Description
The purpose of this paper was to examine whether a less precise (or imprecise) estimate
may increase investors’ confidence and improve investors’ perceptions of fair value reliability. The
main criticism of fair value accounting has been its lack of reliability perceived by investors.
Accounting Research Journal
The joint effects of management incentive and information precision on perceived
reliability in fair value estimates
Ning Du J ohn E. McEnroe Kevin Stevens
Article information:
To cite this document:
Ning Du J ohn E. McEnroe Kevin Stevens , (2014),"The joint effects of management incentive and
information precision on perceived reliability in fair value estimates", Accounting Research J ournal, Vol. 27
Iss 2 pp. 188 - 206
Permanent link to this document:
http://dx.doi.org/10.1108/ARJ -10-2012-0081
Downloaded on: 24 January 2016, At: 21:20 (PT)
References: this document contains references to 33 other documents.
To copy this document: [email protected]
The fulltext of this document has been downloaded 1418 times since 2014*
Users who downloaded this article also downloaded:
Prodosh Simlai, (2014),"Firm characteristics, distress risk and average stock returns", Accounting Research
J ournal, Vol. 27 Iss 2 pp. 101-123 http://dx.doi.org/10.1108/ARJ -06-2012-0046
Etumudon Ndidi Asien, (2014),"Impact of firm-specific characteristics on managers’ identity disclosure",
Accounting Research J ournal, Vol. 27 Iss 2 pp. 150-168 http://dx.doi.org/10.1108/ARJ -03-2013-0010
Shu Hui Wee, Soon Yau Foong, Michael S.C. Tse, (2014),"Management control systems and organisational
learning: the effects of design and use", Accounting Research J ournal, Vol. 27 Iss 2 pp. 169-187 http://
dx.doi.org/10.1108/ARJ -05-2013-0026
Access to this document was granted through an Emerald subscription provided by All users group
For Authors
If you would like to write for this, or any other Emerald publication, then please use our Emerald for
Authors service information about how to choose which publication to write for and submission guidelines
are available for all. Please visit www.emeraldinsight.com/authors for more information.
About Emerald www.emeraldinsight.com
Emerald is a global publisher linking research and practice to the benefit of society. The company
manages a portfolio of more than 290 journals and over 2,350 books and book series volumes, as well as
providing an extensive range of online products and additional customer resources and services.
Emerald is both COUNTER 4 and TRANSFER compliant. The organization is a partner of the Committee
on Publication Ethics (COPE) and also works with Portico and the LOCKSS initiative for digital archive
preservation.
*Related content and download information correct at time of download.
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
The joint effects of management
incentive and information
precision on perceived reliability
in fair value estimates
Ning Du, John E. McEnroe and Kevin Stevens
School of Accountancy and Management Information Systems (MIS),
DePaul University, Chicago, Illinois, USA
Abstract
Purpose – The purpose of this paper was to examine whether a less precise (or imprecise) estimate
may increase investors’ confdence and improve investors’ perceptions of fair value reliability. The
main criticism of fair value accounting has been its lack of reliability perceived by investors.
Design/methodology/approach – A 2 ? 3 randomized experiment was used where management
incentive and information precision are manipulated.
Findings – The results from this study indicate that perceived reliability is jointly affected by
management’s incentives andinformationprecision. Reliabilityratingis the highest for fair value stated
as a point estimate with a specifed confdence level attached to it. Further analysis indicates that higher
perceived reliability is related to its representational faithfulness because participants perceive that a
point estimate with a specifed confdence level better matches uncertainty in measuring future cash
fows.
Originality/value – This is the frst study to examine whether a less precise (or imprecise)
estimate may increase investors’ confdence and improve investors’ perceptions of fair value
reliability. Because of the subjectivity and uncertainty in fair value estimates, less precise fair
value estimates may not be viewed as less reliable. In fact, using a precise format to represent fair
value estimates may not be appropriate (neither reliable nor credible), because a precise point
estimate fails to capture its underlying uncertainty in future cash fows. Aless precise format could
represent a credible choice for fair value because it refects uncertainty and subjectivity and
effectively communicates management’s assessments of variability in future cash fows.
Keywords Uncertainty, Reliability, Fair value, Management bias
Paper type Research paper
Introduction
The past few years have witnessed the global convergence of national accounting
standards and International Accounting Standards [IAS, superseded by International
Financial Reporting Standards (IFRS)]. Since 2005, when European Union countries frst
mandated its use, more than 100 countries have adopted IFRS. Over the past few years,
the US Securities Exchange Commission has taken several steps to acknowledge IFRSs
as a high-quality set of fnancial reporting standards. These steps include elimination of
the reconciliation to USA Generally Accepted Accounting Principles (USA GAAP) by
foreign private issuers using IFRSs, and the issuance of a proposed “roadmap” for the
use of IFRS by US domestic registrants. Although the movement toward a worldwide
single-set of accounting standards has slowed, convergence projects between the
The current issue and full text archive of this journal is available at
www.emeraldinsight.com/1030-9616.htm
ARJ
27,2
188
Accounting Research Journal
Vol. 27 No. 2, 2014
pp. 188-206
©Emerald Group Publishing Limited
1030-9616
DOI 10.1108/ARJ-10-2012-0081
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Financial Accounting Standards Board (FASB) and the International Accounting
Standards Board (IASB, 2011) in the areas of leases, revenue recognition and fnancial
instruments continue.
Recent developments in IFRS tend to move the standards toward the use of fair value
accounting even for non-fnancial assets. If the USA adopts IFRS, then a greater use of
fair value accounting necessarily follows. Because of the uncertainty inherent in fair
value accounting (as compared to historical cost) and the lack of implementation
guidance, fair value accounting remains vulnerable to managerial opportunismthrough
accounting and/or transaction decisions (Ball et al., 2003). Moreover, fair value estimates
are uncertain because predictions must be made of future market value. The reliability
concern is particularly serious in cases where the observable market prices are not
available, and the company must rely on the mark-to-model approach to derive fair
value estimates. Because human judgments are prone to errors and biases, it is not
surprising that fnancial statement users are skeptical of the objectivity and reliability of
fair value information. The lack of reliability may limit the relevance of fair value
estimates, reduce the comparability of fnancial reports worldwide and ultimately
impair decision usefulness.
Despite the on-going debate about its relevance and reliability, fair value accounting
has gained increasing importance. On May 12, 2011, IASB issued IFRS 13, Fair Value
Measurement, and the FASB issued Accounting Standard Update No. 2011-04, Fair
Value Measurement (Topic 820): Amendments to Achieve Common Fair Value
Measurement and Disclosure Requirements in USA GAAP and IFRSs (ASC 820)
(together, hereafter referred to as “the new guidance”). The new guidance results in a
consistent defnition of fair value and requires common measurements of fair value
between IFRS and USAGAAP. In addition to increasing global harmonization, the new
guidance also signifcantly enhances the disclosure requirements for fair value
accounting.
ASC 820 and IFRS 13 are consistent in their frameworks used for fair value
measurement. Both establish a three-level hierarchy for measuring fair value. Level 1
and 2 inputs are mark-to-market inputs and use quoted prices in active markets for
identical or similar assets or liabilities. Level 3 inputs are mark-to-model and require the
frm to use a valuation model to estimate the market value of the assets or liabilities.
Level 3 estimates require considerable management judgment in the selection and
application of valuation techniques, and therefore, require effective internal controls and
knowledgeable independent auditors and other corporate governance mechanisms to
thwart potential abuses. Because of its potential for judgmental bias and errors, whether
fair values derived from the models refect economic reality and indeed represent the
“true” values of an entity become questionable in many investors’ minds.
Given this background, the goal of this study is to understand how to improve
investors’ perception of fair value reliability. To achieve this goal, we must frst
differentiate between real reliability and perceived reliability. Real reliability refers to
the uncertainty or variability inherent in information content, commonly defned as the
closeness between the benchmark (i.e. what it purports to represent) and accounting
representation[1]; in contrast, perceived reliability refers to user’s perceptions or
understanding of disclosed information. Given the high degree of uncertainty and
subjectivity inherent in fair value measurements, accounting standard setters and
preparers may not be able to signifcantly improve the real reliability of information
189
Management
incentive and
information
precision
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
content. However, we argue that a careful choice of presentation format may help
improve the information’s perceived reliability. Indeed, prior research has found that
characteristics of management disclosure, such as precision, venue or time horizon,
affect the disclosure reliability perceived by investors (Koonce et al., 1998; Mercer, 2004).
Disclosure precision, an important dimension of management disclosure, has been
found to affect investors’ judgments and perceptions. Specifcally, imprecise disclosures
are viewed as less reliable/credible than more precise disclosures because imprecision
signals management uncertainty (Baginski et al., 2004; Hirst et al., 1999).
But, because of the subjectivity and uncertainty in fair value estimates, less precise
fair value estimates may not be viewed as less reliable. In fact, using a precise format to
represent fair value estimates may not be appropriate (i.e. may be neither reliable nor
credible), because a precise point estimate fails to capture the underlying uncertainties
in future cash fows. Aless precise format could represent a credible choice for fair value
because it refects uncertainty and subjectivity and effectively communicates
management’s assessments of variability in future cash fows (Budescu and Du, 2007;
Du and Budescu, 2005; Du et al., 2011, 2010). Thus, we examine whether a less precise (or
imprecise) estimate may increase investors’ confdence and improve investors’
perceptions of fair value reliability. In addition, we investigate whether different types
of imprecise estimates may have differential effects on investors’ reliability judgments.
Last, we examine whether management incentives may moderate or accentuate these
effects. To study these questions we use a 2 ? 3 between-subjects randomized
experiment. We manipulate incentive at two levels (gain or loss) and precision at three
levels (a precise point estimate, a point estimate at 75 per cent confdence level or a range
estimate). MBAstudents participate in the experiment and judge the degree of reliability
and bias in the fair value estimate provided by the management. Results from the
experiment showthat participants viewfair value gain to be more biased than fair value
loss regardless of its precision. In addition, we fnd that when the fair value gain is stated
as a point estimate at 75 per cent confdence level, participants believe it is more reliable
than the precise point estimate. The results from this study are timely and provide a
priori evidence to the IASB and FASB in evaluating the effectiveness of fair value
disclosures.
Literature review and hypothesis development
Imprecise information. Improving perceived reliability of fair value estimates has
inspired much research interest in the past. For example, Frederickson et al. (2006)
and Blacconiere et al. (2009) investigated perceived fair value reliability in the
context of employee stock option compensation (SOC) mandated under Statement of
Financial Accounting Standard No. 123R. In particular, they examined disavowals,
a unique form of management voluntary disclosures, which explicitly question the
reliability or usefulness of the mandated information in audited fnancial
statements. The results from their studies suggest that by disclosing the lack of
reliability in fair value estimates in SOC, management disavowals help investors
properly weigh fair value estimates in SOC, and thus, improve investors’
perceptions of the reliability of those estimates.
The reliability concern is particularly germane for mark-to-model assets or liabilities.
For example, the discounted cash fow model is extremely sensitive because even a
small change in inputs may result in large valuation differences. In particular, managers
ARJ
27,2
190
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
must be able to objectively assess future risks and uncertainties and use a discount rate
that includes appropriate risk adjustments. Further, managers must be able to estimate
both the amount and timing of future cash fows. Maines and Wahlen (2006) argue that
reliable accounting information must be able to signal differences in the magnitude as
well as the uncertainty inherent in future cash fows. Although future cash fows may be
highly uncertain and diffcult to measure reliably in the current period, revealing the
degree of uncertainty associated with the values of assets and liabilities forecasted by
cash fow models may provide representationally faithful information about the
uncertainty of cash fow modeling.
Prior experimental evidence suggests that investors prefer the precision of
information to match its underlying uncertainty (Budescu and Du, 2007; Du and
Budescu, 2005; Du et al., 2011, 2010). A precise point estimate refects certainty and
masks the risks and uncertainties inherent in fair value measurement and accordingly
fails the faithful representation criterion. Unlike a precise point estimate, imprecise fair
value estimates are probabilistic and refect variability in future cash fows. The choice
of imprecise fair value estimates is similar to management disavowal and signals
greater uncertainty about the accuracy of unobserved inputs, and by extension, the
accuracy of the fair value estimate (Baginski et al., 1993). Therefore, we argue that by
varying degree of precision in fair value estimates, managers reveal howconfdent they
feel about future cash fow; this signal helps investors properly weigh the reliability of
fair value information. Knowing the limitations of fair value information increases
investors’ confdence in using the information and ultimately improves investors’
perceptions of fair value reliability which improves its decision usefulness. This
assumption leads us to H1:
H1. Investors perceive higher reliability in imprecise fair value estimates than
precise fair value estimates.
Different formats of imprecise information
Imprecise estimates communicate probabilistic information. An estimate may come as a
point estimate with a specifed probability to indicate a confdence level (e.g. managers
are 50, 75 or 90 per cent confdent that the fair value is $100,000) or as an interval
estimate spanning minimum and maximum values (e.g. fair value ranges from $75,000
to $125,000). Winman et al. (2004) suggest that probability estimates and confdence
intervals (ranges) are formally equivalent because high (low) uncertainties can be
expressed by low (high) probability judgments or by wide (narrow) interval (range)
estimates assuming that people’s beliefs are symmetrical. For example, the 75 per cent
confdence interval of $100,000 is equivalent to the interval ranging from $75,000 to
$125,000 if we assume the standard deviation as $21,739 with a normal distribution.
