the credibility consequences of managers’ decisions to ......credibility consequences of...
TRANSCRIPT
The Credibility Consequences of Managers’ Decisions to Provide Warnings about Unexpected Earnings
Molly Mercer Goizueta Business School
Emory University 1300 Clifton Road Atlanta, GA 30322
(404) 727-7079 [email protected]
May 1, 2002 This paper is based on my dissertation completed at the University of Texas at Austin. I thank my dissertation chair, Lisa Koonce, for her support and guidance throughout this project, and my remaining committee members – Keith Brown, Michael Clement, Eric Hirst, and Senyo Tse – for their many helpful suggestions. This paper also benefited greatly from the comments of Jan Barton, Ananda Ganguly, Kevin Jackson, Steve Kachelmeier, Ron King, Bill Kinney, Jay Koehler, Mary Lea McAnally, Grace Pownall, Bill Schwartz, Kristy Towry, Greg Waymire, Alex Yen and seminar participants at the University of Arizona, Emory University, University of Illinois, University of Iowa, University of Minnesota, Santa Clara University, University of Texas, University of Utah, and Washington University. The Center for Business Measurement and Assurance Services and the Eugene and Dora Bonham Memorial Fund provided financial support for this project.
The Credibility Consequences of Managers’ Decisions to Provide Warnings about Unexpected Earnings
Abstract
This study provides a theoretical framework and experimental evidence on how managers’ disclosure decisions affect their credibility with investors. Further, I examine whether investors’ judgments of management credibility are based on different factors in the short- and longer-term. My results show that in the short-term, management disclosure decisions regarding negative news have larger effects on perceived management credibility than disclosure decisions regarding positive news. Specifically, managers who warn investors about unexpected negative news are rewarded with greater credibility increases than managers who warn about unexpected positive news, and managers who fail to warn about unexpected negative news are penalized with greater credibility decreases than managers who fail to warn about unexpected positive news. The results also show that these short-term credibility effects do not persist over time. In the longer-term, managers who report positive earnings news are rated as having higher credibility than managers who report negative earnings news, regardless of their disclosure decisions. Key Words: voluntary disclosure; management credibility; judgment and decision-making
1
I. INTRODUCTION
Firms that are able to credibly communicate information have less disperse and more
accurate analyst forecasts, lower bid-ask spreads, and consequently, a lower cost of capital than
less credible firms (Botosan 1997; Barron et al. 1999; Healy et al. 1999). Recent popular press
articles have detailed how Xerox managers suffered “another blow to their credibility”
(Maremont 2002), how Dupont’s management has made progress in its “battle to regain
credibility with Wall Street” (Warren 2001), and Guess Inc.’s efforts to “repair its badly
damaged credibility with investors” (Bannon 2001). In fact, Dennis Koslowski, CEO of Tyco,
recently noted that a CEO’s job “is all about credibility” (Lublin 2001). Despite the likely
importance of credibility to managers, we have a limited understanding of the factors affecting
management credibility. The extant literature has examined the consequences of having (or not
having) credibility; however, there has been little systematic study regarding how specific
disclosure decisions influence or determine credibility. Identifying determinants of management
credibility is the focus of this study.
I offer a theoretical framework and experimental evidence regarding how management’s
disclosure decisions affect management credibility, where management credibility is defined as
investors’ beliefs about management’s competence and trustworthiness in financial disclosure.
Based on attribution theory and affect-based reasoning theories, I predict that the management
credibility consequences of a disclosure decision will depend on news valence (i.e., whether
news is positive or negative). In addition, I predict that management credibility is based on
different factors in the short-term and long-term.
To test these predictions, I conduct an experiment in which I vary (1) whether
management provides a warning about unexpected earnings, (2) whether unexpected earnings
2
convey positive or negative news relative to the consensus analyst forecast, and (3) the time
elapsing between the earnings announcement and participants’ assessments of management’s
credibility. My experimental approach complements the existing empirical-archival work on
reactions to disclosure decisions in several important ways. Using an experimental approach
allows me to hold constant factors other than management disclosures; such control is especially
helpful when studying the determinants of credibility because firms with high credibility are
fundamentally different in terms of earnings and returns than firms with lesser credibility (Lang
and Lundholm 1993). Experimentation also allows for the collection of data on investor
reactions to specific disclosure decisions. Archival research typically has used financial
analysts’ ratings of a firm’s disclosures (AIMR reports) to measure a firm’s credibility.
However, AIMR reports are annual ratings, and the long time lag between a particular
management disclosure and the subsequent AIMR report makes it difficult to use these reports to
infer the credibility effects of a specific disclosure decision.
The experimental results are generally consistent with my predictions. I find that in the
short-term, management disclosure decisions regarding negative news have larger effects on
perceived management credibility than disclosure decisions regarding positive news.
Specifically, managers who warn investors about unexpected negative news increase their
credibility with investors more than managers who warn about unexpected positive news;
managers who fail to warn about unexpected negative news lose more credibility than managers
who fail to warn about unexpected positive news. Further, these short-term credibility effects do
not persist over time. In the longer-term, perceived management credibility is primarily a
function of whether the firm reports positive or negative news. That is, managers of firms with
positive earnings news are rated as being more trustworthy and competent in financial disclosure
3
than managers of firms with negative earnings news, regardless of their previous disclosure
decisions.
These results should be informative to firm managers who struggle to foresee investors’
reactions to their disclosure decisions. The theoretical literature on disclosure assumes that
managers properly weigh the costs and benefits of various disclosures when making disclosure
decisions (Dye 1985). However, managers cannot properly weigh these costs and benefits if
they do not correctly anticipate investor reactions to their disclosure decisions, and firm
managers often appear surprised by investors’ reactions to their disclosures (McKay 2000). My
study’s results suggest that firms seeking immediate improvements to their credibility should
concentrate on their negative news disclosures, as these disclosures have the greatest effects on
short-term management credibility. In contrast, managers interested in longer-term credibility
should focus on performance rather than disclosure, as the credibility effects of voluntary
disclosure decisions are not long-lived.
The results also have implications for researchers who study voluntary disclosures.
