, if managers are varying their vocal cues to increase persuasiveness, analysts and investors should consider the potential influence of these cues on their judgment and decision making. My study is not without limitations. First, I use the setting of private calls between a manager and an analyst to ensure that a single reporting goal is active in each response. Given prior research showing that audience size can affect managers’ disclosure choices (Durney 2019), it is possible that vocal cues may differ in settings with larger audiences. Second, participants in my experiment respond using a script, rather than providing a spontaneous response. While holding the language constant provides a cleaner test of the effects of reporting goals on vocal cues, it is possible that my results may not generalize to more spontaneous responses. While there is evidence of phonetic differences between scripted and spontaneous speech (Wagner et al. 2015) and syntactic differences between scripted and unscripted responses in earnings calls 42 (Lee 2016), I examine the effects of reporting goals on vocal cues, and it is unclear how these effects would systematically differ between scripted and spontaneous speech. However, firms engage in extensive scripting of public disclosures, including responses to questions asked by analysts in earnings calls (Brown et al. 2019; Bamber and Abraham 2019). Therefore, at a minimum, my results generalize to these types of communications. Finally, I operationalize the investors in my study using recordings from voice actors, rather than confederates or a separate group of participants. Although this choice provides additional control to strengthen causal inferences, it sacrifices some of the richness of a more interactive setting (Kachelmeier 2018). Future experimental and archival research could test the generalizability of these findings in less-controlled, more interactive settings. My study suggests several promising areas for future research in accounting. First, while my experiment demonstrates that managers systematically vary their voices based on their reporting goals, it is unclear whether these changes affect investor judgments. Future research could test these effects by experimentally varying manager vocal cues, either by selecting recordings which differ on these cues or by synthetically manipulating them in Praat (Boersma and Weenink 2019). Future research could also examine how managers vary their vocal cues when discussing other issues, such as earnings restatements or asset impairments, and how these cues interact with the linguistic features studied in the accounting literature. Finally, while my study focuses on communications with investors, future research could explore how managers vary their vocal cues when communicating with other parties. For example, research in auditing has examined how auditors react to managers’ 43 nonverbal cues associated with deception (Bennett and Hatfield 2018). Future research could examine whether managers vary their vocal cues when trying to persuade auditors that an aggressive estimate is reasonable. While a shift from face-to-face to computer-mediated communication may reduce auditors’ ability to “read” their client’s nonverbal cues, it may insulate them from the potentially harmful influence of these cues. Audio data from earnings calls, road shows, and other management presentations are already being used by practitioners (Cohen and Malloy 2012) and are increasingly available to researchers (Mayew and Venkatachalam 2013; Teoh 2018). In this study, I introduce techniques and best practices from the phonetics literature for the processing and analysis of acoustic data, demonstrate how this data can be elicited in a controlled experimental setting, and provide causal evidence on the determinants of manager vocal cues in financial communications. 