IBIMA Publishing
Journal of Eastern Europe Research in Business & Economics
http://www.ibimapublishing.com/journals/JEERBE/jeerbe.html
Vol. 202 !202"# $rticle %& '('')# 0 pages
&*%: 0.+)/202.'('')
Copyright 2012 Vintil Georgeta and Toroap Maria Georgia. This is an open access article distributed
under the Creative Commons ttribution !icense unported ".0# $hich permits unrestricted use# distribution#
and reproduction in any medium# provided that original $or% is properly cited. Contact author& Vintil
Georgeta '(mail& vintilageorgeta)yahoo.*r
Bankruptcy Prediction Model for Listed
Companies in Romania
Vintil Georgeta and Toroap Maria Georgia
Academy of Economic Studies, Bucharest, Romania
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Abstract
,(score model is one o* the most *re-uently used model *or early *inancial *ailure $arning and
considers various *inancial ratios selected as prediction variables.The purpose o* this paper is to
use multivariant discriminant analysis .M/0 to substantiate a score *unction e**ective in
ban%ruptcy ris% prediction o* enterprises on 1omanian economy e2ample. 3n order to discriminate
bet$een ban%rupt and non(ban%rupt in the scoring model $e used relevant *inancial ratios related
to activity# li-uidity# leverage and pro*itability. The $eighting coe**icients established bet$een
independent variables and the ob4ective *unction(score# are determined by using statistical tools. 3n
this conte2t# the article aims to build a scoring *unction in order to identi*y ban%rupt companies#
using a sample o* companies listed on 5ucharest 6toc% '2change. The results in this article can be
used to appraise the e**ectiveness o* applying M/ *inancial *ailure models *or 1omanian
companies# to ma%e an idea about curent and *uture *inancial situation# and ta%e# i* necessary#
corrective measures.
Key words: discriminant analysis# ban%ruptcy# prediction# *inancial ratios# score.
JEL Classification: C"7# G""# G18# G"2# C92.
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
ntroduction
:inancial *ailure prediction is a critical *actor
in developing strong(built capital mar%ets in
most o* the capitalistic countries and
numerous studies in the *ield o* economic
and *inancial analysis *ocus on ban%ruptcy
early $arning. Creditors and investors are
greatly concerned $ith the possibility o*
company;s ban%ruptcy. <hile the lenders are
interested in the credit $orthiness o* the
*irm# shareholders are more heavily involved
$ith pro*its and dividends prospects.
:inancial statements are a valuable source o*
in*ormation *or the decision ma%ers as a
benchmar% *or *uture targets and pro4ections.
Most o* corporate management studies
include a section about $hy companies *all
into ban%ruptcy. 6latter .1=7>0# *or e2ample#
identi*ied eleven *actors as the principal
causes o* corporate decline. The *irst one is
the poor management *actor $hich may
emerge as sheer incompetence or lac% o*
interest in the top management. part *rom
poor management# another ma4or *actor *or
corporate decline is the inade-uate *inancial
control $hich occurs in the absence o* or
inade-uacy o* cash(*lo$ *orecasts# costing
systems and budgetary control. ?ther
*inancial causes o* decline are high gearing#
conservative *inancial policies# the use o*
inappropriate *inancing sources# high cost
structure# adverse movements in commodity
prices and overtrading. 3n addition#
competition bet$een *irms# irresponsiveness
to mar%et demand changes# lac% o* mar%eting
e**ort# launching big pro4ects $ithout prior
@ournal o* 'astern 'urope 1esearch in 5usiness A 'conomics 2
planning and not(properly scrutiniBed
ac-uisitions might potentially give rise to
corporate *ailures as $ell.
3ncluded by theorists in Cpro*essional
applications o* *inancial analysisD or by
practitioners in Eother methods o* *inancial
analysisD# discriminate analysis can be also
seen as a method belonging to the stage o*
maturity o* *inancial analysis. The scoring is a
method that involves internal and e2ternal
diagnostic and re-uires ris% interpretation
*or the investor# company;s creditor# but also
*or the enterprise as a system in *urther
activities. 3t is based on a value 4udgment
development that combines a relevant group
o* *inancial ratios .or variables0.
