Download as pdf or txt
Download as pdf or txt
You are on page 1of 23

European Accounting Review

Vol. 16, No. 2, 277 –298, 2007

Does Pro Forma Reporting Bias


Analyst Forecasts?

PATRIC ANDERSSON & NICLAS HELLMAN



Centre for Economic Psychology, Stockholm School of Economics, Sweden and   Department of
Accounting and Managerial Finance, Stockholm School of Economics, Sweden

ABSTRACT Standard setters put much effort into the development of ‘better’ financial
reporting standards, that is, standards that more accurately capture the economic
substance of business activities. However, the more sophisticated accounting treatments
caused by new standards, and the growing complexity of business activities as such, has
made financial reports more difficult to understand. In response to this situation, some
companies use pro forma reporting, which means that certain complex items required
by financial reporting standards are excluded. This study adopts a user perspective and
investigates how pro forma reporting affects analysts’ judgments in an experimental
setting. On the basis of psychological theory, our hypothesis suggests that analysts’
judgments will be affected by differences in the way company performance is
presented. Our results show that analysts who received both pro forma and Generally
Accepted Accounting Principles (GAAP) information made significantly higher
earnings per share (EPS) forecasts than those who received GAAP information only. It
is argued that positive framing and higher levels of anchor explain this result, which
suggests in turn that analysts’ EPS forecasts can be manipulated by alternative ways of
presenting company performance. Some possible implications of this finding for
standard setters are discussed.

Correspondence Address: Niclas Hellman, Ph.D., Assistant Professor, Department of Accounting


and Managerial Finance, Stockholm School of Economics, Box 6501, 113 83 Stockholm, Sweden.
E-mail: Niclas.Hellman@hhs.se

0963-8180 Print/1468-4497 Online/07/020277–22 # 2007 European Accounting Association


DOI: 10.1080/09638180701390966
Published by Routledge Journals, Taylor & Francis Ltd on behalf of the EAA.
278 P. Andersson & N. Hellman

1. Introduction
There is a general international trend towards increasingly complex accounting.
Pro forma reporting, that excludes some of the complex GAAP-required items,1
offers a way of demystifying complex accounting disclosures to allow for better
communication of a company’s ‘core earnings’ (Bhattacharya et al., 2004). Pro
forma reporting started in the USA in the mid-1990s (Bhattacharya et al.,
2003), and has since been adopted by many companies, including large non-
US companies such as Nokia and British Telecom.
The pro forma adjustments generally lead to improved results (Bradshaw and
Sloan, 2002),2 and critics of pro forma reporting claim that it is simply a way of
increasing profits to make a company look more attractive to analysts and inves-
tors. Doyle et al. (2003) report that expenses excluded by the calculation of pro
forma earnings are negatively related to future cash flows, which indicates that
the items excluded from the pro forma statement are in fact relevant to analysts
and investors. If the users of accounting information do not accept the complex
GAAP reports, but rely on simplified and improved pro forma numbers, their
judgment may be biased. However, experimental research (Frederickson and
Miller, 2004; Elliott, 2006) indicates that professional users (i.e. financial ana-
lysts) are able to ‘see through’ the pro forma adjustments, and their judgment
does not become biased. The main purpose of the present study is to test
whether this finding also holds for settings in which there is substantial difference
between (negative) pro forma and (positive) pro forma earnings. Thirty-six
Swedish financial analysts in an experimental setting were given fourth-quarter
earnings releases for an anonymous listed company, and were asked to forecast
the next year’s earnings per share (EPS) according to Swedish GAAP. One
group of analysts was given GAAP information only, while another group
received both pro forma and GAAP information. The results show that the ana-
lysts who were given both pro forma and GAAP information made significantly
higher EPS (according to GAAP) forecasts, compared to the analysts who
received GAAP information only. If an alternative (non-GAAP) presentation
of a company’s performance draws attention away from GAAP numbers in
such a way as to make sophisticated users change their judgment, as our
results suggest, this represents a potential threat to the public’s trust in account-
ing, in that managers may modify the pro forma reporting in an opportunistic
way. We suggest that standard setters should be concerned about the proliferation
of pro forma earnings and about their presentation and use by various parties.

2. Pro Forma Reporting


2.1. The Phenomenon
The Latin phrase ‘pro forma’ refers to a fictive situation as if certain things either
had, or had not, happened. In the case of pro forma profits, the idea is to present
the net profit figure as if certain events leading to ‘unusual’ items had not
Does Pro Forma Reporting Bias Analyst Forecasts? 279

occurred. Pro forma income statements are included in companies’ quarterly


earnings releases. They often precede the income statement according to
GAAP, and it is common for the written text in the earnings release to emphasize
the pro forma figures. One feature of pro forma reporting is that the unusual
item adjustments persist throughout the whole income statement and affect the
bottom line profit and the EPS calculation. Thus, putting more emphasis on
EBITDA (earnings before interest, taxes, depreciation and amortization) or
non-recurring items in the earnings release is not enough to qualify as pro
forma reporting.
Bhattacharya et al. (2003) examine the use of quarterly pro forma reporting
among US companies between 1998 and 2000. They note that pro forma report-
ing increased dramatically during this period and that the pro forma announcers
were particularly common in the service and high-tech industries. However,
although many companies did go in for pro forma reporting, its occurrence
was still fairly infrequent. Of 1,149 earnings releases in the sample, only 11%
of the companies reported pro forma earnings for four or more quarters during
the relevant period and very few of these cases occurred in consecutive quarters
(ibid., p. 296). In an earlier study, Bradshaw and Sloan (2002) reported that pro
forma profits were significantly higher than GAAP profits, and that the largest pro
forma profit overstatements occurred in the fourth quarter.
Hitherto, pro forma reporting has been subject to little regulation. In 2003, the
Securities and Exchange Commission (SEC) issued a regulation requiring com-
panies reporting on a pro forma basis to disclose a quantitative reconciliation of
the differences compared to GAAP (SEC, 2003). The empirical results presented
in Bowen et al. (2005) suggest that SEC’s recommendation in 2001 regarding pro
forma reporting was already helping to reduce the emphasis on pro forma earn-
ings in 2002. The International Financial Reporting Standards (IFRS) do not
include an adequate standard on the presentation of IFRS financial statements.
On the basis of a review of 65 large European companies who implemented
the IFRS in 2005, Ernst & Young (2006) report:

The widespread use of alternative, non-IFRS measures in companies’


results announcements and presentations suggests that . . . there is a gap
between IFRS and what managements believe is necessary in order to com-
municate to the markets information which enables underlying perform-
ance and sustainable cash flows to be assessed.

(Ernst & Young, 2006, p. 4)

In Sweden, the new listing rules on the Stockholm Stock Exchange require
companies to include three specific GAAP-measured items (turnover, net
profit and earnings per share) at the beginning of their earnings releases
(Stockholmsbörsen, 2003).
280 P. Andersson & N. Hellman

2.2. Why do Companies Report Pro Forma Profits?


Over the last decade, the International Accounting Standards Committee/Board
(IASC/IASB) and the Financial Accounting Standards Board (FASB) have both
issued a variety of complex standards, and the overall collection of standards is
now very extensive under both IFRS and US GAAP. A possible explanation for
the tendency towards pro forma reporting could be that managers and financial
analysts together are seeking to create a simpler and more comprehensible
mode of financial reporting. Harvey Pitt, the former SEC chairman, suggested
in a speech in 2001 that the increasing use of pro forma reporting should be
seen in relation to the growing complexity that is making it more difficult for
users to understand companies’ financial reporting according to GAAP:3

Investors anxious for current, simplified and comprehensible financial


reporting are today more likely to rely on a company’s ‘pro forma’ disclos-
ures than the same company’s meticulously prepared, mandated GAAP
financial disclosures.

