Analysts' Forecast Accuracy in Germany: The Effect of Different
Accounting Principles and Changes of Accounting Principles
Jürgen Ernstberger, Simon Krotter, Christian Stadler
Accounting | Finance | Management | Marketing | Operations and Information Systems
V o l u m e 1 | I s s u e 1 | M a y 2 0 0 8 | w w w . b u s i n e s s - r e s e a r c h . o r g

---

BuR - Business Research
Official Open Access Journal of VHB
Verband der Hochschullehrer für Betriebswirtschaft e.V.
Volume 1 | Issue 1 | May 2008 | 26-53
Analysts’ Forecast Accuracy in Germany:
The Effect of Different Accounting Principles
and Changes of Accounting Principles
Jürgen Ernstberger, Faculty of Business, Economics and Information Systems, University of Regensburg,
E-mail: juergen.ernstberger@wiwi.uni-regensburg.de
Simon Krotter, Corporate Finance Department, Siemens AG, München, E-mail: simon.krotter@siemens.com
Christian Stadler, School of Management, Royal Holloway, University of London, E-mail: c.stadler@rhul.ac.uk
Abstract
This paper assesses the influence of an adoption of IAS/IFRS or US GAAP on the financial analysts’ fore-
cast accuracy in a homogenous institutional framework. Our findings suggest that the forecast accuracy is
higher for estimates based on IFRS or US GAAP data than for forecasts based on German GAAP data.
Moreover, in the year of switching from German GAAP to US GAAP the forecast accuracy is lower than in
other years. The paper contributes to prior research by providing evidence about the usefulness of interna-
tional accounting data and about the adoption effects of a change to such accounting principles.

Keywords: accounting, adoption effect, analysts, analysts’ forecast accuracy, financial analysts, German
GAAP, Germany, HGB, IAS, IFRS, IFRS adoption, learning effect, US GAAP
1. Introduction
reasons: First, many publicly traded German com-
We investigate the impact of different accounting
panies have successively switched to internationally
principles and of a change of accounting principles
accepted accounting principles (IAAP), i.e. IAS/
on financial analysts’ earnings forecast accuracy. IFRS or US GAAP, before 2005. Therefore, the
Our results provide evidence that forecast accuracy
impact of the adoption of new accounting principles
is higher for estimates based on data prepared un-
could be examined while controlling for macroeco-
der internationally accepted accounting principles
nomic and other variables which are subject to
(IAAP), i.e. International Accounting Standards change over time. Second, the national German
(IAS)/International Financial Reporting Standards
GAAP is significantly different to IAS/IFRS and US
(IFRS)1 or United States Generally Accepted Ac-
GAAP which makes the impact of the accounting
counting Principles (US GAAP), than for estimates
principles on the analysts’ accuracy more obvious.
based on German GAAP (also called “Handelsge-
The motivation for this study is the adoption proc-
setzbuch (HGB)”) data. Moreover, in years of the
ess of IFRS in Europe. Since 2005 almost all pub-
adoption of new accounting principles the forecast
licly traded European companies have been re-
accuracy is lower than in other years for companies
quired by the IAS regulation (1606/2002/EC) to
switching from HGB to US GAAP. Germany pro-
prepare consolidated accounts under IFRS.2 For this
vides a unique framework for these analyses for two
transition process we try to make some inferences

from the adoption process in Germany where many
1 The International Financial Reporting Standards (IFRS)
were initially called International Accounting Standards (IAS).
In 2001, they changed name to International Financial Re-
2 Companies, which are publicly traded both in the European
porting Standards (IFRS). We use the term “IAS” when we
Union and on a regulated third-country market and which are
refer to periods before 2001, a combination of both terms
therefore applying other IAAP (especially US GAAP) in their
(“IAS/IFRS”), when we refer to a time period including years
consolidated accounts, are allowed to defer the application of
before 2001 and the term “IFRS” when we refer to more recent
IFRS until 2007. This also holds for companies which only
periods.
have publicly traded debt securities.
26

