Failure prediction from the investors’ view by using financial ratios. Lesson from Romania

117 4, XIX, 2016 Finance DOI: 10.15240/ ul/001/2016-4-009 In oduc ion The economy’s globaliza ion which culmina ed wi h he global i nancial c isis ha e up ed in 2007 has a ec ed he economies a ound he wo ld demons a ing ha he esul s o esea ch s udies on bank up cy isk p edic ion a e insu i cien . Fu he mo e, he s udies conduc ed so a do no p o ide di ec insigh in o he capi al ma ke o in es o s o whom he i s s ep in a company’s ailu e is mani es ed by delis ing he company om he s ock exchange. Acco ding wi h Al man (1968) “in he case o lis ed companies, in a 4 days in e al be o e announcing bank up cy, in es o s lose abou 41% o he capi al in es ed in bank up companies.” The abili y o p edic he companies’ bank up cy om he ea ly s age o delis ing is he main no el y o he p esen s udy. In he specialized li e a u e he e a e ela i ely ew conce ns abou bank up cy’s isk p edic ion om he in es o ’s pe spec i e. Mos s udies e alua ing he bank up cy isk we e s uc u ed o app oach bank up cy om a “legal” pe spec i e acco ding o which companies we e g ouped in o ailing companies ( o which he e is a s a emen in his ega d a he cou ) and he o he g oup consis s o non- bank up companies. The e is hough, ano he g oup o s udies which analyse bank up cy isks om an “economic” pe spec i e acco ding o which he companies we e g ouped in o ailing companies ( ep esen ed by companies wi h low i nancial and economic pe o mance) and non-bank up companies ( ep esen ed by companies ha ha e di e en pe o mance indica o s conside ed o be high). An e en ewe numbe o s udies a e ocused (in he economic app oach) on he in es o s’ angle, acco ding o which companies become bank up beginning wi h he delis ing phase (Ch is idis & G ego y, 2010; Tu ada a agool, 2013; Wang & Campbell, 2010). Ou pape has he ollowing main objec i es: Fi s o all, we aim o iden i y which i nancial indica o s ha e a signi i can impac on he p obabili y o a company o ace bank up cy isks exp essed om he in es o s’ pe spec i e by s udying he impac on he p ospec s o s ocks’ delis ing om he s ock exchange. Secondly, we would like o de e mine o wha ex en a e hese indica o s iden i i ed in o he s udies as signi i can bank up cy signs. Is he e any di e ence be ween he indica o s iden i i ed as ha ing a majo impac om he in es o s’ pe spec i e compa ed o he gene al one? The pape is s uc u ed as ollows. In he nex sec ion (Sec ion 1) we p esen a e iew o he li e a u e in e ms o bank up cy isk whe e he “ ailu e” concep is p esen ed, we p esen he impo ance gi en o i nancial a es in bank up cy isk s udies highligh ing he main me hods used in bank up cy’s isk analysis and i nally we emphasize he ime and space limi s o ailu e p edic ion models ha exis in he li e a u e. Sec ion 2 e e s o he gene al p esen a ion o he elabo a ion me hodology. Fu he , Sec ion 3 is dedica ed o he desc ip ion o he ob ained empi ical esul s, hei alignmen o o he empi ical i ndings and o highligh ing he changes in bank up cy p edic o s. The las pa o he pape (Conclusions) is alloca ed o desc ibing he conclusions and o highligh ing he added alue o his s udy in o de o co e some gaps in he niche o de eloping some bank up cy isk models om he pe spec i e o he capi al ma ke in es o s. FAILURE PREDICTION FROM THE INVESTORS’ VIEW BY USING FINANCIAL RATIOS. LESSON FROM ROMANIA Monica Viole a Achim, So in Nicolae Bo lea, Lucian Vasile Găban EM_4_2016.indd 117EM_4_2016.indd 117 6.12.2016 10:50:186.12.2016 10:50:18 118 2016, XIX, 4 Finance 1. Li e a u e Re iew 1.1 On he “Failu e” Concep Fi s ly i is necessa y ha we cla i y he e m o “ ailu e”. In es iga ing he s udies on bank up cy isk, we can conclude ha co po a e “ ailu e” is app oached om wo pe spec i es: a legal one and an economic one. Unde a legal aspec , a company’s bank up cy occu s when he e is a s ong bank up cy s a emen in a cou . Fo US an ailing i m migh i le ei he o liquida ing i s asse s o o business eo ganisa ion (Al man & Ho chkiss, 2006). Acco ding o Balcaen and Ooghe (2006) he mos s udies used ‘legal’ in e p e a ions o he ele an e ms because o he objec i e’s pe spec i e o disc imina e ailed and non- ailed i ms. Among hese s udies we can men ion he ollowing: Anghel (2002), Al man (1968), Bea e , McNichols and Rhie (2005), Chi and Tang (2006), Pla and Pla (1990; 1991; 2008), Shumway (2001). Unde an economic aspec , a ailu e can be de i ned h ough he company’s i nancial pe o mance ha can be ep esen ed by: “Insu i cien e enues o co e cos s and whe e he a e age e u n on in es men is below he i m’s cos o capi al” (Al man & Ho chkiss, 2006); “nega i e equi y and/o nega i e ea nings” (Robu-Mi oniuc, 2012; Tu ada a