h. scott asay, university of iowa

Do Features That Associate Managers With a Message

Magnify Investors’ Reactions to Narrative Disclosures?*

H. SCOTT ASAY, University of Iowa

ROBERT LIBBY, Cornell University†

KRISTINA M. RENNEKAMP, Cornell University

*The authors would like to acknowledge helpful comments from Tim Bauer, Rob Bloomfield,

Scott Emett, Brooke Elliott, Stephanie Grant, Jessen Hobson, Frank Hodge, Ken Merkley, Mark

Nelson, Mark Peecher, and Hun-Tong Tan and participants at the Cornell University, University

of Illinois, and University of North Carolina Accounting Workshops and 2013 BYU Accounting

Research Symposium.

†Corresponding author: 383 Sage Hall, Cornell University, Ithaca, New York, 14853. Phone:

607-255-3348. Fax: 607-254-4590. Email: [email protected].

Do Features that Associate Managers with a Message

Magnify Investors’ Reactions to Narrative Disclosures?

Abstract

We test whether investors react more strongly to narrative disclosures when the CEO’s presence is more

salient or prominent in the disclosure. In our first experiment, holding all other information constant, we

manipulate whether a disclosure uses more personal pronouns (e.g., “I” and “our” rather than “the

company” and “its”). Again holding all other information constant, our second experiment manipulates

whether a disclosure does or does not contain a photo of the CEO. Both manipulations lead to stronger

reactions from investors in between-subjects tests. That is, when news is positive (negative), including

either more personal pronouns or the CEO’s photo leads to more positive (negative) reactions from

investors. This provides convergent evidence for our hypothesis. We also find that, within-subjects, both

manipulations are perceived as indicating greater association with the message, but participants do not

expect an effect on investment evaluations. In our third experiment, we provide additional evidence that

personal pronoun usage affects investor reactions by increasing the perceived credibility of the disclosure.

Keywords: language, voluntary disclosure, investor judgment, Plain English

This research did not receive any specific grant from funding agencies in the public, commercial, or not-

for-profit sectors.

1

1. Introduction

Managers play a key role in communicating firm performance, and the disclosures they provide

can vary in the extent to which the manager’s presence is salient or prominent in the disclosure (hereafter,

managers’ “association with the message”). For example, some types of disclosures (e.g., conference

calls, letters to shareholders, MD&A, etc.) might associate managers more with the message than other

types of disclosures (e.g., press releases, 10-Ks, etc.), and firms are increasingly using disclosure

mediums that could more closely associate managers with the information in their disclosures (e.g., online

video disclosures as in Elliott et al. 2012, or social media disclosures as in Cade 2016). Even within these

different types of disclosures there is variation in the extent to which disclosure choices associate

managers with the message they are communicating. In this paper, we examine how two such disclosure

choices – a manager’s use of personal pronouns and the inclusion of the manager’s photo in the disclosure

– affect investors’ reactions to the information in the disclosure.

Drawing on prior work in communications and psycholinguistics (Cohn, Mehl, and Pennebaker

2004; Hyland 2005a, b; Pennebaker 2011), we predict that these disclosure choices signal to investors that

managers believe the information in the disclosure, thereby magnifying investors’ reactions to that

information. Understanding these effects is important for several reasons. First, there is a great deal of

variation in practice in the use of personal pronouns and photos. For example, recent work in the

accounting literature documents variation in managers’ use of personal pronouns within firm disclosures

(Gow, Kaplan, Larcker, and Zakolyukina 2015; Brochet, Miller, Naranjo, and Yu 2016). Likewise, the

Letters to Shareholders in the 2015 annual reports of Procter and Gamble, Walmart, and Target all include

photos of their respective CEOs, whereas the Letters to Shareholders in the 2015 annual reports for Ford,

Apple, and Kroger do not. Second, the literature has suggested a variety of reasons for this variation,

including deception, self-serving attribution bias, and manager traits such as overconfidence, narcissism,

and cultural background. Each of these explanations suggests that personal pronouns and photos may be

signals of negative managerial traits. The theory that we rely on instead focuses on investors’ reactions to

these disclosure choices, and suggests that they may magnify investors’ positive or negative reactions to

2

news. Third, prior archival work indicates that personal pronoun and photo usage may be related to firm

performance (Chatterjee and Hambrick 2007; Chen and Loftus 2017), which suggests the possibility of

strategic use. Users would benefit from knowing about the possible effects of these strategies on their

investment evaluations, especially given that psychology research suggests that they may be unaware of

these effects (Pennebaker 2011). Finally, the effects of personal pronouns are of particular interest given

that the SEC describes them as one of the components of Plain English disclosures (SEC 1998). The SEC

requires firms to use Plain English in some disclosures and encourages firms to use Plain English in all

disclosures.1 Our study examines whether doing so may have an effect on users that the SEC has not

considered.

We test our predictions using a series of experiments. In Experiment 1, we use a 2 x 2 between-

subjects design. Within an abbreviated letter to shareholders, we manipulate (1) whether the disclosure

contains good or bad news and (2) whether the CEO refers to the company in first or third person in the

final sentence of the disclosure when discussing future performance expectations (e.g., “I” and “our” vs.

“the company” and “its”). We isolate our manipulation to the final sentence in order to provide a strong

test of our theory without changing other characteristics of the disclosure. Consistent with our predictions,

we find that participants’ reactions to the disclosure are stronger when the disclosure contains more first-

person personal pronouns. That is, participants’ responses are more positive when good news is conveyed

using first-person personal pronouns in the final sentence of the disclosure and more negative when bad

news is conveyed using first-person personal pronouns in the final sentence of the disclosure. Within-

participant analyses indicate that participants do believe that personal pronouns increase managers’

association with the message, but do not anticipate an effect on valuation judgments. Findings of

valuation differences using between-subjects treatments but not using within-subjects treatments suggests

1 See, for example, SEC Rule 421(d), Rules 13a-20 and 15d-20, and SEC Releases No. 33-8998 and 34-38164. As

stated in Release No. 34-38164 “[The SEC’s] ultimate goal is to have all disclosure documents written in plain

English.”

3

that participants may be unaware of the effects of the personal pronouns in our between-subjects test

(Kahneman and Tversky 1996; Libby, Bloomfield, and Nelson 2002).

In Experiment 2, we provide convergent evidence by using a different disclosure choice to

manipulate the extent to which managers associate themselves with the message within the same

abbreviated letter to shareholders. We use a 2 x 2 between-subjects design and manipulate (1) whether the

disclosure contains good or bad news and (2) whether the CEO’s photo is present in the disclosure.

Again, consistent with our predictions, we find that participants’ reactions to the disclosure are stronger

when the CEO’s photo is included. That is, participants’ responses are more positive when a photo of the

CEO is included in a good-news disclosure and more negative when a photo of the CEO is included in a

bad-news disclosure. As in Experiment 1, within-participants analyses indicate that participants believe

that the photo increases the managers’ association with the message but do not anticipate an effect on

valuation judgments, suggesting that participants may be unaware of the effects of the photo in our

between-subjects test.

While we find that investors’ valuation judgments are affected by our manipulations in both

Experiments 1 and 2, we do not find evidence that these reactions are explained by measures of the extent

to which participants are willing to rely on the disclosure or the extent to which they think that

management believes the information in the disclosure, both of which may be thought of as proxies for

disclosure credibility. Similarly, we do not find evidence that these reactions are explained by measures

of management’s trustworthiness or competence, two dimensions of management credibility (Mercer

2004). We conduct a third experiment to provide additional evidence regarding the process underlying

our results. In Experiment 3, we use a 2 x 2 between-subjects design similar to that of Experiment 1.

Specifically, we manipulate (1) whether the disclosure contains good or bad news and (2) whether the

CEO refers to the company in the first or third person. In order to increase our ability to shed light on the

underlying process, we use a shorter disclosure, manipulate the usage of personal pronouns throughout

the disclosure (rather than in only the final sentence), and increase our sample size. Further, we also

collect additional measures to capture alternative processes that might explain our results (psychological

4

distance, perceived management control, and involvement). Results from Experiment 3 indicate that

personal pronouns increase the extent to which participants felt like they could rely on the information in

the disclosure, the extent to which participants felt like management believed the information in the

disclosure, and perceptions of management trustworthiness. These measures, in turn, led to more positive

reactions to good news and more negative reactions to bad news. In contrast, the alternative process

measures do not explain our results. These findings provide additional support for the idea that disclosure

choices that associate managers with the message increase the credibility of the disclosure and magnify

investors’ reactions to the information contained therein.

