dividends and earnings quality in poland
TRANSCRIPT
8172019 Dividends and Earnings Quality in Poland
httpslidepdfcomreaderfulldividends-and-earnings-quality-in-poland 16
wwwe-nansecomUniversity of Information Technology and Management Sucharskiego 2 35-225 Rzeszoacutew
43
Financial Internet Quarterly bdquoe-Finanserdquo 2013 vol 9 | nr 3
42
Financial Internet Quarterly bdquoe-Finanserdquo 2013 vol 9 | nr 3
wwwe-nansecomUniversity of Information Technology and Management Sucharskiego 2 35-225 Rzeszoacutew
DIVIDENDS AND EARNINGS
QUALITY IN POLAND
Mieczysław Kowerski1
Ph D Mieczysław Kowerski Mieczysław Kowerski Zamość University o Management and Administration Akademicka St 4 22-400
Zamość Poland mkowerskiwsziaedupl
Abstract The purpose of this article is to show on the example of Warsaw Stock Exchange Poland (WSE) how in
emerging capital markets dividends provide information about earnings quality as measured by their
persistence In the paper the regressions models of future earnings (in years t + 1 and t + 2) wereapplied on current earnings (in year t) current dividends decision (in year t) and the interaction of
current dividend decision and earnings proposed by D J Skinner and E Soltes (2011) using pooled
cross ndash sectional time ndash series data Aset of 2263 observations coming from the companies listed on
the WSE in 1995-2009 was used for the calculation For estimating the parameters recursive modeling
was used Specic models were estimated using the heteroskedasticity-corrected general least squares
method It was shown that on the WSE the quality of earnings depends more distinctly on a rmrsquos
dividend policy than on the developed markets
Received 05062013 Accepted 11122013
JEL Classication G39
Keywords Quality of earnings earnings persistence dividend policy Warsaw Stock Exchange Performance Measurement
bdquoDividends ell the ruthrdquo (Miller 2006 p 33)
I983150983156983154983151983140983157983139983156983145983151983150
Earnings are the most synthetic measure o the
economic benefits achieved by actions undertaken
by the company (Nowak 2009 p 181) Tat is why
it is a basic measure o business activity evaluation
by shareholders and potential investors Earnings are
also quite ofen the basis or evaluating and rewarding
company managers But the act that it is a synthetic
measure which is in practice the unction o all
business transactions (both positive and negative)
occurring in an enterprise causes a lot o problems
with its unequivocal evaluation On the other hand
delivering a financial result is a complicated process
that involves decision making and as such is one
o the actions most ofen bordering on the side o
lsquocreative accountingrsquo (Wąsowski 2010 p 16) and
may fluctuate over time Consequently the quality o
financial results becomes an issue
Quality earnings may be very differently understood
and measured P Dechow W Ge and C Schrand
define earnings quality as ollows ldquoHigher quality
earnings provide more inormation about the eatures
o a firmrsquos financial perormance that are relevant
to a specific decision made by a specific decision-
makerrdquo (2010 p 344) Te same authors suggest three
categories o quality earnings indicators (Dechow et
al 2010 p 345)
1) properties o earnings
2) investor responsiveness to earnings
3) external indicators o earnings misstatements
Note they propose to proxy properties o earnings by
the ollowing indicators
1) earnings persistence
2) abnormal accruals
3) earnings smoothness
4) asymmetric timeliness and timely loss recognition
5) target beating
A simple model specification estimates earnings
persistence as
Earnings Earningst t t +
= + +1 0 1
α α ε
(1)
In model (1) earnings are typically scaled by assets
although some researchers examine margins (scaled
by sales) or scaled by the number o shares A higher
α 1 implies a more persistent earnings stream
Intuitively the logic behind earnings persistence
being a quality metric is as ollows i firm A has a more
persistent earnings stream than firm B in perpetuity
then in firm A current earnings is a more useul
summary measure o uture perormance A urther
extension o model (1) is to determine whether other
financial statement elements (or variables outside o
the financial statements) are incremental over current
earnings in predicting uture earnings (Dechow et al
2010 p 352)
Earnings Earnings Financial Statements componet + = + +
1 0 1 2α α α nnts
Other ormation
t
t t + +α ε 3
inf (2)
Tese authors also suggest six groups o thedeterminants o earnings quality (Dechow et al
2010 p 379)
1) firm characteristics (most ofen analyzed in researchare firm perormance debt growth and investmentsize)
2) financial reporting practices
3) governance and controls
4) auditors
5) equity market incentives
6) external actors (including capital requirementspolitical processes and tax and non-tax regulation)
DIVIDENDS AS A TOOL FORASSESSING EARNINGS QUALITYFOR DEVELOPED CAPITALMARKETS
Among the actors already mentioned which
determine earnings quality a very important one is
missing as pointed out by D J Skinner and E Soltes
(2011) namely ndash the fi rmrsquos dividend policies
A wealth o literature dating back to at least the
articles by J Lintner (1956) and M Miller and F
Modigliani (1961) points out that dividends are a way
to signal good prospects or high uture earnings by
the management board o a company
Later this idea took the orm o signaling theory
(Bhattacharaya 1979 Myers amp Majlu 1984 John amp
Williams 1985)
Te basis o this theory is the inormation asymmetry
between the management board and the minority
shareholders Minority shareholders usually do
not have the same inormation as management
and majority shareholders Complete inormation
especially about a companyrsquos uture (regarding orexample technologies and production processes)
is not provided by studying the company accounts
Tereore a dividend may be a way to provide
minority shareholders and potential investors
with inormation about the companyrsquos situation
and its uture profits New or increased dividends
are a positive signal about the companyrsquos financial
situation whereas their cancellation or reduction is
a negative signal
According to J Lintner (1956 p 97) dividend policy
is one o a companyrsquos primary financial decisions
Lintner conducted very detailed interviews with
the boards o 28 targeted companies For those
companies he collected financial data between
the years 1947-1953 (196 observations) Tese
interviews show that according to management
dividends are very important or shareholders Note
that shareholders are not so much interested in
a constant level o paid dividends but in a relatively
fixed percentage pay-out Te belie that the market
puts a premium on stability or gradual growth was
strong enough so that most managers sought to avoid
making changes in their dividend rates which mighthave had to be reversed within a year or so (Lintner
1956 p 99) In addition managers reduce dividend
rates very reluctantlydagger
Te results o the interviews carried out by J Lintner
concerning the dividend policy lead to the conclusion
that management will decide not to pay dividends
until they believe that they will be able to pay them
in the uture so that in the uture they will able to
dagger Sometimes in order to lsquodeendrsquo the existing payout ratiothey pay a dividend although the company has recorded a loss
(DeAngelo et al 2008 p130)
8172019 Dividends and Earnings Quality in Poland
httpslidepdfcomreaderfulldividends-and-earnings-quality-in-poland 26
wwwe-nansecomUniversity of Information Technology and Management Sucharskiego 2 35-225 Rzeszoacutew
wwwe-nansecomUniversity of Information Technology and Management Sucharskiego 2 35-225 Rzeszoacutew
4544
Financial Internet Quarterly bdquoe-Finanserdquo 2013 vol 9 | nr 3 Financial Internet Quarterly bdquoe-Finanserdquo 2013 vol 9 | nr 3
achieve adequate (permanent) earnings Lintnerrsquos
conclusions have been confirmed by (Brav Graham
Harvey amp Michaely 2005) recently
It can thereore be hypothesized that dividends
provide inormation about earnings quality measured
by their persistence
Recent accounting scandals seem to confirm the
above hypothesis Although it is relatively easy
or management to lsquofixrsquo current profits and lsquopaintrsquo
a companyrsquos situation it is much more expensive
to pay dividends in order to inorm minority
shareholders and potential uture shareholders o
the good financial situation o the company and
its high level o profits when that level is a result o
lsquocreative accountingrsquo Management may decide on this
operation occasionally especially when that profit
is not a result o actual company perormance but
accounting interventions improving this result only
or a short period (eg through lsquoappropriatersquo booking
o liabilities at the end o the reporting period)
Skinner and Soltes (2011 p 14) in order to investigate
the relationship between dividend policy and earnings
quality suggested earnings linear models in years t +
1 and t + 2 determined by a decision to pay dividends
in year t the rate o assets return in year t and variable
product describing interaction o the decision to pay
dividends in year t and return on assets in year t Note
that earnings in year t t + 1 and t + 2 were related to
the assets at the end o year t ndash 1
E TA D E TA D E TAit it it it it it it it + minus minus minus
( ) = + + ( ) + sdot( ) +1 1 0 1 2 1 3 1
α α α α ε ε it
(3)
and
E TA D E TA D E TAit it it it it it it it + minus minus minus
( ) = + + ( ) + sdot( ) +2 1 0 1 2 1 3 1
α α α α ε ε it
(4)
where
E TAit it
minus1 mdash firmrsquos earnings in year t to total assets in the end o year t ndash 1 and in (return on assets)
E TAit it + minus1 1
mdash firmrsquos earnings in year t +1 in relation to total assets in the end o year t ndash 1 in
E TAit it + minus2 1
mdash firmrsquos earnings in year t +2 in relation to total assets in the end o year t ndash 1 in
Dit
mdash an indicator variable set to 1 i a dividend is paid by firm and in year t and 0 otherwise
D E TAit it it sdot
minus
( )1mdash interaction between D
it and E
it A
it ndash1
I we assume that variable Eit A
it ndash 1 is a measure o
total assets profitability then variable Dit (E
it A
it ndash 1 )
is profitability o companies paying dividends in
year t Tus the authors believe that earnings quality
is determined by their earnings persistence Hence
the above models can be called earnings persistence
models Model (3) indicates how persistent are
earnings achieved by companies in year t in the
ollowing year (t + 1) while the model (4) indicates
how persistent are earnings achieved by companies in
year t two years later (t + 2)
In this regression coefficient α2 measures the
persistence o earnings o all firms (irrespective i the
firm pays a dividend or not) Under the hypothesis
that dividends are inormative about the quality o
reported earnings it is expected that the coefficient
on earnings will be larger or dividend-paying firms
indicating that their earnings are more persistent
(α3 gt 0) Te sum o the coefficients α2 and α3 inorms
us about the earnings persistence o companies
paying dividends (Skinner amp Soltes 2011 p 14)
Skinner and Soltes analyzed a sample including all
non-utility non-financial domestic firms quoted
on the NYSE AMEX and NASDAQ rom 1974 to
2005 Altogether they gathered a total o 123728
observations (Skinner amp Soltes 2011 p 10) Tey
estimated all regressions using the ordinary least
squares method with two-way robust standard errors
(clustered by firm and time) to account or cross-
sectional and time series dependence Models were
estimated separately or the three sub-periods 1974-
1983 1984-1994 and 1994-2005
In models describing earnings persistence in year
t + 1 (model 3) values o estimated coefficients α2
oscillated around 08 (respectively 0781 0812