Empirically, however, these two different forms of estimates have produced
systematically different judgments (Budescu and Du, 2007; Rottenstreich and Tversky,
1997). Prior evidence shows that the interval estimate induces higher error and bias than
the point probability estimate (Klayman et al., 1999; Juslin et al., 2000), perhaps because
individuals have diffculty in inferring the probability or confdence level from an
interval estimate. Unlike an interval estimate where users have to infer managers’
confdence in management’s judgment, a point estimate with a specifed confdence level
191
Management
incentive and
information
precision
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
(e.g. 75 per cent) may lead to an increase in perceived reliability. Given this, we
developed our second hypothesis as follows:
H2. The reliability rating will be the highest when the fair value estimate is
presented as a point estimate with a specifed confdence level.
Management incentives
The IFRS framework includes both relevance and reliability as underlying assumptions
and refers to possible trade-offs between these objectives in paragraph 32 of IAS 1:
“Information may be relevant but so unreliable in nature or representation that its
recognition may be potentially misleading”. Furthermore, IAS 1also states that
information in the fnancial statements is reliable if it is free from material errors and
bias (see paragraph 31). Because fair value estimates, especially those with Level 3
inputs, are based on judgmental calculations of hypothetical values and markets, it can
become diffcult to determine whether those inputs are free of errors and preparer bias.
For example, He et al. (2012) fnd that Chinese managers “cherry pick” their investment
portfolio, and are more likely to sell investment securities for a gain to avoid reporting
fnancial losses after frst IFRS-based China Accounting Standards were adopted in
2007. Prior literature suggests that the degree of bias users perceive in information
reported by management depends on the consistency between the reported information
and the users’ beliefs about management’s reporting incentives (Frederickson et al.,
2006; Hirst et al., 1995; Hodge et al., 2006). Attribution and persuasion theory (Eagly and
Chaiken, 1975; Fiske and Pavelchak, 1986) suggests that individuals consider
incentive-consistent signals to be strategic and biased, and thus less credible (Hirst et al.,
1995; Hodge et al., 2006). Because fair value estimates result in gains or losses that
eventually fow through owner’s equity, managers may strategically use fair value
accounting to produce desired performance outcomes for personal gains. As such, we
believe managers have strong incentive to bias their fair value estimates to a gain
condition; further, this incentive is strong enough to be insensitive to information
precision. In other words, managers, under the infuence of their economic incentives,
may bias a precise point estimate as well as the range or confdence level in the fair value
estimate. The potential for management error and partiality suggests that investors will
be skeptical of fair value approximations resulting in potential gains but not those in
losses. Thus, we expect a main effect of incentive on perceived bias judgment which
leads to our third hypothesis:
H3. Investors perceive higher management bias in fair value estimates leading to
potential gains than potential losses.
Joint effects of management incentives and information precision
When both incentives and precision are present, we expect the two factors may jointly
affect investors’ reliability judgments. Specifcally, the precision of fair value estimate
may affect investors’ reliability judgments differently depending on whether fair value
estimates are incentive consistent (gain) or inconsistent (loss). As a fair value gain is
consistent with meeting managerial incentives, we expect investors to be skeptical of
fair value estimates when they result in gains; in addition, we assume that investors will
actively look for cues, such as information precision, to improve decision usefulness.
Because a point estimate with a specifed confdence level (e.g. 75 per cent) is expressed
ARJ
27,2
192
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
as a precise probability value and clearly communicates the degree of uncertainty in the
future cash fow, we anticipate that investors will prefer this format over a range
estimate. Specifcally, we argue that the degree of improvement in perceived reliability
is the greatest when incentive consistent (gain) fair value is presented imprecisely as a
point estimate quantifed by a specifed confdence level. Thus, our last hypothesis tests
the joint effects of management incentive and information precision on perceived fair
value reliability and is stated as follows:
H4. Apoint fair value estimate with a specifed confdence level is perceived to be
more reliable than a range fair value estimate or a precise point estimate
when this fair value estimate results in potential gains, but the difference in
perceived reliability may disappear when the fair value estimate results in
potential losses.
Methodology
Participants
The participants are 114 MBA students from a large private university. Sixty-three are
males, and 51 are females. Students volunteered to participate in response to in-class
announcements and were randomly assigned to six different experimental conditions.
Demographic information collected at the end of the experiment indicates that most
participants are likely to have the necessary knowledge and experience to complete the
task. The majority of participants have both work and investment experience. On
average, participants have 6.5 years of working experience and 4.5 years of investment
experience.
Procedure and task
In the experiment, we asked the participants to assume that they work in the investment
department of a frm. Their task is to judge the degree of reliability and management
bias in the fair value estimate provided by the management. We provided participants
with a short summary which states that two years ago, Co. A acquired a minority
interest in Co. B, as well as detailed information about the transaction. After reviewing
the provided information, participants answered a series of questions about Company A
and its investment in Company B. Finally, the participants fnished by answering a
short series of manipulation check questions and are fully debriefed. The entire
experiment took 20 minutes, on average, to complete. (See Appendix for a copy of the
experimental instrument.)
Independent variables
We use a 2 (Incentive) ?3 (Precision) between-subjects design. The incentive variable is
manipulated at two levels (gain or loss) and the precision variable at three levels (a
precise point estimate, a point estimate at 75 per cent confdence level and a range
estimate). (The point estimate at 75 per cent confdence level is referred to as the 75 per
cent point hereafter.) Participants are randomly assigned to experimental conditions
and read the following instruction:
Please assume that you work in the investment department of a frm. Your supervisor has
instructed you to assess the fnancial performance of Company A, a 100-year old fnancial
193
Management
incentive and
information
precision
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
boutique frm that through its wholly owned and partially owned subsidiaries is engaged in
fnancial services activities. After extensive research, you prepared a short summary with
some important facts about Company A:
Co. A acquired a minority interest in Co. B two years ago at a price of $60,000 ($140,000). B is
a non-public, technology-based asset management company that is in the process of growing
its business and has the ultimate goal of becoming one of the premier investment advising
companies in the market. However, Company Bhas not yet achieved a positive cash fowin its
month-to-month operations at December 31, 2009.
At the end of 2009, the fair value of B is estimated at $100,000, $40,000 higher (lower) than its
original purchase price of $60,000 ($140,000). Company A adjusts its fnancial statements to
refect this increase (decrease). The investment in Company B is nowcarried at $100,000 as an
asset in the balance sheet, and the increase (decrease) of $40,000 is recognized as an unrealized
holding gain (loss) in its comprehensive income statement.
The fair value of $100,000 is estimated by Company A’s management. Because there is no
publicly observed market-based information about Company B (Company B is not traded in
active market, and there is no company in the marketplace that is comparable to Company B),
Company A’s management used a discounted cash fow analysis as the valuation model. The
model uses projected fnancial information for the next fve years as the inputs and relies
heavily on managers’ assumptions of future cash fow and choice of risk adjusted-discount
rate.
We manipulate management incentive as the direction of comprehensive income change
by holding the fair value constant at $100,000. We vary the historical cost at $60,000 or
$140,000, respectively, to create fair value gain or loss for this investment. In the fair
value gain (loss) condition, participants are told that the interest in Co. B was acquired
two years ago at a price of $60,000 ($140,000) and the 2009 change in other
comprehensive income was mainly attributed to the adjustment of a $40,000 increase
(decrease) in fair value.
We manipulate the precision of fair value at three levels. In the point estimate
condition, the fair value is given as a point estimate of $100,000. In the 75 per cent point
confdence condition, we add the sentence: “Though the fair value is estimated at
$100,000 for B, A’s management indicates that they are only 75 per cent confdent about
this estimate because of the assumptions used”. In the interval estimate condition, we
add the sentence:
Though the fair value is estimated at $100,000 for B, A’s management indicates that this
estimate may range as low as $75,000 and as high as $125,000, depending on the assumptions
used.
These three conditions have the same expected value of $100,000. That is, the midpoint
of the interval estimate of $75,000-$125,000 is $100,000, and the 75 per cent confdence
level of $100,000 is also $75,000-$125,000.
Finally, all participants are told that:
Before you make recommendation to your clients about whether to invest in Company A, your
supervisor would like to knowthe basis for your judgments. Please proceed to the next page to
answer the following questions.
ARJ
27,2
194
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Dependent variables
Dependent variables include perceived reliability and the degree of management bias.
They are based on the following questions:
• For the purpose of making investment-related decisions, how reliable do you fnd
the fair value estimate of $100,000? (1 being not reliable at all, 11 being very
reliable).
• When using the discounted cash fow model, how likely would Company A’s
management intentionally use assumptions that lead to the current valuation of
Company B? (1 being very unlikely, 11 being very likely).
Results
Manipulation checks
To check the incentive (fair value gain or loss) manipulation, we ask participants
whether the fair value estimate of Company B is higher or lower than the historical cost.
All participants checked the correct answers. For the precision manipulation, we ask
two questions. First, we ask participants to check the fair value estimate that is included
in their cases ($100,000, $100,000 at 75 per cent confdence level and $75,000 to $125,000).
Only one participant checked the wrong answer (excluding this participant does not
change the results).
The riskiness of an investment is an important factor in investment decisions. The
degree of risk perceived by participants may affect their reliability judgments. We asked
the participants to assess how risky Company A’s investment in Company B is on an
11-point scale (1 being not risky, 11 being very risky). We conducted an analysis of
variance (ANOVA) on perceived risk and found that participants perceived fair value
loss (mean ?7.82) as signifcantly riskier than fair value gain (mean ?7.07) [F(1, 104) ?
4.28, (p ? 0.05)]. However, the precision factor and its interaction with the incentive
factor are not signifcant.
Multivariate analysis of covariance
As the two dependent variables, perceived reliability and bias ratings, are potentially
correlated, we rely on multivariate analysis of variance (MANOVA) for the initial
analysis. To control for any potentially extraneous infuence on dependent variables, we
include gender, investment experience, work experience and the ratings on perceived
risk as covariates in our analysis. Multivariate analysis of covariance indicates none of
the four control variables is signifcant. Specifcally, for risk, Wilks’ lambda is 0.97 (F ?
1.09, p ? 0.05); for work experience, Wilks’ lambda is 0.99 (F ? 0.04, p ? 0.05); for
investment experience, Wilks’ lambda is 0.99 (F?0.47, p ?0.05); and for gender, Wilks’
lambda is 0.96 (F ? 1.72, p ? 0.05). In addition, MANOVA indicates that the
independent factor of precision is insignifcant (Wilks’ lambda ? 0.96, F ? 0.87, p ?
0.05), but the independent factor of bias (Wilks’ lambda ? 19.13, p ? 0.05) and their
interaction (Wilks’ lambda ? 0,90, F ? 2.45, p ? 0.05) are signifcant. To explain the
signifcant results, we conduct ANOVA in the following section for each of the
dependent variables.
Analysis of variance
Tables I-IV report the means for reliability ratings which are also plotted in Figure 1.
The descriptive statistics showthat participants do not perceive the fair value estimates
195
Management
incentive and
information
precision
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
as very reliable, as most of the ratings are quite low. Means are 5.22 for precise point,
5.56 for 75 per cent point and 5.10 for range. In general, the fair value loss (mean ?6.08)
is perceived signifcantly more reliable than the fair value gain (mean ? 4.50)
[F(1, 103 ?18.70, p ?0.05]. We suspect that the lower reliability in the gain condition
might have been driven by higher perceived management bias. H1 predicts that
investors perceive precise information to be less reliable than imprecise information. To
test H1, we treat the point estimate as the category of precise estimate. Then we consider
the range estimate and the point estimate at 75 per cent confdence level as one category
of imprecise estimate. We then compare the reliability ratings between these two
groups. As presented in Table 1, the mean rating for imprecise estimates (5.42) is
slightly higher than the mean rating for precise point estimate (5.22), but the difference
is not statistically different. Thus, our frst hypothesis is not supported.
H2 predicts that the reliability rating will be the highest when the fair value estimate
is presented as a point estimate at 75 per cent confdence level. Table II shows that the 75
per cent confdence estimate has the highest reliability rating (mean ?5.73), followed by
precise point estimate (mean ? 5.22) and range estimate (mean ? 5.10). However, the
ANCOVA analysis in Table III suggests the effect of precision is not statistically
signifcant [F (1, 103) ?0.39, p ?0.05]. Thus, our data fail to support H2.
H3 predicts that the perceived degree of management bias will be higher in the fair
value gain condition than in the fair value loss condition regardless of the precision of
fair value information. Table V presents the descriptive statistics on management bias,
which are also plotted in Figure 2. The ANOVA indicates the fair value is perceived to
Table I.
Perceived reliability of fair
value estimates rating
(tests of H1, H2 and H4):
mean perceived reliability
rating (test of H1)
Precise Imprecise
Point 75 per cent confdence Range
5.22 5.73 5.10
(0.36) (0.22) (1.99)
n ?36 n ?38 n ?38
5.42 (average)
Notes: The perceived reliability rating is for the following question on an 11-point scale: For the
purpose of making investment-related decisions, how reliable do you fnd the fair value estimate of
$100,000? (1 being not reliable at all, 1 being very reliable). The dependent variable is perceived
reliability; the independent variable is information precision and is manipulated at three levels: point
stands for a precise point estimate; 75 per cent confdence stands for a point estimate at 75 per cent
confdence level; and range stands for a confdence interval; H1 predicts that investors perceive higher
reliability in imprecise fair value estimates than precise fair value estimates; to test H1, we treat the
point estimate as the category of precise estimate (average mean of 5.22); then we consider the range
estimate and the point estimate at 75 per cent confdence level as one category of imprecise estimate
(average mean of 5.42); H2 predicts that investors perceive higher reliability when fair value estimate is
presented as a point estimate with a specifed confdence level (average mean of 5.73) than an interval
range estimate (average mean of 5.10) and a precise point estimate (average mean of 5.22); to test H2, we
compare the average means for the three levels of precision; H4 predicts that a point fair value estimate
with a specifed confdence level is perceived to be more reliable than a range fair value estimate or a
precise point estimate when this fair value estimate results in potential gains; fair value precision has no
effect on perceived reliability when fair value estimate results in potential losses; to test H4, we conduct
simple effect tests to evaluate pairwise differences among the adjusted means using the Holm’s
sequential Bonferroni to control for Type I error across the three pairwise comparisons
ARJ
27,2
196
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
be more biased in the gain condition (mean ?9.06) than in the loss condition (mean ?