Several studies have operationalized the informativeness of a firm’s disclosure policy as the
overall amount of disclosure (e.g., Botosan 1997). This measure may be a rough proxy because
my results indicate that investors do not value all voluntary disclosures equally. Incorporating
information about whether the disclosure conveys positive or negative news should result in
more refined measures. In addition, my results may have implications for archival researchers
who use AIMR reports to measure the actual quality of voluntary disclosure. AIMR reports
capture analysts’ perceptions of disclosure quality rather than actual disclosure quality. To the
extent that AIMR ratings are made via a process similar to that used to evaluate longer-term
credibility, my data suggest these ratings may not be a good proxy for actual disclosure quality.
4
Rather, AIMR ratings are likely to be based largely on firm performance. Lang and Lundholm
(1993) provide some support for this idea, as they find that firms with higher AIMR ratings have
higher earnings and stock returns. They argue that these results may be due to firms’ tendencies
to provide more complete disclosures when they are doing well. In other words, Lang and
Lundholm suggest that firm performance affects a firm’s actual disclosures which, in turn, affect
analysts’ perceptions. My results suggest a simpler explanation. AIMR ratings could be a direct
result of prior firm performance, independent of any actual disclosure differences across firms.
II. THEORY AND HYPOTHESES
Credibility Consequences of Disclosure Decisions
Firm managers face numerous decisions in the process of crafting voluntary disclosures.
They can choose whether to disclose, when to disclose, how much to disclose, and the degree of
disclosure accuracy. In many cases, the quality of a firm’s voluntary disclosures can be
evaluated, either at the time of the disclosure or when the firm’s actual financial results are
revealed. Hence, it is likely that management’s voluntary disclosure decisions will influence
management credibility.1 Several studies suggest that whether and when management makes
disclosures (hereafter termed timeliness), how much management chooses to disclose (hereafter
termed completeness), and whether management is accurate in their disclosures (hereafter termed
accuracy) affect its credibility with investors.
Libby and Tan (1999) demonstrate that managers who provide timely warnings about
unexpected negative earnings are rated as having higher credibility than managers who do not
provide such disclosures. No studies directly examine the management credibility consequences
of more complete disclosure; however, more complete disclosures are associated with positive
stock price effects (Botosan 1997). These results are consistent with a causal argument in which
5
more complete disclosure leads to higher management credibility, which leads to a lower cost of
capital (King et al. 1990). There is direct evidence that disclosure accuracy affects management
credibility. Tan et al. (1999) show that managers who accurately forecast earnings have higher
perceived competence and trustworthiness than managers who significantly overstate or
understate earnings. Also, firms with a reputation for accurate reporting are more likely to
influence analysts and investors in their subsequent disclosures (Williams 1996).
My study focuses on the timeliness of managers’ voluntary disclosures.2 Specifically, I
examine the credibility consequences of managers’ decisions about whether to provide investors
with warnings about unexpected earnings.3 Research in psychology proposes that more timely
disclosure will lead to higher perceived credibility (Giffin 1967; Mayer et al. 1995), and Libby
and Tan’s (1999) results provide support for this proposition in a voluntary disclosure setting.
Thus, prior research suggests that managers who provide warnings about unexpected earnings
will be perceived as more credible than managers who do not provide such disclosures.
Effects of News Valence on the Credibility Consequences of Disclosure Decisions
Overview
This section takes the basic proposition that more timely disclosure will increase
management credibility a step further. Drawing on attribution theory and affect-based reasoning
theories, I hypothesize that all warnings about unexpected earnings are not equal in their effects
on management credibility and that these credibility effects differ in the short- versus longer-
term. Figure 1 provides an overview of my model. My basic proposition is that cognitive
responses to news valence (i.e., whether earnings news is positive or negative) and disclosure
decisions (i.e., whether management provides a warning) will jointly determine management
credibility effects in the short-term. As time passes, however, affective responses to new valence
6
and disclosure decisions will determine management credibility. These short- and longer-term
reactions are described in more detail below.
[Insert Figure 1] Short-term Credibility Consequences
The term ‘attribution theory’ refers to a group of theories from social psychology that
explain how individuals determine why an event has occurred (Fiske and Taylor 1991). I draw
on attribution theory to explain why the short-term credibility consequences of management’s
disclosure decisions depend on whether the earnings news is positive or negative. Specifically, I
predict that news valence will influence both the amount and type of attributions made by
investors, and that these attributions, in turn, affect perceived management credibility.
The attribution literature shows that people are more likely to think about why an event
has occurred when that event is negative (Weiner 1985). Because negative outcomes are more
likely to trigger attributional processing than positive outcomes, I anticipate that investors will
spend more time thinking about negative news disclosure decisions. The attribution literature
also suggests that investors will make different types of attributions regarding negative news
voluntary disclosures and positive news voluntary disclosures.4 Specifically, a disclosure that is
at odds with the manager’s personal incentives is more likely to be attributed to some enduring
dispositional characteristic of the manager (i.e., their honesty or trustworthiness) than a
disclosure that is consistent with the manager’s incentives (Jones and Davis 1965). Because
negative news tends to have an unfavorable impact on managers’ tenure and compensation
(Dimma 1996; Hughes 1997), investors will likely view warnings about negative news as more
inconsistent with managers’ personal incentives than warnings about positive news. Thus, I
expect that investors receiving negative news warnings are more likely to attribute those
7
warnings to management’s honesty and trustworthiness than investors receiving positive news
warnings.
These differences in attributions, in turn, are predicted to affect perceived credibility.
Because investors spend more time thinking about the reasons for negative news disclosure
decisions and negative new disclosures are more likely to result in dispositional attributions,
warnings about negative news are expected to result in larger increases in management
credibility than warnings about positive news. Similarly, a failure to warn about negative news
is predicted to result in a larger decrease in perceived management credibility than a failure to
warn about positive news. This reasoning is summarized in Hypothesis 1 and can be seen
graphically in Panel A of Figure 2.5
H1: In the short-term, providing warnings about negative news will lead to greater increases in management credibility than providing warnings about positive news. Not warning about negative news will lead to greater decreases in management credibility than not warning about positive news.
