Microsoft Word - 2977-11443-1-RV-writer2-new Co De Receive doi:10.5 Abstra Experim expecta the ban likeliho Logit, P all man The res has bee accurac firms w Keywo omparin Analy Depar Tel epartment o Te ed: Decemb 5296/ajfa.v5 ct mental stud ations in cap nkruptcy of ood of manu Probit, and nufacturing f sults showed en81, 80, an cy of Logit with high acc rds: Bankru ng Log ysis Mo rtment of Bu Is l: 98-912-2 Ma of Accountin Si el: 98-916-6 ber 31, 2012 5i1.2977 dies across pital market the compan ufacturing f Multiple-di firms listed d that the p nd % 70 res model and curacy in th uptcy, Logi git, Prob odels in Co Khalili usiness Man lamic Azad 217-2635 akvandi, Sar ng, Science istan and Ba 621-6032 2 Accept URL: http the world t is their abi nies. For thi firms and c iscriminant in Stock Ex redictability spectively. T also Multip he level logi it model, Pr 48 bit and n Predic ompani i Araghi, M nagement, S d University E-mail: m ra (Corresp e and Resear aluchestan, E-mail: sa ted: January p://dx.doi.or d show tha ility in pred is purpose, comparing t analysis. St xchange of y of Logit, The results ple discrim it model. robit model, Asian Multip cting B ies Maryam Science and y, Tehran, Ir m.khaliliarag onding auth rch Branch, Zahedan, Ir aramakvand y 16, 2013 rg/10.5296/a at one of i dicting the s this study the predicta tatistical po Tehran sinc Probit, and also showe minant analy , Multiple-d n Journal of F ple Disc ankrup d Research B ran ghi@gmail.c hor) , Islamic Az ran di@ymail.co Publish ajfa.v5i1.29 investors an tatus of acti tries to pred ability powe opulation of ce 2000-20 Multiple-d ed the highe ysis method discriminant Finance & Ac ISSN 19 2013, Vol. www.macrothi crimina ptcy of Branch com zad Univers om hed: June 1, 977 nd other a ivity consis dict the ban er of 3 met f the study i 10. discriminant er predictab ds identify b t model ccounting 946-052X 5, No. 1 ink.org/ajfa ant sity 2013 activists’ stency or nkruptcy thods of included t models ility and bankrupt 1. Intro Predicti 3 decad its cons very im compan (Tonatiu Unrelia that ma in the s patterns assuran Many fi 1. They 2. Con physica 3. A pa which a Regardi debt. The com causes value de reduces Therefo bankrup commo standard variable stateme 2. Rese Bankrup of the c bankrup others, predict reduce estimate oduction ing bankrup des (Ching– sequences a mportant for ny leads spr uh Pena et a ability from ay discourag same year, s for foreca nce to invest firms get ban y have to sel flicts amon al damage an art of compa are not as im ing these ca mpanies lac increase in ecrease enh s the current ore it seem ptcy of co on techniqu d modeling es used in ents (Etemad earch theor ptcy always community ptcy. But it Manageme bankruptcy the risk of e the comp ptcy has bee –Ching Yeh affect the ec r variety of read of crisi al.2009). survival of ge investme and is out asting firm t in stock m nkrupt annu ll their prop ng creditors nd inventor any value is mportant as ases, bankru cking debt h bankruptcy hances becau t value of th ms necessar mpanies. S ues in bank g technique Building di, 2007 ). ries and bac s has divert . It is very can be claim ent, investor y, not only t f your port pany's finan en a challen h et al.2010 conomy of public and s to other p f the existin ent in the S of stock. T ms that are market with t ually becaus perties with s may dela ry depreciati s spent for l 1 and 2. uptcy cost i have never b y likelihood use of the c he company ry to find Statistical t kruptcy pre es and has these mod ckground ed wide ran difficult to med that it rs, creditors o prevent th tfolio (Etem ncial bankru 49 nging issue 0). Bankrupt the country d commercia parts of the f ng company Stock Excha Therefore w going bank the lower ri se such situ low price. ay liquidati ion increase lawyers’ fee is high. It j been bankru d and reduc costs of bank y, enhancing a model th techniques ediction. Th focused o dels is the nge of indiv o provide a will be influ s, competito he risk of bu