7.06). The degree of bias does not seem to vary greatly for point (mean ? 7.81),
75 per cent point (mean ?8.59) and range (mean ?7.78). The ANOVAresults (Table VI)
indicate a signifcant main effect between gain and loss [F(1, 101) ? 22.50, p ? 0.01].
Neither the precision nor its interaction with the incentive factor is signifcant. This
suggests that participants believe that fair value with gain is signifcantly more biased
than that with loss. These results support H3 (Table VII).
H4 predicts that investors perceive the point estimate with a specifed confdence
interval (e.g. at 75 per cent) to be the most reliable estimate when fair value results in
potential gain. The perceived reliability does not vary when fair value estimates result in
potential loss. Figure 1 shows that in the gain condition, the fair value at 75 per cent
point estimate has the highest rating with a mean of 5.56, very close to the mean rating
of 5.90 for the 75 per cent point estimate in the loss condition. Table II presents the
results for ANOVA. We observe a signifcant interaction between the two independent
variables of precision and incentive [F(2, 103) ?5.76, p ?0.05], as well as a main effect
for direction [F(1, 103) ? 18.70, p ? 0.01]. To interpret the main effect of direction, we
conduct simple effect tests to evaluate pairwise differences among the adjusted means.
The Holm’s sequential Bonferroni procedure was used to control for Type I error across
the three pairwise comparisons. Results of simple effects are presented in Table IV.
Apparently, there is a signifcant difference in the adjusted means between precise point
(mean ?3.63) and 75 per cent point (mean ?5.56) in the gain condition (p ?0.05), but no
signifcant difference in the other comparisons. Though the rating in the range condition
Table II.
Perceived reliability of fair
value estimates rating
(tests of H1, H2 and H4):
mean perceived reliability
rating by gain or loss (test
of H2 and H4)
Condition Point 75 per cent confdence Range
Gain 3.63 (2.19) 5.56 (2.48) 4.30 (1.95)
n ?16 n ?18 n ?20
Loss 6.80 (1.51) 5.90 (1.69) 5.89 (1.71)
n ?20 n ?20 n ?18
Average 5.22 5.73 5.10
Notes: The perceived reliability rating is for the following question on an 11-point scale: For the
purpose of making investment-related decisions, how reliable do you fnd the fair value estimate of
$100,000? (1 being not reliable at all, 1 being very reliable) The dependent variable is perceived
reliability; the independent variable is information precision and is manipulated at three levels: point
stands for a precise point estimate; 75 per cent confdence stands for a point estimate at 75 per cent
confdence level; and range stands for a confdence interval; H1 predicts that investors perceive higher
reliability in imprecise fair value estimates than precise fair value estimates; to test H1, we treat the
point estimate as the category of precise estimate (average mean of 5.22); then we consider the range
estimate and the point estimate at 75 per cent confdence level as one category of imprecise estimate
(average mean of 5.42); H2 predicts that investors perceive higher reliability when fair value estimate is
presented as a point estimate with a specifed confdence level (average mean of 5.73) than an interval
range estimate (average mean of 5.10) and a precise point estimate (average mean of 5.22); to test H2, we
compare the average means for the three levels of precision; H4 predicts that a point fair value estimate
with a specifed confdence level is perceived to be more reliable than a range fair value estimate or a
precise point estimate when this fair value estimate results in potential gains; fair value precision has no
effect on perceived reliability when fair value estimate results in potential losses; to test H4, we conduct
simple effect tests to evaluate pairwise differences among the adjusted means using the Holm’s
sequential Bonferroni to control for Type I error across the three pairwise comparisons
197
Management
incentive and
information
precision
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
is lower than the rating in the 75 per cent point condition, the difference fails to reach
statistical signifcance. This evidence partially supports H4.
Additional analysis
Because of our focus on the effects of information precision, it is important to understand
how sensitive our results are to the confdence level chosen (e.g. 50, 60 or 75 per cent) or
size of the range provided. To this end, we conducted another experiment where we
varied the level of confdence and size of the range. We recruited 45 MBA students and
randomly assigned themto three conditions. We provided themwith the same case as in
the prior experiment but with slight variations. Because our results suggest that
information precision affects reliability and bias judgments when fair value estimates
produce potential gains but not losses, the new experiment solely focuses on the gain
condition. In addition, the newexperiment includes three different levels of imprecision.
For the point estimate with confdence level, we provided participants with a lowered
confdence level of 50 per cent. For the range estimate, we widened the interval to lower
the confdence; specifcally, we provide participants with a wider interval, which ranges
as low as $50,000 and as high as $150,000. Moreover, we include another condition of
imprecise statement by providing a statement that fair values are imprecise without
providing either a confdence level or range; specifcally, we provide participants with
the following statement: “Though the fair value is estimated at $100,000 for B, A’s
management indicates that this number is imprecise”.
Table III.
Perceived reliability of fair
value estimates rating
(tests of H1, H2 and H4):
two-way ANOVA with
perceived reliability rating
as the dependent variable
(tests of H1, H2 and H4)
Source SS df MS F p (2-sided)
Precision 2.93 2 1.47 0.39 0.68
Direction 70.47 1 70.47 18.70 0.00
Precision ?direction 43.43 2 21.72 5.76 0.00
Within cells (error) 508.218 104 3.77
Notes: The perceived reliability rating is for the following question on an 11-point scale: For the
purpose of making investment-related decisions, how reliable do you fnd the fair value estimate of
$100,000? (1 being not reliable at all, 1 being very reliable) The dependent variable is perceived
reliability; the independent variable is information precision and is manipulated at three levels:
point stands for a precise point estimate; 75 per cent confdence stands for a point estimate at 75 per
cent confdence level; and range stands for a confdence interval; H1 predicts that investors
perceive higher reliability in imprecise fair value estimates than precise fair value estimates; to test
H1, we treat the point estimate as the category of precise estimate (average mean of 5.22); then we
consider the range estimate and the point estimate at 75 per cent confdence level as one category
of imprecise estimate (average mean of 5.42); H2 predicts that investors perceive higher reliability
when fair value estimate is presented as a point estimate with a specifed confdence level (average
mean of 5.73) than an interval range estimate (average mean of 5.10) and a precise point estimate
(averagemeanof 5.22); totest H2, wecomparetheaveragemeansfor thethreelevelsof precision; H4predicts
that a point fair value estimate with a specifed confdence level is perceived to be more reliable than a range
fair value estimate or a precise point estimate when this fair value estimate results in potential gains; fair
value precisionhas no effect onperceivedreliabilitywhenfair value estimate results in potential losses; to
test H4, we conduct simple effect tests to evaluate pairwise differences among the adjusted means
using the Holm’s sequential Bonferroni to control for Type I error across the three pairwise
comparisons
ARJ
27,2
198
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
The descriptive statistics are reported in Table VIII. The reliability rating for 50 per cent
confdence level is 4.93, 4.07 for wider range and 4.20 for imprecise verbal statement.
Apparently, participants perceive the point estimate with a specifed 50 per cent
confdence level to be the most reliable, consistent with fndings in the prior experiment.
However, the one-way ANOVA indicates that the difference in means across the three
conditions is not statistically signifcant [F(51,2) ? 0.94, p ? 0.40]. The average
reliability rating is 4.40, signifcantly lower than the reliability rating in the prior
experiment. This suggests that as managers lower their confdence, participants
become more skeptical about the reliability of fair value estimates, and become less
sensitive of information precision. Furthermore, the bias rating for the 50 per cent
confdence level is 6.86, 6.80 for wider range and 7.52 when we introduce the imprecise
verbal statement, with an average rating of 7.06 [F(2,51) ?0.63, p ?0.54]. The average
rating is very similar to the rating in the prior experiment. This suggests the perceived
management bias does not vary as the level of information precision changes.
Conclusions
Fair value accounting is often criticized for its lack of reliability. Because of the high
degree of uncertainty and subjectivity inherent in fair value measurement, not much can
Table IV.
Perceived reliability of fair
value estimates rating
(tests of H1, H2 and H4):
results of simple effects on
perceived reliability (test
of H4)
Condition Mean difference Standard deviation Signifcance
Gain
Point 75 per cent confdence ?1.84 0.68 0.02
Point Range ?0.68 0.65 0.91
75 per cent confdence Range 1.17 0.64 0.22
Loss
Point 75 per cent confdence 1.28 0.63 0.14
Point Range 0.91 0.63 0.46
75 per cent confdence Range ?0.37 0.65 1.00
Notes: The perceived reliability rating is for the following question on an 11-point scale: For the
purpose of making investment-related decisions, how reliable do you fnd the fair value estimate of
$100,000? (1 being not reliable at all, 1 being very reliable) The dependent variable is perceived
reliability; the independent variable is information precision and is manipulated at three levels: point
stands for a precise point estimate; 75 per cent confdence stands for a point estimate at 75 per cent
confdence level; and range stands for a confdence interval; H1 predicts that investors perceive higher
reliability in imprecise fair value estimates than precise fair value estimates; to test H1, we treat the
point estimate as the category of precise estimate (average mean of 5.22); then we consider the range
estimate and the point estimate at 75 per cent confdence level as one category of imprecise estimate
(average mean of 5.42); H2 predicts that investors perceive higher reliability when fair value estimate is
presented as a point estimate with a specifed confdence level (average mean of 5.73) than an interval
range estimate (average mean of 5.10) and a precise point estimate (average mean of 5.22); to test H2, we
compare the average means for the three levels of precision; H4 predicts that a point fair value estimate
with a specifed confdence level is perceived to be more reliable than a range fair value estimate or a
precise point estimate when this fair value estimate results in potential gains; fair value precision has no
effect on perceived reliability when fair value estimate results in potential losses; to test H4, we conduct
simple effect tests to evaluate pairwise differences among the adjusted means using the Holm’s
sequential Bonferroni to control for Type I error across the three pairwise comparisons
199
Management
incentive and
information
precision
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
be done to improve the reliability of its information content, given the profession’s
extant measurement techniques. However, we argue that a careful choice of its
presentation format may improve its perceived reliability in users’ minds. Our evidence
provides some support to this conjecture. In general, participants believe fair value
estimates resulting in gain as less reliable than those in loss, potentially due to higher
management bias perceived in the gain condition. Our results also suggest that, instead
of relying on a precise point estimate, managers may improve the perceived reliability of
fair value estimates in the gain condition by specifying the degree of confdence. This
disclosure can be included in the notes to the fnancial statements or through other
channels.
This experimental investigation, as in the case of others, is limited in several
respects. We manipulate the precision of fair value at three levels, a precise point
3.63
5.56
4.3
6.8
5.9 5.89
3.5
4
4.5
5
5.5
6
6.5
7
Point 75% confident Range
R
e
l
i
a
b
i
l
i
t
y
Gain
Loss
Notes: This figure presents the means for the
dependent variables: perceived management bias
and perceived reliability. The horizontal axis
presents the independent variable at three levels:
point stands for a precise point estimate; 75 per
cent confidence stands for a point estimate at 75
per cent confidence level; and range stands for a
confidence interval. The ratings are presented on
the vertical axis. They are based on the following
questions: For the purpose of making investment-
related decisions, how reliable do you find the
fair value estimate of $100,000? (1 being not
reliable at all, 1 being very reliable). When using
the discounted cash flow model, how likely
would Company A’s management intentionally
use assumptions that lead to the current valuation
of Company B? (1 being very unlikely, 11 being
very likely)
Figure 1.
Various judgments of fair
value: perceived reliability
Table V.
Perceived management
bias rating (test of H3):
bias (mean) (test of H3)
Condition Point 75 per cent confdence Range
Gain 8.62 (2.58) 9.56 (1.54) 9.00 (0.92)
n ?16 n ?18 n ?20
Loss 7.00 (3.01) 7.63 (2.31) 6.56 (2.43)
n ?20 n ?16 n ?18
ARJ
27,2
200
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
estimate of $100,000, a 75 per cent confdence point estimate and an interval estimate
of $75,000-$125,000. The 75 per cent confdence interval is chosen arbitrarily. We
chose this level to allow for a robust test of the theory, and however, results of our
study could be sensitive to the levels chosen for the precision variable. A critical
assumption that we use here is that participants’ beliefs are symmetric (uniform or
normal). To the extent that investors’ beliefs are skewed, the interval estimate and
the 75 per cent confdence level may not be equivalent. In addition, the experimental
design uses an investment level of $100,000. We did not provide participants with
information about materiality. To the extent that participants do not believe this
level is material, the results may be unreliable. Furthermore, we need to be careful in
generalizing our fndings because of our marginally signifcant results, which could
be due to our relatively small sample size. Future work may employ a larger sample
8.62
9.56
9
7
7.63
6.56
6
6.5
7
7.5
8
8.5
9
9.5
10
Point 75% confident Range
B
i
a
s
Gain
Loss
Notes: This figure presents the means for the
dependent variables: perceived management bias
and perceived reliability. The horizontal axis
presents the independent variable at three levels:
point stands for a precise point estimate; 75 per
cent confidence stands for a point estimate at 75
per cent confidence level; and range stands for a
confidence interval. The ratings are presented on
the vertical axis. They are based on the following
questions: For the purpose of making investment-
related decisions, how reliable do you find the
fair value estimate of $100,000? (1 being not
reliable at all, 1 being very reliable). When using
the discounted cash flow model, how likely
would Company A’s management intentionally
use assumptions that lead to the current valuation
of Company B? (1 being very unlikely, 11 being
very likely)
Figure 2.
Various judgments of fair
value: perceived
management bias
Table VI.