[Insert Figure 2]
Longer-term Credibility Consequences
I also propose that the short-term credibility consequences predicted in Hypothesis 1 may
not hold over time. As time passes subsequent to a disclosure decision, investors’ memories of
the detailed information underlying their credibility assessments will fade. When investors try to
recall these details, they will reconstruct prior events to be consistent with their overall affective
reaction to the events (Clore et al. 1994; Kida and Smith 1995).6
To explain the longer-term credibility consequences of disclosure decisions, I draw on an
affect-based model of financial decision-making proposed by Kida and Smith (1995). Kida and
Smith argue that when investors experience an event, both the event and the overall affective
reaction to that event are encoded in memory. Further, they posit that affective reactions create
8
stronger memory traces than the specific events underlying those reactions, so as time passes, the
specific details about the event become less accessible in memory than the overall affective
reaction (Damasio 1994). As a result, when people try to remember the event, their memories
are reconstructed to be consistent with that overall affective reaction. If Kida and Smith are
correct about the role that affect plays in memory, investors’ overall affective reactions should
influence their memories of management’s disclosure decisions and, consequently, their longer-
term assessments of management’s credibility.
Although affect-based reasoning theories predict that the longer-term credibility
consequences of a disclosure decision will be determined by investors’ overall affective
reactions, they do not specify whether an investor’s overall affective reaction will be determined
by news valence or management’s disclosure decision. Investors are hypothesized to experience
affective reactions to both the earnings news and the disclosure decision. In some cases, these
affective reactions will be in conflict, such as when managers report negative news (which
results in negative affect) but provide a warning about the news (which results in positive affect).
Kida and Smith (1995) argue that in cases where underlying components of an event promote
conflicting affective responses, the stronger of the affective reactions will determine investors’
overall affective reaction. However, this research does not provide an unambiguous prediction
about which reaction – news valence or disclosure decision – will be stronger. I therefore
discuss longer-term credibility effects under both scenarios.
Some evidence suggests that investors’ affective reactions to earnings news will be
stronger than their affective reactions to the related disclosure decisions. Unexpected earnings
news appears to have a greater impact on stock market reactions than the associated disclosure
decisions (Kasznik and Lev 1995). To the extent that investors ultimately care most about their
9
stock market returns, reactions to the earnings news will be stronger than reactions to the related
disclosure decision. As shown in Panel B of Figure 2, if investors experience stronger affective
reactions to earnings news, the long-term credibility consequences of disclosure decisions will be
determined primarily by news valence. That is, managers of firms with negative earnings news
will be rated as having lower credibility than managers of firms with positive earnings news,
regardless of whether investors were warned about the news.
Alternatively, overall affective reactions may be determined primarily by disclosure
decisions rather than news valence. Koehler and Gershoff (2002) show that when people believe
that they have been betrayed, they experience stronger affective reactions than justified by their
actual loss. Therefore, investors’ affective reactions to disclosure decisions may be stronger than
their reactions to the earnings news if they feel betrayed by managers’ nondisclosures. If
affective reactions to the disclosure decision are stronger, disclosure decisions will be the
primary determinant of investors’ longer-term credibility judgments. As a result, managers who
provide warnings about unexpected news will be perceived as more credible than managers who
do not provide warnings, regardless of whether those disclosures relate to positive earnings news
or negative earnings news.
To summarize, I expect that as time passes, investors’ memories of the detailed
information underlying their beliefs about management credibility will fade. When investors
attempt to recall these details, they will reconstruct events to be consistent with their initial
overall affective reaction to the events. As a result, I predict that in the longer-term, investors’
perception of management credibility will be determined by their overall affective reaction to the
news valence and disclosure decision. However, because the affect-based reasoning literature
does not specify whether investors’ overall affective reactions will be determined primarily by
10
news valence or disclosure decisions, I do not make predictions about the primary determinant of
those reactions.
H2: As time passes subsequent to management’s disclosure decisions, investors’ overall affective reactions will determine their perceptions of management’s credibility.
III. EXPERIMENT Participants
To test my hypotheses, I conducted an experiment with 244 MBA students at a large state
university. Participants were recruited via an email notice sent to all MBA students and were
paid a flat wage of $10 plus the chance to win a random lottery. On average, study participants
had six years of work experience and had completed three accounting and two finance classes;
nearly all (94%) had invested in common stocks or common stock mutual funds.
Experimental Design and Task
The experiment is based on a 2 x 2 x 2 between-subjects design, with Disclosure
Decision (Warning, No Warning), News Valence (Positive News, Negative News), and Time
Elapsed since the disclosure decision (Short-term, Longer-term) as independent variables.
Figure 3 provides an overview of the experimental procedures. Participants were asked to
assume the role of a member of an investment club that had recently purchased the common
stock of Dentex, a company in the dental supply industry. Each participant was given
background information about the company, including brief excerpts from management’s
discussion and analysis and historical financial statement data. Participants also were informed
that the consensus analyst forecast for the company’s upcoming quarterly earnings was $0.52.
After reviewing this information, participants provided pre-test assessments of management
credibility.
11
[Insert Figure 3]
Half of the participants received a voluntary disclosure from management stating that
actual quarterly earnings were expected to differ from the consensus forecast. Specifically,
participants in the Positive [Negative] News condition received the following management
disclosure.
On March 15, 2000, Dentex management voluntarily issued the following press release:
In a presentation to industry analysts today, Dentex, Inc. CEO Murray Levine indicated that the company currently expects earnings per share will be $0.60 [$0.44] for the quarter ending March 31, 2000. This estimate is $0.08 above [below] the current consensus analyst forecast of $0.52.
The other half of participants did not receive a voluntary disclosure.
Subsequently, all participants answered a number of demographic questions regarding
their investing experience, work experience, and prior coursework. All participants then were
provided with the firm’s actual quarterly earnings. Actual earnings were reported as $0.60 to
participants in the Positive News condition and $0.44 to participants in the Negative News
condition. Thus, I manipulated whether investors were warned (i.e., disclosure timeliness) and
held constant the accuracy and completeness of disclosure. After viewing the actual earnings
information, all participants answered several manipulation check and affective reaction
questions.
