iwan editan

8/13/2019 Iwan Editan

http://slidepdf.com/reader/full/iwan-editan 1/17

SIMPOSIUM NASIONAL AKUNTANSI VISurabaya, 16 – 17 Oktober 2003

BANKRUPTCY PREDICTION MODEL WITH ZETAc OPTIMALCUT-OFF SCORE TO CORRECT TYPE I ERRORS

MOHAMAD IWANYogyakarta

Abstract This research examines financial ratios that distinguish between bankrupt and non-bankrupt companies and

make use of those distinguishing ratios to build a one-year prior to bankruptcy prediction model. This research alsocalculates how many times the type I error is more costly compared to the type II error. The costs of type I and type IIerrors (cost of misclassification errors) in conjunction to the calculation of prior probabilities of bankruptcy and non-bankruptcy are used in the calculation of the ZT! c optimal cut-off score. The bankruptcy prediction result using ZT!c optimal cut-off score is compared to the bankruptcy prediction result using a cut-off score which does notconsider neither cost of classification errors nor prior probabilities as stated by "air et al. (#$$%) and for later purposeswill be referred to "air et al. optimum cutting score. &omparison between the prediction results of both cut-off scoresis purported to determine the better cut-off score between the two' so that the prediction result is more conseratie

and minimies expected costs' which may occur from classification errors.This is the first research in Indonesia that incorporates type I and II errors and prior probabilities ofbankruptcy and non-bankruptcy in the computation of the cut-off score used in performing bankruptcy prediction.arlier researches gae the same weight between type I and II errors and prior probabilities of bankruptcy and non-bankruptcy' while this research gies a greater weigh on type I error than that on type II error and prior probability ofnon-bankruptcy than that on prior probability of bankruptcy.

This research has successfully attained the following results* (#) type I error is in fact +$'%, times morecostly compared to type II error' () ratios distinguish between bankrupt and non-bankrupt groups' (,) financialratios proed to be effectie in predicting bankruptcy' () prediction using ZT! c optimal cut-off score predicts morecompanies filing for bankruptcy within one year compared to prediction using "air et al. optimum cutting score' (+) !lthough prediction using "air et al. optimum cutting score is more accurate' prediction using ZT! c optimal cut-offscore proed to be able to minimie cost incurred from classification errors. BACKGROUND

Uncertain condition in Indonesian’s economy nowadays put firms in the risk of experiencingfinancial distress or even bankruptcy. Prediction error towards the continuity of an entity in the future cancause severe loss. There are two types of errors that may occur namely type I error and type II error. If theprediction performed is a prediction about whether a firm is to file for bankruptcy or not in the future thentype I error can be comprehended as predicting a firm not to file for bankruptcy while in fact the firm doesfile for bankruptcy. Type II error on the contrary is predicting a firm to file for bankruptcy while in fact the firmdoes not file for bankruptcy. !lthough both types of prediction errors inflict a certain amount of financialloss type I errors inflict a greater financial loss compared to that of type II errors. Therefore predictionre"uires a cut#off score which classifies a company in to either the bankrupt group or the non#bankruptgroup with a minimum cost of classification errors.

$itherto there are no theories that affirm definitely what financial ratios must be used in predicting

bankruptcy. %atios used in predicting bankruptcy can vary among different researches. It is merely thesub&ective consideration of the researcher followed by statistical verification on financial ratios applied.

This research uses the formula of !ltman $aldeman dan 'arayanan ()*++, to calculate the-T!c optimal cut#off score. /inancial ratios developed by 0achfoed1 ()**2, are used to distinguishbetween bankrupt and non#bankrupt groups. The distinguishing financial ratios are eventually opted basedon statistical testing to build the prediction model. The prediction model in this research is developed in thesame way of !vianti (3444, that is by applying the two#group discriminant analysis.

! cut above or advantage of this research is that it incorporates the type I and II errors and priorprobabilities of bankruptcy and non#bankruptcy in the computation of -T! c optimal cut#off score thus theprediction result is expected to be able to minimi1e cost incurred from classification errors compared to theprediction result if using the cut#off score as the one stated in $air et al. ()**5, and for later purposes willbe referred to $air et al. optimum cutting score.

114




SESIBakru!t"y Pre#$"t$o Mo#e% &$t'

(eta" O!t$)a% *ut+O S"ore To *orre"t Ty!e I -rror.

PROBLEM FORMULATION AND RESEARCH OBJECTIVEThis research investigates what financial ratios distinguish between bankrupt and non#bankrupt

companies and makes use of those distinguishing ratios to build a one#year prior to bankruptcy prediction

model. This research also calculates how many times type I error is more costly compared to type II error.The costs of type I and type II errors (cost of misclassification errors, in con&unction to the calculation ofprior probabilities of bankruptcy and non#bankruptcy is used in the calculation of the -T! c optimal cut#offscore. The bankruptcy prediction result using -T!c optimal cut#off score is then compared to thebankruptcy prediction result using $air et al. optimum cutting score to determine the better cut#off score toapply in terms of less costly in predicting bankruptcy.

RESEARCH LIMITATIONS6ome limitations exist in this research which are7). Previous researches in general define bankruptcy as legal bankruptcy and this definition is applied

as the dependent variable (8eaver )*99:)*95; !ltman )*95; <hlson )*54; -ain )**2 etc,.$owever this research applies the stock based insolvency definition and uses negative e"uity as its

dependent variable (!vianti 3444,. In conse"uence the term bankruptcy in this research meanshaving a negative e"uity.

3. The determination of the cost of classification errors is obtained from a sample of two state ownedbanks (8ank 8'I dan 8ank 8%I, and two private banks (8ank 'iaga dan 8ank =anamon, for theyear of )**9 )**+ )*** 3444. Thus there is a probability that the computation of the cost ofclassification errors does not represent the entire population as a whole.

>. 'o hold out sample is employed in this research; hence prediction is limited to the original sample.The reason for not employing a hold out sample is because of time consideration and lack of data which until the point this research was concluded was not available.

RESEARCH BENEFITSThis research is deemed to have the following benefits7

). This research is expected to become a basis to opt for the cut#off score to be used in conductingbankruptcy prediction.

3. This research is expected to benefit stakeholders of a company in assessing and making short#term decisions regarding the company.

>. This research is expected to become a sub&ect of information and: or consideration fordevelopments in further researches.

2. This research is expected to start and open a discourse of research in Indonesia regardingbankruptcy prediction that incorporates cost of classification errors and prior probabilities of bankruptcyand non#bankruptcy.

THEORITICAL BASIS%atios are among the most popular and widely used tools of financial analysis (8ernstein and

?ild )**5,. The result of the calculation of financial ratios obtained from a set of financial statements isable to determine the economic ability of a company (0achfoed1 )**@ in !vianti 3444,. 0achfoed1 ()**2,in !vianti (3444, states that financial ratios can be used to predict future events by associating financialratios with economic phenomena’s. Prediction is conducted to reduce future uncertainties. /inancial ratioscan be used to predict future bankruptcy by developing a bankruptcy prediction model. The modeldeveloped in this research is purported to represent financial conditions of companies and to predict whether companies will file for bankruptcy or not (!vianti 3444,.

