Tone at the top, corporate irresponsibility and the Enron emails

Kieran Taylor-Neu (Department of Accounting, University of Alberta, Alberta, Canada)
Abu S. Rahaman (Accounting Area, Schulich School of Business, York University, Toronto, Canada)
Gregory D. Saxton (Accounting Area, Schulich School of Business, York University, Toronto, Canada)
Dean Neu (Accounting Area, Schulich School of Business, York University, Toronto, Canada)

Accounting, Auditing & Accountability Journal

ISSN: 0951-3574

Article publication date: 10 October 2024

Issue publication date: 16 December 2024

757

Abstract

Purpose

This study aims to examine whether senior Enron executive emails celebrated, or at least left a space for, corporate irresponsibility. Engaging with prior organizational-focused research, we investigate how corporate emails sent by senior executives help constitute Enron by communicating to employees senior management’s stance about important topics and social characters.

Design/methodology/approach

The study analyzes the 527,356 sentences contained in 144,228 emails sent by Enron senior executives and other employees in the three-year period (1999–2001) before the company’s collapse. Sentences are used as the base-level speech unit because we are interested in identifying the tone and emotions expressed about specific topics and stakeholders. Tone is measured using Loughran and McDonald’s (2016) financial dictionary approach, and emotion is measured using Mohammad and Turney’s (2013) NRC word-emotion lexicon. Least Absolute Shrinkage and Selection Operator (LASSO) regressions are used to explore the determinants of senior management tone and emotions.

Findings

The analysis illustrates that while both senior executives and other employees utilized email to help accomplish task-related activities, they employed different evaluative tones to talk about key topics and stakeholders. Also important is what is left unsaid, with a “spiral of silence” emanating from senior management that created a space for corporate irresponsibility.

Originality/value

Combining advanced computerized textual analysis with qualitative techniques, we analyze a unique dataset to explore micro details involved in using email to communicate a tone at the top. The findings illustrate how what is said or not said by senior management contributes to the constitution of an organization.

Keywords

Citation

Taylor-Neu, K., Rahaman, A.S., Saxton, G.D. and Neu, D. (2024), "Tone at the top, corporate irresponsibility and the Enron emails", Accounting, Auditing & Accountability Journal, Vol. 37 No. 9, pp. 336-364. https://doi.org/10.1108/AAAJ-12-2023-6792

Publisher

:

Emerald Publishing Limited

Copyright © 2024, Kieran Taylor-Neu, Abu S. Rahaman, Gregory D. Saxton and Dean Neu

License

Published by Emerald Publishing Limited. This article is published under the Creative Commons Attribution (CC BY 4.0) licence. Anyone may reproduce, distribute, translate and create derivative works of this article (for both commercial and non-commercial purposes), subject to full attribution to the original publication and authors. The full terms of this licence may be seen at http://creativecommons.org/licences/by/4.0/legalcode


Introduction

The collapse of Enron remains a pivotal event in global corporate history – an event that continues to raise critical questions about the interplay between corporate governance, ethics and transparency. Over two decades after its dramatic failure in 2001, analysts continue to debate how such a debacle was possible (Iren and Kim, 2023; Penikis et al., 2023). As Arnold and de Lange observed (2004, p. 751), the overarching reasons behind Enron's collapse may never be fully revealed, a testament to the complexity of large corporate systems. This unresolved puzzle continues to captivate scholars, urging a deeper exploration of the role of governance, transparency and leadership in corporate malfeasance.

The current study seeks to advance our understanding by examining the “tone at the top” (Gunz and Thorne, 2015) communicated by senior executive emails. Building on prior organizational-focused research, the current study has four motivations. First, following Jonsson’s (1998, p. 411) appeal for researchers to appreciate that “managers work with words” (Craig and Amernic, 2004, p. 813), the study considers whether the evaluative tone used by Enron’s senior managers in their emails (i.e. words or silence) condoned irresponsible corporate behavior. Did the email commentary by senior managers leave a space for the trading department to keep extensive and detailed double sets of accounting records? Did the emails celebrate generating profit at all costs? And did the emails acknowledge that there were legal and ethical constraints on corporate behavior? In effect, did the tone at the top go beyond “words” to create a spiral of silence as a corporate communication strategy?

The second motivation is to better understand whether the tone at the top changed depending on the size of the internal audience and what was happening to the company at that moment in time. Recent research on professional associations suggests that the tone at the top is attuned to the characteristics of the internal audience and that there may be shifts in emphasis and tone depending on both the target audience and external events (cf. Goodrick et al., 2020; Neu and Saxton, 2023). These findings motivate us to consider whether the tone used by senior executives changed over time as the company neared collapse. It also encourages us to examine whether the tone used varied depending on the size of the audience and, hence, the publicness of the utterance. Building upon previous research (cf. Braun et al., 2019; Markus, 1994), we assume that the tone used may vary depending on whether the email is being sent to one or two colleagues or is addressed to a much larger internal audience.

Third, the study is motivated to examine the emotions expressed in senior management emails and whether these emotions change over time. Whereas tone reflects an evaluative opinion, emotion is a conceptually distinct form of communication that involves the expression of a subjective evaluation about the speech topics and social actors (see Goodrick et al., 2020; Bakhtin, 2010). Previous linguistic-focused research (Bakhtin, 1994, pp. 167–168) proposes that utterances are directed at specific organizational audiences and potentially reflect how senior executives feel about these topics and social actors. At the same time, prior institutional research reminds us that speech communication can be used strategically to elicit specific audience emotions (Goodrick et al., 2020, pp. 747–748; Jarvis et al., 2019, p. 1360; Edgar et al., 2022, p. 46; Riaz et al., 2023) and that the suppression of emotion by the speaker can, itself, be strategic (Jarvis et al., 2019, p. 1361). For example, in their analysis of how language is used to convey profit warnings, Edgar et al. (2022, p. 28) identify three distinct stages: “denials, defiance, and desperation.” They suggest that these stages, seen as “communication artifacts,” can be strategically employed based on management’s intentions and emotional state (cf. Riaz et al., 2023). Similarly, Postaire and Puyou (2024, p. 840) propose that “accounting documents prompt emotional reactions” and provide “an affective technology” for senior executives to elicit desired emotions from organizational stakeholders (see also Roszkowska-Menkes et al., 2024). The Enron email corpus provides us with an opportunity to examine whether the emotions expressed in senior executive utterances paralleled the company’s financial rollercoaster as it went from hitting a stock price high in December 2000 to bankruptcy in December 2001 and whether these emotions led or lagged the emotions expressed by other employees.

Finally, we are motivated to understand whether what was left unsaid constituted an important part of the tone at the top. Specifically, we seek to understand whether senior executive silence on key topics was a deliberate management tool for constructing a “firewall” for maintaining organizational secrets, for curtailing the likelihood of whistleblowing and for creating a “spiral of silence” that opened a space in which corporate irresponsibility could take place (Noelle-Neumann, 1984; Noelle-Neumann and Petersen, 2004).

The email corpus that was seized by the Federal Energy Regulatory Commission (FERC) shortly after the failure provides us with a unique window into Enron’s internal corporate communications and is the primary dataset that we use for the analysis in the current study. The Enron email corpus is the only legal, publicly available and ethically useable archive of a public company’s email communications. Recent corporate scandals such as those involving Volkswagen, Odebrecht and Uber indicate that corporate emails may contain traces of both illegal activity and an irresponsible corporate mindset, but these email corpuses are not in the public domain (cf. Chavkin, 2019; Davies et al., 2022; Shepardson, 2016) [1]. For these reasons, the Enron email corpus provides us with a rare glimpse inside a corporation that was involved in illegal and irresponsible activities.

Using sentences as our unit of analysis and advanced computerized textual analysis, the current study seeks to identify the tones and emotions expressed about specific topics and stakeholders in the 527,356 sentences contained in 144,228 non-duplicate, original emails sent by Enron senior executives and other employees. We focus on the three-year period (1999–2001) before the company’s collapse. Our methodology uses R ‘s text processing module to prepare the data and measures tone and emotion through established financial and emotion dictionaries (cf. Loughran and McDonald, 2011; Mohammad and Turney, 2013). We employ LASSO regressions to analyze senior executive tone and emotions, complemented by qualitative analysis.

The analysis illustrates that while both senior executives and other employees utilized email to help accomplish task-related activities, the evaluative tone used to talk about a key external stakeholder—the FERC—and topics like deregulation, the market, Democrats and Republicans—differed. The analysis also shows that the emotions expressed by both senior executives and other employees changed in the periods leading up to the collapse of the energy giant, with the emotions expressed by senior executives having a slightly different pattern than for other employees. We also find that very little was said about the trading division’s activities, suggesting that on certain topics senior executive silence was a purposeful corporate communicative strategy designed to ensure organizational opacity and to diminish the likelihood of whistleblowing.

The rest of the paper is organized as follows. After providing background information on the Enron case, we turn to our theoretical framing, covering the concept of tone and its importance in modeling ethical leadership as well as how tone is diffused within organizations; here we also highlight key insights from Bakhtin (2010, 2013) along with the concept of silence from the communication and organizational behavior literature (Brinsfield, 2013; Noelle-Neumann, 1984). We then turn to our data and methods before presenting our findings. The final section presents our discussion and conclusions.

Background on the Enron collapse

More than 20 years after the failure of Enron in 2001, commentators are still asking how this could have happened. Enron’s bankruptcy sent shock waves through the US financial system and major equity markets around the world (Constable, 2021; Curwen, 2021; Silverstein, 2013; Thomas, 2002). The collapse was rapid and seemingly unexpected. At the end of 2000, Enron was the darling of Wall Street, and its shares were up 87% over the previous year and trading at 70 times earnings. In spring 2001, financial analysts started to question Enron’s accounting practices and the stock price came under increasing scrutiny (see Thomas, 2002). On August 14, 2001, CFO Jeffery Skilling publicly resigned and on October 16, 2001, it was announced that there would be massive restatements of the financial statements for the years 1997–2000. These restatements reduced earnings for the previous four-year period by $613 million (23% of reported profits for the period) and wiped out $1.2 billion (10%) in balance sheet equity (Healy and Palepu, 2003). Finally, on December 2, 2001, Enron filed for bankruptcy, ending the financial rollercoaster that went from a stock price high on December 31, 2000, to bankruptcy proceedings the following year [2].

