Determinants of working capital management

DETERMINANTS OF WORKING CAPITAL MANAGEMENT Sonia Baños-Caballe o, [email p o ec ed], Uni e sidad de Mu cia Ped o J. Ga cía Te uel. pj e [email protected], Uni e sidad de Mu cia Ped o Ma ínez Solano, pm[email p o ec ed], Uni e sidad de Mu cia ABSTRACT This pape analyzes he de e minan s o wo king capi al managemen (WCM) o a sample o Spanish i ms du ing he pe iod 1997-2004. We ind ha i ms ha e a a ge in es men in wo king capi al and ha hey ake decisions in o de o achie e his. The esul s appea o suppo he hypo hesis ha he wo king capi al compe es wi h in es men in ixed asse s o he unds o he i ms when hey ha e inancial cons ain s. Finally, we also ind ha WCM depends on ba gaining powe and o he inancial ac o s such as he a ailabili y o in e nal inance, cos o inancing and inancial cons ain s. KEYWORDS: Wo king capi al, Ne T ade Cycle, dynamic panel da a, ma ke impe ec ions, endogenei y. Acknowledgemen s: This esea ch is pa o he P ojec ECO2008-06179/ECON inanced by he Resea ch Agency o he Spanish go e nmen . The au ho s also acknowledge suppo om Fundación CajaMu cia. 1. In oduc ion Since he seminal wo k by Modigliani and Mille (1958) showing ha a i m´s inancial s uc u e is i ele an o in es men , he li e a u e on in es men decisions has been enla ged by many heo e ical and empi ical con ibu ions. I has shown ha in he p esence o ma ke impe ec ions, i ms may p e e one sou ce o unds o e ano he because o a wide cos -wedge be ween in e nal and ex e nal unding sou ces. Mye s and Majlu (1984) show ha i ms p esen a p e e ence o in e nal o e ex e nal unds, and in he case o ex e nal unds, a company p e e s deb be o e equi ies. In ac , Fazza i, Hubba d and Pe e sen, (1988) es he inancing hie a chy hypo hesis and sugges ha he i ms´ in es men may depend on inancial ac o s such as he a ailabili y o in e nal inance, access o capi al ma ke s o cos o inancing. The in es men ha i ms make in cu en asse s, as well as he cu en liabili ies used, ep esen s he main sha e o i ems on a i m´s balance shee . In ac , in ou sample hey ep esen mo e han 50 pe cen o he i ms´ o al asse s and o al liabili ies. Decisions abou how much o in es in ecei able accoun s and in en o ies, and how much c edi o accep om supplie s, a e e lec ed in he managemen o i ms’ wo king capi al. This may ha e an impo an impac on he p o i abili y and liquidi y o he i m (Shin and Soenen, 1998), so i ms ha e o e alua e he ade-o be ween expec ed p o i abili y and liquidi y isk be o e deciding he wo king capi al policy o adop . Despi e he impo ance o wo king capi al managemen o he p o i abili y o he i ms (Smi h, 1980; Soenen, 1993; Jose, Lancas e and S e ens, 1996; Shin and Soenen, 1998; Deloo , 2003; and Ga cia and Ma inez, 2007), he e has been li le wo k on he empi ical de e minan s o he wo king capi al managemen . Chiou, Cheng and Wu (2006) analyze i m cha ac e is ics and mac oeconomic ac o s ha migh a ec wo king capi al managemen o companies om Taiwan. Kieschnick, Laplan e and Moussawi (2006) also s udy he de e minan s o WCM, in his case, o a sample o US companies. Howe e , hese s udies ha e se e al limi a ions. Fi s , hey do no con ol o endogenei y. Second, none o hese exis ing empi ical s udies has shown he possible exis ence o a a ge le el o he measu es o wo king capi al managemen , e en hough i s exis ence appea s o be e iden . Se e al p e ious s udies (Nadi i, 1969; Eme y, 1984: Blinde , 1986, among o he s) ha e shown ha i ms ha e a ge o op imal le els o hei indi idual