Participants in the Short-term condition answered additional questions immediately after
viewing the actual earnings information. Specifically, they provided assessments of
management’s credibility, answered questions about their attributional processing, and indicated
the degree to which they would be willing to rely on a subsequent forecast issued by
management. Longer-term participants answered this same set of questions approximately two
weeks after completing the experimental materials.7
12
Credibility Measures
The primary dimensions of management credibility are management competence and
management trustworthiness. To capture these dimensions, I combined questions from two
widely accepted source credibility scales: McCroskey (1966) and Leathers (1992). The
Appendix shows the six questions that were chosen from these scales. Three competence
questions focus on investors’ perceptions of the competence, knowledge, and qualifications of
management for providing financial disclosures. Three trustworthiness questions focus on
investors’ perceptions of management’s trustworthiness, honesty, and truthfulness.
As noted previously, archival accounting research has demonstrated the important
consequences of management credibility. Because my paper focuses on the determinants of
credibility, my primary dependent variable is perceived management credibility. That is, I focus
on how management’s disclosure decisions affect investors’ perceptions of management
credibility, rather than on how differences in credibility judgments affect earnings predictions or
investment decisions.8
However, to ensure that participants’ credibility ratings have behavioral consequences, I
also examine whether differences in perceived credibility are reflected in subsequent behavior.
Managers care about their credibility because investors are more likely to rely on disclosures
from highly credible managers (and greater reliance on management disclosures results in a
reduction in information asymmetry and its associated costs) (King et al. 1990). Therefore, in
addition to testing the effects of disclosure decisions on perceived management credibility, I
examine the effects of these decisions on participants’ willingness to rely on subsequent
management disclosures. Specifically, after participants viewed the experimental manipulations,
13
they were shown a management earnings forecast for a subsequent quarter and asked how
willing they would be to rely on this forecast when predicting earnings.9
IV. RESULTS
Manipulation and Other Checks
Responses to the manipulation check questions indicated that all manipulations were
successful. In the Warning condition, 98% of participants correctly indicated that they received
a warning, and 98% of participants in the No Warning condition correctly indicated that they did
not receive a warning. For the News Valence manipulation, 100% [99%] of the participants in
the Positive [Negative] News condition correctly indicated the sign of the unexpected earnings
news.
To rule out the possibility that I not only manipulated news valence (positive or negative)
but also the expectedness of the news, participants were asked to evaluate the expectedness of
the earnings information. A possible alternative explanation for my hypothesized results is that
participants were more surprised by the negative earnings news than the positive earnings news.
However, the results reveal that participants in the Negative News condition were not
significantly more surprised by the unexpected earnings than participants in the Positive News
condition (t=0.47, p=0.64), thereby ruling out this explanation.
Credibility Differences in the Short- and Longer-term
Participants’ responses to the credibility questions indicated that the three competence
questions and the three trustworthiness questions each captured a separate underlying construct,
with Cronbach’s alphas of 0.73 and 0.89, respectively. To simplify the presentation of the
results, I formed composite ratings of competence and trustworthiness by using the mean of
14
participants’ responses to the questions measuring each construct. The composite measures are
subsequently referred to as ‘management competence’ and ‘management trustworthiness.’10
To examine whether disclosure decisions and news valence have different effects on
management credibility in the short- and longer-term, I conducted three-way ANCOVAs on
management competence and management trustworthiness. For each ANCOVA, News Valence,
Disclosure Decision, and Time Elapsed were independent variables and participants’ pre-test
credibility assessments were included as a covariate.11 Both ANCOVAs showed significant
three-way interaction terms (F=3.91, p<0.05 for competence; F=5.43, p<0.05 for
trustworthiness), suggesting that the credibility effects of news valence and disclosure decisions
differ in the short- and longer-term. Consequently, I analyze these short- and longer-term effects
separately in the following sections.
Short-term Effects
Hypothesis 1 predicts that in the short-term, providing warnings about negative news will
lead to greater credibility gains than providing warnings about positive news, and not providing
warnings about negative news will lead to greater credibility losses than not providing warnings
about positive news. To test Hypothesis 1, I conducted separate two-way ANCOVAs on
competence and trustworthiness, with News Valence and Disclosure Decision as independent
variables and investors’ pre-test credibility judgments as covariates. As seen in Table 1, the
ANCOVA for perceived management trustworthiness shows a significant Disclosure
Decision*News Valence interaction (F=10.42, p<0.01). Simple effects analyses show that,
consistent with Hypothesis 1, managers are rewarded with greater trustworthiness for warnings
about negative news (µ=4.61) than for warnings about positive news (µ=4.14) (F=3.43, p=0.07).
In addition, managers are accorded lower trustworthiness ratings for not providing warnings
15
about negative news (µ=3.19) than for not warning about positive news (µ=3.84) (F=7.43,
p<0.01). For perceived management competence, managers who warned were rated as being
more competent than those who didn’t warn, as evidenced by the significant main effect for
Disclosure Decision (F=27.30; p<0.01). However, the results show an insignificant Disclosure
Decision*News Valence interaction term (F=1.32, p=0.17), suggesting that the differences in
perceived competence between the Warning and No Warning conditions were not significantly
larger for negative news than for positive news. A graphical summary of the trustworthiness and
competence results is provided in Panel A of Figure 4.
[Insert Table 1 & Figure 4]
Taken together, these results suggest that investors differentiate between the two
credibility dimensions. That is, the valence of unexpected earnings and the decision to warn
about unexpected earnings jointly affect investors’ perceptions of management trustworthiness,
but not management competence. Prior accounting literature on management credibility
presumes that credibility is a one-dimensional construct. My results suggest that not only are
there two dimensions to credibility, but disclosure decisions may have different effects on these
dimensions.
To ensure that differences in perceived management credibility are reflected in
investors’ reliance on subsequent management disclosures, I examine the effects of positive and
negative news disclosure decisions on participants’ willingness to rely on a management
earnings forecast in a subsequent quarter. The results of an ANOVA with Disclosure Decision
and News Valence as independent variables and subsequent reliance as the dependent variable
show a significant Disclosure Decision*News Valence interaction term (F=20.78, p<0.01).
Follow-up analyses confirm that participants are more willing to rely on a subsequent disclosure
16
when managers had previously warned about negative news (µ=5.48) than when managers had
warned about positive news (µ=5.01) (F=3.21, p=0.08). In addition, participants are less likely
to rely on a subsequent disclosure when management failed to warn about negative news
(µ=3.57) than when management failed to warn about positive news (µ=5.05) (F=24.50, p<0.01).