Our findings complement related accounting research that examines disclosure attributes that

magnify investor reactions through perceptions of either disclosure credibility or management credibility

(Jennings 1987; Mercer 2004). Our findings suggest that managers can enhance the credibility of their

disclosures by increasing their personal salience or prominence in the disclosure. These findings have

potentially important implications for preparers and users of disclosures, as managers might seek to

influence investors’ reactions to firm communications by selectively adopting disclosure choices that

more closely associate themselves with, or distance themselves from, the information in disclosures.

Consistent with this idea, recent archival evidence suggests that the use of personal pronouns and CEO

photos are related to firm performance (e.g., Chatterjee and Hambrick 2007; Chen and Loftus 2017).

Managers might also opt for disclosure mediums that influence their association with the information in

their disclosures (e.g., online video vs. textual disclosures as in Elliott et al. 2012) and make other choices

within a disclosure medium to associate themselves or distance themselves from the message (e.g.,

personal vs. corporate Twitter accounts).

Our findings also complement archival studies in accounting, finance, and management and have

important implications for researchers who use CEO photos and personal pronouns as proxies for

conceptual variables. For example, personal pronouns and CEO photos have both been used as proxies for

narcissism (Chatterjee and Hambrick 2007) and overconfidence (Pennebaker 2011; Schrand and Zechman

2012). In addition, personal pronouns have been used as a proxy for deception (Larcker and Zakolyukina

5

2012) and self-serving attribution bias (Li 2011; Chen and Loftus 2017). While much of the prior work

tends to view these disclosure choices as indicators of negative managerial traits, our results suggest that

users of disclosures do not view them as uniformly negative. Instead, these disclosure choices lead

participants to respond more strongly to the news in a disclosure, which results in either a more positive

or negative reaction.

Finally, our results should also be of interest to regulators who have emphasized the importance

of disclosure language (e.g., SEC 1998). As part of the Plain English initiative, the SEC encourages firms

to more frequently use personal pronouns to increase disclosure clarity (SEC 1998). However, little is

known about how personal pronoun usage might influence investors’ interpretation of, or reaction to,

disclosures. Our findings suggest one potentially unanticipated effect of encouraging an increase in

personal pronoun usage – personal pronouns may cause users to react more strongly to information in a

disclosure.

The remainder of this paper proceeds as follows. Section 2 provides background information and

develops our hypotheses. Sections 3 and 4 describe methods and results for Experiments 1 and 2,

respectively. Section 5 describes our supplemental experiment that tests the robustness of our personal

pronoun results. Section 6 concludes.

2. Background and Development of Hypotheses

The impact of narrative disclosures

Qualitative, non-numeric information makes up an important part of accounting disclosures.

Narratives in disclosures allow firms to provide additional detail and explanations for firm performance

and to convey additional information about firm strategy and goals for the future. The literature

examining the characteristics of narrative disclosures has grown recently, in part due to the electronic

availability of textual data and tools that allow these data to be analyzed efficiently (for a review, see Li

2010a). We classify research on the qualitative characteristics of disclosures as primarily focusing on

disclosure content or disclosure style. Content refers to what information is explicitly stated (e.g., “Sales

6

in consumer markets increased”). Style refers to how that information is communicated and has little

impact on the literal meaning of communications.

Disclosure content

Many of the studies of narrative disclosure characteristics examine what these disclosures

explicitly state by developing measures based on nouns, adjectives, regular verbs, and adverbs. Since

linguists commonly refer to these types of words as “content” words, we refer to these studies as

examining disclosure content choices. Content words label objects or actions, typically have a culturally

shared meaning, and are “absolutely necessary to convey an idea to someone else” (Pennebaker 2011,

21). If content choices are removed from a communication, the communication becomes meaningless.

Existing studies of disclosure content choices focus on attributes including disclosure tone (e.g., Kothari,

Li, and Short 2009; Feldman, Govindaraj, Livnat, and Segal 2010; Li 2010b; Davis, Piger, and Sedor

2012; Davis and Tama-Sweet 2012; Demers and Vega 2014; Huang, Teoh, and Zhang 2014), language

vividness (Hales, Kuang, and Venkataraman 2011), and disclosure quantity (e.g., Merkley 2014; see

Merkl-Davies and Brennan 2007 and Li 2010a for recent reviews).

Disclosure style

Recent studies have also begun to examine how disclosures are communicated, which we refer to

as disclosure style choices. Relative to disclosure content choices, disclosure style choices have little

effect on the literal meaning of the disclosure.2 Style choices that have received attention in recent studies

include disclosure attributes such as disclosure readability (Li 2008; You and Zhang 2009; Miller 2010;

Rennekamp 2012; Tan et al. 2014; Tan et al. 2015), disclosure medium (Elliott et al. 2012), and emphasis

(Bowen, Davis, and Matsumoto 2005; Elliott 2006). This literature suggests that style choices can have

important effects on user decisions. For example, empirical evidence suggests that bad news disclosures

tend to be longer and less readable than good news disclosures (Li 2008). Further, longer, less readable

2 That is, the meaning of communications can still largely be understood if style choices are removed from a

communication. Our distinction between content choices and style choices is adapted from Pennebaker’s (2011)

classification of content words and style words.

7

disclosures are associated with lower trading and weaker reactions (You and Zhang 2009; Miller 2010;

Rennekamp 2012; Tan et al. 2014; Tan et al. 2015) and increase investors’ propensity to seek out and rely

on outside sources of information about the firm (Asay et al. 2017).

Recent work has also examined the usage of personal pronouns and CEO photos. Concerning

personal pronoun usage, research suggests that their use might reflect managerial narcissism (Chatterjee

and Hambrick 2007), deception (Larcker and Zakolyukina 2012), or managers’ susceptibility to self-

serving attribution bias (Li 2011; Chen and Loftus 2017). The prevalence of CEO photos has been

interpreted as indicating managerial overconfidence (Schrand and Zechman 2012) or managerial

narcissism (Chatterjee and Hambrick 2007; Olsen, Dworkis, and Young 2013). While these studies have

treated personal pronouns and CEO photos as proxies for managerial traits, we are interested in

examining how users react to these style choices in an experimental setting that controls for underlying

firm characteristics.

Disclosure credibility and association with the message

Prior work argues that users’ reactions to a disclosure depend, in part, on the disclosure’s

credibility (Jennings 1987; Mercer 2004), which Mercer (2004) defines as “investors’ perceptions of the

believability of a particular disclosure” (p. 186). This literature suggests that the market responds more

strongly to more credible disclosures (Hutton and Stocken 2009; Ng, Tuna, and Verdi 2013) and that

disclosure credibility is influenced by what is communicated as well as by how it is communicated. For

example, disclosures are more credible if they provide more precise information (Hassell et al. 1988;

Hirst, Koonce, and Miller 1999; Hirst, Koonce, and Venkataraman 2007) or additional supporting

information (Hutton, Miller, and Skinner 2003; Baginski, Hassell, and Kimbrough 2004). Holding

constant what is explicitly stated in the disclosure, disclosure credibility is also influenced by the

disclosure’s time horizon (Pownall et al. 1993), medium (Elliott et al. 2012), and the credibility of the

individual or firm providing the disclosure (Hirst, Koonce, and Miller 1999; Hodge, Hopkins, and Pratt

2006).

8

We suggest that the extent to which managers choose to associate themselves with a disclosure

can play a similar role. When managers are highly associated with a disclosure, they assume greater

responsibility for its content. For example, statements made by managers can be directly attributed to

them and presumably reflect their views or beliefs. In contrast, when statements are made by “the

company,” responsibility is more diffuse. Because disclosures that turn out to be inaccurate (e.g.,

restatements, missed forecasts, etc.) elicit negative market reactions and can damage managers’

reputations (Desai, Hogan, and Wilkins 2006; Hutton and Stocken 2009), managers may be hesitant to

make their presence salient or prominent in the disclosure unless they are highly confident in the accuracy

of its content. This suggests that managers are able to signal the extent to which they believe a

disclosure’s content by choosing to associate themselves with (or distance themselves from) that

disclosure. As a result, disclosures that feature the manager more prominently might elicit stronger

reactions from investors by increasing their willingness to rely on the information therein.