componen s such as accoun s ecei able, in en o ies and accoun s payable. Mo eo e , unde impe ec capi al ma ke s, i ms ha e o e alua e he ade-o be ween cos s and bene i s o main aining a la ge in es men in wo king capi al. In his pape we a emp o ex end he empi ical esea ch on wo king capi al managemen in a numbe o ways. Fi s , in con as o he exis ing empi ical s udies, his pape adop s a dynamic amewo k, which assumes ha i ms ha e a a ge in es men in wo king capi al and ha hey adjus hei cu en in es men g adually o e ime because o adjus men cos s. Second, we use panel da a because i allows us o con ol o unobse able he e ogenei y, making i possible o exclude biases de i ing om he exis ence o indi idual e ec s. Thi d, we use wo-s ep GMM es ima o o a oid he p oblem o possible endogenei y. Fo example, se e al s udies ha e shown how he measu es o he wo king capi al managemen a ec p o i abili y and i ms´ sales. Finally, we p esen empi ical e idence o a sample o Spanish i ms in he con ex o he con inen al model, which is cha ac e ised by less-de eloped capi al ma ke s (La Po a, Lopez de Silanes, Shlei e and Vishny, 1997,1998), low in es o p o ec ion and high concen a ion o owne ship. I allows ou esul s o be compa ed wi h o he s ob ained o companies wi h di e en inancial sys ems. Following Shin and Soenen (1998), we use he Ne T ade Cycle (NTC) as a measu e o wo king capi al managemen . NTC indica es he numbe o “days sales” he i m has o inance i s wo king capi al unde ce e is pa ibus condi ions. Ou indings indica e ha i ms ha e a a ge Ne T ade Cycle and hey adjus hei cu en Ne T ade Cycle o hei a ge g adually o e ime because o adjus men cos s. Mo eo e , he speed o adjus men is ela i ely quick, which may be because being in disequilib ium is cos ly o his sample. On he o he hand, he esul s epo ed in his s udy sugges ha i ms ha a e capable o gene a ing mo e in e nal unds ha e a longe cycle. The esul s also lend suppo o Fazza i and Pe e sen´s (1993) a gumen ha wo king capi al compe es wi h he in es men in ixed asse s o he unds o he i ms when hey su e inancial cons ain s. The hypo hesis ha companies wi h a g ea e ba gaining powe ollow a mo e agg essi e wo king capi al policy is also suppo ed by he indings o his pape . Finally, ou esul s sugges ha Ne T ade Cycle also depends on o he inancial ac o s such as he cos o inancing and inancial cons ain s. Thus, we ob ain ha g ow h oppo uni ies, p obabili y o inancial dis ess and cos o ex e nal inancing nega i ely a ec NTC. Howe e , we do no ind suppo o he hypo hesis ha le e age in luences he measu es o wo king capi al managemen . The emainde o his pape is o ganized as ollows. The nex sec ion de elops he hypo heses and e iews he p e ious s udies on wo king capi al managemen . In sec ion 3 we desc ibe he empi ical model and da a. We p esen ou esul s in sec ion 4 and ela e hem o ea lie indings. Finally, he main conclusions a e p esen ed in Sec ion 6. 2. Theo e ical amewo k and hypo heses. In pe ec capi al ma ke s, in es men decisions o a i m a e independen o i s inancial si ua ion (Modigliani and Mille , 1958). Since he e is no capi al a ioning, i ms can always ob ain ex e nal inancing wi hou p oblem, so he i ms´ in es men should be d i en only by expec ed u u e p o i abili y and, he e o e, should no be a ec ed by he a ailabili y o in e nal unds. Howe e , in impe ec ma ke s, he i ms´ in es men may depend on inancial ac o s such as he a ailabili y o in e nal inance, access o capi al ma ke s o cos o inancing (Fazza i, Hubba d and Pe e sen, 