The purpose o* this article is to revie$ and
e2amine the main early $arning ban%ruptcy
approaches# particularly the F, modelsG that
are utiliBed as valuable symptoms o*
potential *ailure. The article consists o* *ive
parts. The 3ntroduction is *ollo$ed by
sections covering !iterature 1evie$#
Methodology# 1esults and Conclusions.
Literature Re!iew
6cience(based models *or ban%ruptcy
prediction have been developed *or the *irst
time in H.6. in the GI0s by <.J. 5eaver and
'.3. ltman. 5eaver *ound that several
indicators could success*ully distinguish a
sample o* ban%rupt companies to another
one $ithout *inancial di**iculties# obtaining
conclusive results *or a period o* up to *ive
years be*ore the onset o* ban%ruptcy. 5eaver
e2amined *inancial ratios independently#
$ithout ta%ing into account the e2isting lin%s
bet$een them# resulting in a univariate or
one(dimensional analysis. Je has# ho$ever#
the ground $or% *or multivariate analysis#
$hich $as developed by ltman and by other
economists# $hose outcome is to develop a
scoring model based on a combination o*
rates that distinguish the most ris%y
businesses compared to the healthy ones.
'.3. ltman developed in 1=I7# a multivariate
analysis o* ban%ruptcy# combining *ive
*inancial ratios in a single *unction#
populariBed as the score or ,(score model.
6ubse-uently# this model $as improved and
published under the name C,eta nalysisD#
standing at the base o* other ban%ruptcy
prediction models.
The ,(score model suggested by ltman is
based on discriminant analysis# $hich is used
to develop models o* classi*ication and
prediction o* observations belonging to
certain groups determined a priori. To this
end# the discriminant analysis builds a
classi*ier based on a set o* observations and
indicators characteristic *or these
observations. 3n the case o* ltman model the
set o* observations is represented by a
number o* companies classi*ied by the author
in solvent and insolvent# and the considered
indicators are certain *inancial ratios based
upon the *inancial situation o* companies is
analyBed.
?hlson .1=700 uses a logit model# $ith less
restrictive assumptions than those ta%en by
the M/ approach. ,mi4e$s%i .1=7>0 adopts
a probit approach that is also based on
accounting data but uses a di**erent set o*
independent variables. ll o* these
approaches predict *uture ban%ruptcy based
on accounting ratios e2tracted *rom publicly
available *inancial statements.
6hum$ay .20010 proposed a discrete(time
haBard model to predict a company;s
ban%ruptcy using both accounting and
mar%et variables. The main di**erence
bet$een this model and the static logit model
is that the haBard model can be estimated
$ithin the logit *rame$or% $hile using the
entire li*e span o* in*ormation .all company(
years0 *or each *irm. 5y contrast# the static
logit model incorporates only one *irm(year
*or each observation .i.e.# each observation
consists o* a single set o* variables observed
at a single point in time0.
" @ournal o* 'astern 'urope 1esearch in 5usiness A 'conomics
nother re*erence o* the ban%ruptcy
prediction literature *ocuses on mar%et(
based in*ormation. mong others# Jillegeist
.200>0 has developed a 56M(Krob
ban%ruptcy prediction model that is based on
the 5lac%L6cholesLMerton option pricing
model. Their results indicate that the 56M(
Krob model outper*orms the models o*
ltman .1=I70 and ?hlson .1=700 in a series
o* tests.
1ecent papers ta%e into consideration also
various *irm(characteristics that may be
use*ul additional predictors o* *uture
ban%ruptcy are ta%en. :or e2ample /enis et
al. .1==80 measure corporate diversi*ication
by the number o* business segments. 5eaver
et al. .20090 propose that# other things e-ual#
large *irms have a smaller probability o*
ban%ruptcy and that a part o* this
e2planation is related to corporate
diversi*ication. That is# corporate
diversi*ication and *irm(siBe are t$o *irm(
characteristics that may help to predict
*uture ban%ruptcy. Ver$i4meren et al. .20100
remar% that companies $ith strong interest
in employee $ell(being reduce the chance o*
ban%ruptcy.