In looking back at what had happened, Time magazine observed that the collapse of Enron eroded the trust of investors and effectively “changed American business forever” (Constable, 2021). In an earlier article on the collapse of the energy giant, Thomas (2002, p. 41) wrote that, “when a company looks too good to be true, it usually is.” Prior to its demise, Enron was one of North America’s largest corporations and was feted as a highly successful and visionary company (Healy and Palepu, 2003; Thomas, 2002) [3]. After its collapse, many commentators suggested that inadequate auditing and inadequate internal accounting (Kotb et al., 2020) as well as a corporate culture that valued risk-taking—and sometimes illegal activity—were contributing factors (Healy and Palepu, 2003; Sims and Brinkmann, 2003; Thomas, 2002). Subsequent analysis of Enron’s internal documents and congressional testimony tended to corroborate these claims, depicting Enron and its employees as acolytes of “cowboy capitalism” (Kolb, 2007) who kept double sets of accounting records to hide what they were doing (McCullough, 2002, 2004). Stories circulated about Enron energy traders holding the California energy market hostage with audio recordings of these same traders saying things like “let them freeze in the dark” and “burn baby, burn” (Roberts, 2004). Other media stories emphasized how individual and corporate greed resulted in thousands of retirees losing their life savings when Enron collapsed (Silverstein, 2013). These commentaries painted a picture of a company out of control and with an irresponsible senior management team that failed to rein in, and indeed celebrated, the pursuit of corporate and individual profit by any means (Constable, 2021; Curwen, 2021).

The collapse of Enron prompted research aimed at understanding the operational practices that led to the company's downfall and investigating the broader social consequences of its failure (Baker, 2003; Sridharan et al., 2002). Prior accounting research, for example, showed that Enron engaged in deceptive financial reporting practices, utilizing US Generally Accepted Accounting Principles to obscure the reality of its financial situation, effectively prioritizing form over substance (Baker and Hayes, 2004). In Baker and Hayes’ (2004) view, “had the concept of substance over form been applied at Enron, investors and creditors would have been provided with a more realistic view of the company’s financial position and its results of operations” (p. 767). In a subsequent paper, Baker and Hayes (2005, p. 5) propose that “astute financial analysis would have revealed the instability of Enron Corp. business model” and possibly could have averted the collapse of the company. Previous research in accounting has also emphasized the significant challenge faced by financial analysts in identifying and exposing the financial misconduct at Enron, particularly within its trading division (cf. Ohlrogge et al., 2024). This difficulty was primarily due to the substantial information imbalance and the self-serving actions of key figures, including top executives, which obscured the company's illegal activities from its stakeholders (Arnold and de Lange, 2004). Twenty years after the scandal, Enron “has left us all with the surreal feeling that the enormity of the ramifications may never be fully revealed” (Arnold and de Lange, 2004, p. 751).

Theoretical framing: tone at the top in executive emails

We now turn to our theoretical framing. In the first subsection, we review the literature on the tone at the top and its importance in organizing processes within organizations, and in the second section, we cover how tone can be communicated by what senior executives say and do not say in their emails to other employees.

Importance of tone at the top

The phrase “the tone at the top” draws attention to the importance of ethical leadership within organizations (Craig and Amernic, 2020, p. 109; Edgar et al., 2022, p. 32), a form of leadership that many commentators suggest is much needed within an increasingly complex and increasingly morally ambiguous world (Gunz and Thorne, 2015). Schwartz and colleagues, for example, cite the speech that then-President George W. Bush delivered regarding the corporate failures of Enron and WorldCom. In this speech, Bush states that “the scandals …. have hurt the stock market. And worst of all, they are hurting the millions of people who depend on the integrity of businesses for their livelihood and their retirement, for their peace of mind and their financial well-being” (Bush, quoted in Schwartz et al., 2005, pp. 79–80). Given this context, leadership involves communicating that ethics and social responsibility are core organizational values (Ribeiro et al., 2022).

In line with these ideas, we define the phrase “the tone at the top” in linguistic terms (cf. Bakhtin, 2013, pp. 84–85) as referring to an evaluative opinion by senior management about important topics and social actors that is communicated to internal and external audiences [4]. From this vantage point, the tone at the top is important because it frames (Arivdsson and Sabelfeld, 2023, p. 166) and impacts the activities of lower-level staff. Schwartz et al., for example, state that everyone “within the firm look to the top for guidance” (2005, p. 87). Similarly, Soltani (2014) cites the Treadway Commission who state that the “tone set by top management—the corporate environment or culture within which financial reporting occurs—is the most important factor contributing to the integrity of the financial reporting process” (quoted on p. 256). Rose and colleagues reiterate this vantage point, stating that previous “research demonstrates that the tone at the top is a key factor” that directs organizational behavior at the lower level (Rose et al., 2021, p. 295).

Unsurprisingly, the tone at the top is communicated through both words and actions. Brown et al., note that ethical leadership involves “the demonstration of normatively appropriate conduct through personal actions and interpersonal relationships, and the promotion of such conduct to followers through two-way communication” (2005, p. 59). Tone is a “nuanced and layered concept” that can be communicated via written documents such as the inclusion of a CEO letter in an annual report to shareholders, a sustainability report, or enshrined in action control tools such as formal corporate policies, procedures and codes of ethics (Amernic and Craig, 2013; Craig and Amernic, 2020; Fisher et al., 2020, p. 77). Tone can also be communicated through omission, what is not said by initiating and actively promoting a “spiral of silence” (Noelle-Neumann, 1984; Noelle-Neumann and Petersen, 2004). From this vantage point, leaving certain things unsaid may be a key management strategy when the stakes are high (Brinsfield, 2013).

Finally, tone can be communicated via the actions of senior executives who proactively communicate a positive tone via CSR disclosures and initiatives (Michelon and Parbonetti, 2012). In some cases, like the narratives in CEO letters to shareholders, tone can be strategically employed as a means for CEOs to “sustain self-legitimacy” (Craig and Amernic, 2020, p. 106). However, if senior executives’ actions are seen as inappropriate or unfair, they can communicate a negative tone which may undermine the company’s internal ethical culture and the external social responsibility reputation (cf. Kim and Choi, 2018). For instance, Schwartz et al. note how “Enron’s board of directors voted three times to suspend conflict of interest provisions in Enron’s code of ethics to permit CFO Andrew Fastow to establish and operate entities that transacted business with Enron” (2005, p. 85). It is these congruous and incongruous words and deeds that communicate the “ethical tone for the organization” (Soltani, 2014, p. 255).

The tone at the top is also crucial for defining the relationship between the organization and its wider stakeholders and thus for achieving meaningful corporate social responsibility (Cooper and Owen, 2007). Communications about external stakeholders are important because they signal and direct how employees should behave toward others (Cho et al., 2015a, b; Comoli et al., 2023; Deegan, 2019; Deegan and Rankin, 1996; Gray, 2006; Patten, 2020; Rahaman et al., 2004). However, if the tone at the top is silent on strategic relationships with key external stakeholders, this silence could risk antagonizing them and potentially call into question the legitimacy of the organization (Neu et al., 1998; Oliver, 1991). The literature suggests that strategic legitimacy is seen as “controllable,” making communication channels and managerial tone at the top crucial aspects of what some may refer to as corporate “self-laudatory” legitimation processes (Aerts and Cormier, 2008; see also Guthrie and Parker, 1989). Interdependency in this context, whether for resource dependency or legitimacy reasons places limits on corporate behaviors, including how organizational members speak about external stakeholders [5]. These limits involve, at minimum, not speaking too badly about external stakeholders. From this vantage point, a negative tone from the top regarding external stakeholders encourages lower-level employees to think, and possibly act, in a similar manner.

To conclude, previous research illustrates that senior executive communications disseminate the tone at the top to lower-level employees. This tone, which takes the form of both silence as well as evaluative and expressive utterances, matters because senior executive communications not only signal appropriate ways of speaking and acting but also limit and constrain the domain of acceptable behaviors. Sometimes these acceptable behaviors, emanating from the tone at the top, would suggest that employees simply remain silent about corporate activities, thus leaving no traces in any emails or other communication mediums. It is for these reasons that the tone at the top is important.

Email and the tone at the top

The current study complements and extends prior tone at the top research by drawing upon the ideas of Bakhtin (1994, 2010, 2013) and others to foreground the micro details involved in using email to communicate a tone at the top. In the remainder of this section, we employ our linguistic definition of tone at the top to highlight five salient aspects of senior executive email communications within organizations.

First, senior executives construct normative narratives for organizational audiences that are informed by and reflect their worldview and ethical stance (Craig and Amernic, 2020, p. 112). Narratives are constructed from the ground up since it is “value judgments” that determine “the very selection of the verbal material and the form of the verbal whole” (Bakhtin, 2010, p. 165). Normative narratives are assembled by packaging together different subjects and tones into an overall narrative (Bakhtin, 2013, pp. 117–123). This assembling allows the speaker to simultaneously develop a line of argument and communicate a worldview and ethical stance [6].

Previous organizational-focused research makes these same observations. Farias et al. (2021) suggest that management narratives from both loss-making and profitable companies involve impression management and context-specific storytelling (see also Moreno et al., 2019; Quinn et al., 2023; Hamza et al., 2023). These narratives focus on field-specific topics and social characters that convey the organization’s core values and interactions with key stakeholders (cf. García-Rosell, 2019; Neu et al., 2022). Winkler (2011, p. 654) notes that normative narratives are built around “particular understandings of individuals and groups, how they are inter-related and how they should behave.” These narratives impose “a logical structure on what might otherwise appear as unconnected, isolated events” (Preuss and Dawson, 2009, p. 138) and often contain a “moral to the story” (Farias et al., 2021, p. 129), influencing audiences’ perceptions of topics and characters (cf. Stevens, 2008). Communications from the top thus articulate an ethical stance (Kockelman, 2009) that impacts the views and actions of lower-level employees.

Second, evaluative opinions about topics and social characters are brought to life via intonation and rhetorical style; the latter often impacted by the anticipated level of scrutiny in the institutional environment (Aerts and Yan, 2017; Arivdsson and Sabelfeld, 2023). Here, we use “intonation” not just in its traditional linguistic sense of pitch variation in live speech, but more broadly to encompass the overall tone, choice of words and sentence structure that senior executives use to convey their worldview and ethical stance. Bakhtin (2013, p. 85), for example, proposes that intonation infuses a world view and ethical stance into otherwise neutral observations. Because of this, he notes that “intonation can be thoroughly understood only when one is in touch with the assumed value judgments” that underpin an utterance (2010, p. 165) [7]. As previous research shows, intonation is added to evaluative opinions by using adjectives, verbs and adverbs to tell a tone-infused normative narrative about a topic or social character (cf. Audi et al., 2016; Buchholz et al., 2018; Liu and Baker, 2016) [8]. In organizational research, Buchholz et al. (2018) propose that management narcissism is often communicated through an abnormally “optimistic tone” and that stakeholders need to be mindful of overly optimistic reporting language and tone from senior management.

Third, senior executive communication contains emotional content, but this content is ambiguous because communication is “double-faced” in that it is “oriented in two directions: with respect to the listener as ally or witness and with respect to the object of the utterance” (Bakhtin, 1994, p. 168). In its audience-focused face, utterances are brought to life not only by intonation, since it is a tone that “scolds, or caresses, denigrates or magnifies” (p. 168), but also by emotional content (p. 168). As an example, senior management’s communication on sustainability often involves evaluative opinions that reflect the socio-political context, civil society activism and other “participants” and are expressed in tones that are persuasive to these groups (Arivdsson and Sabelfeld, 2023). And in its topic-oriented face, emotion can also enter into the utterance conveying “the speaker's subjective emotional evaluation of the referentially semantic content” (Bakhtin, 2013, p. 84; see also Aerts and Yan, 2017, p. 414).