This view is supported by prior communications research suggesting that self-mention (i.e.,

greater use of personal pronouns) is a rhetorical device that associates authors with their arguments

(Hyland 2001). Moreover, this literature suggests that self-mention emphasizes that the author believes

what they are saying (Hyland 2005a, b). Consistent with this idea, prior psycholinguistics research argues

that personal pronouns are reliable markers of a speaker’s belief in their message (Pennebaker 2011).

Prior psycholinguistics literature also finds that individuals with a desire to distance themselves from an

event use fewer first-person pronouns (Cohn, Mehl, and Pennebaker 2004).

Because the inclusion of a photo associates a manager more closely with a disclosure message,

the presence or absence of a manager’s photo could also reflect the manager’s degree of belief in the

information. Further, photos help individuals envision communicators actually speaking (Jameson 2000),

so a CEO’s photo may serve as an important cue to investors that the words represent the CEO’s

interpretation of the information and therefore their belief in the validity of the message. The

psycholinguistics, managerial narcissism, and managerial overconfidence literatures are largely silent on

how investors would respond to variation in managers’ pronoun or photo usage.

9

In accounting, our predictions are consistent with prior literature showing that style choices can

magnify investor reactions to disclosures. Sedor (2002), Rennekamp (2012), and Tan et al. (2014, 2015)

focus on the effects of style changes in the message. Elliott et al. (2012) show that video disclosures

increase participants’ reactions to the disclosure, even beyond the effects of a photo. They predict and

find that video increases participants’ reactions to the disclosure by increasing the salience of the CEO’s

perceived trustworthiness. Although Elliott et al. (2012) do not focus on how their video manipulations

associate managers with the disclosure, video disclosures may further associate a manager with a message

above and beyond the inclusion of a photo in a textual disclosure.3 Thus, our predictions may also offer an

interesting interpretation of results in prior literature.

In summary, we suggest that style choices that more closely associate managers with the

information in their disclosures (e.g., greater use of personal pronouns or the inclusion of a CEO’s photo)

will magnify investors’ reactions to firm disclosures.

HYPOTHESIS 1. Greater use of personal pronouns in a narrative disclosure will

magnify investor reactions to the news in the disclosure.

HYPOTHESIS 2. The inclusion of a CEO’s photo in a narrative disclosure will magnify

investor reactions to the news in the disclosure.

Disclosure design goals

Our experiments present participants with a short letter to shareholders and ask them to provide a

valuation judgment based on background information and the letter. Our first major goal in the design of

the letter is to vary the sign of the news without varying other content or style elements. We manipulate

the sign of news by substituting antonyms for the important nouns, adjectives, and adverbs used in the

3Our manipulation and contribution differs from Elliott et al. (2012) in several important respects. First, Elliott et al.

(2012) expose all participants to a photo of the CEO in their study, to ensure that all participants are familiar with

the CEO’s appearance. Second, the use of audio and video allows participants to make inferences about verbal and

non-verbal cues, which could include information about a firm’s future performance (Mayew and Venkatachalam

2012), and convey more than our construct of interest, association with the message. Finally, their theory relies more

on users’ “rich sensory experience provided by video”, and conscious interpretations of the trustworthiness of

managers. In contrast, our manipulation lacks the richness of video and our predictions do not rely on sensory

experience.

10

letter to indicate the sign of the news (above-below; higher-lower; strength-weakness; positively-

negatively; increased-decreased; grow-slow; satisfied-dissatisfied; improvement-deterioration). Our

second goal is to vary personal pronoun use (Experiments 1 and 3) or photo inclusion (Experiment 2)

without affecting disclosure content (nouns, adjectives, regular verbs, and adverbs) or other style

attributes, such as readability (SEC 1998). To accomplish this goal, we change only the last sentence of

the letter in Experiment 1 and include/exclude a single photo with no changes in the wording of the letter

in Experiment 2. In Experiment 3, we manipulate personal pronoun usage throughout a shorter disclosure,

increase our sample size, and collect other process measures in order to increase our ability to provide

insight about the process underlying our results. The stimuli are kept short and require only 3 to 4 minutes

to read and evaluate.

3. Experiment 1

Method

Design

In Experiment 1, we use a 2 x 2 between-subjects design, and manipulate (1) whether an

abbreviated letter to shareholders for a fictitious soft drink company conveys good or bad news about the

firm and (2) the rate of personal pronoun usage in the letter. Specifically, in the final sentence of the letter

to shareholders containing good (bad) news,4 the CEO indicates one of the following (italics added here

to highlight the pronoun manipulations; see Appendix 1 for illustrations of our Good News conditions):

“The Company is satisfied (dissatisfied) with its accomplishments in 2012, and expects further

improvement (deterioration) in operations in the near future.”

“I am satisfied (dissatisfied) with our accomplishments in 2012, and I expect further improvement

(deterioration) in operations in the near future.”

4 In the bad news condition, each of the words in parentheses in the examples is substituted for the immediately

preceding word.

11

The CEO’s photo is not included in any of the disclosures in Experiment 1, and the language used

throughout the disclosure is identical throughout the disclosure until the final sentence.5

Participants

Participants in Experiment 1 are 138 individuals recruited from Amazon’s Mechanical Turk

platform. Since our task requires that participants be able to read and understand English and pay careful

attention to the task, we recruit participants who are native English speakers and have an approval rate of

at least 95 percent.6 On average, participants are 32 years old and have 11 years of full-time work

experience. Fifty-five percent are male. Participants take an average of 3.2 minutes to complete the task

and receive $0.25 of compensation.

Procedures

Participants are randomly assigned to one of the four experimental conditions. Participants begin

the study by reading background information about a fictitious company (Cooper Soda Co.). They then

provide an initial judgment about the appropriate common stock valuation for the firm. Next, they view

the abbreviated letter to shareholders that contains our manipulations. After reviewing the letter,

participants again indicate their beliefs about the appropriate common stock valuation of the firm. They

then answer several debriefing and demographic questions. Once they move on to the next step, they are

prohibited from changing their answers to earlier parts of the task.

Dependent measure

Following prior literature, we elicit valuation judgments by asking participants to indicate on a

101-point scale what they believe to be an appropriate common stock valuation for the firm, ranging from

0 (“low”) to 100 (“high”) (Koonce and Lipe 2010). Participants make this judgment before and after

5 The abbreviated letter to shareholders is adapted from Rennekamp (2012).

6 We initially recruited 150 participants for Experiment 1. To prevent Workers from having multiple accounts, AMT

requires U.S. based workers to provide their social security number or individual tax identification number. In

addition, AMT prevents Workers from participating multiple times on the same task, and Qualtrics (which we use to

distribute our survey) uses cookies to prevent participants from participating multiple times in the same study. 10

participants were excluded for being repeat participants from pilot testing, and two participants were excluded for

being non-Native English speakers. Results from testing our predicted interaction are still significant if we include

all 150 participants.

12

viewing the letter to shareholders that contains our manipulations, and the difference between the two

measures serves as our primary dependent measure. This change measure captures the direction and

magnitude of participants’ reactions to the disclosure. Eliciting our dependent measure in this way is

designed to reduce noise in participants’ final judgments by starting them from a similar baseline of

beliefs, which should help us better detect the effects of our manipulations.

Follow-up questions

Following the valuation judgment, participants respond to several additional between-subjects

questions intended to measure whether pronoun use increases the extent to which participants believe (1)

that they could rely on the disclosure (“I felt like I could rely on the information in the Letter to

Shareholders”), (2) that the manager believes the disclosure’s message (“I felt like the management of

Cooper Soda really believed the information in the Letter to Shareholders”), and (3) that the manager is

competent (“How competent or incompetent do you believe the management of Cooper Soda to be?”) and

(4) that the manager is trustworthy (“How trustworthy or untrustworthy do you believe the management

of Cooper Soda to be?”). The first two measures test whether our manipulations affect participants’

perceptions about credibility of the disclosure itself (investors’ willingness to rely on the statements and

the strength of managers’ belief in the information), whereas the third and fourth measures test whether

our manipulations affect participants’ perceptions of the credibility of the management delivering the

information (competence and trustworthiness) (e.g., Mercer 2005; Koonce and Lipe 2010).