1988). Unde his si ua ion, he wo king capi al le el held by companies may also be sensi i e o hese inancial ac o s. The Ne T ade Cycle (NTC) and Cash Con e sion Cycle (CCC) a e he mos popula measu es o wo king capi al managemen used in p e ious wo ks, due o he c i icism o s a ic measu es (Gi man, 1974; Kama h, 1989). Bo h o hese measu es a e a dynamic measu e o ongoing liquidi y managemen , and hey a e closely co ela ed. The CCC is calcula ed as (accoun s ecei ables/sales)*365 + (in en o ies/cos o sales)*365 - (accoun s payable/pu chases)*365 and shows he ime lag be ween expendi u e o he pu chase o aw ma e ials and he collec ion o sales o inished goods (Deloo , 2003). So, he longe his ime lag, he la ge he in es men in wo king capi al. The Ne T ade Cycle (NTC) is basically equal o he CCC, bu he h ee componen s (accoun s ecei able, in en o ies and accoun s payable) a e exp essed as a pe cen age o sales, so indica ing he numbe o “days sales” he i m has o inance i s wo king capi al (Shin and Soenen, 1998). Acco ding o hese au ho s, his las measu e p o ides an easy es ima e o addi ional inancing needs wi h ega d o wo king capi al exp essed as a unc ion o he p ojec ed sales g ow h. They suppo ed he use o he NTC o measu e wo king capi al managemen e iciency, since all h ee componen s a e exp essed as a pe cen age o sales which is mo e use ul. Inc easing hese cycles may posi i ely a ec i ms´ p o i abili y o wo easons. Fi s , i may inc ease i ms´ sales (Blinde and Maccini,1991; Smi h, 1987; Eme y, 1987; Deloo and Jege s, 1996; Pe e sen and Rajan, 1997; and Ng, Smi h and Smi h, 1999). These wo ks show ha i ms´ sales inc ease when hey inc ease hei in es men in in en o ies o ade c edi g an ed. Second, Ng e al., (1999) and Wilne (2000) also demons a e ha i ms may ge impo an discoun s o ea ly paymen s which educe hei supplie inancing. Howe e , his bene i has o o se he cos s o a la ge in es men in wo king capi al when i ms ope a e unde impe ec capi al ma ke s. Fi s , i ms ha e a inancing cos . Second, he main cos o holding a highe wo king capi al le el is he oppo uni y cos , because a i m may o go o he mo e p oduc i e in es men s in o de o hold ha le el. Finally, and acco ding o Soenen (1993), longe cycles migh also lead companies o bank up cy. Hence, unde impe ec capi al ma ke s, companies may ha e an op imal Ne T ade Cycle ha balances he cos s and bene i s o main aining i and which maximizes hei alue. In addi ion, since a longe cycle indica es a need o addi ional capi al, i may depend on agency cos s, asymme ic in o ma ion and inancial dis ess, because hese lead o a highe cos o inancing ex e nal and c edi a ioning. Asymme ic In o ma ion and Agency con lic s Asymme ic in o ma ion and agency cos s could lead o ei he unde in es men o o e in es men . On he one hand, gi en he limi ed liabili y o sha eholde s, hey migh ca y ou iskie in es men p ojec s (p oblem o o e in es men ), because sha eholde s would bene i om he i m’s highe alue, while c edi o s would su e he possible losses (Jensen and Meckling, 1976). On he o he hand, he con lic be ween sha eholde s and c edi o s, acco ding o Mye s (1977), can also lead o a p oblem o unde in es men , because gi en he p io i y o c edi o s in case o bank up cy, sha eholde s may decide no o ca y ou o o abandon in es men p ojec s wi h a posi i e ne p esen alue when he ne p esen alue o he in es men is less han he amoun o deb issued. Consequen ly, i ms ha e o pay a isk p emium, which esul s in a highe cos o ex e nal sou ces o unds. In