44 APPENDIX A – AUDIO PROMPT AND RESPONSE SCRIPTS This appendix presents the scripts used in the experiment. The scripts in Panels A, B, and C were provided to voice actors to create the audio prompts for each recording. Panel A presents the script for the voice actor playing the IRO, used as the prompts Practice condition. Panels B and C present the scripts given to two voice actors playing analysts, used as prompts for the Action and Inaction Goal conditions, with the ordering of the two actors counterbalanced across participants. Panel D presents the script used by participants for all three recordings. Panel A. Practice Condition Script “Ok, before you talk to investors, let’s practice your response. Even though you’re using a script, it’s important to sound unscripted when responding to questions. So, for this practice, just focus on reading the response in a conversational way. When you’re ready, try reading it out loud.” Panel B. Analyst Script 1 “Hey, thanks for taking my call. Just a couple questions. First, on the Vital Brewing rollout, I know it's still early and you highlighted some of the challenges in the press release, but can you talk a bit more about what’s driving the first quarter performance? And following up on that, you’ve talked about the opportunities you see with Vital. In light of the soft Q1 results and a competitive industry backdrop, can you give a bit of color on how to think about the full year outlook?” Panel C. Analyst Script 2 “Hey, appreciate you taking the time. Just a few questions, really focusing on the Vital Brewing expansion. This is the first quarter where we’re seeing it hit the financials, so I was hoping you could give a bit of clarity on some of the challenges in the quarter? And this is related, but the Q1 results for Vital were pretty similar to industry trends. Can you just give some context on how we should think about opportunities with the brand and how that relates to the full-year outlook?” Panel D. Response Script "Sure, I’m happy to speak to that. So, in the first quarter, we definitely saw some challenges with Vital Brewing. There were some unexpected headwinds with the rollout, and those showed up in our first quarter earnings. You know, with any new brand, there are some growing pains. We’re ramping up production at Vital, and that’s honestly been a difficult transition. So, our first quarter results were a bit worse than expected, but I would think about this as a temporary adjustment. Now, looking at the full year, we’re excited about the Vital brand. We’re seeing big opportunities for growth, which is why we’ve raised our full year guidance. Now that we’ve increased production at Vital, we’re ready to start competing aggressively in our target markets. We've already made some investments to build brand awareness, and the early signs are very promising. Overall, we built a strong foundation in the first quarter, and Galaxy is well-positioned for growth for the rest of the year." 45 APPENDIX B – EXPERIMENTAL MANIPULATIONS This appendix presents descriptions provided to participants in the experiment. Panel A presents the analyst in the Action Goal condition. In addition to the information below, participants are told that their goal is to persuade the investor “to make an investment in Galaxy” and to respond to his question in a way that “plays up the positive full-year outlook.” Panel B presents the analyst in the Inaction Goal condition. In addition to the information below, participants are told that their goal is to persuade the investor “not to sell his investment in Galaxy” and to respond to his question in a way that “downplays the weak performance in the first quarter.” The ordering of the investor name, investment firm, and image are held constant. However, I counterbalance the ordering of all information related to the manipulation. Panel A. Action Goal Condition Panel B. Inaction Goal Condition 46 APPENDIX C – VARIABLE DESCRIPTIONS Variable Description Mean of fundamental frequency in hertz (Hz), using pitch range settings of 75-250Hz for male speakers and 100-300Hz for female Mean Pitch speakers. Frequency observations outside 2.5 median absolute deviations (Leys et al. 2013) excluded. Standard deviation of fundamental frequency in hertz (Hz), using pitch range settings of 75-250Hz for male speakers and 100-300Hz for St. Dev. Pitch female speakers. Frequency observations outside 2.5 median absolute deviations (Leys et al. 2013) excluded. Mean of Intensity in decibels (dB) of the voiced portions of Mean Volume recordings. Standard deviation of Intensity in decibels (dB) of the voiced portions St. Dev. Volume of recordings. Words per minute, calculated by dividing the number of words in the Speech Rate script (held constant at 164) by the duration of the recording. 