Results

Between-subjects tests

Our expectation is that investors will react more positively (negatively) to good (bad) news when

managers associate themselves with the disclosure by using personal pronouns at a higher rate in the

disclosure. Descriptive statistics for our primary dependent measure are presented in panel A of Table 1.7

<INSERT TABLE 1 HERE>

7 There are no significant differences in initial valuation judgments (i.e., before participants view the disclosure

containing our manipulations) across conditions (p=0.861, two-tailed).

13

Consistent with an effective manipulation of news valence, we find that participants adjust their

valuations positively after viewing the good news disclosure (p < 0.001, one-tailed, untabulated) and

negatively after viewing the bad news disclosure (p < 0.001, one-tailed). In addition, we find that this

effect is magnified when the CEO uses personal pronouns at a higher rate in the disclosure (p = 0.034,

one-tailed) (see panel B of Table 1).8 Further, we also find that the pronouns lead to both more negative

reactions for the bad news disclosure (p = 0.097, one-tailed), and more positive reactions for the good

news disclosure (p = 0.099, one-tailed), as indicated in panel C of Table 1.

Potential mediators

Participants perceive that management believes the bad news more than the good news (p =

0.003, two-tailed), that they can rely more on the bad news than good news (p < 0.001, two-tailed), and

that management is more competent in the good news condition than the bad (p < 0.001, two-tailed).

News does not affect the perceived trustworthiness of management (p = 0.360, two-tailed). In sum, the

sign of the news affects our measure of the strength of the manager’s belief in the message, participants’

willingness to rely on the information in the disclosure, and one of the two measures of manager

credibility. However, pronoun usage does not have a significant main effect on any of these possible

mediators: how strongly they agree that management believes the message being conveyed, how strongly

they believe that they can rely on the information, or perceptions of management’s competence and

trustworthiness (p = 0.656, p = 0.441, p = 0.528, and p = 0.310, two-tailed, respectively). Personal

pronoun usage also does not interact with news valence to affect any of these additional measures (all p >

0.409, two-tailed).9 Thus, while the overall pattern of results is consistent with H1, these measures do not

shed much light on the theoretical process underlying these results.

8 Note that all reported one-tailed p-values associated with tests using the F-distribution are one-tailed equivalents,

given our directional predictions. 9 After viewing the disclosures that included our manipulations, we also asked participants about their perceptions of

the firm’s future earnings, as well as the desirability of the firm overall as an investment. While correlated with our

primary valuation judgments (ρ >0.85 across both measures), these judgments are noisier in that we did not elicit a

pre-manipulation judgment to serve as a control. For both the measure of perceptions of future earnings and

14

Within-subjects tests

We next use a within-subjects study to test whether our pronoun manipulation operationalizes our

construct of interest and to examine whether participants consciously expect the manipulation to affect

valuation judgments. Participants for the within-subjects tests are also recruited from AMT (with the

requirement that they did not complete our primary experiments) and are randomly assigned to either the

good news or bad news condition. To facilitate comparison, we present participants with the disclosure

containing more personal pronouns side-by-side with the disclosure containing fewer personal pronouns.

The order of the two letters is varied and has no effect on the results. Participants rate their agreement

with the statement, “The letter that includes more personal pronouns (e.g., “I” and “our”) more closely

associates the CEO with the message contained in the letter” (on a 6-point scale ranging from 1 = strongly

disagree to 6 = strongly agree). Values of 4, 5, and 6 on the scale represent responses in agreement with

the statement. 98 percent of participants agree that the disclosure containing more pronouns more closely

associates the CEO with the message in the disclosure (p < 0.001, one-tailed). We interpret these findings

as supporting the validity of our operationalization of association with the message. In addition,

participants agree, on average, that the two disclosures would lead to the same valuation judgments by

investors (p < 0.001, one-tailed). 72 percent of the participants believe that investors would make the

same valuation in both cases. This suggests that the majority are not aware of the effects of personal

pronouns on investment evaluations, even when variation in pronoun usage is made obvious to them by

the side-by-side presentation.

Summary

In summary, we find that personal pronoun use magnifies participants’ reactions to the valence of

the message in a disclosure. Personal pronouns even enhance the effect of bad news, which is judged by

our participants to be more credible than good news. Participants’ beliefs about disclosure credibility and

management credibility do not mediate the effects on valuation judgments, and participants in our within-

desirability of the firm as an investment, results are directionally consistent with the valuation measure, although

insignificant.

15

subjects analysis do not expect personal pronoun use to affect investment evaluations. Together, this

evidence suggests that the between-subjects effect may be subconscious, consistent with prior research

suggesting that individuals often lack awareness of others’ or even their own use of personal pronouns

(Pennebaker 2011).

4. Experiment 2

Method

Design and procedures

In Experiment 2, we use a 2 x 2 between-subjects design and manipulate (1) whether an

abbreviated letter to shareholders conveys good or bad news about the firm and (2) whether the letter

contains a photo of the CEO (see Appendix 2 for illustrations of our Good News conditions). All other

aspects of the task and procedure are identical to Experiment 1.

The photo we use portrays a smiling CEO. To ensure that this is consistent with CEO photos in

practice, we examine CEO photos contained in the letters to shareholders for firms in the S&P 500. We

rank firms by their change in earnings for the most recent year (scaled by market cap) and categorize the

100 firms in the top (bottom) quintile as a “high performer” (“low performer”). On average, high

performers have a positive change in earnings while low performers have a negative change in earnings.

From this sample of 200 firms, we then collect the first CEO photo (if any) contained in their letter to

shareholders.

After collecting photos, two independent raters who are unaware of our hypotheses look at each

photo from our sample of S&P 500 firms individually, as well as the CEO photo we use in our study, and

indicate how happy each CEO appears in the photo on a 9-point scale (1 = “Very Unhappy”, 5 =

“Neutral”, and 9 = “Very Happy”; Intra-class correlation = 0.81). Performance does not affect the

likelihood of issuing a letter to shareholders (p = 0.480, two-tailed), but low performers are more likely to

include a photo in their letter to shareholders (p = 0.051, two-tailed). On average, CEOs are rated above

the midpoint (mean = 6.39; median = 6.5; p < 0.001, two-tailed), suggesting they generally look happy in

their photos. This result holds for low performers (mean = 6.43; median = 6.5; p < 0.001, two-tailed) and

16

high performers (mean = 6.34; median = 6.5; p < 0.001, two-tailed), and this difference is not significant

(p = 0.660, two- tailed). Finally, both raters rate our CEO as a “7”, which is about one-half of one

standard deviation above the mean rating for both high and low performers in our S&P 500 sample.10

Overall, these results suggest our photo is well within the normal range for CEO photos that appear in

letters to shareholders following both good and bad performance.

Participants

Participants in Experiment 2 are 104 Workers recruited from Amazon’s Mechanical Turk

platform.11

We again restrict our sample to Workers within the United States who are native English

speakers and have an approval rate of at least 95 percent. On average, participants are 32 years old and

have 12 years of full-time work experience. 53 percent are female. Participants take an average of 3.5

minutes to complete the task and receive $0.25 of compensation.

Results


Our expectation is that investors will react more positively to good news and more negatively to

bad news when the manager associates himself with the disclosure by including his photo in the

disclosure. Descriptive statistics for our primary dependent measure are presented in panel A of Table 2.12



valuations positively after viewing the good news disclosure (p < 0.001, one-tailed) and negatively after

viewing the bad news disclosure (p < 0.001, one-tailed). In addition, we find that this effect is magnified

10

The photo that we use in our study is presented randomly within the sample of real photos from S&P 500 firms.