madi, 2007 ) uptcy, becau Asian for many sc tcy is impo y. In fact, b al organizat financial sy y in Stock E ange, becau with the de kruptcy, Ca sk (Rekabd ations: ing the ass es. e, court cos ust occurs f upt. So, fina ce earnings. kruptcy. Inc g its capital hat can pre are consid his model n signs bu informatio viduals, orga an accurate uenced ban ors, and leg urning their ). Hence, T use in case n Journal of F cientific stu rtant in fina ankruptcy p tions. Becau stem and ca Exchange is use perhaps esign and p an be gave ar et al,200 sets. Then st, and organ for the com ancial provis As a resul creasing ban costs (Briga edict the fi dered as th has been usiness failu n containe anizations a definition kruptcy phe gal institutio r capital, bu They are th of bankrup Finance & Ac ISSN 19 2013, Vol. www.macrothi udies during ancial studi prediction h use the fail ause system s one of the bankrupt c propose app e the invest 06 ). the probab nizational e mpanies whi sions throug lt, the likeli nkruptcy lik am, 2003 ). financial cr he most ba used from ure. Genera ed in the f and the gen for interest enomena m ons. Investo ut its use as he ways tha ptcy, stock ccounting 946-052X 5, No. 1 ink.org/ajfa g the last ies since has been lure of a mic crisis e factors company propriate tors this bility of expenses ich have gh debts ihood of kelihood isis and asic and m classic ally, the financial eral part t groups more than ors with a tool to at could price of compan taken in impact to corp bankrup Theref process Comme For pre viewpo strategi models likeliho Bankrup intellige univaria discrim adjustm Artifici machin techniqu and the Bankrup risk.(Fir Komija model t multiva liabilitie power al(2011 The res model h t-2, resp years t, a resear regressi third re way the did a r compar nies sharply nto conside on lending porate loan ptcy. fore, accura s and the Co ercial units eventive me int, tools pr c actions an is one of th ood of bankr ptcy predic ence, and t ate and m minant analy ment process al neural n es, Sion-ba ues eoretical m ptcy theor rouziyan et ani et al (20 to predict t ariate probi es,gross pro than comp )compared sults showe have better pectively, % t-1 and t-2, rch titled" ion to class espectively % e logit mod research tit red two mod y decreases eration wide decisions fo ns demand ate and tim ontinuance o managers a easures wh redicting fin nd avoided he techniqu ruptcy estim tion models theoretical. multivariate. ysis models, ses. networks, g ased reaso models is a ry of gam al,2010) 005) condu the econom t and logit ofit to sale pany econo Ohlson log d Ohlson lo than perfor %96, %91 , respective Accounting sify the com %85.1, %87 del to discr tled "Bankr del of Logit (Rasoulzad ely in Acco for creditors they shou mely predict of credit ins also can be ich prevent nancial ban from bankr ues and tool mate with co s can be cla Statistical . Multivari , linear prob enetic algo ning and also includ mbling, Ca ucted a rese mic bankrupt t models w s and net p omic bank git analysis ogit analysi rmance. Oh and %89, ly, %85, % g Data and mpanies stu 7.6 and %8 riminate ana ruptcy pred t and Multip 50 deh,2000). P ounting and s and also th uld be abl tion of ban stitutions (A informed t t bankruptc nkruptcy pro ruptcy(Arab ls to predict ompositions assified in th models th iate statist bability, log orithms, rec fuzzy logi ding criterio ash manage earch titled tcy of comp with explana profit to cu kruptcy in s and multip is model th hlson model While the 74and 76% d the Predic udied. His m 82.6.Hough alysis techn diction: an ple-discrimi Asian Predicting f d financial heir profitab le to predi nkruptcy ha Arabmazar e timely of th cy (Etemad ovides this bmazar et a t the future s a group of hree groups hemselves a ical model git, probit,th ursive Afra ic, are com on analysis ement theo "The cond panies in Ir atory variab urrent debt Iran. (Ko ple discrim an multiple prediction fulmer mod 6.(Vadiee et ction of Bu model accu hton in their niques.