Definii!n" !f B#n$%&'c(8ankruptcy can be classified in to two categories which are stock based insolvency and legal

bankruptcy (!vianti 3444,. Previous researches apply the legal bankruptcy definition as its dependentvariable. The term bankrupt in this research as in (!vianti 3444, applies the stock based insolvencydefinition. ! company is said to be bankrupt (stock based insolvency, if a company experiences lack oftemporary li"uidity and continues to have a larger book value of liabilities than assets thus the e"uity

115






becomes negative (including minority interest in the subsidiary’s net assets,. In such circumstance acompany is said to be bankrupt from the e"uity perspective (%oss et al. )**> and 8righam A Bapensky)**> in !vianti 3444,. The reason why this definition is employed is because data of publicly held

companies in Indonesia that are legally bankrupt are very hard to get hold of if any. The Commerce Courtin Indonesia was formed in )**5 and until the point this research was conducted there were only very fewbankruptcy appeals. Conse"uently it is very difficult to find data regarding companies having legallybankrupt status.

!ccording to !ct 'umber 2:)**2 an institution is stated to be bankrupt by the &udgment of court ifthe debtor retain two or more creditors and does not pay at least one overdue and collectible debt.

=ifferent researchers often define the term bankruptcy differently and due to certain conditions andlimitations thus the term bankrupt employed in this research is companies that have negative e"uities andfor that reason other definitions are not cited in this research.

PREVIOUS LITERATURE AND HYPOTHESES DEVELOPMENTThere have been two types of bankruptcy prediction studies. The first (e.g. 8eaver )*99, looks at

the relation between individual accounting numbers or ratios and bankruptcy (the univariate approach,. Theother uses several ratios to predict bankruptcy (the multivariate approach,. The univariate approach usesone ratio at a time to predict failure. It is likely that different ratios reflect different aspects of the firm’sfinancial position so better predictions can be obtained by using combinations of ratios instead of one ratio./or this reason the multivariate approach "uickly supplanted the univariate approach (?atts and-immerman )*59,.

6everal researches were conducted in the bankruptcy prediction domain (companies predicted donot include banking and financial sector companies, as the followings 8eaver ()*99; )*95a; )*95b, !ltman()*95; )*+>, !ltman and Dorris ()*+9, !ltman and 0cBough ()*+2, !ltman $aldeman and 'arayanan()*++, =eakin ()*+3, Dibby ()*+@, 8lum ()*+2, dmister ()*+3, ?ilcox ()*+>, 0oyer ()*++, Dev()*+), <hlson ()*54, 6chiedler ()*5), 6cott ()*5), =ambolena A Ehoury ()*54, -mi&ewski ()*52,0ensah ()*52, Bentry et al. ()*5+, 8arniv A %aveh ()*5*, Platt A Platt ()**4, -ain ()**2, %ichardson

et al. ()**5, Din et al. ()***, 6etyorini A $alim ()***, and !vianti (3444,.8eaver ()*99, used the univariate approach and the main findings of the research was that

accounting data in forms of financial ratios have the ability to predict failure for at least five years prior to thefailure. 8eaver called this approach as a profile analysis.

!ltman ()*95, used the multivariate discriminant analysis (0=!, to predict bankruptcy. There wasa limitation in !ltman’s model because the prediction accuracy for predictions over than two years prior tobankruptcy became awfully low compared to the prediction of one and two years prior to bankruptcy.

!ltman $aldeman and 'arayanan ()*++, also developed a bankruptcy prediction model usingthe 0=!. <ne distinguishing point of this research amongst others is the determination of the cut#off score.<ther bankruptcy prediction researches give the same weight between type I and type II errors and priorprobabilities of bankruptcy and non#bankruptcy. That approach is in contrast with this research that usesthe -T!c optimal cut#off score which gives a greater weight on type I error than type II error and prior

probability of non bankruptcy than prior probability of bankruptcy. <ne of !ltman’s essential findings is thattype I error is >@ times more costly than type II error. Therefore the cut#off score must weigh type I errorand type II error differently.

<hlson ()*54, used the logistic regression analysis and developed > bankruptcy predictionmodels; prediction models for one two and one or two years prior to bankruptcy. The sampling methodused by <hlson ()*54, was proportional with the population. <hlson ()*54, also considered the publicationdate of financial statements because other researches assumed that financial statements for the year ofbankruptcy were published before bankruptcy filings occurred resulting in overstatement of predictionpower.

!vianti (3444, built bankruptcy prediction models using > different methods. ach method wasused to build > bankruptcy prediction models#one year two years and three years prior to bankruptcy;hence * models were successfully developed. The methods used to build the models were Dinier=iscriminant !nalysis Dinier =iscriminant !nalysis combined with Principal Component !nalysis andDogistic %egression. =ifferent financial ratios were employed as the predictor variable for each year and

116






each method. %esults showed that the linier discriminant model was superior to the other two methods inperforming predictions for one and two years prior to bankruptcy. !s for the prediction of > years prior tobankruptcy the logistic regression model was superior to the other two models.

Previous researches have proven that financial ratios can be used to build a bankruptcy predictionmodel and use different financial ratios in order to predict bankruptcy. It has been proven that different setsof data will result in different models (!ltman ()*95, !ltman $aldeman and 'arayanan ()*++, <hlson()*54, !vianti (3444,. 6ince financial ratios differ along with different researches that use different data itis necessary here to determine what financial ratios differ from bankrupt and non#bankrupt companiesaccording to the data used in this research (prediction can only be conducted if any characteristics of theob&ect being predicted do exist which in this case the ob&ect are bankrupt and non#bankrupt companiesthus the characteristics that are expected to differ from bankrupt and non#bankrupt companies are financialratios of both groups,. <n the basis of the argument above the following alternative hypothesis isformulated7

$)7 /inancial ratios differ between bankrupt and non#bankrupt companies.If financial ratios that distinguish between bankrupt and non#bankrupt companies exist then those

financial ratios will be used to develop a one#year prior to bankruptcy prediction model. This step isperformed to further understand whether those financial ratios can be used to predict bankruptcybeforehand. !s a result the following alternative hypothesis is formulated7

$37 /inancial ratios can be used to predict bankruptcy beforehand.