(Pitt, 2001)

The argument that pro forma reporting satisfies the need of investors and analysts
for simplified reporting geared to the core earnings has received support in
market-based accounting research. Bradshaw and Sloan (2002, p. 64) conclude
that investors display a growing preference for the modified versions of profits
reported by the analyst tracking services compared to the profits reported accord-
ing to GAAP. Moreover, earnings response coefficients and regression R 2’s are
both higher for pro forma profits than for GAAP profits (Bradshaw and Sloan,
2002, p. 65; Bhattacharya et al., 2003, p. 316). As regards the use of pro
forma information for forecasting purposes, Bhattacharya et al. (2003) regressed
the revision of analysts’ one-quarter-ahead earnings forecasts on forecast errors
based on different earnings metrics. Their results suggest that pro forma earnings
have significantly greater explanatory power than earnings according to GAAP
(ibid., p. 309), which suggests that analysts regard pro forma earnings to be
more informative than GAAP earnings.
However, the use of pro forma reporting also involves some risks connected with
management’s behaviour and the quality of financial reporting. In preparing the pro
forma income statement, the items that are excluded in the GAAP counterpart are
referred to as ‘unusual items’, ‘non-core business items’, ‘non-cash items’, etc.
The managers’ official argument for this procedure is that it helps external stake-
holders to compare the underlying earnings capacity across companies. However,
the common anecdotal story is that managers seek to manipulate the external stake-
holders’ view of the company by excluding such items. As pro forma reporting is
highly dependent on management’s opinion of what is classified as ‘unusual’ or
‘non-core’, there is a substantial risk that the pro forma reporting will be
Does Pro Forma Reporting Bias Analyst Forecasts? 281

manipulated in such a way as to allow the company to meet earnings projections.


This issue, as to whether or not managers behave opportunistically, has been
examined on the basis of archival data, and the results are mixed. Bhattacharya
et al. (2004, p. 36) report that a very large proportion of the pro forma companies
in their sample actually change the definition of pro forma earnings from one
period to another, which suggests manipulative behaviour. In a similar vein,
Doyle et al. (2003) examine the predictive value of expenses that are excluded
from pro forma earnings, and find that although some expenses (‘special items’)
are generally unrelated to future cash flows, the so-called ‘other exclusions’ are
very predictive of future negative cash flows (ibid., p. 148). Bowen et al. (2005)
investigate the factors associated with the emphasis placed on pro forma earnings
compared to GAAP earnings. They conclude that managers are using specific
emphasis in the earnings press release as a disclosure tool that has an influence
on investors. In a comparable study, on the other hand, Bradshaw and Sloan
(2002) examined management discussions in earnings releases, and they report
that although managers have stepped up their efforts to emphasize pro forma
profits, the empirical results do not give any indication of the relative displacement
of GAAP earnings in the earnings announcements (ibid., p. 63). In a similar vein,
Johnson and Schwartz (2001) show that the magnitude of the market premium
for pro forma firms is unrelated to the specific characteristics of the pro forma
disclosures, and argue that investors are not misled by pro forma profits.
If managers do in fact modify pro forma reporting in an opportunistic way, as
some of the above research findings suggest, then there is a potential threat here
to the public’s trust in accounting. FASB (2002) has also expressed concern that
pro forma reporting may undermine the quality of financial reporting. However,
there would be less of a problem if investors were able to ‘see through’ any
attempt at exaggerating earnings. This question has been subjected to experimen-
tal research, and is also the focus of the present paper. Thus, we will now leave
the pro forma phenomenon as such, and will consider instead whether or not this
type of reporting affects the judgments made by its users.

2.3. The Impact of Pro Forma Reporting on Analyst Judgments


An important issue connected with pro forma reporting concerns the extent to which
it influences the judgments and decisions of investors, financial analysts, forecasters
and other actors in the financial market. Up to now, to the best of our knowledge,
two experimental studies have investigated this issue empirically. Frederickson and
Miller (2004) conducted an experiment involving two groups of participants: 46
MBA students (who acted as proxies for non-professionals) and 34 experienced
analysts. The participants’ main task was to study a particular case, appraise
the fair value of the company’s stock and then to provide an explanation in
writing of the reasoning behind their judgment. They were given background infor-
mation and an earnings announcement from a hypothetical technology-oriented
company in the medical sector. The announcement was manipulated so as to
282 P. Andersson & N. Hellman

report GAAP earnings only in one version and both pro forma and GAAP earnings
in the other. Apart from different levels of annual percentage increase (8.8% as
against 0.8%), the pro forma earnings were slightly higher than the GAAP earnings
($1.68 as against $1.24 per share). The case was mailed to the analysts, but was
given to the students in small groups. On a basis of evidence from psychological
research on decision-making and expertise, Frederickson and Miller (2004) hypoth-
esized that non-professionals would predict higher stock prices when using
announcements including both pro forma and GAAP earnings than they would
do if the announcements only included GAAP earnings. As regards the financial
analysts, Frederickson and Miller formulated an explorative research question
instead of a formal hypothesis about the effect of pro forma reporting on analysts’
judgments, since earlier research had generated mixed evidence regarding the ten-
dency for accounting experts to be the victim of cognitive bias (e.g. Shanteau, 1989;
Smith and Kida, 1991; Maines, 1995). Frederickson and Miller (2004) found, in line
with their hypothesis, that the 22 MBA students who were exposed to pro forma
plus GAAP earnings announcements made higher stock price appraisals than did
the 24 students who received GAAP information only. However, this effect
could be explained by differences in the valuation models applied and by uninten-
tional cognitive effects, rather than by the non-professionals relying on pro forma
earnings information because they perceived it as being informative. Regarding
the explorative research question, it was found that the appraised stock prices
given by the participating analysts were similar regardless of the experimental
conditions. In other words, the analysts were not influenced by the presence of
pro forma reporting. Frederickson and Miller (2004) suggested that this finding
was due to the tendency on the part of the participating analysts to rely on
well-defined valuation models.
Elliott (2006) casts further light on the effects of pro forma reporting, investi-
gating two issues: (1) whether it is the presence of pro forma information or the
particular emphasis of such information that affects non-professionals; and (2)
the extent to which the judgment of analysts and non-professionals is affected
when there is a quantitative reconciliation between the GAAP and pro forma
earnings calculations. On a basis of behavioural research showing that salient
information tends to capture people’s attention regardless of its usefulness,
Elliott hypothesized that non-professionals would arrive at a different judgment
when the emphasis was on pro forma earnings. She also predicted that a specifi-
cation of the pro forma adjustments (the reconciliation) would reduce the effect
of pro forma reporting on the judgment of non-professionals, whereas it would
not affect the judgment of analysts. To test these hypotheses Elliott conducted
an experiment involving 89 MBA students (representing non-professional inves-
tors) and 55 experienced analysts. The participants read a press release regarding
a hypothetical company in the technology sector. The release consisted of finan-
cial information for the first quarter of the last two years. The participants were
asked to rate the company’s earnings performance (on a 100 points scale ranging
from very weak to very strong) and then to decide on the amount of money to
Does Pro Forma Reporting Bias Analyst Forecasts? 283