---

BuR - Business Research
Official Open Access Journal of VHB
Verband der Hochschullehrer für Betriebswirtschaft e.V.
Volume 1 | Issue 1 | May 2008 | 26-53
companies switched to IAS/IFRS or US GAAP be-
2. Literature review
fore 2005.
Many studies have analyzed the accuracy of ana-
2.1 Drivers of analyst forecast accuracy
lysts’ forecasts and its determinants. Most of the
In the last few years, there has been an explosion of
prior research has been conducted for the US (e.g.
research examining analyst forecasts. One can iden-
Lys and Soo 1995; Mikhail, Walther, and Willis tify interacting analyst-specific and firm-specific
1997; Alford and Berger 1999; Clement 1999; Jacob,
factors that drive analyst accuracy (see Figure 1).
Lys, and Neale 1999; Duru and Reeb 2002; Irvine
Concerning the individual analyst, it is known
2004; Gu and Wang 2005; Lin and Yang 2006) or
amongst other things that he or she tends to be
the UK (e.g. Acker, Horton, and Tonks 2002). We
rather optimistic in the way that his or her forecasts
analyze the German market similar to Capstaff, are systematically upward biased (e.g. Easterwood
Paudyal, and Rees (1998), Wallmeier (2005), Daske
and Nutt 1999) and are revised rather gradually (e.g.
(2005), and Bessler and Stanzel (2007). However,
Bartov, Givoly, and Hayn 2002). In addition, ana-
only the study by Daske (2005) examining the pe-
lysts certainly may exhibit different skills which may
riod 1993–2002 explicitly controls for the type of
emanate from their experience, workload, or risk
accounting principles applied. Similarly, our study
tolerance. On the other side, a firm’s characteristics
considers the impact of differences in the account-
also drive analyst forecast accuracy, e.g. company
ing regimes applied, but also delivers evidence on
size, industry or country the company operates in,
analysts’ forecast accuracy for a more recent time
or its regulatory environment (e.g. Das and Sauda-
period, i.e. the years 1998–2004.
garan 1998; Higgins 1998). Furthermore, manage-
Our study contributes to prior research as it consid-
ment actions may influence forecast accuracy di-
ers accounting principles and accounting principle
rectly or indirectly. On the one hand, several studies
changes as control variables. Thus, we are able to
indicate the management of earnings and the guid-
document in contrast to Daske (2005) that the im-
ance of forecasts towards the consensus (Bannister
plication of an international accounting regime and Newman 1996; Degeorge, Patel, and Zeck-
(IAS/IFRS and US GAAP) is associated with a
hauser 1999; Matsumoto 2002; Bartov, Givoly, and
higher forecast accuracy of financial analysts. Since
Hayn 2002; Abarbanell and Lehavy 2003; Hutton
1998 German companies have been allowed by law
2005; Burgstahler and Eames 2006; for Germany
to choose to prepare their consolidated accounts in
see Bessler and Stanzel 2007). On the other hand,
accordance with national GAAP, IAS/IFRS, or US
through issuing its own forecasts, management tries
GAAP. Therefore, Germany provides a unique to influence analysts’ expectations (e.g. Williams
framework for a comparative analysis within a ho-
1996; Bamber and Cheon 1998; Lennox and Park
mogeneous institutional background. This allows us
2006). Thus, both the management of earnings and
to effectively control for institutional factors (e.g.
expectations jointly drive the analysts’ consensus
regulatory requirements and the enforcement sys-
gradually down to beatable analyst forecasts, which
tem) which prove to be important determinants of
may favor equity issuances or insider trading (e.g.
analysts’ forecast quality (Hope 2003a; Hope Richardson, Teoh, and Wysocki 2004).
2003b; Hope 2004; Barniv, Myring, and Thomas
2005).
Figure 1: Drivers of analyst forecast
Our paper is arranged as follows: Section 2 gives an
accuracy
overview of related previous studies. Section 3 states
the research hypotheses. Section 4 explains the
Dr
D iv
r ers
iv
of
o
variables and models used in the analyses. Section 5
an
a aly
a s
ly t f
t oreca
o
s
reca t ac
t
cu
ac rac
ra y
c
gives an overview of the sample selection process
and of the data sources. Sections 6 and 7 present
and discuss the descriptive and regression results.
x An
A al
n ys
al t
ys -
t sp
s ecific
ec
x
Firm
r -sp
m
e
-sp cific
Section 8 provides the results of sensitivity analyses
Behavior
Fi
F rm
r characteris
r tics
and Section 9 concludes.
Skilll
Fi
F rm
r actio
a
n
ctio