Jillegeist et al. .200>0 compare the
per*ormance o* their 56M(Krob model
against the ltman and ?hlson models in a
series o* in(sample and out(o*(sample tests#
concluding that the 56M model outper*orms
the accounting(based models. 6imilarly#
Chava and @arro$ .200>0 e2amine the
relative per*ormance o* 6hum$ay;s haBard
model against the ltman and ,mi4e$s%i
models# concluding that the haBard model
outper*orms static logit models.
1omania has e2pressed also an interest in
obtaining a synthetic tool to *orecast the ris%
o* ban%ruptcy in ban%s and businesses. 3n
this regard# should be mentioned& the 5 6core
:unction .1==70# developed by /.
5ileMteanu# Model 3 .1==70 built by 3voniciu
.similar to 5 6core :unction0# Model .20020
3on nghel;s outcome on the 1omanian
economy.
5an%ruptcy prediction models using M/ in
1omanian economy conte2t are also
highlighted in latest studies .Vintil and
Toroap# 20110 in order to achieve
discrimination bet$een ban%rupt and non(
ban%rupt in the scoring model based on
relevant *inancial ratios related to activity
.stoc% rotation# receivable collection# debt
payment#assets rotation0# li-uidity.current
li-uidity -uic% li-uidity0# leverage .e-uity
debt# cash(*lo$ to debt# debt to total assets0
and pro*itability .return on sales# return on
assets# return on e-uity# return on revenue0.
nother recent study *or ban%ruptcy ris%
estimation .rmeanu# Vintil et al.# 20120
achieves eligible results $ith the built score
model based on *inancial ratios *or the
1omanian economy;s *rame$or%.
6coring models *ramed so *ar have the
disadvantage that can be applied only in
economies $here the statistical study $as
conducted .or branch or sector analyBed0#
their use cannot be generaliBed territorially.
There$ith# periods mar%ed by economic
instability change the correlations captured
by the score *unction developed# $hich limits
the temporal use o* these models# re-uiring a
re(enactment at regular intervals.
Researc" Met"odolo#y
The purpose o* this paper is to use
discriminant analysis to substantiate a score
*unction e**ective in ban%ruptcy ris%
prediction o* enterprises on 1omanian
economy e2ample. :or achieving
discrimination bet$een ban%rupt and non(
ban%rupt in the scoring model $e used
relevant *inancial ratios related to activity
.stoc% rotation# receivable collection# debt
payment#assets rotation0# li-uidity.current
li-uidity -uic% li-uidity0# leverage .e-uity
debt# cash(*lo$ to debt# debt to total assets0
and pro*itability .return on sales# return on
assets# return on e-uity# return on revenue0.
@ournal o* 'astern 'urope 1esearch in 5usiness A 'conomics >
The research $as conducted on a sample o*
90 companies listed on the 5ucharest 6toc%
'2change# out o* $hich 28 $ere used to build
the score *unction and 2" *or a posterior
analysis in determining the success rate o* it.
This analysis is based on *inancial
in*ormation e2tracted *rom annual reports o*
companies .balance sheet# pro*it and loss# the
nne2es to the *inancial statements0 *or the
year 2007.
The sample considered in building the score
*unction includes 28 1omanian companies
both *rom public and private out o* $hich 1"
companies $ithout *inancial problems and
1> companies are ban%rupt or in di**iculty .it
$as opened the insolvency proceeding0
according to in*ormation posted on the
Ministry o* :inance $ebsite.
/etailed analysis o* the sample allo$ed
appraisal o* the obvious di**erences bet$een
the t$o groups o* enterprises. :or this
analysis there have been used the average
and median o* the *inancial ratios# most
relevant being the median rates $hich
eliminate unusual values encountered in
some cases .see Table 10.