invest in the company (ranging from $0 to $5,000). The press release was
manipulated as regards the presence of pro forma information, the emphasis on
pro forma information and the presence of a reconciliation of the differences
between the GAAP and the pro forma measures. The GAAP earnings were nega-
tive and the pro forma earnings slightly positive, respectively (i.e. a loss of $1.02
as against a profit of $0.01 per share). Elliott found that non-professionals tended
to rate earnings performance higher and to invest more money when they were
exposed to the press release emphasizing pro forma earnings. This tendency
was weaker when the same group read the press release that included a reconci-
liation. The analysts revealed no significant difference in judgment between the
press release emphasizing pro forma earnings and the other that included GAAP
information only. However, contrary to her hypothesis, Elliott (2006) reported
that analysts who had examined a press release involving a reconciliation gave
higher ratings to earnings performance and were willing to invest more money
in the company concerned than the analysts who were confronted by press
releases giving GAAP information only or those putting more emphasis on pro
forma earnings than on GAAP earnings. In her view, this somewhat paradoxical
finding could be explained by the tendency of the analysts to regard pro forma
earnings as more reliable and less misleading when reconciliation was present.
In sum, earlier studies provide consistent empirical evidence that pro forma
reporting does not seem to have any effect on analysts’ judgments, while the
judgments of non-professional investors may be so affected.4 However, the evi-
dence should be considered in light of the following factors which, in our view,
limit its validity and reliability. First, Frederickson and Miller (2004) rely on a
case in which pro forma earnings do not differ substantially from GAAP earnings.
The difference per share was $0.44, corresponding to about 1.7% of the mean
stock price appreciation made by the analysts. Bigger discrepancies could have
consequences for the analysts’ judgments. Elliott’s (2006) case material included
a somewhat bigger difference ($1.03 per share), but then there was also a ten-
dency towards higher analyst rankings for earnings performance in the GAAP
press release compared to the case of the pro forma earnings press release.5
Further, neither Frederickson and Miller (2004) nor Elliott (2006) analysed the
case of pro forma profits and GAAP losses in pro forma reports,6 although
such combinations are not uncommon in practice, according to the results of
studies of pro forma reporting based on archival data:

. . . the mean GAAP earnings is a net loss of 14.7 cents per share, while the
mean pro forma earnings is a net income of 8.5 cents per share. This result
is consistent with the notion expressed in the popular press that most items
excluded from GAAP income to arrive at the pro forma number are
expenses that decrease income. Moreover, it is consistent with the results
of prior academic research examining actual pro forma press releases . . .

(Bhattacharya et al., 2004, p. 32)


284 P. Andersson & N. Hellman

Secondly, prompting analysts to estimate fair stock values for a company


(Frederickson and Miller, 2004) or to decide on the amount of money they are
prepared to invest in a firm (Elliott, 2006) is an indirect way of capturing these
actors’ predictions, and such interventions are problematic in the context of
task realism and differences in individual attitudes to risk.7 Moreover, when per-
forming the tasks the participants may have considered the general condition of
the stock market and the economy at the time when they were taking part in the
experiment. For example, the analysts participating in the study described in
Frederickson and Miller (2004) were asked to evaluate the financial information
of a technology-oriented company for the year 2001. It is plausible that,
consciously or unconsciously, their judgment was affected by the poor state of
the stock market in 2001. Thirdly, the participants of the two studies were not
being monitored when they were performing the tasks, which mean that
there is some uncertainty as to whether the participants acted in line with their
instructions. The possibility that they employed decision-aids or that the
responses were produced by someone else, cannot be safely ruled out. Fourthly,
the material used in the two studies involved fictitious firms in the technology
sector. Consequently, the results might not allow for generalizations about
other industries, associated with different financial characteristics. It is these
factors that motivated the writing of the present paper.

2.4. Rationale and Contributions of the Present Paper


Our paper proceeds from prior research on the effects of pro forma reporting on
the judgments made by financial analysts. Basing our argument on psychological
theory, we suggest that financial analysts will be affected by exposure to a com-
bination of negative GAAP earnings and positive pro forma earnings. This is not
an unusual combination in practice, as we have already noted. Given a substantial
discrepancy between GAAP and pro forma earnings, we suggest that the forecast
of earnings made by analysts confronted by pro forma and GAAP information
will be higher than those made by analysts who receive GAAP information only.
Moreover, earlier studies have relied on dependent measures that raise ques-
tions about reliability and validity. In an attempt to increase the external validity
of the dependent variable, we employ another measure based on earnings fore-
casts. By asking analysts to provide earnings estimates for the coming year we
hope to eliminate confounding factors such as individual risk propensity, incen-
tives and task realism that are present in prior research (Frederickson and Miller,
2004; Elliott, 2006). We also seek to improve the reliability of the whole under-
taking by conducting a monitored experiment with the analysts rather than the
non-monitored experiments of analysts reported in Frederickson and Miller
(2004) and Elliott (2006). Finally, following Elliott’s (2006) suggestion, our
case concerns an authentic but anonymous firm working in an established indus-
try rather than in the technology sector.
Does Pro Forma Reporting Bias Analyst Forecasts? 285

In summary, our paper investigates how analysts’ earnings forecasts are


affected in a situation where the discrepancy between pro forma and GAAP earn-
ings is substantial and involves a pro forma profit and a GAAP loss. Further, we
hope to improve the validity and reliability in relation to earlier studies. Another
aim of the paper is to investigate whether the presence or the absence of pro
forma information affects the strength of the confidence that analysts place in
their earnings forecasts. This aspect of the subject has not been addressed in Fre-
derickson and Miller (2004) or Elliott (2006).
In contrast to earlier studies, the empirical material for the present paper con-
sists of the responses of European (or, more precisely, Swedish) financial analysts
rather than of US analysts. However, we feel that this divergence will not con-
taminate our results, as empirical evidence indicates that Swedish and US ana-
lysts are not too dissimilar. Using questionnaire data, Olbert (1994) reported
no significant differences between Swedish and US analysts when it came to
the frequency with which they used fundamental analysis for stock valuation pur-
poses, how often they used p/e ratios for stock valuation purposes, how often
they used earnings per share as a forecast variable, or how highly they ranked
profit and loss statements as a source of information. Any disagreement
between the findings of our study and those of earlier research may not thus
depend on differences in the nationality of the participants.
In the next section we will describe the theoretical foundation for our hypoth-
esis, namely that analysts can be expected to make higher EPS (according to
GAAP) forecasts if they receive pro forma and GAAP information compared
with their receipt of GAAP information only.