agool, 2013); “ educ ions in di idends, iola ions o deb co enan s” (Tu ada a agool, 2013); “going p i a e o a publicly lis ed company” which is simila o delis ing he companies’ sha es (Ch is idis & G ego y, 2010; Ohlson,1980; Tu ada a agool, 2013; Wang & Campbell, 2010). On he o he hand, he e a e a ious s udies which op o “economic” in e p e a ion when de e mining he wo ypes o companies: bank up o non-bank up ; as ollows: Bea e (1966; 2005), Ch is idis and G ego y (2010), Ohlson (1980), Robu-Mi oniuc (2012), Tu ada a agool (2013), Wang and Campbell (2010). Acco ding o Al man and Ho chkiss (2006) ou common e ms a e widely used in bank up cy s udies, namely: ailu e, insol ency, de aul and bank up cy. We can add he e he e m o “ i nancial dis ess” which is simila wi h he economic ailu e (Tu ada a agool, 2013) which means ha he company has i nancial p oblems bu is no in bank up cy ye . As we can see, he e a e a wide ange o applica ions o he e m “ ailu e”, acco ding o he speci i c objec i es o each s udy and he speci i c needs o a ious decision-make s. In his pape , he concep o co po a e “ ailu e” is used om he in es o s’ pe spec i e, o whom he delis ing o he company om he s ock ma ke is synonymous wi h bank up cy i sel because hei in es men is comp omised, a ading pla o m wouldn’ be exis ing anymo e (Wang & Campbell, 2010). In o he wo ds, in ou s udy, bo h business “ ailu e” and “delis ing” a e used in e changeably. 1.2 The Role o Financial Indica o s in P edic ing he Financial Dis ess The in e na ional accoun ing egula ion consis s in In e na ional Financial Repo ing S anda ds. The i nancial s a emen s’ objec i e is o p o ide in o ma ion abou he i nancial posi ion, pe o mance and changes in he en i y’s i nancial posi ion ha a e use ul o a wide ange o use s in making economic decisions. I goes wi hou saying ha he i nancial in o ma ion om he i nancial and accoun ing s a us is designed o highligh he company’s i nancial condi ion. Mo eo e , se e al s udies conduc ed by Bea e , Co eia and McNichols (2010) highligh ha “ he i nancial s a emen s ha e been used o mo e han 100 yea s o assess i nancial dis ess’ likelihood”. F om he i nancial in o ma ion’s ca ego y, i nancial a es a e he mos commonly used in bank up cy isk assessmen because i is belie ed ha hei use compa ed o ha o he indica o s’ absolu e le els p o ides a gene al deg ee o applicabili y o he companies. By using i nancial a ios, he limi s gene a ed by he companies’ size a e he eby signi i can ly educed. The p e e ence o using i nancial a es in bank up cy isk assessmen has been mani es ed since he s udies o Bea e (1966) and Al man (1968). In his uni a ia e analysis, Bea e (1966) iden i i ed he cash l ow indica o o o al deb a io o be ex emely sensi i e o a company’s i nancial condi ion. Subsequen ly, Al man (1968) in oduced he mul i a ia e analysis and iden i i ed i e ep esen a i e i nancial a ios o he i nancial condi ion: wo king capi al/ o al asse s; o al e ained ea nings/ o al asse s; ea nings be o e in e es and axes/ o al asse s; ma ke alue o equi y / book alue o o al deb ; sales/ o al asse s. EM_4_2016.indd 118EM_4_2016.indd 118 6.12.2016 10:50:196.12.2016 10:50:19 119 4, XIX, 2016 Finance Ano he pionee o bank up cy isk assessmen namely Ohslon (1980) in oduced he Logi model in his analysis and ound ha a company’s size, p o i abili y and liquidi y ep esen a nega i e co ela ion wi h he ailu e p obabili y. Also, he ound ha he company’s gea ing is posi i ely co ela ed wi h he ailu e p obabili y. Also, Chen and Shime da (1981) e iewed 26 a icles ha classi i ed 65 i nancial a ios inco po a ed in p edic i e s udies be ween 1966 and 1975, and selec ed 41 i nancial a ios ha we e conside ed o be impo an . Emphasizing he impo ance o using i nancial a ios in he analysis, hey ound ha , by using he i nancial a ios, he accu acy o he p edic ion o a company’s bank up cy exceeds 90%. Anyway, we can classi y he business pe o mance as ollows (Achim & Bo lea, 2014): he accoun ing-based measu e o i nancial pe o mance ( ep esen ed by Re u n on asse s, Re u n on equi y, Le e age a io; Flexibili y and so on) and he ma ke -based measu e o i nancial pe o mance ( ep esen ed by Ma ke capi aliza ion; P ice o book a io; Ma ke o Book a io, P ice Ea nings a io, Di idend Yield a io, Tobin’s Q and so on). E en i he e ec o a iables o he han hose accoun ing-based measu e on he p obabili y o bank up cy is al eady p o en (by mac oeconomic a iables, ma ke -based measu e a iables, co po a e go e nance a iables, e c.), ecen esea ch e ealed ha i nancial a iables a e s ill conside ed signi i can a iables o he company’s pe o mance, p o iding he majo i y in l uence on bank up cy isk p obabili y (Achim & Bo lea, 2012; 2013; Ag a al & Ta l e , 2008; Bea e e al., 2005; Ka as & Režňáko á, 2014; Tu ada a agool, 2013). Mo eo e , he e ec s o non-accoun ing based measu e a iables ul ima ely s ill e l ec in i nancial pe o mances (inc ease o sales, inc ease o ne income, inc ease o weal h and so on). These esul s a e also suppo ed by he i ndings o Bea e e al. (2005) which e l ec ha “ma ke - based a iables a e no a subs i u e o he accoun ing-based in o ma ion bu a he a p oxy o he p edic i e powe a ainable by cap u ing he o al mix o in o ma ion, including bo h he i nancial s a emen and non- i nancial s a emen in o ma ion.” The e o e, he ma ke - based measu e a iables include many o he in l uences o he han e l ec ing he in e nal pe o mance and ul ima ely hey also go- back wi hin he alue o accoun ing based- measu e. So, based on he li e a u e e iew, he accoun ing-based measu e a iables i nally seem o be he bes in e l ec ing he business pe o mances. 1.3 Re iew o Used Me hodologies The de elopmen o p edic i e business ailu e models was he subjec o nume ous esea che s’ s udies. Since 1968, he p ima y me hods ha ha e been used o model de elopmen a e he mul i a ia e disc iminan analyses (MDA), de eloped by Al man (1968; 1970; 2005; 2006) and Bea e (1966; 1968; 2005; 2010). Ohlson (1980) c i icizes he MDA, especially he es ic i e s a is ical equi emen s imposed by he model and in oduces o he i s ime he logis ic eg ession me hod mean o be e p edic he company’s ailu e. The bene i s o he logis ic eg ession me hod a e subsequen ly ecognized by many au ho s, he majo ad an age being ha while he MDA model calcula es a bank up cy sco e using a linea unc ion, he Logi model p edic s ha p obabili y as a ”dicho omous dependen a iable ha is a unc ion o a ec o o explana o y a iables” (Aziz & Da , 2006). In hei la ge su ey, Bello a y e al. (2007) ound ha he mul iple disc iminan analysis is he mos common me hod o p edic ing bank up cy isk, being used in 36% o he in es iga ed s udies. On he second place (wi h 25%) a e he Logi and P obi models. La e , o he me hods suppo ed he de elopmen o bank up cy isk p edic ion models as al e na i es o he wo models men ioned abo e: neu al ne wo ks, haza d model, dis ance o de aul cox eg ession. 1.4 Limi s o Failu e P edic ion Models The in e es in de eloping some bank up cy isk models was mani es ed by esea che s om he whole wo ld; hey would be applied no only o de eloped coun ies bu also o he eme ging coun ies, oo. Fo de eloped coun ies, among he mos impo an bank up cy isk s udies, we can men ion: Bea e (1966), Al man (1968; 2005; 2006), Ohlson (1980), Shumway (2001) ( o US), Ch is idis and G ego y (2010) ( o UK). Fo de eloping coun ies om Cen al and Eas e n Eu ope, we can men ion: Dominiak and Mazu kiewicz (2011) (Poland), Ša lija and Jege (2011) (C oa ia); Dun and B ads ee EM_4_2016.indd 119EM_4_2016.indd 119 6.12.2016 10:50:196.12.2016 10:50:19 120 2016, XIX, 4 Finance (2014), Ka as and Režňáko á (2013; 2014), Pi o a (2011) (Czech Republic); Elenko and File a (2006) (Bulga ia); Sze e in and László (2014) (Hunga y); Ugu lu and Aksoy (2006) (Tu key); Anghel (2002), Siminica (2010), Robu and Mi oniuc (2012) (Romania). Fo de eloping coun ies om Asia we no e he s udies o Wang and Campbell (2010) (China), Sun a uk (2010; 2013) and Tu ada a agool (2013) (Thailand); Lee W-C (2007) and Lee M-C (2014) (Taiwan). Despi e he e o s o globaliza ion and con e gence he economies a ound he wo ld, he e a e s ill la ge dispa i ies be ween he wo ld’s coun ies’ na ional economies ha make impossible he exis ence o a global model o assessing bank up cy isk. A complex su ey conduc ed by Pla and Pla (2008) in h ee global egions namely Asia (including Aus alia), Eu ope and he US highligh s ha in e na ional di e ences in accoun ing ules, lending p ac ices, managemen skills’ le els and legal equi emen s ha e de e mined he ejec ion o he null hypo hesis ha assumed ha a single global model would explain i nancial dis ess on each egion in a ou o a ully elaxed model which c ea ed indi idual i nancial dis ess models o each egion. Some simila s udies we e conduc ed by Lai inen and Su as (2013) on a sample o 30 Eu opean coun ies and hei i ndings highligh ed he signi i can di e ences in he shape and powe o p edic ing bank up cy isk models in l uenced by cha ac e is ics speci i c o each coun y, like: economic en i onmen , company s a us classi i ca ion and coding sys ems, legisla ion and cul u e. On he same le el, Al man and Ho chkiss (2006) also highligh ed he mos impo an di e ences be ween eme ging and de eloping ma ke s, like cu ency ulne abili y, indus y isk, and compe i i e posi ion and ake hese ac o s in o accoun o de elop a bank up cy isk assessmen model speci i cally o eme ging economies unde