Extant research has picked up on these two faces of emotion and noted that the unfiltered expression of emotion by senior executives can undermine role norms because “corporations require emotional display and deep acting to sustain their self-definitions which are related to their goals” (Flam, 2002, p. 92). Similarly, senior executives use emotion, or its absence, to shape the cognitive and emotional investments of other organizational participants (Jarvis et al., 2019, p. 1382). Strategizing with emotions may require them to either express or suppress emotional content, depending on the circumstances and audience (Delmestri and Goodrick, 2016, p. 234). However, there are limits to this control. The expression of subjective emotions is an immediate affective reaction that sometimes precedes thought (Massumi, 1996, p. 228). Employees, bombarded with internal and external information, are likely to have affective reactions to events. Enron's financial troubles were front-page news from March to December 2001, leading to emotional reactions among both employees and senior executives due to the uncertainty and speculation regarding the consequences of the scandal (Thomas, 2002). This said, it remains unclear whether senior executives could control their emotions as expected or whether employee emotions changed over time.

Fourth, prior research suggests that the email topic as well as the size of the audience influences both the mode of communication and the message that senior executives communicate. Markus (1994), finds that senior executives and their assistants view email as “the primary medium of internal work-related communication, appropriate for any task” (p. 519), since emails have a large reach and save time. This said, Markus also notes that senior executives do not use email for confidential, private or delicate information nor for dealing with feelings or emotions (p. 519). To the list of these exceptions, we would also add that emails are unlikely to be used for morally unsavory or borderline illegal activities since email traces persist. In the case of the Volkswagen emissions scandal, senior Volkswagen executives attempted to destroy problematic emails and files (McLaughlin et al., 2017) and refused to provide internal email correspondence to regulators. Similarly, Enron traders used private cellphones to talk amongst themselves on the trading floor (Nix et al., 2021, p. 797). Following this prior research, we assume that the audience size mediates the tone used by senior executives since the larger the audience, the less private the conversation.

Fifth, prior organizational communication research suggests several key aspects of what is left unsaid in corporate emails (Bruneau, 1973; Brinsfield, 2013; Christian, 2020). To start, this silence can be strategic and can also take multiple forms: psycholinguistic, interactive or socio-cultural silence (see Bruneau, 1973 for a detailed discussion). And it is often subject to multiple interpretations, akin to nonverbal cues in verbal communicative interactions. For Bruneau, communicative silence does not mean that nothing is being said; the decision to speak or not can be dictated by the individual’s assessment of what action is expected or acceptable at a moment in time, including the prevailing rituals and tone at the top – what is often referred to as “silencing by authority” (Bruneau, 1973, p. 37). Prior research also suggests that in some settings purposeful silence develops over time and coexists with verbal and written utterances (Christian, 2020). Through their expressive utterances, leaders can either model behaviors that lead to a culture of silence or a culture of dialogue (Maxfield, 2016; Gatti and Poli, 2023). Organizational participants often pick up on a culture that valorizes and perhaps even rewards silence, leading to a “spiral of silence” within the organization (Noelle-Neumann, 1984; Noelle-Neumann and Petersen, 2004). Brinsfield (2013) proposes that such silence can (among other motivations) be relational or defensive, allowing the employee to avoid harming a relationship with senior management or having a fear of reprisal for speaking up. This often leads to the production of widespread “facades of conformity” across the organization (Maxfield, 2016; Hewlin, 2003). In the current study, we acknowledge that the Enron email corpus that we analyze may not necessarily capture the nuances of silence that are important aspects of the tone at the top. That being said, we propose that managerial silence on important topics or issues may have contributed to the corporate irresponsibility that occurred at Enron.

In summary, our theoretical framing motivates us to examine three inter-related research questions. Research question #1 (RQ1) is our primary research question and focuses on whether Enron’s senior executives communicate a tone about key topics and social actors that differ from other employees. Research question #2 (RQ2) adds a degree of specificity to RQ1 by analyzing whether the tone at the top varies by the size of the email audience as well as by what was happening to Enron at that moment in time. Finally, research question #3 (RQ3) examines whether the emotions expressed by senior executives and other employees changed in the periods immediately before the company’s collapse.

Data and method

Sample

To address our research questions, we use the Enron case as our focus due to its significant and enduring impact on the field of corporate governance and corporate communication (Penikis et al., 2023). Enron’s extensive, publicly available email dataset – known as the Enron Corpus – offers a unique opportunity for in-depth textual analysis that is not possible in other, more recent cases since it is the only currently available, legal corpus of internal corporate email communications (Connor, 2015). The Enron case allows for a granular exploration of corporate communication strategies, providing insights that are both rich in detail and broad in relevance. This approach not only enhances the specificity and depth of our analytical findings but also makes our study a valuable reference for researchers looking to apply similar methodologies to other financial market scandals (see Boddy, 2023). By focusing on a well-documented and critically examined event, our research contributes to a deeper understanding of corporate communication of tone at the top.

We use the version of the email corpus seized by the FERC and subsequently organized and edited by academic researchers as the dataset for the current study. More specifically, we start from the Ruhe (2016) version of the Enron email corpus, which contains 203,648 emails that were gathered from the inboxes of 149 Enron employees [9]. The corpus is an archive of emails both sent by senior executives as well as emails sent to senior executives that these individuals did not delete. We organized the Ruhe data through a series of editing steps. First, we removed duplicate emails that were stored in more than one senior executive inbox, emails sent from non-Enron email addresses and emails that contained less than 30 characters. These eliminations resulted in a dataset of 144,228 non-duplicate emails that were sent from an “@enron” email address.

Data preparation

Our first data preparation step was to divide the emails into sentences since this is our unit of analysis. This resulted in a dataset that consists of 527,356 sentences. As argued earlier, sentences are used as the base-level speech unit because we are interested in identifying the tone and emotions expressed about specific topics and stakeholders. Following from Bakhtin (2013, p. 91), we recognize that a complete email will enlist multiple sentences—often with different tones—to build a narrative. Focusing on sentence-level data allows us to identify the tone and emotion expressed within a single utterance. This approach is reductionist in that it reduces the corpus to individual sentences and thus loses the more holistic meaning of the entire email. At the same time, it foregrounds the micro-level tones and emotions that are expressed about certain topics and stakeholders.

The next data preparation step was to use the text processing module in R to generate a document-term matrix that consists of all the unstemmed words in the corpus [10]. This document-term matrix allowed us to generate a frequency listing of all the words, from which we identified external stakeholder nouns as well as subject nouns. This listing included subject nouns such as regulation and deregulation about which we expect senior executives to have strong evaluative opinions, as well as subject nouns like gas and electricity, etc. that denoted specific aspects of Enron’s business operations. The latter group of subject nouns are included in the analysis because these nouns both act as a control variable and allow us to see if there are distinctions in the evaluative opinions across these two groups. Figure 1 contains a frequency listing of the stakeholder and subject nouns within the email corpus.

Methodology, variable creation and analysis plan

As mentioned in the preceding sections, our theoretical framing is oriented around five conceptual distinctions: (1) the sender of the email (senior executive vs other employees), (2) the tone articulated, (3) the emotions expressed, (4) the audience size and (5) the time period during which the email was sent.

To identify whether the email sender was a senior executive, we started from the functional position information contained in the Ruhe data set and then refined the listing by using Creamer et al.’s (2022) employee classification list as well as additional publicly available information such as LinkedIn [11]. This information was used to construct a senior executive variable (senior) that was coded “1” if the email sender was a senior executive (CEOs, Presidents, Executive VPs, Managing Directors, VPs) and “0” otherwise. Within the corpus of sentences, 13.0% of the sentences were sent by senior executives.

Tone was measured using the financial dictionary approach of Loughran and McDonald (2011, 2016), which is the dominant approach within the accounting and finance literature (cf. Gandhi et al., 2019; Johnman et al., 2018; Kearney and Liu, 2014). As Loughran and McDonald note, other, more general measures of tone “substantially misclassify words when gauging tone in financial applications” (2011, p. 36). The analysis module within R that we use generates a negativity measure that starts from Loughran and McDonald’s word list and “counts the number of words associated with a (negative) sentiment scaled by the total number of words in the document” (Loughran and McDonald, 2016, p. 11) [12]. To ensure that the calculated negativity score does not count the focal stakeholder and subject words included in Figure 1, we first remove these words from the corpus.

The provided analyses use three different versions of Loughran and McDonald’s negativity score as our dependent variable. The first measure (Neg.i) converts the LM negativity score into a 1/0 indicator variable. This measure distinguishes between sentences that express a negative tone from those that do not, providing us with a baseline as to whether senior executives are more likely to enunciate a negative or non-negative tone about a topic or stakeholder. The second measure (Neg.i20) and third measure (Neg.i10) attempt to take into account the degree of negativity that is being articulated by examining the upper distribution tail of the negativity score. Neg.i20 is coded 1 if the LM negativity score is greater than the 20% upper tail cutoff (0.111), whereas Neg.i10 is coded 1 if the LM negativity score is greater than the 10% upper tail cutoff (0.166). In contrast to the Neg.i measure, Neg.i20 and Neg.i10 allow us to consider whether senior executives are more likely to use somewhat negative or really negative tones when talking about certain topics and stakeholders. The use of three different measures provides more information about the relationships between the dependent and independent variables.

Emotion is measured using the NRC word-emotion association lexicon (Mohammad and Turney, 2013). The authors “compiled a large English-term emotion association lexicon,” noting eight basic emotions: anger, fear, anticipation, trust, surprise, sadness, joy and disgust as “prototypical emotions” (p. 438). Like other available word-emotion lexicons in R, it was developed by manually annotating large amounts of textual data (cf. Mohammad, 2022). This lexicon has been used in previous research and provides results that are similar to other word-emotion lexicons such as EmosenticNet and DepecheMood (Kušen et al., 2017). While different lexicons measure varying numbers of emotions, almost all contain a core group of emotions consisting of anger, fear, joy and sadness. In the analyses that follow, we report the results for three of the core emotions (joy, fear and sadness) [13]. Our dependent variables Joy, Fear and Sadness are binary variables coded as “1” for sentences that contain one or more words from, respectively, the joy, fear and sadness dictionaries, otherwise “0.”

Audience size, Recipients, is measured by counting the number of email recipients for a given email message. On average, emails sent by senior executives had 4.2 recipients. Our assumption is that emails with more recipients are more “public” compared to email with fewer recipients.

To assess whether tone and emotion changed over time, we divided the data into four periods. Period1 contains all emails sent prior to January 1, 2001. As noted earlier, on December 31, 2000, Enron shares were trading at 70 times earnings and up 87% over the previous year. Period2 consists of emails sent between January 1, 2001, and August 14, 2001, which was the date that Jeffery Skilling resigned. Period3 contains emails sent between August 15, 2001, and October 16, 2001. On October 16th, an announcement was made that Enron would have to restate its previous earnings. Period4 consists of emails sent between October 17, 2001, and December 2, 2001, which was the day that Enron filed for bankruptcy. Table 1 provides mean values on the aforementioned tone and emotion variables across the four periods.