3. Theory and Hypothesis Development


Pro forma reporting is related to certain constructs appearing in research on cog-
nitive psychology and behavioural decision-making. In particular, this type of
reporting is associated with the notion of framing, which refers essentially to
the way decision situations are represented or perceived by decision-makers
(Tversky and Kahneman, 1981; Kahneman and Tversky, 1984). In an oft-cited
article, Kahneman and Tversky (1979) found that participants’ preferences dif-
fered depending on whether the alternative was framed as a win or a loss situ-
ation. Similarly, people tend to evaluate choices differently when the amounts
concerned are expressed in different currencies or in terms of nominal as
against real values (Soman, 2004). Evidence of the impact of framing on
human judgment has also been documented by Camerer (2000), using field
data. Similarly, pro forma reporting can be expected to create framing effects,
given that the discrepancy between pro forma earnings and GAAP earnings is
obvious or is at least perceived as discernible.
Tversky and Kahneman (1974) found that people tend to rely on a variety of
heuristics when making their judgments. One heuristic assumes that individuals
consider initial values (anchors) and then adjust their judgments from this base.
286 P. Andersson & N. Hellman

On the whole, individuals tend to make insufficient adjustments, leading in turn to


biased judgments (Chapman and Johnson, 2002). This anchoring heuristic is very
relevant in forecasting tasks such as the prediction of future earnings (cf. Ayton,
1998). Psychological research has shown that if you give someone a number – it
may or may not be informative – this will have a remarkable impact on the way
they respond to a question requiring a numerical answer, since, consciously or
unconsciously, subjects will anchor their responses to that number (Chapman
and Johnson, 2000). If people are confronted by more than one anchor, they
may try to consider all the values or they may stick to the one which they
guess to be the most relevant (Whyte and Sebenius, 1997). It seems likely that
their judgments will vary more than if they had only been offered a single
anchor. As regards pro forma profits, analysts who get both adjusted and
GAAP-based information (i.e. two anchors) will probably arrive at completely
different estimates compared to those of their peers who only have access to
EPS based on GAAP (i.e. a single anchor). Thus, starting from the psychological
concepts of framing and anchoring, it is reasonable to assume that pro forma
earnings will have a considerable impact on the predictions of financial analysts.
There are two objections to this line of argument. First, previous studies
(Frederickson and Miller, 2004; Elliott, 2006) of pro forma seem to suggest
that no influence exists. As noted above, there are some weak points. In particu-
lar, there is some doubt as to whether these studies were able to stimulate framing
and anchoring effects. It should be remembered that they employed experimental
material where the discrepancy between pro forma and GAAP earnings per share
was $0.44 (Frederickson and Miller, 2004) and $1.03 (Elliott, 2006), respect-
ively. From a cognitive psychological point of view, the two types of earnings
measures could be perceived as fairly similar. Anchoring research generally
relies on experimental tasks in which the anchors differ substantially
(Chapman and Johnson, 2002), as they do for example in Whyte and Sebenius
(1997) – between $8 and $12. Thus, previous pro forma studies may have
relied on measures that were not psychologically dissimilar. In this context, it
should be noted that there was a tendency towards greater impact of the pro
forma information on the analysts’ judgments in Elliott’s (2006) study compared
to the study made by Frederickson and Miller (2004). This tendency may perhaps
be related to the somewhat bigger difference between the anchors in Elliott’s
(2006) study.
A second objection is that analysts are experts, and should thus be less suscep-
tible to judgmental bias. Analysts have acquired extensive knowledge and devel-
oped methods for analysing accounting information (Bouwman et al., 1987).
However, research has found mixed evidence regarding the so-called expertise
argument. Smith and Kida (1991) reviewed a great many studies that investigated
whether professional auditors are victims of heuristics and bias. Their review
suggests that auditors do tend to be victims, but that tendency is weaker when
they face tasks that are job-related and familiar. Shanteau (1989) came to the
same conclusion. On the other hand, Smith and Kida (1991) could not rule out
Does Pro Forma Reporting Bias Analyst Forecasts? 287

the possibility that some audit decisions were associated with anchoring and
adjustment. It should be noted that auditors’ tasks differ from those of financial
analysts, which suggests that tendencies observed in one profession may not
necessarily apply to another. Further, there is a body of research showing that
expert judgment is fragile, even in the case of accustomed tasks (e.g. Camerer
and Johnson, 1991; Andersson, 2005; Tetlock, 2005). For example, despite con-
siderable experience, managers tend to be the victims of anchoring effects
(Lovallo and Kahneman, 2003). Anchoring has also been claimed to partly
explain overreactions and underreactions in analysts’ earnings forecasts (Amir
and Ganzach, 1998). Moreover, length of experience is also weakly related to
performance (Ericsson and Lehmann, 1996). Thus, even if financial analysts
are regarded as experts, they may still be sensitive to framing and anchoring
effects.
Despite the conflicting evidence as to whether or not experts are victims of
judgmental heuristics, previous studies (Frederickson and Miller, 2004; Elliott,
2006) have argued that analysts will not be influenced by pro forma information,
due to their expertise. Elliott (2006) for example found that analysts were not
affected by the emphasis on pro forma earnings and attributed this to their invest-
ment-related expertise and familiarity with the task. However, one manipulation
in Elliott’s experiments with her analyst subjects (the manipulation involving a
quantitative reconciliation between pro forma and GAAP information), yields
statistically significant results that are inconsistent with her hypothesis, thus
casting some doubt on the expertise argument that is used to support both of
the hypotheses related to analysts. Elliott attributed this inconsistency to a
greater perceived reliability regarding pro forma earnings as compared to
GAAP, when reconciliation is added. An alternative interpretation could be
that this manipulation stimulated framing and anchoring effects.8
Thus, we dispute the expertise argument and argue that the impact of pro forma
earnings is related primarily to the psychological theory of anchoring. In testing
this theory, the anchors used should not be too similar, in other words there needs
to be a marked difference between pro forma earnings and GAAP. One real-
world example of this could concern firms that are reporting pro forma earnings
at the same time as GAAP losses. Such firms also tend to give greater emphasis to
pro forma earnings in their press releases, compared with other companies
(Bowen et al., 2005, p. 1013), which motivates an emphasis on pro forma earn-
ings in the case design. In light of the above, the following hypothesis was
formulated.

H1: Earnings releases including both pro forma and GAAP-based infor-
mation will be associated with forecasts of higher EPS than earnings
releases that include GAAP-based information only.

As well as testing this hypothesis, we also look at whether pro forma reporting
affects the amount of confidence that financial analysts place in their forecasts,
288 P. Andersson & N. Hellman

an issue that has not been explored in previous studies. Intuitively, one would
expect that the conflicting information in earnings releases based on pro forma
and GAAP statements would inspire less confidence than earnings releases
based on GAAP alone. On the one hand, research shows that access to additional
information makes people more confident, even though the additional infor-
mation does not necessarily lead to more accurate predictions (Oskamp, 1982).
On the other hand, judgment research in financial accounting suggests that the
effect of providing professionals with additional information can be mixed (cf.
Maines, 1995). In view of these rather imprecise findings, we decided to
include a research question about the effects of pro forma reporting on confi-
dence, rather than formulating a hypothesis on the issue. For this reason, we
will also look into the level of confidence associated with earnings releases
including both pro forma and GAAP-based information and compare this with
the level of confidence associated with earnings releases including GAAP infor-
mation only.