47 APPENDIX D – PREPARED QUESTIONS FOR SEMI-STRUCTURED INTERVIEWS Question 1 How do you prepare managers for communicating with investors? How do you think about the specific audience when preparing managers to Question 2 communicate with investors? Leaving aside what the executive is going to say, how do you think about delivery Question 3 style when speaking with investors? Imagine your CEO is about to take two phone calls: one from a current shareholder who is considering selling their shares, and one from a prospective investor who is Question 4 considering investing in the company. Again leaving aside what they should say, do you advise them differently on how to say it in these two calls? 48 REFERENCES Aboody, D., & Kasznik, R. (2000). CEO stock option awards and the timing of corporate voluntary disclosures. Journal of Accounting and Economics, 29(1), 73-100. Amel-Zadeh, A., Scherf, A., & Soltes, E. F. (2019). Creating firm disclosures. Journal of Financial Reporting. Anilowski, C., Feng, M., & Skinner, D. J. (2007). Does earnings guidance affect market returns? The nature and information content of aggregate earnings guidance. Journal of Accounting and Economics, 44(1-2), 36-63. Asay, H. S., Libby, R., & Rennekamp, K. (2018). Firm performance, reporting goals, and language choices in narrative disclosures. Journal of Accounting and Economics, 65(2-3), 380-398. Baayen, R. H., Davidson, D. J., & Bates, D. M. (2008). Mixed-effects modeling with crossed random effects for subjects and items. Journal of Memory and Language, 59(4), 390-412. Bachorowski, J. A., & Owren, M. J. (2008). Vocal expressions of emotion. Handbook of Emotions, 3, 196-210. Bamber, M., & Abraham, S. (2019). On the “Realities” of Investor‐Manager Interactivity: Baudrillard, Hyperreality, and Management Q&A Sessions. Contemporary Accounting Research, 37(2), 1290-1325. Bennett, G. B., & Hatfield, R. C. (2018). Staff auditors' proclivity for computer- mediated communication with clients and its effect on skeptical behavior. Accounting, Organizations and Society, 68, 42-57. Bhojraj, S., Blacconiere, W. G., & D'Souza, J. D. (2004). Voluntary disclosure in a multi-audience setting: An empirical investigation. The Accounting Review, 79(4), 921-947. 49 Blankespoor, E. (2018). Firm communication and investor response: A framework and discussion integrating social media. Accounting, Organizations and Society, 68, 80-87. Blankespoor, E., deHaan, E., & Marinovic, I. (2020). Disclosure processing costs, investors’ information choice, and equity market outcomes: A review. Journal of Accounting and Economics, 101344. Blankespoor, E., Hendricks, B. E., & Miller, G. S. (2017). Perceptions and price: Evidence from CEO presentations at IPO roadshows. Journal of Accounting Research, 55(2), 275-327. Boersma, P., & Weenink, D. (2019). Praat: doing phonetics by computer (Version 6.0. 52)[Windows]. Brown, L. D., Call, A. C., Clement, M. B., & Sharp, N. Y. (2015). Inside the “black box” of sell‐side financial analysts. Journal of Accounting Research, 53(1), 1- 47. Brown, L. D., Call, A. C., Clement, M. B., & Sharp, N. Y. (2016). The activities of buy-side analysts and the determinants of their stock recommendations. Journal of Accounting and Economics, 62(1), 139-156. Brown, L. D., Call, A. C., Clement, M. B., & Sharp, N. Y. (2019). Managing the narrative: Investor relations officers and corporate disclosure. Journal of Accounting and Economics, 67(1), 58-79. Burgoon, J. K., Birk, T., & Pfau, M. (1990). Nonverbal behaviors, persuasion, and credibility. Human Communication Research, 17(1), 140-169. Burgoon, J. K., Guerrero, L. K., & Floyd, K. (2016). Nonverbal Communication. Routledge. Cade, N. L. (2018). Corporate social media: How two-way disclosure channels influence investors. Accounting, Organizations and Society, 68, 63-79. 50 Cade, N. L., Koonce, L., & Mendoza, K. I. (2020). Using video to disclose forward- looking information: the effect of nonverbal cues on investors’ judgments. Review of Accounting Studies, 25(4), 1444-1474. Cesario, J., & Higgins, E. T. (2008). Making message recipients “feel right” how nonverbal cues can increase persuasion. Psychological Science, 19(5), 415- 420. Cesario, J., Higgins, E. T., & Scholer, A. A. (2008). Regulatory fit and persuasion: Basic principles and remaining questions. Social and Personality Psychology Compass, 2(1), 444-463. Cianci, A. M., & Falsetta, D. (2008). Impact of investors’ status on their evaluation of positive and negative, and past and future information. Accounting & Finance, 48(5), 719-739. Clor‐Proell, S. M., & Maines, L. A. (2014). The impact of recognition versus