27

---

BuR - Business Research
Official Open Access Journal of VHB
Verband der Hochschullehrer für Betriebswirtschaft e.V.
Volume 1 | Issue 1 | May 2008 | 26-53
2.2 Analyst forecast accuracy and
(1999) document a positive association between
accounting data
forecast accuracy and the quality of disclosures in
This study deals with the actions of management
the Management Discussion and Analysis (MD&A)
that relate to the firm’s accounting practices and
and especially certain forward-looking information
disclosure policy. Our work combines two strands of
in the MD&A.
research: studies on the impact of accounting data
Besides disclosures, specific properties of account-
and studies on the impact of the adoption of IAAP.
ing standards and their impact on forecast accuracy
Financial analysts are frequently regarded as so-
have been examined. In a cross-country study Basu,
phisticated processors of financial information and
Hwang, and Jan (1998) find that forecast accuracy
often taken as representatives of the market is lower in countries with accounting regimes hav-
(Revsine, Collins, and Johnson 2001). Furthermore,
ing the following properties: higher relative use of
evidence suggests that financial statements are an
current value accounting (with revaluations passing
important source of information for analysts when
through the income statement), less relative use of
determining their forecasts (Acker, Horton, and accruals, and less choice between accounting meth-
Tonks 2002; Peek 2005).
ods. However, for a sample of 18 countries Hope
The influences of (changes in) accounting standards
(2004) finds contrary results. Controlling for varia-
and disclosures on the forecast accuracy have been
tions in the enforcement system between different
widely examined. For example, Lang and Lundholm
countries and for variations in disclosure levels
(1996) show that a disclosure score for US compa-
between different firms, he shows that the relative
nies in the period 1985–1989 is positively associated
extent of choice is negatively associated and the
with the number of analysts following (i.e. coverage)
relative extent of accrual accounting is positively
as well as forecast accuracy, and is negatively asso-
associated with forecast accuracy. The impact of
ciated with forecast dispersion as well as variability
conservatism on the forecast accuracy of financial
of forecast revisions. However, the category of in-
analysts is investigated by Mensah, Song, and Ho
vestor relations, in particular, shows significant (2004). They analyzed three measures of conserva-
relationships, whereas annual financial statements
tism and find that a higher level of conservatism
as well as other annual published information are
leads to higher forecast errors of financial analysts.
mostly not significant. The findings of Lang and
Ashbaugh and Pincus (2001) compare different
Lundholm (1996) are confirmed in an international
accounting principles with reference to the forecast
study by Hope (2003a). Analyzing 1,309 firm-year
accuracy of financial analysts and set IAS as a
observations from 22 countries he observes that
benchmark. For a cross-country sample of 80 com-
companies’ disclosure levels (as well as the level of
panies having adopted IAS between 1990 and 1993,
enforcement) are positively associated with the they show that analysts’ forecast errors are posi-
forecast accuracy of analysts. Several other studies
tively related to the differences between various
examine the association between both the level of
domestic GAAP and IAS. Moreover, they find that
and an increase in the quality of disclosures with
forecast accuracy improves after the adoption of
forecast errors (Higgins 1998; Chang, Khanna, and
IAS. Cuijpers and Buijink (2005) investigated the
Palepu 2000; Ang and Ciccone 2001; Acker, Hor-
determinants and consequences of applying IAAP
ton, and Tonks 2002; Hope 2003b; Vanstraelen,
for 114 companies from 12 European countries (in-
Zarzeski, and Robb 2003; Hope 2004) and find
cluding Germany) in 1999. They show that IAAP
similar results for various countries.
application has a positive impact on analysts follow-
In particular the predictive value of disclosures ing. However, they also document an increased
seems to have a positive impact on properties of
analyst forecast dispersion effect for companies
analysts’ earnings forecasts. For example, the in-
applying IAAP. A similar study was conducted by
formation provided in segment reports which aims
Daske (2005) for the period between 1993 and 2002
at helping users of financial statements to evaluate
in Germany. He finds a lower accuracy and higher
the present and future performance of a company is
dispersion, but no significant difference in the vola-
found to enhance analysts’ earnings forecasts tility of analysts’ earnings forecasts based on
(Baldwin 1984; Swaminathan 1991; Hussain 1998;
IAS/IFRS or US GAAP in comparison to those
Lobo, Kwon, and Ndubizu 1998; Behn, Nichols, and
based on German GAAP. However, in contrast to
Street 2002). Similarly, Barron, Kile, and O’Keefe
our study, he does not control for the complexity of
28

---

BuR - Business Research
Official Open Access Journal of VHB
Verband der Hochschullehrer für Betriebswirtschaft e.V.
Volume 1 | Issue 1 | May 2008 | 26-53
the forecasting task measured by the beta of a com-
ies about Germany examines the capital market
pany as we do. He also provides evidence that the
impacts of the IAS/IFRS adoption, the other part
level of differences between earnings under the accounting or disclosure quality of the IAS/IFRS
IAS/IFRS or US GAAP and under German GAAP as
in comparison to German GAAP. Concerning the
well as the level of guidance provided by companies
capital market impacts, Leuz and Verrecchia (2000)
concerning the transition process has impacts on
as well as Gassen and Sellhorn (2006) document
the forecast accuracy and forecast dispersion in the
lower information asymmetry, but no decrease in
period of switching to another accounting regime.
volatility after the adoption of IAS/IFRS. Compar-
Reasons for these results being different to ours
ing IAS and US GAAP adopters in Germany, Leuz
might be the different time period investigated and
(2003) shows that there seems to be no significant
the different methodology.
difference in terms of information asymmetry
Other studies examine the effect of changes of ac-
measured by bid-ask spreads and share turnover.
counting methods on forecast accuracy. For the US,
Finally, Daske (2006) was not able to document
Brown, Richardson, and Schwager (1987) find only
that the expected cost of equity capital has de-
a slight impact on forecast accuracy for accounting
creased after the adoption of IAS/IFRS and US
changes in 1976. The impact is smaller when addi-
GAAP.
tional disclosures, like pro-forma adjustments, are
Concerning the accounting or disclosure quality,
provided. Elliott and Philbrick (1990) investigated
Hung and Subramanyam (2007) find only little
accounting changes between 1976 and 1984. Their
evidence for a higher value relevance of IAS in com-
results suggest that the forecast accuracy turns out
parison to German GAAP for a sample of 80 com-
to be lower in years when accounting changes occur
panies that adopted IAS between 1998 and 2002.
without prior disclosures. The findings are con-
Moreover, they provide evidence that German
firmed for the Netherlands by Peek (2005). The
GAAP is more conservative and less fair-value ori-
forecast accuracy is lower after changes in account-
entated than IAS. Bartov, Goldberg, and Kim (2005)
ing procedures affecting earnings before extraordi-
document a higher value relevance for positive earn-
nary items. These change effects depend on the ings based on IAS and US GAAP than for those
disclosures prior to the change and the type of based on German GAAP. However, no difference is
change.
found either between the three accounting regimes
The impact of the adoption of IAS/IFRS has been
for negative earnings or between IAS and US GAAP
analyzed by several studies using different perspec-
based earnings. In addition, Gassen and Sellhorn
tives and different methods. In a cross-country (2006) show that companies applying IAS/IFRS
study including Germany, Barth, Landsman, and
have more persistent, less predictable and more
Lang (2007) document a slight decrease in the cost
conditional conservative earnings than companies
of capital after the adoption of IAS/IFRS and a
applying German GAAP. Focusing especially on
higher accounting quality of IAS/IFRS in compari-
disclosures, Daske and Gebhardt (2006) provide
son to domestic GAAP, concerning several measures
evidence that disclosure quality has increased after
such as timely loss recognition or value relevance.
the adoption of IAS/IFRS or US GAAP by German
Other studies use reconciliations of net income companies.
and/or shareholders’ equity to evaluate differences
As the review of the existing literature reveals, there
between the IAS/IFRS and other accounting princi-
is still no recent large-scale study on the comparison
ples and the impacts of these differences (e.g. Harris
between IAS/IFRS and domestic GAAP which ex-
and Muller 1999; Beckman, Brandes, and Eierle amines their impact on the analysts’ forecast accu-
2007). Most studies assess the impact of the adop-
racy within the same institutional setting. The goal
tion of IAS/IFRS in a single country. Studies have
of this study is to fill this research gap and to pro-
been conducted, e.g. for Switzerland (Auer 1999),
vide insights into the impacts of the different prop-
Finland (Kinnunen, Niskanen, and Kasanen 2000),
erties of IAS/IFRS, US GAAP, and German GAAP
and Kuwait (El Shamy and Al-Qenae 2005).
on forecast accuracy in Germany. As Gebhardt, Lee,
and Swaminathan (2001) argue, higher forecast
2.3 Impacts of the adoption of IAAP
accuracy could be associated with higher accounting
Most studies on the impact of an adoption of quality and with lower implied cost of capital.
IAS/IFRS focus on Germany. One part of the stud-
Therefore, the research method used is related to
29