T#)*e +, Pe%cen#e !f c!%%ec*( c*#""if(in !f . )#n$%&'c( '%e/ici!n 0!/e*" &"in 1e0&*i2#%i#e #''%!#c1

'ecessary to stress on are researches of 8eaver ()*99, !ltman ()*95, <hlson ()*54, and!vianti (3444,. 'either of those considered on prior probabilities of bankruptcy and non#bankruptcy nor cost

of classification errors (cost occurring from type I and II errors,. !ccording to -mi&ewski ()*5>, in table )above if type I error is given a greater weight than type II error then the percentage of correctly predictingthe bankrupt companies also becomes greater while the percentage of correctly predicting the non#bankrupt companies becomes smaller. !n increase of correctly predicting bankrupt companies means thatthe amount of companies predicted to be non#bankrupt from those companies supposed to be predicted asbankrupt companies becomes smaller. <n the contrary a decrease of correctly predicting non#bankruptcompanies means that the amount of companies predicted to be bankrupt from those companies supposedto be predicted as non#bankrupt becomes greater. Thus giving a greater weight on type I error than type IIerror will predict more companies as bankrupt companies and less companies as non#bankrupt companies.!ccordingly the following alternative hypotheses is formulated7

$>7 The percentage of predicting bankrupt companies will become larger by giving a greater weight on type I error than type II error compared to the percentage of predicting bankrupt

companies of a prediction that gives the same weight on type I and type II errors.The greater the amount of companies predicted to be bankrupt indicate the lesser the amount of

type I errors. Thus expected cost of classification errors can be minimi1ed since type I error is much more

Researches and Year Type I Cost= Type I Cost=2xType II Cost

Type I Cost=20xType II Cost

Type ICost=38xType II

Cost

NB B TS NB B TS NB B TS NB B TS

Beaer !1"66# ""$% 36$1 "8$4 ""$2 43$1 "8$1 "1$8 86$1 "1$% "1$8 86$1 "1$%

&'t(an !1"68# ""$8 6$" "%$" ""$3 33$3 "8$0 88$8 84$% 88$8 83$3 "3$1 83$5

B')( !1"%4# ""$8 18$1 "8$" ""$4 38$" "8$2 "5$" %%$8 "5$6 "0$6 8%$5 "0$1

&'t(an et a'$ !1"%%# ""$8 4$2 "%$" "8$" 38$" "%$% "1$4 81$" "1$2 84$0 "4$4 84$3

*a(+o'ena,-ho)ry!1"80# ""$% 26$4 "8$3 "8$" 4%$2 "%$% "5$% %3$6 "5$3 83$% "1$6 83$"

.h'son !1"80# ""$8 8$3 "8$0 "8$8 40$3 "%$% "3$5 84$% "5$3 8"$6 "3$1 8"$%/(es !1"83# ""$% "$% "80 ""$3 3%$5 "8$1 "5$5 %5$0 "5$1 86$0 "1$2 86$0

Note NB Non7+anr)pt co(panes B Banr)pt co(panes

TS Tota' Sa(p'eSo)rce /(es !1"83 Ta+e' 5dan 6# pp$32733 n oster !1"86#

11%






costly compared to type II error. Conse"uently a model with a cut#off score that gives a greater weight ontype I error than type II error is better in terms of more cost effective than a model that gives the same weight on type I and type II errors. /or this reason the following alternative hypotheses is formulated7

$27 8ankruptcy prediction with a cut#off score that incorporates prior probability of bankruptcy andnon#bankruptcy and type I and type II errors will cut costs occurring from prediction errors.

RESEARCH METHODOLOGY SAMPLE AND DATAS#0'*e !f B&i*/in 1e B#n$%&'c( P%e/ici!n M!/e*

The sampling method employed in building the bankruptcy prediction model is the matched pairsampling method based on industrial sectors and si1e of companies. Industrial sectors are determinedbased on the categories in the Indonesian Capital 0arket =irectory 3444 and 344) while the si1e ofcompanies are determined based on the total asset average in accordance with 8apepam’s regulationsub&ect on /oreign Capital Investment: =omestic Capital Investment, regarding the criteria’s of si1es ofcompanies based on the amounts of total assets. !verages of total assets that are used to determine thesi1es of companies are obtained from the Indonesian Capital 0arket =irectory 3444 and 344).

6teps in determining sample are as follows7). =etermine bankrupt companies in years of )*** and 3444 and trace their financial statements one

year back which are years of )**5 for bankrupt companies of )*** and )*** for bankruptcompanies of 3444.

3. =etermine non#bankrupt companies that match bankrupt companies determined earlier. 'on#bankrupt companies are selected using industrial sectors and si1es as the criteria of matchedpairs.

The followings are lists of bankrupt and non#bankrupt companies used in building the one#year prior tobankruptcy prediction model.

Table 2. Sample of bankrupt and non-bankrupt companies

No. Bankrupt Companies Non-bankrupt Companies

1 9T Ctatah Ind)str :ar(er T+ 9T &nea Ta(+an; T+

2 9T 9rasdha &nea Na;a T+ 9T :)'t Bntan; Indonesa T+

3 9T Tex(aco <aya T+ 9T Ten Indonesa +er Corporaton !TIIC.# T+

4 9T 9o'ysndo a 9erasa T+ 9T B)d &cd <aya T+

5 9T 9anca >rata(a Sat T+ 9T 9)dad 9rest;e ?(ted T+

6 9T 99 9ere+)nan ?ondon S)(atra Indonesa T+ 9T &stra &r;o ?estar T+

% 9T &'ter &+ad T+ 9T T(ah T+

8 9T *ao(as &+ad T+ 9T Sar @)sada T+

" 9T S:&RT Corporaton T+ 9T :ayora Indah T+

10 9T &r;o 9antes T+ 9T 9anasa Indosyntec T+

11 9T 9r(arndo &sa InArast)ct)re T+ 9T Sepat) Bata T+

12 9T S)rya *)(a Ind)str T+ 9T S)(a'ndo ?estar <aya T+

13 9T S)ra+aya &;)n; Ind)str 9)'p T+ 9T S)par(a T+

14 9T To+a 9)'p ?estar !9T Indrayon ta(a# T+ 9T aar S)rya >esa T+

15 9T Tr 9o'yta Indonesa T+ 9T ?a)tan ?)as T+

16 9T &r;ha -arya 9r(a Ind)stry T+ 9T >ahana <aya 9erasa !9T &@&RI# T+

1% 9T :)'a Ind)strndo T+ 9T S)rya Toto Indonesa T+

18 9T Tex(aco 9erasa n;neern; T+ 9T -o(ats) Indonesa T+

1" 9T T -a+e' Indonesa!-a+e'(eta' Indonesa#T+ 9T S)( Indoa+e' !I-I Indah -a+e' Indonesa# T+

20 9T aah T)n;;a' T+ 9T &stra Internatona' T+

21 9T Indo(o+' S)ses Internatona' T+ 9T &stra .toparts T+

22 9T T 9etroche( Ind)stres T+ 9T nted Tractor T+

23 9T Cp)tra *ee'op(ent T+ 9T B)t Sent)' T+

24 9T *har(a'a Int'and T+ 9T *)ta 9ert T+ 25 9T <aarta Seta+)d 9roperty T+ 9T <aarta Internatona' @ote' , *ee'op(ent T+

26 9T -aasan Ind)stry <a+a+ea T+ 9T <aya Rea' 9roperty T+

118






2% 9T ?ppo ?and *ee'op(ent T+ 9T Bare'and *ee'op(ent !'an; Rea'ty# T+

28 9T :odern'and Rea'ty T+ 9T S)rya Se(esta Intern)sa T+

2" 9T S)rya(as *)ta :a()r T+ 9T S)((arecon &;)n; T+

S#0'*e &"e/ in /ee%0inin '%i!% '%!)#)i*iie"Prior probabilities of bankruptcy are calculated using the numbers of bankrupt companies divided

by total companies listed on the Fakarta 6tock xchange for years of )**9 )**+ )*** and 3444 whilethe calculation of prior probabilities of non#bankruptcy are the numbers of non#bankrupt companies dividedby total companies listed on the Fakarta 6tock xchange for years of )**9 )**+ )*** and 3444.S#0'*e &"e/ in c#*c&*#in c!" !f c*#""ific#i!n e%%!%" 3('e I #n/ ('e II e%%!%"4

The sample of banks used in computing cost of classification errors are 8ank 8'I 8ank 8%I 8ank'iaga and 8ank =anamon. !ll data used are data from years )**9 )**+ )*** and 3444. !ssuming thatbanking characteristics in Indonesia is similar the two state#owned and two private#owned banksaforementioned are expected to represent the whole banking population.