4. Methodology
4.1. Rationale for the Chosen Experimental Approach
To test our hypothesis (H1), we decided to adopt an experimental approach. This
scientific method, which is commonly used in (Anglo-American) accounting
research (cf. Libby et al., 2002), enables a reliable evaluation of causes and
effects in controlled environments. For the purpose of the present paper, such
an approach made it possible to observe analysts’ forecasting behaviour under
exposure to two types of earnings releases, namely, GAAP-based or pro forma
information. Further, given the experimental approach, it was possible to
ensure that all analysts had equal access to the information (except in so far as
it was manipulated for experimental purposes) and to control for any contami-
nating factors. For instance, if analysts had responded on different occasions,
the possibility cannot be ruled out that their judgment might have been affected
by the vagaries of the daily (business) news. Similarly, incentives arising from
institutional circumstances also tend to distort analysts’ forecasts (e.g. Hunton
and McEwen, 1997). Another factor, which hampers the findings presented in
Frederickson and Miller (2004) and Elliott (2006), could be that the analysts
may employ a variety of decision aids or, may simply hand over the task to
someone not fully qualified to deliver responses on their behalf.
Further, the experimental approach has advantages lacking in other research
methods such as interviews or the econometric analysis of archival data, as it
allows for the separate examination of effects associated with the financial report-
ing. First, in real-life listed companies do not publish two separate earnings
releases that are sent to two groups of analysts. Second, in making their earnings
forecasts many analysts tend to take note of their colleagues’ estimates, so that
their own estimates do not deviate too far from these (Trueman, 1994). Third,
Does Pro Forma Reporting Bias Analyst Forecasts? 289

some analysts may possess private information enabling them to make more
reliable earnings forecasts.

4.2. Participants
Thirty-six financial analysts (29 of them males) who had enrolled in an executive
education programme in advanced finance were asked to make their own evalu-
ation of an earnings release. Their mean age was 34.8 years. The main occupation
of around 81% of them concerned corporate analyses and stock investment, while
the rest of them were engaged in stockbroking and giving financial advice. Their
average experience of the tasks concerned was 6.1 years.

4.3. Study Material


The study material consisted of two types of earnings release. One type included
accounting data based on pro forma and GAAP calculations, while the second
type involved GAAP-based accounting data only. The two types will be referred
to below as the pro forma report and the GAAP report, respectively. Following
the field, both types of earnings releases have been based here on figures from
a real-life listed company, but the identity of the company has been concealed.
This procedure guaranteed both realism and an ecologically valid task.
Both types of earnings releases opened with a brief summary of sales, income
before tax, earnings per share and dividends. See Table 1 for details.9 A few brief
sentences then described the business and market conditions of the company con-
cerned. These sentences were identical in the two types of earnings release. After
this brief summary, the pro forma report went on to present the pro forma income
statement, the GAAP income statement and the balance sheet. The GAAP report,
in contrast, proceeded to the GAAP income statement and the balance sheet only.
Regardless of the type of earnings release, both the income statements and the
balance sheet all spanned three successive years. Differing as regards their
content, the GAAP report ran to three pages and the pro forma report to four.

4.4. Experimental Design and Task


The type of earnings release represented the independent variable. Thus, this
between-groups measure had two levels: GAAP report and pro forma report.
The study material was followed by a two-page questionnaire prompting the
participants to make an individual assessment of the financial status of the
company. First, participants were asked to predict EPS according to GAAP for
the current year, indicating their predictions on an 11-point scale ranging from
250 SEK to 50 SEK with a midpoint denoting zero earnings. Note that this
approach to measuring the effects on judgment of pro forma information
differs from the approach adopted in earlier studies on the subject (Frederickson
and Miller, 2004; Elliott, 2006). In our view, asking participants explicitly to
290 P. Andersson & N. Hellman

Table 1. Summary of the contents of the study material: GAAP and pro forma earnings
releases
GAAP earnings release Pro forma earnings release

Page 1: Introductory summary Page 1: Introductory summary


†Sales fell by 11% to 53,358 (59,746) †Sales (pro forma) fell by 16% to
million. 49,342 (58,640) million.
† Net income was 26,892 million, †Net income (pro forma) was 3,258
which was a strong improvement million, which was a strong
compared over previous year improvement compared to the
(211,022 million). previous year (22,102 million).
†Earnings per share were 222.46, †Earnings per share (pro forma)
compared to 234.50 in the previous increased to 10.47, compared to
year. 26.11 in the previous year.
†The board suggests dividend (3.00); †The board suggests dividend (3.00);
unchanged from the previous year. unchanged from the previous year.

Page 2: GAAP consolidated income Page 2: Pro forma income statement


statement
†Gross margin: 17,482 million. †Gross margin: 16,578 million.
†Operating income: 23,626 million. †Operating income: 4,014 million.
†Income before financial items: †Income before financial items: 3,258
26,892 million. million.
†Net income: 24,970 million. †Net income: 2,317 million.

Page 3: GAAP consolidated balance Page 3: GAAP consolidated income


sheet statement
†Total assets: 62,186 million. †See column to the left.
†Stockholders’ equity: 22,478 million.
†Long-term liabilities: 19,446 million. Page 4: GAAP consolidated balance sheet
†Current liabilities: 12,379 million. †See column to the left.

Note: The two earnings releases were written in Swedish and all amounts in the two reports were given
in Swedish kronor (SEK). On 18 February 2007, 100 SEK corresponded to about $14.22 or E10.82.

predict EPS guarantees greater external validity, since it avoids some of the pro-
blems associated with other approaches (see Sections 2.3 and 2.4). Second, after
noting their responses, the participants indicated their level of confidence on a 6-
point scale ranging from ‘very unsure’ to ‘very sure’. There were then seven
empty lines where the participants could comment on their predictions.
Largely, the purpose of these lines was to check the extent to which the subjects
had observed and interpreted the content of the independent variable (cf. Libby
et al., 2002). Third, the participants were asked to evaluate the degree of
realism in the material and whether the information given was sufficient as a
basis for economic forecasting. These questions represented an attempt to
measure the external validity of the study material. Finally, there were questions
about demographics, the motivation for participating in the study and the time
devoted to the experiment.10
Does Pro Forma Reporting Bias Analyst Forecasts? 291

Before running the experiment, the study material was pre-tested on four Ph.D.
students specializing in accounting and financial analysis. On the whole, they felt
that the instructions were clear and that the contents of the study material were
realistic.

4.5. Procedure
The two types of earnings release were randomly distributed among the partici-
pants at a lecture on advanced finance. The participants were instructed to make
an individual examination of the study material, and then to answer a question-
naire. As an incentive they were each given the reward of their choice: one
cinema ticket or three lottery tickets worth approximately $10 (64% chose the
lottery tickets), if their completed questionnaire were drawn.
A total of 16 participants read the GAAP report, while 20 studied the pro forma
report. The majority of the participants (61%) said that they felt motivated to
undertake the task. On average, it took about 19 minutes to complete the task
and to answer the questionnaire. Despite exposure to different parts of the
material, the two groups did not differ significantly when it came to the (self-
reported) time devoted to the task.