0835) and were significant at the 001 level which
according to the authors means that the profits are
airly persistent Furthermore these parameters
confirm the results o R G Sloanrsquos study (1996) who
received the value o the coefficient α2 = 084 Tere
is evidence o a modest increase in persistence over
time with the coefficient on earnings increasing
rom 078 in the earlier sub-period (1974 through
1983) to 084 in the most recent sub-period (1994
through 2005) Coefficients on a variable which
is the product o a decision o dividend paying
and earnings in relation to the value o assets
(Dit (E
it A
it ndash 1 )) are positive and statistically significant
in all three sub-periods (respectively 0031 0080
and 0064) It means that profits are more persistent
or dividend payers Since 1984 the sum o α2 and α3
coefficients or dividend payers has been around 090
Also similar results are provided by models
describing earnings persistence achieved in year t in
two years later (model 4) However coefficient values
on variables describing earnings α2 are a little lower
than in the previous models In this case the sum o
coefficients on the variables describing earnings ordividend payers in the period 1994ndash2005 is 0730
+ 0099 = 0829 (Skinner amp Soltes 2011 p 15) It
is worth emphasizing that the estimated models
are characterized by a high degree o explanation
Adjusted coefficients o determination (Adj R 2)
values or earnings persistence models o year t + 1
range rom 063 to 070 and or earnings persistence
models or year t + 2 are slightly worse ranging rom
043 to 054
Te estimation results have shown that dividend
payers have higher persistence (and thus quality) o
earnings than those not paying dividends and this
relationship does not depend on the level o dividends
paid
METHODS PROPOSED FOR
ASSESSING EARNINGS QUALITY
OF COMPANIES LISTED ON THE
WARSAW STOCK EXCHANGE
D983137983156983137
Te Warsaw Stock Exchange (WSE) is the most
dynamically growing market in Central and EasternEurope (Warsaw Stock Exchange Wiener Boumlrse
Prague Stock Exchange Budapest Stock Exchange
Bucharest Stock Exchange Bulgarian Stock
Exchange) Te WSE is the regional leader in terms o
key market ratios such as the value o equities trading
the number o domestic and oreign companies
the number o IPOs and since 2009 capitalization
which in the end o 2009 was 105 billion euro and
in the end o 2010 142 billion euro Te GDP share
o capitalization o domestic companies leapt rom
31 in the end o 2009 to 38 in the end o 2010 At
the end o 2009 the WSE Main List comprised 397
companies (354 domestic and 25 oreign) and at the
end o 2010 the number o quoted firms increased to
400 (373 domestic and 27 oreign)Dagger2
Te database o all domestic companies listed on
the Warsaw Stock Exchange rom 1995 to 2009 sect was
a starting point or the calculation It must be
considered that only the companies whose shares
were listed on the stock exchange throughout the
entire year were taken into account From the set o
domestic companies national investment unds were
excluded due to their different method o financial
reporting Some companies were removed which
in act were recorded throughout the year but were
excluded rom the stock exchanges in the first hal o
the next year4Moreover companies with negative
equity values and companies with zero net revenues
rom sales o products services goods and materials(not engaged in any operating activities in a certain
year) were excluded rom the calculation
With the development o the stock exchange
the number o companies admitted to the study
each year increased In 1995 the study included
44 companies while in 2008 293 companies In
this way cross-section datasets or 14 years were
obtained Every year this set consists o different
numbers o observations and can be analyzed or
each year separately Also annual data rom all years
can be combined and a set o pooled cross-sectional
time-series data can be obtained In total this set
consists o 2263 observations (companies ndash years)
It should be emphasized that in this pooled set each
observation ought to be treated as a separate entity
Te propensity to pay dividends o companies listed
on the Warsaw Stock Exchange is much lower than
in developed capital markets (Bartram Brown How
amp Verhoeven 2009 DeAngelo H DeAngelo L amp
Skinner 2008 von Eije amp Megginson 2008 Denis amp
Osobov 2008) But on the Warsaw Stock Exchange
the characteristic or developed capital market
process o lsquodisappearing dividendsrsquo (Fama amp French2001) was not observed Te research results suggest
the decision to equalize dividends and capital profits
tax rate rom 2004 and systematic CI reduction was
beneficial to increasing the share o companies paying
dividends in the total number o listed companies
Tis allowed companies to allocate larger earnings to
dividends (Kowerski 2010)
Dagger Warsaw Stock Exchange Annual Report 2010
httpwwwgpwplraporty_roczne (accessed 4 February 2012)sect Data rom Notoria Service httpirnotoriapl (accessed4 February 2012)
Such companies usually did not submit reports to Notoria
Service
8172019 Dividends and Earnings Quality in Poland
httpslidepdfcomreaderfulldividends-and-earnings-quality-in-poland 36
wwwe-nansecomUniversity of Information Technology and Management Sucharskiego 2 35-225 Rzeszoacutew
wwwe-nansecomUniversity of Information Technology and Management Sucharskiego 2 35-225 Rzeszoacutew
4746
Financial Internet Quarterly bdquoe-Finanserdquo 2013 vol 9 | nr 3 Financial Internet Quarterly bdquoe-Finanserdquo 2013 vol 9 | nr 3
Figure 1 Changes in the number of companies under survey from 1996 to 2009 and share of dividend payers
Source Own calculations
Te dividend policy o companies listed on the Warsaw
Stock Exchange has become more and more similar
to the behavior o companies in developed capital
markets Dividend value not only in current prices
but also in constant prices is growing rapidly In 2009
the average dividend payout made by a company
listed on the Warsaw Stock Exchange amounted to
843 million zloty and it was in current prices twenty
six times higher while in constant prices five times
higher than the average payout in 1992 But the
relation o dividends to GDP remains very low and
does not exceed 1 Also we can obser ve an increase
o payout concentration mdash a relatively small number
o major companies pay increasing dividends whichrepresent a significant part o all payouts Companies
pay an increasing share o profits which makes the
dividend payout ratio increase the dividend yield
ratio also increases
From 1996 to 2009 shares o companies paying
dividends underwent multidirectional changes Tey
were particularly high (above 40) rom 1996 to
1997 Ten they ell to a minimum in 2002 (215)
From 2003 to 2006 they increased again to 375
Since 2007 shares o dividend-paying companies
have been alling
M983141983156983144983151983140 983151983142 983141983155983156983145983149983137983156983145983151983150
Te method and models suggested and discussed by
Skinner and Soltes in the previous chapter were used
to test earnings quality o domestic companies listed
on the Warsaw Stock Exchange
According to model (3) a relation o earnings in year
t + 1 to value o total assets at the end o year t ndash 1 is
a unction o dividend in year t earnings in year t to
the value o total assets at the end o year t ndash 1 and the
product o two previous variables Tis means that to
calculate the values o dependent and independent
variables it is necessary to have data about companies
listed or the successive three years In a baseline
collection not all companies met this criteria
thereore only 1481 observations could be included
in the study According to model (4) earnings in year
t + 2 to the value o total assets at the end o year t ndash 1
is a unction o dividend in year t earnings in year t to
the value o total assets at the end o year t ndash 1 and the
product o the two previous variables Consequently
this means that to calculate the values o dependent
and independent variables it is necessary to have data
on companies listed or the successive our years Tis
limited the initial collection to 1195 observations
In both sets o data there are single outlier
observations o a dependent variable Observations
o this type can significantly change the final result
o the analysis and their disregard can be atal Te
simplest but quite effective method o lsquocopingrsquo with
outliers is to remove them rom the collection o data
under consideration which increases the robustness
o the estimated coefficients Estimators obtained in
this way are called lsquorobust estimatorsrsquo and estimated
models that can be called lsquorobust regressionrsquo denote
a set o estimation techniques which are less sensitive
than the ordinary least squares to the effect o
possible influential observations (Baldau amp Santos
Silva 2009 p 2)
In the present study observations or which
dependent variable values were lower thanndash100 or higher than 100 were removed As
a result the output set o observations or model 3
was reduced to 1468 while or model 4 to 1174
It must be emphasized that the method used or
selection o companies in the models (3) and (4)
especially or robust estimation can cause a sample
selection bias (Heckman 1976) with companies o
a slightly better economic and financial situation
Firms with negative equity and those that were not
listed or did not meet any o the criteria or creating
the output database respectively by three or our
successive years were removed rom samples On the
other hand the situation o companies excluded rom
the calculation usually so considerably deviates rom
the vast majority o companies included in samples
that their presence not only would not h elp to explain
changes in earnings quality but also could darken the
depiction o the phenomenondaggerdagger
Because o the act that the number o observations
in particular years are different and still there is no
inormation as to what time period (how many years)
observations should be selected recursive modeling
was used consisting o estimating consecutive modelso with increasingly shorter series emerging by
removing the oldest data every year (Charemza amp
Deadman 1997 p 62-65)
Te best method or estimating coefficients o
models (3) and (4) proved to be a heteroskedasticity-
corrected general least squares method
daggerdagger In the case o companies with negative equities some
indicators incorrectly inorm about the situation o the company
For example when such a company records a negative earning
(which is very probable) then the rate o return on equity is positive
mdash which could indicate a good inancial situation
THE RESULTS OF THEESTIMATION OF EARNINGSPERSISTENCE MODELS OFCOMPANIES LISTED ON THEWARSAW STOCK EXCHANGE
In 10 models o earnings persistence in the ollowing
year using recursive modeling and starting rom the
initial set o observations (1481 observations rom
1997 to 2008) coefficients on variable Dt were negative
and statistically insignificant A detailed analysis o
results leads to the conclusion that a catalysis effect
occurred (Hellwig 1977) which caused the lack o
coincidence o coefficients on Dt variable (Hellwig
1976)DaggerDagger Correlation coefficients between Et +1A
t ndash 1
and Dt variables are positive and coefficients on D
t
variable are negative
DaggerDagger Coeicient αi ulills the coincidence rule when sign αi = sign
ri where ri is correlation coeicient between dependent variable
and i-th independent variable
8172019 Dividends and Earnings Quality in Poland
httpslidepdfcomreaderfulldividends-and-earnings-quality-in-poland 46
wwwe-nansecomUniversity of Information Technology and Management Sucharskiego 2 35-225 Rzeszoacutew
wwwe-nansecomUniversity of Information Technology and Management Sucharskiego 2 35-225 Rzeszoacutew
4948