---

BuR - Business Research
Official Open Access Journal of VHB
Verband der Hochschullehrer für Betriebswirtschaft e.V.
Volume 1 | Issue 1 | May 2008 | 26-53
both the accounting quality of IAS/IFRS and US
(2007). Peek (2005) argues and shows empirically
GAAP and the capital market impact of the adoption
that a higher fair value-orientation decreases fore-
of these accounting regimes. However, it has to be
cast accuracy, because earnings under historical cost
noted that high analyst forecast accuracy does not
accounting are more reliable and verifiable (e.g. Ijiri
necessarily imply high accounting quality, because
and Noel 1984; Knutson 1992) and less volatile than
analysts’ forecast accuracy is a complex attribute
under current cost accounting. However, it has to be
which is – as indicated above – shaped by a set of
considered that according to IAS/IFRS and US
potentially competing incentives and by the institu-
GAAP only a part of the fair value changes, e.g. from
tional environment.
trading financial assets, has an impact on net in-
come. The other part of changes can (e.g. revalua-
3. Hypotheses
tions of property, plant, and equipment according to
In this study, we focus on the question of how ana-
IAS/IFRS as well as revaluations of available-for-
lysts cope with the adoption of IAS/IFRS or US
sale financial assets according to IAS/IFRS until
GAAP. Thereby, we are interested in the short-term
2003), or must (e.g. revaluations of available-for-
effects a change of the accounting principles and in
sale financial assets according to US GAAP and
the long-term implications of the adoption of more
according to IFRS since 2004) be recognized di-
investor-oriented accounting principles in a country
rectly in equity. Therefore, the fair value-orientation
with traditionally rather stakeholder-oriented ac-
could provide forward-looking information to finan-
counting principles.
cial analysts without influencing reported and fore-
The first hypothesis of our study concerns the fore-
casted earnings measures.
cast accuracy of financial analysts for earnings per
Furthermore, the lower level of accrual accounting
share based on the different accounting regimes.
in German GAAP compared to US GAAP (Nobes
The increase in the quantity and quality of disclo-
and Parker 1998; Basu, Hwang, and Jan 1998; Hope
sures after the adoption of IAS/IFRS or US GAAP
2004) or IAS/IFRS supposedly influences forecast
(Daske and Gebhardt 2006) should c.p. lead to a
accuracy. Capitalization and amortization provides
higher forecast accuracy as studies on the influence
useful information about future profitability and,
of disclosures on forecast accuracy suggest (Lang
therefore, should improve forecast accuracy (Peek
and Lundholm 1996; Higgins 1998; Chang, Khanna,
2005). This would c.p. imply a higher forecast accu-
and Palepu 2000; Ang and Ciccone 2001; Acker,
racy for financial statements based on IAAP. How-
Horton, and Tonks 2002; Hope 2003a; Hope ever, a higher level of accrual accounting provides
2003b; Hope 2004). On the other hand, based on
possibilities of earnings management which might
the results by Mensah, Song, and Ho (2004) the
have a detrimental effect on the forecast accuracy.
more conditional conservative earnings of IAS/IFRS
The volatility of earnings is another important factor
vis-à-vis German GAAP found by Gassen and Sell-
in explaining the forecast accuracy of financial ana-
horn (2006) should c.p. decrease forecast accuracy
lysts (Ashbaugh and Pincus 2001; Peek 2005).
for companies switching from German GAAP to Estimating earnings with a higher volatility might
IAS/IFRS.
not allow incorporating simple models to extrapo-
Moreover, German GAAP is less fair value-oriented
late previous earnings trends. Thus, a higher volatil-
than US GAAP and IAS/IFRS. For example, Ger-
ity of earnings is likely to result in lower predictabil-
man GAAP does not allow upward revaluations of
ity of earnings which leads to a lower forecast accu-
certain types of financial assets at fair value in con-
racy and a higher forecast bias of financial analysts
trast to US GAAP and IAS/IFRS. According to (e.g., Lys and Soo 1995; Das, Levine, and Sivarama-
IAS/IFRS even property, plant, and equipment and
krishnan 1998). As the higher fair value-orientation
investment property can be measured at fair value
of IAS/IFRS and US GAAP in comparison to Ger-
(IAS16.31).3 The higher fair value-orientation of man GAAP might lead to a higher volatility of earn-
IAS/IFRS in comparison to German GAAP is em-
ings, c.p. a lower forecast accuracy could be ex-
pirically validated by Hung and Subramanyam pected for companies applying IAAP.
Finally, the lower extent of choices in US GAAP than