5. Results
For the participant groups as a whole the average EPS forecast was 25.83 SEK
(SD ¼ 12.96). Table 2 shows clearly that the two types of earnings releases
yielded substantially different EPS forecasts. Participants examining the GAAP
report predicted a lower EPS (M ¼ 213.13 SEK) than those who were given
the combined report of pro forma and GAAP-based information (M ¼ 0.00
SEK). Note that the forecasts submitted by this second participant group were
more widely dispersed and covered a greater range (40) than those made by

Table 2. Descriptive statistics regarding EPS forecasts and the confidence in such forecasts
Type of earnings release
GAAP Pro forma
information and GAAP information
(n ¼ 16) (n ¼ 20)

Mean EPS forecast (SD) 213.13 (7.04) 0.00 (13.76)


Median EPS forecast (Min: Max) 210 (220: 0) 10 (230: 10)
Mean confidence in EPS forecast 2.25 (1.24) 2.70 (1.26)
(SD)
Median confidence in EPS forecast 2 (1: 5) 2 (1: 6)
(Min: Max)

Note: The EPS forecast was measured on an 11-point scale ranging from ‘ 2 50 SEK’ (1) to ‘50 SEK’
(11), while confidence was rated on a 6-point scale ranging from ‘very unsure’ to ‘very sure’.
292 P. Andersson & N. Hellman

the group receiving GAAP-based information only. A statistical test (Levene’s


test for equality of variances) revealed that the dispersion differed significantly
between the groups (F ¼ 12.22, p , 0.001). The measure of EPS forecast was
recoded as a variable, which had 11 points reflecting the 11 possible forecasts
(i.e. 250, 240, 230, etc.). This variable was then used in the statistical analyses.
Independent samples t-test (based on the assumption of unequal variance)
showed that the mean difference of 21.31 (SD error ¼ 0.35) was significant
(t(29.4) ¼ 23.70, p , 0.001, Cohen’s d ¼ 21.24). A non-parametric test (i.e.
Mann – Whitney U) confirmed this result. There was thus support for the hypoth-
esis that earnings releases including both pro forma and GAAP-based infor-
mation lead to higher forecasts of EPS than earnings releases involving
GAAP-based information only.
The participants also rated the confidence they felt in their own EPS forecasts.
Across the participant groups as a whole the mean level of such confidence was
2.50 (SD ¼ 1.25), which meant that the average participant felt ‘rather uncertain’
about his or her forecast. As Table 2 shows, those who were given the GAAP-
based report were slightly less confident than those given the combined pro
forma and GAAP report (mean ¼ 2.25 vs. 2.70). Independent samples t-test
showed that the mean difference of 20.45 (SD error ¼ 21.30) was not signifi-
cant (t(34) ¼ 21.07, n.s., Cohen’s d ¼ 20.36). Non-parametric tests (i.e.
Mann – Whitney U) also confirmed that there was no significant difference
among participant groups as regards confidence. Non-parametric correlation ana-
lyses were also run for each participant group, and these showed that confidence
was unrelated to EPS forecast (rs ¼ 20.04 and 20.08, n.s.).
It should be remembered that the participants were able to comment on and
motivate their EPS forecasts. This information would indirectly reflect the way
they had interpreted the study material. In all, nine of the participants who had
read the GAAP report and seven of those who had read the pro forma report
chose to take up the offer. The former group seemed concerned about three
issues: (1) the shortage of information, (2) the management of costs, and (3) the
assumption of market development. Comments from the group looking at pro
forma earnings were difficult to classify since they took up so many different
issues. One participant for instance, talked about the expected market development,
while another discussed the possibility of the company repurchasing its own stock.
And, interestingly, only one participant wrote that he knew little about pro forma
reporting, which perhaps suggests that most (at least of the participants receiving
the pro forma report) were familiar to some extent with this phenomenon.
It could perhaps be expected that financial analysts of varying experience
would react differently to the two types of financial report. To test this assump-
tion, the sample was divided into two groups: 21 less experienced participants
(with less than the 4.8-year median) and 15 of their more experienced colleagues
(with more than the 4.8-year median). These two groups then allowed compari-
sons to be made within the two experimental groups, that is, those confronted by
GAAP reports as against pro forma reports. The nine participants with less
Does Pro Forma Reporting Bias Analyst Forecasts? 293

experience and the seven with more experience who all received the GAAP
report predicted more or less the same level of EPS (M ¼ 212.22 as against
214.29). Similarly, there was no substantial difference between the predictions
of the 12 less experienced and the 8 more experienced participants who were
given the report combining both pro forma and GAAP-based information
(M ¼ 23.33 as against 5.00). The application of non-parametric tests (i.e.
Mann – Whitney U), which was motivated by the limited numbers in the two
length-of-experience subgroups revealed non-significant differences only.11
In order to make a simultaneous evaluation of the effects of the experimental
manipulation (GAAP as against pro forma) and level of experience (less vs.
more), a multiple regression model was tested with EPS forecast as the dependent
variable (MS ¼ 8.24, F(2) ¼ 6.43, p , 0.001). The model had a significant pre-
dictor, which was the experimental manipulation (standardized beta ¼ 0.52,
t ¼ 3.49, p , 0.001) and a non-significant predictor, which was the level of
experience (standardized beta ¼ 0.14, t ¼ 0.95, n.s.). Accounting for about
24% of the variance concerning the EPS forecasts, the regression model gave
further support for the hypothesis.
Finally, after they had made their forecasts, the participants were asked to
comment on the study material. Of those who received the GAAP report, one par-
ticipant considered the material to be unrealistic, while the equivalent figure for
the pro forma group was five. Eight in the former group and 11 in the latter con-
sidered the material to be realistic. The rest of the participants said they were
doubtful about the realism of the study material. At the same time, many partici-
pants (50 and 65%, respectively) wanted additional information and found the
material as it stood to be insufficient for economic forecasting. Overall, the per-
ceived lack of unrealism and the demand for additional information seem reason-
able, given the need for simplification in experimental settings. Nevertheless, the
response to the debriefing questions did not suggest that the study material was
perceived as incorrect or misleading. Moreover, statistical tests (i.e. t-tests,
Mann – Whitney U) showed that the responses to the debriefing questions were
unrelated to the EPS forecasts.

6. Discussion and Conclusions


The purpose of this study was to investigate the impact of pro forma reporting on
the judgment of financial analysts, and particularly on their earnings forecasts. In
line with our hypothesis, the participants who examined the pro forma earnings
release predicted EPS (according to GAAP) to be significantly higher than those
who received the report based on GAAP information only. This result has been
interpreted here as a consequence of the more positive framing and the higher
anchor level generated by the pro forma report compared to that generated by
the GAAP report. This result differs from the findings in Frederickson and
Miller (2004) and Elliott (2006), where financial analysts reached similar judg-
ments regardless of the type of earnings release. We believe that this difference
294 P. Andersson & N. Hellman