his sense, he pecking o de heo y o Mye s (1984) s a es ha i ms gi e p io i y o esou ces gene a ed in e nally o e deb and new equi y. Following his line, S igli z and Weiss (1981) and Mye s and Majlu (1984) sugges ed ha a sho age o in e nally gene a ed unds may lead o i m unde in es men . This idea has been suppo ed by se e al ea lie s udies ha ha e demons a ed ha he amoun o co po a e in es men is a ec ed by i s in e nal inancing (Fazza i e al., 1988; Ca pen e , 1995; Kadapakkam, Kuma and Riddick, 1998; Hoshi, Kashyap and Scha s ein., 1991; Hadlock, 1998; Clea y, 1999; Moyen, 2004). The empi ical e idence sugges s i migh be due, as we commen ed abo e, o he highe isk p emium demanded because o asymme ic in o ma ion and he agency con lic s. G eenwald e al., (1984), on he o he hand, sugges ha asymme ic in o ma ion may also esul in c edi a ioning in compe i i e ma ke s, which migh also a ec he le el o i ms´ in es men . Hence, and aking in o accoun hese hypo hesis, in e nal unds should also posi i ely in luence he i ms´ wo king capi al in es men , as is demons a ed by Fazza i and Pe e sen (1993). G ow h oppo uni ies is ano he a iable ha migh a ec wo king capi al managemen , because i ms wi h highe g ow h pe spec i es ha e mo e se e e agency con lic s be ween c edi o s and sha eholde s (Mye s, 1977) and la ge asymme ic in o ma ion due o hei alue being la gely de e mined by hese g ow h pe spec i es (Mye s and Majlu , 1984). Thus, we would expec hese companies o ha e a sho e cycle. The esul s ob ained by he empi ical e idence, howe e , lead o opposi e conclusions abou he e ec s o his a iable. Kieschnick e al., (2006) show ha his a iable posi i ely in luences Cash Con e sion Cycle o US i ms, while Chiou e al., (2006) do no ind any ela ion be ween hese a iables. These p oblems, on he o he hand, a e also sensi i e o le e age, acco ding o he exis ing li e a u e. The esul s ob ained by K ishnawami e al., (1999) appea o indica e ha agency con lic s be ween sha eholde s and c edi o s dec ease also wi h p i a e deb , which p edomina es in Spain. In his line, o he ex en ha deb a io ac s as a p oxy o he abili y o he i ms o ob ain deb i would be expec ed ha i ms wi h highe le e age (g ea e abili y o aise deb ) will hold mo e in es men in wo king capi al. Fi m size is ano he ac o in luencing hese agency cos s. The agency p oblem be ween sha eholde s and c edi o s is expec ed o be a enua ed by size (Smi h and Wa ne , 1979), since smalle i ms su e mo e se e e asymme ic in o ma ion be ween inside s and ou side s (Jo dan e al., 1998; and Be ge e al., 2001), due o he ac ha less public in o ma ion is a ailable o hem. Thus, his ac o would be expec ed o posi i ely in luence he leng h o his cycle. Howe e , we should also men ion ha his a iable has also been associa ed in he li e a u e wi h he i m´s ba gaining powe wi h i s supplie s and cus ome s, showing ha la ge i ms ha e a g ea e ba gaining powe , so hey migh ha e a sho e NTC. Acco ding o Long e al. (1993), Lee and S owe (1993), and Pike, Cheng, C a ens and Lamminmaki (2005), smalle i ms ha e o ex end mo e c edi o gua an ee hei p oduc s, gi en hei lowe epu a ions. In addi ion, hey a e o e ed less ade c edi (Niskanen and Niskanen, 2006). Wi h ega d o he esul s ob ained om he empi ical e idence, hey lead o con a y conclusions abou he e ec o his a iable on he measu es o wo king capi al managemen (Moss and S ine (1993); Jose e al., (1996); Kieschnick e al., (2006) and Chiou e al., (2006). Finally, we also expec in es men in ixed asse