Thus, negative news disclosure decisions have greater effects on subsequent reliance than
positive news disclosure decisions, suggesting that management credibility affects investors’
willingness to rely on management disclosures.12
Longer-term Effects
Hypothesis 2 predicts that investors’ longer-term credibility assessments will be
determined by their affective, rather than cognitive, reactions. Immediately after experiencing
the experimental manipulations, participants provided assessments of their affect related to the
disclosure decision and the news valence. These assessments were made on comparable 7-point
Likert-type scales. I calculated the strength of affect related to the disclosure decision as the
absolute value of the difference between the participant’s affective response and the scale
midpoint; the strength of affect related to the earnings news was calculated in the same manner.
I then calculated the relative strength of participants’ affective responses by subtracting
disclosure decision affect strength from earnings news affect strength. If affective reactions to
news valence are stronger than affective reactions to disclosure decisions, the mean of this
relative affect measure will be positive. On the other hand, if affective reactions to disclosure
decisions are stronger, the mean of the relative affect measure will be negative. I find that mean
relative affect is positive and significantly different than zero (t=2.32, p=0.02), suggesting that
affective responses to news valence are stronger than affective responses to disclosure
decisions.13
17
Because the above data suggest that investors experience stronger affective responses to
news valence than to disclosure decisions, I expect news valence to be the primary determinant
of management credibility in the longer-term. In statistical terms, this suggests that two-way
ANCOVAs for competence and trustworthiness will show only a significant main effect for
News Valence. As seen in Table 2, the data support this prediction. Both ANCOVAs show a
significant News Valence main effect (F=14.67, p<0.01 for trustworthiness; F=20.78, p<0.01 for
competence). Furthermore, both ANCOVAs show insignificant effects for Disclosure Decision
(F=0.00, p=0.99 for trustworthiness; F=0.39, p=0.53 for competence). Panel B of Figure B
provides a summary of these results. The results suggest that there is little longer-term
credibility benefit from providing warnings about unexpected earnings. Rather, in the longer-
term, perceived management credibility is mainly a function of firm performance, with managers
of firms that report positive earnings news being rated as having higher credibility than managers
of firms that report negative earnings news.
[Insert Table 2]
Based on the management credibility results, I expect that in the longer-term, news
valence also will determine reliance on a subsequent disclosure. Consistent with this prediction,
an ANOVA with reliance on a subsequent forecast as the dependent variable and News Valence
and Disclosure Decision as independent variables shows a significant main effect for News
Valence (F=3.81, p<0.05) and an insignificant main effect for Disclosure Decision (F=0.03,
p=0.87). In the longer-term, Positive News participants are more willing to rely on a subsequent
management disclosure than Negative News participants, and management’s prior disclosure
decision does not affect subsequent reliance.14
18
To summarize, my results indicate that the short-term credibility benefits of providing
warnings about unexpected earnings do not necessarily persist in the longer-term. I find that
longer-term perceptions of management’s competence and trustworthiness are driven primarily
by news valence. Specifically, managers of firms with negative earnings news are rated as
having lower credibility, and managers of firms with positive earnings news are rated as having
higher credibility, regardless of whether investors were warned about unexpected news. In
addition, these differences in perceived management credibility manifest themselves in reliance
on subsequent disclosures.
V. CONCLUSIONS
Firm managers are continually faced with decisions about the extent to which they should
voluntarily disclose unexpected news to analysts and investors. However, the extant literature
provides little guidance about the factors that affect investors’ reactions to managers’ voluntary
disclosure decisions. My study provides evidence on the short- and longer-term credibility
consequences of managers’ decisions to warn investors about unexpected earnings.
My results suggest that managers seeking immediate improvements to their credibility
should concentrate on their negative news disclosures. Specifically, I find that disclosure
decisions regarding negative news have larger short-term effects on management credibility than
do positive news disclosure decisions. That is, managers who provide warnings about
unexpected negative news receive larger boosts in perceived credibility than managers who
provide warnings about unexpected positive news. Managers who fail to warn about negative
news suffer larger decreases in perceived credibility than managers who fail to warn about
unexpected positive news. These differences in perceived management credibility have
19
important consequences, as I find that credibility influences investors’ reliance on subsequent
management disclosures.
My results also show that the credibility effects described above do not persist in the
longer-term. Rather, management credibility is primarily a function of the valence of news
disclosed. In the longer-term, managers of firms with positive earnings news are rated as having
higher credibility than managers of firms with negative earnings news, regardless of managers’
disclosure decisions. This result is interesting because of the uncertainty regarding whether
managers are better off providing warnings about unexpected negative news. Kasznik and Lev
(1995) and Libby and Tan (1999) find that firms who provide warnings are penalized with larger
immediate stock price reactions than firms that do not provide such warnings. Despite this
apparent penalty for warning about negative news, firms are still more likely to provide warnings
about unexpected negative (versus positive) news (Skinner 1994). Why might firms continue to
warn investors about negative news despite the negative stock price consequences? One
proposed explanation is that the immediate valuation benefits associated with failing to warn
investors about negative news are overwhelmed by the longer-term credibility costs of not
warning (Kasznik and Lev 1995). Although negative news disclosure decisions have significant
effects on managers’ credibility in the short-term, I show that these credibility benefits for
warning about negative news do not persist in the longer-term. Thus, my study provides
evidence that long-term credibility concerns likely do not explain managers’ greater willingness
to warn about negative news.
My study suggests several avenues for future research. I examine investors’ reactions to
voluntary disclosures in only one context – management’s decisions to provide warnings about
unexpected earnings. However, my findings likely generalize to other voluntary disclosure
20
decisions. For example, I would expect variations in the completeness of voluntary disclosures
about negative news to have larger short-term credibility effects than variations in the
completeness of positive news disclosures. Further, my study focuses on the formation of
investors’ beliefs about management credibility. That is, I examine how voluntary disclosure
decisions influence management credibility in a scenario where investors do not have a long
history of interactions with management and thus do not have strongly held beliefs about
management’s credibility. The effects of disclosure decisions on management credibility may
differ in cases where investors have strongly held prior beliefs (Ross and Lepper 1980; Koehler
1993). For example, investors who strongly believe that management is credible may interpret a
lack of disclosure differently than investors who believe management is not credible. Future
research can examine how strong prior beliefs about management influence the credibility
consequences of disclosure decisions.