(Hou investigati inant analys n Journal of F financial ba field. Such bility. Befor ict possibil as a signific et al, 2007 ). he risk of b di,2007). Fr possibility al,2007).Ban state of the f financial ra , statistical are divided l is comp he total cum az, rough s mposed art s of sheet ory and t ditional like ran". The r bles curren have maxim mijani et minant analy e discrimina accuracy is del predicti t al,2011)H usiness Fail uracy was fo r study conc ughton 1994, on of cash sis models. T Finance & Ac ISSN 19 2013, Vol. www.macrothi ankruptcy h research h re creditors lity of com cant impact bankruptcy rom a mana to adoption nkruptcy pr e company' atio. modeling, a d into two posed of M mulative and sets, suppor tificial inte / entropy theories of elihood of results show nt assets to imum of pr al 2005, )Va ysis fulmer ant analysis s for years t ion accurac Houghton (19 lure" using or the first cluded that 4)Aziz et a h flow Mo The results ccounting 946-052X 5, No. 1 ink.org/ajfa has been has great respond mpany’s t on the adopt to agement n timely rediction s in that artificial groups, Multiple d Partial rt vector elligence theory, f credit optimal wed four current redictive adiee et models. s Fulmer t, t-1and cy is for )994 did logistic years to a better l (1988) dels" to showed better p discrim The res compar overall to 89 p sample importa size, in and Pro (Lennox Thus, th Exchan bankrup 3. Meth In this probit, busines the rank likeliho followin Where, paramet p(z) wil resultin event o include Probit m the valu to Logi than cum performance minant analy ults showed re with desi accuracy be percent (Az including ant bankrup ndustrial sec obit models x, 1999). his study to nge, compa ptcy such L hodology study, we and multip ss failures. B king of eve ood in a def ng formula: Xi (i=1,… ters of the m ll tend to ze ng likelihoo or not, predi a range of model is use ues 0 and 1. t models. B mulative lo e of logistic sis. d better perf igned mode etween 79% ziz et al,198 949 Englis ptcy factors ction, and e s, compared o predict the aring the re ogit, Probit use to pre le discrimin By allocatin ery sample finite group. : , n) shows model. P (z) ero and if Z od equal to0 icted likelih values betw ed when the . (Such as b But, the form gistic funct c regression formance of el based on 9 to %92 an 88).Lennox( sh compani include pr economic cy d with disc e bankruptcy esults of t t, and Multip dict corpor nant analys ng some w company. . Success or 11 e independen ) likelihood Z move into 0.5.Logit an hood of the ween 0 and e dependent ankruptcy a mer uses cu ion. , a P a 51 n model, th f designed m n multiple d nd was accu (1999)exam ies since 1 rofitability, ycles. The criminant an y likelihood the applica ple-discrim rate bankru sis. Logit m weights to in This ranki r failure lik 1 e nt variables d is a numb +∞, p(z) w analysis inst event. Thu 1. t variable,w and non-ban umulative li a = Asian e model wa model by th discriminan uracy of dis mined the re 1978-1994. financial le results show nalysis in p d of manufa tion of 3 minant analy uptcy from t model has w ndependent ing is used kelihood in t 1 ... s, and a an er between will tend to tead of pre us dependen was qualitati nkruptcy)Pr ikelihood fu a n Journal of F as designed he logistic re t analysis. criminant a easons for t Based on everage, cas wed higher predicting c acturing firm statistical sis. three statist wide applica variables, t for determ this model i nd bi (i=1,… 0 and 1.If 1.when Z i edicting wh nt variable c ve and be a robit models unction whi Finance & Ac ISSN 19 2013, Vol. www.macrothi d based on m egression m Logit mode analysis betw the bankrup Lennox, th ash flows, c r accuracy o companies' ms in Tehra models to tical model ations in pr this model mining mem is calculate …, n) are es f Z move in is equal to z hat will be could encom able to prov s are mostly ich is norma ccounting 946-052X 5, No. 1 ink.org/ajfa multiple model, in els have ween 73 ptcy of a he most company