in IAS/IFRS or German GAAP (Nobes and Parker
3 For a complete list of the use of fair value in IFRS see Cairns
1998) could influence forecast accuracy. First, more
(2006).
choices and less discretion could be aligned with
30

---

BuR - Business Research
Official Open Access Journal of VHB
Verband der Hochschullehrer für Betriebswirtschaft e.V.
Volume 1 | Issue 1 | May 2008 | 26-53
higher forecast accuracy, because it may improve
attach to the incorporation of public and of private
the ability of companies to manage earnings to-
information into the development of their forecasts.
wards the analysts’ earnings forecasts. Several stud-
As Barron, Kim, Lim, and Stevens (1998) argue the
ies indicate such earnings management behavior
forecast error of financial analysts can be divided
(Bannister and Newman 1996; Degeorge, Patel, and
into a common component resulting from errors in
Zeckhauser 1999; Matsumoto 2002; Abarbanell and
public information and an idiosyncratic component
Lehavy 2003).
resulting from errors in private information. If the
In contrast, choices increase complexity and uncer-
adoption of IAS/IFRS or US GAAP increases the
tainty of analysts (Ashbaugh and Pincus 2001) and
errors in public information (Ashbaugh and Pincus
therefore impair forecast accuracy (Hope 2004). 2001) or makes analysts focus more on public
Furthermore, earnings management may have ob-
rather than on private information even though the
jectives other than meeting analysts’ earnings fore-
error in public information is high, forecast accuracy
casts, e.g. to report a non-negative result (Hayn
might deteriorate.
1995), to increase share price before a stock transac-
Moreover, a change in the accounting principles
tion (Dechow, Sloan, and Hutton 1996; Teoh, applied may impair the possibility to extrapolate
Welch, and Wong 1998), or to meet contractual
earnings trends as a restatement is only mandatory
provisions, such as in short-term bonus contracts
for one year prior to the adoption. This could nega-
that are tied to accounting measures (Holthausen,
tively influence the ability of analysts to forecast
Larcker, and Sloan 1995). These objectives might
future earnings (Peek 2005). However, companies
worsen the predictability of earnings for analysts as
might have adjusted several choices in their finan-
they are usually not known by company outsiders.
cial statements under the previous accounting prin-
Empirically, the degree of earnings management ciples (Daske 2005) or in the financial statements
seems to be largely the same across accounting under the newly adopted accounting principles to
principles in Germany (Van Tendeloo and Van-
smooth the adoption process. Whereas the first
straelen 2005; Goncharov 2005). Thus, the extent
smoothing strategy should be possible for all com-
of choices of different accounting principles should
panies adopting IAAP, companies switching to US
not have a significant impact on the forecast accu-
GAAP could not follow the second smoothing strat-
racy across accounting principles in Germany.
egy because of the relatively low extent of choices in
In summary, the effect of applying different ac-
US GAAP already mentioned above.
counting regimes on the forecast accuracy of finan-
Another argument for lower forecast accuracy after
cial analysts is not obvious. After the adoption of
the adoption of other accounting principles can be
IAS/IFRS or US GAAP the analysts should receive
derived from the functional fixation theory (Hand
more (externally verified) information about the 1990). According to this theory individuals tend to
company’s financial position which enables them to
retain their decision-making process after a change
build their prediction on a larger information set.
in the accounting principles providing the informa-
Similarly, more informative accounting methods are
tion for their decisions. When the adoption of a new
likely to enhance analysts’ forecast accuracy without
accounting regime requires modifications, the deci-
seriously biasing the net income to be forecasted.
sion-making process used beforehand might no
Therefore, we state the following hypothesis:
longer be suitable and consequently forecast accu-
Hypothesis 1: Forecast accuracy is higher for racy could deteriorate. Again, the smoothing strate-
companies applying IAS/IFRS and US GAAP than
gies mentioned above could mitigate this effect.
for those applying German GAAP.
Nevertheless, there is a potential reason for an in-
If a company has to or can change its measurement
crease in forecast accuracy in the years of the adop-
method when new accounting principles are tion of IAS/IFRS and US GAAP. Prior research
adopted, this may hamper forecast accuracy. Brown
shows that the motivation for accounting changes is
(1983) as well as Elliott and Philbrick (1990) pro-
to improve the informativeness of financial report-
vide evidence for this hypothesis for US companies.
ing (e.g., Bartov and Bodnar 1996). Therefore, fore-
Furthermore, a change from prudent accounting
cast accuracy should c.p. increase in the year of
principles to accounting principles that purport to
adoption, especially when additional disclosures,
provide a true and fair view of the financial position
like reconciliations, are provided.
of company could change the importance analysts
31