Financial Internet Quarterly bdquoe-Finanserdquo 2013 vol 9 | nr 3 Financial Internet Quarterly bdquoe-Finanserdquo 2013 vol 9 | nr 3
Table 1 The results of estimation of earnings persistence models in yeart + 1with heteroskedasticity-corrected general l east squares method
Source Own calculations in GREL (Corttrell amp Luchetti 2010)
Table 2 The results of estimation of earnings persistence models in yeart + 2with heteroskedasticity-corrected general l east squares method
Source Own calculations in GREL (Corttrell amp Luchetti 2010)
Only in the models or 2007-2008 and or 2008
parameters on the Dt variable were coincident but
also statistically insignificant Tereore it was
decided to reject the Dt variable and to estimate
models o earnings persistence in the ollowing year
dependent on the two other variables
Figure 2 Changes in values of coefficients in earnings persistence modelsin the following year (t + 1) estimated by recursive modeling method
Source Own calculations
Estimated by a recursive method coefficients are
characterized by very high stability Te model
estimated on data rom the years 2002 to 2008 (1039
observations) is characterized by the highest value o
the adjusted determination coefficient although this
value (02000) is much lower than in Skinner-Soltes
modelssectsect Tis model will serve or a more detailed
analysis
Estimated in this model coefficient α2 is 0467 and
thus is about 034 lower than estimated by Skinner-
Soltes coefficient α2 or the US Stock Exchanges
Furthermore estimated coefficient α3 is 0416 and ismany times higher than coefficient α3 estimated by
sectsect On the other hand the value o F statistic indicates the
signiicance o the multiple correlation coeicient R
( p = 300E-29) and thus the overall signiicant eect o both
variables on the earnings persistence in the ollowing year A
relatively low determination coeicient value in cross-section
models and cross-time models estimated on large sets o micro
data is quite common (Gruszczyński 2002 p 55)
It should be emphasized however that other earningspersistence models in the ollowing year obtained rom recursive
modeling are characterized by eatures similar to the discussed
model
Skinner-Soltes or the US Stock Exchangesdaggerdaggerdagger Tis
means that in Warsaw earnings o companies paying
dividends are more persistent than companies not
paying dividends than in New York In Warsaw
in year t the increase o return on total assets o
the company paying the dividend by 1 percentage
point caused the increase o earnings value in year
t + 1 to the value o total assets in year t ndash 1 by 0883
percentage points while in the case o a company not
paying dividends only by 0416 percentage points On
the New York Stock Exchanges the difference in avor
o companies paying dividends in the test period di dnot exceed 008 o a percentage point
Applying similar procedures as in the case o earnings
persistence models in the ollowing year models o
earnings persistence two years later were estimated
In this case removing 21 outliers provided models
that can be used to assess the phenomenon As in the
case o earnings persistence models in the ollowing
year again coefficients on the Dt variable proved to
daggerdaggerdagger O course comparisons o results obtained or the WarsawStock Exchange and or New York Stock Exchanges should be
treated with caution mainly because o the much smaller number
o observations on WSE
8172019 Dividends and Earnings Quality in Poland
httpslidepdfcomreaderfulldividends-and-earnings-quality-in-poland 56
wwwe-nansecomUniversity of Information Technology and Management Sucharskiego 2 35-225 Rzeszoacutew
wwwe-nansecomUniversity of Information Technology and Management Sucharskiego 2 35-225 Rzeszoacutew
5150
Financial Internet Quarterly bdquoe-Finanserdquo 2013 vol 9 | nr 3 Financial Internet Quarterly bdquoe-Finanserdquo 2013 vol 9 | nr 3
be not coincidental Tereore models with two
exogenous variables were estimated Te estimated
models o earnings persistence two years later have
much lower quality than the models o earnings
persistence in the ollowing yearDaggerDaggerDagger Te highest
value o the adjusted determination coe fficient does
not exceed 012 Estimated values o coefficients α2
are significantly lower than in previous models
whereas the values o coefficients α3 are rising as
the number o observations decreases Starting rom
a model estimated on data rom 2001-2008 with
the exception o a model estimated on data rom
2005-2008 estimated coefficients o α3 are higher
than estimated coefficients o α2 For example in
2003-2008 the increase o return on total assets
o the company paying the dividend in year t by 1percentage point caused an increase in earnings in
year t + 2 to the value o assets in year t ndash 1 by 0787
percentage points while in the case o a company not
paying dividends the increase was only about 0324
percentage points Tese results are even stronger
DaggerDaggerDagger Also estimated by Skinner and Soltes models o earnings
persistence two years later were characterized by slightly lower
quality than earnings persistence models in the ollowing year still
the dierences were small
support or the argument that companies paying
dividends have higher earnings persistence quality
Te impact o dividend policy on improving earnings
persistence has been particularly evident in recent
years
C983151983150983139983148983157983155983145983151983150983155
Te study has shown that it is more clearly visible
on the Warsaw Stock Exchange than on developed
capital markets that companies paying dividends are
characterized by higher quality o earnings measured
by their persistence It could be said that Warsaw
Stock Exchange is ull justification or the motto o
this study that lsquodividends tell the truthrsquo mdash in this
case about the quality o a companyrsquos profit From
the other side the results o the presented studies are
biased with the small samples So the calculations
should be repeated in subsequent years with the
urther development o the Warsaw Stock Exchange
and the increase o the number o quoted stocks (the
increase o the number o observations)
Te presented method o evaluation o earnings
quality can be recommended to other emerging
markets
R983141983142983141983154983141983150983139983141983155
Baldau M amp Santos Silva J M C (2009 January) Onthe use o robust regression in econometrics University ofEssex Discussion Paper Series 664
Bartram S M Brown P How J C Y amp Verhoeven P(2009) Agency Conflicts and Corporate Payout PoliciesA Global Studyrdquo Retrieved romhttpssrncomabstract=1068281
Bhattacharaya S (1979) Imperect InormationsDividend Policy and lsquoTe Bird in the Handrsquo Fallacy Bell
Journal of Economics 10(1) 259-270
Brav A Graham J R Harvey C R amp Michaely R(2005 September) Payout policy in the 21st century
Journal of Financial of Economics 77(3) 483-527
Charemza W W amp Deadman D F (1997) New
Directions in Econometric Practice (2nd ed) Lyme NHEdward Elgar
Cottrell A amp Lucchetti R bdquoJackrdquo (2010) Gretl UserrsquosGuide Retrieved romhttpwwwgnuorglicenses-dlhtml
DeAngelo H DeAngelo L amp Skinner D J (2008)Corporate Payout Policy Foundations and rendsreg inFinance 3(2-3) 95ndash287
Dechow P Ge W amp Schrand C (2010 December)Understanding earnings quality a review o the proxiestheir determinants and their consequences Journal of
Accounting and Economics 50(2-3) 344ndash401
Denis D amp Osobov I (2008 July) Why do firms paydividends International evidence on the determinantso dividend policy Journal of Financial Economics 89(1)62-82
von Eije H amp Megginson W (2008 August) Dividendsand share repurchases in the European Union Journal ofFinancial Economics 89(2) 347-374
Fama E F amp French K F (2001 April) Disappearingdividends changing firm characteristics or lowerpropensity to pay Journal of Financial Economics 60(1)3-43
Gruszczyński M (2002) Modele i prognozy zmiennych jakościowych w finansach i bankowości Warszawa OficynaWydawnicza Szkoły Głoacutewnej Handlowej w Warszawie
Heckman J J (1976) Te common structure o statisticalmodels o truncation sample selection and a limited
dependent variables and a simple estimator or suchmodels Annales of Economic and Social Measurement 5(4) 120-137
Hellwig Z (1976) Przechodniość relacji skorelowaniazmiennych losowych i płynące stąd wnioskiekonometryczne Przegląd Statystyczny 23(1) 3-20
Hellwig Z (1977) Eekt katalizy w modeluekonometrycznym jego wykrywanie i usuwanie PrzeglądStatystyczny 24(2) 179-191
John K amp Williams J (1985 September) DividendsDilution and axes A Signaling Equilibrium Te Journalof Finance 40(4) 1053-1070
Kowalewski O Stetsyuk I amp alavera O (2007) Docorporate governance and ownership determine dividendpolicy in Poland Bank i Kredyt 38(11-12) 60-86
Kowerski M (2010) Wpływ opodatkowania na p olitykędywidend spoacutełek handlowych Ekonomista 3 409-422
Lintner J (1956 May) Distribution o Incomes oCorporation Among Dividends Retained Earnings andaxes American Economic Review 46(2) 97-113
Miller L (2006) Te Single Best Investment CreatingWealth with Dividend Growth Chicago IndependentPublishers Group
Miller M amp Modigliani F (1961 October) DividendPolicy Growth and Valuation o Shares Te Journal ofBusiness 34(4) 411-433
Myers S C amp Majlu N S (1984 June) Corporatefinancing and investment decisions when firms haveinormation that investors do not have Journal ofFinancial Economics 13(2) 187-221
Nowak E (2009) Wynik finansowy przedsiębiorstwa InK Czubakowska W Gabrusewicz amp E Nowak (Eds)Przychody mdash Koszty mdash Wynik finansowy przedsiębiorstwa Warszawa Polskie Wydawnictwo Ekonomiczne
Skinner D J amp Soltes E (2011 March) What DoDividends ell Us About Earnings Quality Review of
Accounting Studies 16(1) 1-28
Sloan R G (1996 July) Do Stock Prices Fully ReflectInormation in Accruals and Cash Flows About Future
Earnings Te Accounting Review 71(3) 289-315
Wąsowski W (2010) Kreatywna rachunkowość Fałszowanie sprawozdań finansowych Warszawa Difin
8172019 Dividends and Earnings Quality in Poland
httpslidepdfcomreaderfulldividends-and-earnings-quality-in-poland 66
Reproduced with permission of the copyright owner Further reproduction prohibited without
permission
8172019 Dividends and Earnings Quality in Poland
httpslidepdfcomreaderfulldividends-and-earnings-quality-in-poland 26
wwwe-nansecomUniversity of Information Technology and Management Sucharskiego 2 35-225 Rzeszoacutew
wwwe-nansecomUniversity of Information Technology and Management Sucharskiego 2 35-225 Rzeszoacutew
4544
Financial Internet Quarterly bdquoe-Finanserdquo 2013 vol 9 | nr 3 Financial Internet Quarterly bdquoe-Finanserdquo 2013 vol 9 | nr 3
achieve adequate (permanent) earnings Lintnerrsquos
conclusions have been confirmed by (Brav Graham
Harvey amp Michaely 2005) recently
It can thereore be hypothesized that dividends
provide inormation about earnings quality measured
by their persistence
Recent accounting scandals seem to confirm the
above hypothesis Although it is relatively easy
or management to lsquofixrsquo current profits and lsquopaintrsquo
a companyrsquos situation it is much more expensive
to pay dividends in order to inorm minority
shareholders and potential uture shareholders o
the good financial situation o the company and
its high level o profits when that level is a result o
lsquocreative accountingrsquo Management may decide on this
operation occasionally especially when that profit
is not a result o actual company perormance but
accounting interventions improving this result only
or a short period (eg through lsquoappropriatersquo booking
o liabilities at the end o the reporting period)
Skinner and Soltes (2011 p 14) in order to investigate
the relationship between dividend policy and earnings
quality suggested earnings linear models in years t +
1 and t + 2 determined by a decision to pay dividends