he Eme ging Ma ke sco e (EMS) Model. A limi ed ca ego y aimed o c ea e he models om he in es o s’ pe spec i e. Thei demands a e di e en om hose o o he decision-make s, he e o e he esul s could no be used o a gene al pu pose, bu o a speci i c one. In his ega d, Wang and Campbell (2010) in hei s udy on Chinese Publicly T aded Companies analysed which i nancial indica o s ha e a signi i can impac on a company’s delis ing s a e om he s ock. They ound ha nega i e own equi y and a nega i e ne income o he las wo yea s a e he wo mos in l uen ial a iables in ailu e p edic ion om he in es o s’ pe spec i e. By using his pe spec i e, Ch is idis and G ego y (2010), in hei su ey on he companies lis ed on he London S ock Exchange, iden i i ed eigh a iables as being signi i can o he companies’ isk ailu e, such as: liquidi y indica o s (wo king capi al o e o al asse s, quick asse s o e cu en asse s), p o i abili y (change in ne income, a dummy a iable equal o one i ne income was nega i e o he las wo yea s, ea nings be o e in e es and ax o e sha e capi al), cash l ow, unds- l ow and le e age (measu ed by o al liabili ies o e o al asse s). Fo he Asian s ock ma ke , mo e speci i cally o Thailand’s s ock ma ke , Tu ada a agool (2013) iden i i ed eigh s a is ically signi i can a ios such as: h ee a ios (cu en asse s o cu en liabili ies, quick asse s o cu en liabili ies and wo king capi al o o al asse s) in he liquidi y g oup; wo a ios (sales o o al asse s and sales o in en o y) in he u no e g oup; wo a ios (ea nings be o e in e es and ax o in e es and o al equi y o o al liabili ies ) in he le e age g oup and one a io (ne income o o al asse s) in he p o i abili y g oup. In es iga ing he abo e esul s, i canno be es ablished ha he a iables would ha e a signi i can impac on s ock ma ke delis ing. Such limi a ion o bank up cy isk models’ applica ion in space and ime is highligh ed by many au ho s such as Balcaen and Ooghe (2006) o Cî ciuma u (2011). Fo ins ance, Cî ciuma u (2011) also ema ked ha e en o he same economy, he pe iods o economic ins abili y a ec he ailu e p edic ions’ esul s and he e o e i is necessa y o egula ly upda e he models in o de o cap u e he new economic and i nancial condi ions. We can add he e ha e en i he bank up cy p edic ion models a e c ea ed speci i cally om he in es o s’ pe spec i e, hey signi i can ly di e om coun y o coun y, om one wo king me hodology o ano he , om he pe iod in which hese models we e c ea ed (i hey we e c ea ed in a pe iod o economic g ow h, he esul s no longe apply o hose om he pe iod o economic c isis) e c. 2. Me hodology 2.1 Sample and Da a In his s udy, he choice o he wo g oups o companies is made om he in es o s’ EM_4_2016.indd 120EM_4_2016.indd 120 6.12.2016 10:50:196.12.2016 10:50:19 121 4, XIX, 2016 Finance poin o iew on he s ock ma ke . As long as a company is delis ed, i becomes wo hless o he in es o because a ading pla o m no longe exis s. E en i he company, being delis ed, will con inue o ope a e o a pe iod o ime, “ he sha eholde s ha e essen ially los hei in es men ” (Wang & Campbell, 2010). Fo ou s udy a i m is iden i i ed as ailed i i is delis ed om he Bucha es S ock Exchange. O he wise, i he i m is lis ed, i is iden i i ed as non- ailed. The s a is ic popula ion consis s in 88 la ge non- i nancial companies ha a e aded on Bucha es S ock Exchange a he end o 2013. Banks and o he i nancial ins i u ions a e excluded om he s udy because hei i nancial s a emen s a e p epa ed on a di e en basis om hose o indus ial i ms (B yan , 1997; Cha i ou, Neophy ou, & Cha alambous, 2004; Flagg, Gi oux, & Wiggins J ., 1991; He & Kama h, 2006; Ohlson, 1980; Tu ada a agool, 2013). Fu he , he non- i nancial companies belong o di e en a ea o business such as: indus y, comme ce, cons uc ion and accommoda ion. F om he sample o 88 non- i nancial companies, we ha e iden i i ed 65 lis ed companies and 21 delis ed companies. The companies lis ing on he Bucha es S ock Exchange is classi i ed on h ee ca ego ies (Tie 1, Tie 2 and Tie 3) acco ding o hei pe o mances, as ollows (Bucha es S ock Exchange, “Issue ’s guide o S ock and bonds”, 2010): Tie 1 includes he companies which ha e an equi y alue a leas o 30 million EURO in he las i nancial yea and ha e ob ained a ne p o i in he las wo yea s o ac i i y. We ha e 16 non- i nancial companies, a he end o 2013. Tie 2 includes he companies which ha e an equi y alue a leas o 2 million EURO in he he las i nancial yea . We ha e 48 non- i nancial companies, a he end o 2013. Tie 3 includes he companies which ha e an equi y alue a leas o 1 million EURO in he he las i nancial yea . We ha e only one companies, a he end o 2013. A special ca ego y o companies aded on he Bucha es S ock Exchange is