---

BuR - Business Research
Official Open Access Journal of VHB
Verband der Hochschullehrer für Betriebswirtschaft e.V.
Volume 1 | Issue 1 | May 2008 | 26-53
Despite the plausibility of the opposing view, we
forecast accuracy as dependent variable on the vari-
believe that the most likely effect of an adoption of
ables that are examined and on several control vari-
new accounting principles is that forecast accuracy
ables as independent variables.
is lower for the year of change, because financial
Similar to other studies on the determinants of fore-
analysts might not be able to cope with such a great
cast accuracy (e.g., Lang, Lins, and Miller 2003;
change in an important data source. As smoothing
Hope 2004), we define forecast accuracy of a com-
strategies are more likely for companies switching
pany’s earnings per share (EPS) as follows:
from German GAAP to IFRS we state the following

research hypothesis:
−
EPS

Actual
−
EPS

forecasted
Median


Hypothesis 2: In the year of adoption of new
FA_MEDIAN =
of

middle

at the

price
Stock

accounting principles, forecast accuracy is lower
month
forecast

the
than in other periods, especially for companies
switching from German GAAP to US GAAP.
We have made two modifications in comparison to
According to Markov and Tamayo (2006), financial
previous definitions. First, we use the median of the
analysts learn about the parameters of companies’
estimates during a specified period as a consensus
earnings processes and thereby improve their fore-
forecast and not the mean, because the median is
casts. Such a relationship is suggested by studies at
less sensitive to outliers. In order to control for the
the individual analyst level showing that the firm-
influence of this modification on the result, we con-
specific experience of an analyst is positively associ-
ducted a sensitivity analysis using the mean of the
ated with her forecast accuracy (Mikhail, Walther,
estimates alternatively and found no material differ-
and Willis 1997; Clement 1999; Clement, Rees, and
ences in the results (see Section 8 for details). Sec-
Swanson 2003). Moreover, Markov and Tamayo ond, we do not use the median of an entire fiscal
(2006) argue that unanticipated shocks to the earn-
year, but rather of each month before the report
ings process may bias the learning process by rein-
date, i.e. the date when the actual EPS is reported by
troducing uncertainty. Subsequently, financial ana-
the company. This enables us to control for the ef-
lysts need to learn again about the determinants of
fect that forecasts closer to the report date are usu-
the earnings process. A change in the accounting
ally better than earlier forecasts by using a control
principles as an important information basis for the
variable. Due to this modification we use the stock
earnings forecasts could be seen as such a shock. In
price at the middle of the forecast month as deflator
the second year of applying IAAP the financial fig-
to facilitate comparisons across observations.
ures of the first year of adopting IAS/IFRS or US
GAAP and of at least one year of comparison pre-
4.2 Independent variables
ceding the year of adopting a new accounting re-
The first hypothesis is tested by including dummy
gime are available to financial analysts. Assuming
variables for the type of accounting regime applied
that this information is already sufficient for ana-
into the regression. Taking German GAAP as a ref-
lysts in order to evaluate the impact of adopting
erence group, we use IFRS that takes the value 1 if
international accounting regimes on the financial
the actual EPS is based on IAS/IFRS and 0 other-
statements of a company we state the following
wise as well as US that takes the value 1 if the actual
hypothesis:
EPS is based on US GAAP and 0 otherwise. Assum-
Hypothesis 3: In the year after the adoption of
ing that hypothesis 1 is true, these dummy variables
new accounting principles forecast accuracy is not
should be significant and positive in a regression on
different to that of other periods (except for the
the forecast accuracy.
periods of adoption).
To test the second hypothesis, we define the dummy
variable ADOPT which takes the value 1 when the
4. Variables and models
forecast is made in a month of the year a company
adopted IFRS or US GAAP and 0 otherwise. Using
this variable as well as lagged accounting regime
4.1 Dependent variable
dummies indicating the type of accounting princi-
The study investigates the influence of different ples applied in the previous year (HGB-1, IFRS-1,
accounting regimes on the forecast accuracy of fi-
US-1), we construct interaction terms to indicate
nancial analysts. Therefore, we regress a measure of
what type of switch of the accounting principles
32