disclosure on financial information: A preparer's perspective. Journal of Accounting Research, 52(3), 671-701. Cohen, L., & Malloy, C. (2011). Business Intelligence Advisors (BIA), Inc. Finding the hidden meaning in corporate disclosures. Harvard Business School Case, 212-031. Durney, M. T. (2019). The Effect of Audience Size on Managers’ Private Disclosures. Available at SSRN: https://ssrn.com/abstract=3469816 or http://dx.doi.org/10.2139/ssrn.3469816 Einhorn, E. (2007). Voluntary disclosure under uncertainty about the reporting objective. Journal of Accounting and Economics, 43(2-3), 245-274. Ekman, P., & Friesen, W. V. (1969). Nonverbal leakage and clues to deception. Psychiatry, 32(1), 88-106. Elliott, W. B., Hodge, F. D., & Sedor, L. M. (2012). Using online video to announce a restatement: Influences on investment decisions and the mediating role of trust. The Accounting Review, 87(2), 513-535. 51 Fischer, P. E., & Verrecchia, R. E. (2000). Reporting bias. The Accounting Review, 75(2), 229-245. Fraccaro, P. J., O'Connor, J. J., Re, D. E., Jones, B. C., DeBruine, L. M., & Feinberg, D. R. (2013). Faking it: deliberately altered voice pitch and vocal attractiveness. Animal Behaviour, 85(1), 127-136. Freeman, V. (2014). Hyperarticulation as a signal of stance. Journal of Phonetics, 45, 1-11. Goudbeek, M., & Scherer, K. (2010). Beyond arousal: Valence and potency/control cues in the vocal expression of emotion. The Journal of the Acoustical Society of America, 128(3), 1322-1336. Guyer, J. J., Fabrigar, L. R., & Vaughan-Johnston, T. I. (2019). Speech rate, intonation, and pitch: Investigating the bias and cue effects of vocal confidence on persuasion. Personality and Social Psychology Bulletin, 45(3), 389-405. Hall, J. A. (1980). Voice tone and persuasion. Journal of Personality and Social Psychology, 38(6), 924. Hall, J. A., Horgan, T. G., & Murphy, N. A. (2019). Nonverbal communication. Annual Review of Psychology, 70, 271-294. Hamilton, E. L., & Winchel, J. (2019). Investors' processing of financial communications: A persuasion perspective. Behavioral Research in Accounting, 31(1), 133-156. Harris, L. L., Hobson, J. L., & Jackson, K. E. (2016). The effect of investor status on investors' susceptibility to earnings fixation. Contemporary Accounting Research, 33(1), 152-171 Hellbernd, N., & Sammler, D. (2016). Prosody conveys speaker’s intentions: Acoustic cues for speech act perception. Journal of Memory and Language, 88, 70-86. Higgins, E. T. (2000). Making a good decision: value from fit. American Psychologist, 55(11), 1217. 52 Higgins, E. T. (2005). Value from regulatory fit. Current Directions in Psychological Science, 14(4), 209-213. Hobson, J. L., Mayew, W. J., & Venkatachalam, M. (2012). Analyzing speech to detect financial misreporting. Journal of Accounting Research, 50(2), 349-392. Hughes, S. M., Mogilski, J. K., & Harrison, M. A. (2014). The perception and parameters of intentional voice manipulation. Journal of Nonverbal Behavior, 38(1), 107-127. Jiang, X., & Pell, M. D. (2017). The sound of confidence and doubt. Speech Communication, 88, 106-126. Kachelmeier, S. J. (2018). Testing auditor-client interactions without letting auditors and clients fully interact: Comments on Bennett and Hatfield (2018). Accounting, Organizations and Society, 68, 58-62. Ladefoged, P. (2003). Phonetic Data Analysis: An Introduction to Fieldwork and Instrumental Techniques. Wiley-Blackwell. Ladefoged, P., & Johnson, K. (2014). A Course in Phonetics. Nelson Education. Lee, J. (2016). Can investors detect managers' lack of spontaneity? Adherence to predetermined scripts during earnings conference calls. The Accounting Review, 91(1), 229-250. Lee, A. Y., & Aaker, J. L. (2004). Bringing the frame into focus: the influence of regulatory fit on processing fluency and persuasion. Journal of Personality and Social Psychology, 86(2), 205. Leys, C., Ley, C., Klein, O., Bernard, P., & Licata, L. (2013). Detecting outliers: Do not use standard deviation around the mean, use absolute deviation around the median. Journal of Experimental Social Psychology, 49(4), 764-766. Li, F. (2008). Annual report readability, current earnings, and earnings persistence. Journal of Accounting and Economics, 45(2-3), 221-247. 