ep esen ed by delis ed companies. Delis ing means he emo al o a lis ed company om he Bucha es S ock Exchange, olun a ily o in olun a ily, because i is no anymo e in compliance wi h he lis ing equi emen s o he Bucha es S ock Exchange (no longe quali i ed o lis ing, in none o he abo e men ioned ca ego ies). A he end o 2013 he e a e 21 delis ed companies. Based on economic and s a is ical easons, we will wi hhold in he “Lis ed” ca ego y only hose companies ha mee he highes pe o mance c i e ia. Mo e speci i cally, om he o al sample o lis ed companies (65 companies) we e ain only hose lis ed in Tie 1 (namely, 16 companies). The s a is ical easons on which he decision o es ablishing he samples o bank up cy isk’s p edic ion analysis elied, aim he balance be ween he wo samples. Al hough he e is a common de aul e s’ sha e o less han 10% in he en i e da abase, his imbalance can lead o la ge de aul p edic ion e o s. Ma qués, Ga cía and Sánchez (2013) men ioned in hei s udy he p oblem o imbalanced da a and ound ha he use o esampling me hods could consis en ly imp o e he pe o mance gi en by he o iginal imbalanced da a. Many s udies ocused on hese aspec s and used balanced samples in hei isk ailu e s udies: Bea e (1966; 1968) used 79/79, Al man (1968) used 33/33, Deakin (1972) used 32/32, Bei and Liu (2005) used 31/31, Hossa i (2007) used 247/247, Ugu lu and Aksoy (2006) used 27/27, He and Kama (2006) used 20/20, Tu ada a agool (2013) used 14/14. The economic easons o choosing he wo samples a e based on he ac ha he Tie 1 consis s o hose companies ha mee he mos igo ous pe o mance equi emen s and unde hese ci cums ance he likelihood ha i nancial pe o mance a iables o be mo e sensi i e o he wo non- ailu e/ ailu e s a es, is highe . Finally, ou s udy will be conduc ed based on wo samples: a non- ailed sample which consis s in 16 lis ed companies and a ailed companies sample which consis s in 21 delis ed companies. Ou wo samples a e homogeneous, on he one hand, in e ms o he companies’ size (we conside only la ge companies) and, on he o he hand, in e ms o he pe o med ac i i y (we only sample he companies wi h an economic p o i le and he e o e he i nancial ins i u ions a e no included). The da a we e collec ed om he companies’ annual i nancial s a emen s, be ween 2002 and 2012, which a e a ailable on he si e o he Bucha es S ock Exchange (www.b b. o). Fo he delis ed i ms, i nancial da a up o wo yea s EM_4_2016.indd 121EM_4_2016.indd 121 6.12.2016 10:50:196.12.2016 10:50:19 122 2016, XIX, 4 Finance p io o delis ing we e used in ou analysis. Fo he lis ed i ms, we used he i nancial da a om he las 3 yea s, namely be ween 2010 and 2012. The esul was o 90 obse a ions. 2.2 Me hod o Wo k The dependen a iable Dn is a bina y a iable: Dn = 1 o he s ock delis ed companies; Dn = 0 o he s ock lis ed companies. Since he dependen a iable is bina y, he bina y models we e used o iden i ying he delis ing p obabili y (bank up cy isk), which was modelled o depend on eg esso s. The index unc ion o mula ion explains an unobse ed con inuous andom a iable y* (la en a iable), bu all we obse e is he bina y a iable y, which akes he 1 o 0 alue acco ding o whe he y* c osses a h eshold o no . Di e en dis ibu ions o y* lead o di e en bina y ou come models. Le y* be an unobse ed a iable. The eg ession model o y* is he index unc ion model (Came on & T i edy, 2009): y* = X'β + u (1) whe e he eg esso ec o X is a K × 1 column ec o wi h j h en y Xj, he pa ame e ec o β is a K × 1 column ec o wi h j h en y βj, and he e o ec o u is a K × 1 column ec o wi h j h en y uj. Le a ec o o da a deno ed as Xi = (X1i ,..., Xki) om N obse a ions. Then X'β = β1X1 + β2X2 + ... + βkXk . The model (1) canno be es ima ed because y* is no obse ed. We ha e y = { 1 i y* > 0, 0 i y* ≤ 0, (2) whe e he ze o h eshold is a no maliza ion. Gi en he la en - a iable models (1) and (2) we ha e P (y = 1/X) = F(X'β) (3) whe e F(X'β) is he cumula i e dis ibu ion unc ion (c.d. .) o –u. We ob ain he P obi model i u is s anda d no mally dis ibu ed and he Logi model i u is logis ically dis ibu ed. Gi en he model (3), a mo e ele an o m wi h condi ional p obabili y is gi en by pi ≡ P [yi = 1/ X] = F(Xi β) (4) Fo bina y models, he maximum likelihood es ima o (MLE) is na u al es ima o , because he densi y is unambiguously he Be noulli. Fo a sample (yi, Xi), i = 1,...,N, o N independen obse a ions, ML es ima ion, β ^ , maximizes he associa ed log-likelihood unc ion. The β ^ is ob ained by i e a i e me hods and is asymp o ically no mally dis ibu ed. The Wald es and he likelihood- a io (LR) es a e used o p oduce he es s a is ics and p- alue o a es o he signi i cance o indi idual coe i cien s, he con i dence in e als o indi idual coe i cien s, and