---

BuR - Business Research
Official Open Access Journal of VHB
Verband der Hochschullehrer für Betriebswirtschaft e.V.
Volume 1 | Issue 1 | May 2008 | 26-53
occurred in the respective fiscal period, i.e. from
4.3 Control variables
German GAAP to IAS/IFRS (ADOPT * IFRS * Following previous studies, we include several con-
HGB-1), from US GAAP to IAS/IFRS (ADOPT *
trol variables into the regression. TIME refers to the
IFRS * US-1), from German GAAP to US GAAP
number of days between the report date and the
(ADOPT * US * HGB-1) or from IFRS to US GAAP
date of the consensus forecast. Figure 2 illustrates
(ADOPT * US * IFRS-1). Under the assumption that
the calculation of the variable TIME. In a certain
hypothesis 2 applies, at least the interaction term
year (2003 and 2004 in the example) we focus on
indicating a switch from German GAAP to US GAAP
consensus estimates during the time period between
should be significant and have a negative sign in a
two report dates (15/03/2003 – 15/03/2004 and
regression on forecast accuracy.
15/03/2004 – 15/02/2005 in the example) regard-
Moreover, we use the interaction terms ADOPT-1 *
less of the length of this period (12 months and 11
IFRS as well as ADOPT-1 * US for testing hypothesis
months in the example). The single forecasts of
3. A value of 1 for these interaction terms means
analysts during this period are valid for a maximum
that IFRS or US GAAP, respectively, was adopted in
of 105 days (as defined by IBES) or until the earlier
the previous year, i.e. the observation refers to the
of the report date or the withdrawing of the forecast
second fiscal year of applying IAS/IFRS or US
by the respective analyst. The consensus forecast of
GAAP. Hypothesis 3 implies that the forecast accu-
all valid single forecasts is calculated at the middle
racy for such years is not significantly different in
of all months within the period between two report
comparison to other periods (except for the periods
dates (e.g. at 15/06/2003 and 15/09/2004 in the
of adoption) and thus these interactions terms example). The period between the date of the con-
should not be significant in a regression on forecast
sensus forecast and the report date is defined as
accuracy.
TIME for this consensus forecast (15/03/2004 –
Figure 2: Illustration of calculating the variable TIME

12 months
11 months
Ea
E rni
rn n
i g
n s
g rep
re o
p rt
o da
d te
15/
15 03/
0 2
3/ 00
2 3
00
15/0
/ 3/2
3/ 004
15/
1 0
5/ 2/
0 20
2/ 05
20
t
Fisca
i
l period end date
31/1
1/ 2/2002
31/
31 12/
1 20
2/ 03
20
31/1
/ 2/2
2/ 004
12 month
t s
12 months

TIME
TI
in
i dep
e en
e den
d
t varia
i ble
b
274
27 days
153 days
y
IBES
E mont
S
hly consen
onse su
s s for
o e
r ca
c st
a
15/
15 06/
0 20
6/ 03
20
15/
15 09/
0 2
9/ 004
2
(in
(i c
n lud
c
e
lud s
e al
a ll si
s ng
i le
ng fo
f re
o c
re a
c st
s s
t mad
m e
ad
in
i th
n
e
th last
s 1
t 05 d
5 a
d ys
y )
s
Ea
E rn
r in
i g
n s
g rep
re o
p rt
o da
d te
t
15/0
/ 3/2
3/ 003
15/
15 03/
0 2
3/ 00
2 4
00
15/0
5/ 2/2005
t
Fiscal
Fisc per
pe iod en
e d
n da
d te
31/
3 1
1/ 2/
1 20
2/ 02
20
31/1
1/ 2/2003
31/1
31/ 2/
1 20
2/ 04
20
33