=ata from year )**5 are not included in the determination of the -T!c optimal cut#off score

(calculation of prior probabilities and cost of classification errors, for the reason that in year )**5 theaverage credit interest rate and the Central 8ank Certificate interest rate boosted very high in comparison with previous and subse"uent years because of the economic crisis thus no new loan at that time wasgiven. The unusual high level of interest rate in year )**5 is deemed to be able to distort the calculation of-T!c optimal cut#off score.

IDENTIFICATION AND OPERATIONAL DEFINITION OF VARIABLES OF PREDICTION MODELThis research uses two types of variables that are dependent variables and independent variables.

The dependent variables are categories (non#metric, that are bankrupt companies and non#bankruptcompanies. The independent variables used are metric variables that are financial ratios used by <u andPenman ()*5*, 0achfoed1 ()**2, and !vianti (3444,. Disted below are the financial ratios used.

T#)*e 5, Fin#nci#* %#i!" &"e/ ! )&i*/ 1e !ne-(e#% '%i!% ! )#n$%&'c( '%e/ici!n 0!/e*No. Financial Ratios Abbreviation

Short Term Liui!it" #I$

1$ Cash to c)rrent 'a+'tes !D1# CC?

2$ Cash A'o to c)rrent 'a+'tes !D2# CC?

3$ E)c assets to c)rrent 'a+'tes !D3# E&C?

4$ C)rrent assets to c)rrent 'a+'tes !D4# C&C?

Lon% Term Solvenc" #II$

5$ C)rrent assets to tota' 'a+'tes !D5# C&T?

6$ Net orth and 'on; ter( de+t to Axed assets !D6# N>?T*&

%$ Net orth to Axed assets !D%# N>&

&ro'itabilit" #III$

8$ .peratn; nco(e to arnn;s +eAore taxes !D8# .INBT"$ arnn;s +eAore taxes to sa'es !D"# BTS

10$ ross proAt to sa'es !D10# 9S

11$ .peratn; nco(e to sa'es !D11# .IS

12$ Net nco(e to sa'es !D12# NIS

&ro!uctivit" #I($

13$ Inentory to orn; capta' !D13# I>C

14$ Cost oA ;oods so'd to nentory !D14# C.SI

15$ Sa'es to F)c assets !D15# SE&

16$ Sa'es to cash !D16# SC

1%$ Sa'es to acco)nts recea+'es !D1%# S&R

18$ Cash A'o to tota' assets !D18# CT&

1"$ C)rrent assets to tota' assets !D1"# C&T&

20$ E)c assets to nentory !D20# E&I

21$ Inentory to sa'es !D21# IS

11"






22$ Sa'es to tota' assets !D22# ST&

23$ >orn; capta' to tota' assets !D23# >CT&

In!ebt ness #($

24$ Tota' 'a+'tes to c)rrent assets !D24# T?C&

25$ .peratn; nco(e to tota' 'a+'tes !D25# .IT?

26$ C)rrent 'a+'tes to tota' assets !D26# C?T&

Investment Intensiveness #(I$

2%$ Cash A'o to tota' 'a+'tes !D2%# CT?

28$ Sa'es to Axed assets !D28# S&

2"$ >orn; capta' to tota' assets !D2"# >CT&

30$ C)rrent assets to sa'es !D30# C&S

31$ E)c assets to tota' assets !D31# E&T&

32$ Net orth to sa'es !D32# N>S

33$ >orn; capta' to sa'es !D33# >CS

34$ Inentory to tota' assets !D34# IT&

35$ Cash A'o to sa'es !D35# CS

Levera%e #(II$36$ Net orth to tota' assets !D36# N>T&

3%$ C)rrent 'a+'tes to nentory !D3%# C?I

38$ Tota' 'a+'tes to tota' assets !D38# T?T&

Return on Investment #(III$

3"$ arnn;s +eAore taxes to net orth !D3"# BTN>

40$ Net nco(e to Axed assets !D40# NI&

41$ Net nco(e to net orth !D41# NIN>

42$ arnn;s +eAore taxes to tota' assets !D42# BTT&

43$ Net nco(e to tota' assets !D43# NIT&

Euit" #I)$

44$ Sa'es to c)rrent 'a+'tes !D44# SC?

45$ Net nco(e to tota' 'a+'tes !D45# NIT?46$ C)rrent 'a+'tes to net orth !D46# C?N>

4%$ Net orth to tota' 'a+'tes !D4%# N>T?

THE BANKRUPTCY PREDICTION MODELThe bankruptcy prediction model is built by using the two#group discriminant analysis because the

dependent variables are non#metric and the independent variables are metric.$) is tested by performing the ?ilks Dambda test which is a test of e"uality of group means of financial

ratios of both groups (bankrupt and non#bankrupt,. This test is performed to examine whether distinguishingratios exist between the two groups. The significance level in this research is @G for the / critical value which means that financial ratios having significance level under @G are ratios that distinguish betweenbankrupt groups and non#bankrupt groups. If any financial ratio does have a significance value under @G

then a discriminant function can be developed.6ubse"uently the financial ratios that distinguish between bankrupt and non#bankrupt groups are

selected to come up with the best variate (linier combination, by applying the stepwise selection algorithm./inancial ratios that constitute the best linier combination are the final financial ratios that are going to beused as independent variables in the discriminant function which is the bankruptcy prediction model. Theprediction using the discriminant function is designed to test $3.

DESCRIPTON OF HAIR6 e #*, OPTIMUM CUTTING SCORE! cutting score is necessary in performing prediction if using discriminant analysis. ! cutting score is

the criterion (score, against which individual’s discriminant score is &udged to determine into which groupthe individual should be classified. Those entities whose - scores are below this score are assigned to onegroup while those whose scores are above it are classified in the other group ($air et al. )**5,.

!ccording to $air et al. ()**5, the optimal cutting score will differ depending on whether the si1es ofthe groups are e"ual or une"ual. 6ince this research applies the matched pair sampling method by which

120






the number of members in each group are e"ual thus the formula to compute the cutting score or the $airet al. optimum cutting score is as follow7

2

B A

CE

Z Z Z

+=

where-C H Critical cutting score value for e"ual group si1es-! H Centroid for group ! (bankrupt,-8 H Centroid for group 8 (non#bankrupt,

Centroid’s of each group (-! and -8, is computed by calculating the average of the - discriminantscores of each group.