he es s o o e all signi i cance. In he s a is ics li e a u e a e y common in e p e a ion o he coe i cien s is in e ms o ma ginal e ec s. We a e in e es ing in de e mining he ma ginal e ec o change in a eg esso on he condi ional p obabili y ha y = 1. Fo gene al model (3) and change in he j h eg esso , assumed o be con inuous, his is ∂P (yi = 1/Xi ) = F ,(Xi β)βj. ∂ xij (5) The ma ginal e ec s di e wi h he poin o e alua ion Xi. In measu ing ma ginal e ec s we calcula e he change in he p obabili y P (y = 1) when eg esso s change by one uni . Ma ginal e ec s o logi a iables a e es ima ed. A measu e o goodness o i in he linea eg ession model is R2. The gena aliza ions o nonlinea models a e called pseudo-R2. This measu e is no always compu able bu i is o he bina y ou come model. This yields he R2 measu e o bina y models p oposed by McFadden. We used a good guide (Came oon & T i edy, 2009) ha explains how an econome ics compu e package such as S a a, may pe o m eg ession analysis o a quali a i e bina y a iable. Based on in es iga ing he a o esaid i ndings, a he ini ial s age, we conside ed 12 i nancial a ios. These a iables a e g ouped as ollows: A. P o i abili y 1. Re u n on asse s = Ne esul s / To al asse s (ROA) 2. Re u n on equi y = Ne esul s / Own Equi y (ROE) 3. Ne p o i ma gin = Ne esul s / Sales (NPM) B. Le e age 4. Le e age a io = Deb / Sha eholde ’s equi y (LEV) 5. S abili y a io = Engaged capi al / (Own equi y + Liabili ies) (STAB) EM_4_2016.indd 122EM_4_2016.indd 122 6.12.2016 10:50:196.12.2016 10:50:19 123 4, XIX, 2016 Finance C. Liquidi y and sol ency 6. Cu en liquidi y a io = Cu en asse s/ cu en liabili ies (CL) 7. Flexibili y a io = Ne wo king capi al/ To al asse s (FLEX) 8. Cash- l ow/Ne esul s (CWR) 9. Gene al sol ency = To al asse s/To al deb s (SOLV) D. Ac i i y 10. Asse s u no e = Sales/To al asse s (ASTU) 11. Cu en asse s u no e = Sales/Cu en asse s (CASTU) 12. Cu en deb s u no e = Sales/Cu en deb s (CDTU) The co ela ion esul s be ween he selec ed a iables a e p o ided in he ollowing able: Wi h a ew excep ions, which we will conside , Table 1 shows no s ong co ela ion among he en a iables which a e men ioned abo e. They can be used oge he as po en ially explana o y a iables in any o de o in any combina ion. F om he selec ed a iables, in e ms o he in o ma ion’s impac , h ee models a e conside ed o be ele an o he p obabili y o s ock exchange delis ing. I is well known ha ei he Logi and P obi models can be used o iden i ying he a iables ha a e signi i can and ha e p edic i e powe , because hey ha e simila shapes o cen al alues o F(.) bu di e in he ails as F(.) app oaches 0 o 1. 3. Empi ical Resul s and Discussions Empi ical esul s o Model 1, Model 2 and Model 3 a e p o ided in he ollowing. The McFadden’s R2 ma ginal e ec s and he pe cen age o co ec ly classi i ed obse a ions p o ide he explana o y powe o a iables in models. Model 1 The esul s o he Logi model a e p esen ed in Table 2. All eg esso s, FLEX, ASTU and CASTU a e s a is ically di e en om ze o a he 0.05 le el. The null hypo hesis ha he coe i cien s o FLEX, ASTU and CASTU a e ze o is ejec ed a he 0.05 le el. This is con i med by he LR es . The sign o he coe i cien is also he sign o he ma ginal e ec p esen ed in Table 3. The a iables FLEX and CASTU ha e nega i e coe i cien s which means ha inc easing he l exibili y and also he u no e in cu en asse s conduc o a dec ease o he p obabili y o delis ed P (Dn = 1). Bo h coe i cien s ha e an expec ed signs (nega i es) meaning ha by assu ing an adequa e alue o wo king capi al and a high a io o eplacing he cu en asse s by u no e ac i i y, a well e i ciency in ope a ing ac i i y is pe o med. To he con a y, he a iable ASTU has a posi i e coe i cien , means ha inc easing he u no e in o al asse s conduc o an inc ease o he ASTU SOLV FLEX LEV STAB CL ROE NPM CASTU CDTU CWR ROA ASTU 1.0000 SOLV –0.2375 1.0000 FLEX –0.0834 0.4034 1.0000 LEV 0.0282 –0.1392 –0.0381 1.0000 STAB –0.2812 0.4932 0.8176 0.0171 1.0000 CL –0.1746 0.7191 0.5288 –0.1046 0.3730 1.0000 ROE 0.0636 0.0802 0.1509 –0.8210 0.0110 0.1045 1.0000 NPM 0.2374 0.0376 0.0436 0.0342 0.0454 –0.0726 –0.0291 1.0000 CASTU 0.4965 0.1032 –0.2161 0.0974 –0.0438 –0.3046 –0.1060 0.2146 1.0000 CDTU 0.3071 0.6486 0.4163 0.1043 0.4103 0.5649 0.1111 0.2434 0.3610 1.0000 CWR 0.3433 –0.0825 0.0169 0.0044 –0.0239 –0.0441 –0.0294 0.0280 0.1868 0.0755 1.0000 ROA 0.1310 0.1122 0.4706 –0.0042 0.4088 0.0888 0.1559 0.5755 0.0113 0.3122 –0.0327 1.0000 Sou ce: own calcula ions Tab. 1: Co ela ion ma ix EM_4_2016.indd 123EM_4_2016.indd 123 6.12.2016 10:50:196.12.2016 10:50:19 124 2016, XIX, 4 Finance p obabili y o delis ing. By aking accoun he abo e commen s, i is means ha he small alue o he i xed asse