s o in luence Ne T ade Cycle. The p oblems o asymme ic in o ma ion and agency con lic be ween sha eholde s and c edi o s a e expec ed o be mo e se ious o companies wi h lowe ixed in es men and g ea e in angible asse s, because hese la e asse s canno be easily alued by po en ial ex e nal in es o s and hese i ms would ha e a lowe liquida ion alue o hei asse s. Thus, sha eholde s migh decide o ca y ou g ea e isk p ojec s in hese i ms i he liquida ion alue o hei asse s is lowe han he deb alue. Mo eo e , asse angibili y inc eases he alue ha can be ecap u ed by c edi o s in bank up cy. Thus, we should expec a posi i e ela ion be ween his a iable and he Ne T ade Cycle. Howe e , when i ms ope a e unde impe ec capi al ma ke s, hey bea inancial cons ain s, so his a iable migh compe e wi h he wo king capi al o i m´s capi al in his si ua ion, as is epo ed by Fazza i and Pe e sen (1993) and Kieschnick e al. (2006). Financial dis ess We also in oduce he a iable p obabili y o inancial dis ess because he agency con lic s commen ed abo e a e mo e p onounced o inancially dis ess i ms. The cos s o inancial dis ess a ise when he i m canno mee i s paymen obliga ions ei he in he sho o he long e m. This can a ec he Ne T ade Cycle o i ms, since companies wi h a g ea e p obabili y o inancial dis ess ha e mo e di icul ies in ob aining capi al. Gi en ha a longe cycle indica es a need o addi ional capi al, hese i ms migh ha e a sho e NTC. 3. Me hod and Da a 3.1. Me hod Following he heo ies desc ibed in he p e ious sec ion and conside ing he cos s and bene i s o keeping wo king capi al, we assume ha i ms ha e a a ge Ne T ade Cycle. Fi ms´cu en Ne T ade Cycle may no always equal hei desi ed cycle and, hence, i ms migh ake ime o adjus om ac ual o he desi ed cycle. This can be o se e al easons. Nadi i (1969), o ins ance, sugges s ha i ms canno always es ima e hei sales accu a ely and wi h ce ain y, and hence nei he hei pu chases; hey do no accu a ely an icipa e changes in mone a y policy o in he a es o de aul and bad deb s on hei ade c edi ; and he disco e y and collec ion o delinquen accoun s ake ime and in ol e cos s which may be dis ibu ed o e ime. Like Shin and Soenen (1998), we use he Ne T ade Cycle as dependen a iable, which is calcula ed by he ollowing exp ession: NTC= (accoun s ecei ables/ sales)*365 + (in en o ies/sales)*365 - (accoun s payable/sales)*365. Thus, i p o ides an easy es ima e o addi ional inancing needs wi h ega d o wo king capi al exp essed as a unc ion o he p ojec ed sales g ow h. Wi h ega d o he independen a iables, he capaci y o gene a e in e nal unds is p oxied by he a iable cash low (CFLOW), de ined as he a io o ea nings be o e in e es and ax plus dep ecia ion o sales. We use wo p oxies o measu e he g ow h oppo uni ies. GROWTH1 is calcula ed by he a io ma ke - o- book alue o asse s ((ma ke alue o equi y + ma ke alue o deb ) / o al asse s), while GROWTH2 is de ined as he a io ma ke - o-book alue o equi y (ma ke alue o equi y / book alue o equi y). The a io o o al deb o e o al asse s (LEV) is used as p oxy o he le e age. We use he na u al loga i hm o asse s (SIZE1) and he na u al loga i hm o sales (SIZE2) o measu e he size. The in es men in ixed asse s (FA) o he i m is measu ed by he a io angible ixed asse s o e o al asse s. The likelihood o inancial dis ess (ZSCORE) is calcula ed acco ding o he e-es ima ion o Al man´s (1968) model ca ied ou by Be gley e al., (1996), gi en by he ollowing exp ession: ZSCOREi = 0,104*X1 + 1,010*X2 + 0,106*X3 + 0,003*X4 + 