$able %: A!era#e &A'() and Median &M*+) of t"e ,inancial Ratios for t"e $wo (roups +on-
Bankrupt.Bankrupt
,inancial Ratios
+-B B
A'( M*+ A'( M*+
Current li-uidity >.>= 1.72 0.I8 0.I9
Nuic% li-uidity >.01 1."9 0.>I 0."I
'-uity debt 0.1I 0.01 (0.1= 0.0"
Cash(*lo$ to debt 1.2" 0.00 (0.01 0.00
/ebt to asset 2=O "0O =1O 7"O
6toc% rotation .days0 "= >9 8" 81
1eceivable collection .days0 1"8 108 12> 7=
Total debt payment .days0 1I1 128 80= >II
6upplier payment .days0 >> 27 128 119
sset rotation 0.81 0.I" 0.I8 0.I1
1eturn on sales 1IO =O ("0O (17O
1eturn on e-uity =O 7O (1"2O ("2O
1eturn on asset 10O 10O (72O (11O
1eturn on revenue 1>O IO ("9O (21O
.?$n calculations0
:ollo$ing the selection step *or the
discrimination o* the t$o categories $e %ept
the *ollo$ing *ive *inancial variables& current
li-uidity# return on asset# return on revenues#
debt to asset# total debt payment.
Current !i-uidity P Current ssets Q Current
!iabilities
1eturn on sset P Gross Kro*it Q .'-uity R
!ong Term !iabilities0
1eturn on 1evenues P Set Kro*it Q Total
1evenues
/ebt to sset P Total /ebt Q Total ssets
9 @ournal o* 'astern 'urope 1esearch in 5usiness A 'conomics
Total /ebt Kayment .days0 P .Total /ebt Q
Turnover0 T "I0 .
Researc" Results
The model has the *ollo$ing *unction&
,P:unction.1entability o*
1evenues#1entability o* ssets# /ebt to sset#
Current !i-uidity# Total /ebt Kayment0.
The model $hich describes the relationship
bet$een the rentability o* incomes#
rentability o* assets# leverage global rate#
li-uidity ratio and payment obligations is a
strong and linear relationship. The model is
valid.Three parameters are nor statistical
signi*icant# but the model could be revised.
The /urbin(<atson test is 1.8># $hich means
that there is indecision# it is recommending
to accept the positive autocorrelation o*
residuals. The value *or coe**icient o*
determination is I7O# $hich means that the
model is e2plained in I7O o* the e2ogenous
variables. :or chec%ing the homos%edasticity
hypothesis *or this model $ill be using <hite
test. <hite(test involves the *ollo$ing steps&
3nitial model parameter estimation and
calculation o* estimated residual variableU
5uild an au2iliary regression based on
presumption that it is a relationship
bet$een the s-uare error values#
e2ogenous variables included in the initial
model and the s-uare o* its values. 3n the
model presents. heteros%edasticity. 3n our
vision# the model can be applied *or
prognosis# because it $ill be revised and
it can be added more data and more
variables *or increasing the validity o* this
model.
3n our opinion i* the value o* , is included in
the interval .(1UR10# the company $ill be in
ban%ruptcy.