of Logit failures an Stock predict ls, logit, redicting predicts mbership d by the stimated nto ∞, zero, the actually mpass to vide only y similar al rather Multipl into dis among For suc prerequ context identifie contain this me achieve Classifi Multiva point, it the dis Equatio :T where ‘ companx : T :Coe u :is th The firs corpora had mo 3.1 Var To iden principa large nu several the who highest collecti This ap variable Varianc called t analyzin le-discrimin stinct group the groups ch predictio uisite for MD of ratio-b ed: the firs s companie thod, a com ed score and ication w ariable-discr t was regard scriminant on 1 gives The value th ‘ k’ represe nies. ‘ fkm’ The value fo efficients as he constant st time Altm ate failure.T deling more riables ntify the m al compone umber of in principal c ole variance amount in ons of varia proach will e with each ce explained the specific ng principa nant analysi ps with diff and predict ons, some DA is that based mode st group in es that are s mpany was d the maxim was done riminant an ded as a non function, a the general hat the disc nts either th is also know or the financ sociated wi term or the man (1968) This study a e than one v most importa ent analysis nitial variab omponents e. At first Sp n total var ables that ha l apply to se factor is cal d by each fa c value. Th al compone s is a multiv ferent quali t the likelih quantitativ a data item els for sign ncludes com still a going attributed to mum similar based o nalysis mode n-bankrupt c and the fin specificatio u u x criminant fu he group of wn as the sc cial ratio ‘i’ ith each fina intercept. suggested m lot attentio variable. ant financia was used. bles into a are to revie pss, selects c iance. This ave the high elect the thir lled factor l factor equal he first spec ents, the va 52 variable me ities. This s hood of a co ve independ m could be c naling corp mpanies tha g concern. o a bankrup rities. on optimu el. If the sco company.T nancial rati on of the Mu x unction gen f collapsed core. ’ for compan ancial ratio multiple dis on because f al ratios fo In principa small num ewed so tha combination s collection hest contribu rd, fourth an lodding and to the squa cific value alues over Asian ethod in whi study aimed ompany’s de dent variab classified in porate colla at have col The data it pt or non-ba um short ore of a com he mathema ios are cal MDA-based m⋯ u nerates for c companies ny ‘ m’ in g ‘x ’. scriminant a for the first or selecting al componen mber of new at the new v n of variable n makes fi ution in expl nd subseque d their Value are of the fa of the high 1 were reg n Journal of F ich the phen d to recogn ependency t bles were u nto two or m apse, two g lapsed and tem is there ankrupt gro cut point mpany was atical equat lled discrim model. u x company ‘ or the grou group ‘ k ’. analysis me time in ban main varia nt analysis, w variables variables ha es that their irst factor. laining the r ent factors. e fluctuates actor loddin hest and is garded and Finance & Ac ISSN 19 2013, Vol. www.macrothi nomena are nize the dif to a specifi used.Thus, more group groups are d the secon efore a com oup accordin t, identifi lower than tion is referr minating va m’ in grou up of non-c ethod for pr nkruptcy pr ables of th convert all that the sa ave a large r correlation Second F remaining v Correlation between -1 ng .This var greater tha d used as th ccounting 946-052X 5, No. 1 ink.org/ajfa e divided fferences c group. a basic s. In the readily d group mpany.In ng to its ed for shortcut red to as ariables. up ‘ k ’, ollapsed redicting rediction e study, lowing a ame first share of ns are the Factor is variance. n of each and +1. riance is an 1 . In he most signific proport or 90 p was ide Indepen -Return -Debt ra -Debt c -Collect -Invento -Debt to -Produc -Depen 4. Resu Statistic Exchan 134 no Busines the half bankrup 1. The c 2. Their 3. Finan 4.They 2000- 2 4.1. log Logit re cant specifi ional of the percent, to e entified. ndent