0,169*X5 whe e X1 = Wo king capi al / To al asse s; X2 =Re ained ea nings / To al asse s; X3 = Ne ope a ing p o i s /To al asse s; X4 = Ma ke alue o capi al / Book alue o deb ; X5 = Sales / To al asse s. Thus, a highe ZSCORE implies a lowe p obabili y o insol ency. Finally, aking in o accoun he p e ious s udies on he de e minan s o wo king capi al managemen and he heo ies desc ibed in sec ion 2, we also in oduce he a iables cos o ex e nal inance (FCOST) and p o i abili y (PRO) as independen a iables. We expec i ms wi h a highe cos o ex e nal inance o hold a smalle NTC, since he cos o unds in es ed in i is highe . This cos is also measu ed by wo p oxies. The i s (FCOST1) is calcula ed by he a io inancial cos s/( o al deb - accoun s payable). In he second one (FCOST2) we do no elimina e accoun s payable o he o al deb . On he o he hand, he empi ical e idence demons a es ha e u n also nega i ely a ec s measu es o wo king capi al managemen . The a ios ea nings be o e in e es and axes o e o al asse (PRO1) and ea nings be o e in e es and axes o e sales (PRO2) a e used in ou analysis as p oxies o his a iable. Thus, Ne T ade Cycle can be explained by he ollowing a iables: NTC*= (in e nal esou ces, g ow h op ions, le e age, size, ixed asse s, p obabili y o inancial dis ess, cos o ex e nal inance, p o i abili y) Fi ms may be e y di e en om each o he and he e a e o he cha ac e is ics ha migh in luence hei Ne T ade Cycle ha a e di icul o measu e o ha d o ob ain, and which a e no in ou model. The e o e, we use panel da a, because i allows us con ol o unobse able he e ogenei y, making i possible o exclude biases de i ing om he exis ence o indi idual e ec s (Hsiao, 1985). Addi ionally, we can also include ime e ec s o cap u e he in luence o economic ac o s ha may also a ec he leng h o Ne T ade Cycle. Finally, i allows us examine a pa ial adjus men model o con i m whe he i ms pu sue a a ge Ne T ade Cycle. I he e is a a ge Ne T ade Cycle, i ms should ake he app op ia e s eps o achie e i . Howe e , adjus men is no immedia e because i ms ha e o bea cos s o adjus men , so hey will adjus hei cu en NTC acco ding o he ollowing exp ession:  10) 1,, *( 1,, < < − − = − − γ γ i NTC i NTC i NTC i NTC  (1) whe e NTCi, is he Ne T ade Cycle in he pe iod , and NTC*i, is he a ge Ne T ade Cycle, which is es ima ed om he ollowing equa ion: i i i i i i i i i i PROFCOSTZSCOREFA SIZELEVGROWTHCFLOWNTC ,,8,7,6,5 ,4,3,2,10, * εββββ β β β β β +++++ + + ++= (2) whe e i, ε is a andom dis u bance and k β a e he unknown pa ame e s o be es ima ed. The exp ession )*( 1,, − − i i NTCNTC is he adjus men equi ed o each he i m’s a ge NTC, and he coe icien  γ measu es he speed o adjus men , which is in e sely ela ed o adjus men cos s, and akes alues be ween 0 and 1. I 0= γ , hen 1,, − = i i NTCNTC , and he cu en Ne T ade Cycle emains as in he p e ious pe iod, indica ing ha companies bea high adjus men cos s. I , in con as , 1= γ , hen i i NTCNTC ,, * = , and i ms immedia ely adjus hei Ne T ade Cycle o hei a ge . I we subs i u e equa ion (2) in o equa ion (1) and we include he unobse able he e ogenei y and he ime dummy a iables, he cu en NTC is de e mined by he ollowing exp ession: i i i i i i i i i i i i PROFCOSTZSCOREFA SIZELEVGROWTHCFLOWNTCNTC ,,8,7,6,5 ,4,3,2,11,, υληδδδδ δ δ δ δ ρ α +++++++ + + + ++= −(3) whe e i ikk and ,,0 ;);1(; γε υ γβ δ γ ρ γβ α = =− = = Pa ame e i ηcap u es o he cha ac e is ics o i ms which a e no obse able bu which ha e a signi ican impac on he leng h o he Ne T ade Cycle. These change