s cause p oblem o disin es men , conduc ing o inc ease o he p obabili y o delis ing. We will commen in de ails all hese esul s a e we un all he conside ed models. Fo he i ed Logi model he McFadden’s R2 is 0.3585. Table 2 shows a good model o he impac o FLEX, ASTU and CASTU on P (Dn = 1). Table 3 p o ides an es ima e o he ma ginal e ec a X = X. Among he h ee s ock exchange delis ing de e minan s, l exibili y a io (FLEX) shows ha i has he highes nega i e ma ginal impac on dependen a iable (Dn = 1) and asse s u no e (ASTU) has a high posi i e ma ginal impac . An impo an poin o no e is ha o he ma ginal o all h ee Logi a iables i.e. FLEX, ASTU and CASTU on Dn, s ock exchange delis ing is s a is ically signi i can wi h a signi i can le el o 0.015, 0.00 and espec i ely 0.003 and hey con ibu e wi h 52.59% o he p obabili y o s ock exchange delis ing. The ma ginal e ec o he model’s p edic i e powe on s ock exchange delis ing is o 0.5259 i.e. 52.59%. One measu e o goodness o i is he pe cen age o co ec ly classi i ed obse a ions. Fo he i ed Logi model, we ob ain Table 4. Table 4 compa es i ed and ac ual alues. The pe cen age o co ec ly speci i ed alues is o 83.33%. In his Table, 5 obse a ions a e misclassi i ed as 1 (s ock exchange delis ed) when he co ec classi i ca ion is 0 (s ock exchange lis ed), and 10 alues a e misclassi i ed as 0 (s ock exchange lis ed) when he co ec alue is 1 (s ock exchange delis ed). In he ollowing, we p esen Model 2 and Model 3 compa ed o Model 1, o show he insigni i can impac o some a iables conside ed a he ini ial s age. Model 2 Asse s u no e (ASTU) ep esen s he de e mining ac o o he p obabili y o s ock exchange delis ing. Table 1 shows no s ong co ela ion among ASTU and CDTU. The esul s o he Logi model wi h ASTU and CDTU Dn Coe . Robus S d. E . z P>|z| [95% Con . In e al] FLEX –3.66313 1.508794 –2.43 0.015 –6.62032 –0.70595 ASTU 4.38998 1.232050 3.56 0.000 1.97521 6.80475 CASTU –1.08176 0.369810 –2.93 0.003 –1.80658 –0.35695 _cons –0.89814 0.491670 –1.83 0.068 –1.86181 0.06552 Numbe o obs. = 90 Wald chi2(3) = 12.93 P ob > chi2 = 0.0048 Log pseudolikelihood = –39.891217 Pseudo R2 = 0.3585 Sou ce: own calcula ions Va iable dy/dx S d. E . z P>|z| [95% Con . In e al] X FLEX –0.91332 0.36985 –2.47 0.014 –1.63822 –0.18842 0.07215 ASTU 1.09454 0.29883 3.66 0.000 0.50885 1.68024 0.84165 CASTU –0.26971 0.09048 –2.98 0.003 –0.44704 –0.09238 2.24492 y = P (Dn) (p edic ) = 0.525912 Sou ce: own calcula ions Tab. 2: Logis ic eg ession Tab. 3: Ma ginal e ec s a e Logi EM_4_2016.indd 124EM_4_2016.indd 124 6.12.2016 10:50:206.12.2016 10:50:20 125 4, XIX, 2016 Finance as he eg esso s a e p esen ed in Table 5. The null hypo hesis ha he coe i cien s o ASTU and CDTU a e ze o is ejec ed a he 0.05 le el. The sign o he coe i cien is also he sign o he ma ginal e ec p esen ed in Table 6. As in he model 1, he coe i cien o a iable ASTU is also posi i e (we commen on his). As o he a iable CDTU, he nega i e coe i cien has an economic signi i cance, meaning ha a highe abili y o a company o pay i s deb s, he smalle is he p obabili y o delis ing. McFadden’ R2 is 0.2960. Table 5 shows a good model o he impac o eg esso s on he delis ing p obabili y P (Dn = 1), bu a smalle han ha om Model 1. The ma ginal e ec s o wo Logi a iables i.e. ASTU and CDTU on Dn, is s a is ically signi i can and hey con ibu e wi h a p obabili y o 51.83 % o s ock exchange delis ing. The pe cen age o co ec ly speci i ed alues (p esen ed in Tab. 7) is wi h 80% smalle han ha om he i s model. The p edic i e powe o he model on he s ock exchange delis ing doesn’ change signi i can ly. Logis ic model o Dn T ue Classi i ed D ~D To al + 32537 –104353 To al 42 48 90 Co ec ly classi i ed 83.33% Sou ce: own calcula ions Dn Coe . Robus S d. E . z P>|z| [95% Con . In e al] ASTU 3.44703 1.02342 3.37 0.001 1.44115 5.45291 CDTU –0.36271 0.15072 –2.41 0.016 –0.65812 –0.06731 _cons –1.53414 0.47610 –3.22 0.001 –2.46729 –0.60099 Numbe o obs. = 90 Wald chi2(2) = 11.83 P ob > chi2 = 0.0027 Log pseudolikelihood = –43.774824 Pseudo R2 = 0.2960 Sou ce: own calcula ions Va iable dy/dx S d. E . z P>|z| [95% Con . In e al] X ASTU 0.86060 0.25191 3.42 0.001 0.366874 1.35433 0.84161 CDTU –0.09055 0.03737 –2.42 0.015 –0.163803 –0.01731 3.56632 y = P (Dn) (p edic ) = 0.518330 Sou ce: own calcula ions Tab. 4: The pe cen age o co ec ly classi i ed obse a ions Tab. 5: Logis ic eg ession Tab. 6: Ma ginal e ec s a e logi EM_4_2016.indd 125EM_4_2016.indd 125 6.12.2016 10:50:206.12.2016 10:50:20 132 2016, XIX, 4 Finance Risk Models Based on Financial S a emen In o ma ion: Compa isons ac oss Eu opean Coun ies. Jou nal o Finance and Economics, 1(3), 1-26. doi:10.12735/j e. 1i3p01. Ma qués, A., Ga cía, V., & Sánchez, J. S. (2013). On