Perceived management
bias rating (test of H3):
two-way ANOVA with
bias rating as the
dependent variable (test of
H3)
Source SS df MS F p (2-sided)
Precision 18.00 2 9.00 1.81 0.17
Direction 112.35 1 112.35 22.55 0.00
Precision ?direction 3.08 2 1.54 0.31 0.74
Within cells (error) 503.13 102 4.98
201
Management
incentive and
information
precision
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
to further validate our results. Finally, this study uses MBA students rather than
real investors. MBA students may not have strong knowledge of management
incentives and may interpret fair value estimates differently. Future studies should
explore how real market participants react to differences in formats of reporting fair
market values.
Table VII.
Perceived management
bias rating (test of H3):
results of simple effects
(test of H3)
Condition Mean difference Standard deviation Signifcance
Gain
Point 75 per cent confdence ?1.08 0.78 0.51
Point Range ?0.38 0.75 1.00
75 per cent confdence Range 0.71 0.74 1.00
Loss
Point 75 per cent confdence ?0.69 0.75 1.00
Point Range 0.44 0.73 1.00
75 per cent confdence Range 1.13 0.77 0.44
Notes: H3 predicts the perceived degree of management bias is higher in a fair value gain
condition than in a fair value loss condition regardless of the precision of fair value information; to
test this hypothesis, we asked the following question on an 11-point scale: “When using the
discounted cash fow model, how likely would Company A’s management intentionally use
assumptions that lead to the current valuation of Company B?” (1 being very unlikely, 11 being
very likely); the dependent variable is the degree of management bias; the independent variable,
information precision, is manipulated at three levels: point stands for a precise point estimate, 75
per cent confdence stands for a point estimate at 75 per cent confdence level and range stands for
a confdence interval
Table VIII.
Additional analysis
Rating 50 per cent confdence Wider range Imprecise verbal
Reliability
Mean 4.93 4.07 4.20
Standard deviation 1.73 2.09 1.78
Bias
Mean 6.86 6.80 7.52
Standard deviation 2.51 2.76 1.76
N 14 15 25
Notes: The new experiment intends to understand how sensitive our results are to the confdence
level chosen (e.g., 50, 60 or 75 per cent) or size of the range provided; it solely focuses on the gain
condition and includes three different levels of imprecision; for the point estimate with confdence
level, we provided participants with a lowered confdence level of 50 per cent; for the range
estimate, we widened the interval to lower the confdence; specifcally, we provide participants
with a wider interval, which ranges as low as $50,000 and as high as $150,000; moreover, we
include another condition of imprecise statement by providing a statement that fair values are
imprecise without providing either a confdence level or range; specifcally, we provide
participants with the following statement: “Though the fair value is estimated at $100,000 for B,
A’s management indicates that this number is imprecise”
ARJ
27,2
202
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Note
1. In its glossary of terms, Concepts Statement 2 (SFAC 2) of US GAAP defnes reliability as the
quality of information that assures that information is reasonably free from error or bias and
faithfully represents what it purports to represent. With respect to measures, it states that
“*t]he reliability of a measure rests on the faithfulness with which it represents what it
purports to represent, coupled with an assurance for the user, which comes through
verifcation, that it has that representational quality” (paragraph 59). Thus, the principal
components of reliability are representational faithfulness and verifability (see The FASB’S
Conceptual Framework: Relevance and Reliability by Financial Accounting Standards
Advisory Council September 2004). In October 2010 the FASB issued a new concept
statement, SFAC 8, which replaces SFAC 1 and 2.
References
Ball, R., Robin, A. and Wu, J.S. (2003), “Incentives versus standards: properties of accounting
income in four East Asian countries, and implications for acceptance of IAS”, Journal of
Accounting and Economics, Vol. 36 Nos 1/3, pp. 235-270.
Baginski, S.P., Hassell, J.M. and Kimbrough, M.D. (2004), “Why do managers explain their
earnings forecasts?”, Journal of Accounting Research, Vol. 42 No. 1, pp. 1-29.
Blacconiere, W.G., Frederickson, J.R., Johnson, M.F. and Lewis, M.F. (2009), “Do voluntary
disclosures that disavow the reliability of mandated fair value information refect
legitimate concerns about reliability?”, Working paper, IN University, IN.
Budescu, D.V. and Du, N. (2007), “The coherence and consistency of investors’ probability
judgments”, Management Science, Vol. 53 No. 11, pp. 1731-1744.
Du, N. and Budescu, D.V. (2005), “The effects of imprecise probabilities and outcomes in
evaluating investment options”, Management Science, Vol. 51 No. 12, pp. 1791-1803.
Du, N., Budescu, D.V., Shelley, M. and Omer, T. (2011), “The appeal of vague fnancial forecasts”,
Organizational Behavior and Human Decision Processes, Vol. 114 No. 2, pp. 179-189.
Du, N., Lin, L. and McEnroe, J. (2010), “Is the truth the problem?”, The CPA Journal, Vol. 80 No. 1,
pp. 6, 8, 10-11.
Eagly, A.H. and Chaiken, S. (1975), “An attribution analysis of the effect of communicator
characteristics on opinion change: the case of communicator attractiveness”, Journal of
Personality and Social Psychology, Vol. 32, pp. 136-144.
Fiske, S.T. and Pavelchak, M.A. (1986), “Category-based versus piecemeal-based affective
responses: developments in schema-triggered affect”, in Sorrentino, R.M. and Higgins, E.T.
(Eds), Handbook of Motivation and Cognition: Foundations of Social Behavior, Guilford
Press New York, NY, pp. 167-203.
Frederickson, J., Hodge, F. and Pratt, J. (2006), “The evolution of stock option accounting:
disclosure, voluntary recognition, mandated recognition and management disavowals”,
The Accounting Review, Vol. 81 No. 5, pp. 1073-1093.
He, X.J., Wong, T.J. and Young, D.Q. (2012), “Challenges for implementation of fair value
accounting in emerging markets: evidence from China”, Contemporary Accounting
Research, Vol. 29 No. 2, pp. 538-562.
Hirst, D.E., Koonce, L. and Simko, P.J. (1995), “Investor reactions to fnancial analysts’ research
reports”, Journal of Accounting Research, Vol. 33 No. 3, pp. 335-351.
Hirst, D.E., Koonce, L. and Miller, J. (1999), “The joint effect of management’s prior forecast
accuracy and the form of its fnancial forecasts on investor judgment”, Journal of
Accounting Research, Vol. 37 No. 1, pp. 101-124.
203
Management
incentive and
information
precision
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Hodge, F., Hopkins, P. and Pratt, J. (2006), “Management reporting incentives and classifcation
credibility: the effects of reporting discretion and reputation”, Accounting, Organizations,
and Society, Vol. 31 No. 7, pp. 623-634.
International Accounting Standard Board (IASB) (2011), Presentation of Financial Statements,
International Accounting Standard (IAS) 1, London.
Juslin, P., Winman, A. and Olsson, H. (2000), “Naive empiricism and dogmatism in confdence
research: a critical examination of the hard-easy effect”, Psychology Review, Vol. 107 No. 2,
pp. 384-396.
Klayman, J., Soll, J., Gonzalez-Vallejo, C. and Barlas, S. (1999), “Overconfdence: it depends on
how, what, and whom you ask”, Organizational Behavior and Human Decision
Processes, Vol. 79 No. 3, pp. 216-247.
Maines, L. and Wahlen, J. (2006), “The nature of accounting information reliability: inferences
from archival and experimental research”, Accounting Horizons, Vol. 20 No. 4, pp. 399-425.
Mercer, M. (2004), “How do investors assess the credibility of management disclosures?”,
Accounting Horizons, Vol. 18 No. 3, pp. 185-196.
Rottenstreich, Y. and Tversky, A. (1997), “Unpacking, repacking, and anchoring: advances in
support theory”, Psychology Review, Vol. 104 No. 2, pp. 406-415.
Winman, A., Hansson, P. and Juslin, P. (2004), “Subjective probability intervals: how to reduce
overconfdence by interval evaluation”, Journal of Experimental Psychology, Learning,
Memory, Cognition, Vol. 30 No. 6, pp. 1167-1175.
Further reading
Budescu, D.V. and Wallsten, T.S. (1995), “Processing linguistic probabilities: general principles
and empirical evidence”, in Busemeyer, J., Medin, D.L. and Hastie, R. (Eds), Decision Making
from a Cognitive Perspective, Academic Press, San Diego, CA, pp. 275-318.
Du, N. (2009), “Do investors react differently to range and point management earnings forecasts?”,
The Journal of Behavioral Finance, Vol. 10 No. 4, p. 195.
Financial Accounting Standards Board (FASB) (1978), “Objectives of fnancial reporting by
business enterprises”, Statement of Financial Accounting Concepts No. 1, FASB,
Norwalk, CT.
Financial Accounting Standards Board (FASB) (2006), “Fair value measurements”, Statement of
Financial Accounting Standards No. 157, FASB, Norwalk, CT.
Financial Accounting Standards Board (FASB) (2007), “The fair value option for fnancial assets
and fnancial liabilities”, Statement of Financial Accounting Standards No. 159, FASB,
Norwalk, CT.
Jorion, P. (2002), “Howinformative are value-at-risk disclosures?”, The Accounting Review, Vol. 77
No. 4, pp. 911-931.
Kahneman, D. and Tversky, A. (1979), “Prospect theory: an analysis of decision under risk”,
Econometrica, Vol. 47 No. 2, pp. 263-292.
Kennedy, J., Mitchell, T. and Sefcik, S.E. (1998), “Disclosure of contingent environmental
liabilities: some unintended consequences?”, Journal of Accounting Research, Vol. 36,
pp. 257-277.
Libby, R., Nelson, M.W. and Hunton, J. (2006), “Recognition versus disclosure, auditor tolerance
for misstatement, and the reliability of stock-compensation and lease information”, Journal
of Accounting Research, Vol. 44 No. 3, pp. 533-560.
ARJ
27,2
204
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Libby, R., Tan, H.T. and Hunton, J.E. (2006), “Does the form of management’s earnings guidance
affect analysts’ earnings forecasts?”, The Accounting Review, Vol. 81 No. 1, pp. 207-225.
Wallsten, T.S. (1990), “Measuring vague uncertainties and understanding their use in decision
making”, in von Furstenberg, G.M. (Ed.), Acting Under Uncertainty, Kluwer, pp. 377-398.
Zwick, R. and Wallsten, T.S. (1989), “Combining stochastic and linguistic inexactness: theory and
experimental evaluation of four fuzzy probability models”, International Journal of
Man-Machine Studies, Vol. 30, pp. 69-111.
Appendix
Fair value gain (loss) in point estimate
Please assume you work in the investment department of a frm. Your supervisor has instructed
you to assess the fnancial performance of Company A, a 100-year old fnancial boutique frmthat,
through its wholly owned and partially owned subsidiaries, is engaged in fnancial services
activities. After extensive research, you prepared a short summary with some important facts
about Company A.
• Co. A acquired a minority interest in Co. B two years ago at a price of $60,000 ($140,000). B
is a non-public, technology-based asset management company that is in the process of
growing its business and has the ultimate goal of becoming one of the premier investment
advising companies in the market. However, B has not yet achieved a positive cash fow in
its month-to-month operations at December 31, 2009.
• At the end of 2009, the fair value of Bis estimated at $100,000, $40,000 higher (lower) than its
original purchase price of $60,000 ($140,000). Company Aadjusts its fnancial statements to
refect this increase (decrease). The investment in B is nowcarried at $100,000 as an asset in
the balance sheet, and the increase (decrease) of $40,000 is recognized as unrealized holding
gain (loss) in its comprehensive income statement.
• The fair value of $100,000 is estimated by A’s management. Because there is no publicly
observed market-based information about B (B is not traded in active markets), and there is
no company in the marketplace that is comparable to B, Co. A’s management used the
Discounted Cash Flow Analysis as the valuation model. The model uses projected fnancial
information for the next fve years as inputs and relies heavily on managers’ assumptions of
future cash fow and risk adjusted-discount rate.
Before you make recommendation to your clients about whether to invest in Company A, your
supervisor would like to know your judgments. Please proceed to the next page to answer the
questions.
(For range estimate, the following sentence is added: “Though the fair value is estimated at
$100,000 for B, A’s management indicates that this estimate may range as low as $75,000 and as
high as $125,000 depending on the assumptions used”.)
(For 75 per cent confdence point estimate, the following sentence is added: “Though the fair
value is estimated at $100,000 for B, A’s management indicates that they are only 75 per cent
confdent about this estimate because of the assumptions used”.)
Please answer the following three questions:
(1) When using the discounted cash fowmodel, howlikely would Company A’s management
intentionally use assumptions that lead to the current valuation of Company B?
(Very unlikely) 1 2 3 4 5 6 7 8 9 10 11 (Very likely)
(2) For the purpose of making investment-related decisions, how reliable do you fnd the fair
value estimate of $100,000?
(Not reliable at all) 1 2 3 4 5 6 7 8 9 10 11 (Very reliable)
205
Management
incentive and
information
precision
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
(3) How risky is company a’s the investment in company B?
(Not at all risky) 1 2 3 4 5 6 7 8 9 10 11 (Very risky)
Additional questions:
(1) Please indicate your gender (check one).
Male _______ Female _______
(2) How much working experience do you have?
____________ years __________ months
(3) How much investment experience do you have?
____________ years __________ months
(4) Compared to the historical cost, the fair value estimate of company B is:
higher ____ lower ____ (check one)
(5) Which estimate best describes the fair value provided? (check one)
$100,000 _____ $100,000 at 75 per cent confdent _____ $75,000-$125,000 _____
Corresponding author
Ning Du can be contacted at: [email protected]
To purchase reprints of this article please e-mail: [email protected]
Or visit our web site for further details: www.emeraldinsight.com/reprints
ARJ
27,2
206
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
doc_551487061.pdf
The purpose of this paper was to examine whether a less precise (or imprecise) estimate
may increase investors’ confidence and improve investors’ perceptions of fair value reliability. The
main criticism of fair value accounting has been its lack of reliability perceived by investors.