21
ENDNOTES
1 As noted previously, management credibility is defined as investors’ beliefs about management’s competence and trustworthiness in providing financial disclosures (Hovland et al. 1953, 21; Giffin 1967). 2 In practice, the timeliness, accuracy, and completeness of voluntary disclosures may not be independent. For example, more timely management disclosures may be less accurate, because these disclosures are released significantly before the release of actual financial results. However, given my experimental design, I am able to manipulate timeliness and hold constant accuracy and completeness. 3 I choose to examine managers’ disclosure decisions in this context for several reasons. First, there appears to be substantial real-world variation in managers’ decisions to provide earnings warnings. For example, Skinner (1994) finds that firms provide warnings about large unexpected negative earnings news 25% of the time and warnings about large unexpected positive earnings news 7% of the time. In addition, prior studies on investors’ reactions to earnings warnings (e.g., Kasznik and Lev 1995) are central to the development of my hypotheses regarding the longer-term credibility effects of disclosure decisions. 4 Attribution theory provides guidance on the types of attributions made subsequent to negative news disclosures and positive news disclosures, but does not specify the types of attributions that will be made in non-disclosure situations. Hence, I do not make predictions about the types of attributions made when warnings are not provided. 5 My predictions regarding the credibility effects of disclosure decisions do not differ for the two dimensions of management credibility – competence and trustworthiness. In my experiment, as in many real-world situations, participants do not know whether management fails to warn because management was mistaken about the firm’s prospects (which should affect perceived competence) or because they were intentionally trying to mislead (which should affect perceived trustworthiness). Arguably, this uncertainty will cause management’s warning decisions to affect both competence and trustworthiness. In further support of this view, prior research finds that attributes such as competence and trustworthiness tend to move together (Nathan and Tippins 1990). That is, managers who are perceived as having one positive [negative] attribute are more likely to be perceived as having other positive [negative] attributes. 6 Historically, the term “affect” has been used to refer to various constructs, including evaluations, moods, and emotions. Evaluations refer to simple positive or negative feelings associated with a stimulus, and moods refer to general affective states (typically positive or negative) that are not targeted at a particular stimulus. Emotions refer to more fleeting affective states such as anger, jealousy, or serenity, that go beyond simple positive or negative feelings (Fiske and Taylor 1991). Consistent with numerous recent articles that examine the role of affect on decision-making processes (e.g., Finucane et al. 2000), I focus on affect as an evaluation. Thus, in this study, affect refers to a positive or negative emotional response associated with an event. 7 Pilot testing confirmed that two weeks was an appropriate time delay for this study. Two weeks was a sufficient amount of time for reconstruction to be necessary, but not so long that participant response rate was threatened. 8 Asking participants to provide earnings predictions or make investment decisions in addition to the credibility judgments could contaminate those credibility judgments. For example, study participants might have made investment decisions based on their personal investment philosophy, and then justified those decisions via their assessments of credibility (e.g., decisions to sell could be justified with low credibility judgments). Given the paucity of research on the determinants of credibility, I felt it was most important to accurately measure investors’ credibility judgments. 9 A positive news management earnings forecast was chosen for this task because prior research shows that management credibility is most important for increasing the believability of subsequent positive news disclosures (Williams 1996). 10 Results for the individual competence and trustworthiness questions are similar to those reported for the composite measures. 11 ANCOVA results are presented for simplicity. Inferences do not change when the data are analyzed using multivariate tests. In addition, results are presented using participants’ pre-test credibility assessments as covariates, but analyses using difference scores lead to similar conclusions. 12 In addition to documenting the credibility effects associated with managers’ disclosure decisions, my analyses provide insights into why these effects occur. Recall that news valence was expected to affect the amount and type of attributions resulting from management’s disclosure decisions. I measured the amount of attributional processing
22
by videotaping a subset of participants and recording the time they spent completing the materials. I find that participants in the Negative News condition spent significantly more time completing the materials (µ=17.92 minutes) than participants in the Positive News condition (µ=16.01 minutes) (t=1.69; p<0.05), consistent with negative news disclosure decisions triggering more attributional processing than positive news disclosure decisions. I also find evidence that positive and negative news disclosures result different types of attributions. Specifically, participants who were warned about negative news were more likely to attribute the disclosure to management’s honesty (Mann-Whitney, p<0.05) and less likely to attribute to management’s personal interests (Mann-Whitney, p<0.05) than participants who were warned about positive news. Overall, the attribution results provide strong support for my theoretical predictions about the short-term effects of disclosure decisions. 13 Additional analyses confirm this finding. Participants were asked whether they would feel more negatively overall about the management of a firm that 1) reported actual earnings below the consensus forecast and warned about those unexpected earnings or 2) reported actual earnings above the consensus forecast and did not warn about those unexpected earnings. The majority of participants (63%) stated that they would feel more negatively about management that reported negative news but warned about that news, providing further evidence that news valence is a stronger determinant of overall affect than disclosure decisions. 14 An alternative explanation for this result is that investors are more willing to rely on a subsequent positive news disclosure for a company that previously reported positive news. In other words, the greater inherent plausibility of the subsequent disclosure for Positive News participants, rather than Positive News participants’ beliefs about management credibility, could produce the main effect of News Valence on reliance. To rule out this explanation, I conduct an ANCOVA using perceptions of management’s competence and trustworthiness as covariates. The effects of News Valence on subsequent reliance are eliminated (F=0.00, p=0.97) in this analysis, suggesting that the effects of News Valence on subsequent reliance are caused by participants’ perceptions of management’s credibility.
23
REFERENCES Bannon, L. 2001. Guess takes broad steps to regain trust of investors. The Wall Street Journal
(April 17): B6. Baron, R. and D. Kenny. 1986. The moderator-mediator variable distinction in social
psychological research: conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology 51(6): 1173-1182.
Barron, O., C. Kile, and T. O’Keefe. 1999. MD&A quality as measured by the SEC and
analysts’ earnings forecasts. Contemporary Accounting Research 16(1): 75-109. Botosan, C. 1997. Disclosure level and the cost of equity capital. The Accounting Review 72
(3): 323-349. Clore, G., N. Schwarz, and M. Conway. 1994. Affective causes and consequences of social
information processing. In R.Wyer and T. Srull (Eds.), Handbook of Social Cognition, (Vol. 1, pp. 323-417). Lawrence Erlbaum Associates: Hillsdale, NJ.