The analyses that follow use a variety of statistical techniques conducted in R, particularly, LASSO regression algorithms (Tibshirani, 1996) to explore the relationship between the dependent and independent variables. For both our analyses of tone and emotion, we view the 1/0 tone or emotion variable as the dependent variable and the presence of individual focal words both alone and interacted with senior as the independent variables. Our core variables of interest are the interaction terms that measure the interaction between the speaker and the focal words for topics and stakeholders because we expect that senior executives—compared to other employees—will speak differently about certain topics and stakeholders. We utilize a LASSO GLM regression rather than the more common logistic regression because LASSO algorithms apply a penalty function to variable coefficients as a way of correcting for potential multi-collinearity in the independent variables as well as potential over-fitting of the data [14]. One of the unique features of LASSO regressions is that regression results only include coefficients for variables that are significantly related to the dependent variable since the penalty function sets non-significant coefficients to zero (Hastie et al., 2021, p. 6).

In the analysis that follows, we first provide the LASSO GLM regression results for the relationships among tone scores and the speaking patterns of senior executives. We then complement these quantitative analyses with inductive analyses and descriptive details of the data, examining specific examples of both the vocabularies as well as the types of tone-infused sentences used by senior executives. Finally, we consider the use of emotion. The combination of statistical and descriptive analyses helps us to better understand how the tone at the top was expressed within Enron’s emails.

Findings

Tone at the top

The three columns in Table 2 report the LASSO regression results for the three different specifications of the negativity-dependent variable. Within each column, we first report the focal interaction effects, the main effects and the control variables. We are primarily interested in the focal interaction effects because these variables tell us whether all else being equal, senior executives were more likely to use different tones when speaking about certain topics and stakeholders. The columnar format of Table 2 also allows us to read across the columns to see whether senior executives were more likely to use negative, somewhat negative or extremely negative tones when talking about specific topics and stakeholders. As mentioned previously, this approach helps us to assess the sensitivity of the results to the way that the dependent variable is specified. All three regressions include additional control variables for the period in which the email was sent.

The reported results for the focal interactions are mostly consistent with our expectations. For FERC, senior executive enunciations were more likely to be negative (as seen in the positive coefficient on ferc × senior) regardless of the negativity measure used, whereas for deregulation the enunciations were non-negative. Consistent with our expectations, senior executives were also less likely to speak negatively about customers. Similarly, senior executives were more likely to speak negatively about the democrats but less likely to speak negatively about the republicans. This said, there was no difference in how senior executives and other employees talked about regulation in any of the three models.

Table 2 also reports results for both the impact of audience size (recipients) as well as for tonal differences across temporal periods. For audience size, as indicated by the coefficient on recipients × senior, senior executives were more likely to use negative and somewhat negative but not extremely negative tones as the audience size increased. These results are consistent with the idea that it is acceptable to articulate negative but not-too-negative opinions, especially when the email is public. The negativity results for the different periods add additional detail to this statement. More specifically, the results show that, during periods 1 and 2, senior executives tended to not use extremely negative tone but that this changed in periods 3 and 4 as the situation at Enron started to unravel. Thus, while the use of an extremely negative tone in periods 1 and 2 appeared to be strategic, the use of extremely negative tones in the later periods could be said to be emotive in that it was expressing senior executives’ “subjective emotional evaluation” (Bakhtin, 2013, p. 84) regarding what was happening to the company. We return to the implications of this observation in the discussion section.

In sum, the regression results presented in Table 2 illustrate that the sentence-level utterances of senior executives communicate a tone at the top that is different from other employees. Compared to the enunciations of other employees, senior executives were more likely to speak negatively about FERC and democrats but less negatively about the SEC, deregulation, investors, the market, customers and republicans. This mode of speaking is consistent with a pro-market normative worldview that emphasizes the benefits of deregulated markets and Republicans who champion this worldview [15].

Descriptive detail

The LASSO regression results are interesting because they suggest that tonal utterances by senior executives lead other employees on certain topics. At the same time, it is important to acknowledge that there are aspects of tone that are not visible in the aggregate statistical results. First, the results do not distinguish between task-oriented sentences pertaining to senior executive responsibilities and more general evaluative utterances. Second, tone is communicated not only by how senior executives talk about these important topics and stakeholders but also by how they speak more generally. For example, was the ethos of “cowboy” capitalism that previous commentaries mention (Kolb, 2007) evident in senior executive email communications? Finally, did the mode of speaking leave a space for illegal and/or irresponsible activity in that senior executives did not speak strongly enough about certain topics? The current section considers the vocabulary used by senior executives as well as specific sentence-level data to better understand these three aspects.

Our starting point for this portion of the analysis is the tidytext suite of textual analysis tools found within R. Tidytext allows us to identify and count the words found in the 70,539 sentences written by senior executives. The results show that the vocabulary used in senior executive emails was quite large (44,863 words) but that the top 1,000 words accounted for almost 60% of the email vocabulary. Furthermore, 20% of the total vocabulary consisted of words that appeared less than 25 times in the senior executive email corpus. While these infrequent words undoubtedly communicate a tone at the top, we assume that the effective communication of tone to other employees depends more on patterned, repetitive modes of speaking. To shed light on such repeated patterns, Figure 2 provides a wordcloud of the top 200 words used in senior executive emails.

The wordcloud in Figure 2 is consistent with the idea that emails are chiefly a task-oriented mode of communication. For example, the wordcloud contains both task-oriented subject words such as development (2,441 instances), power (1,517), risk (1,329) and business (1,458). It also contains organizing words such as meeting (2,914), call (2,783), mail (1,824), send (1,524), email (1,146) and forward (1,437) [16]. Interestingly, the top 200 vocabulary words, by themselves, are almost exclusively tonally neutral in that there are no strongly positive nor strongly negative words other than the word good (1,526). In this regard, the primary vocabulary used by senior executives was business-like.

When we move down a level and consider the vocabulary used in sentences about the federal regulator, FERC, we see a similar emphasis on business-like words. The word cloud in Figure 3 summarizes the 100 most-prevalent words found in these FERC-centric sentences. An inductive analysis of these sentences, meanwhile, shows that the majority were task-oriented, often talking about specific situations and how FERC is expected to act, and how Enron can respond. At the same time, there were a small number of sentences that included strong expressive phrasings such as “price gouging,” or it “will be used against us,” or “heading into an administrative morass.” This latter group of sentences tended to be less specific in their focus.

To highlight how a normative worldview and ethical stance are communicated within senior executive emails, Table 3 shows a sample of illustrative sentences. The sentences reported in Table 3 suggest that a portion of the negativity expressed by senior executives was “part of the job” in that senior executives were tasked with dealing with FERC and responding to their regulatory concerns. At the same time, senior executives sometimes slipped their worldview and ethical stance into the communication stream via the use of more abstract sentences.

In contrast to senior executive enunciations about FERC, enunciations about deregulation used a slightly different vocabulary and were both more positive and more abstract. These sentences tended to express a belief in deregulation. For example: “deregulation worked, the problems in California can be directly traced to continued regulatory interference in the market.” Or: “we are talking to xxxx about writing a ‘what if’ no deregulation piece for us and hopefully others.”

Whereas enunciations about FERC and deregulation were about the business of Enron and how it was/should be regulated, senior executive enunciations about the democrats and republicans were about how the two political parties—and politics more generally—has the potential to impact Enron’s business activities. This group of sentences sometimes enlisted descriptive phrasings—“an avenue of escape from the dungeon of irrelevancy” or “are weak kneed and spineless”—to communicate an evaluative opinion. In contrast to the illustrative sentences about FERC, sentences about democrats/republicans appear to be more general and less task-related.

Taken together, the sentences reported in Table 3 illustrate how a tone at the top is communicated. On one level, this tone is consistent with a cowboy capitalism worldview in that senior executives enunciate a belief in deregulation and use a negative tone when talking about FERC. At the same time, the tone is mostly business-like in that senior executives tend not to use inflammatory words when talking about FERC nor do they use these words on a general basis [17]. We might suggest that the tone from the top was a polite version of cowboy capitalism where swearing and other forms of disparagement were not prevalent.

The preceding analysis brings to the fore how senior executives communicated a tone at the top. At the same time, did these enunciations of tone leave a space for illegal activity and/or irresponsible activity? Although we cannot answer this directly, the vocabulary used by senior executives also highlights what they did not talk about. On the one hand, senior executives did not use accounting terminology in an attempt to guide and control the organization. Accounting words were used infrequently in senior executive emails in that no accounting words appeared in the top 100 vocabulary terms. Accounting terminology like revenue (452 times), expenses (113), profit (444) and control (477) were used infrequently.

On the other hand, words denoting “responsibility” were also mostly absent. The phrase social responsibility (0) and the words ethics (8) and morals (10) appeared very infrequently, if at all, within senior executive emails. In this regard, senior executives did not use accounting words to guide and control the organization nor did they exercise ethical leadership and send “a clear, consistent message that ethics and social responsibility are valued in the firm” (Sama and Shoaf, 2008, p. 44 quoted in Weber, 2010, p. 168). Once again, this latter set of omissions appears to be consistent with the laissez-faire aspects of cowboy capitalism.

The above analysis complements the previously provided LASSO regression results by illustrating the vocabularies and types of sentences that senior executives used to communicate a tone at the top. In the aggregate, the vocabularies were primarily task-oriented, yet the sentences were often evaluative in that they communicated a worldview about important topics and key stakeholders. The selected sentences also draw attention to the variation in speaking styles within and across the different topics, including how some sentences appear to be related to a specific task-related situation whereas others appear to be a more general enunciation of a senior executive worldview. While the vocabulary lists and the illustrative sentences are unable to summarize the results across over 500,000 sentences in the way that the LASSO regression results can, they do help us to visualize the sentence-level data and help foreground other key aspects of senior executive tonal communications.

Expressing emotion

Table 4 reports the LASSO regression results for the expression of joy, fear and sadness. The table parallels Table 2 in that we include the same independent variables. What differs is that we are primarily interested in how the expression of emotions by senior executives changed across time periods. In this regard, the other independent variables can be viewed as control variables.

Table 4 results illustrate that the expression of emotion by senior executives varied across the temporal periods compared to other employees. In Period 1, senior executives were more likely to express joy compared to other employees, whereas in Periods 3 and 4, senior executives were less likely to express joy. In terms of fear, senior executives were less likely to express fear compared to other employees as the company started to unravel in Periods 2–4. In terms of sadness, senior executives were less likely to express sadness compared to other employees in Periods 2–4. These results suggest that the expression of emotion by senior executives, like the enunciation of tone, differed from other employees across the different periods.