in year t the rate o assets return in year t and variable
product describing interaction o the decision to pay
dividends in year t and return on assets in year t Note
that earnings in year t t + 1 and t + 2 were related to
the assets at the end o year t ndash 1
E TA D E TA D E TAit it it it it it it it + minus minus minus
( ) = + + ( ) + sdot( ) +1 1 0 1 2 1 3 1
α α α α ε ε it
(3)
and
E TA D E TA D E TAit it it it it it it it + minus minus minus
( ) = + + ( ) + sdot( ) +2 1 0 1 2 1 3 1
α α α α ε ε it
(4)
where
E TAit it
minus1 mdash firmrsquos earnings in year t to total assets in the end o year t ndash 1 and in (return on assets)
E TAit it + minus1 1
mdash firmrsquos earnings in year t +1 in relation to total assets in the end o year t ndash 1 in
E TAit it + minus2 1
mdash firmrsquos earnings in year t +2 in relation to total assets in the end o year t ndash 1 in
Dit
mdash an indicator variable set to 1 i a dividend is paid by firm and in year t and 0 otherwise
D E TAit it it sdot
minus
( )1mdash interaction between D
it and E
it A
it ndash1
I we assume that variable Eit A
it ndash 1 is a measure o
total assets profitability then variable Dit (E
it A
it ndash 1 )
is profitability o companies paying dividends in
year t Tus the authors believe that earnings quality
is determined by their earnings persistence Hence
the above models can be called earnings persistence
models Model (3) indicates how persistent are
earnings achieved by companies in year t in the
ollowing year (t + 1) while the model (4) indicates
how persistent are earnings achieved by companies in
year t two years later (t + 2)
In this regression coefficient α2 measures the
persistence o earnings o all firms (irrespective i the
firm pays a dividend or not) Under the hypothesis
that dividends are inormative about the quality o
reported earnings it is expected that the coefficient
on earnings will be larger or dividend-paying firms
indicating that their earnings are more persistent
(α3 gt 0) Te sum o the coefficients α2 and α3 inorms
us about the earnings persistence o companies
paying dividends (Skinner amp Soltes 2011 p 14)
Skinner and Soltes analyzed a sample including all
non-utility non-financial domestic firms quoted
on the NYSE AMEX and NASDAQ rom 1974 to
2005 Altogether they gathered a total o 123728
observations (Skinner amp Soltes 2011 p 10) Tey
estimated all regressions using the ordinary least
squares method with two-way robust standard errors
(clustered by firm and time) to account or cross-
sectional and time series dependence Models were
estimated separately or the three sub-periods 1974-
1983 1984-1994 and 1994-2005
In models describing earnings persistence in year
t + 1 (model 3) values o estimated coefficients α2
oscillated around 08 (respectively 0781 0812
0835) and were significant at the 001 level which
according to the authors means that the profits are
airly persistent Furthermore these parameters
confirm the results o R G Sloanrsquos study (1996) who
received the value o the coefficient α2 = 084 Tere
is evidence o a modest increase in persistence over
time with the coefficient on earnings increasing
rom 078 in the earlier sub-period (1974 through
1983) to 084 in the most recent sub-period (1994
through 2005) Coefficients on a variable which
is the product o a decision o dividend paying
and earnings in relation to the value o assets
(Dit (E
it A
it ndash 1 )) are positive and statistically significant
in all three sub-periods (respectively 0031 0080
and 0064) It means that profits are more persistent
or dividend payers Since 1984 the sum o α2 and α3
coefficients or dividend payers has been around 090
Also similar results are provided by models
describing earnings persistence achieved in year t in
two years later (model 4) However coefficient values
on variables describing earnings α2 are a little lower
than in the previous models In this case the sum o
coefficients on the variables describing earnings ordividend payers in the period 1994ndash2005 is 0730
+ 0099 = 0829 (Skinner amp Soltes 2011 p 15) It
is worth emphasizing that the estimated models
are characterized by a high degree o explanation
Adjusted coefficients o determination (Adj R 2)
values or earnings persistence models o year t + 1
range rom 063 to 070 and or earnings persistence
models or year t + 2 are slightly worse ranging rom
043 to 054
Te estimation results have shown that dividend
payers have higher persistence (and thus quality) o
earnings than those not paying dividends and this
relationship does not depend on the level o dividends
paid
METHODS PROPOSED FOR
ASSESSING EARNINGS QUALITY
OF COMPANIES LISTED ON THE
WARSAW STOCK EXCHANGE
D983137983156983137
Te Warsaw Stock Exchange (WSE) is the most
dynamically growing market in Central and EasternEurope (Warsaw Stock Exchange Wiener Boumlrse
Prague Stock Exchange Budapest Stock Exchange
Bucharest Stock Exchange Bulgarian Stock
Exchange) Te WSE is the regional leader in terms o
key market ratios such as the value o equities trading
the number o domestic and oreign companies
the number o IPOs and since 2009 capitalization
which in the end o 2009 was 105 billion euro and
in the end o 2010 142 billion euro Te GDP share
o capitalization o domestic companies leapt rom
31 in the end o 2009 to 38 in the end o 2010 At
the end o 2009 the WSE Main List comprised 397
companies (354 domestic and 25 oreign) and at the
end o 2010 the number o quoted firms increased to
400 (373 domestic and 27 oreign)Dagger2
Te database o all domestic companies listed on
the Warsaw Stock Exchange rom 1995 to 2009 sect was
a starting point or the calculation It must be
considered that only the companies whose shares
were listed on the stock exchange throughout the
entire year were taken into account From the set o
domestic companies national investment unds were
excluded due to their different method o financial
reporting Some companies were removed which
in act were recorded throughout the year but were
excluded rom the stock exchanges in the first hal o
the next year4Moreover companies with negative
equity values and companies with zero net revenues
rom sales o products services goods and materials(not engaged in any operating activities in a certain
year) were excluded rom the calculation
With the development o the stock exchange
the number o companies admitted to the study
each year increased In 1995 the study included
44 companies while in 2008 293 companies In
this way cross-section datasets or 14 years were
obtained Every year this set consists o different
numbers o observations and can be analyzed or
each year separately Also annual data rom all years
can be combined and a set o pooled cross-sectional
time-series data can be obtained In total this set
consists o 2263 observations (companies ndash years)
It should be emphasized that in this pooled set each
observation ought to be treated as a separate entity
Te propensity to pay dividends o companies listed
on the Warsaw Stock Exchange is much lower than
in developed capital markets (Bartram Brown How
amp Verhoeven 2009 DeAngelo H DeAngelo L amp
Skinner 2008 von Eije amp Megginson 2008 Denis amp
Osobov 2008) But on the Warsaw Stock Exchange
the characteristic or developed capital market
process o lsquodisappearing dividendsrsquo (Fama amp French2001) was not observed Te research results suggest
the decision to equalize dividends and capital profits
tax rate rom 2004 and systematic CI reduction was
beneficial to increasing the share o companies paying
dividends in the total number o listed companies
Tis allowed companies to allocate larger earnings to
dividends (Kowerski 2010)
Dagger Warsaw Stock Exchange Annual Report 2010
httpwwwgpwplraporty_roczne (accessed 4 February 2012)sect Data rom Notoria Service httpirnotoriapl (accessed4 February 2012)
Such companies usually did not submit reports to Notoria
Service
8172019 Dividends and Earnings Quality in Poland
httpslidepdfcomreaderfulldividends-and-earnings-quality-in-poland 36
wwwe-nansecomUniversity of Information Technology and Management Sucharskiego 2 35-225 Rzeszoacutew
wwwe-nansecomUniversity of Information Technology and Management Sucharskiego 2 35-225 Rzeszoacutew
4746
Financial Internet Quarterly bdquoe-Finanserdquo 2013 vol 9 | nr 3 Financial Internet Quarterly bdquoe-Finanserdquo 2013 vol 9 | nr 3
Figure 1 Changes in the number of companies under survey from 1996 to 2009 and share of dividend payers
Source Own calculations
Te dividend policy o companies listed on the Warsaw
Stock Exchange has become more and more similar
to the behavior o companies in developed capital
markets Dividend value not only in current prices
but also in constant prices is growing rapidly In 2009
the average dividend payout made by a company
listed on the Warsaw Stock Exchange amounted to
843 million zloty and it was in current prices twenty
six times higher while in constant prices five times
higher than the average payout in 1992 But the
relation o dividends to GDP remains very low and
does not exceed 1 Also we can obser ve an increase
o payout concentration mdash a relatively small number
o major companies pay increasing dividends whichrepresent a significant part o all payouts Companies
pay an increasing share o profits which makes the
dividend payout ratio increase the dividend yield
ratio also increases
From 1996 to 2009 shares o companies paying
dividends underwent multidirectional changes Tey
were particularly high (above 40) rom 1996 to
1997 Ten they ell to a minimum in 2002 (215)
From 2003 to 2006 they increased again to 375
Since 2007 shares o dividend-paying companies
have been alling
M983141983156983144983151983140 983151983142 983141983155983156983145983149983137983156983145983151983150
Te method and models suggested and discussed by
Skinner and Soltes in the previous chapter were used
to test earnings quality o domestic companies listed
on the Warsaw Stock Exchange
According to model (3) a relation o earnings in year
t + 1 to value o total assets at the end o year t ndash 1 is
a unction o dividend in year t earnings in year t to
the value o total assets at the end o year t ndash 1 and the
product o two previous variables Tis means that to
calculate the values o dependent and independent
variables it is necessary to have data about companies
listed or the successive three years In a baseline
collection not all companies met this criteria
thereore only 1481 observations could be included
in the study According to model (4) earnings in year
t + 2 to the value o total assets at the end o year t ndash 1
is a unction o dividend in year t earnings in year t to
the value o total assets at the end o year t ndash 1 and the
product o the two previous variables Consequently
this means that to calculate the values o dependent
and independent variables it is necessary to have data
on companies listed or the successive our years Tis
limited the initial collection to 1195 observations
In both sets o data there are single outlier
observations o a dependent variable Observations
o this type can significantly change the final result
o the analysis and their disregard can be atal Te
simplest but quite effective method o lsquocopingrsquo with
outliers is to remove them rom the collection o data
under consideration which increases the robustness
o the estimated coefficients Estimators obtained in
this way are called lsquorobust estimatorsrsquo and estimated
models that can be called lsquorobust regressionrsquo denote
a set o estimation techniques which are less sensitive
than the ordinary least squares to the effect o
possible influential observations (Baldau amp Santos
Silva 2009 p 2)
In the present study observations or which
dependent variable values were lower thanndash100 or higher than 100 were removed As
a result the output set o observations or model 3