The two raters are unaware that the photo used in our study is not of a real CEO. 11

We initially recruited 120 participants for our task, although 121 participated. 16 participants have the same IP

address or AMT ID as participants from pilot testing, and these participants are excluded from analyses. We further

exclude one participant that identifies as a non-native English speaker because our task requires that participants be

able to read and interpret a disclosure. Results from testing for our predicted interaction are still significant if all

participants are included. 12

There are no significant differences in initial valuation judgments (i.e., before participants view the disclosure


17

when the CEO’s photo is included in the disclosure (p = 0.005, one-tailed) (see panel B of Table 2).

Further, as indicated in panel C of Table 2, we find that the photo leads to more positive reactions for the

good news disclosure (p = 0.098, one-tailed), but more negative reactions for the bad news disclosure (p =

0.009, one-tailed).

Mediators

News valence significantly affects only the two management credibility measures (p < 0.002,

two-tailed). Consistent with Experiment 1, the photo does not affect participants’ perceptions about how

strongly management believes the message being conveyed (p=0.395, two-tailed). Similarly, we find no

evidence that the photo increases their rating of the degree to which they could rely on the information (p

= 0.921, two-tailed), or perceptions of management’s competence or trustworthiness (p = 0.504 and p =

0.566, two-tailed, respectively), the two common components of perceived credibility of management in

prior research. The photo also does not interact with news valence to affect any of these additional

measures (all p > 0.365, two-tailed).13

Within-Subjects Tests

We next use a within-subjects study to test whether our photo manipulation operationalizes our

construct of interest and to examine whether participants consciously expect the manipulation to affect

valuation judgments. Participants for the within-subjects tests are also recruited from AMT (with the

requirement that they did not complete our primary experiments) and are randomly assigned to either the

good news or bad news condition. As in our personal pronoun study, we present participants with the

disclosure containing a photo side-by-side with the disclosure without the photo. Participants rate their

agreement with the statement, “The letter that includes the photo of the CEO more closely associates the

CEO with the message contained in the letter” (on a 6-point scale ranging from 1 = strongly disagree to 6

= strongly agree). Values of 4, 5, and 6 on the scale represent responses in agreement with the statement.

13

Consistent with the results for the main dependent variable, we find that the photo magnifies participants’

perceptions of future earnings (p = 0.046, one-tailed). The photo does not affect the perceived desirability of the

company as an investment opportunity (p = 0.872, two-tailed).

18

89 percent of the participants agree that the disclosure containing the CEO’s photo more closely

associates the CEO with the message in the disclosure (p < 0.001, one-tailed). We interpret these findings

as supporting the validity of our operationalization of association with the message. In addition,

participants agree, on average, that the two disclosures would lead to the same valuation judgments by

investors (p < 0.001, one-tailed). 69 percent of the participants believe that investors would make the

same valuation in both cases. This suggests that the majority are not aware of the effects of CEO photos

on investment evaluations, even when their usage is made obvious to them.

Summary

These results complement our results from Experiment 1. Specifically, we operationalize our

construct of interest (association with the message) using a very different style choice. Consistent with

our predictions, we document a similar pattern of results, as each style choice magnifies participants’

reactions to good and bad news disclosures. As in Experiment 1, responses to our follow-up questions do

not provide additional insight with respect to the process underlying these results. The within-subjects

tests and the mediation analysis using follow-up questions may reflect participants’ lack of insight into

the underlying process (Griffith et al. 2015).

5. Experiment 3

Our primary results from Experiments 1 and 2 are consistent with our predictions. However,

participants’ responses to follow-up questions did not provide additional insight into the underlying

process in either experiment. There are at least three possible explanations for this inconsistency. First,

it’s possible that participants have relatively little self-insight about the cognitive processes underlying

their judgments (Nisbett and Wilson 1977; Pennebaker 2011; Griffith et al. 2015). Second, the subtle

nature of our manipulations may limit our ability to observe the effects on our measures of the intervening

process. And third, it’s possible that our primary results are driven by an alternative process. For example,

managers’ association with the message might operate by changing participants’ psychological distance

from the firm or by changing perceived management control and involvement. We shed some additional

light on this issue in Experiment 3.

19

Method

Design and procedures

In Experiment 3, we provide an additional test of our theory using a 2 x 2 experiment. As in

Experiment 1, we manipulate (1) whether an abbreviated letter to shareholders conveys good or bad news

about the firm and (2) the rate of personal pronoun usage in the letter. In order to increase our ability to

provide insight about the process underlying our results, we use a shorter disclosure and manipulate

personal pronoun usage throughout that disclosure (rather than in only the final sentence) (see Appendix

3). We also increase our sample size and collect process measures to capture participants’ psychological

distance from the firm, perceptions of management control, and perceptions of management involvement.

Experimental procedures are otherwise identical to Experiments 1 and 2.

Follow-up questions

Following the valuation judgment, participants respond to several additional between-subjects

questions. As in Experiments 1 and 2, participants indicate the extent to which (1) they could rely on the

disclosure, (2) they believe that the manager believes the disclosure’s message (3) they believe that the

manager is competent, and (4) that the manager is trustworthy. In addition, participants rate the extent to

which they “believe the CEO has control over the performance of Cooper Soda” and the extent to which

they “believe the CEO is involved in the operations of Cooper Soda.” Participants are also asked to

estimate (in miles) the distance from their current location to Cooper Soda’s headquarters in Portland

Oregon. The difference between their estimate and the actual distance calculated from their IP address

serves as a measure of psychological distance (Elliott, Rennekamp, and White 2015). The order of these

follow-up questions is randomized between subjects.

Participants

Participants in Experiment 3 are 302 Workers recruited from Amazon’s Mechanical Turk

platform. We again restrict our sample to Workers within the United States who are native English

20

speakers and have an approval rate of at least 95 percent.14

On average, participants are 37.2 years old.

Fifty-six percent are female. Participants take an average of 4.92 minutes to complete the task and receive

$0.50 of compensation.

Results


Our expectation is that investors will react more positively to good news and more negatively to

bad news when the manager associates himself with the disclosure. Descriptive statistics for our primary

dependent measure are presented in panel A of Table 3.15



valuations positively after viewing the good news disclosure (p < 0.001, one-tailed) and negatively after

viewing the bad news disclosure (p < 0.001, one-tailed). In addition, we find that this effect is magnified

when personal pronouns are used at a higher rate in the disclosure (p = 0.069, one-tailed) (see panel B of

Table 3). As indicated in panel C of Table 3, we find that reactions to the good news disclosure are

directionally more positive when the disclosure contains more personal pronouns, though this difference

is not significant (p = 0.256, one-tailed). In contrast, reactions to the bad news disclosure are more

negative when the disclosure contains more personal pronouns (p = 0.075, one-tailed).

Mediators

Disclosure Credibility: Consistent with the idea that greater use of personal pronouns increases

disclosure credibility, we find that greater use of personal pronouns increases the extent to which

participants felt they could rely on the disclosure (p = 0.036, two-tailed) and the extent to which

participants felt like the manager believed the information in the disclosure (p = 0.072, two-tailed). When

news is good, participants’ valuation judgments are positively associated with each of these measures

14

We exclude one Worker who is not a native English speaker. Inferences are unchanged if the Worker is included. 15

There are no significant differences in initial valuation judgments (i.e., before participants view the disclosure


21

(both p < 0.001, two-tailed). In contrast, when news is bad, participants’ valuation judgments are

negatively associated with each of these measures (both p < 0.001, two-tailed). A principal component

analysis for these two measures indicates that they load on a single factor (eigenvalue = 1.784, variance

explained = 89.2%), and the news contained in the disclosure moderates the effect of this factor on

participants’ valuation judgments (F(5, 295) = 37.18; p < 0.001, two-tailed). A path analysis confirms that

greater pronoun usage increases valuation judgments through this factor when news is good (z = 1.77; p =

0.077, two-tailed) and decreases valuation judgments through this factor when news is bad (z = -1.86; p =

0.062, two-tailed) (see Figure 1).16

These findings suggest that greater use of personal pronouns affects

participants’ valuation judgments by increasing disclosure credibility.