variab n on equity ( atio over ratio tion period ory turnove o equity rati ct to workin dent variab ults cal populati nge of Tehra on- bankrupt ss Law in Ir f of their c pt companie companies s r fiscal year ncial inform should hav 2010. git model egression m c values.To e change tot express initi bles of the s (ROE) er io ng capital ra le was the l ion of the s an since 200 t companies ran, based o capital must es was base should be m r should end mation of the ve 10 succ model fitness o choose t tal percenta ial changes. study includ atio ikelihood o study includ 00-2010. To s were used on which th t declare ba d on the fol manufacturin d in Septem e companie cessive yea s was tested 53 the number age cumulat Thus After de: of bankruptc ded all man o measure m d. Bankrupt he firms wit ankruptcy o llowing con ng. mber. s should be ars of activ d based on th1 1 Asian r of main tive so that r examining cy or non- b nufacturing model fitnes tcy measure th minimum or capital lo nditions: accessible. vity in Stoc he followine e ε n Journal of F variables c the this ma g 22 financi bankruptcy o companies ss, the data o e in this stud m accumulat oss. Sample ck Exchang ng equation: Finance & Ac ISSN 19 2013, Vol. www.macrothi can be con ain compone ial ratios, 7 occurrence. s accepted i of 55 bank dy was Act ted loss, eq e selection ge of Tehra ccounting 946-052X 5, No. 1 ink.org/ajfa nsidered ents, 80 7 factors in Stock rupt and 141 of qual with of non- an since Table 1 Table 2 0.000 Table 3 RO Inv Co Pro De De De Co Accord conclud coeffici gives th 48 Table 4 Observed B Percent Estimat bankrup .1 Regressi Coeffi Nagel 0.273 .2 Significa 3. Logistic r OE ventory turnov llection period oduct of work bt ratio bt to equity ra bt coverage ra nstant ing to likeli ded that the ient it identi he following7.667 0.00 4. To accura Bankruptcy tion accurac pt companie ion statistics icient of deter kerke ant regressio Sig. regression c ver d ing capital rat atio atio ihood value e model is ifies %27 o g formula: 024 0.0.005X ε ately estima 0 1 cy was %81 es shown in s Co Co rmination 0. on test 7 coefficients Sig. 10.000 10.09 10.761 10.927tio 10.005 10.004 10.691 10.054 e in significa statistically f distributio 003 0ε ate of bankr Estimation 0 133 35 1. For non-b n Table4. 54 oefficient of d ox & Snell .191 df 40.055 Walddf 14.261.000 2.8211.000 0.0921.000 0.0081.000 7.9831.000 8.2141.000 0.1581.000 3.7231.000 ance test (P y significan on. Table 3 .0004X ruptcy n of bankrupt 1 1 2 bankrupt co Asian determination 5 Std Statistic 0.064 0.0 0.02 0.08 0.03 0.04 0.08 253 =0.000) sho nt and due shows logis 0.001X cy 1 1 20 mpanies, it n Journal of F -2 Log like 187.898 tandard deviat 006 002 001 007 078 080 012 52.742 own in Tabl to its resu stic regressi 0.220X Percentag 99.3 36.4 81.0 was %99 a Finance & Ac ISSN 19 2013, Vol. www.macrothi elihood C Coeffiction 0.024- 0.003 0.0004 0.001 0.220 0.228- 0.005- 487.667 le 1 and 2, i ulted determ ion coeffici 0.228X ge of accuracy and it was % ccounting 946-052X 5, No. 1 ink.org/ajfa Chi-square cient 7 it can be mination ents and y %36 for 4.2 Pro Probit r Table 5 0.000 Table 6 Intercept ROE Inventory tu Collection p Product of w Debt ratio Debt to equ Debt covera Accord conclud coeffici Table 7 obit model regression m 5. Significan 6. Probit reg urnover period working capit uity ratio age ratio ing to like ded that th ients and giv1 0 7. To accura Obser Percen model was m nt regressio Sig. 7 gression coe al ratio lihood valu he model i ves the follo135 00.0742X ately estima rved Bankrupt nt measured ba 1 on test 7 efficients Sig. 0.279 0.000 0.059 0.924 0.892 0.012 0.009 0.724 ue in signif is statistica owing form0.01020.0024X ate of bankr Estim 0 1330 tcy 371 55 ased on the df 34.984 Wal df 1.1711 13.9101 3.5571 0.0091 0.0181 6.3431 6.8561 0.1241 ficance test ally signific mula: 0.0019ε ruptcy mation of ban 1 1 18 Asian following e 4 ld Statistic 1 0 0 0 0 0 0 0 t (P=0.000) cant. Table 0.0001X Per nkruptcy 99 338 80. n Journal of F equation: ε Standard d 124.7689 0.0027 0.0010 0.0007 0.0040 0.0363 0.0283 0.0069 ) shown in e 6 shows X 0.0005 rcentage of ac 0 Finance & Ac ISSN 19 2013, Vol. www.macrothi deviation 1- 0. 