was reduced to 1468 while or model 4 to 1174
It must be emphasized that the method used or
selection o companies in the models (3) and (4)
especially or robust estimation can cause a sample
selection bias (Heckman 1976) with companies o
a slightly better economic and financial situation
Firms with negative equity and those that were not
listed or did not meet any o the criteria or creating
the output database respectively by three or our
successive years were removed rom samples On the
other hand the situation o companies excluded rom
the calculation usually so considerably deviates rom
the vast majority o companies included in samples
that their presence not only would not h elp to explain
changes in earnings quality but also could darken the
depiction o the phenomenondaggerdagger
Because o the act that the number o observations
in particular years are different and still there is no
inormation as to what time period (how many years)
observations should be selected recursive modeling
was used consisting o estimating consecutive modelso with increasingly shorter series emerging by
removing the oldest data every year (Charemza amp
Deadman 1997 p 62-65)
Te best method or estimating coefficients o
models (3) and (4) proved to be a heteroskedasticity-
corrected general least squares method
daggerdagger In the case o companies with negative equities some
indicators incorrectly inorm about the situation o the company
For example when such a company records a negative earning
(which is very probable) then the rate o return on equity is positive
mdash which could indicate a good inancial situation
THE RESULTS OF THEESTIMATION OF EARNINGSPERSISTENCE MODELS OFCOMPANIES LISTED ON THEWARSAW STOCK EXCHANGE
In 10 models o earnings persistence in the ollowing
year using recursive modeling and starting rom the
initial set o observations (1481 observations rom
1997 to 2008) coefficients on variable Dt were negative
and statistically insignificant A detailed analysis o
results leads to the conclusion that a catalysis effect
occurred (Hellwig 1977) which caused the lack o
coincidence o coefficients on Dt variable (Hellwig
1976)DaggerDagger Correlation coefficients between Et +1A
t ndash 1
and Dt variables are positive and coefficients on D
t
variable are negative
DaggerDagger Coeicient αi ulills the coincidence rule when sign αi = sign
ri where ri is correlation coeicient between dependent variable
and i-th independent variable
8172019 Dividends and Earnings Quality in Poland
httpslidepdfcomreaderfulldividends-and-earnings-quality-in-poland 46
wwwe-nansecomUniversity of Information Technology and Management Sucharskiego 2 35-225 Rzeszoacutew
wwwe-nansecomUniversity of Information Technology and Management Sucharskiego 2 35-225 Rzeszoacutew
4948
Financial Internet Quarterly bdquoe-Finanserdquo 2013 vol 9 | nr 3 Financial Internet Quarterly bdquoe-Finanserdquo 2013 vol 9 | nr 3
Table 1 The results of estimation of earnings persistence models in yeart + 1with heteroskedasticity-corrected general l east squares method
Source Own calculations in GREL (Corttrell amp Luchetti 2010)
Table 2 The results of estimation of earnings persistence models in yeart + 2with heteroskedasticity-corrected general l east squares method
Source Own calculations in GREL (Corttrell amp Luchetti 2010)
Only in the models or 2007-2008 and or 2008
parameters on the Dt variable were coincident but
also statistically insignificant Tereore it was
decided to reject the Dt variable and to estimate
models o earnings persistence in the ollowing year
dependent on the two other variables
Figure 2 Changes in values of coefficients in earnings persistence modelsin the following year (t + 1) estimated by recursive modeling method
Source Own calculations
Estimated by a recursive method coefficients are
characterized by very high stability Te model
estimated on data rom the years 2002 to 2008 (1039
observations) is characterized by the highest value o
the adjusted determination coefficient although this
value (02000) is much lower than in Skinner-Soltes
modelssectsect Tis model will serve or a more detailed
analysis
Estimated in this model coefficient α2 is 0467 and
thus is about 034 lower than estimated by Skinner-
Soltes coefficient α2 or the US Stock Exchanges
Furthermore estimated coefficient α3 is 0416 and ismany times higher than coefficient α3 estimated by
sectsect On the other hand the value o F statistic indicates the
signiicance o the multiple correlation coeicient R
( p = 300E-29) and thus the overall signiicant eect o both
variables on the earnings persistence in the ollowing year A
relatively low determination coeicient value in cross-section
models and cross-time models estimated on large sets o micro
data is quite common (Gruszczyński 2002 p 55)
It should be emphasized however that other earningspersistence models in the ollowing year obtained rom recursive
modeling are characterized by eatures similar to the discussed
model
Skinner-Soltes or the US Stock Exchangesdaggerdaggerdagger Tis
means that in Warsaw earnings o companies paying
dividends are more persistent than companies not
paying dividends than in New York In Warsaw
in year t the increase o return on total assets o
the company paying the dividend by 1 percentage
point caused the increase o earnings value in year
t + 1 to the value o total assets in year t ndash 1 by 0883
percentage points while in the case o a company not
paying dividends only by 0416 percentage points On
the New York Stock Exchanges the difference in avor
o companies paying dividends in the test period di dnot exceed 008 o a percentage point
Applying similar procedures as in the case o earnings
persistence models in the ollowing year models o
earnings persistence two years later were estimated
In this case removing 21 outliers provided models
that can be used to assess the phenomenon As in the
case o earnings persistence models in the ollowing
year again coefficients on the Dt variable proved to
daggerdaggerdagger O course comparisons o results obtained or the WarsawStock Exchange and or New York Stock Exchanges should be
treated with caution mainly because o the much smaller number
o observations on WSE
8172019 Dividends and Earnings Quality in Poland
httpslidepdfcomreaderfulldividends-and-earnings-quality-in-poland 56
wwwe-nansecomUniversity of Information Technology and Management Sucharskiego 2 35-225 Rzeszoacutew
wwwe-nansecomUniversity of Information Technology and Management Sucharskiego 2 35-225 Rzeszoacutew
5150
Financial Internet Quarterly bdquoe-Finanserdquo 2013 vol 9 | nr 3 Financial Internet Quarterly bdquoe-Finanserdquo 2013 vol 9 | nr 3
be not coincidental Tereore models with two
exogenous variables were estimated Te estimated
models o earnings persistence two years later have
much lower quality than the models o earnings
persistence in the ollowing yearDaggerDaggerDagger Te highest
value o the adjusted determination coe fficient does
not exceed 012 Estimated values o coefficients α2
are significantly lower than in previous models
whereas the values o coefficients α3 are rising as
the number o observations decreases Starting rom
a model estimated on data rom 2001-2008 with
the exception o a model estimated on data rom
2005-2008 estimated coefficients o α3 are higher
than estimated coefficients o α2 For example in
2003-2008 the increase o return on total assets
o the company paying the dividend in year t by 1percentage point caused an increase in earnings in
year t + 2 to the value o assets in year t ndash 1 by 0787
percentage points while in the case o a company not
paying dividends the increase was only about 0324
percentage points Tese results are even stronger
DaggerDaggerDagger Also estimated by Skinner and Soltes models o earnings
persistence two years later were characterized by slightly lower
quality than earnings persistence models in the ollowing year still
the dierences were small
support or the argument that companies paying
dividends have higher earnings persistence quality
Te impact o dividend policy on improving earnings
persistence has been particularly evident in recent
years
C983151983150983139983148983157983155983145983151983150983155
Te study has shown that it is more clearly visible
on the Warsaw Stock Exchange than on developed
capital markets that companies paying dividends are
characterized by higher quality o earnings measured
by their persistence It could be said that Warsaw
Stock Exchange is ull justification or the motto o
this study that lsquodividends tell the truthrsquo mdash in this
case about the quality o a companyrsquos profit From
the other side the results o the presented studies are
biased with the small samples So the calculations
should be repeated in subsequent years with the
urther development o the Warsaw Stock Exchange
and the increase o the number o quoted stocks (the
increase o the number o observations)
Te presented method o evaluation o earnings
quality can be recommended to other emerging
markets
R983141983142983141983154983141983150983139983141983155
Baldau M amp Santos Silva J M C (2009 January) Onthe use o robust regression in econometrics University ofEssex Discussion Paper Series 664
Bartram S M Brown P How J C Y amp Verhoeven P(2009) Agency Conflicts and Corporate Payout PoliciesA Global Studyrdquo Retrieved romhttpssrncomabstract=1068281
Bhattacharaya S (1979) Imperect InormationsDividend Policy and lsquoTe Bird in the Handrsquo Fallacy Bell
Journal of Economics 10(1) 259-270
Brav A Graham J R Harvey C R amp Michaely R(2005 September) Payout policy in the 21st century
Journal of Financial of Economics 77(3) 483-527
Charemza W W amp Deadman D F (1997) New
Directions in Econometric Practice (2nd ed) Lyme NHEdward Elgar
Cottrell A amp Lucchetti R bdquoJackrdquo (2010) Gretl UserrsquosGuide Retrieved romhttpwwwgnuorglicenses-dlhtml
DeAngelo H DeAngelo L amp Skinner D J (2008)Corporate Payout Policy Foundations and rendsreg inFinance 3(2-3) 95ndash287
Dechow P Ge W amp Schrand C (2010 December)Understanding earnings quality a review o the proxiestheir determinants and their consequences Journal of
Accounting and Economics 50(2-3) 344ndash401
Denis D amp Osobov I (2008 July) Why do firms paydividends International evidence on the determinantso dividend policy Journal of Financial Economics 89(1)62-82
von Eije H amp Megginson W (2008 August) Dividendsand share repurchases in the European Union Journal ofFinancial Economics 89(2) 347-374
Fama E F amp French K F (2001 April) Disappearingdividends changing firm characteristics or lowerpropensity to pay Journal of Financial Economics 60(1)3-43
Gruszczyński M (2002) Modele i prognozy zmiennych jakościowych w finansach i bankowości Warszawa OficynaWydawnicza Szkoły Głoacutewnej Handlowej w Warszawie
Heckman J J (1976) Te common structure o statisticalmodels o truncation sample selection and a limited
dependent variables and a simple estimator or suchmodels Annales of Economic and Social Measurement 5(4) 120-137
Hellwig Z (1976) Przechodniość relacji skorelowaniazmiennych losowych i płynące stąd wnioskiekonometryczne Przegląd Statystyczny 23(1) 3-20
Hellwig Z (1977) Eekt katalizy w modeluekonometrycznym jego wykrywanie i usuwanie PrzeglądStatystyczny 24(2) 179-191
John K amp Williams J (1985 September) DividendsDilution and axes A Signaling Equilibrium Te Journalof Finance 40(4) 1053-1070
Kowalewski O Stetsyuk I amp alavera O (2007) Docorporate governance and ownership determine dividendpolicy in Poland Bank i Kredyt 38(11-12) 60-86
Kowerski M (2010) Wpływ opodatkowania na p olitykędywidend spoacutełek handlowych Ekonomista 3 409-422
Lintner J (1956 May) Distribution o Incomes oCorporation Among Dividends Retained Earnings andaxes American Economic Review 46(2) 97-113
Miller L (2006) Te Single Best Investment CreatingWealth with Dividend Growth Chicago IndependentPublishers Group