ac oss companies bu emain ixed o a gi en i m h ough ime. The a iable λ ,on hecon a y, is a ime dummy ha changes in ime bu is equal o all i ms in each o he ime pe iods conside ed. This pa ame e is designed o cap u e he in luence o economic ac o s ha also may a ec he leng h o his Cycle, bu ha i ms canno con ol. Finally, he pa ame e s i, υ a e andom dis u bances. We use he ins umen al a iable es ima ion me hod o a oid he p oblem o endogenei y, which appea s o be e iden in ou analysis, as se e al s udies ha e shown. Fo example, he wo king capi al managemen migh a ec p o i abili y (Jose e al., 1996; Shin and Soenen, 1998; Deloo , 2003; and Ga cia and Ma inez, 2007) and i ms´ sales (Blinde and Maccini, 1991; Smi h, 1987; Eme y, 1987; Deloo and Jege s, 1996; Pe e sen and Rajan, 1997; and Ng e al., 1999). I we do no con ol o endogenei y, i migh se iously a ec he es ima ion esul s. A ellano and Bond (1991) p oposed he applica ion o he Gene al Me hod o Momen (GMM) o he equa ion in i s di e ences. We use wo- s ep GMM es ima o because, al hough he es ima o o ins umen al a iables in one s age is always consis en , i he dis u bances show he e oskedas ici y, he es ima ion in wo s ages inc eases e iciency. 3.2. Da a The da a o his analysis we e ob ained om h ee sou ces o in o ma ion. Fi s , da a om inancial s a emen s ha e been aken om he SABI (Ibe ian Balance Shee s Analysis Sys em) da abase, which was de eloped by Bu eau Van Dijk. Second, he ma ke alue o equi y was ex ac ed om CNMV (Spanish Secu i y Exchange Commission). Finally, G oss Domes ic p oduc da a we e collec ed om he Bank o Spain. Ou da a consis s o non- inancial Spanish i ms lis ed on he Spain S ock Exchange o he pe iod 1997- 2004. We ha e selec ed i ms whose in o ma ion is a ailable o a leas i e consecu i e yea s be ween 1997 and 2004. F om his, we ob ained a panel comp ising 502 obse a ions co esponding o 60 i ms. Table 1 Summa y o S a is ics Mean S d. De Min Median Max NTC 115.19 96.506 -29.73 91.46 590.91 CFLOW 0.1687 0.1279 -0.053 0.1303 0.7371 GROWTH1 1.3836 0.7360 0.5758 1.1650 5.5831 GROWTH2 2.074 2.2875 0.1546 1.4696 2.0257 LEV 0.5833 0.1619 0.078 0.5986 0.9521 ASSETS 4,276,179 11,700,000 14,882 403,551 91,800,000 SALES 2,447,147 6,076,944 3,471 362,130.5 44,000,000 AF 0.5059 0.2172 0.0711 0.4967 0.9872 ZSCORE 0.3035 0.1575 0.0179 0.2899 0.7285 FCOST1 0.0593 0.0411 0.0048 0.050 0.3772 FCOST2 0.0411 0.02767 0.0025 0.0363 0.2206 PRO1 0.0706 0.0509 -0.1222 0.0633 0.3181 PRO2 0.1094 0.1070 -0.1443 0.0861 0.6975 GDP 0.0382 0.0079 0.024 0.036 0.05 No es: NTC ep esen s he Ne T ade Cycle; CFLOW he cash lows gene a ed by he i m; GROWTH 1 and GROWTH2 he g ow h oppo uni ies; LEV he le e age; ASSETS he o al asse s in housands o eu os; SALES he sales in housands o eu os; AF he in es men in ixed asse s; ZSCORE he p obabili y o inancial dis ess; FCOST1 and FCOST2 he cos o ex e nal inance; PRO1 and PRO2 he p o i abili y; and GDP he G oss Domes ic P oduc g ow h. Table 1 summa izes he desc ip i e s a is ics o ou sample. I we obse e he mean alues, we can see ha he mean Ne T ade Cycle in ou sample is 115.19 days. These i ms ha e had posi i e g ow h pe spec i es du ing he selec ed esea ch pe iod and an in es men in ixed asse s o e he o al asse s o 51.6%. They gene a e a cash low o 16.9% o e sales and p esen 58.3% o deb o e o al asse s. Thei p o i abili y o e asse s and sales a e 7% and 10.9%, espec i ely. 4. Resul s Table 2 shows he esul s om eg essing Ne T ade Cycle on he di e en a iables explained abo e. To con i