he sui abili y o esampling echniques o he class imbalance p oblem in c edi sco ing. Jou nal o he Ope a ional Resea ch Socie y, 64(7), 1060-1070. doi:10.1057/jo s.2012.120. Ohlson, J. A. (1980). Financial Ra ios and he P obabilis ic P edic ion o Bank up cy. Jou nal o Accoun ing Resea ch, 18(1), 109- 131. doi:10.2307/2490395. Pla , H. D., & Pla , M. B. (2008). Financial dis ess compa ison ac oss h ee global egions. Jou nal o Risk and Financial Managemen , 1(1), 129-162. doi:10.3390/j m1010129. Pi o á, K. (2011). Possibili ies o he Al man Ze a model applica ion o Czech i ms. E&M Ekonomie a Managemen , 14(3), 66-76. Robu, M., Mi oniuc, M., & Robu, I. B. (2012). A p ac ical model o es ing he hypo hesis o “going-conce n” in he i nancial audi mission o Romanian lis ed companies. Re is a Audi Financia , 2, 13-23. Ša lija, N., & Jege , M. (2011). Compa ing i nancial dis ess p edic ion models be o e and du ing ecession. C oa ian Ope a ional Resea ch Re iew, 2(1), 133-142. Siminica, M. (2010). Financial diagnosis. Publishing House SITECH. Shumway, T. (2001), Fo ecas ing Bank up cy Mo e Accu a ely: A Simple Haza d Model. Jou nal o Business, 74(1), 101-124. doi:10.1086/209665. Sze e in, E. K., & Laszlo, K. (2014). The E i ciency o Bank up cy Fo ecas Models in he Hunga ian SME Sec o . Jou nal o Compe i i eness, 6(2), 56-73. doi:10.7441/ joc.2014.02.05. Todea, A., & Laza , D. (2012). Global C isis and Rela i e E i ciency: empi ical e idence om Cen al and Eas e n Eu opean S ock Ma ke s. The Re iew o Finance and Banking, 4(1), 45-52. Tu ada a agool, S. (2013). The ole o i nancial a ios in signalling i nancial dis ess: e idence om Thai lis ed companies (Doc o al disse a ion). Lismo e: Sou he n C oss Uni e si y (h p://epubs.scu.edu.au/cgi/ iewcon en .cgi?a icle=1291&con ex = heses). Ugu lu, M., & Aksoy, H. (2006). P edic ion o co po a e i nancial dis ess in an eme ging ma ke : he case o Tu key. C oss Cul u al Managemen : An In e na ional Jou nal, 13(4), 277-295. doi:10.1108/13527600610713396. Wang, Y., & Campbell, M. (2010). Financial a ios and he p edic ion o bank up cy: The Ohlson model applied o Chinese publicly aded companies. The Jou nal o O ganiza ional Leade ship and Business, 17(1), 334-338. Assoc. P o . Phd. habil. Monica Viole a Achim Babeş-Bolyai Uni e si y Facul y o Economics and Business Adminis a ion [email p o ec ed] [email p o ec ed] Assoc. P o . Phd. So in Nicolae Bo lea Phd. Wes Vasile Goldis Uni e si y Facul y o Economics, In o ma ics and Enginee ing [email p o ec ed] Economis Phd. Lucian Vasile Găban “1 Decemb ie 1918” Uni e si y Facul y o Economics Alba Iulia [email p o ec ed] EM_4_2016.indd 132EM_4_2016.indd 132 6.12.2016 10:50:216.12.2016 10:50:21 133 4, XIX, 2016 Finance Abs ac FAILURE PREDICTION FROM THE INVESTORS’ VIEW BY USING FINANCIAL RATIOS. LESSON FROM ROMANIA Monica Viole a Achim, So in Nicolae Bo lea, Lucian Vasile Găban The pu pose o ou s udy is o iden i y which i nancial indica o s ha e a signi i can impac on he p obabili y o Romanian companies’ bank up cy isk om he in es o s’ poin o iew by s udying he impac on he p obabili y o sha es delis ing om he s ock exchange. The esea ch is conduc ed on a sample o 16 ailed and 21 non- ailed non- i nancial companies lis ed on he Bucha es S ock Exchange be ween 2002 and 2012. The Logi analysis is used o iden i ying he a iables ha a e signi i can and ha e p edic i e powe on dis ess likelihood. By using 12 main i nancial a ios, we es ima e h ee al e na i e Logi models o de e mining hei signs, signi i cance, p edic i e powe , e i ciency o i es s. The i s model p o ides he highes explana o y powe . Th ee a iables such as Flexibili y a io (FLEX), Asse s u no e (ASTU) and Cu en asse s u no e (CASTU) a e ound o be signi i can de e minan s o s ock exchange delis ing. These h ee a iables p o ide 52.59% o co ec p edic ion o bank up cy isk. The pe cen age o co ec ly classi i ed obse a ions o he i ed Logi model is o 83.33%. Mo eo e , his esea ch a emp s o e eal he changes ha may appea among bank up cy p edic o s gi en ha he bank up cy isk model is de eloped om he in es o s’ poin o iew and no om ha o a simple decision-making pe son. Fo a s ock ma ke in es o , bank up cy al eady s a s a he s age o delis ing he company because he in es men was s ongly comp omised, whe he o i con inues i s ac i i y o no . O ien a ion owa ds in es o s when p edic ing bank up cy isk is he main elemen o o iginali y ha ou esea ch adds o he scien i i c achie emen s in bank up cy, un il his momen . Key Wo ds: Delis ing, ailu e p edic ion, i nancial a ios, Logi model. JEL Classi i ca ion: C25, C52, G33. DOI: 10.15240/ ul/001/2016-4-009 EM_4_2016.indd 133EM_4_2016.indd 133 6.12.2016 10:50:216.12.2016 10:50:21