Miller M amp Modigliani F (1961 October) DividendPolicy Growth and Valuation o Shares Te Journal ofBusiness 34(4) 411-433
Myers S C amp Majlu N S (1984 June) Corporatefinancing and investment decisions when firms haveinormation that investors do not have Journal ofFinancial Economics 13(2) 187-221
Nowak E (2009) Wynik finansowy przedsiębiorstwa InK Czubakowska W Gabrusewicz amp E Nowak (Eds)Przychody mdash Koszty mdash Wynik finansowy przedsiębiorstwa Warszawa Polskie Wydawnictwo Ekonomiczne
Skinner D J amp Soltes E (2011 March) What DoDividends ell Us About Earnings Quality Review of
Accounting Studies 16(1) 1-28
Sloan R G (1996 July) Do Stock Prices Fully ReflectInormation in Accruals and Cash Flows About Future
Earnings Te Accounting Review 71(3) 289-315
Wąsowski W (2010) Kreatywna rachunkowość Fałszowanie sprawozdań finansowych Warszawa Difin
8172019 Dividends and Earnings Quality in Poland
httpslidepdfcomreaderfulldividends-and-earnings-quality-in-poland 66
Reproduced with permission of the copyright owner Further reproduction prohibited without
permission
8172019 Dividends and Earnings Quality in Poland
httpslidepdfcomreaderfulldividends-and-earnings-quality-in-poland 36
wwwe-nansecomUniversity of Information Technology and Management Sucharskiego 2 35-225 Rzeszoacutew
wwwe-nansecomUniversity of Information Technology and Management Sucharskiego 2 35-225 Rzeszoacutew
4746
Financial Internet Quarterly bdquoe-Finanserdquo 2013 vol 9 | nr 3 Financial Internet Quarterly bdquoe-Finanserdquo 2013 vol 9 | nr 3
Figure 1 Changes in the number of companies under survey from 1996 to 2009 and share of dividend payers
Source Own calculations
Te dividend policy o companies listed on the Warsaw
Stock Exchange has become more and more similar
to the behavior o companies in developed capital
markets Dividend value not only in current prices
but also in constant prices is growing rapidly In 2009
the average dividend payout made by a company
listed on the Warsaw Stock Exchange amounted to
843 million zloty and it was in current prices twenty
six times higher while in constant prices five times
higher than the average payout in 1992 But the
relation o dividends to GDP remains very low and
does not exceed 1 Also we can obser ve an increase
o payout concentration mdash a relatively small number
o major companies pay increasing dividends whichrepresent a significant part o all payouts Companies
pay an increasing share o profits which makes the
dividend payout ratio increase the dividend yield
ratio also increases
From 1996 to 2009 shares o companies paying
dividends underwent multidirectional changes Tey
were particularly high (above 40) rom 1996 to
1997 Ten they ell to a minimum in 2002 (215)
From 2003 to 2006 they increased again to 375
Since 2007 shares o dividend-paying companies
have been alling
M983141983156983144983151983140 983151983142 983141983155983156983145983149983137983156983145983151983150
Te method and models suggested and discussed by
Skinner and Soltes in the previous chapter were used
to test earnings quality o domestic companies listed
on the Warsaw Stock Exchange
According to model (3) a relation o earnings in year
t + 1 to value o total assets at the end o year t ndash 1 is
a unction o dividend in year t earnings in year t to
the value o total assets at the end o year t ndash 1 and the
product o two previous variables Tis means that to
calculate the values o dependent and independent
variables it is necessary to have data about companies
listed or the successive three years In a baseline
collection not all companies met this criteria
thereore only 1481 observations could be included
in the study According to model (4) earnings in year
t + 2 to the value o total assets at the end o year t ndash 1
is a unction o dividend in year t earnings in year t to
the value o total assets at the end o year t ndash 1 and the
product o the two previous variables Consequently
this means that to calculate the values o dependent
and independent variables it is necessary to have data
on companies listed or the successive our years Tis
limited the initial collection to 1195 observations
In both sets o data there are single outlier
observations o a dependent variable Observations
o this type can significantly change the final result
o the analysis and their disregard can be atal Te
simplest but quite effective method o lsquocopingrsquo with
outliers is to remove them rom the collection o data
under consideration which increases the robustness
o the estimated coefficients Estimators obtained in
this way are called lsquorobust estimatorsrsquo and estimated
models that can be called lsquorobust regressionrsquo denote
a set o estimation techniques which are less sensitive
than the ordinary least squares to the effect o
possible influential observations (Baldau amp Santos
Silva 2009 p 2)
In the present study observations or which
dependent variable values were lower thanndash100 or higher than 100 were removed As
a result the output set o observations or model 3
was reduced to 1468 while or model 4 to 1174
It must be emphasized that the method used or
selection o companies in the models (3) and (4)
especially or robust estimation can cause a sample
selection bias (Heckman 1976) with companies o
a slightly better economic and financial situation
Firms with negative equity and those that were not
listed or did not meet any o the criteria or creating
the output database respectively by three or our
successive years were removed rom samples On the
other hand the situation o companies excluded rom
the calculation usually so considerably deviates rom
the vast majority o companies included in samples
that their presence not only would not h elp to explain
changes in earnings quality but also could darken the
depiction o the phenomenondaggerdagger
Because o the act that the number o observations
in particular years are different and still there is no
inormation as to what time period (how many years)
observations should be selected recursive modeling
was used consisting o estimating consecutive modelso with increasingly shorter series emerging by
removing the oldest data every year (Charemza amp
Deadman 1997 p 62-65)
Te best method or estimating coefficients o
models (3) and (4) proved to be a heteroskedasticity-
corrected general least squares method
daggerdagger In the case o companies with negative equities some
indicators incorrectly inorm about the situation o the company
For example when such a company records a negative earning
(which is very probable) then the rate o return on equity is positive
mdash which could indicate a good inancial situation
THE RESULTS OF THEESTIMATION OF EARNINGSPERSISTENCE MODELS OFCOMPANIES LISTED ON THEWARSAW STOCK EXCHANGE
In 10 models o earnings persistence in the ollowing
year using recursive modeling and starting rom the
initial set o observations (1481 observations rom
1997 to 2008) coefficients on variable Dt were negative
and statistically insignificant A detailed analysis o
results leads to the conclusion that a catalysis effect
occurred (Hellwig 1977) which caused the lack o
coincidence o coefficients on Dt variable (Hellwig
1976)DaggerDagger Correlation coefficients between Et +1A
t ndash 1
and Dt variables are positive and coefficients on D
t
variable are negative
DaggerDagger Coeicient αi ulills the coincidence rule when sign αi = sign
ri where ri is correlation coeicient between dependent variable
and i-th independent variable
8172019 Dividends and Earnings Quality in Poland
httpslidepdfcomreaderfulldividends-and-earnings-quality-in-poland 46
wwwe-nansecomUniversity of Information Technology and Management Sucharskiego 2 35-225 Rzeszoacutew
wwwe-nansecomUniversity of Information Technology and Management Sucharskiego 2 35-225 Rzeszoacutew
4948
Financial Internet Quarterly bdquoe-Finanserdquo 2013 vol 9 | nr 3 Financial Internet Quarterly bdquoe-Finanserdquo 2013 vol 9 | nr 3
Table 1 The results of estimation of earnings persistence models in yeart + 1with heteroskedasticity-corrected general l east squares method
Source Own calculations in GREL (Corttrell amp Luchetti 2010)
Table 2 The results of estimation of earnings persistence models in yeart + 2with heteroskedasticity-corrected general l east squares method
Source Own calculations in GREL (Corttrell amp Luchetti 2010)
Only in the models or 2007-2008 and or 2008
parameters on the Dt variable were coincident but
also statistically insignificant Tereore it was
decided to reject the Dt variable and to estimate
models o earnings persistence in the ollowing year
dependent on the two other variables
Figure 2 Changes in values of coefficients in earnings persistence modelsin the following year (t + 1) estimated by recursive modeling method
Source Own calculations
Estimated by a recursive method coefficients are
characterized by very high stability Te model
estimated on data rom the years 2002 to 2008 (1039
observations) is characterized by the highest value o
the adjusted determination coefficient although this
value (02000) is much lower than in Skinner-Soltes
modelssectsect Tis model will serve or a more detailed
analysis
Estimated in this model coefficient α2 is 0467 and
thus is about 034 lower than estimated by Skinner-
Soltes coefficient α2 or the US Stock Exchanges
Furthermore estimated coefficient α3 is 0416 and ismany times higher than coefficient α3 estimated by
sectsect On the other hand the value o F statistic indicates the
signiicance o the multiple correlation coeicient R
( p = 300E-29) and thus the overall signiicant eect o both
variables on the earnings persistence in the ollowing year A
relatively low determination coeicient value in cross-section
models and cross-time models estimated on large sets o micro
data is quite common (Gruszczyński 2002 p 55)
It should be emphasized however that other earningspersistence models in the ollowing year obtained rom recursive
modeling are characterized by eatures similar to the discussed
model
Skinner-Soltes or the US Stock Exchangesdaggerdaggerdagger Tis
means that in Warsaw earnings o companies paying
dividends are more persistent than companies not
paying dividends than in New York In Warsaw
in year t the increase o return on total assets o
the company paying the dividend by 1 percentage
point caused the increase o earnings value in year
t + 1 to the value o total assets in year t ndash 1 by 0883
percentage points while in the case o a company not
paying dividends only by 0416 percentage points On
the New York Stock Exchanges the difference in avor
o companies paying dividends in the test period di dnot exceed 008 o a percentage point
Applying similar procedures as in the case o earnings
persistence models in the ollowing year models o
earnings persistence two years later were estimated
In this case removing 21 outliers provided models
that can be used to assess the phenomenon As in the
case o earnings persistence models in the ollowing
year again coefficients on the Dt variable proved to
daggerdaggerdagger O course comparisons o results obtained or the WarsawStock Exchange and or New York Stock Exchanges should be
treated with caution mainly because o the much smaller number
o observations on WSE
8172019 Dividends and Earnings Quality in Poland
httpslidepdfcomreaderfulldividends-and-earnings-quality-in-poland 56
wwwe-nansecomUniversity of Information Technology and Management Sucharskiego 2 35-225 Rzeszoacutew
wwwe-nansecomUniversity of Information Technology and Management Sucharskiego 2 35-225 Rzeszoacutew
5150
Financial Internet Quarterly bdquoe-Finanserdquo 2013 vol 9 | nr 3 Financial Internet Quarterly bdquoe-Finanserdquo 2013 vol 9 | nr 3
be not coincidental Tereore models with two
exogenous variables were estimated Te estimated
models o earnings persistence two years later have