---

BuR - Business Research
Official Open Access Journal of VHB
Verband der Hochschullehrer für Betriebswirtschaft e.V.
Volume 1 | Issue 1 | May 2008 | 26-53
15/06/2003 = 274 days and 15/02/2005 – TECDAX in 2003. SDAX (called SMAX until 2002)
15/09/2004 = 153 days).
is the small-cap segment. A consensus forecast of a
As already mentioned, forecasts issued earlier are
month is given the value 1 for one of these variables
likely to be less accurate than forecasts issued closer
if the company for which the forecast is made be-
to the time earnings are announced. The reason for
longs to the respective market segment in this
this is that the analysts have less information avail-
month (on the respective forecast date), and 0 oth-
able and thus higher uncertainty about a company’s
erwise. To our knowledge, these dummy variables
results in a fiscal year. Accordingly, this relationship
have never been used before in an empirical study
is documented by several studies (e.g., Brown, on analysts’ forecast accuracy. Nevertheless, we
Richardson, and Schwager 1987; Lys and Soo 1995;
believe that they are good proxies for the informa-
Das and Saudaragan 1998; Jacob, Lys, and Neale
tion environment of a German listed company, be-
1999; Duru and Reeb 2002). We expect a negative
cause these market indices are characterized by
sign for this variable in the regression on forecast
additional information requirements (e.g. concern-
accuracy.
ing quarterly reports). Moreover, the DAX requires
Similar to several previous studies (e.g. Lang and
a higher free float than the other market segments
Lundholm 1996; Hope 2003a; Hope 2003b), we use
(German Stock Exchange 2007). Finally, the visibil-
firm size as a control variable. The variable MCAP
ity of the DAX is highest followed by the MDAX, the
represents the market capitalization at the middle of
TECDAX, and the SDAX. The New Market also had
the month for which the consensus forecast is made.
a very high visibility during its existence, especially
We use the natural logarithm of this variable as
in the years 1998–2001, and had additional infor-
MCAP itself is heavily skewed. Log(MCAP) is con-
mation requirements. Therefore, we expect that
sidered to be a proxy for the information environ-
forecast accuracy for companies belonging to one of
ment of a company, as large companies are more
these market segments is higher than for stocks
likely to provide additional information to the public
belonging to none of these segments. Furthermore,
or more private information (Jaggi and Jain 1998)
forecast accuracy is probably highest for the DAX
to the analysts than smaller companies. This would
followed by the New Market and the MDAX.
imply a positive coefficient on log(MCAP). Firm size
As we will show in Section 6 only one of the last
is also seen as a proxy for other company-specific
three controls (i.e. log(MCAP), COVERAGE, or the
factors, like management incentives for which pre-
index dummy variables) is included in the analyses
dictions are unclear (Hope 2003a).
to avoid multicollinearity.
COVERAGE stands for the number of analysts fol-
Plumlee (2003) finds that the complexity of the
lowing a company in a month. This variable is a
forecasting task is negatively associated with fore-
proxy for the intensity of competition (Lys and Soo
cast accuracy. We expect the complexity of the fore-
1995) and thus for the incentives to forecast accu-
casting task and, thus, the forecast errors to be posi-
rately (Hope 2003a). As the forecast accuracy of the
tively correlated with a stock’s risk. As the consen-
consensus forecast should be higher when more sus forecast of a stock is often viewed as the surro-
analysts estimate, we expect a positive estimated
gate for market expectations about that company,
coefficient on COVERAGE.
we add BETA as a variable covering systematic risk
Moreover, we define dummy variables for the mem-
to proxy for difficulties in forecasting risky results.
bership in one of the market segments of the Ger-
BETA is the fundamental or predicted beta of a
man Stock Exchange. Taking companies not in-
company in the investigated period. In financial
cluded in a stock index as a reference group, we
theory, beta is a gauge of the expected response of
differentiate between the following indices: DAX,
the stock to the overall market portfolio. The pre-
MDAX, TECDAX, NEW_MARKET, and SDAX. dicted beta is derived from Barra’s German equity
DAX stands for the German blue-chip segment risk model, which is a multifactor risk model includ-
comprising the 30 largest and most actively traded
ing 12 risk attributes (e.g. leverage, earnings vari-
German companies. MDAX is the mid-cap segment,
ability ability, growth, and liquidity) plus industry
and TECDAX (founded in 2003) is the technology
exposure. Table 1 gives an overview of all risk attrib-
sector segment. NEW_MARKET stands for the utes used in Barra’s German equity risk model.
“Neuer Markt”, which was the German market seg-
These risk attributes are operationalized by calculat-
ment for technology stocks and was replaced by the
ing 43 variables, so-called descriptors. We do not
34

---

Document Outline

• ernstberger3.pdf
  • Analysts Forecast Accuracy in Germany: The Effect of Different Accounting Principles and Changes of Accounting Principles
  • Abstract
  • Keywords
  • 1. Introduction
  • 2. Literature review
    • 2.1 Drivers of analyst forecast accuracy
      • Figure 1: Drivers of analyst forecast accuracy
    • 2.2 Analyst forecast accuracy and accounting data
    • 2.3 Impacts of the adoption of IAAP
  • 3. Hypotheses
  • 4. Variables and models
    • 4.1 Dependent variable
    • 4.2 Independent variables
    • 4.3 Control variables
    • Figure 2: Illustration of calculating the variable TIME
    • Table 1: Factors in BarraАs German equity risk model
    • 4.4 Models
  • 5. Sample
    • 5.1 Restriction to Germany
    • 5.2 Sample period
    • 5.3 Variables and data sources
    • Table 2: Overview of variables and of data sources
    • 5.4 Sample selection
    • Table 3: Descriptive statistics
  • 6. Descriptive statistics
    • Table 4: Differences between variables for HGB observations and IAS/IFRS as well as US GAAP observations
    • Table 5: Descriptive statistics for the years and accounting principles applied
    • Table 6: Correlations
  • 7. Regression results
    • Table 7: Regression results for models (1a)-(1c)
    • Table 8: Regression results for models (2a)-(2c)
    • Table 9: Regression results for models (2a)-(2c) using aggregated data
  • 8. Sensitivity analyses
    • 8.1 Aggregated yearly data
    • 8.2 Self-selection
    • Table 10: Regression results for the propensity of applying IAS/IFRS and US GAAP
    • Table 11: Regression results for models (2a)(2c) including the Inverse Mills Ratio
    • 8.3 Further sensitivity analyses
  • 9. Conclusions
  • Acknowledgments
  • References
  • Biographies

---