DESCRIPTION OF ZETAc OPTIMAL CUT-OFF SCORE-T!c optimal cutoff score is obtained from the following formula7

-T!c 002

01

CF

CF'n=

where") H Prior probability of bankruptcy"3 H Prior probability of non#bankruptcyCi H cost of type I errorCii H cost of type II errorPrior probability of bankruptcy ("), is computed by dividing the number of bankrupt companies with

the total companies listed in F6 for each year. Prior probability of non#bankruptcy (" 3, is computed bydividing the number of non#bankrupt companies with the total companies for each year. 8oth priorprobabilities of bankruptcy and non#bankruptcy are computed for years of )**9 )**+ )*** and 3444.

Type I error is analogous to that of an accepted loan that defaults and the type II error to a re&ectedloan that would have resulted in a successful payoff. Thus Type I and type II errors (C i and Cii, are

computed by the following formulas7

=

oAA#7!char;ed 'osses'oanross

recoered 'osses'oanoA &(o)nt71 C

Type II (Cii, is computed by the following formula7Cii H r J ir H effective interest rate on the loan which is computed as follow7

+

= I?8x

T?

?8I?R x

T?

?R r

C% H Doans given in %upiahC/ H Doans given in /oreign currencies (after converted into %upiah,IC% H !verage interest rate of loans given in %upiah

IC/ H !verage interest rate of loans given in /oreign currenciesTC H Total credit distributed (Total loans given in %upiah and /oreign currencies,.

i H effective opportunity cost for the bank (one year average of the >4 days Central 8ankCertificate interest rate for years )**9 )**+ )*** and 3444,.

The -T!c optimal cut#off score gives a greater weight on type I error than type II error and priorprobabilities of non#bankruptcy than prior probability of bankruptcy. Type I error is given a greater weightcompared to type II error because type I error is more costly than type II error. Prior probability of non#bankruptcy is also given a greater weight than prior probability of bankruptcy since the probability of acompany to file for bankruptcy is much more smaller than the probability of a company to not file forbankruptcy.

In the formula above ") is multiplied by C i to incorporate the cost of type I error with the probabilityof bankruptcy to understand the probability of type I error to occur. /urthermore "3 is also multiplied by Cii

to incorporate the cost of type II error with the probability of non#bankruptcy to understand the probability oftype II error to occur. In view of that -T! c optimal cut#off score is a cut#off score used in bankruptcyprediction based on the consideration of probabilities of occurrences of type I and II errors.

121






Comparison of the one#year prior to bankruptcy prediction results between predictions using $airet al. optimum cutting score and predictions using -T!c optimal cut#off score is designed to test $>.

The prediction results between predictions using $air et al. optimum cutting score and predictions

using -T!c optimal cut#off score will be different. The differences of the prediction results are difference inthe occurrences of type I and type II errors by each prediction using different cut#off scores. Therefore $ 2

can be tested.

RESEARCH FINDINGS=escribed below are the research findings that are organi1ed in a systematically order. The

findings are described based on the steps first conducted in the research.

TEST OF E7UALITY OF GROUP MEANS OF FINANCIAL RATIOS OF BOTH GROUPS 3BANKRUPTAND NON-BANKRUPT4

%esult of the test of e"uality of group means using ?ilks Dambda showed that 33 financial ratiosamong the 2+ financial ratios tested have significance levels under @G as shown in table 2 below.

8ased on the test result below $) that indicated that distinguishing ratios exist between bankruptand non#bankrupt groups is accepted.

T#)*e 8, Sinific#n*( /iffe%en fin#nci#* %#i!" )e9een )#n$%&' #n/ n!n-)#n$%&' %!&'"

(ariables*ilks+

Lamb!aF Si%.

Cash to c)rrent 'a+'tes !D1# 0$846 "$801 0$003

E)c assets to c)rrent 'a+'tes !D3# 0$845 "$883 0$003

C)rrent assets to c)rrent 'a+'tes !D4# 0$%"8 13$6"5 0$001

C)rrent assets to tota' 'a+'tes !D5# 0$80" 12$%%5 0$001

arnn;s +eAore taxes to sa'es !D"# 0$"24 4$445 0$04

ross proAt to sa'es !D10# 0$"2% 4$258 0$044 Net nco(e to sa'es !D12# 0$"24 4$445 0$04

C)rrent assets to tota' assets !D1"# 0$"0" 5$434 0$024

>orn; capta' to tota' assets !D23# 0$%48 18$213 0

Tota' 'a+'tes to c)rrent assets !D24# 0$%65 16$5" 0

.peratn; nco(e to tota' 'a+'tes !D25# 0$868 8$205 0$006

C)rrent 'a+'tes to tota' assets !D26# 0$%5" 1%$18 0

>orn; capta' to tota' assets !D2"# 0$%48 18$213 0

E)c assets to tota' assets !D31# 0$886 6$"44 0$011

Net orth to tota' assets !D36# 0$4"4 55$346 0

Tota' 'a+'tes to tota' assets !D38# 0$4"8 54$401 0

Net nco(e to Axed assets !D40# 0$"04 5$%54 0$02

arnn;s +eAore nco(e taxes to tota' assets !D42# 0$%61 1%$005 0

Net nco(e to tota' assets !D43# 0$%1" 21$082 0

Sa'es to c)rrent 'a+'tes !D44# 0$884 %$06" 0$01

Net nco(e to tota' 'a+'tes !D45# 0$81" 11$"26 0$001

Net orth to tota' 'a+'tes !D4%# 0$%85 14$%58 0

STEPWISE SELECTION ALGORITHMThe stepwise selection algorithm is performed in order to obtain the best variate (linier

combination, from the 33 distinguishing financial ratios above which will be used in the linier discriminantfunction. %esults of the stepwise selection algorithm showed that only two financial ratios are best to beused as independent variables in the discriminant function that are net worth to total assets (>9,' and net worth to total liabilities (2+,.

DISCRIMINANT FUNCTION BUILT

122






The Canonical =iscriminant /unction Coefficient obtained for the one#year prior to bankruptcyprediction can be seen in table @ below.

T#)*e :, C#n!nic#* Di"c%i0in#n F&nci!n C!efficien)ncton

1

D36 10$086

D4% 7$824

!Constant# 72$060

Unstandardi1ed coefficients

The one#year prior to bankruptcy prediction model built is as follow7Z ; -<,=>=? +=,=@>5> =,@<88.

where

- H =iscriminant index (classification score,>9 H net worth to total assets2+ H net worth to total liabilities

COMPUTATION OF HAIR6 ET AL, OPTIMUM CUTTING SCOREIn exhibit ) the - score for each company has been computed thus the centroids or the average

of - scores for each group can be computed as follows7Centroid for bankrupt group H #).32+Centroid for non#bankrupt group H ).)+@

$air et al. optimum cutting score2

1$1%51$24%7=

+

H -=,=5>

COMPUTATION OF ZETAC OPTIMAL CUT-OFF SCORETable 9 below shows the computation of prior probability of bankruptcy (" ), and prior probability of

non#bankruptcy ("3,.