much lower quality than the models o earnings
persistence in the ollowing yearDaggerDaggerDagger Te highest
value o the adjusted determination coe fficient does
not exceed 012 Estimated values o coefficients α2
are significantly lower than in previous models
whereas the values o coefficients α3 are rising as
the number o observations decreases Starting rom
a model estimated on data rom 2001-2008 with
the exception o a model estimated on data rom
2005-2008 estimated coefficients o α3 are higher
than estimated coefficients o α2 For example in
2003-2008 the increase o return on total assets
o the company paying the dividend in year t by 1percentage point caused an increase in earnings in
year t + 2 to the value o assets in year t ndash 1 by 0787
percentage points while in the case o a company not
paying dividends the increase was only about 0324
percentage points Tese results are even stronger
DaggerDaggerDagger Also estimated by Skinner and Soltes models o earnings
persistence two years later were characterized by slightly lower
quality than earnings persistence models in the ollowing year still
the dierences were small
support or the argument that companies paying
dividends have higher earnings persistence quality
Te impact o dividend policy on improving earnings
persistence has been particularly evident in recent
years
C983151983150983139983148983157983155983145983151983150983155
Te study has shown that it is more clearly visible
on the Warsaw Stock Exchange than on developed
capital markets that companies paying dividends are
characterized by higher quality o earnings measured
by their persistence It could be said that Warsaw
Stock Exchange is ull justification or the motto o
this study that lsquodividends tell the truthrsquo mdash in this
case about the quality o a companyrsquos profit From
the other side the results o the presented studies are
biased with the small samples So the calculations
should be repeated in subsequent years with the
urther development o the Warsaw Stock Exchange
and the increase o the number o quoted stocks (the
increase o the number o observations)
Te presented method o evaluation o earnings
quality can be recommended to other emerging
markets
R983141983142983141983154983141983150983139983141983155
Baldau M amp Santos Silva J M C (2009 January) Onthe use o robust regression in econometrics University ofEssex Discussion Paper Series 664
Bartram S M Brown P How J C Y amp Verhoeven P(2009) Agency Conflicts and Corporate Payout PoliciesA Global Studyrdquo Retrieved romhttpssrncomabstract=1068281
Bhattacharaya S (1979) Imperect InormationsDividend Policy and lsquoTe Bird in the Handrsquo Fallacy Bell
Journal of Economics 10(1) 259-270
Brav A Graham J R Harvey C R amp Michaely R(2005 September) Payout policy in the 21st century
Journal of Financial of Economics 77(3) 483-527
Charemza W W amp Deadman D F (1997) New
Directions in Econometric Practice (2nd ed) Lyme NHEdward Elgar
Cottrell A amp Lucchetti R bdquoJackrdquo (2010) Gretl UserrsquosGuide Retrieved romhttpwwwgnuorglicenses-dlhtml
DeAngelo H DeAngelo L amp Skinner D J (2008)Corporate Payout Policy Foundations and rendsreg inFinance 3(2-3) 95ndash287
Dechow P Ge W amp Schrand C (2010 December)Understanding earnings quality a review o the proxiestheir determinants and their consequences Journal of
Accounting and Economics 50(2-3) 344ndash401
Denis D amp Osobov I (2008 July) Why do firms paydividends International evidence on the determinantso dividend policy Journal of Financial Economics 89(1)62-82
von Eije H amp Megginson W (2008 August) Dividendsand share repurchases in the European Union Journal ofFinancial Economics 89(2) 347-374
Fama E F amp French K F (2001 April) Disappearingdividends changing firm characteristics or lowerpropensity to pay Journal of Financial Economics 60(1)3-43
Gruszczyński M (2002) Modele i prognozy zmiennych jakościowych w finansach i bankowości Warszawa OficynaWydawnicza Szkoły Głoacutewnej Handlowej w Warszawie
Heckman J J (1976) Te common structure o statisticalmodels o truncation sample selection and a limited
dependent variables and a simple estimator or suchmodels Annales of Economic and Social Measurement 5(4) 120-137
Hellwig Z (1976) Przechodniość relacji skorelowaniazmiennych losowych i płynące stąd wnioskiekonometryczne Przegląd Statystyczny 23(1) 3-20
Hellwig Z (1977) Eekt katalizy w modeluekonometrycznym jego wykrywanie i usuwanie PrzeglądStatystyczny 24(2) 179-191
John K amp Williams J (1985 September) DividendsDilution and axes A Signaling Equilibrium Te Journalof Finance 40(4) 1053-1070
Kowalewski O Stetsyuk I amp alavera O (2007) Docorporate governance and ownership determine dividendpolicy in Poland Bank i Kredyt 38(11-12) 60-86
Kowerski M (2010) Wpływ opodatkowania na p olitykędywidend spoacutełek handlowych Ekonomista 3 409-422
Lintner J (1956 May) Distribution o Incomes oCorporation Among Dividends Retained Earnings andaxes American Economic Review 46(2) 97-113
Miller L (2006) Te Single Best Investment CreatingWealth with Dividend Growth Chicago IndependentPublishers Group
Miller M amp Modigliani F (1961 October) DividendPolicy Growth and Valuation o Shares Te Journal ofBusiness 34(4) 411-433
Myers S C amp Majlu N S (1984 June) Corporatefinancing and investment decisions when firms haveinormation that investors do not have Journal ofFinancial Economics 13(2) 187-221
Nowak E (2009) Wynik finansowy przedsiębiorstwa InK Czubakowska W Gabrusewicz amp E Nowak (Eds)Przychody mdash Koszty mdash Wynik finansowy przedsiębiorstwa Warszawa Polskie Wydawnictwo Ekonomiczne
Skinner D J amp Soltes E (2011 March) What DoDividends ell Us About Earnings Quality Review of
Accounting Studies 16(1) 1-28
Sloan R G (1996 July) Do Stock Prices Fully ReflectInormation in Accruals and Cash Flows About Future
Earnings Te Accounting Review 71(3) 289-315
Wąsowski W (2010) Kreatywna rachunkowość Fałszowanie sprawozdań finansowych Warszawa Difin
8172019 Dividends and Earnings Quality in Poland
httpslidepdfcomreaderfulldividends-and-earnings-quality-in-poland 66
Reproduced with permission of the copyright owner Further reproduction prohibited without
permission
8172019 Dividends and Earnings Quality in Poland
httpslidepdfcomreaderfulldividends-and-earnings-quality-in-poland 46
wwwe-nansecomUniversity of Information Technology and Management Sucharskiego 2 35-225 Rzeszoacutew
wwwe-nansecomUniversity of Information Technology and Management Sucharskiego 2 35-225 Rzeszoacutew
4948
Financial Internet Quarterly bdquoe-Finanserdquo 2013 vol 9 | nr 3 Financial Internet Quarterly bdquoe-Finanserdquo 2013 vol 9 | nr 3
Table 1 The results of estimation of earnings persistence models in yeart + 1with heteroskedasticity-corrected general l east squares method
Source Own calculations in GREL (Corttrell amp Luchetti 2010)
Table 2 The results of estimation of earnings persistence models in yeart + 2with heteroskedasticity-corrected general l east squares method
Source Own calculations in GREL (Corttrell amp Luchetti 2010)
Only in the models or 2007-2008 and or 2008
parameters on the Dt variable were coincident but
also statistically insignificant Tereore it was
decided to reject the Dt variable and to estimate
models o earnings persistence in the ollowing year
dependent on the two other variables
Figure 2 Changes in values of coefficients in earnings persistence modelsin the following year (t + 1) estimated by recursive modeling method
Source Own calculations
Estimated by a recursive method coefficients are
characterized by very high stability Te model
estimated on data rom the years 2002 to 2008 (1039
observations) is characterized by the highest value o
the adjusted determination coefficient although this
value (02000) is much lower than in Skinner-Soltes
modelssectsect Tis model will serve or a more detailed
analysis
Estimated in this model coefficient α2 is 0467 and
thus is about 034 lower than estimated by Skinner-
Soltes coefficient α2 or the US Stock Exchanges
Furthermore estimated coefficient α3 is 0416 and ismany times higher than coefficient α3 estimated by
sectsect On the other hand the value o F statistic indicates the
signiicance o the multiple correlation coeicient R
( p = 300E-29) and thus the overall signiicant eect o both
variables on the earnings persistence in the ollowing year A
relatively low determination coeicient value in cross-section
models and cross-time models estimated on large sets o micro
data is quite common (Gruszczyński 2002 p 55)
It should be emphasized however that other earningspersistence models in the ollowing year obtained rom recursive
modeling are characterized by eatures similar to the discussed
model
Skinner-Soltes or the US Stock Exchangesdaggerdaggerdagger Tis
means that in Warsaw earnings o companies paying
dividends are more persistent than companies not
paying dividends than in New York In Warsaw
in year t the increase o return on total assets o
the company paying the dividend by 1 percentage
point caused the increase o earnings value in year
t + 1 to the value o total assets in year t ndash 1 by 0883
percentage points while in the case o a company not
paying dividends only by 0416 percentage points On
the New York Stock Exchanges the difference in avor
o companies paying dividends in the test period di dnot exceed 008 o a percentage point
Applying similar procedures as in the case o earnings
persistence models in the ollowing year models o
earnings persistence two years later were estimated
In this case removing 21 outliers provided models
that can be used to assess the phenomenon As in the
case o earnings persistence models in the ollowing
year again coefficients on the Dt variable proved to
daggerdaggerdagger O course comparisons o results obtained or the WarsawStock Exchange and or New York Stock Exchanges should be
treated with caution mainly because o the much smaller number
o observations on WSE
8172019 Dividends and Earnings Quality in Poland
httpslidepdfcomreaderfulldividends-and-earnings-quality-in-poland 56
wwwe-nansecomUniversity of Information Technology and Management Sucharskiego 2 35-225 Rzeszoacutew
wwwe-nansecomUniversity of Information Technology and Management Sucharskiego 2 35-225 Rzeszoacutew
5150
Financial Internet Quarterly bdquoe-Finanserdquo 2013 vol 9 | nr 3 Financial Internet Quarterly bdquoe-Finanserdquo 2013 vol 9 | nr 3
be not coincidental Tereore models with two
exogenous variables were estimated Te estimated
models o earnings persistence two years later have
much lower quality than the models o earnings
persistence in the ollowing yearDaggerDaggerDagger Te highest
value o the adjusted determination coe fficient does
not exceed 012 Estimated values o coefficients α2
are significantly lower than in previous models
whereas the values o coefficients α3 are rising as
the number o observations decreases Starting rom
a model estimated on data rom 2001-2008 with
the exception o a model estimated on data rom
2005-2008 estimated coefficients o α3 are higher
than estimated coefficients o α2 For example in
2003-2008 the increase o return on total assets
o the company paying the dividend in year t by 1percentage point caused an increase in earnings in
year t + 2 to the value o assets in year t ndash 1 by 0787
percentage points while in the case o a company not
paying dividends the increase was only about 0324
percentage points Tese results are even stronger
DaggerDaggerDagger Also estimated by Skinner and Soltes models o earnings
persistence two years later were characterized by slightly lower
quality than earnings persistence models in the ollowing year still
the dierences were small
support or the argument that companies paying
dividends have higher earnings persistence quality
Te impact o dividend policy on improving earnings
persistence has been particularly evident in recent