As reflected in the coefficients on the other interaction terms, senior executives were more likely to express both joy and fear when talking about the FERC as well as fear when talking about regulation. Furthermore, senior executives were also more likely to express joy and fear – and less likely to express sadness – as the audience size increased.

To better illustrate how the expression of emotion changed over time, we include Figure 4, which reports how the expression of fear, joy and sadness changed over time for senior executives as well as other employees. Figure 4 first calculates the average amount of emotional expression for each of these emotions and then subtracts the average from the actual amount for each of the periods. In this way, Figure 4 shows how the use of these three emotional registers changed for senior executives and other employees over time.

Figure 4 results are mostly consistent with what we would expect. For both groups, utterances became less joyful and sadder as time progressed. For other employees, utterances also became more fearful in Periods 2–4 compared to Period 1. This finding is not surprising because other employees were likely uncertain as to what was happening and were fearful about what might happen to their jobs. However, in contrast to other employees, utterances by senior executives were more fearful in Period 2 but not in Periods 3 and 4. This result raises questions as to whether the expression of fear by senior executives was strategic in that they managed to control their fear (through remaining silent or inexpressive) even as the company approached bankruptcy. We return to this topic in the discussion section that follows.

Discussion and conclusion

We end with a discussion of our findings followed by the study’s contributions, limitations and future research directions before a brief conclusion. The study has examined the tone that Enron’s senior executives communicated and the emotions expressed in their emails in the three years immediately preceding the company’s collapse. We started by defining the tone at the top as an evaluative opinion by senior executives about important topics and social actors that are communicated to internal (and external) audiences. We then distinguished between tone and emotions, where emotions refer to an expressive, subjective evaluation about speech topics and social actors. We find that tone and emotions are expressed in diverse ways: verbally, via written emails and through silence. These concepts were then used to examine how variations in (1) the speaker, (2) the audience and (3) the temporal moment impacted what senior executives communicated in their emails. The results indicate that senior executives not only communicated differently than other employees about important topics and social actors but also that senior executive utterances varied depending on the audience size and the temporal moment when the utterance occurred. What was unsaid by management in their emails, especially regarding the activities of the trading division, may have been a purposeful part of the tone at the top. Management’s communication strategy might have resulted in a spiral of silence among lower-level employees and external industry analysts who did not have the confidence to speak up about corporate irresponsibility.

These findings contribute to our theoretical and empirical understandings in four important ways. First, the study complements previous organizational research on narratives (Farias et al., 2021; Neu et al., 2022) by illustrating how senior executive enunciations are “assembled” through the bundling together of tonally-variable sentences about different stakeholders and topics. As we have argued throughout the study, the construction of senior executive narratives depends on such assembling activities since the use of different sentences facilitates the construction of nuanced narratives that have a moral to the story (Farias et al., 2021) that communicate to employees how they should think and feel about specific external stakeholders and topics.

Second, the study builds on research on the contextual aspects of executive communication (cf. Goodrick et al., 2020; Firoozi and Ku, 2023) in illustrating how senior executive communications vary depending on the sentence subject, the topic, the audience size and the temporal moment. Moreover, the results illustrate that, in most cases, senior executives may speak negatively, but not too negatively, about important subjects and social actors – or even remain silent.

Third, the study highlights the importance of distinguishing between the enunciation of tone (i.e. an evaluative opinion) and the expression of emotion (i.e. the expression of a subjective evaluation). While the data cannot distinguish between the affective and strategic aspects of emotional content, the results suggest that both tone and emotions must be considered. We also acknowledge, as proposed by Hart (2001, p. 45) that tone could encompass a broader range of dimensions, including certainty, optimism and realism, among others.

Fourth, the results show that leadership is communicated by both what is said and how it is said but also by what is left unsaid. As the previous section illustrated, senior executives rarely used accounting vocabulary to guide and control the organization. Furthermore, there was no clear and consistent message about ethics or social responsibility. This is not to say that senior executives did not talk about these important topics in other communication mediums but rather that they did not insert these topics into email communication in the ways that they inserted comments about deregulation. Stated differently, ethics and social responsibility were constituted as “non-topics” within Enron’s email communications. Senior executives’ silence on these critical topics may have been a purposeful part of the tone at the top that other Enron employees picked up on as an element of the company’s façade of conformity. This leads us to conclude that tone is communicated by both what is said and what is suppressed.

While the aforementioned insights contribute to our theoretical and empirical understandings of senior executive communications, we acknowledge the study’s limitations. Notably, we examine senior executive emails from a single company. This dataset is more than twenty years old and thus may not be representative of other more recent cases of potential corporate irresponsibility such as Volkswagen, Odebrecht and Uber. At the same time, the Enron email dataset is the only currently available, legal corpus of internal corporate email communications (Connor, 2015). Furthermore, there is anecdotal evidence that the corporate email communications of Enron’s senior executives may not be all that different from other corporations such as Volkswagen (McLaughlin et al., 2017; Shepardson, 2016), Odebrecht (Chavkin, 2019), Uber (Davies et al., 2022) and Sony (Connor, 2015). Thus, despite the limitations of the Enron data, we believe the analysis contributes to our understanding of how internal senior executive emails communicate key messages about what is important to the organization. We hope future research can build upon these findings when additional, newer datasets become available.

Relatedly, we made the research choice to conduct a Big Data quantitative analysis of senior executive email communications. As a result, our research choice does not allow us to provide a fine-grained reading of senior executive and corporate narratives such as in Moreno et al. (2019), Craig and Amernic (2020), Quinn et al. (2023) and Arivdsson and Sabelfeld (2023). Finally, while the current study documents the salience of both tone and emotion, available empirical measures of emotion are noisy and have difficulty distinguishing between evaluative opinions and expressive evaluations within a sentence since both forms of enunciation sometimes use the same word to either enunciate tone and/or emotion (Mohammad, 2022). For these reasons, additional fine-grained qualitative research on the Enron dataset and other internal forms of corporate communication has the potential to complement and extend the aggregate-level findings of the current study.

In conclusion, the current study illustrates how Enron’s senior executives communicated a tone from the top. This tone was mostly silent about the need to act in socially responsible ways, but it also did not explicitly celebrate risk-taking nor did it exhort employees to maximize profit regardless of the means. In this regard, the results seem to suggest that senior executive email communications left a space for, but did not condone, corporate irresponsibility. Discussions about these important yet sensitive topics were, apparently, left to other more private communication mediums.

Figures

Topic prevalence

Figure 1

Topic prevalence

Word cloud of senior management communication

Figure 2

Word cloud of senior management communication

Word cloud of senior executive communication regarding the federal energy regulatory commission

Figure 3

Word cloud of senior executive communication regarding the federal energy regulatory commission

Changes in emotions across periods for senior management and other employees

Figure 4

Changes in emotions across periods for senior management and other employees

Tone and emotion descriptive statistics

Period1 (Jan 1, 1999–
Dec 31, 2000)
Period2 (Jan 1, 2001–
Aug 14, 2001)
Period3 (Aug 15, 2001–
Oct 16, 2001)
Period4 (Oct 17, 2001–
Dec 2, 2001)
Tone
Negativity0.0540.0540.0560.058
Emotion
Fear0.1810.2240.2310.253
Joy0.2510.2480.2310.241
Sadness0.1550.1870.1830.241
N178,032208,96856,83183,525

Note(s): Table 1 shows mean values on the tone and emotion variables in each of the four periods indicated above. The last row (N) indicates the distribution of the 527,356 discrete sentences in our Enron email database across the four periods

Source(s): Table created by authors

LASSO-based estimates of topical/stakeholder predictors of negative sentiment by senior management

Neg.i (negative)Neg.i20 (somewhat negative)Neg.i10 (extremely negative)
Intercept0.3940.2280.103
Focal interactions
ferc × senior0.0200.0150.006
nyse × senior−0.241
sec × senior−0.056−0.123−0.024
government × senior −0.014−0.014
regulation × senior
deregulation × senior−0.204−0.055
investor × senior−0.0480.007−0.017
market × senior0.007−0.015−0.011
customers × senior−0.005−0.008−0.002
public × senior0.035 −0.012
republicans × senior−0.041−0.103−0.032
democrats × senior0.059 0.047
recipients × senior0.0010.001−0.001
period1xsenior0.008−0.001−0.003
period2xsenior0.003−0.001−0.002
period3xsenior−0.072 0.009
period4xsenior−0.0600.0130.016
Main effects
Senior −0.009
Ferc0.104−0.025−0.030
Nyse−0.012−0.080−0.035
sec0.1870.0410.010
Government0.127−0.009−0.017
regulation0.101−0.019−0.018
Deregulation0.0590.005−0.008
Investor0.150 −0.015
Market0.112−0.005−0.015
Customers0.087−0.022−0.023
Public0.082−0.051−0.030
Republicans0.1160.004−0.011
Democrats0.170−0.010−0.011
Recipients0.0510.0210.005
Control variables
trading0.026−0.065−0.039
power0.092−0.032−0.029
Energy0.076−0.058−0.038
Company0.063−0.047−0.032
Gas0.036−0.075−0.044
business0.062−0.066−0.041
Electricity0.1110.002−0.014
period10.002−0.016−0.007
period2 −0.011−0.005
period3−0.0130.0020.002
period4−0.0010.0190.007
trading × senior0.0140.006−0.011
power × senior −0.018−0.007
energy × senior0.005−0.032−0.010
company × senior0.010−0.012−0.011
gas × senior 0.006
electricity × senior −0.0290.004

Note(s): Table shows LASSO-based estimates of the relationship between senior management (senior) and the topics and stakeholders. We interact with each of the variables with senior to determine whether senior executives are more or less likely to express negativity about a given topic or stakeholder. Our dependent variables Neg.i, Neg.i20 and Neg.i10 are binary variables coded as “1” for sentences that are, respectively, negative, somewhat negative and very negative. Because Neg.i, Neg.i20 and Neg.i10 are binary dependent variables, our LASSO-based estimates are built off logistic regression. Variables not significantly related to negativity (such as government × senior in the first model) are assigned a coefficient of 0 during LASSO regression and thus coefficient values are not shown

Source(s): Table created by author

Quotes – Enron: illustrative sentences

FERC
the ferc nor the partnerships will accept these charges unless we have the data to send the costs to them
we do not have the option of reducing our certificated capacity without filing an abandonment at the ferc in which we would have to demonstrate that the abandonment would not impair our abilities to meet our firm obligations
we have examined joskow and ferc’s analysis on the price gouging issue and we believe there are several major flaws in their methodologies
by copy of this email i will ask our outside counsel bob madden to comment on how he thinks the ferc ruling can and will be used against us
on point, i think that at the end of the day ferc will not deal with the negative ctc directly
If the intent of the soft cap is to move toward cost justification for bids above $150/MWh, then FERC is headed into an administrative morass
De-regulation
deregulation worked, the problems in California can be directly traced to continued regulatory interference in the market
however the goal is to produce a high level analysis, i. e. deregulation is good and can work, command and control regulation is bad and we know that it does not work
we are talking to xxxx about writing a ‘what if’ no deregulation piece for us and hopefully others
Democrats and republicans
republicans meanwhile sense that the crisis gives them an avenue of escape from the dungeon of irrelevancy to which they had been exiled by heavy losses in the last three elections
you and your right-wing friends know the democrats are weak kneed and spineless
they view it as unfair and punitive which it is and are frustrated that the democrats would include it simply to make the deal harvey proof