T#)*e >, C!0'&#i!n !f P%i!% P%!)#)i*i( !f B#n$%&'c( 3+4 #n/ N!n-B#n$%&'c( 3<4

,ear Bankrupt Non-

bankrupt

Companies

Total

Companies

Companies

1""6 1 210 211 0$004%4 0$""526

1""% 15 212 22% 0$06608 0$"33"2

1""" " 225 234 0$03846 0$"6154

2000 24 230 254 0$0"44" 0$"0551

&era;e oA F1 and F2 0$050"43 0$"4"058

Table 9 shows that prior probability of non#bankruptcy is much more greater compared to priorprobability of bankruptcy. There is a @.4*2> G of probability of bankruptcy to happen while the probabilityof a company to not file for bankruptcy is *2.*4@5 G.

Table + below shows the type I and type II errors in metric forms happening in year )**9 )**+)*** and 3444 in 8ank 'iaga 8ank =anamon 8ank 8'I and 8ank 8%I.

T#)*e ., T('e I #n/ ('e II e%%!%" in 0e%ic f!%0" !f e#c1 )#n$ in (e#%" !f+>6 +.6 +6 #n/ <===

Name o' Banks ,ear Ci Cii

123






Ban Na;a 1""6 0$"41 0$012

Ban *ana(on 1""6 1$000 0$02"

Ban BNI 1""6 0$%4" 70$003

Ban BRI 1""6 0$618 0$056

Ban Na;a 1""% 0$838 0$045

Ban *ana(on 1""% 1$000 0$081

Ban BNI 1""% 0$%48 70$002

Ban BRI 1""% 0$%38 0$045

Ban Na;a 1""" 0$""" 70$084

Ban *ana(on 1""" 0$"54 70$050

Ban BNI 1""" 0$""4 70$00%

Ban BRI 1""" 0$"83 0$016

Ban Na;a 2000 0$""% 70$015

Ban *ana(on 2000 0$"85 0$014

Ban BNI 2000 0$"%" 0$036

Ban BRI 2000 0$%38 0$065

&era;e oA C dan C /.01 /./2

/rom table + above how costly type I error is compared to type II error can be computed bydividing the average value of Ci with the average value of Cii hence the result is :,@5 (4.5*):4.4)@,.

!ccordingly the cost incurred by the occurrence of type I error is @*.5> times more than the costincurred by the occurrence of type II error. The considerable cost that must be incurred if prediction errorsoccur moreover if type I error occurs reason for the determination of a new cut#off score that is -T!c

optimal cut#off score. Therefore the -T!c optimal cut#off score where the calculation will be elaboratedbelow is more tight in predicting meaning that more companies will be predicted as bankrupt while lesscompanies will be predicted as non#bankrupt compared to the prediction result that utili1es $air et al.optimum cutting score. In other words prediction using -T!c optimal cut#off score is more conservativeand more cost effective compared to prediction using $air et al. optimum cutting score.

=escription of the calculation of ")Ci and "3Cii for each bank and each year of )**9 )**+ )***and 3444 is shown in table 5 below.

8ased on table * below averages of ")Ci and "3Cii for years )**9 )**+ )*** and 3444respectively are 4.429 and 4.4)2. !ccordingly -T!c optimal cut#off score can be computed as

0140

0460'n e"ualling +,+.

T#)*e @, C#*c&*#i!n" !f +Ci #n/ <Cii f!% e#c1 )#n$ in (e#%" !f +>6 +.6 +6 #n/ <===Name o' Banks ,ear Ci Cii Ci Cii

Ban Na;a 1""6 0$"41 0$012 0$005 0$""5 0$004 0$012

Ban *ana(on 1""6 1$000 0$02" 0$005 0$""5 0$005 0$02"

Ban BNI 1""6 0$%4" 70$003 0$005 0$""5 0$004 70$003Ban BRI 1""6 0$618 0$056 0$005 0$""5 0$003 0$056

Avera%e o' Ci !an Cii 'or "ear 113 /.//4 /./5

Ban Na;a 1""% 0$838 0$045 0$066 0$"34 0$055 0$042

Ban *ana(on 1""% 1$000 0$081 0$066 0$"34 0$066 0$0%6

Ban BNI 1""% 0$%48 70$002 0$066 0$"34 0$04" 70$002

Ban BRI 1""% 0$%38 0$045 0$066 0$"34 0$04" 0$042

Avera%e o' Ci !an Cii 'or "ear 116 /./22 /./51

Ban Na;a 1""" 0$""" 70$084 0$038 0$"62 0$038 70$081

Ban *ana(on 1""" 0$"54 70$050 0$038 0$"62 0$03% 70$048

Ban BNI 1""" 0$""4 70$00% 0$038 0$"62 0$038 70$00%

Ban BRI 1""" 0$"83 0$016 0$038 0$"62 0$038 0$016

Avera%e o' Ci !an Cii 'or "ear 111 /./50 -/./5/Ban Na;a 2000 0$""% 70$015 0$0"4 0$"06 0$0"4 70$013

Ban *ana(on 2000 0$"85 0$014 0$0"4 0$"06 0$0"3 0$013

124






Ban BNI 2000 0$"%" 0$036 0$0"4 0$"06 0$0"2 0$032

Ban BRI 2000 0$%38 0$065 0$0"4 0$"06 0$0%0 0$05"

Avera%e o' Ci !an Cii 'or "ear /// /./06 /./5

T#)*e , A2e%#e" !f +Ci #n/ <Cii f!% (e#%" +>6 +.6 +6 #n/ <===,ear Ci Cii

1""6 0$004 0$023

1""% 0$055 0$03"

1""" 0$038 70$030

2000 0$08% 0$023

Avera%e /./43 /./4

PREDICTION RESULTS USING BOTH HAIR6 e #*, OPTIMUM CUTTING SCORE AND ZETAC OPTIMAL

CUT-OFF SCOREThe one year prior to bankruptcy prediction using either $air et al. optimum cutting score or

-T!c optimal cut#off score was performed by means of the Kif functionL. 8ankrupt companies were given a4 (1ero, code while non#bankrupt companies were given a ) (one, code. /or predictions using $air et al.optimum cutting score the Kif functionL made was if the - discriminant score of a company is smaller than J4.4>9 then that company will be categori1ed in the bankrupt group (code 4, and if incorrect will becategori1ed in the non#bankrupt group (code ),. !s for the prediction using -T! c optimal cut#off score theKif functionL made was if the - discriminant score of a company is smaller than ).)*5 then that company willbe categori1ed in the bankrupt group (code 4, and if incorrect will be categori1ed in the non#bankrupt group(code ),.

In table )4 and )) below prediction results using $air et al. optimum cutting score and -T! c

optimal cut#off score are shown respectively (see exhibit ) for details of prediction results,.