53 Li, F. (2010). Textual analysis of corporate disclosures: A survey of the literature. Journal of Accounting Literature, 29, 143-165. Libby, R., Bloomfield, R., & Nelson, M. W. (2002). Experimental research in financial accounting. Accounting, Organizations and Society, 27(8), 775-810. Libby, R., & Emett, S. A. (2014). Earnings presentation effects on manager reporting choices and investor decisions. Accounting and Business Research, 44(4), 410- 438. Loughran, T., & McDonald, B. (2016). Textual analysis in accounting and finance: A survey. Journal of Accounting Research, 54(4), 1187-1230. Mayew, W. J., Parsons, C. A., & Venkatachalam, M. (2013). Voice pitch and the labor market success of male chief executive officers. Evolution and Human Behavior, 34(4), 243-248. Mayew, W. J., Sethuraman, M., & Venkatachalam, M. (2020). Individual Analysts' Stock Recommendations, Earnings Forecasts, and the Informativeness of Conference Call Question and Answer Sessions. The Accounting Review, 95(6), 311-337. Mayew, W. J., & Venkatachalam, M. (2012). The power of voice: Managerial affective states and future firm performance. The Journal of Finance, 67(1), 1- 43. Mayew, W. J., & Venkatachalam, M. (2013). Speech analysis in financial markets. Foundations and Trends in Accounting, 7(2), 73-130. Mehrabian, A., & Williams, M. (1969). Nonverbal concomitants of perceived and intended persuasiveness. Journal of Personality and Social Psychology, 13(1), 37. Mercer, M. (2004). How do investors assess the credibility of management disclosures?. Accounting Horizons, 18(3), 185-196. Miller, N., Maruyama, G., Beaber, R. J., & Valone, K. (1976). Speed of speech and persuasion. Journal of Personality and Social Psychology, 34(4), 615. 54 Mitchell, R. L., & Ross, E. D. (2013). Attitudinal prosody: What we know and directions for future study. Neuroscience & Biobehavioral Reviews, 37(3), 471-479. Nelson, M. W., Elliott, J. A., & Tarpley, R. L. (2002). Evidence from auditors about managers' and auditors' earnings management decisions. The Accounting Review, 77(s-1), 175-202. Price, S. M., Seiler, M. J., & Shen, J. (2017). Do investors infer vocal cues from CEOs during quarterly REIT conference calls?. The Journal of Real Estate Finance and Economics, 54(4), 515-557. Rogers, J. L., & Stocken, P. C. (2005). Credibility of management forecasts. The Accounting Review, 80(4), 1233-1260. Rogers, J. L., & Van Buskirk, A. (2013). Bundled forecasts in empirical accounting research. Journal of Accounting and Economics, 55(1), 43-65. Scherer, K. R. (2003). Vocal communication of emotion: A review of research paradigms. Speech Communication, 40(1-2), 227-256. Scherer, K. R. (2013). Vocal markers of emotion: Comparing induction and acting elicitation. Computer Speech & Language, 27(1), 40-58. Smith, S. M., & Shaffer, D. R. (1991). Celerity and cajolery: Rapid speech may promote or inhibit persuasion through its impact on message elaboration. Personality and Social Psychology Bulletin, 17(6), 663-669. Stocken, P. C. (2013). Strategic accounting disclosure. Foundations and Trends in Accounting, 7(4), 197-291. Svec, J. G., & Granqvist, S. (2010). Guidelines for selecting microphones for human voice production research. American Journal of Speech-Language Pathology, 19(4), 356-368. Teoh, S. H. (2018). The promise and challenges of new datasets for accounting research. Accounting, Organizations and Society, 68, 109-117. 55 Van Zant, A. B., & Berger, J. (2020). How the voice persuades. Journal of Personality and Social Psychology, 118(4), 661-682. Verrecchia, R. E. (2001). Essays on disclosure. Journal of Accounting and Economics, 32(1-3), 97-180. Vogel, A. P., Maruff, P., Snyder, P. J., & Mundt, J. C. (2009). Standardization of pitch-range settings in voice acoustic analysis. Behavior Research Methods, 41(2), 318-324. Vrij, A., Hartwig, M., & Granhag, P. A. (2019). Reading lies: nonverbal communication and deception. Annual Review of Psychology, 70, 295-317. Wagner, P., Trouvain, J., & Zimmerer, F. (2015). In defense of stylistic diversity in speech research. Journal of Phonetics, 48, 1-12. Weinstein, N., Zougkou, K., & Paulmann, S. (2018). You ‘have’ to hear this: Using tone of voice to motivate others. Journal of Experimental Psychology: Human Perception and Performance, 44(6), 898. Witz (2020). How Algorithm-Based Investing Affects Firm Disclosures and Decisions. Working Paper. Woodall, W. G., & Burgoon, J. K. (1983). Talking fast and changing attitudes: A critique and clarification. Journal of Nonverbal Behavior, 8(2), 126-142. Xu, Y. (2010). In defense of lab speech. Journal of Phonetics, 38(3), 329-336. 56