Note(s): Table shows illustrative sentences that provide a descriptive overview of the different types of senior executive utterances that underpin the results reported in Table 2. These quotes illustrate how a normative worldview and ethical stance are communicated within the senior executive emails. In line with prevailing best practices, sentences in the email database are made lowercase as part of the text processing required for the analyses (see Li and Haque, 2019; Neu and Saxton, 2023)

Source(s): Table created by authors

LASSO-based estimates of topical/stakeholder predictors of emotions by senior management

JoyFearSadness
Intercept0.1910.1440.139
Focal interactions
ferc × senior0.0080.002
nyse × senior−0.082−0.073−0.115
sec × senior0.115−0.0420.133
government × senior0.035 0.001
regulation × senior0.0650.0280.031
deregulation × senior−0.102 −0.078
investor × senior−0.042−0.160−0.077
market × senior0.029 0.007
customers × senior−0.032
public × senior−0.015 −0.017
republicans × senior−0.012−0.0140.256
democrats × senior−0.182 0.077
recipients × senior0.0010.017−0.001
period1 × senior0.003 0.007
period2 × senior −0.010−0.002
period3 × senior−0.017−0.037−0.018
period4 × senior−0.015−0.053−0.043
Main effects
Senior−0.016 −0.013
Ferc0.0120.093−0.043
Nyse0.014
Sec0.0010.0620.076
Government0.0270.7940.071
Regulation0.0160.1350.068
Deregulation0.0560.0620.075
Investor0.1210.1810.181
Market0.0390.1070.054
Customers0.0750.0460.055
Public0.0740.1000.073
Republicans0.0980.1070.161
Democrats0.1400.1880.180
Recipients−0.012−0.0010.008
Control variables
trading0.0140.0540.031
power0.0400.0620.056
Energy0.0580.0950.051
company0.0430.0620.071
Gas0.0550.0150.006
business0.0860.0650.042
electricity0.0230.0740.072
period10.013−0.014−0.017
period20.005
period3−0.0070.008
period4−0.0030.0260.035
trading × senior0.001 −0.005
power × senior0.0070.025
energy × senior0.006 −0.012
company × senior0.033−0.003−0.020
gas × senior−0.053 0.005
electricity × senior0.0890.0600.065

Note(s): Table shows LASSO-based estimates of the relationship between senior management (senior) and the topics and stakeholders. We interact each of the variables with senior to determine whether senior executives are more or less likely to express emotions about a given topic or stakeholder. Our dependent variables Joy, Fear and Sadness are binary variables coded as “1” for sentences that contain one or more words from respectively, the joy, fear and sadness dictionaries described in the methods section. Because Joy, Fear and Sadness are binary dependent variables, our LASSO-based estimates are built off logistic regression. Variables not significantly related to emotions (such as government × senior in the Fear regression in the second column) are assigned a coefficient of 0 during LASSO regression and thus coefficient values are not shown

Source(s): Table created by authors

Notes

1.

The Sony email hack data is in the public domain but there are important ethical issues and data issues regarding the use of this data (Connor, 2015).

2.

See Healy and Palepu (2003, p. 4) for a timeline of critical events in the Enron scandal.

3.

Enron famously won Fortune magazine’s “America’s most innovative company” award for an unprecedented six years between 1996 and 2001.

4.

Our definitional choice is informed by a broad literature that proposes the essential connection between managerial ethical values and tone at the top (for example, see Ewelt-Knauer et al., 2022; Warren et al., 2015).

5.

The distinction between the interdependency and accountability perspectives hinges on whether there is an instrumental or moral responsibility to external stakeholders. Oliver (1991) notes an organization might respond to the demands of external stakeholders by providing a “one off” disclosure without entering into a longer-term, repetitive accountability relationship.

6.

Bakhtin is careful to state that an utterance is not necessarily the same as a sentence since an utterance concludes when the speaker has finished speaking about a particular topic. At the same time, he notes that, as a grammatical form, sentences allow the speaker to package together different subjects and tones into an overall utterance (2013, p. 81).

7.

Bakhtin states that it is “the commonness of assumed basic value judgments [among participants] that constitutes the canvas upon which living human speech embroiders the designs of intonation” (2010, p. 166).

8.

Aerts and Yan (2017, p. 407) suggest that these different components co-occur and thus are used to communicate a tone and style. Li and Haque (2019, p. 1690) introduce the notion of “hedges” as a way to communicate tone and propose that rhetorical hedging in written narratives is likely a more “deliberate choice of tactics to influence stakeholder perceptions.”

9.

While there are a number of versions of the Enron email dataset, we use the Ruhe (2016) version, which is readily available in R format and which provides arguably the cleanest version of the email database: http://www.ahschulz.de/enron-email-data.

10.

Stemming is a process in natural language processing that reduces words to their root form to simplify text analysis (e.g. converting “discussing” to “discuss”).

11.

Creamer et al.’s (2022) position listing builds upon Shetty and Adibi’s (2004) original employee classification by using signatures, content, internal traders’ list and documents released during the bankruptcy proceedings to classify unknown employee positions more accurately.

12.

A complete list of the Loughran and McDonald negative dictionary words we employed is available online at https://sraf.nd.edu/loughranmcdonald-master-dictionary/

13.

We conducted the same analysis for anger but have not included those results here. The findings for anger mirrored those for fear for both senior executives and other employees.

14.

We use the “best lambda” as the penalty function which, according to Hastie et al. (2021), produces the most robust results.

15.

We also ran the LASSO regression without any interaction terms to assess the overall tone of the email stream. The coefficients for all of the independent variables—except NYSE—were negative, which indicates that the overall tone, for almost all variables, was negative.

16.

The numbers in parentheses show the number of times that each word was used by senior executives.

17.

A review of the words that appeared less than 25 times included: clueless (1), bureaucratic (1), dumb (5), inefficient (11) and morass (1). The number of times senior executives used these words (i.e. the numbers in parentheses) indicates that such explicit tonal words were used infrequently by senior executives.

References

Aerts, W. and Cormier, D. (2008), “Media legitimacy and corporate environmental communication”, Accounting, Organizations and Society, Vol. 24 No. 1, pp. 1-27, doi: 10.1016/j.aos.2008.02.005.

Aerts, W. and Yan, B. (2017), “Rhetorical impression management in the letter to shareholders and institutional setting: a metadiscourse perspective”, Accounting, Auditing and Accountability Journal, Vol. 30 No. 2, pp. 404-432, doi: 10.1108/aaaj-01-2015-1916.

Amernic, J. and Craig, R. (2013), “Leadership discourse, culture, and corporate ethics: CEO-speak at news corporation”, Journal of Business Ethics, Vol. 118 No. 2, pp. 379-394, doi: 10.1007/s10551-012-1506-0.

Arivdsson, S. and Sabelfeld, S. (2023), “Adaptive framing of sustainability in CEO letters”, Accounting, Auditing and Accountability Journal, Vol. 36 No. 9, pp. 161-199, doi: 10.1108/aaaj-11-2019-4274.

Arnold, B. and de Lange, P. (2004), “Enron: an examination of agency problems”, Critical Perspectives on Accounting, Vol. 15 Nos 6-7, pp. 751-765, doi: 10.1016/j.cpa.2003.08.005.

Audi, R., Loughran, T. and McDonald, B. (2016), “Trust, but verify: MD&A language and the role of trust in corporate culture”, Journal of Business Ethics, Vol. 139 No. 3, pp. 551-561, doi: 10.1007/s10551-015-2659-4.

Baker, C.R. (2003), “Investigating Enron as a public private partnership”, Accounting, Auditing and Accountability Journal, Vol. 16 No. 3, pp. 446-466, doi: 10.1108/09513570310482327.

Baker, C.R. and Hayes, R. (2004), “Reflecting form over substance: the case of Enron Corp”, Critical Perspectives on Accounting, Vol. 15 Nos 6-7, pp. 767-785, doi: 10.1016/j.cpa.2002.08.001.

Baker, C.R. and Hayes, R. (2005), “The fallout: was Enron an Accounting failure?”, Managerial Finance, Vol. 31 No. 9, pp. 5-28, doi: 10.1108/03074350510769839.

Bakhtin, M.M. (1994), The Bakhtin Reader: Selected Writings of Bakhtin, Medvedev, and Voloshinov, New York, edited by Voloshinov, V.N., Medvedev, P.N., Morris, P and Arnold, E.

Bakhtin, M.M. (2010), The Dialogic Imagination: Four Essays, University of Texas Press, Austin, TX.

Bakhtin, M.M. (2013), Speech Genres and Other Late Essays, University of Texas Press, Austin, TX.

Boddy, C. (2023), “Insights into the Bernie Madoff financial market scandal which identify new opportunities for business market researchers”, International Journal of Market Research, Vol. 66 No. 1, pp. 149-167, doi: 10.1177/14707853231173260.

Braun, S., Hernandez Bark, A., Kirchner, A., Stegmann, S. and Van Dick, R. (2019), “Emails from the boss—curse or blessing? Relations between communication channels, leader evaluation, and employees' attitudes”, International Journal of Business Communication, Vol. 56 No. 1, pp. 50-81, doi: 10.1177/2329488415597516.

Brinsfield, C.T. (2013), “Employee silence motives: investigation of dimensionality and development of measures”, Journal of Organizational Behavior, Vol. 34 No. 5, pp. 671-697, doi: 10.1002/job.1829.

Brown, M.E., Treviño, L.K. and Harrison, D.A. (2005), “Ethical leadership: a social learning perspective for construct development and testing”, Organizational Behavior and Human Decision Processes, Vol. 97 No. 2, pp. 117-134, doi: 10.1016/j.obhdp.2005.03.002.

Bruneau, T.J. (1973), “Communicative silences: forms and functions”, Journal of Communication, Vol. 23 No. 1, pp. 17-46, doi: 10.1111/j.1460-2466.1973.tb00929.x.

Buchholz, F., Jaeschke, R., Lopatta, K. and Maas, K. (2018), “The use of optimistic tone by narcissistic CEOs”, Accounting, Auditing and Accountability Journal, Vol. 31 No. 2, pp. 531-562, doi: 10.1108/aaaj-11-2015-2292.

Chavkin, S. (2019), “Leak exposes millions of dollars in new payments in Odebrecht cash-for-contracts scandal”, available at: https://www.icij.org/investigations/bribery-division/leak-exposes-millions-of-dollars-in-new-payments-in-odebrecht-cash-for-contracts-scandal (accessed 12 December 2023).