Table )4 elucidates that 35 bankrupt companies from the overall of 3* bankrupt companies werecorrectly classified and 32 non#bankrupt companies from the overall of 3* non#bankrupt companies werecorrectly classified. Thus the canonical linier discriminant model using $air et al. optimum cutting scorepredicted bankrupt companies as many as >> companies and non#bankrupt companies as many as 3@companies. It also can be concluded that ) type I error and @ type II errors occurred from the prediction.

T#)*e +=, P%e/ici!n %e"&*" 9i1 H#i%6 e #*, !'i0&0 c&in "c!%e

Classi'ications &re!icte! 7roup 8embership

Bankrupt Non-bankrupt Total

Count

Bankrupt 0 1

Non-bankrupt 2 4 1

9

Bankrupt 13.2 5.2 //:/

Non-bankrupt 6. 0.0 //:/

Table )) elucidates that 3* bankrupt companies from the overall of 3* bankrupt companies werecorrectly classified and )@ non#bankrupt companies from the overall of 3* non#bankrupt companies werecorrectly classified. Thus the canonical linier discriminant model using -T!c optimal cut#off scorepredicted bankrupt companies as many as 2> companies and non#bankrupt companies as many as )@companies. It also can be concluded that no type I error and )2 type II errors occurred from the prediction.

T#)*e ++, P%e/ici!n %e"&*" 9i1 ZETAc !'i0#* c&-!ff "c!%e

Classi'ications &re!icte! 7roup 8embership

Bankrupt Non-bankrupt Total

125






Count

Bankrupt 1 / 1

Non-bankrupt 4 2 1

9Bankrupt //./ /./ //./

Non-bankrupt 40.5 2.6 //./

It is understood from the prediction tables above (tables )4 and )), that predictions using $air etal. optimum cutting score is more accurate than predictions using -T!c optimal cut#off score. $oweverpredictions using $air et al. optimum cutting score will imperil more because by putting the same weightbetween type I error and type II error will result in the same opportunity towards the occurrence of type Ierror and type II error. !lthough more misclassifications were made in predictions using -T!c optimal cut#off score it is more safe or conservative because the chance of a certain company to be predicted asbankrupt becomes greater (from >> companies predicted as non#bankrupt using $air et al. optimum cuttingto 2> companies predicted as bankrupt using -T!c optimal cut#off score, and the chance of a certain

company to be predicted as non#bankrupt becomes smaller (from 3@ companies predicted as non#bankruptusing $air et al. optimum cutting to )@ companies predicted as bankrupt using -T!c optimal cutoff score,.Therefore $3 that indicated that financial ratios could be used to predict bankruptcy before hand is

accepted based on both prediction results using $air et al. optimum cutting score (35 bankrupt companiesfrom the overall of 3* bankrupt companies were correctly predicted and 32 non#bankrupt companies fromthe overall of 3* non#bankrupt companies were also correctly predicted, and -T!c optimal cut#off score(all bankrupt companies were correctly predicted and )@ non#bankrupt companies from the overall of 3*non#bankrupt companies were also correctly predicted,

$> indicated that the percentage of predicting bankrupt companies will become larger by giving agreater weight on type I than type II errors compared to the percentage of predicting bankrupt companies ofa prediction that gives the same weight on type I and type II errors. Predictions using $air et al. optimumcutting score that puts the same weight between type I errors and type II errors predict bankrupt companies

as many as >> companies while predictions using -T!c optimal cut#off score that puts a greater weight ontype I errors as much of @*5> times more costly than type II errors predict bankrupt companies as many as22 companies. Conse"uently $> is also accepted.

$2 indicated that predictions using -T!c optimal cut#off score would cut expected costs incurredfrom prediction errors. The cost cut even though the prediction results were not as accurate as predictionresults using $air et al. optimum cutting score can be seen from the success of -T!c optimal cut#off scorein deleting ) type I error that occurred in the prediction using $air et al. optimum cutting score. 6incetype I error is @*5> times more costly compared to type II error -T! c optimal cut#off score thatsuccessfully deleted the type I error even though only as many as ) type I error compared to $air et al.optimum cutting score is better to use in predicting bankruptcy in terms of more cost effective. !lthoughpredictions using -T!c optimal cut#off score result in more type II errors compared to predictions using$air et al. optimum cutting score (from @ type II errors using $air et al. optimum cutting score to )2 type II

errors using -T!c optimal cut#off score, predictions using -T! c optimal cut#off score is still more costeffective than predictions using $air et al. optimum cutting score.

The cost cut incurred in use of -T!c optimal cut#off score compared to the use of $air et al.optimum cutting score can be seen as follows7

-T!c optimal cutoff score $air et al. optimum cutting score

Type I error H 4 Type I error H )Type II errors H )2 Type II errors H @

Type I error H @*.5> type II error

126






Cost cut H @*.5> type II error J * type II errorsH @4.5> type II errors

8ased on the calculation above the use of -T!c optimal cut#off score in comparison with the useof $air et al. optimum cutting score will cut costs as much as @4.5> times of the cost of type II error. /orthis reason predictions of one#year prior to bankruptcy using -T!c optimal cut#off score will cut expectedcosts that may occur from predictions errors and is more conservative in performing predictions. Thus $ 2

has been accepted.

CONCLUSIONSIn this research four hypotheses have been developed and tested. The conclusions are as

follows7

). /inancial ratios do in fact differ between bankrupt and non#bankrupt companies. The ?ilksDambda statistical test proved that 33 ratios differ significantly in the level of significance of @G

between bankrupt and non#bankrupt companies.3. Two financial ratios from the overall of 33 financial ratios that distinguish bankrupt and non#

bankrupt companies can be used as variables of predictions in the one#year prior to bankruptcyprediction model. The two financial ratios belong to the Deverage and "uity group which are thenet worth to total assets ratio (>9, and the net worth to total liabilities ratio (2+,.

>. Predictions using $air et al. optimum cutting score correctly predicted 35 bankrupt companiesfrom the overall of 3* bankrupt companies and 32 non#bankrupt companies from the overall of 3*non#bankrupt companies. !s predictions using -T!c optimal cut#off score correctly predicted allbankrupt companies and )@ non#bankrupt companies from the overall of 3* non#bankruptcompanies. Thus it is proven that financial ratios can be used to predict bankruptcy beforehand.

2. Predictions using -T!c optimal cut#off score that puts a greater weight on type I error as much as@*.5> times more costly than type II error and a greater weight on prior probability of non#

bankruptcy (*2.*4@5 G, than prior probability of bankruptcy (@.4*2> G, predicted bankruptcompanies as many as 2> companies from the overall sample of @5 companies and predicted )@non#bankrupt companies from the overall sample of @5 companies. !s predictions using $air et al.optimum cutting score predicted bankrupt companies as many as >> companies from the overallsample of @5 companies and predicted 3@ non#bankrupt companies from the overall sample of @5companies./rom the one#year prior to bankruptcy prediction results it can be concluded that the use of -T! c

optimal cut#off score although not as accurate as predictions using $air et al. optimum cuttingscore is still better to use in terms of more conservative. In view of that a chance of a certaincompany to be predicted as bankrupt becomes greater and as non#bankrupt becomes smaller. Inother words the probabilty of occurrences of type I errors are minimali1ed.