Damasio, A. 1994. Descartes’ Error: Emotion, Reason, and the Human Brain. Avon: New
York. Dimma, W. 1996. Executive compensation gets its share. Business Quarterly 61(1): 77-80. Dye, R. 1985. Disclosure of nonproprietary information. Journal of Accounting Research 23:
123-145. Finucane, M., A. Alkakami, P. Slovic, and S. Johnson. 2000. The affect heuristic in judgments
of risks and benefits. Journal of Behavioral Decision Making 13: 1-17. Fiske, S. and S. Taylor. 1991. Social Cognition. McGraw-Hill: New York. Giffin, K. 1967. The contribution of studies of source credibility to a theory of interpersonal
trust in the communication process. Psychological Bulletin 68(2): 104-120. Healy, P., A. Hutton, and K. Palepu. 1999. Stock performance and intermediation changes
surrounding sustained increases in disclosure. Contemporary Accounting Research 16(3): 485-520.
Hovland, C., I. Janis, and H. Kelley. 1953. Communication and Persuasion. Yale University
Press: New Haven, Connecticut. Hughes, J. 1997. How does your pay stack up? Financial Executive 13(3): 24-26. Jones, E. and K. Davis. 1965. From acts to disposition: The attribution process in person
perception. In L. Berkowitz (Ed.), Advances in Experimental Social Psychology (Vol. 2, pp. 219-266). Academic Press: New York.
24
Kasznik, R. and B. Lev. 1995. To warn or not to warn: management disclosures in the face of
an earnings surprise. The Accounting Review 70: 113-153. Kida, T. and J. Smith. 1995. The encoding and retrieval of numerical data for decision making
in accounting contexts: Model development. Accounting, Organizations, and Society 20 (7/8): 585-610.
King, R., G. Pownall, and G. Waymire. 1990. Expectations adjustment via timely management
forecasts: review, synthesis, and suggestions for future research. Journal of Accounting Literature 9: 113-144.
Koehler, J. 1993. The influence of prior beliefs on scientific judgments of evidence quality.
Organizational Behavior and Human Decision Processes 56: 28-55. Koehler, J. and A. Gershoff. 2002. Betrayal aversion: When agents of protection cause the very
harm they are supposed to prevent. Forthcoming in Organizational Behavior and Human Decision Processes.
Lang, M. and R. Lundholm. 1993. Cross-sectional determinants of analyst ratings of corporate
disclosures. Journal of Accounting Research 31(2): 246-271. Leathers, D. 1992. Successful Nonverbal Communications: Principles and Applications.
MacMillan: New York. Libby, R. and H. Tan. 1999. Analysts’ reactions to warnings of negative earnings surprises.
Journal of Accounting Research 37(2): 415-435. Lublin, J. 2001. CEO’s guide to survival – Shared traits of leadership help chiefs of Sunbeam,
Polaroid, and Tyco endure. The Wall Street Journal (February 27): B1. Maremont, M. 2002. Xerox says SEC unit disagrees with the way it accounts for certain lease
transactions. The Wall Street Journal (January 8): B6. Mayer, R., J. Davis, and F. D. Schoorman. 1995. An integrative model of organizational trust.
Academy of Management Review 20(3): 709-734. McCroskey, J. 1966. Scales for the measurement of ethos. Speech Monographs 33: 65-72. McKay, B. 2000. Street seeks real thing on Coke outlook. The Wall Street Journal (January
17): C1. Nathan, B. and N. Tippins. 1990. The consequences of halo “error” in performance ratings: A
field study of the moderating effects of halo on test validation results. Journal of Applied Psychology 75(3): 290-296.
25
Ross, L. and M. Lepper. 1980. The perseverance of beliefs: Empirical and normative considerations. New Directions for Methodology of Social and Behavioral Science 4: 17-36.
Skinner, D. 1994. Why firms voluntarily disclose bad news. Journal of Accounting Research
32(1): 38-60. Tan, H., R. Libby, and J. Hunton. 1999. Analysts’ reactions to earnings preannouncement
strategies. Working paper, Nanyang Technological University. Warren, S. 2001. DuPont’s net falls. The Wall Street Journal (April 25): B15. Weiner, B. 1985. “Spontaneous” causal thinking. Psychological Bulletin 97(1): 74-84. Williams, P. 1996. The relation between a prior earnings forecast by management and analyst
response to a current management forecast. The Accounting Review 71(1): 103-113.
26
FIGURE 1
OVERVIEW OF THE CREDIBILITY CONSEQUENCES OF DISCLOSURE DECISIONS
This figure provides an overview of my theoretical framework, which indicates that investors experience both cognitive and affective reactions subsequent to a disclosure decision. I predict that cognitive reactions are the primary determinant of management credibility in the short-term. In the longer-term, however, management credibility is determined by investors’ affective reactions.
News Valence X
Disclosure Decision
Cognitive Reactions
Short-term Perceived
Management Credibility
Longer-term Perceived
Management Credibility
Affective Reactions
27
FIGURE 2 PREDICTIONS
Panel A: Short-term Credibility Judgments
Panel B: Longer-term Credibility Judgments
If News ValenceDetermines Overall Affect
No Warning Warning
Disclosure Decision
Perc
eive
d M
anag
emen
t C
redi
bilit
y
No Warning WarningDisclosure Decision
Perc
eive
d M
anag
emen
t C
redi
bilit
y
Positive NewsNegative News
If Disclosure Decision Determines Overall Affect
No Warning Warning
Disclosure Decision
Perc
eive
d M
anag
emen
t C
redi
bilit
y
Positive NewsNegative News
28
FIGURE 3
SEQUENCE OF EXPERIMENTAL PROCEDURES
Information Provided Questions Asked Instructions
Background Information
Pre-test Credibility Assessments
Warning (for participants in the Warning condition)
Demographic Questions
Announcement of Actual Earnings
Manipulation Check Questions
FOR ALL
PARTICIPANTS
Affective Reaction Questions
Post-test Credibility Assessments
Attribution Questions
FOR
SHORT-TERM PARTICIPANTS
ONLY Reliance on a Subsequent Forecast Question
Two Weeks Later
Post-test Credibility Assessments
Attribution Questions
FOR
LONGER-TERM PARTICIPANTS
ONLY Reliance on a Subsequent Forecast Question
29
FIGURE 4
SUMMARY OF MANAGEMENT CREDIBILITY RESULTS
Panel A: Short-term Credibility Judgments
Panel B: Longer-term Credibility Judgments
This figure depicts the short- and longer-term credibility effects of disclosure decisions. To measure short-term credibility, participants assessed management’s competence and trustworthiness immediately after viewing the experimental manipulations. To measure longer-term credibility, participants answered the competence and trustworthiness questions approximately two weeks after viewing the manipulations. See Table 1 for a description of the experimental manipulations and the competence and trustworthiness measures.