$able /: *escripti!e 0tatistics for 0tudied ndicators
@ournal o* 'astern 'urope 1esearch in 5usiness A 'conomics I
$able 1: $"e 2"ite $est
8 @ournal o* 'astern 'urope 1esearch in 5usiness A 'conomics
$able 3: E4cel 5utput
$able 6: E!iews 5utput
@ournal o* 'astern 'urope 1esearch in 5usiness A 'conomics 7
,i# %7 $"e Jar8ue-Bera $est
$able 9: $"e Actual: ,itted 'alues and Residuals
obs ctual :itted 1esidual 1esidual Klot
1 0.02000 (0.092=8 0.082=8 V . V W . V
2 0.01000 (0.1017I 0.1117I V . V W . V
" 0.01000 (0.027I" 0.0"7I" V . VW . V
> 0.09000 (0.00709 0.09709 V . VW . V
9 0.07000 (0.0>81" 0.1281" V . V W. V
I 0.1=000 0.121"9 0.0I7I9 V . V W . V
8 0.09000 0.022I> 0.028"I V . VW . V
7 0.0=000 0.0"0"8 0.09=I" V . VW . V
= 0.01000 0.0I29= (0.0929= V . WV . V
10 0.0>000 0.07>"0 (0.0>>"0 V . WV . V
11 (0.8"000 (0.>"78I (0.2=12> V W . V . V
12 (0.1>000 (0."1I9" 0.18I9" V . V W V
1" (0.2"000 (0."210I 0.0=10I V . V W . V
1> 0.00000 (0.1=18= 0.1=18= V . V W V
19 (0.22000 (0.1>0>= (0.08=91 V . W V . V
1I (0.1=000 (0.1"800 (0.09"00 V . WV . V
18 (0.II000 (0.10==8 (0.9900" VW . V . V
17 0.00000 (0.1I92> 0.1I92> V . V W V
1= (0.0>000 (0.0=>17 0.09>17 V . VW . V
20 (0.22000 (0.000>8 (0.21=9" V W. V . V
21 0.00000 (0.0""8= 0.0""8= V . VW . V
22 (0.12000 (0.0199I (0.10>>> V . W V . V
2" (0.18000 (0.27888 0.11888 V . V W. V
Conclusions
Corporate *inancial *ailure prediction is o*
critical importance *or mangers# sta%eholders
and other parties related. 3n the literature#
many researchers have *ocused on the
*inancial ratios o* corporations *or *ailure
prediction.
Given the current economic situation# trying
to build a ban%ruptcy prediction *unction
score *or 1omanian companies is a real
challenge.
= @ournal o* 'astern 'urope 1esearch in 5usiness A 'conomics
3n our vision# the model can be applied *or
prognosis# because it $ill be revised and it
can be added more data and more variables
*or increasing the validity o* this model. 3n
our opinion i* the value o* , is included in the
interval .(1UR10# the company $ill be in
ban%ruptcy.
/iscriminant analysis as relevant method in
the arsenal o* economic and *inancial
analysis tools $ill become a necessity in the
ne2t period# *or understanding and applying
the economic reality in predicting speci*ic
-uestions o* interest to users o* *inancial and
economic in*ormation. This may prove
e**ective *or& *urther research ban%ruptcy
prediction in companies $ith speci*ic
application to economic sectors or small and
medium enterprisesU 4udging companies
operational activity by auditors and
accountantsU using score *unctions to provide
in*ormation to investors concerned about
*inding the most pro*itable investments# or
solutions *or their port*olio in order to earn
an optimal overall return(ris% per shareU
using discriminant analysis in the
macroeconomics# in areas such as analysis
and prediction o* success vs. *ailure o*
speci*ic economic policies on the
development o* disadvantaged areasU
implementing economic programs *or the
development o* certain industries or sectors.
References
ltman# '. 3. .1=I70. E:inancial 1atios#
/iscriminant nalysis and the Krediction o*
Corporate 5an%ruptcy#E Journal of ,inance
2"# 97=LI0=.
nghel# 3. .20000. G5an%ruptcy Krediction in
1omanian 'nterprises#G Economic -ribune#
Bucharest.
rmeanu# 6. /.# Vintil# G.# Moscalu# M.#
:ilipescu# M. ?. A !aBar# K. .20120. EHsing
Nuantitative /ata nalysis Techni-ues *or
5an%ruptcy 1is% 'stimation *or
Corporations#E -heoretical an. $pplie.
Economics# Volume T3T .20120# So. 1.9II0#
=8(112.
5ileMteanu# G. .1==70. G/iagnosis# 1is% and
5usiness '**iciency#G /irton 0ublishing 1ouse#
-imi2oara.