@. The use of -T!c optimal cut#off score in comparison with the use of $air et al. optimum cutting

score will cut costs as much as @4.5> times of the cost of type II error. /or this reason predictionsof one#year prior to bankruptcy using -T!c optimal cut#off score is more safe cost effective andconservative in performing predictions.

RECCOMENDATIONS). Up to now independent variables used to distinguish between bankrupt and non#bankrupt

companies and then for performing bankruptcy predictions are merely from financial statementdata published by companies. There is a great possibility that market data such as stock pricemarket capitali1ation of a company macro#economic condition etc can be used to predictbankruptcy beforehand. /urther research is expected to incorporate market data in buildingbankruptcy prediction models.

3. The sample of bankrupt and non#bankrupt companies in building the one#year prior to bankruptcyprediction model is limited to companies listed on F6 in years of )*** and 3444. /urther research

12%






is expected to use wider range of pooled data so that the model built better represents the realcondition of the population.

>. The sample of banks in the computation of -T!c optimal cut#off score only uses two state#owned

banks and two private owned banks. It is expected that a specific research is conducted toinvestigate the cost occurring from type I errors and type II errors so that the computation of-T!c optimal cut#off score becomes more accurate as conducted by !ltman ()*++b, in hisexceptional research K/ending error costs for commercial banks* 0ome conceptual and empiricalissues' 1ournal of &ommercial 2ank /ending <ctoberL.

2. This research did not use a hold out sample because of time consideration and lack of data./urther research is expected to use a hold out sample (a different set of data than the data used inbuilding the bankruptcy prediction model, in order to validate the model and cut#off score built withthe intention that the results of the research also becomes more valid.

REFERENCES

!ltman .I %. $aldeman and P. 'arayanan. K ZT! !nalysis* ! new model to identify bankruptcy risk ofcorporations3 Fournal of 8anking and /inance Fune)**+ pp. 3*#@2.

!vianti Ilya. L4odel 5rediksi 6epailitan miten di 2ursa fek 1akarta 7engan 4enggunakan Indikator-Indikator 6euangan3 =isertasi 6#> Program Pasca 6ar&ana Universitas Pad&ad&aran 8andung3444.

8aridwan -aki. Intermediate !ccounting. disi Eetu&uh. 8P/ Yogyakarta)**3.8ernstein Deopold !. Fohn F. ?ild. 8inancial 0tatement !nalysis* Theory' !pplication' and Interpretation .

6ixth edition. 0cBraw#$ill International )**5./!68 0tatement of 8inancial !ccounting &oncepts 9o.)#<b&ectives of /inancial %eporting by 8usiness

nterprises ()*+5,./oster B. 8inancial 0tatement !nalysis. 6econd edition. Prentice J$all Inc. )*59$air Fr. Foseph /. %olph . !nderson %onald D. Tatham and ?illiam C.8lack 4ultiariate 7ata

!nalysis /ifth dition Prentice $all International Inc )**5.Ikatan !kuntan Indonesia. 0tandar !kuntansi 6euangan per # !pril ::. 6alemba mpat. 3443.Disetyati ni. !nalisis Daporan Eeuangan 6ebagai !lat Prediksi Eebangkrutan 8ank. Tesis 6#3

Program Pasca 6ar&ana Universitas Bad&ah 0ada Yogyakarta 3444.Distari Putu. K !nalisa ;asio 6euangan 0ebagai !lat 5rediksi 6ebangkrutan3 6kripsi %eguler

!kuntansi /akultas konomi Universitas Bad&ah 0ada 3443.<hlson F. !. K8inancial ;atios and the 5robabilistic 5rediction of 2ankruptcy3' Fournal of !ccounting

%esearch 6pring )*54 pp. )4*#)>).?atts %oss Dand Ferold D. -immerman 5ositie !ccounting Theory 'ew Fersey7 Prentice $all

International )*59.?ilkinson Foseph ?. !ccounting Information 0ystems* ssential &oncepts and !pplications Fohn

?iley A 6ons /ourth edition 3444.

E;hibit Hair, et

al ZETAc

C!/e #Constant$

/./03)53

-/.04)46 < optimum optimal

cutting cutoff

128






score score

0 72060 00"% 0108 7116% 0 0

0 72060 0036 003% 71%26 0 0

0 72060 0128 014% 708"1 0 00 72060 0083 00"0 71301 0 0

0 72060 0002 0002 72045 0 0

0 72060 0144 0168 70%50 0 0

0 72060 024" 0332 0180 1 0

0 72060 0144 0168 70%51 0 0

0 72060 01%5 0212 704%1 0 0

0 72060 0015 0015 71"22 0 0

0 72060 0102 0114 71121 0 0

0 72060 01%3 020" 7048" 0 0

0 72060 0086 00"4 712%0 0 0

0 72060 0100 0111 71146 0 0

0 72060 0065 006" 71462 0 0

0 72060 0036 003% 71%28 0 00 72060 000" 000" 71"%% 0 0

0 72060 0065 006" 71464 0 0

0 72060 0042 0043 716%6 0 0

0 72060 0086 00"4 712%3 0 0

0 72060 000% 0008 71""1 0 0

0 72060 00"5 0104 711"3 0 0

0 72060 0134 0154 7083" 0 0

0 72060 008% 00"5 71260 0 0

0 72060 0085 00"3 712%8 0 0

0 72060 0001 0001 72051 0 0

0 72060 00"2 0101 71218 0 0

0 72060 0154 0181 70661 0 00 72060 00"1 00"8 71218 0 0

1 72060 0%31 2%18 30%4 1 1

1 72060 0385 0626 1308 1 1

1 72060 0163 01"5 705%2 0 0

1 72060 0304 043% 0645 1 0

1 72060 02"% 0422 0585 1 0

1 72060 0522 10"2 2305 1 1

1 72060 0802 4043 26"4 1 1

1 72060 084" 5636 1862 1 1

1 72060 04%% 0"11 1""8 1 1

1 72060 0058 0061 7152% 0 0

1 72060 06"0 2223 3065 1 11 72060 0305 0438 0653 1 0

1 72060 0533 1141 23%5 1 1

1 72060 0241 031% 0108 1 0

1 72060 05"6 14%% 2%3% 1 1

1 72060 022" 02"8 0008 1 0

1 72060 0253 033" 0214 1 0

1 72060 0836 5084 21%" 1 1

1 72060 0880 %320 0%82 1 0

1 72060 0186 021% 70365 0 0

1 72060 0135 0156 70826 0 0

1 72060 0641 1%82 2"32 1 1

1 72060 03%1 05"0 11"% 1 1

1 72060 026% 0364 0332 1 0

1 72060 0505 1020 21"3 1 1

1 72060 0502 100% 21%1 1 1

12"



SIMPOSIUM NASIONAL AKUNTANSI VISurabaya 16 – 17 Oktober 2003



1 72060 0355 0552 10%1 1 0

1 72060 0145 01%0 70%34 0 0

1 72060 0424 0%35 1608 1 1

0 = +anr)pt1 = non7+anr)pt

130

iwan editan

Documents