[INSERT FIGURE 1]

Management Credibility: Greater use of personal pronouns also increases the extent to which

participants view management as trustworthy (p = 0.030, two-tailed). Pronoun use does not affect

participants’ perceptions of management’s competence (p = 0.453, two-tailed). When news is good,

participants’ valuation judgments are positively associated with participants’ perceptions of management

trustworthiness (p = 0.011, two-tailed). However, when news is bad, participants’ valuation judgements

are not associated with their perceptions of management trustworthiness (p = 0.982, two-tailed).

Management Control and Involvement: Personal pronoun use does not affect the extent to

which participants believe the CEO has control over firm performance (p = 0.568, two-tailed) or the

extent to which participants believe the CEO is involved in the operations of the firm (p = 0.978, two-

tailed).

Psychological Distance: Greater use of personal pronouns also marginally increases the

perceived distance between participants’ current location and the firm’s headquarters (p = 0.087, two-

tailed), suggesting that greater personal pronoun usage might lead to greater psychological distance.

16

Following Hayes (2013), we estimate these conditional indirect effects using a moderated mediation model based

on 10,000 bootstrap samples.

22

However, this measure is not associated with participants’ valuation judgments when news is good (p =

0.217, two-tailed) or when news is bad (p = 0.746).

Summary

Overall, these results provide additional support for our predictions. Across three experiments, we

generally find that investors react more strongly to disclosures that feature the manager more

prominently.17

In addition, the results of the process measures in Experiment 3 are supportive of the idea

that managers’ association with the message increases the perceived credibility of the disclosure, which in

turn leads to stronger reactions to the information in the disclosure.

6. Conclusion

We predict that disclosure choices that more closely associate managers with the message in their

disclosures should magnify investors’ reactions to disclosures. Consistent with this theory, we find that

participants in within-subjects tests interpret both first-person pronoun use and inclusion of a CEO photo

as indicating greater association with the message. Between-subjects, we find that investors’ reactions to

disclosures are magnified when those disclosures use first-person personal pronouns at a higher rate or

contain a CEO’s photo, leading investors to react more positively when news is good, but more

negatively when news is bad. Within-subjects tests indicate that participants do not believe that their

investment decisions would be affected by either the use of pronouns or photos in the disclosures. The

combination of effects using between-subjects treatments but not using what are normally more powerful

within-subjects treatments, as has been demonstrated here, suggests that the between-subjects differences

are unintentional (Kahneman and Tversky 1996; Libby, Bloomfield, and Nelson 2002; Libby,

17

This conclusion is further supported by a supplemental experiment in which we again manipulate whether a

disclosure contains good or bad news and the use of personal pronouns. The disclosure consists of a short forward-

looking claim about the likely outcome of a company lawsuit. Participants are 50 accounting students from a top

undergraduate program, the dependent variable is a direct likelihood measure of belief as opposed to a valuation

judgment, and the first-person pronoun used is “we” (rather than “I”). We again find that increased use of personal

pronouns magnifies participants’ reactions to the disclosure (p < 0.001, one-tailed). Specifically, we find that

pronoun usage decreases participants’ likelihood assessments when news is bad (p = 0.010, one-tailed), but

increases likelihood assessments when news is good (p = 0.009, one-tailed). These results suggest that the results

from Experiments 1 and 3 are not specific to the AMT participants, the valuation dependent variable, or the use of

“I” versus “we.”

23

Rennekamp, and Seybert 2015). The process measures collected in Experiment 3 provide additional

support for the idea that managers’ association with the message increases disclosure credibility and

generally do not provide support for other potential mediators (e.g., management credibility, management

control and involvement, or psychological distance).

Our paper makes several contributions. First, these findings are related to prior research on the

inclusion of credibility-enhancing devices in disclosures. For example, prior research has examined

disclosures that include more forward-looking verifiable statements (Hutton, Miller, and Skinner 2003),

greater disaggregation of forecast components (Hirst, Koonce, and Venkataraman 2007; Lansford, Lev,

and Tucker 2013; Merkley, Bamber, and Christensen 2013), or less self-serving attributions (Baginski,

Hassell, and Hillison 2000; Baginski, Hassell, and Kimbrough 2004). While the judgment effects we

predict are similar to that of these credibility-enhancing devices, our paper differs from prior studies in

that most of the credibility-enhancing devices examined in these studies involve content, rather than style

choices. Further, our participants do not appear to consciously react to our pronoun and photo

manipulations.

Second, prior work in accounting, finance, and management has used CEO photos and personal

pronouns as proxies for unobservable management characteristics like narcissism and overconfidence,

and these characteristics have been interpreted predominantly as negative. Our findings suggest that

investors do not react to personal pronouns and photos as negative indicators of narcissism and

overconfidence, as those would likely predict a negative main effect on investment decisions, not the

interaction we find. Similarly, while photos with certain characteristics may elicit a positive or negative

response, these characteristics should also result in a main effect, and do not explain our interaction

pattern of results. Our pattern of results also cannot be explained by investors reacting to perceptions that

managers’ attributions in the disclosure were self-serving, given that we hold attributions constant across

both good and bad news conditions. Our theory and evidence presents a more nuanced view of the effect

of these style choices – style choices that more closely associate managers with their message magnify

investors’ reactions to disclosures.

24

Third, the SEC explicitly recommends the use of personal pronouns in firm disclosures (SEC

1998), suggesting a positive view of personal pronouns. Our results suggest one potential unintended

consequence of this recommendation. Because this style choice increases managers’ association with the

message, increasing personal pronoun usage might also magnify non-professional investors’ reactions to

disclosures. If managers’ use of personal pronouns or photos does not reflect the strength of their belief in

the message conveyed by the disclosure, this increased reaction may be unwarranted.

Finally, our paper also contributes to the broader literature on how style choices in corporate

disclosures affect investors’ judgments and decisions. Prior literature has either measured (see Li 2010a)

or manipulated (Sedor 2002; Elliott et al. 2012; Rennekamp 2012; Tan et al. 2014, 2015; Elliott,

Rennekamp, and White 2015; Asay et al. 2017) features like readability, concreteness, and/or disclosure

medium. We contribute to this literature by demonstrating that style choices that associate management

with the message have similar effects on non-professional investors’ judgments. A fruitful direction for

future research is to examine other style differences that may affect association with the message. For

example, managers could more closely associate themselves with a written disclosure message by

mentioning their names more frequently in the disclosure, including more direct quotes, or more clearly

indicating themselves as the speaker or narrator of the message. Managers’ association with the message

may also be influenced by disclosure medium and disclosure venue. For example, verbal disclosures (e.g.,

online videos or conference calls) may increase association with the message to a greater extent than

written disclosures, and disclosures provided via social media may increase association with the message

by making the disclosure seem more like a direct communication from management (e.g., a CEO vs.

corporate Twitter account). Thus, future work might extend our results by providing additional evidence

on features that associate management with the information in their disclosures. We leave it to future

research to investigate how investors react to other features that more closely associate managers with a

disclosure message, and what might be driving these responses.

Our disclosures were designed to be short, and our participants’ evaluations can best be described

as initial impressions. Given that initial impressions can be formed very quickly, and can have strong and

25

lasting impressions on subsequent evaluation of information (Russo, Meloy, and Wilks 2000; Wilks

2002), our results suggest that very subtle changes in disclosure features could have relatively strong

effects on investors’ reactions to information in the disclosure. However, the exact nature of their effects

on interpretation of more complex disclosures is a topic for future research.

26

Appendix 1 – Good News Conditions for Experiment 1*

Fewer Personal Pronouns: More Personal Pronouns:

* The bolded words in the final sentence of the disclosures were not bolded in the actual experiment. Emphasis has been added here to highlight the subtle

differences between the two disclosures.

27

Appendix 2 – Good News Conditions for Experiment 2

Photo Absent: Photo Present:

28

Appendix 3 – Good News Conditions for Experiment 3*

Fewer Personal Pronouns: More Personal Pronouns:

29

REFERENCES

Asay, S., W. B. Elliott, and K. Rennekamp. 2017. Disclosure readability and the sensitivity of investors’

valuation judgments to outside information. The Accounting Review 92 (4): 1-25.

Baginski, S. P., J. M. Hassell, and W. A. Hillison. 2000. Voluntary causal disclosures: tendencies and

capital market reaction. Review of Quantitative Finance and Accounting 15 (4): 371-389.