4.333.85
4.42
3.42
No Warning Warning
Disclosure Decision
Perc
eive
d M
anag
emen
t C
ompe
tenc
e
4.143.84
4.61
3.19
No Warning Warning
Disclosure DecisionPe
rcei
ved
Man
agem
ent
Trus
twor
thin
ess
Positive NewsNegative News
4.584.37
3.73.9
No Warning Warning
Disclosure Decision
Perc
eive
d M
anag
emen
t C
ompe
tenc
e
4.194.1
3.533.44
No Warning Warning
Disclosure Decision
Perc
eive
d M
anag
emen
t Tr
ustw
orth
ines
s
Positive NewsNegative News
30
TABLE 1
SHORT-TERM CREDIBILITY JUDGMENTSa
Panel A: Descriptive Statistics – Mean (Standard Deviation) Ratings
Management Competenceb Management Trustworthinessb
Positive News
Negative News
Positive News
Negative News
No Warning
3.85
(0.91) n=32
3.42
(0.99) n=32
3.64
(0.97) n=64
3.84
(0.72) n=32
3.19
(1.21) n=32
3.51
(1.04) n=64
Disclosure Decision
Warning
4.33 (0.80) n=32
4.42 (0.84) n=30
4.38 (0.82) n=62
4.14 (0.78) n=32
4.61 (0.83) n=30
4.37 (0.83) n=62
4.09 (0.88) n=64
3.91 (1.05) n=62
3.99 (0.76) n=64
3.88 (1.26) n=62
Panel B: ANCOVA Results
Management Competence Management Trustworthiness
df
Mean Square
F- Statistic
p-value
df
Mean Square
F- Statistic
p-value
Pre-test Ratingc
1
11.54
16.37
.000
1
26.12
42.84
.000
Disclosure Decision 1 19.24 27.30 .000 1 27.60 45.27 .000 News Valence 1 0.83 1.18 .279 1 0.23 0.37 .543 Disclosure Decision*News Valence 1 1.32 1.88 .173 1 6.36 10.42 .002
a This table provides an analysis of participants’ short-term credibility judgments. Half of the participants were provided with a voluntary disclosure from management stating that actual quarterly earnings were expected to differ from the consensus analyst forecast. Specifically, participants in the Positive [Negative] News condition were told that earnings per share were expected to be $0.08 above [below] the analyst forecast. The other half of participants did not receive any voluntary disclosure. Subsequently, all participants were provided with the firm’s actual quarterly earnings, which were $0.08 above [below] the consensus forecast for Positive [Negative] News participants. Immediately after viewing these materials, participants provided their assessments of management’s competence and trustworthiness. b Management competence was measured as the mean of participants’ responses to three questions regarding the competence, knowledge, and qualifications of management for providing financial disclosures. Management trustworthiness was measured as the mean of participants’ responses to three questions regarding management’s trustworthiness, honesty, and truthfulness. All credibility questions were answered on 7-point scales, and responses were standardized so that higher ratings always indicated greater perceived competence/trustworthiness.
c Participants’ pre-test ratings of management competence and management trustworthiness were included as covariates in the management competence ANCOVA and the management trustworthiness ANCOVA, respectively.
31
TABLE 2
LONGER-TERM CREDIBILITY JUDGMENTSa
Panel A: Descriptive Statistics – Mean (Standard Deviation) Ratings
Management Competence Management Trustworthiness
Positive News
Negative News
Positive News
Negative News
No Warning
4.37
(0.69) n=24
3.90
(0.92) n=29
4.11
(0.85) n=53
4.10
(0.63) n=24
3.44
(0.92) n=29
3.74
(0.86) n=53
Disclosure Decision
Warning
4.58 (0.79) n=28
3.70 (0.79) n=26
4.16 (0.90) n=54
4.19 (0.74) n=28
3.53 (0.96) n=26
3.87 (0.91) n=54
4.48 (0.74) n=52
3.80 (0.86) n=55
4.15 (0.69) n=52
3.48 (0.93) n=55
Panel B: ANCOVA Results
Management Competence Management Trustworthiness
df
Mean Square
F- Statistic
p-value
df
Mean Square
F- Statistic
p-value
Pre-test Ratingb
1
2.84
4.50
.036
1
5.06
7.93
.006
Disclosure Decision 1 0.00 0.00 .985 1 025 0.39 .532 News Valence 1 13.10 20.78 .000 1 9.37 14.67 .000 Disclosure Decision*News Valence 1 1.13 1.79 .184 1 0.00 0.00 .987
a This table provides an analysis of participants’ longer-term credibility judgments. See Table 1 for a description of the experimental manipulations. Approximately two weeks after viewing these manipulations, participants provided assessments of management’s competence and trustworthiness. b Participants’ pre-test ratings of management competence and management trustworthiness were included as covariates in the management competence ANCOVA and the management trustworthiness ANCOVA, respectively.
32
APPENDIX MANAGEMENT CREDIBILITY QUESTIONS
Panel A: Competence Questionsa
1. I believe that Dentex management is very competent at providing financial disclosures. 2. I believe that Dentex management has little knowledge of the factors involved in providing
useful disclosures. 3. I believe that few people are as qualified as Dentex management to provide useful financial
disclosures about Dentex. Panel B: Trustworthiness Questionsa
4. I believe that Dentex management is very trustworthy. 5. I believe that Dentex management is very honest. 6. I believe that Dentex management may not be truthful in their financial disclosures. a Participants were asked to evaluate each sentence on a 7-point scale with endpoints labeled ‘Strongly Disagree’ and ‘Strongly Agree.’