0- 0- 0- 0- 0. 0. n Table5, it s probit re 5X 0.09 ccuracy ccounting 946-052X 5, No. 1 ink.org/ajfa Chi-square Coefficient 135.00 .0102 0.0019 0.0001 0.0005 0.0915 .0742 .0024 t can be gression 15X Estimat bankrup 4.3 Mu Multipl Table 8 Table 9 Sig. 0.243 Table 1 ROE Inventory t Collection Product to Debt ratio Debt to equ Debt cover (Constant) Regardi signific 8 and Table formula Canonical C 0.220 tion accura pt companie ultiple-discr le-discrimin 8. Table eig 9. Table me 10. Table no turnover period working capit uity ratio r ratio ing the lik cance level, 9. 10 shows a: Correlation acy was 80 es, shown in riminant mo nant model w genvalue fun eaningful tes df 7 on-standard tal ratio kelihood va it can be sa non-standa C %0 . For no n Table7. odel was examin v nction st C 9 dized coeffic alue in sig aid that the m ardized coe Cumulative % 100.0 56 on-bankrupt ned based on b b Chi-square 9.130 cient Focal gnificance model is sta efficient Fo Asian t companie n the follow b xj functions test (P=0.2 atistically in ocal functio % of Varianc 100.0 n Journal of F s, it was % wing equatio Wilks' Lam 0.951 243) and c nsignificant ons and gi ce Finance & Ac ISSN 19 2013, Vol. www.macrothi %99 and % on: mbda Function 1 0.001- 0.001 0.003 0.001 0.121 0.011- 0.015- 55.071- comparing t, as shown ives the fo Eigenvalue 0.051a ccounting 946-052X 5, No. 1 ink.org/ajfa 33% for it with in Table ollowing Table 1 Observed Percent Estimat was %8 Table 1 regressi Multiple Table. Multipl regressi Multipl Conclu In this analysis model w Logit m model f indicati The res corpora compan non-ban bankrup showed model. 55.070 11. To accu Bankruptcy tion accurac 84 for non-b 12. Ratio t ion. discriminant 12 show le-discrimin ion is not h le-discrimin usion study we c s models to was %81, fo model, Could from 55 ba ive high pow sults of Pro ate bankrup nies is %9 nkrupt com pt companie d that Probit 7 0.0010.015X urately estim Estim 0 112 0 35 1 cy in the sa bankrupt co test for com analysis Data e ws the Ra nant regress higher. Also nant regressi compared p o predict b or identify n d predict fro ankrupt com wer logit mo obit model c ptcy has % 99 and for mpanies wer es has been t model to 0.001 mate of bank mation of ban ample is sh ompanies an mparing th envelopment a atio test f ion.The res o,Z test res ion and DE redicting p ankruptcy.T non-bankrup om 134 non mpany, have odel in pred confirm tha 80% .accurac r bankrupt re able to p n able to co identify ban 57 0.003X kruptcy nkruptcy 1 22 20 hown in Ta nd %36 for he models 69.8 72.0analysis for compa sults show th sults don’t EA. ower of Lo The results pt companie n-bankrupt c e been able diction of co at can predi cy of the P companies predict corr orrectly pred nkruptcy fir Asian X 0.001X able11. Esti r bankrupt c of efficien Accura 8% 0% aring the hat the accu confirm a s ogit, probit showed th es %99 and companies, to predict b orporate ban ict power th Probit mod s is 33 pe rectly 133 dict bankru rms, no sig n Journal of F X 0.121 Percen 83.6 36.4 69.8 imation acc ompanies. cy and Mu p- value acy 0.325 models o uracy of Mu significant d and multiv hat predictin d %36 for ba 134 compa bankruptcy nkruptcy. his model i dels to iden ercent.This companies uptcy 18 co nificant dif Finance & Ac ISSN 19 2013, Vol. www.macrothi X 0.011 ntage of accur curacy was ultiple-discr Z t 0.4 of efficien ultiple-discr difference b variate discr ng power o ankrupt com anies correct 20 compan in the predi ntify non-b model fro and also f ompany.The fference in t ccounting 946-052X 5, No. 1 ink.org/ajfa 1X racy %70. It riminant test 453 cy and riminant between riminant of Logit mpanies. tly. This nies that iction of bankrupt om 134 from 55 e results the logit The res about % corpora This m correctl recogni perform and pre has a