Accounting Research Journal
The joint effects of management incentive and information precision on perceived
reliability in fair value estimates
Ning Du J ohn E. McEnroe Kevin Stevens
Article information:
To cite this document:
Ning Du J ohn E. McEnroe Kevin Stevens , (2014),"The joint effects of management incentive and
information precision on perceived reliability in fair value estimates", Accounting Research J ournal, Vol. 27
Iss 2 pp. 188 - 206
Permanent link to this document:
http://dx.doi.org/10.1108/ARJ -10-2012-0081
Downloaded on: 24 January 2016, At: 21:20 (PT)
References: this document contains references to 33 other documents.
To copy this document: [email protected]
The fulltext of this document has been downloaded 1418 times since 2014*
Users who downloaded this article also downloaded:
Prodosh Simlai, (2014),"Firm characteristics, distress risk and average stock returns", Accounting Research
J ournal, Vol. 27 Iss 2 pp. 101-123 http://dx.doi.org/10.1108/ARJ -06-2012-0046
Etumudon Ndidi Asien, (2014),"Impact of firm-specific characteristics on managers’ identity disclosure",
Accounting Research J ournal, Vol. 27 Iss 2 pp. 150-168 http://dx.doi.org/10.1108/ARJ -03-2013-0010
Shu Hui Wee, Soon Yau Foong, Michael S.C. Tse, (2014),"Management control systems and organisational
learning: the effects of design and use", Accounting Research J ournal, Vol. 27 Iss 2 pp. 169-187 http://
dx.doi.org/10.1108/ARJ -05-2013-0026
Access to this document was granted through an Emerald subscription provided by All users group
For Authors
If you would like to write for this, or any other Emerald publication, then please use our Emerald for
Authors service information about how to choose which publication to write for and submission guidelines
are available for all. Please visit www.emeraldinsight.com/authors for more information.
About Emerald www.emeraldinsight.com
Emerald is a global publisher linking research and practice to the benefit of society. The company
manages a portfolio of more than 290 journals and over 2,350 books and book series volumes, as well as
providing an extensive range of online products and additional customer resources and services.
Emerald is both COUNTER 4 and TRANSFER compliant. The organization is a partner of the Committee
on Publication Ethics (COPE) and also works with Portico and the LOCKSS initiative for digital archive
preservation.
*Related content and download information correct at time of download.
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
The joint effects of management
incentive and information
precision on perceived reliability
in fair value estimates
Ning Du, John E. McEnroe and Kevin Stevens
School of Accountancy and Management Information Systems (MIS),
DePaul University, Chicago, Illinois, USA
Abstract
Purpose – The purpose of this paper was to examine whether a less precise (or imprecise) estimate
may increase investors’ confdence and improve investors’ perceptions of fair value reliability. The
main criticism of fair value accounting has been its lack of reliability perceived by investors.
Design/methodology/approach – A 2 ? 3 randomized experiment was used where management
incentive and information precision are manipulated.
Findings – The results from this study indicate that perceived reliability is jointly affected by
management’s incentives andinformationprecision. Reliabilityratingis the highest for fair value stated
as a point estimate with a specifed confdence level attached to it. Further analysis indicates that higher
perceived reliability is related to its representational faithfulness because participants perceive that a
point estimate with a specifed confdence level better matches uncertainty in measuring future cash
fows.
Originality/value – This is the frst study to examine whether a less precise (or imprecise)
estimate may increase investors’ confdence and improve investors’ perceptions of fair value
reliability. Because of the subjectivity and uncertainty in fair value estimates, less precise fair
value estimates may not be viewed as less reliable. In fact, using a precise format to represent fair
value estimates may not be appropriate (neither reliable nor credible), because a precise point
estimate fails to capture its underlying uncertainty in future cash fows. Aless precise format could
represent a credible choice for fair value because it refects uncertainty and subjectivity and
effectively communicates management’s assessments of variability in future cash fows.
Keywords Uncertainty, Reliability, Fair value, Management bias
Paper type Research paper
Introduction
The past few years have witnessed the global convergence of national accounting
standards and International Accounting Standards [IAS, superseded by International
Financial Reporting Standards (IFRS)]. Since 2005, when European Union countries frst
mandated its use, more than 100 countries have adopted IFRS. Over the past few years,
the US Securities Exchange Commission has taken several steps to acknowledge IFRSs
as a high-quality set of fnancial reporting standards. These steps include elimination of
the reconciliation to USA Generally Accepted Accounting Principles (USA GAAP) by
foreign private issuers using IFRSs, and the issuance of a proposed “roadmap” for the
use of IFRS by US domestic registrants. Although the movement toward a worldwide
single-set of accounting standards has slowed, convergence projects between the
The current issue and full text archive of this journal is available at
www.emeraldinsight.com/1030-9616.htm
ARJ
27,2
188
Accounting Research Journal
Vol. 27 No. 2, 2014
pp. 188-206
©Emerald Group Publishing Limited
1030-9616
DOI 10.1108/ARJ-10-2012-0081
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Financial Accounting Standards Board (FASB) and the International Accounting
Standards Board (IASB, 2011) in the areas of leases, revenue recognition and fnancial
instruments continue.
Recent developments in IFRS tend to move the standards toward the use of fair value
accounting even for non-fnancial assets. If the USA adopts IFRS, then a greater use of
fair value accounting necessarily follows. Because of the uncertainty inherent in fair
value accounting (as compared to historical cost) and the lack of implementation
guidance, fair value accounting remains vulnerable to managerial opportunismthrough
accounting and/or transaction decisions (Ball et al., 2003). Moreover, fair value estimates
are uncertain because predictions must be made of future market value. The reliability
concern is particularly serious in cases where the observable market prices are not
available, and the company must rely on the mark-to-model approach to derive fair
value estimates. Because human judgments are prone to errors and biases, it is not
surprising that fnancial statement users are skeptical of the objectivity and reliability of
fair value information. The lack of reliability may limit the relevance of fair value
estimates, reduce the comparability of fnancial reports worldwide and ultimately
impair decision usefulness.
Despite the on-going debate about its relevance and reliability, fair value accounting
has gained increasing importance. On May 12, 2011, IASB issued IFRS 13, Fair Value
Measurement, and the FASB issued Accounting Standard Update No. 2011-04, Fair
Value Measurement (Topic 820): Amendments to Achieve Common Fair Value
Measurement and Disclosure Requirements in USA GAAP and IFRSs (ASC 820)
(together, hereafter referred to as “the new guidance”). The new guidance results in a
consistent defnition of fair value and requires common measurements of fair value
between IFRS and USAGAAP. In addition to increasing global harmonization, the new
guidance also signifcantly enhances the disclosure requirements for fair value
accounting.
ASC 820 and IFRS 13 are consistent in their frameworks used for fair value
measurement. Both establish a three-level hierarchy for measuring fair value. Level 1
and 2 inputs are mark-to-market inputs and use quoted prices in active markets for
identical or similar assets or liabilities. Level 3 inputs are mark-to-model and require the
frm to use a valuation model to estimate the market value of the assets or liabilities.
Level 3 estimates require considerable management judgment in the selection and
application of valuation techniques, and therefore, require effective internal controls and
knowledgeable independent auditors and other corporate governance mechanisms to
thwart potential abuses. Because of its potential for judgmental bias and errors, whether
fair values derived from the models refect economic reality and indeed represent the
“true” values of an entity become questionable in many investors’ minds.
Given this background, the goal of this study is to understand how to improve
investors’ perception of fair value reliability. To achieve this goal, we must frst
differentiate between real reliability and perceived reliability. Real reliability refers to
the uncertainty or variability inherent in information content, commonly defned as the
closeness between the benchmark (i.e. what it purports to represent) and accounting
representation[1]; in contrast, perceived reliability refers to user’s perceptions or
understanding of disclosed information. Given the high degree of uncertainty and
subjectivity inherent in fair value measurements, accounting standard setters and
preparers may not be able to signifcantly improve the real reliability of information
189
Management
incentive and
information
precision
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
content. However, we argue that a careful choice of presentation format may help
improve the information’s perceived reliability. Indeed, prior research has found that
characteristics of management disclosure, such as precision, venue or time horizon,
affect the disclosure reliability perceived by investors (Koonce et al., 1998; Mercer, 2004).
Disclosure precision, an important dimension of management disclosure, has been
found to affect investors’ judgments and perceptions. Specifcally, imprecise disclosures
are viewed as less reliable/credible than more precise disclosures because imprecision
signals management uncertainty (Baginski et al., 2004; Hirst et al., 1999).
But, because of the subjectivity and uncertainty in fair value estimates, less precise
fair value estimates may not be viewed as less reliable. In fact, using a precise format to
represent fair value estimates may not be appropriate (i.e. may be neither reliable nor
credible), because a precise point estimate fails to capture the underlying uncertainties
in future cash fows. Aless precise format could represent a credible choice for fair value
because it refects uncertainty and subjectivity and effectively communicates
management’s assessments of variability in future cash fows (Budescu and Du, 2007;
Du and Budescu, 2005; Du et al., 2011, 2010). Thus, we examine whether a less precise (or
imprecise) estimate may increase investors’ confdence and improve investors’
perceptions of fair value reliability. In addition, we investigate whether different types
of imprecise estimates may have differential effects on investors’ reliability judgments.
Last, we examine whether management incentives may moderate or accentuate these
effects. To study these questions we use a 2 ? 3 between-subjects randomized
experiment. We manipulate incentive at two levels (gain or loss) and precision at three
levels (a precise point estimate, a point estimate at 75 per cent confdence level or a range
estimate). MBAstudents participate in the experiment and judge the degree of reliability
and bias in the fair value estimate provided by the management. Results from the
experiment showthat participants viewfair value gain to be more biased than fair value
loss regardless of its precision. In addition, we fnd that when the fair value gain is stated
as a point estimate at 75 per cent confdence level, participants believe it is more reliable
than the precise point estimate. The results from this study are timely and provide a
priori evidence to the IASB and FASB in evaluating the effectiveness of fair value
disclosures.
Literature review and hypothesis development
Imprecise information. Improving perceived reliability of fair value estimates has
inspired much research interest in the past. For example, Frederickson et al. (2006)
and Blacconiere et al. (2009) investigated perceived fair value reliability in the
context of employee stock option compensation (SOC) mandated under Statement of
Financial Accounting Standard No. 123R. In particular, they examined disavowals,
a unique form of management voluntary disclosures, which explicitly question the
reliability or usefulness of the mandated information in audited fnancial
statements. The results from their studies suggest that by disclosing the lack of
reliability in fair value estimates in SOC, management disavowals help investors
properly weigh fair value estimates in SOC, and thus, improve investors’
perceptions of the reliability of those estimates.
The reliability concern is particularly germane for mark-to-model assets or liabilities.
For example, the discounted cash fow model is extremely sensitive because even a
small change in inputs may result in large valuation differences. In particular, managers
ARJ
27,2
190
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
must be able to objectively assess future risks and uncertainties and use a discount rate
that includes appropriate risk adjustments. Further, managers must be able to estimate
both the amount and timing of future cash fows. Maines and Wahlen (2006) argue that
reliable accounting information must be able to signal differences in the magnitude as
well as the uncertainty inherent in future cash fows. Although future cash fows may be
highly uncertain and diffcult to measure reliably in the current period, revealing the
degree of uncertainty associated with the values of assets and liabilities forecasted by
cash fow models may provide representationally faithful information about the
uncertainty of cash fow modeling.
Prior experimental evidence suggests that investors prefer the precision of
information to match its underlying uncertainty (Budescu and Du, 2007; Du and
Budescu, 2005; Du et al., 2011, 2010). A precise point estimate refects certainty and
masks the risks and uncertainties inherent in fair value measurement and accordingly
fails the faithful representation criterion. Unlike a precise point estimate, imprecise fair
value estimates are probabilistic and refect variability in future cash fows. The choice
of imprecise fair value estimates is similar to management disavowal and signals
greater uncertainty about the accuracy of unobserved inputs, and by extension, the
accuracy of the fair value estimate (Baginski et al., 1993). Therefore, we argue that by
varying degree of precision in fair value estimates, managers reveal howconfdent they
feel about future cash fow; this signal helps investors properly weigh the reliability of
fair value information. Knowing the limitations of fair value information increases
investors’ confdence in using the information and ultimately improves investors’
perceptions of fair value reliability which improves its decision usefulness. This
assumption leads us to H1:
H1. Investors perceive higher reliability in imprecise fair value estimates than
precise fair value estimates.
Different formats of imprecise information
Imprecise estimates communicate probabilistic information. An estimate may come as a
point estimate with a specifed probability to indicate a confdence level (e.g. managers
are 50, 75 or 90 per cent confdent that the fair value is $100,000) or as an interval
estimate spanning minimum and maximum values (e.g. fair value ranges from $75,000
to $125,000). Winman et al. (2004) suggest that probability estimates and confdence
intervals (ranges) are formally equivalent because high (low) uncertainties can be
expressed by low (high) probability judgments or by wide (narrow) interval (range)
estimates assuming that people’s beliefs are symmetrical. For example, the 75 per cent
confdence interval of $100,000 is equivalent to the interval ranging from $75,000 to
$125,000 if we assume the standard deviation as $21,739 with a normal distribution.
Empirically, however, these two different forms of estimates have produced
systematically different judgments (Budescu and Du, 2007; Rottenstreich and Tversky,
1997). Prior evidence shows that the interval estimate induces higher error and bias than
the point probability estimate (Klayman et al., 1999; Juslin et al., 2000), perhaps because
individuals have diffculty in inferring the probability or confdence level from an
interval estimate. Unlike an interval estimate where users have to infer managers’
confdence in management’s judgment, a point estimate with a specifed confdence level
191
Management
incentive and
information
precision
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
(e.g. 75 per cent) may lead to an increase in perceived reliability. Given this, we
developed our second hypothesis as follows:
H2. The reliability rating will be the highest when the fair value estimate is
presented as a point estimate with a specifed confdence level.