years
C983151983150983139983148983157983155983145983151983150983155
Te study has shown that it is more clearly visible
on the Warsaw Stock Exchange than on developed
capital markets that companies paying dividends are
characterized by higher quality o earnings measured
by their persistence It could be said that Warsaw
Stock Exchange is ull justification or the motto o
this study that lsquodividends tell the truthrsquo mdash in this
case about the quality o a companyrsquos profit From
the other side the results o the presented studies are
biased with the small samples So the calculations
should be repeated in subsequent years with the
urther development o the Warsaw Stock Exchange
and the increase o the number o quoted stocks (the
increase o the number o observations)
Te presented method o evaluation o earnings
quality can be recommended to other emerging
markets
R983141983142983141983154983141983150983139983141983155
Baldau M amp Santos Silva J M C (2009 January) Onthe use o robust regression in econometrics University ofEssex Discussion Paper Series 664
Bartram S M Brown P How J C Y amp Verhoeven P(2009) Agency Conflicts and Corporate Payout PoliciesA Global Studyrdquo Retrieved romhttpssrncomabstract=1068281
Bhattacharaya S (1979) Imperect InormationsDividend Policy and lsquoTe Bird in the Handrsquo Fallacy Bell
Journal of Economics 10(1) 259-270
Brav A Graham J R Harvey C R amp Michaely R(2005 September) Payout policy in the 21st century
Journal of Financial of Economics 77(3) 483-527
Charemza W W amp Deadman D F (1997) New
Directions in Econometric Practice (2nd ed) Lyme NHEdward Elgar
Cottrell A amp Lucchetti R bdquoJackrdquo (2010) Gretl UserrsquosGuide Retrieved romhttpwwwgnuorglicenses-dlhtml
DeAngelo H DeAngelo L amp Skinner D J (2008)Corporate Payout Policy Foundations and rendsreg inFinance 3(2-3) 95ndash287
Dechow P Ge W amp Schrand C (2010 December)Understanding earnings quality a review o the proxiestheir determinants and their consequences Journal of
Accounting and Economics 50(2-3) 344ndash401
Denis D amp Osobov I (2008 July) Why do firms paydividends International evidence on the determinantso dividend policy Journal of Financial Economics 89(1)62-82
von Eije H amp Megginson W (2008 August) Dividendsand share repurchases in the European Union Journal ofFinancial Economics 89(2) 347-374
Fama E F amp French K F (2001 April) Disappearingdividends changing firm characteristics or lowerpropensity to pay Journal of Financial Economics 60(1)3-43
Gruszczyński M (2002) Modele i prognozy zmiennych jakościowych w finansach i bankowości Warszawa OficynaWydawnicza Szkoły Głoacutewnej Handlowej w Warszawie
Heckman J J (1976) Te common structure o statisticalmodels o truncation sample selection and a limited
dependent variables and a simple estimator or suchmodels Annales of Economic and Social Measurement 5(4) 120-137
Hellwig Z (1976) Przechodniość relacji skorelowaniazmiennych losowych i płynące stąd wnioskiekonometryczne Przegląd Statystyczny 23(1) 3-20
Hellwig Z (1977) Eekt katalizy w modeluekonometrycznym jego wykrywanie i usuwanie PrzeglądStatystyczny 24(2) 179-191
John K amp Williams J (1985 September) DividendsDilution and axes A Signaling Equilibrium Te Journalof Finance 40(4) 1053-1070
Kowalewski O Stetsyuk I amp alavera O (2007) Docorporate governance and ownership determine dividendpolicy in Poland Bank i Kredyt 38(11-12) 60-86
Kowerski M (2010) Wpływ opodatkowania na p olitykędywidend spoacutełek handlowych Ekonomista 3 409-422
Lintner J (1956 May) Distribution o Incomes oCorporation Among Dividends Retained Earnings andaxes American Economic Review 46(2) 97-113
Miller L (2006) Te Single Best Investment CreatingWealth with Dividend Growth Chicago IndependentPublishers Group
Miller M amp Modigliani F (1961 October) DividendPolicy Growth and Valuation o Shares Te Journal ofBusiness 34(4) 411-433
Myers S C amp Majlu N S (1984 June) Corporatefinancing and investment decisions when firms haveinormation that investors do not have Journal ofFinancial Economics 13(2) 187-221
Nowak E (2009) Wynik finansowy przedsiębiorstwa InK Czubakowska W Gabrusewicz amp E Nowak (Eds)Przychody mdash Koszty mdash Wynik finansowy przedsiębiorstwa Warszawa Polskie Wydawnictwo Ekonomiczne
Skinner D J amp Soltes E (2011 March) What DoDividends ell Us About Earnings Quality Review of
Accounting Studies 16(1) 1-28
Sloan R G (1996 July) Do Stock Prices Fully ReflectInormation in Accruals and Cash Flows About Future
Earnings Te Accounting Review 71(3) 289-315
Wąsowski W (2010) Kreatywna rachunkowość Fałszowanie sprawozdań finansowych Warszawa Difin
8172019 Dividends and Earnings Quality in Poland
httpslidepdfcomreaderfulldividends-and-earnings-quality-in-poland 66
Reproduced with permission of the copyright owner Further reproduction prohibited without
permission
8172019 Dividends and Earnings Quality in Poland
httpslidepdfcomreaderfulldividends-and-earnings-quality-in-poland 56
wwwe-nansecomUniversity of Information Technology and Management Sucharskiego 2 35-225 Rzeszoacutew
wwwe-nansecomUniversity of Information Technology and Management Sucharskiego 2 35-225 Rzeszoacutew
5150
Financial Internet Quarterly bdquoe-Finanserdquo 2013 vol 9 | nr 3 Financial Internet Quarterly bdquoe-Finanserdquo 2013 vol 9 | nr 3
be not coincidental Tereore models with two
exogenous variables were estimated Te estimated
models o earnings persistence two years later have
much lower quality than the models o earnings
persistence in the ollowing yearDaggerDaggerDagger Te highest
value o the adjusted determination coe fficient does
not exceed 012 Estimated values o coefficients α2
are significantly lower than in previous models
whereas the values o coefficients α3 are rising as
the number o observations decreases Starting rom
a model estimated on data rom 2001-2008 with
the exception o a model estimated on data rom
2005-2008 estimated coefficients o α3 are higher
than estimated coefficients o α2 For example in
2003-2008 the increase o return on total assets
o the company paying the dividend in year t by 1percentage point caused an increase in earnings in
year t + 2 to the value o assets in year t ndash 1 by 0787
percentage points while in the case o a company not
paying dividends the increase was only about 0324
percentage points Tese results are even stronger
DaggerDaggerDagger Also estimated by Skinner and Soltes models o earnings
persistence two years later were characterized by slightly lower
quality than earnings persistence models in the ollowing year still
the dierences were small
support or the argument that companies paying
dividends have higher earnings persistence quality
Te impact o dividend policy on improving earnings
persistence has been particularly evident in recent
years
C983151983150983139983148983157983155983145983151983150983155
Te study has shown that it is more clearly visible
on the Warsaw Stock Exchange than on developed
capital markets that companies paying dividends are
characterized by higher quality o earnings measured
by their persistence It could be said that Warsaw
Stock Exchange is ull justification or the motto o
this study that lsquodividends tell the truthrsquo mdash in this
case about the quality o a companyrsquos profit From
the other side the results o the presented studies are
biased with the small samples So the calculations
should be repeated in subsequent years with the
urther development o the Warsaw Stock Exchange
and the increase o the number o quoted stocks (the
increase o the number o observations)
Te presented method o evaluation o earnings
quality can be recommended to other emerging
markets
R983141983142983141983154983141983150983139983141983155
Baldau M amp Santos Silva J M C (2009 January) Onthe use o robust regression in econometrics University ofEssex Discussion Paper Series 664
Bartram S M Brown P How J C Y amp Verhoeven P(2009) Agency Conflicts and Corporate Payout PoliciesA Global Studyrdquo Retrieved romhttpssrncomabstract=1068281
Bhattacharaya S (1979) Imperect InormationsDividend Policy and lsquoTe Bird in the Handrsquo Fallacy Bell
Journal of Economics 10(1) 259-270
Brav A Graham J R Harvey C R amp Michaely R(2005 September) Payout policy in the 21st century
Journal of Financial of Economics 77(3) 483-527
Charemza W W amp Deadman D F (1997) New
Directions in Econometric Practice (2nd ed) Lyme NHEdward Elgar
Cottrell A amp Lucchetti R bdquoJackrdquo (2010) Gretl UserrsquosGuide Retrieved romhttpwwwgnuorglicenses-dlhtml
DeAngelo H DeAngelo L amp Skinner D J (2008)Corporate Payout Policy Foundations and rendsreg inFinance 3(2-3) 95ndash287
Dechow P Ge W amp Schrand C (2010 December)Understanding earnings quality a review o the proxiestheir determinants and their consequences Journal of
Accounting and Economics 50(2-3) 344ndash401
Denis D amp Osobov I (2008 July) Why do firms paydividends International evidence on the determinantso dividend policy Journal of Financial Economics 89(1)62-82
von Eije H amp Megginson W (2008 August) Dividendsand share repurchases in the European Union Journal ofFinancial Economics 89(2) 347-374
Fama E F amp French K F (2001 April) Disappearingdividends changing firm characteristics or lowerpropensity to pay Journal of Financial Economics 60(1)3-43
Gruszczyński M (2002) Modele i prognozy zmiennych jakościowych w finansach i bankowości Warszawa OficynaWydawnicza Szkoły Głoacutewnej Handlowej w Warszawie
Heckman J J (1976) Te common structure o statisticalmodels o truncation sample selection and a limited
dependent variables and a simple estimator or suchmodels Annales of Economic and Social Measurement 5(4) 120-137
Hellwig Z (1976) Przechodniość relacji skorelowaniazmiennych losowych i płynące stąd wnioskiekonometryczne Przegląd Statystyczny 23(1) 3-20
Hellwig Z (1977) Eekt katalizy w modeluekonometrycznym jego wykrywanie i usuwanie PrzeglądStatystyczny 24(2) 179-191
John K amp Williams J (1985 September) DividendsDilution and axes A Signaling Equilibrium Te Journalof Finance 40(4) 1053-1070
Kowalewski O Stetsyuk I amp alavera O (2007) Docorporate governance and ownership determine dividendpolicy in Poland Bank i Kredyt 38(11-12) 60-86
Kowerski M (2010) Wpływ opodatkowania na p olitykędywidend spoacutełek handlowych Ekonomista 3 409-422
Lintner J (1956 May) Distribution o Incomes oCorporation Among Dividends Retained Earnings andaxes American Economic Review 46(2) 97-113
Miller L (2006) Te Single Best Investment CreatingWealth with Dividend Growth Chicago IndependentPublishers Group
Miller M amp Modigliani F (1961 October) DividendPolicy Growth and Valuation o Shares Te Journal ofBusiness 34(4) 411-433
Myers S C amp Majlu N S (1984 June) Corporatefinancing and investment decisions when firms haveinormation that investors do not have Journal ofFinancial Economics 13(2) 187-221
Nowak E (2009) Wynik finansowy przedsiębiorstwa InK Czubakowska W Gabrusewicz amp E Nowak (Eds)Przychody mdash Koszty mdash Wynik finansowy przedsiębiorstwa Warszawa Polskie Wydawnictwo Ekonomiczne
Skinner D J amp Soltes E (2011 March) What DoDividends ell Us About Earnings Quality Review of
Accounting Studies 16(1) 1-28
Sloan R G (1996 July) Do Stock Prices Fully ReflectInormation in Accruals and Cash Flows About Future
Earnings Te Accounting Review 71(3) 289-315
Wąsowski W (2010) Kreatywna rachunkowość Fałszowanie sprawozdań finansowych Warszawa Difin
8172019 Dividends and Earnings Quality in Poland
httpslidepdfcomreaderfulldividends-and-earnings-quality-in-poland 66
Reproduced with permission of the copyright owner Further reproduction prohibited without
permission
8172019 Dividends and Earnings Quality in Poland
httpslidepdfcomreaderfulldividends-and-earnings-quality-in-poland 66
Reproduced with permission of the copyright owner Further reproduction prohibited without
permission