Cho, C.H., Laine, M., Roberts, R.W. and Rodrigue, M. (2015a), “Organized hypocrisy, organizational facades, and sustainability reporting”, Accounting, Organizations and Society, Vol. 40 No. 1, pp. 78-94, doi: 10.1016/j.aos.2014.12.003.

Cho, C.H., Michelon, G., Patten, D.M. and Roberts, R.W. (2015b), “CSR disclosure: the more things change…?”, Accounting, Auditing and Accountability Journal, Vol. 28 No. 1, pp. 14-35, doi: 10.1108/aaaj-12-2013-1549.

Christian, T. (2020), “Organizational silence: why employees don't speak up”, Medium, available at: https://medium.com/@tiani1999/organizational-silence-why-employees-dont-speak-up-a01c7dbc3fb9 (accessed 7 April 2024).

Comoli, M., Tettamanzi, P. and Murgolo, M. (2023), “Accounting for ESG under disruptions: a systematic literature network analysis”, Sustainability, Vol. 15 No. 8, p. 6633, doi: 10.3390/su15086633.

Connor, J.D. (2015), “The Sony hack: data and decision in the contemporary studio”, Media Industries Journal, Vol. 2 No. 2, doi: 10.3998/mij.15031809.0002.203.

Constable, S. (2021), “How the Enron scandal changed American business forever”, available at: https://time.com/6125253/enron-scandal-changed-american-business-forever/ (accessed 12 December 2023).

Cooper, S.M. and Owen, D.L. (2007), “Corporate social reporting and stakeholder accountability: the missing link”, Accounting, Organizations and Society, Vol. 32 Nos 7-8, pp. 649-667, doi: 10.1016/j.aos.2007.02.001.

Craig, R. and Amernic, J. (2004), “Enron discourse: the rhetoric of resilient capitalism”, Critical Perspectives on Accounting, Vol. 15 Nos 6-7, pp. 813-852, doi: 10.1016/j.cpa.2002.12.001.

Craig, R. and Amernic, J. (2020), “Autobiographical vignettes in annual report CEO letters as a lens to understand how leadership is conceived and enacted”, Accounting, Auditing and Accountability Journal, Vol. 33 No. 1, pp. 106-123, doi: 10.1108/aaaj-12-2018-3774.

Creamer, G.G., Stolfo, S.J., Creamer, M., Hershkop, S. and Rowe, R. (2022), “Discovering organizational hierarchy through a corporate ranking algorithm: the Enron case”, Complexity, Vol. 2022 No. 1, 8154476, doi: 10.1155/2022/8154476.

Curwen, L. (2021), “The collapse of Enron and the dark side of business”, available at: https://www.bbc.com/news/business-58026162 (accessed 12 December 2023).

Davies, H., Goodley, S., Lawrence, F., Lewis, P. and O'Carroll, L. (2022), “Uber broke laws, duped police and secretly lobbied governments, leak reveals”, available at: https://www.theguardian.com/news/2022/jul/10/uber-files-leak-reveals-global-lobbying-campaign (accessed 12 December 2023).

Deegan, C. (2019), “Legitimacy theory: despite its enduring popularity and contribution, time is right for a makeover”, Accounting, Auditing and Accountability Journal, Vol. 23 No. 1, pp. 2307-2329.

Deegan, C. and Rankin, M. (1996), “Do Australian companies report environmental news objectively? An analysis of environmental disclosures by firms prosecuted successfully by the Environmental Protection Authority”, Accounting, Auditing and Accountability Journal, Vol. 9 No. 2, pp. 50-67, doi: 10.1108/09513579610116358.

Delmestri, G. and Goodrick, E. (2016), “Looking away: denial and emotions in institutional stability and change”, in Gehman, J., Lounsbury, M. and Greenwood, R. (Eds), How Institutions Matter, Emerald Group Publishing, pp. 233-271.

Edgar, V.C., Brennan, N.M. and Power, S.B. (2022), “The language of profit warnings: a case of denial, defiance, desperation and defeat”, Accounting, Auditing and Accountability Journal, Vol. 35 No. 9, pp. 28-56, doi: 10.1108/aaaj-04-2020-4519.

Ewelt-Knauer, C., Schwering, A. and Winkelmann, S. (2022), “Doing good by doing bad: how tone at the top and tone at the bottom impact performance-improving noncompliant behavior”, Journal of Business Ethics, Vol. 175 No. 3, pp. 609-624, doi: 10.1007/s10551-020-04647-6.

Farias, C., Seremani, T. and Fernández, P.D. (2021), “Popular culture, moral narratives and organizational portrayals: a multimodal reflexive analysis of a reality television show”, Journal of Business Ethics, Vol. 171 No. 2, pp. 211-226, doi: 10.1007/s10551-020-04460-1.

Firoozi, M. and Ku, C.H. (2023), “Corporate accountability during crisis in the digitized era”, Accounting, Auditing and Accountability Journal, Vol. 36 No. 3, pp. 933-964, doi: 10.1108/aaaj-04-2020-4509.

Fisher, R., Van Staden, C.J. and Richards, G. (2020), “Watch that tone: an investigation of the use and stylistic consequences of tone in corporate accountability disclosures”, Accounting, Auditing and Accountability Journal, Vol. 33 No. 1, pp. 77-105, doi: 10.1108/aaaj-10-2016-2745.

Flam, H. (2002), “Corporate emotions and emotions in corporations”, The Sociological Review, Vol. 50 No. 2_suppl, pp. 90-112, doi: 10.1111/j.1467-954x.2002.tb03593.x.

Gandhi, P., Loughran, T. and McDonald, B. (2019), “Using annual report sentiment as a proxy for financial distress in US banks”, Journal of Behavioral Finance, Vol. 20 No. 4, pp. 424-436, doi: 10.1080/15427560.2019.1553176.

García-Rosell, J.C. (2019), “A discursive perspective on corporate social responsibility education: a story co-creation exercise”, Journal of Business Ethics, Vol. 154 No. 4, pp. 1019-1032, doi: 10.1007/s10551-016-3399-9.

Gatti, M. and Poli, S. (2023), “Accounting to promote ethical tolerance: the case of the Italian prostitution model (1860)”, Accounting, Auditing and Accountability Journal, Vol. 36 No. 9, pp. 309-339, doi: 10.1108/aaaj-08-2022-5956.

Goodrick, E., Jarvis, L.C. and Reay, T. (2020), “Preserving a professional institution: emotion in discursive institutional work”, Journal of Management Studies, Vol. 57 No. 4, pp. 735-774, doi: 10.1111/joms.12535.

Gray, R. (2006), “Social, environmental and sustainability reporting and value creation. Whose value? Whose creation?”, Accounting, Auditing and Accountability Journal, Vol. 19 No. 6, pp. 793-819, doi: 10.1108/09513570610709872.

Gunz, S. and Thorne, L. (2015), “Introduction to special issue on tone at the top”, Journal of Business Ethics, Vol. 126, pp. 1-2, doi: 10.1007/s10551-013-2035-1.

Guthrie, J. and Parker, L. (1989), “Corporate social reporting: a rebuttal of legitimacy theory”, Accounting and Business Research, Vol. 19 No. 76, pp. 343-352, doi: 10.1080/00014788.1989.9728863.

Hamza, M., Mezgani, N. and Jarboui, A. (2023), “CSR as an impression management strategy: the joint effect of disclosure tone management and earnings management”, Sustainability Accounting, Management and Policy Journal, Vol. 14 No. 6, pp. 1126-1149, doi: 10.1108/sampj-08-2022-0423.

Hart, R.P. (2001), “Redeveloping DICTION: theoretical considerations”, in West, M.D. (Ed.), Theory, Method and Practice in Computer Content Analysis, Springer, New York, pp. 43-60.

Hastie, T., Qian, J. and Tay, K. (2021), “An Introduction to glmnet”, available at: https://glmnet.stanford.edu/articles/glmnet.html (accessed 12 December 2023).

Healy, P.M. and Palepu, K.G. (2003), “The fall of Enron”, Journal of Economic Perspectives, Vol. 17 No. 2, pp. 3-26, doi: 10.1257/089533003765888403.

Hewlin, P.E. (2003), “And the award for best actor goes to…Facades of conformity in organizational settings”, Academy of Management Review, Vol. 28 No. 4, pp. 633-642, doi: 10.2307/30040752.

Iren, P. and Kim, M.S. (2023), “How harsh should the legislation be to prevent financial crimes?”, in Concepts and Cases of Illicit Finance, IGI Global, pp. 37-50.

Jarvis, L.C., Goodrick, E. and Hudson, B.A. (2019), “Where the heart functions best: reactive–affective conflict and the disruptive work of animal rights organizations”, Academy of Management Journal, Vol. 62 No. 5, pp. 1358-1387, doi: 10.5465/amj.2017.0342.

Johnman, M., Vanstone, B.J. and Gepp, A. (2018), “Predicting FTSE 100 returns and volatility using sentiment analysis”, Accounting and Finance, Vol. 58 No. S1, pp. 253-274, doi: 10.1111/acfi.12373.

Jonsson, S. (1998), “Relate managerial accounting research to managerial work”, Accounting, Organizations and Society, Vol. 23 No. 4, pp. 411-434, doi: 10.1016/s0361-3682(97)00018-4.

Kearney, C. and Liu, S. (2014), “Textual sentiment in finance: a survey of methods and models”, International Review of Financial Analysis, Vol. 33, pp. 171-185, doi: 10.1016/j.irfa.2014.02.006.

Kim, S. and Choi, S.M. (2018), “Congruence effects in post-crisis CSR communication: the mediating role of attribution of corporate motives”, Journal of Business Ethics, Vol. 153 No. 2, pp. 447-463, doi: 10.1007/s10551-016-3425-y.

Kockelman, P. (2009), “The complexity of discourse”, Journal of Quantitative Linguistics, Vol. 16 No. 1, pp. 1-39, doi: 10.1080/09296170802514146.

Kolb, R.W. (Ed.) (2007), Encyclopedia of Business Ethics and Society, Sage Publications, London.

Kotb, A., Elbardan, H. and Halabi, H. (2020), “Mapping of internal audit research: a post-Enron structured literature review”, Accounting, Auditing and Accountability Journal, Vol. 33 No. 8, pp. 1969-1996, doi: 10.1108/aaaj-07-2018-3581.

Kušen, E., Cascavilla, G., Figl, K., Conti, M. and Strembeck, M. (2017), “Identifying emotions in social media: comparison of word-emotion lexicons”, 2017 5th International Conference on Future Internet of Things and Cloud Workshops (FiCloudW), pp. 132-137.

Li, Z. and Haque, S. (2019), “Corporate social responsibility employment narratives: a linguistic analysis”, Accounting, Auditing and Accountability Journal, Vol. 32 No. 6, pp. 1690-1713, doi: 10.1108/aaaj-10-2016-2753.

Liu, H. and Baker, C. (2016), “Ordinary aristocrats: the discursive construction of philanthropists as ethical leaders”, Journal of Business Ethics, Vol. 133 No. 2, pp. 261-277, doi: 10.1007/s10551-014-2394-2.