Management incentives
The IFRS framework includes both relevance and reliability as underlying assumptions
and refers to possible trade-offs between these objectives in paragraph 32 of IAS 1:
“Information may be relevant but so unreliable in nature or representation that its
recognition may be potentially misleading”. Furthermore, IAS 1also states that
information in the fnancial statements is reliable if it is free from material errors and
bias (see paragraph 31). Because fair value estimates, especially those with Level 3
inputs, are based on judgmental calculations of hypothetical values and markets, it can
become diffcult to determine whether those inputs are free of errors and preparer bias.
For example, He et al. (2012) fnd that Chinese managers “cherry pick” their investment
portfolio, and are more likely to sell investment securities for a gain to avoid reporting
fnancial losses after frst IFRS-based China Accounting Standards were adopted in
2007. Prior literature suggests that the degree of bias users perceive in information
reported by management depends on the consistency between the reported information
and the users’ beliefs about management’s reporting incentives (Frederickson et al.,
2006; Hirst et al., 1995; Hodge et al., 2006). Attribution and persuasion theory (Eagly and
Chaiken, 1975; Fiske and Pavelchak, 1986) suggests that individuals consider
incentive-consistent signals to be strategic and biased, and thus less credible (Hirst et al.,
1995; Hodge et al., 2006). Because fair value estimates result in gains or losses that
eventually fow through owner’s equity, managers may strategically use fair value
accounting to produce desired performance outcomes for personal gains. As such, we
believe managers have strong incentive to bias their fair value estimates to a gain
condition; further, this incentive is strong enough to be insensitive to information
precision. In other words, managers, under the infuence of their economic incentives,
may bias a precise point estimate as well as the range or confdence level in the fair value
estimate. The potential for management error and partiality suggests that investors will
be skeptical of fair value approximations resulting in potential gains but not those in
losses. Thus, we expect a main effect of incentive on perceived bias judgment which
leads to our third hypothesis:
H3. Investors perceive higher management bias in fair value estimates leading to
potential gains than potential losses.
Joint effects of management incentives and information precision
When both incentives and precision are present, we expect the two factors may jointly
affect investors’ reliability judgments. Specifcally, the precision of fair value estimate
may affect investors’ reliability judgments differently depending on whether fair value
estimates are incentive consistent (gain) or inconsistent (loss). As a fair value gain is
consistent with meeting managerial incentives, we expect investors to be skeptical of
fair value estimates when they result in gains; in addition, we assume that investors will
actively look for cues, such as information precision, to improve decision usefulness.
Because a point estimate with a specifed confdence level (e.g. 75 per cent) is expressed
ARJ
27,2
192
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
as a precise probability value and clearly communicates the degree of uncertainty in the
future cash fow, we anticipate that investors will prefer this format over a range
estimate. Specifcally, we argue that the degree of improvement in perceived reliability
is the greatest when incentive consistent (gain) fair value is presented imprecisely as a
point estimate quantifed by a specifed confdence level. Thus, our last hypothesis tests
the joint effects of management incentive and information precision on perceived fair
value reliability and is stated as follows:
H4. Apoint fair value estimate with a specifed confdence level is perceived to be
more reliable than a range fair value estimate or a precise point estimate
when this fair value estimate results in potential gains, but the difference in
perceived reliability may disappear when the fair value estimate results in
potential losses.
Methodology
Participants
The participants are 114 MBA students from a large private university. Sixty-three are
males, and 51 are females. Students volunteered to participate in response to in-class
announcements and were randomly assigned to six different experimental conditions.
Demographic information collected at the end of the experiment indicates that most
participants are likely to have the necessary knowledge and experience to complete the
task. The majority of participants have both work and investment experience. On
average, participants have 6.5 years of working experience and 4.5 years of investment
experience.
Procedure and task
In the experiment, we asked the participants to assume that they work in the investment
department of a frm. Their task is to judge the degree of reliability and management
bias in the fair value estimate provided by the management. We provided participants
with a short summary which states that two years ago, Co. A acquired a minority
interest in Co. B, as well as detailed information about the transaction. After reviewing
the provided information, participants answered a series of questions about Company A
and its investment in Company B. Finally, the participants fnished by answering a
short series of manipulation check questions and are fully debriefed. The entire
experiment took 20 minutes, on average, to complete. (See Appendix for a copy of the
experimental instrument.)
Independent variables
We use a 2 (Incentive) ?3 (Precision) between-subjects design. The incentive variable is
manipulated at two levels (gain or loss) and the precision variable at three levels (a
precise point estimate, a point estimate at 75 per cent confdence level and a range
estimate). (The point estimate at 75 per cent confdence level is referred to as the 75 per
cent point hereafter.) Participants are randomly assigned to experimental conditions
and read the following instruction:
Please assume that you work in the investment department of a frm. Your supervisor has
instructed you to assess the fnancial performance of Company A, a 100-year old fnancial
193
Management
incentive and
information
precision
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
boutique frm that through its wholly owned and partially owned subsidiaries is engaged in
fnancial services activities. After extensive research, you prepared a short summary with
some important facts about Company A:
Co. A acquired a minority interest in Co. B two years ago at a price of $60,000 ($140,000). B is
a non-public, technology-based asset management company that is in the process of growing
its business and has the ultimate goal of becoming one of the premier investment advising
companies in the market. However, Company Bhas not yet achieved a positive cash fowin its
month-to-month operations at December 31, 2009.
At the end of 2009, the fair value of B is estimated at $100,000, $40,000 higher (lower) than its
original purchase price of $60,000 ($140,000). Company A adjusts its fnancial statements to
refect this increase (decrease). The investment in Company B is nowcarried at $100,000 as an
asset in the balance sheet, and the increase (decrease) of $40,000 is recognized as an unrealized
holding gain (loss) in its comprehensive income statement.
The fair value of $100,000 is estimated by Company A’s management. Because there is no
publicly observed market-based information about Company B (Company B is not traded in
active market, and there is no company in the marketplace that is comparable to Company B),
Company A’s management used a discounted cash fow analysis as the valuation model. The
model uses projected fnancial information for the next fve years as the inputs and relies
heavily on managers’ assumptions of future cash fow and choice of risk adjusted-discount
rate.
We manipulate management incentive as the direction of comprehensive income change
by holding the fair value constant at $100,000. We vary the historical cost at $60,000 or
$140,000, respectively, to create fair value gain or loss for this investment. In the fair
value gain (loss) condition, participants are told that the interest in Co. B was acquired
two years ago at a price of $60,000 ($140,000) and the 2009 change in other
comprehensive income was mainly attributed to the adjustment of a $40,000 increase
(decrease) in fair value.
We manipulate the precision of fair value at three levels. In the point estimate
condition, the fair value is given as a point estimate of $100,000. In the 75 per cent point
confdence condition, we add the sentence: “Though the fair value is estimated at
$100,000 for B, A’s management indicates that they are only 75 per cent confdent about
this estimate because of the assumptions used”. In the interval estimate condition, we
add the sentence:
Though the fair value is estimated at $100,000 for B, A’s management indicates that this
estimate may range as low as $75,000 and as high as $125,000, depending on the assumptions
used.
These three conditions have the same expected value of $100,000. That is, the midpoint
of the interval estimate of $75,000-$125,000 is $100,000, and the 75 per cent confdence
level of $100,000 is also $75,000-$125,000.
Finally, all participants are told that:
Before you make recommendation to your clients about whether to invest in Company A, your
supervisor would like to knowthe basis for your judgments. Please proceed to the next page to
answer the following questions.
ARJ
27,2
194
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Dependent variables
Dependent variables include perceived reliability and the degree of management bias.
They are based on the following questions:
• For the purpose of making investment-related decisions, how reliable do you fnd
the fair value estimate of $100,000? (1 being not reliable at all, 11 being very
reliable).
• When using the discounted cash fow model, how likely would Company A’s
management intentionally use assumptions that lead to the current valuation of
Company B? (1 being very unlikely, 11 being very likely).
Results
Manipulation checks
To check the incentive (fair value gain or loss) manipulation, we ask participants
whether the fair value estimate of Company B is higher or lower than the historical cost.
All participants checked the correct answers. For the precision manipulation, we ask
two questions. First, we ask participants to check the fair value estimate that is included
in their cases ($100,000, $100,000 at 75 per cent confdence level and $75,000 to $125,000).
Only one participant checked the wrong answer (excluding this participant does not
change the results).
The riskiness of an investment is an important factor in investment decisions. The
degree of risk perceived by participants may affect their reliability judgments. We asked
the participants to assess how risky Company A’s investment in Company B is on an
11-point scale (1 being not risky, 11 being very risky). We conducted an analysis of
variance (ANOVA) on perceived risk and found that participants perceived fair value
loss (mean ?7.82) as signifcantly riskier than fair value gain (mean ?7.07) [F(1, 104) ?
4.28, (p ? 0.05)]. However, the precision factor and its interaction with the incentive
factor are not signifcant.
Multivariate analysis of covariance
As the two dependent variables, perceived reliability and bias ratings, are potentially
correlated, we rely on multivariate analysis of variance (MANOVA) for the initial
analysis. To control for any potentially extraneous infuence on dependent variables, we
include gender, investment experience, work experience and the ratings on perceived
risk as covariates in our analysis. Multivariate analysis of covariance indicates none of
the four control variables is signifcant. Specifcally, for risk, Wilks’ lambda is 0.97 (F ?
1.09, p ? 0.05); for work experience, Wilks’ lambda is 0.99 (F ? 0.04, p ? 0.05); for
investment experience, Wilks’ lambda is 0.99 (F?0.47, p ?0.05); and for gender, Wilks’
lambda is 0.96 (F ? 1.72, p ? 0.05). In addition, MANOVA indicates that the
independent factor of precision is insignifcant (Wilks’ lambda ? 0.96, F ? 0.87, p ?
0.05), but the independent factor of bias (Wilks’ lambda ? 19.13, p ? 0.05) and their
interaction (Wilks’ lambda ? 0,90, F ? 2.45, p ? 0.05) are signifcant. To explain the
signifcant results, we conduct ANOVA in the following section for each of the
dependent variables.
Analysis of variance
Tables I-IV report the means for reliability ratings which are also plotted in Figure 1.
The descriptive statistics showthat participants do not perceive the fair value estimates
195
Management
incentive and
information
precision
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
as very reliable, as most of the ratings are quite low. Means are 5.22 for precise point,
5.56 for 75 per cent point and 5.10 for range. In general, the fair value loss (mean ?6.08)
is perceived signifcantly more reliable than the fair value gain (mean ? 4.50)
[F(1, 103 ?18.70, p ?0.05]. We suspect that the lower reliability in the gain condition
might have been driven by higher perceived management bias. H1 predicts that
investors perceive precise information to be less reliable than imprecise information. To
test H1, we treat the point estimate as the category of precise estimate. Then we consider
the range estimate and the point estimate at 75 per cent confdence level as one category
of imprecise estimate. We then compare the reliability ratings between these two
groups. As presented in Table 1, the mean rating for imprecise estimates (5.42) is
slightly higher than the mean rating for precise point estimate (5.22), but the difference
is not statistically different. Thus, our frst hypothesis is not supported.
H2 predicts that the reliability rating will be the highest when the fair value estimate
is presented as a point estimate at 75 per cent confdence level. Table II shows that the 75
per cent confdence estimate has the highest reliability rating (mean ?5.73), followed by
precise point estimate (mean ? 5.22) and range estimate (mean ? 5.10). However, the
ANCOVA analysis in Table III suggests the effect of precision is not statistically
signifcant [F (1, 103) ?0.39, p ?0.05]. Thus, our data fail to support H2.
H3 predicts that the perceived degree of management bias will be higher in the fair
value gain condition than in the fair value loss condition regardless of the precision of
fair value information. Table V presents the descriptive statistics on management bias,
which are also plotted in Figure 2. The ANOVA indicates the fair value is perceived to
Table I.
Perceived reliability of fair
value estimates rating
(tests of H1, H2 and H4):
mean perceived reliability
rating (test of H1)
Precise Imprecise
Point 75 per cent confdence Range
5.22 5.73 5.10
(0.36) (0.22) (1.99)
n ?36 n ?38 n ?38
5.42 (average)
Notes: The perceived reliability rating is for the following question on an 11-point scale: For the
purpose of making investment-related decisions, how reliable do you fnd the fair value estimate of
$100,000? (1 being not reliable at all, 1 being very reliable). The dependent variable is perceived
reliability; the independent variable is information precision and is manipulated at three levels: point
stands for a precise point estimate; 75 per cent confdence stands for a point estimate at 75 per cent
confdence level; and range stands for a confdence interval; H1 predicts that investors perceive higher
reliability in imprecise fair value estimates than precise fair value estimates; to test H1, we treat the
point estimate as the category of precise estimate (average mean of 5.22); then we consider the range
estimate and the point estimate at 75 per cent confdence level as one category of imprecise estimate
(average mean of 5.42); H2 predicts that investors perceive higher reliability when fair value estimate is
presented as a point estimate with a specifed confdence level (average mean of 5.73) than an interval
range estimate (average mean of 5.10) and a precise point estimate (average mean of 5.22); to test H2, we
compare the average means for the three levels of precision; H4 predicts that a point fair value estimate
with a specifed confdence level is perceived to be more reliable than a range fair value estimate or a
precise point estimate when this fair value estimate results in potential gains; fair value precision has no
effect on perceived reliability when fair value estimate results in potential losses; to test H4, we conduct
simple effect tests to evaluate pairwise differences among the adjusted means using the Holm’s
sequential Bonferroni to control for Type I error across the three pairwise comparisons
ARJ
27,2
196
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
be more biased in the gain condition (mean ?9.06) than in the loss condition (mean ?