pe has a lo bankrup Referen Arabma Accoun Aziz, A flow M Brigam Ching-C set and Applica Etemad monthly Firouzia bankrup Stock E Hought of Prio http://dx Komija the econ Lennox DEA http://dx Rasoulz compan Rekabd in predi ults of the M %70. Accur ate bankrupt model could ly. The resu ition of ban mance comp edictive pow erformance ower predic pt companie nces azar, M., & ntant month A., Emanuel odels. Jour m,W. (2003). Ching, Y., D d support ation, 37(2), di, H., & De y, 5, 40. an, M., Jav ptcy predict Exchange. R ton, K. A. ( ors and th x.doi.org/10 ani, A., & S nomic bank x, C. (1999) approache x.doi.org/10 zadeh, M. nies. Tadbir dar,G. H., C iction of ban Multiple dis racy of the tcy and %36 d predict on ults show th nkrupt comp pared to pro wer of logist near to log ctive power es probit mo Akbari, M ly, 200, 34 l, D., & Law nal of Mana . Financial Der Jang, C vector mac , 1535-1541 ehkordi, H. vid, D., & N tion and co Review of ac 1994). Acco he Age of 0.2307/2490 Saadatfar, J. kruptcy of co ). Identifyin es. Journ 0.1016/S014 (2000). Ap monthly, 12 hinipardaz, nkruptcy. Jo scriminant a e Discrimin 6.4 for the nly 112 co hat Multipl panies in c obit model. tic regressio git model bu r in compar odel has bet . (2007)Pre wson, H. (1 agement Stu managemen Ch., & Ming chines for 1. (2007). A r Najmodini, ompared wit ccounting an ounting Dat f Data. Jo 0717 (2005). Th ompanies in ng failing c al of E 48-6195(99 pplication A 20, 100. R., & Solim ournal of Ap 58 analysis mo nant analys prediction o ompanies fr e discrimin compared to The resear on is higher ut in lower re to two o tter perform edicting corp 988). Bank udies, 25(5) nt. Translate g, Fu, H. (2 business review of b N. (2010). th Altman Z nd auditing ta and the P ournal of he condition n Iran. Jour companies: Economics 9)00009-0 Altman mo mani, B. (2 pplied Math Asian odel showed sis models of corporate rom 134 no nant analysi o logit mod rch results s r than the ot level, Mult other model mance. porate bank ruptcy pred ), 437-419. ed by Abdeh 2010). A hyb failure pre bankruptcy p Application Z model Co g in Iran, 65 Prediction o Accounting nal likelihoo rnal of Hum A re-evalua and B odel at stat 006). The D hematics, 12 n Journal of F d that accura %83.6 for e bankruptcy on-bankrup s model ha del level an show that th ther two mo iple discrim ls. But in c kruptcy usin diction:an in h, H, Pishbo brid approa ediction. Ex prediction m n of genetic ompanies li , 100. of Business g Research od of optim an Sciences ation of the Business, tus determi Discriminan 2, 25-15. Finance & Ac ISSN 19 2013, Vol. www.macrothi acy of this m r the distin cy. pt firms, to as high accu nd has muc he overall a odels. Prob minant analy contrast, to ng neural ne nvestigation ord,PP21 ach of DEA xpert Syste models. Acc c algorithm isted on the Failure: the h, 22(1), 3 mal model to s, 57, 28-3. e Logit, Pro 51(4), 3 ination ban nt analysis o ccounting 946-052X 5, No. 1 ink.org/ajfa model is ction of predict uracy in ch better accuracy it model ysis also identify etworks, n of cash A ، rough em with countant ms in the e Tehran e Setting 361-368. o predict obit, and 347-364. nkruptcy of mixed Tonatiu Mexico Vadiee, analysis research Copyri Copyrig This ar Creativ uh Pena, C. o Working P M., & Mi s and multip h, 13, 23-1. ight Disclai ght reserved rticle is an e Commons Serafin Ma Paper. Bankr resmaeeli, ple discrimi . imer d by Maryam n open-acce s Attributio artinez, & J. ruptcy Pred H. (2011). inant analys m Khalili A ess article d n license (h 59 . Bolanle, A diction: A Co Predicting sis Fulmer A Araghi and S distributed http://creativ Asian A. (2009). L omparison o bankruptcy And compa Sara Makva under the vecommons n Journal of F Learning Tec of Some Sta y by using re them. Ac ndi. terms and s.org/license Finance & Ac ISSN 19 2013, Vol. www.macrothi chniques. B atistical Mo ohlson logi ccounting a d conditions es/by/3.0/). ccounting 946-052X 5, No. 1 ink.org/ajfa Banco de odels. it model nd audit s of the