5arbier# . A Kroutat# @. .1==00. Traite
Krati-ue /e !;analyse :inanciere !;usage
/es 5an-ues# Karis.
5eaver# <. J. .1=II0. :inancial 1atios s
Kredictors o* :ailure# 'mpirical 1esearch in
ccounting# 6elected 6tudies#6upplement to
Journal of $ccounting Research.
5eaver# <. J.# McSichols# M. :. A 1hie# @. <.
.20090. EJave :inancial 6tatements 5ecome
!ess 3n*ormativeX 'vidence *rom the bility
o* :inancial 1atios to Kredict 5an%ruptcy#E
Re3iew of $ccounting 4tu.ies 10# ="L122.
Chava# 6. A @arro$# 1. . .200>0. E5an%ruptcy
Krediction $ith 3ndustry '**ects#E Re3iew of
,inance 7# 9"8L9I=.
Cohen# '. A 6auriel# . .1==00. nalyse
:inanciere(?utils 'T pplications# Economic
0ublishing 1ouse# 0aris.
/enis# /. @.# /iane# Y. /. A tulya# 6. .1==80.
Egency Kroblems# '-uity ?$nership# and
Corporate /iversi*ication#E Journal of ,inance
92# 1"9L1I0.
Jillegeist# 6. .# Yeating# '. Y.# Cram# /. K. A
!undstedt# Y. G. .200>0. Essessing the
Krobability o* 5an%ruptcy#E Re3iew of
$ccounting 4tu.ies =# 9L">.
3voniciu# K. .1==70. Gnalysis o* 5an%ruptcy
1is% through 6core Method#G :inance# 5an%s#
3nsurance MagaBine# Bucharest# So. >.
!achenbruch# K. . .1=890. /iscriminant
nalysis# 1afner# 56.
?hlson# @. . .1=700. E:inancial 1atios and the
Krobabilistic Krediction o* 5an%ruptcy#E
Journal of $ccounting Research 17# 10=L1"1.
6hum$ay# T. .20010. E:orecasting
5an%ruptcy More ccurately& 6imple
JaBard Model#E Journal of Business 8># 101L
12>.
@ournal o* 'astern 'urope 1esearch in 5usiness A 'conomics 10
6latter# 6. K. .1=7>0. Corporate 1ecovery&
6uccess*ul Turnaround 6trategies and Their
3mplementation# Richar. 7la8 0te 9t..#
4ingapore# 0enguin Boo:s.
Tudorel# . .20000. Gnalysis o* 6tructural
Changes through 1egression Models#G
Romanian 4tatistical Re3iew V. >=# Sr. 8# K.
2"(">.
Tudorel# . .200=0. Gpplications in
'conometrics#G Economic 0ublishing.
Tudorel# .# 3acob# .# 3luBia# .# 6tancu# 6. A
Tusa# '. .20070. G3ntroduction in
'conometrics Hsing 'vie$s#G Economic
0ublishing.
Ver$i4meren# K. A /er$all# @. .20100.
E'mployee <ell(5eing# :irm !everage# and
5an%ruptcy 1is%#E Journal *f Ban:ing &
,inance "># =9I(=I>.
Vintil# G. A Toroap# M. G. .20110. 5uilding a
6coring Model *or 5an%ruptcy 1is%
Krediction on Multiple /iscriminant nalysis#
the 3nternational Con*erence CKresent 3ssues
o* Global 'conomyD( 7th 'dition# pril 1Ith(
18th# nnals o* the C?vidiusD Hniversity#
'conomic 6ciences 6eries Volume T3# 3ssue 1
Q2011# 227"(2278.
Voineagu# V.# Titan# '. A 6erban# 1. .200I0.
GTheory and 'conometric Kractice#G /eteor
0ress.
,mi4e$s%i# M. '. .1=7>0. EMethodological
3ssues 1elated to the 'stimation o* :inancial
/istress Krediction Models#E Journal of
$ccounting Research 22 .6uppl.0# 9=L72.
Reproduced with permission of the copyright owner. Further reproduction prohibited without
permission.