7.06). The degree of bias does not seem to vary greatly for point (mean ? 7.81),
75 per cent point (mean ?8.59) and range (mean ?7.78). The ANOVAresults (Table VI)
indicate a signifcant main effect between gain and loss [F(1, 101) ? 22.50, p ? 0.01].
Neither the precision nor its interaction with the incentive factor is signifcant. This
suggests that participants believe that fair value with gain is signifcantly more biased
than that with loss. These results support H3 (Table VII).
H4 predicts that investors perceive the point estimate with a specifed confdence
interval (e.g. at 75 per cent) to be the most reliable estimate when fair value results in
potential gain. The perceived reliability does not vary when fair value estimates result in
potential loss. Figure 1 shows that in the gain condition, the fair value at 75 per cent
point estimate has the highest rating with a mean of 5.56, very close to the mean rating
of 5.90 for the 75 per cent point estimate in the loss condition. Table II presents the
results for ANOVA. We observe a signifcant interaction between the two independent
variables of precision and incentive [F(2, 103) ?5.76, p ?0.05], as well as a main effect
for direction [F(1, 103) ? 18.70, p ? 0.01]. To interpret the main effect of direction, we
conduct simple effect tests to evaluate pairwise differences among the adjusted means.
The Holm’s sequential Bonferroni procedure was used to control for Type I error across
the three pairwise comparisons. Results of simple effects are presented in Table IV.
Apparently, there is a signifcant difference in the adjusted means between precise point
(mean ?3.63) and 75 per cent point (mean ?5.56) in the gain condition (p ?0.05), but no
signifcant difference in the other comparisons. Though the rating in the range condition
Table II.
Perceived reliability of fair
value estimates rating
(tests of H1, H2 and H4):
mean perceived reliability
rating by gain or loss (test
of H2 and H4)
Condition Point 75 per cent confdence Range
Gain 3.63 (2.19) 5.56 (2.48) 4.30 (1.95)
n ?16 n ?18 n ?20
Loss 6.80 (1.51) 5.90 (1.69) 5.89 (1.71)
n ?20 n ?20 n ?18
Average 5.22 5.73 5.10
Notes: The perceived reliability rating is for the following question on an 11-point scale: For the
purpose of making investment-related decisions, how reliable do you fnd the fair value estimate of
$100,000? (1 being not reliable at all, 1 being very reliable) The dependent variable is perceived
reliability; the independent variable is information precision and is manipulated at three levels: point
stands for a precise point estimate; 75 per cent confdence stands for a point estimate at 75 per cent
confdence level; and range stands for a confdence interval; H1 predicts that investors perceive higher
reliability in imprecise fair value estimates than precise fair value estimates; to test H1, we treat the
point estimate as the category of precise estimate (average mean of 5.22); then we consider the range
estimate and the point estimate at 75 per cent confdence level as one category of imprecise estimate
(average mean of 5.42); H2 predicts that investors perceive higher reliability when fair value estimate is
presented as a point estimate with a specifed confdence level (average mean of 5.73) than an interval
range estimate (average mean of 5.10) and a precise point estimate (average mean of 5.22); to test H2, we
compare the average means for the three levels of precision; H4 predicts that a point fair value estimate
with a specifed confdence level is perceived to be more reliable than a range fair value estimate or a
precise point estimate when this fair value estimate results in potential gains; fair value precision has no
effect on perceived reliability when fair value estimate results in potential losses; to test H4, we conduct
simple effect tests to evaluate pairwise differences among the adjusted means using the Holm’s
sequential Bonferroni to control for Type I error across the three pairwise comparisons
197
Management
incentive and
information
precision
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
is lower than the rating in the 75 per cent point condition, the difference fails to reach
statistical signifcance. This evidence partially supports H4.
Additional analysis
Because of our focus on the effects of information precision, it is important to understand
how sensitive our results are to the confdence level chosen (e.g. 50, 60 or 75 per cent) or
size of the range provided. To this end, we conducted another experiment where we
varied the level of confdence and size of the range. We recruited 45 MBA students and
randomly assigned themto three conditions. We provided themwith the same case as in
the prior experiment but with slight variations. Because our results suggest that
information precision affects reliability and bias judgments when fair value estimates
produce potential gains but not losses, the new experiment solely focuses on the gain
condition. In addition, the newexperiment includes three different levels of imprecision.
For the point estimate with confdence level, we provided participants with a lowered
confdence level of 50 per cent. For the range estimate, we widened the interval to lower
the confdence; specifcally, we provide participants with a wider interval, which ranges
as low as $50,000 and as high as $150,000. Moreover, we include another condition of
imprecise statement by providing a statement that fair values are imprecise without
providing either a confdence level or range; specifcally, we provide participants with
the following statement: “Though the fair value is estimated at $100,000 for B, A’s
management indicates that this number is imprecise”.
Table III.
Perceived reliability of fair
value estimates rating
(tests of H1, H2 and H4):
two-way ANOVA with
perceived reliability rating
as the dependent variable
(tests of H1, H2 and H4)
Source SS df MS F p (2-sided)
Precision 2.93 2 1.47 0.39 0.68
Direction 70.47 1 70.47 18.70 0.00
Precision ?direction 43.43 2 21.72 5.76 0.00
Within cells (error) 508.218 104 3.77
Notes: The perceived reliability rating is for the following question on an 11-point scale: For the
purpose of making investment-related decisions, how reliable do you fnd the fair value estimate of
$100,000? (1 being not reliable at all, 1 being very reliable) The dependent variable is perceived
reliability; the independent variable is information precision and is manipulated at three levels:
point stands for a precise point estimate; 75 per cent confdence stands for a point estimate at 75 per
cent confdence level; and range stands for a confdence interval; H1 predicts that investors
perceive higher reliability in imprecise fair value estimates than precise fair value estimates; to test
H1, we treat the point estimate as the category of precise estimate (average mean of 5.22); then we
consider the range estimate and the point estimate at 75 per cent confdence level as one category
of imprecise estimate (average mean of 5.42); H2 predicts that investors perceive higher reliability
when fair value estimate is presented as a point estimate with a specifed confdence level (average
mean of 5.73) than an interval range estimate (average mean of 5.10) and a precise point estimate
(averagemeanof 5.22); totest H2, wecomparetheaveragemeansfor thethreelevelsof precision; H4predicts
that a point fair value estimate with a specifed confdence level is perceived to be more reliable than a range
fair value estimate or a precise point estimate when this fair value estimate results in potential gains; fair
value precisionhas no effect onperceivedreliabilitywhenfair value estimate results in potential losses; to
test H4, we conduct simple effect tests to evaluate pairwise differences among the adjusted means
using the Holm’s sequential Bonferroni to control for Type I error across the three pairwise
comparisons
ARJ
27,2
198
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
The descriptive statistics are reported in Table VIII. The reliability rating for 50 per cent
confdence level is 4.93, 4.07 for wider range and 4.20 for imprecise verbal statement.
Apparently, participants perceive the point estimate with a specifed 50 per cent
confdence level to be the most reliable, consistent with fndings in the prior experiment.
However, the one-way ANOVA indicates that the difference in means across the three
conditions is not statistically signifcant [F(51,2) ? 0.94, p ? 0.40]. The average
reliability rating is 4.40, signifcantly lower than the reliability rating in the prior
experiment. This suggests that as managers lower their confdence, participants
become more skeptical about the reliability of fair value estimates, and become less
sensitive of information precision. Furthermore, the bias rating for the 50 per cent
confdence level is 6.86, 6.80 for wider range and 7.52 when we introduce the imprecise
verbal statement, with an average rating of 7.06 [F(2,51) ?0.63, p ?0.54]. The average
rating is very similar to the rating in the prior experiment. This suggests the perceived
management bias does not vary as the level of information precision changes.
Conclusions
Fair value accounting is often criticized for its lack of reliability. Because of the high
degree of uncertainty and subjectivity inherent in fair value measurement, not much can
Table IV.
Perceived reliability of fair
value estimates rating
(tests of H1, H2 and H4):
results of simple effects on
perceived reliability (test
of H4)
Condition Mean difference Standard deviation Signifcance
Gain
Point 75 per cent confdence ?1.84 0.68 0.02
Point Range ?0.68 0.65 0.91
75 per cent confdence Range 1.17 0.64 0.22
Loss
Point 75 per cent confdence 1.28 0.63 0.14
Point Range 0.91 0.63 0.46
75 per cent confdence Range ?0.37 0.65 1.00
Notes: The perceived reliability rating is for the following question on an 11-point scale: For the
purpose of making investment-related decisions, how reliable do you fnd the fair value estimate of
$100,000? (1 being not reliable at all, 1 being very reliable) The dependent variable is perceived
reliability; the independent variable is information precision and is manipulated at three levels: point
stands for a precise point estimate; 75 per cent confdence stands for a point estimate at 75 per cent
confdence level; and range stands for a confdence interval; H1 predicts that investors perceive higher
reliability in imprecise fair value estimates than precise fair value estimates; to test H1, we treat the
point estimate as the category of precise estimate (average mean of 5.22); then we consider the range
estimate and the point estimate at 75 per cent confdence level as one category of imprecise estimate
(average mean of 5.42); H2 predicts that investors perceive higher reliability when fair value estimate is
presented as a point estimate with a specifed confdence level (average mean of 5.73) than an interval
range estimate (average mean of 5.10) and a precise point estimate (average mean of 5.22); to test H2, we
compare the average means for the three levels of precision; H4 predicts that a point fair value estimate
with a specifed confdence level is perceived to be more reliable than a range fair value estimate or a
precise point estimate when this fair value estimate results in potential gains; fair value precision has no
effect on perceived reliability when fair value estimate results in potential losses; to test H4, we conduct
simple effect tests to evaluate pairwise differences among the adjusted means using the Holm’s
sequential Bonferroni to control for Type I error across the three pairwise comparisons
199
Management
incentive and
information
precision
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
be done to improve the reliability of its information content, given the profession’s
extant measurement techniques. However, we argue that a careful choice of its
presentation format may improve its perceived reliability in users’ minds. Our evidence
provides some support to this conjecture. In general, participants believe fair value
estimates resulting in gain as less reliable than those in loss, potentially due to higher
management bias perceived in the gain condition. Our results also suggest that, instead
of relying on a precise point estimate, managers may improve the perceived reliability of
fair value estimates in the gain condition by specifying the degree of confdence. This
disclosure can be included in the notes to the fnancial statements or through other
channels.
This experimental investigation, as in the case of others, is limited in several
respects. We manipulate the precision of fair value at three levels, a precise point
3.63
5.56
4.3
6.8
5.9 5.89
3.5
4
4.5
5
5.5
6
6.5
7
Point 75% confident Range
R
e
l
i
a
b
i
l
i
t
y
Gain
Loss
Notes: This figure presents the means for the
dependent variables: perceived management bias
and perceived reliability. The horizontal axis
presents the independent variable at three levels:
point stands for a precise point estimate; 75 per
cent confidence stands for a point estimate at 75
per cent confidence level; and range stands for a
confidence interval. The ratings are presented on
the vertical axis. They are based on the following
questions: For the purpose of making investment-
related decisions, how reliable do you find the
fair value estimate of $100,000? (1 being not
reliable at all, 1 being very reliable). When using
the discounted cash flow model, how likely
would Company A’s management intentionally
use assumptions that lead to the current valuation
of Company B? (1 being very unlikely, 11 being
very likely)
Figure 1.
Various judgments of fair
value: perceived reliability
Table V.
Perceived management
bias rating (test of H3):
bias (mean) (test of H3)
Condition Point 75 per cent confdence Range
Gain 8.62 (2.58) 9.56 (1.54) 9.00 (0.92)
n ?16 n ?18 n ?20
Loss 7.00 (3.01) 7.63 (2.31) 6.56 (2.43)
n ?20 n ?16 n ?18
ARJ
27,2
200
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
estimate of $100,000, a 75 per cent confdence point estimate and an interval estimate
of $75,000-$125,000. The 75 per cent confdence interval is chosen arbitrarily. We
chose this level to allow for a robust test of the theory, and however, results of our
study could be sensitive to the levels chosen for the precision variable. A critical
assumption that we use here is that participants’ beliefs are symmetric (uniform or
normal). To the extent that investors’ beliefs are skewed, the interval estimate and
the 75 per cent confdence level may not be equivalent. In addition, the experimental
design uses an investment level of $100,000. We did not provide participants with
information about materiality. To the extent that participants do not believe this
level is material, the results may be unreliable. Furthermore, we need to be careful in
generalizing our fndings because of our marginally signifcant results, which could
be due to our relatively small sample size. Future work may employ a larger sample
8.62
9.56
9
7
7.63
6.56
6
6.5
7
7.5
8
8.5
9
9.5
10
Point 75% confident Range
B
i
a
s
Gain
Loss
Notes: This figure presents the means for the
dependent variables: perceived management bias
and perceived reliability. The horizontal axis
presents the independent variable at three levels:
point stands for a precise point estimate; 75 per
cent confidence stands for a point estimate at 75
per cent confidence level; and range stands for a
confidence interval. The ratings are presented on
the vertical axis. They are based on the following
questions: For the purpose of making investment-
related decisions, how reliable do you find the
fair value estimate of $100,000? (1 being not
reliable at all, 1 being very reliable). When using
the discounted cash flow model, how likely
would Company A’s management intentionally
use assumptions that lead to the current valuation
of Company B? (1 being very unlikely, 11 being
very likely)
Figure 2.
Various judgments of fair
value: perceived
management bias
Table VI.