Baginski, S. P., J. M. Hassell, and M. Kimbrough. 2004. Why do managers explain their earnings

forecasts? Journal of Accounting Research 42 (1): 1-29.

Bowen, R. M., A. K. Davis, and D. A. Matsumoto. 2005. Emphasis on pro forma versus GAAP earnings

in quarterly press releases: Determinants, SEC intervention, and market reactions. The

Accounting Review 80 (4): 1011-1038.

Brochet, F., Miller, G., Naranjo, P., and G. Yu. 2016. Managers’ cultural background and disclosure

attributes. Working paper.

Cade, N. 2016. Corporate social media: How two-way disclosure channels influence investors. Working

paper.

Chatterjee, A. and D. C. Hambrick. 2007. It’s all about me: Narcissistic chief executive officers and their

effects on company strategy and performance. Administrative Science Quarterly 52 (3): 351-386.

Chen, Z. and S. Loftus 2017. Managers’ Self-inclusive language in conference calls: Multi-method

evidence. Working paper. Available at: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2950702

Cohn, M. A., M. R. Mehl, and J. W. Pennebaker. 2004. Linguistic markers of psychological change

surrounding September 11, 2001. Psychological Science 15 (10): 687-693.

Davis, A. K., J. M. Piger, and L. M. Sedor. 2012. Beyond the numbers: Measuring the information

content of earnings press release language. Contemporary Accounting Research 29 (3): 845-868.

Davis, A. K. and I. Tama-Sweet. 2012. Managers’ use of language across alternative disclosure outlets:

Earnings press releases versus MD&A. Contemporary Accounting Research 29 (3): 804-837.

Demers, E. and C. Vega. 2014. The impact of credibility on the Pricing of managerial textual content.

Working paper, The University of Virginia and the Board of Governors of the Federal Reserve

System. Available at http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1153450.

Desai, H., C. E. Hogan, and M. S. Wilkins. 2006. The reputational penalty for aggressive accounting:

Earnings restatements and management turnover. The Accounting Review 81 (1): 83-112.

Elliott, W. B. 2006. Are investors influenced by pro forma emphasis and reconciliation in earnings

announcements? The Accounting Review 81 (1): 113-133.

Elliott, W. B., F. D. Hodge, and L. M. Sedor. 2012. Using online video to announce a restatement:

Influences on investor decisions and the mediating role of trust. The Accounting Review 87 (2):

513-535.

Elliott, W. B., K. Rennekamp, and B. White. 2015 Does concrete language in disclosures increase

willingness to invest? Review of Accounting Studies 20 (2): 839-865.

https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2950702

http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1153450

30

Feldman, R., S. Govindaraj, J. Livnat, and B. Segal. 2010. Management’s tone change, post earnings

announcement drift and accruals. Review of Accounting Studies 15 (4): 915-953.

Gow, I.D., Kaplan, S.N., Larcker, D.F., and A. Zakolyukina. 2015. CEO personality and firm policies.

Working paper.

Griffith, E., Kadous, K., and D. Young. 2015. How insights from the “new” JDM research can improve

auditor judgment: Fundamental research questions and methodological advice. Auditing: A

Journal of Practice & Theory 35 (2): 1-22.

Hales, J., X. Kuang, and S. Venkataraman. 2011. Who believes the hype? An experimental examination

of how language affects investor judgments. Journal of Accounting Research 49 (1): 223-255.

Hassell, J. M., R. H. Jennings, and D. Lasser. 1988. Management earnings forecasts: Their usefulness as a

source of firm-specific information to security analysts. Journal of Financial Research 11

(Winter): 303-319.

Hayes, A.F. 2013. Introduction to mediation, moderation, and condition process analysis: A regression-

based approach. Guilford Press.

Hirst, D. E., L. Koonce, and J. Miller. 1999. The joint effect of management’s prior forecast accuracy and

the form of its financial forecasts on investor judgment. Journal of Accounting Research 37 (1):

101-124.

Hirst, D. E., L. Koonce, and S. Venkataraman. 2007. How disaggregation enhances the credibility of

management earnings forecasts. Journal of Accounting Research 45 (4): 811-837.

Hodge, F., P. Hopkins, and J. Pratt. 2006. Management reporting incentives and classification credibility:

The effects of reporting discretion and reputation. Accounting, Organizations and Society 31 (7):

623-634.

Horton, J. J., D. G. Rand, and R. J. Zeckhauser. 2011. The online laboratory: Conducting experiments in a

real labor market. Experimental Economics 14 (1): 399-425.

Huang X., S. H. Teoh, and Y. Zhang. 2014. Tone management. The Accounting Review 89 (3): 1083-

1113.

Hutton, A., G. Miller, and D. Skinner. 2003. The role of supplementary statements with management

earnings forecasts. Journal of Accounting Research 41 (5): 867-890.

Hutton, A., and P.C. Stocken. 2009. Prior forecasting accuracy and investor reaction to management

earnings forecasts. Working paper.

Hyland, K. 2001. Humble servants of the discipline? Self-mention in research articles. English for

Specific Purposes 20 (3): 207-226.

Hyland, K. 2005a. Metadiscourse: Exploring Interaction in Writing. London: Continuum.

Hyland, K. 2005b. Stance and engagement: a model of interaction in academic discourse. Discourse

Studies 7 (2): 173-192.

Jameson, D. 2000. Telling the investment story: a narrative analysis of shareholder reports. Journal of

Business Communication 37 (1): 7-38.

31

Jennings, R. 1987. Unsystematic security price movements, management earnings forecasts, and revisions

in consensus analyst earnings forecasts. Journal of Accounting Research 25 (1): 90-110.

Kahneman, D., & Tversky, A. 1996. On the reality of cognitive illusions. Psychological Review 103 (3):

582–591.

Koonce, L. and M. Lipe. 2010. Earnings trend and performance relative to benchmarks: How consistency

influences their joint use. Journal of Accounting Research 48 (4): 859-884.

Kothari, S. P., X. Li, and J. E. Short. 2009. The effect of disclosures by management, analysts, and

business press on cost of capital, return volatility, and analyst forecasts: A study using content

analysis. The Accounting Review 84 (5): 1639-1670.

Lansford, B., B. Lev, and J. W. Tucker. 2013. Causes and consequences of disaggregating earnings

guidance. Journal of Business Finance and Accounting 40 (1-2): 26-54.

Larcker, D. F. and A. A. Zakolyukina. 2012. Detecting deceptive discussions in conference calls. Journal

of Accounting Research 50 (2): 495-540.

Li, F. 2008. Annual report readability, current earnings, and earnings persistence. Journal of Accounting

and Economics 45 (2-3): 221-247.

Li, F. 2010a. Textual analysis of corporate disclosures: A survey of the literature. Journal of Accounting

Literature 29 (1): 143-165.

Li, F. 2010b. The information content of forward-looking statements in corporate filings – A naïve

Bayesian machine learning approach. Journal of Accounting Research 48 (5): 1049-1102.

Li, F. 2011. Managers’ self-serving attribution bias and corporate financial policies. Working paper.

Libby, R., R. Bloomfield, and M. W. Nelson. 2002. Experimental research in financial accounting.

Accounting, Organizations and Society 27 (8): 775–810

Libby, R., K. Rennekamp, and N. Seybert. 2015. Regulation and the interdependent roles of managers,

auditors and directors on earnings management and accounting choice. Accounting,

Organizations and Society 47 (November): 25-42.

Mayew, W. J. and M. Venkatachalam. 2012. The power of voice: Managerial affective states and future

firm performance. The Journal of Finance 67 (1): 1-43.

Mercer, M. 2004. How do investors assess the credibility of management disclosures? Accounting

Horizons 18 (3): 185-196.

Mercer, M. 2005. The fleeting effects of disclosure forthcomingness on management’s reporting

credibility. The Accounting Review 80 (2): 723-744.

Merkl-Davies, D. M. and N. Brennan. 2007. Discretionary disclosure strategies in corporate narratives:

incremental information or impression management? Journal of Accounting Literature 26 (1):

116-194.

Merkley, K. J., L. S. Bamber, and T. E. Christensen. 2013. Detailed management earnings forecasts: do

analysts listen? Review of Accounting Studies 18 (2): 479-521.