in results can be explained primarily by the effects of framing and anchoring,


which have been emphasized in the present study. In our experiment, there
was a significant profit in the pro forma statement (10.47 SEK or $1.37 per
share) and a significant loss in the GAAP statement (222.46 SEK or $ 2 2.94
per share). In Frederickson and Miller (2004), the comparable figures were
$1.68 (pro forma profit) as against $1.24 (GAAP profit) and in Elliott (2006)
the comparable figures were $0.01 (pro forma profit) as against $ 2 1.02
(GAAP loss). Further, the difference in results may be due to the use of a
more valid dependent variable (earnings forecasts) and greater reliability (mon-
itored experiment).
Regarding the question of possible effects on analysts’ confidence in their fore-
casts, we found that when comparing with the perceived confidence in the GAAP
report, the pro forma report did not diminish or increase the participants’ confi-
dence in their own forecasts, even though it contained more – albeit contradic-
tory – information. This observation is consistent with the findings of earlier
research (cf. Maines, 1995). One might speculate that the greater volume of infor-
mation in the pro forma report raised the level of confidence, but that this effect
was cancelled out by the contradictory details in the report concerned. Another
interpretation could be that both participant groups were uncertain about their
forecasts, because of the lack of additional information.
As in all research, the present study has its limitations. Admittedly, the number
of participants could have been greater, but it is in fact much the same as in
Frederickson and Miller (2004). It should also be emphasized that recruiting pro-
fessionals to take part in scientific studies is generally difficult (cf. Libby et al.,
2002). Attempts were in fact made to re-run the present experiment with another
group of financial analysts, but it was unfortunately impossible to get access to
more subjects. Another potential limitation concerns the study material. First,
the material could have been more detailed and/or more extensive, and thus
also more authentic. Secondly, the participants could have assessed more than
just one company thus enabling stronger results. On the other hand, such an
approach would have needed more time and might have interfered with the finan-
cial analysts’ normal work. Finally, a within-subject research design might have
been more powerful than the employed between-subjects setting. With such a
design, participants would evaluate a single company, making forecasts over
several periods with or without pro forma earnings reporting.12 It must be
stressed, however, that the setting adopted here is in line with the procedures
common in behavioural accounting studies (e.g. Libby et al., 2002), and
resembles that adopted in prior studies on pro forma reporting (Frederickson
and Miller, 2004; Elliott, 2006).
Standard setters put considerable effort into developing ‘better’ financial
reporting standards, assuming that capital market participants will benefit.13
Over time, this effort has made accounting standards increasingly complex.
Pro forma reporting, which excludes some of the complex GAAP-required
items, offers a simplified way of communicating the company’s ‘core earnings’.
Does Pro Forma Reporting Bias Analyst Forecasts? 295

Earlier research on analysts’ forecast revisions indicates that analysts find pro
forma earnings to be significantly more informative, providing a more stable
measure of future profitability than GAAP earnings (Bhattacharya et al., 2003,
p. 310). However, if pro forma numbers draw attention away from GAAP in a
way that affects users’ judgments, this may have adverse effects for some user
groups. Earlier studies have reported that the judgments of non-sophisticated
users tend to be affected by pro forma reporting, whereas sophisticated users’
judgments remain unaffected (Frederickson and Miller, 2004; Elliott, 2006;
Allee et al., 2007). In contrast, our results show that sophisticated users will
also be affected when there is a substantial difference between GAAP earnings
and pro forma earnings. This should make standard setters concerned. Analysts
have a legitimate demand for financial statements that constitute a good basis
for forecasting. However, the increasingly complex financial statements pro-
duced under IFRS and US GAAP have been unable to satisfy this demand and
therefore analysts seek more forecasting-relevant, pro forma accounting infor-
mation. This observation implies a questioning of the standards as such.
Further, our study shows that such alternative (non-GAAP) measures may bias
analysts’ judgment. If pro forma accounting becomes the dominant source of
information for analysts, they may gradually lose their trust in accounting,
since earlier research indicates that pro forma accounting numbers are not com-
parable across companies and not consistently reported over time (Doyle et al.,
2003; Bhattacharya et al., 2004).
We believe that there is room for more experimental research on pro forma
reporting. The results reported in Elliott (2006) regarding the role of quantitative
reconciliation between pro forma and GAAP needs to be examined further.
In view of the results of the present paper, it would be interesting to include
reconciliation in a setting where the discrepancy between pro forma and
GAAP earnings is substantial, involving a pro forma profit and a GAAP loss.
In addition, the results reported in Bhattacharya et al. (2004), namely, that pro
forma reporting appears to be manipulated by companies to meet analysts’ expec-
tations, hold interesting possibilities for future experimental research.

Acknowledgements
Both authors contributed equally to this work, the order of authorship is arbitrary.
Helpful and valuable comments from Salvador Carmona and two anonymous
reviewers are gratefully acknowledged. The first author wishes to thank the
research foundation of Jan Wallander and Tom Hedelius for financial support.

Notes
1
Ordinary net profit, calculated according to Generally Accepted Accounting Principles
(GAAP), is adjusted for items that are in some sense ‘unusual’. This could include restructuring
charges, write-downs and impairments, research and development expenses, merger and
296 P. Andersson & N. Hellman

acquisition costs, stock compensation expense, goodwill amortization, etc. (Bradshaw and
Sloan, 2002).
2
Bradshaw and Sloan (2002) reported this empirical result on the basis of a comparison between
GAAP earnings and pro forma earnings as registered by analyst tracking services. Thus, they do
not use the companies’ pro forma earnings measures. However, the analyst tracking services
base their pro forma measures on companies’ specifications of what is referred to as non-
recurring or non-cash expenses.
3
Quotations from the speech are included in Bhattacharya et al. (2003, pp. 286–288).
4
These results reported in Frederickson and Miller (2004) and Elliott (2006) also agree with
those of a recent study based on archival data by Allee et al. (2007).
5
According to the descriptive statistics in Elliott (2006), analysts who received the press release
emphasizing pro forma earnings gave slightly higher earnings performance ratings than ana-
lysts who received GAAP information only (31.94 compared to 28.82 on the 100-point
scale). The pro forma earnings press release with reconciliation caused analysts to give signifi-
cantly higher earnings performance rankings compared to the press release including GAAP
information only.
6
Elliott (2006) reports a case in which the GAAP loss of $1.02 is transformed into a pro forma
profit close to zero ($0.01).
7
Analysts’ stock price appraisals could be expected to be preceded by analyst forecasts of a par-
ticular valuation attribute, such as earnings. Furthermore, the appraisal will involve subjective
risk considerations. In the case of investment decisions, analysts do not typically make such
decisions. The analysts’ earnings performance ranking (Elliott, 2006) is likely to be the depen-
dent variable that comes closest to the analyst’s primary real-world task of making earnings
forecasts.
8
In a study based on archival data, Doyle et al. (2003, p. 159) reported that some of the items
excluded from pro forma reports (denoted ‘special items’) are unrelated to future cash flows
from operations, whereas other excluded items (denoted ‘other exclusions’) have a pronounced
negative relation with future cash flows from operations. It is possible that analysts use the
reconciliation to learn more about the type of items affecting the pro forma reporting. If
there are a lot of ‘special items’, this might be a reason for viewing the company more posi-
tively than if the items were part of the group of ‘other exclusions’.
9
The complete research instrument can be obtained from the corresponding author.
10
There were also additional questions prompting the participants to make subjective probability
forecasts with respect to the likelihood of higher sales, lower costs, improved income or a stron-
ger financial position, as well as indicating their confidence in the forecasts concerned.
11
As suggested by one of our reviewers, an alternative procedure would involve splitting the
sample into three subgroups so that the participating analysts were categorized as limited, mod-
erately or highly experienced (n ¼ 19, 13, 16). Such a procedure was also adopted. A Kruskal–
Wallis test showed that the subgroups did not show significantly different reactions to the types
of financial report. Nor were the differences between the limited and highly experienced ana-
lysts significant.
12
We thank an anonymous referee for suggesting this design.
13
We thank an anonymous reviewer for helpful comments on the implications of our results for
standard setters.