Perceived management
bias rating (test of H3):
two-way ANOVA with
bias rating as the
dependent variable (test of
H3)
Source SS df MS F p (2-sided)
Precision 18.00 2 9.00 1.81 0.17
Direction 112.35 1 112.35 22.55 0.00
Precision ?direction 3.08 2 1.54 0.31 0.74
Within cells (error) 503.13 102 4.98
201
Management
incentive and
information
precision
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
to further validate our results. Finally, this study uses MBA students rather than
real investors. MBA students may not have strong knowledge of management
incentives and may interpret fair value estimates differently. Future studies should
explore how real market participants react to differences in formats of reporting fair
market values.
Table VII.
Perceived management
bias rating (test of H3):
results of simple effects
(test of H3)
Condition Mean difference Standard deviation Signifcance
Gain
Point 75 per cent confdence ?1.08 0.78 0.51
Point Range ?0.38 0.75 1.00
75 per cent confdence Range 0.71 0.74 1.00
Loss
Point 75 per cent confdence ?0.69 0.75 1.00
Point Range 0.44 0.73 1.00
75 per cent confdence Range 1.13 0.77 0.44
Notes: H3 predicts the perceived degree of management bias is higher in a fair value gain
condition than in a fair value loss condition regardless of the precision of fair value information; to
test this hypothesis, we asked the following question on an 11-point scale: “When using the
discounted cash fow model, how likely would Company A’s management intentionally use
assumptions that lead to the current valuation of Company B?” (1 being very unlikely, 11 being
very likely); the dependent variable is the degree of management bias; the independent variable,
information precision, is manipulated at three levels: point stands for a precise point estimate, 75
per cent confdence stands for a point estimate at 75 per cent confdence level and range stands for
a confdence interval
Table VIII.
Additional analysis
Rating 50 per cent confdence Wider range Imprecise verbal
Reliability
Mean 4.93 4.07 4.20
Standard deviation 1.73 2.09 1.78
Bias
Mean 6.86 6.80 7.52
Standard deviation 2.51 2.76 1.76
N 14 15 25
Notes: The new experiment intends to understand how sensitive our results are to the confdence
level chosen (e.g., 50, 60 or 75 per cent) or size of the range provided; it solely focuses on the gain
condition and includes three different levels of imprecision; for the point estimate with confdence
level, we provided participants with a lowered confdence level of 50 per cent; for the range
estimate, we widened the interval to lower the confdence; specifcally, we provide participants
with a wider interval, which ranges as low as $50,000 and as high as $150,000; moreover, we
include another condition of imprecise statement by providing a statement that fair values are
imprecise without providing either a confdence level or range; specifcally, we provide
participants with the following statement: “Though the fair value is estimated at $100,000 for B,
A’s management indicates that this number is imprecise”
ARJ
27,2
202
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Note
1. In its glossary of terms, Concepts Statement 2 (SFAC 2) of US GAAP defnes reliability as the
quality of information that assures that information is reasonably free from error or bias and
faithfully represents what it purports to represent. With respect to measures, it states that
“*t]he reliability of a measure rests on the faithfulness with which it represents what it
purports to represent, coupled with an assurance for the user, which comes through
verifcation, that it has that representational quality” (paragraph 59). Thus, the principal
components of reliability are representational faithfulness and verifability (see The FASB’S
Conceptual Framework: Relevance and Reliability by Financial Accounting Standards
Advisory Council September 2004). In October 2010 the FASB issued a new concept
statement, SFAC 8, which replaces SFAC 1 and 2.
References
Ball, R., Robin, A. and Wu, J.S. (2003), “Incentives versus standards: properties of accounting
income in four East Asian countries, and implications for acceptance of IAS”, Journal of
Accounting and Economics, Vol. 36 Nos 1/3, pp. 235-270.
Baginski, S.P., Hassell, J.M. and Kimbrough, M.D. (2004), “Why do managers explain their
earnings forecasts?”, Journal of Accounting Research, Vol. 42 No. 1, pp. 1-29.
Blacconiere, W.G., Frederickson, J.R., Johnson, M.F. and Lewis, M.F. (2009), “Do voluntary
disclosures that disavow the reliability of mandated fair value information refect
legitimate concerns about reliability?”, Working paper, IN University, IN.
Budescu, D.V. and Du, N. (2007), “The coherence and consistency of investors’ probability
judgments”, Management Science, Vol. 53 No. 11, pp. 1731-1744.
Du, N. and Budescu, D.V. (2005), “The effects of imprecise probabilities and outcomes in
evaluating investment options”, Management Science, Vol. 51 No. 12, pp. 1791-1803.
Du, N., Budescu, D.V., Shelley, M. and Omer, T. (2011), “The appeal of vague fnancial forecasts”,
Organizational Behavior and Human Decision Processes, Vol. 114 No. 2, pp. 179-189.
Du, N., Lin, L. and McEnroe, J. (2010), “Is the truth the problem?”, The CPA Journal, Vol. 80 No. 1,
pp. 6, 8, 10-11.
Eagly, A.H. and Chaiken, S. (1975), “An attribution analysis of the effect of communicator
characteristics on opinion change: the case of communicator attractiveness”, Journal of
Personality and Social Psychology, Vol. 32, pp. 136-144.
Fiske, S.T. and Pavelchak, M.A. (1986), “Category-based versus piecemeal-based affective
responses: developments in schema-triggered affect”, in Sorrentino, R.M. and Higgins, E.T.
(Eds), Handbook of Motivation and Cognition: Foundations of Social Behavior, Guilford
Press New York, NY, pp. 167-203.
Frederickson, J., Hodge, F. and Pratt, J. (2006), “The evolution of stock option accounting:
disclosure, voluntary recognition, mandated recognition and management disavowals”,
The Accounting Review, Vol. 81 No. 5, pp. 1073-1093.
He, X.J., Wong, T.J. and Young, D.Q. (2012), “Challenges for implementation of fair value
accounting in emerging markets: evidence from China”, Contemporary Accounting
Research, Vol. 29 No. 2, pp. 538-562.
Hirst, D.E., Koonce, L. and Simko, P.J. (1995), “Investor reactions to fnancial analysts’ research
reports”, Journal of Accounting Research, Vol. 33 No. 3, pp. 335-351.
Hirst, D.E., Koonce, L. and Miller, J. (1999), “The joint effect of management’s prior forecast
accuracy and the form of its fnancial forecasts on investor judgment”, Journal of
Accounting Research, Vol. 37 No. 1, pp. 101-124.
203
Management
incentive and
information
precision
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Hodge, F., Hopkins, P. and Pratt, J. (2006), “Management reporting incentives and classifcation
credibility: the effects of reporting discretion and reputation”, Accounting, Organizations,
and Society, Vol. 31 No. 7, pp. 623-634.
International Accounting Standard Board (IASB) (2011), Presentation of Financial Statements,
International Accounting Standard (IAS) 1, London.
Juslin, P., Winman, A. and Olsson, H. (2000), “Naive empiricism and dogmatism in confdence
research: a critical examination of the hard-easy effect”, Psychology Review, Vol. 107 No. 2,
pp. 384-396.
Klayman, J., Soll, J., Gonzalez-Vallejo, C. and Barlas, S. (1999), “Overconfdence: it depends on
how, what, and whom you ask”, Organizational Behavior and Human Decision
Processes, Vol. 79 No. 3, pp. 216-247.
Maines, L. and Wahlen, J. (2006), “The nature of accounting information reliability: inferences
from archival and experimental research”, Accounting Horizons, Vol. 20 No. 4, pp. 399-425.
Mercer, M. (2004), “How do investors assess the credibility of management disclosures?”,
Accounting Horizons, Vol. 18 No. 3, pp. 185-196.
Rottenstreich, Y. and Tversky, A. (1997), “Unpacking, repacking, and anchoring: advances in
support theory”, Psychology Review, Vol. 104 No. 2, pp. 406-415.
Winman, A., Hansson, P. and Juslin, P. (2004), “Subjective probability intervals: how to reduce
overconfdence by interval evaluation”, Journal of Experimental Psychology, Learning,
Memory, Cognition, Vol. 30 No. 6, pp. 1167-1175.
Further reading
Budescu, D.V. and Wallsten, T.S. (1995), “Processing linguistic probabilities: general principles
and empirical evidence”, in Busemeyer, J., Medin, D.L. and Hastie, R. (Eds), Decision Making
from a Cognitive Perspective, Academic Press, San Diego, CA, pp. 275-318.
Du, N. (2009), “Do investors react differently to range and point management earnings forecasts?”,
The Journal of Behavioral Finance, Vol. 10 No. 4, p. 195.
Financial Accounting Standards Board (FASB) (1978), “Objectives of fnancial reporting by
business enterprises”, Statement of Financial Accounting Concepts No. 1, FASB,
Norwalk, CT.
Financial Accounting Standards Board (FASB) (2006), “Fair value measurements”, Statement of
Financial Accounting Standards No. 157, FASB, Norwalk, CT.
Financial Accounting Standards Board (FASB) (2007), “The fair value option for fnancial assets
and fnancial liabilities”, Statement of Financial Accounting Standards No. 159, FASB,
Norwalk, CT.
Jorion, P. (2002), “Howinformative are value-at-risk disclosures?”, The Accounting Review, Vol. 77
No. 4, pp. 911-931.
Kahneman, D. and Tversky, A. (1979), “Prospect theory: an analysis of decision under risk”,
Econometrica, Vol. 47 No. 2, pp. 263-292.
Kennedy, J., Mitchell, T. and Sefcik, S.E. (1998), “Disclosure of contingent environmental
liabilities: some unintended consequences?”, Journal of Accounting Research, Vol. 36,
pp. 257-277.
Libby, R., Nelson, M.W. and Hunton, J. (2006), “Recognition versus disclosure, auditor tolerance
for misstatement, and the reliability of stock-compensation and lease information”, Journal
of Accounting Research, Vol. 44 No. 3, pp. 533-560.
ARJ
27,2
204
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Libby, R., Tan, H.T. and Hunton, J.E. (2006), “Does the form of management’s earnings guidance
affect analysts’ earnings forecasts?”, The Accounting Review, Vol. 81 No. 1, pp. 207-225.
Wallsten, T.S. (1990), “Measuring vague uncertainties and understanding their use in decision
making”, in von Furstenberg, G.M. (Ed.), Acting Under Uncertainty, Kluwer, pp. 377-398.
Zwick, R. and Wallsten, T.S. (1989), “Combining stochastic and linguistic inexactness: theory and
experimental evaluation of four fuzzy probability models”, International Journal of
Man-Machine Studies, Vol. 30, pp. 69-111.
Appendix
Fair value gain (loss) in point estimate
Please assume you work in the investment department of a frm. Your supervisor has instructed
you to assess the fnancial performance of Company A, a 100-year old fnancial boutique frmthat,
through its wholly owned and partially owned subsidiaries, is engaged in fnancial services
activities. After extensive research, you prepared a short summary with some important facts
about Company A.
• Co. A acquired a minority interest in Co. B two years ago at a price of $60,000 ($140,000). B
is a non-public, technology-based asset management company that is in the process of
growing its business and has the ultimate goal of becoming one of the premier investment
advising companies in the market. However, B has not yet achieved a positive cash fow in
its month-to-month operations at December 31, 2009.
• At the end of 2009, the fair value of Bis estimated at $100,000, $40,000 higher (lower) than its
original purchase price of $60,000 ($140,000). Company Aadjusts its fnancial statements to
refect this increase (decrease). The investment in B is nowcarried at $100,000 as an asset in
the balance sheet, and the increase (decrease) of $40,000 is recognized as unrealized holding
gain (loss) in its comprehensive income statement.
• The fair value of $100,000 is estimated by A’s management. Because there is no publicly
observed market-based information about B (B is not traded in active markets), and there is
no company in the marketplace that is comparable to B, Co. A’s management used the
Discounted Cash Flow Analysis as the valuation model. The model uses projected fnancial
information for the next fve years as inputs and relies heavily on managers’ assumptions of
future cash fow and risk adjusted-discount rate.
Before you make recommendation to your clients about whether to invest in Company A, your
supervisor would like to know your judgments. Please proceed to the next page to answer the
questions.
(For range estimate, the following sentence is added: “Though the fair value is estimated at
$100,000 for B, A’s management indicates that this estimate may range as low as $75,000 and as
high as $125,000 depending on the assumptions used”.)
(For 75 per cent confdence point estimate, the following sentence is added: “Though the fair
value is estimated at $100,000 for B, A’s management indicates that they are only 75 per cent
confdent about this estimate because of the assumptions used”.)
Please answer the following three questions:
(1) When using the discounted cash fowmodel, howlikely would Company A’s management
intentionally use assumptions that lead to the current valuation of Company B?
(Very unlikely) 1 2 3 4 5 6 7 8 9 10 11 (Very likely)
(2) For the purpose of making investment-related decisions, how reliable do you fnd the fair
value estimate of $100,000?
(Not reliable at all) 1 2 3 4 5 6 7 8 9 10 11 (Very reliable)
205
Management
incentive and
information
precision
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
(3) How risky is company a’s the investment in company B?
(Not at all risky) 1 2 3 4 5 6 7 8 9 10 11 (Very risky)
Additional questions:
(1) Please indicate your gender (check one).
Male _______ Female _______
(2) How much working experience do you have?
____________ years __________ months
(3) How much investment experience do you have?
____________ years __________ months
(4) Compared to the historical cost, the fair value estimate of company B is:
higher ____ lower ____ (check one)
(5) Which estimate best describes the fair value provided? (check one)
$100,000 _____ $100,000 at 75 per cent confdent _____ $75,000-$125,000 _____
Corresponding author
Ning Du can be contacted at: [email protected]
To purchase reprints of this article please e-mail: [email protected]
Or visit our web site for further details: www.emeraldinsight.com/reprints
ARJ
27,2
206
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
2
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
doc_551487061.pdf