Loughran, T. and McDonald, B. (2011), “When is a liability not a liability? Textual analysis, dictionaries, and 10-Ks”, The Journal of Finance, Vol. 66 No. 1, pp. 35-65, doi: 10.1111/j.1540-6261.2010.01625.x.

Loughran, T. and McDonald, B. (2016), “Textual analysis in accounting and finance: a survey”, Journal of Accounting Research, Vol. 54 No. 4, pp. 1187-1230, doi: 10.1111/1475-679x.12123.

Markus, M.L. (1994), “Electronic mail as the medium of managerial choice”, Organization Science, Vol. 5 No. 4, pp. 502-527, doi: 10.1287/orsc.5.4.502.

Massumi, B. (1996), The Autonomy of Affect. Deleuze: A Critical Reader, Blackwell Publishing, Oxford.

Maxfield, D. (2016), “How a culture of silence eats away at your company”, Harvard Business Review, Vol. 7, pp. 2-5.

McCullough, R. (2002), Congestion Manipulation in ISO California, McCullough Research, Portland, Oregon, 29 November.

McCullough, R. (2004), Reading Enron's Scheme Accounting Materials, McCullough Research, Portland, Oregon, 28 June.

McLaughlin, D., Harris, A. and Cronin Fisk, M. (2017), “VW officials destroyed files, e-mails as diesel scheme unraveled”, available at: https://www.bloomberg.com/news/articles/2017-01-12/vw-officials-destroyed-files-e-mails-as-diesel-scheme-unraveled (accessed 12 December 2023).

Michelon, G. and Parbonetti, A. (2012), “The effect of corporate governance on sustainability disclosure”, Journal of Management and Governance, Vol. 16 No. 3, pp. 477-509, doi: 10.1007/s10997-010-9160-3.

Mohammad, S.M. (2022), “Best practices in the creation and use of emotion lexicons”, arXiv preprint arXiv:2210.07206.

Mohammad, S.M. and Turney, P.D. (2013), “Crowdsourcing a word–emotion association lexicon”, Computational Intelligence, Vol. 29 No. 3, pp. 436-465, doi: 10.1111/j.1467-8640.2012.00460.x.

Moreno, A., Jones, M.J. and Quinn, M. (2019), “A longitudinal study of the textual characteristics in the chairman's statements of Guinness: an impression management perspective”, Accounting, Auditing and Accountability Journal, Vol. 32 No. 6, pp. 1714-1741, doi: 10.1108/aaaj-01-2018-3308.

Neu, D. and Saxton, G.D. (2023), “Building ethical narratives: the audiences for AICPA editorials”, Journal of Business Ethics, Vol. 182 No. 4, pp. 1055-1072, doi: 10.1007/s10551-021-05003-y.

Neu, D., Warsame, H. and Pedwell, K. (1998), “Managing public impressions: environmental disclosures in annual reports”, Accounting, Organizations and Society, Vol. 23 No. 3, pp. 265-282, doi: 10.1016/s0361-3682(97)00008-1.

Neu, D., Saxton, G.D., Everett, J. and Rahaman, A. (2022), “The centrality of ethical utterances within professional narratives”, Accounting History, Vol. 27 No. 1, pp. 75-94, doi: 10.1177/10323732211040272.

Nix, A., Decker, S. and Wolf, C. (2021), “Enron and the California energy crisis: the role of networks in enabling organizational corruption”, Business History Review, Vol. 95 No. 4, pp. 765-802, doi: 10.1017/s0007680521001008.

Noelle-Neumann, E. (1984), The Spiral of Silence: Public Opinion – Our Social Skin, Chicago University Press, Chicago.

Noelle-Neumann, E. and Peterson, T. (2004), “The spiral of silence and the social nature of man”, in Kaid, L.L. (Ed.), Handbook of Political Communication, Lawrence Erlbaum, London, pp. 339-356.

Ohlrogge, F., Hardies, K. and Claeys, A.-S. (2024), “Investor reactions to apologies for financial misconduct”, Accounting, Organizations and Society, Vol. 112, 101511, doi: 10.1016/j.aos.2023.101511.

Oliver, C. (1991), “Strategic responses to institutional processes”, Academy of Management Review, Vol. 16 No. 1, pp. 145-179, doi: 10.5465/amr.1991.4279002.

Patten, D.M. (2020), “Seeking legitimacy”, Sustainability Accounting, Management and Policy Journal, Vol. 11 No. 6, pp. 1009-1021, doi: 10.1108/sampj-12-2018-0332.

Penikis, S., Relford, J., Hakim, D. and Sebastian, G. (2023), “Uncovering wrongdoing in the Enron email corpus”, Georgia Institute of Technology, Working Paper, available at: https://www.researchgate.net/publication/370050311_Uncovering_Wrongdoing_in_the_Enron_Email_Corpus

Postaire, M. and Puyou, F.-R. (2024), “Restor(y)ing commitment to a failing organization: how narratives and forecasts mitigate anxiety”, Accounting, Auditing and Accountability Journal, Vol. 37 No. 3, pp. 840-865, doi: 10.1108/aaaj-03-2021-5204.

Preuss, L. and Dawson, D. (2009), “On the quality and legitimacy of green narratives in business: a framework for evaluation”, Journal of Business Ethics, Vol. 84 No. S1, pp. 135-149, doi: 10.1007/s10551-008-9693-4.

Quinn, M., Moreno, A. and Bhatta, B. (2023), “Accountability and accounting for fisheries: six decades of reporting by the Electricity Supply Board of Ireland, 1935-1993”, Accounting, Auditing and Accountability Journal, Vol. 36 No. 6, pp. 1606-1636, doi: 10.1108/aaaj-12-2021-5573.

Rahaman, A.S., Lawrence, S. and Roper, J. (2004), “Social and environmental reporting at the VRA: institutionalised legitimacy or legitimation crisis?”, Critical Perspectives on Accounting, Vol. 15 No. 1, pp. 35-56, doi: 10.1016/s1045-2354(03)00005-4.

Riaz, U., Burton, B. and Fearfull, A. (2023), “Emotional propensities and the contemporary Islamic banking industry”, Critical Perspectives on Accounting, Vol. 94, 102449, doi: 10.1016/j.cpa.2022.102449.

Ribeiro, N., Duarte, P.A., Filipe, R. and David, R. (2022), “Does authentic leadership stimulate organizational citizenship behaviors? The importance of affective commitment as a mediator”, Sustainability Accounting, Management and Policy Journal, Vol. 13 No. 2, pp. 320-340, doi: 10.1108/sampj-11-2019-0423.

Roberts, J. (2004), Enron Traders Caught on Tape, CBS News, available at: https://www.cbsnews.com/news/enron-traders-caught-on-tape/ (accessed 1 June 2004)

Rose, A.M., Rose, J.M., Suh, I., Thibodeau, J., Linke, K. and Norman, C.S. (2021), “Why financial executives do bad things: the effects of the slippery slope and tone at the top on misreporting behavior”, Journal of Business Ethics, Vol. 174 No. 2, pp. 291-309, doi: 10.1007/s10551-020-04609-y.

Roszkowska-Menkes, M., Aluchna, M. and Kamisnki, B. (2024), “True transparency or mere decoupling? The study of selective disclosure in sustainability reporting”, Critical Perspectives on Accounting, Vol. 98, 102700, doi: 10.1016/j.cpa.2023.102700.

Ruhe, A.H. (2016), “Enron data”, available at: http://www.ahschulz.de/enron-email-data/

Sama, L.M. and Shoaf, V. (2008), “Ethical leadership for the professions: fostering a moral community”, Journal of Business Ethics, Vol. 78 Nos 1-2, pp. 39-46, doi: 10.1007/s10551-006-9309-9.

Schwartz, M.S., Dunfee, T.W. and Kline, M.J. (2005), “Tone at the top: an ethics code for directors?”, Journal of Business Ethics, Vol. 58 Nos 1-3, pp. 79-100, doi: 10.1007/s10551-005-1390-y.

Shepardson, D. (2016), “VW blasted for shielding emissions documents from U.S. probe”, available at: https://www.reuters.com/article/us-volkswagen-emissions-emails-idUSKBN0UM1WZ20160108/ (accessed 12 December 2023).

Shetty, J. and Adibi, J. (2004), “The Enron email dataset database schema and brief statistical report”, Information sciences institute technical report, University of Southern California, Vol. 4 No. 1, pp. 120-128.

Silverstein, K. (2013), “Enron, ethics and today's corporate values”, available at: https://www.forbes.com/sites/kensilverstein/2013/05/14/enron-ethics-and-todays-corporate-values/?sh=214341c55ab8 (accessed 12 December 2023).

Sims, R.R. and Brinkmann, J. (2003), “Enron ethics (Or: culture matters more than codes)”, Journal of Business Ethics, Vol. 45 No. 3, pp. 243-256, doi: 10.1023/a:1024194519384.

Soltani, B. (2014), “The anatomy of corporate fraud: a comparative analysis of high profile American and European corporate scandals”, Journal of Business Ethics, Vol. 120 No. 2, pp. 251-274, doi: 10.1007/s10551-013-1660-z.

Sridharan, U.V., Dickes, L. and Royce, C.W. (2002), “The social impact of business: enron”, American Journal of Business, Vol. 17 No. 2, pp. 11-22.

Stevens, B. (2008), “Corporate ethical codes: effective instruments for influencing behavior”, Journal of Business Ethics, Vol. 78 No. 4, pp. 601-609, doi: 10.1007/s10551-007-9370-z.

Thomas, C.W. (2002), “The rise and fall of Enron”, Journal of Accountancy, Vol. 193 No. 4, pp. 41-52.

Tibshirani, R. (1996), “Regression shrinkage and selection via the LASSO”, Journal of the Royal Statistical Society: Series B (Methodological), Vol. 58 No. 1, pp. 267-288, doi: 10.1111/j.2517-6161.1996.tb02080.x.

Warren, D.E., Peytcheva, M. and Gaspar, J.P. (2015), “When ethical tones at the top conflict: adapting priority rules to reconcile conflicting tones”, Business Ethics Quarterly, Vol. 25 No. 4, pp. 559-582, doi: 10.1017/beq.2015.40.

Weber, J. (2010), “Assessing the ‘tone at the top’: the moral reasoning of CEOs in the automobile industry”, Journal of Business Ethics, Vol. 92 No. 2, pp. 167-182, doi: 10.1007/s10551-009-0157-2.

Winkler, I. (2011), “The representation of social actors in corporate codes of ethics. How code language positions internal actors”, Journal of Business Ethics, Vol. 101 No. 4, pp. 653-665, doi: 10.1007/s10551-011-0762-8.

Acknowledgements

The authors wish to thank the Social Sciences and Humanities Research Council, Canada and the Haskayne School of Business, University of Calgary, Canada for funding this project.

Corresponding author

Abu S. Rahaman can be contacted at: arahaman@schulich.yorku.ca

Related articles