References
Allee, M. J., Bhattacharya, N., Black, E. L. and Christensen, T. E. (2007) Pro forma disclosure and
investor sophistication: external validation of experimental evidence using archival data,
Accounting, Organizations and Society, 32(3), pp. 201 –222.
Amir, E. and Ganzach, Y. (1998) Overreaction and underreaction in analysts’ forecasts, Journal of
Economic Behavior & Organization, 37(3), pp. 333–347.
Does Pro Forma Reporting Bias Analyst Forecasts? 297

Andersson, P. (2005) How well do financial experts perform? A review of empirical research on per-
formance of analysts, day-traders, forecasters, fund managers, investors, and stockbrokers, The
ICFAI Journal of Behavioral Finance, 2, pp. 44 –61.
Ayton, P. (1998) How bad is human judgment?, in: G. Wright and P. Goodwin (Eds) Forecasting with
Judgment, pp. 237–267 (Chichester: Wiley).
Bhattacharya, N., Black, E. L., Christensen, T. E. and Larson, C. R. (2003) Assessing the relative
informativeness and permanence of pro forma earnings and GAAP operating earnings,
Journal of Accounting and Economics, 36(1–3), pp. 285 –319.
Bhattacharya, N., Black, E. L., Christensen, T. E. and Mergenthaler, R. D. (2004) Empirical evidence
on recent trends in pro forma reporting, Accounting Horizons, 18(1), pp. 27 –43.
Bouwman, M. J., Frishkoff, P. A. and Frishkoff, P. (1987) How do financial analysts make decisions?
A process model of the investment screening decision, Accounting, Organizations and Society,
12(1), pp. 1 –29.
Bowen, R. M., Davis, A. K. and Matsumoto, D. K. (2005) Emphasis on pro forma versus GAAP earn-
ings in quarterly press releases: determinants, SEC intervention, and market reactions, The
Accounting Review, 80(4), pp. 1011–1038.
Bradshaw, M. T. and Sloan, R. G. (2002) GAAP versus the Street: an empirical assessment of two
alternative definitions of earnings, Journal of Accounting Research, 40(1), pp. 41 –66.
Camerer, C. F. (2000) Prospect theory in the wild, in: D. Kahneman and A. Tversky (Eds) Choices,
Values, and Frames, pp. 288–300 (New York: Cambridge University Press).
Camerer, C. F. and Johnson, E. J. (1991) The process–performance paradox in expert judgment: how
can experts know so much and predict so badly?, in: K. A. Ericsson and J. Smith (Eds) Toward a
General Theory of Expertise: Prospects and Limits, pp. 195–217 (New York: Cambridge
University Press).
Chapman, G. B. and Johnson, E. J. (2002) Incorporating the irrelevant: anchors in judgments of
belief and value, in: T. Gilovich, D. Griffin and D. Kahneman (Eds) Heuristics and
Biases: The Psychology of Intuitive Judgment, pp. 120–138 (New York: Cambridge University
Press).
Doyle, J. T., Lundholm, R. J. and Soliman, M. T. (2003) The predictive value of expenses excluded
from pro forma earnings, Review of Accounting Studies, 8(2/3), pp. 145–174.
Elliott, W. B. (2006) Are investors influenced by pro forma emphasis and reconciliations in earnings
announcements?, The Accounting Review, 81(1), pp. 113 –133.
Ericsson, K. A. and Lehmann, A. C. (1996) Expert and exceptional performance: evidence of maximal
adaptation to task constraints, Annual Review of Psychology, 47, pp. 273 –305.
Ernst & Young (2006) IFRS: Observations on the Implementation of IFRS. Available
at: http://www.ey.com/Global/download.nsf/International/IFRS_Implementation_of_IFRS_
Master/$file/EY_IFRS_ImplementationOfIFRS_Master.pdf
Financial Accounting Standards Board (FASB) (2002) Reporting Information about the Financial
Performance Reporting by Business Enterprises: Focusing on the Form and Content of Financial
Statements. Available at: http://www.fasb.org/proposals/performance.pdf
Frederickson, J. R. and Miller, J. S. (2004) The effects of pro forma earnings disclosures on analysts’
and nonprofessional investors’ equity valuation judgments, The Accounting Review, 79(3),
pp. 667–686.
Hunton, J. E. and McEwen, R. A. (1997) An assessment of the relation between analysts’ earnings
forecast accuracy, motivational incentives and cognitive information search strategy, The
Accounting Review, 72(4), pp. 497 –515.
Johnson, W. B. and Schwartz, W. C. (2001) Are investors misled by pro forma earnings?, Working
Paper, University of Iowa. Available at: www.ssrn.com
Kahneman, D. and Tversky, A. (1979) Prospect theory: an analysis of decision under risk, Econome-
trica, 47(2), pp. 263 –291.
Kahneman, D. and Tversky, A. (1984) Choices, values and frames, American Psychologist, 39,
pp. 341–50.
298 P. Andersson & N. Hellman

Libby, R., Bloomfield, R. and Nelson, M. W. (2002) Experimental research in financial accounting,
Accounting, Organizations and Society, 27(8), pp. 775 –810.
Lovallo, D. and Kahneman, D. (2003) Delusion of success – how optimism undermines executives’
decisions, Harvard Business Review, 81(7), pp. 56–63.
Maines, L. A. (1995) Judgment and decision-making research in financial accounting: a review and
analysis, in: R. H. Ashton and A. H. Ashton (Eds) Judgment and Decision-Making in Accounting
and Auditing, pp. 76–101 (New York: Cambridge University Press).
Olbert, L. (1994) Stock valuation methods of financial analysts in a thin stock market in Sweden, with
comparisons to the United Kingdom and the United States, International Journal of Accounting,
29(2), pp. 123 –135.
Oskamp, S. (1982) Overconfidence in case-study judgments, in: D. Kahneman, P. Slovic and
A. Tversky (Eds) Judgment under Uncertainty: Heuristics and Biases, pp. 287–293
(New York: Cambridge University Press).
Pitt, H. (2001) Remarks before the AICPA governing council, Miami Beach, FL, 22 October.
Securities and Exchange Commission (SEC) (2003) Final Rule: Conditions for Non-GAAP Financial
Measures, Release No. 33– 8176, 34– 47226, FR–65.
Shanteau, J. (1989) Cognitive heuristics and biases in behavioral auditing: review, comments and
observations, Accounting, Organizations and Society, 14(1/2), pp. 165–177.
Smith, J. and Kida, T. (1991) Heuristics and biases: expertise and task realism in auditing, Psychologi-
cal Bulletin, 109(3), pp. 472–489.
Soman, D. (2004) Framing, loss aversion, and mental accounting, in: D. J. Koehler and N. Harvey
(Eds) Blackwell Handbook of Judgment and Decision Making, pp. 379 –398 (Oxford:
Blackwell).
Stockholmsbörsen (2003) Börsregler 2003/2004 – Handledning till noteringsavtal och noteringskrav
[Stock Exchange Rules 2003/2004 – Manual to the Listing Contract and Listing Requirements],
Stockholmsbörsens skriftserie No. 2.
Tetlock, P. E. (2005) Expert Political Judgment: How Good Is It? How Can We Know? (Princeton,
NJ: Princeton University Press).
Trueman, B. (1994) Analysts’ forecasts and herding behavior, Review of Financial Studies, 7(1),
pp. 97 –124.
Tversky, A. and Kahneman, D. (1974) Judgment under uncertainty: heuristics and biases, Science,
185, pp. 1124–1131.
Tversky, A. and Kahneman, D. (1981) The framing of decision and the psychology of choice, Science,
211, pp. 453–458.
Whyte, G. and Sebenius, J. K. (1997) The effect of multiple anchors on anchoring in individual and
group judgment, Organizational Behavior and Human Decision Processes, 69(1), pp. 75– 85.

You might also like