m he obus ness o ou esul s we p esen he es ima ion o equa ion (3) using al e na i e p oxies o some independen a iables. In addi ion, in column (6) we es ima e he model including he G oss Domes ic P oduc g ow h and elimina ing he ime dummies o a oid he mul icollinea i y p oblem, since hese dummies should cap u e ha in o ma ion. The m2 s a is ic and Hansen es also a e p esen ed. The m2 s a is ic indica es he e is no second-o de se ial co ela ion, and Hansen Tes shows he absence o co ela ion be ween ins umen s and e o e m. We also p esen he Va iance In la ion Fac o (VIF) o each independen a iable. Ou VIF es s a e lowe han 5, so he e is no mul icollina i y p oblem in ou sample (S udenmund, 1997). In all es ima ions we con ol o indus y e ec s. Table 2 De e minan s o Ne T ade Cycle (1) (2) (3) (4) (5) (6) VIF NTCi -1 0.3705*** (26.95) 0.3794*** (26.33) 0.3510*** (29.22) 0.4173*** (34.07) 0.4050*** (33.40) 0.4003*** (42.20) 1.58 CFLOW 237.44*** (4.82) 181.31*** (3.04) 175.26*** (3.29) 206.61*** (3.25) 332.85*** (2.99) 207.81*** (5.03) 2.86 GROWTH1 -12.7981*** (-3.75) - -15.4135*** (-5.58) -12.0809*** (-2.83) -20.99*** (-5.64) -16.7198*** (-5.76) 1.23 GROWTH2 - -4.6588*** (-2.77) - - - - LEV 2.1479 (0.07) 27.9474 (0.80) 9.1483 (0.25) 34.7871 (1.03) 46.4511 (1.30) 9.1026 (0.47) 2.12 SIZE1 -8.7338*** (-3.00) -10.1825*** (-2.59) - -7.7511*** (-3.00) -10.5788*** (-3.81) -8.9595*** (-4.96) 2.29 SIZE2 - - -13.5618*** (-4.83) - - - - AF -101.90*** (-4.25) -93.845*** (-3.34) -77.041*** (-2.99) -51.963* (-1.78) -73.2563*** (-3.00) -96.6829*** (-5.31) 3.24 ZSCORE 184.28*** (6.44) 207.09*** (6.80) 181.01*** (6.01) 150.06*** (4.85) 110.17*** (4.25) 127.6166*** (8.02) 2.80 FCOST1 -345.13*** (-5.91) -354.43*** (-5.31) -322.34*** (-6.52) - -317.89*** (-5.21) -272.628*** (-9.47) 1.07 FCOST2 - - - -270.48*** (-3.60) - - PRO1 -362.98*** (-4.15) -354.04*** (-3.61) -263.50** (-2.53) -226.87** (-2.59) - -287.14*** (-3.89) 2.21 PRO2 - - - -249.05** (-2.30) - GDP - - - - - 438.66*** (4.58) m2 -0.58 -0.52 -0.59 -0.49 -0.51 -0.51 Hansen Tes 36.15 (221) 42 (221) 41.08 (221) 38.47 (221) 40.24 (221) 43.75 (223) Obse a ions 442 442 442 442 442 442 442 No es: NTC ep esen s he Ne T ade Cycle; CFLOW he cash lows gene a ed by he i m; GROWTH 1 and GROWTH2 he g ow h oppo uni ies; LEV he le e age; SIZE1 and SIZE2 he size; AF he in es men in ixed asse s; ZSCORE he p obabili y o inancial dis ess; FCOST1 and FCOST2 he cos o ex e nal inance; PRO1 and PRO2 he p o i abili y; and GDP he G oss Domes ic P oduc g ow h. Z s a is ic in b acke s. * Indica es signi icance a 10% le el, ** indica es signi icance a 5% le el, *** indica es signi icance a 1%. le el m2 is a se ial co ela ion es o second-o de using esiduals o i s di e ences, asymp o ically dis ibu ed as N(0,1) unde null hypo hesis o no se ial co ela ion. Hansen es is a es o o e -iden i ying es ic ions dis ibu ed asymp o ically unde null hypo hesis o alidi y o ins umen s as Chi-squa ed. Deg ees o eedom in b acke s. VIF ep esen s he Va iance In la ion Fac o o each independen a iable. The esul s show ha he coe icien o he lagged Ne T ade Cycle is posi i e and signi ican a he 1% le el in all he es ima ions made. Hence, ou hypo hesis on he dynamic na u e o wo king capi al managemen is no ejec ed and we con ibu e o he exis ing li e a u e on wo king capi al managemen , inding ha companies ha e a a ge Ne T ade Cycle and ollow an adjus men p ocess o each his a ge . In addi ion, his coe icien is oughly 0.4 in all he es ima ions made, indica ing a ela i ely quick