32

Merkley, K. J. 2014. Narrative disclosure and earnings performance: Evidence from R&D disclosures.

The Accounting Review 89 (2): 725-757.

Miller, B. 2010. The effects of reporting complexity on small and large investor trading. The Accounting

Review 85 (6): 2107-2143.

Ng, J., I. Tuna, and R. Verdi. 2013. Management forecast credibility and underreaction to news. Review

of Accounting Studies 18 (4): 956-986.

Nisbett, R. E. and T. D. Wilson. 1977. Telling more than we can know: Verbal reports on mental

processes. Psychological review 84 (3): 231-259.

Olsen, K., K. K. Dworkis and S. M. Young. 2013. CEO narcissism and accounting: A picture of profits.

Journal of Management Accounting Research 26 (2014): 243-267.

Pennebaker, J. W. 2011. The Secret Life of Pronouns: What Our Words Say about Us. New York, NY:

Bloomsbury Press.

Pownall, G., C. Wasley, and G. Waymire. 1993. The stock price effects of alternative types of

management earnings forecasts. The Accounting Review 68 (4): 896-912.

Rennekamp, K. 2012. Processing fluency and investors’ reactions to disclosure readability. Journal of

Accounting Research 50 (5): 1319-1354.

Russo, J. E., M. G. Meloy, and T. J. Wilks. 2000. Predecisional distortion of information by auditors and

salespersons. Management Science 46 (1): 13-27.

Schrand, C. and S. L. C. Zechman. 2012. Executive overconfidence and the slippery slope to financial

misreporting. Journal of Accounting and Economics 53 (1-2): 311-329.

Securities and Exchange Commission (SEC). 1998. A Plain English Handbook: How to Create Clear

SEC Disclosure. Washington, DC: SEC Office of Investor Education and Assistance. Retrieved

from the SEC’s website: http://www.sec.gov/pdf/handbook.pdf.

Sedor, L. M. 2002. An explanation for unintentional optimism in analysts’ earnings forecasts. The


Tan, H-T., E. Y. Wang, and B. Zhou. 2014. When the use of positive language backfires: The joint effect

of tone, readability and investor sophistication on earnings judgments. Journal of Accounting

Research 52 (1): 273-302.

Tan, H-T., E. Y. Wang, and B. Zhou. 2015. How does readability influence investors' judgments?

Consistency of benchmark performance matters. The Accounting Review 90 (1): 371-393.

Wilks, T. J. 2002. Predecisional distortion of evidence as a consequence of real-time audit review. The


You, H. and X. Zhang. 2009. Financial reporting complexity and investor underreaction to 10-K

information. Review of Accounting Studies 14 (4): 559-586.

http://www.sec.gov/pdf/handbook.pdf

33

FIGURE 1

Indirect Effect of Personal Pronoun Usage (through Disclosure Credibility) on Change in Valuations

(Experiment 3)

This figure displays the observed coefficients and (standard errors) from a moderated mediation model (Hayes 2013)

for Experiment 3. Pronoun Usage is coded as -1 when the disclosure contained fewer personal pronouns and 1 when

the disclosure contained more personal pronouns. News is coded as -1 when the disclosure contained bad news and

1 when the disclosure contained good news. Disclosure credibility is a factor estimated from a principal component

analysis on participants’ ratings of (1) the extent to which they felt they could rely on the disclosure and (2) the

extent to which they felt like the manager believed the information in the disclosure. These two measures load on a

single factor (eigenvalue = 1.784, variance explained = 89.2%).

Pronoun

Usage

Disclosure

Credibility

(Factor)

Change in

Valuation

News

Disclosure

Credibility

x

News

0.159

(0.077)

p = 0.038

-0.512

(0.549)

p = 0.351

-0.605

(0.742)

p = 0.415

17.379

(0.749)

p < 0.001

3.590

(0.549)

p < .001

34

TABLE 1

Descriptive Statistics and Tests of H1 (Experiment 1)

Panel A: Descriptive Statistics – Change in Valuation

Fewer Personal

Pronouns

More Personal

Pronouns Row Total

n

Mean

[std dev]

n

Mean

[std dev]

n

Mean

[std dev]

Good News 35 16.14

[9.14] 34

19.82

[13.37] 69

17.96

[11.21]

Bad News 34 -22.82

[13.46] 35

-26.54

[11.39] 69

-24.71

[12.50]

Column Total 69 -3.06

[22.69] 69

-3.70

[26.28]

Panel B: ANOVA Model of Change in Valuation

Source of Variation SS df MS F-stat p-value

News 62791.317 1 62791.317 449.8798 <0.001a

Personal Pronouns 0.013 1 0.013 0.0001 0.992b

News x Personal Pronouns 472.206 1 472.206 3.3832 0.034a

Error 18702.854 134 139.574

Panel C: Follow-up Tests of Simple Effects

Source of Variation df F-stat p-value

Effect of personal pronouns given good news. 1 1.674 0.099a

Effect of personal pronouns given bad news. 1 1.709

0.097a

This table presents descriptive statistics (Panel A) and tests of H1 (Panels B and C) for Experiment 1, where we use

a 2x2 between-subjects design and manipulate (1) whether the disclosure conveys good or bad news and (2) whether

the CEO uses personal pronouns in the last sentence of the disclosure. Before receiving the disclosure containing our

manipulations, participants provide judgments on a 101-point scale about the appropriate valuation for the firm (0 =

“Low” to 100 = “High”). Participants again provide the judgment after receiving the disclosure containing our

manipulations. This table summarizes the change between the initial and final judgment. a One-tailed equivalents, given our directional predictions.

b Two-tailed.

35

TABLE 2



CEO Photo

Absent

CEO Photo

Present Row Total

n

Mean

[std dev]

n

Mean

[std dev]

n

Mean

[std dev]

Good News 27 15.81

[9.95] 28

19.54

[10.55] 55

17.71

[10.34]

Bad News 24 -17.88

[14.15] 25

-25.24

[6.89] 49

-21.63

[11.55]


[20.78] 53

-1.58

[24.27]



News 39871.256 1 39871.256 352.2126 <0.001a

CEO Photo 85.997 1 85.997 0.7597 0.386b

News x CEO Photo 795.873 1 795.873 7.0305 0.005a

Error 11324.558 101 112.100



Effect of CEO photo given good news. 1 1.681 0.098a

Effect of CEO photo given bad news. 1 5.867

0.009a


a 2x2 between-subjects design and manipulate (1) whether the disclosure conveys good or bad news and (2) whether

a photo of the CEO is included with the disclosure. Before receiving the disclosure containing our manipulations,

participants provide judgments on a 101-point scale about the appropriate valuation for the firm (0 = “Low” to 100 =

“High”). Participants again provide the judgment after receiving the disclosure containing our manipulations. a One-tailed equivalents, given our directional predictions.

b Two-tailed.

36

TABLE 3



Fewer Personal

Pronouns

More Personal

Pronouns Row Total

n

Mean

[std dev]

n

Mean

[std dev]

n

Mean

[std dev]

Good News 76 10.30

[13.22] 75

11.80

[10.69] 151

11.05

[12.01]

Bad News 76 -22.16

[17.13] 74

-25.46

[14.12] 151

-23.79

[15.75]


[22.31] 149

-6.70

[22.47]



News 91434.074 1 91434.074 467.331 <0.001a

Personal Pronouns 61.230 1 61.230 0.313 0.576b

News x Personal Pronouns 433.194 1 433.194 2.214 0.069a

Error



Effect of personal pronouns given good news. 1 0.433 0.256a

Effect of personal pronouns given bad news. 1 2.089 0.075a


a 2x2 between-subjects design and manipulate (1) whether the disclosure conveys good or bad news and (2) the rate

of personal pronoun usage in the disclosure. Before receiving the disclosure containing our manipulations,

participants provide judgments on a 101-point scale about the appropriate valuation for the firm (0 = “Low” to 100 =

“High”). Participants again provide the judgment after receiving the disclosure containing our manipulations. This

table summarizes the change between the initial and final judgment. a One-tailed equivalents, given our directional predictions.

b Two-tailed.

h. scott asay, university of iowa

Documents