How does working capial managemen affec SMEs profiabiliy? Absrac This paper analyzes he relaion beween working capial managemen and profiabiliy for small and medium-sized firms by conrolling for unobservable heerogeneiy and possible endogeneiy. Unlike previous sudies, we examine a non-linear relaion beween hese wo variables. Our resuls show ha here is a non monoonic (concave) relaionship beween working capial level and firm profiabiliy, which indicae ha SMEs have an opimal working capial level ha maximizes heir profiabiliy. Keywords: Cash Conversion Cycle; working capial; profiabiliy; SMEs. JEL classificaion: G30, G31, G32. 1. Inroducion The idea ha working capial managemen affecs he firm s profiabiliy and risk is generally acceped and has recenly received considerable aenions. Firs, Smih (1980) suggess ha working capial managemen is imporan because of is effecs on a firm s profiabiliy and risk, and consequenly is value. Specifically, a more aggressive working capial policy (low invesmen in working capial) is associaed wih a higher 1
reurn and risk, while a conservaive working capial policy (high invesmen in working capial) supposes a lower reurn and risk. These effecs on profiabiliy and risk, herefore, sugges ha firms migh have an opimal working capial level ha balances he coss and benefis of holding working capial and maximizes heir profiabiliy. Indeed, Deloof (2003) suggess ha firms migh have an opimal level of working capial ha maximizes heir value. The lieraure has demonsraed ha here exis opimal levels of is individual componens, such as accouns receivable (Emery 1984a; Nadiri 1969), invenories (Ouyang, Teng, Chuang, and Chuang 2005) and accouns payable (Nadiri 1969). However, previous research on working capial managemen and firm performance (Jose, Lancaser and Sevens 1996; Shin and Soenen 1998; Wang 2002; Deloof 2003; and Garcia-Teruel and Marinez-Solano 2007; among ohers), analyzes a linear relaionship beween invesmen in working capial and firm s profiabiliy. The findings indicae ha he lower he invesmen in working capial he more profiabiliy, ignoring, for insance, he higher risk of loss of sales and inerrupions in he producion process, relaed wih low levels of working capial. The relaion beween working capial and firm s profiabiliy migh, herefore, be concave raher han linear and hese opposing effecs migh be capured wih a quadraic relaionship. This paper analyzes he relaionship beween invesmen in working capial and profiabiliy aking ino accoun he possible non-lineariies of he working capial managemen-profiabiliy relaionship in order o es his risk and reurn rade-off beween differen working capial sraegies. In addiion, as a robusness check of our resuls, we also employ a differen approach, moivaed by Tong s (2008) mehodology in order o es his possible quadraic relaion. We use a sample of small and medium sized Spanish firms o do his for several reasons. Firsly, mos of he previous sudies 2
have basically focused heir analyses on large firms. Secondly, SMEs have imporan financial consrains (Whied 1992; Fazzari and Peersen 1993; and Audresch and Elson 1997) and difficulies in obaining funding in he long-erm capial markes (Walker 1989; Peersen and Rajan 1997; and Scholens 1999), so an efficien working capial managemen is paricularly imporan for hem (Peel and Wilson 1996; Peel, Wilson and Howorh 2000). In his line, Grablowsky (1984) and Kargar and Blumenhal (1994) sugges ha working capial managemen may be crucial for he survival and growh of small companies. Thirdly, he ineres in sudying Spanish firms sems from he fac ha hey operae in a banking oriened financial sysem where capial markes are less developed and banks play an imporan role (Schmid and Tyrell 1997), so our resuls may also be of ineres for oher SMEs esablished in counries wih similar financial sysems, as is he case of mos of he European Counries. Hence, hese firms have fewer alernaives for obaining exernal financing, which makes hem more dependen on rade credi. Indeed, Demirguc-Kun and Maksimovic (2002) sugges ha hese firms gran more rade credi o heir cusomers, and a he same ime receive more finance from heir own suppliers. Finally, he esimaion mehod was seleced in order o avoid unobservable heerogeneiy and possible endogeneiy, because if we do no conrol for hese problems, i migh seriously affec he esimaion resuls. Specifically, we use panel daa and he Generalized Mehod of Momens (GMM) o esimae our models. The resuls obained confirm our hypohesis ha here is an invered U-shaped relaionship beween working capial and profiabiliy. The relaion beween hese variables is posiive when firms hold low levels of working capial, and becomes negaive for higher levels of working capial. I allows us o confirm no only he greaer 3
profiabiliy effec, bu also he greaer risk effec for firms wih low levels of working capial. The remainder of his paper is organized as follows. Secion 2 oulines heoreical links beween working capial policy and profiabiliy. Secion 3 describes he model employed o analyze he relaionship beween working capial and firm s performance and he hypoheses o be esed. In Secion 4, we describe he mehodology and daa used. The resuls are discussed in Secion 5. A robusness check is presened in Secion 6. Finally, Secion 7 concludes he paper. 2. Working capial policy and profiabiliy. As noed in he inroducion, he way in which a firm manages is working capial can have a significan impac on boh he risk (risk of loss of business and inerrupions of producion process) and profiabiliy. Specifically, working capial managemen pracices ha end o enhance he profiabiliy end o increase his risk, and conversely hose pracices ha reduce he risk end o decrease he expeced performance. On he one hand, since an addiional invesmen in invenories or accouns receivable is usually associaed wih greaer sales, a posiive relaion beween working capial and profiabiliy migh be expeced. Larger invenories can preven inerrupions in he producion process and loss of business due o scarciy of producs, and hey can reduce supply coss and price flucuaions (Blinder and Maccini 1991). In addiion, hey allow firms beer service for heir cusomers and avoid high producion coss arising from high flucuaions in producion (Schiff and Lieber 1974). Graning rade credi also simulaes sales because i allows buyers o verify produc and services qualiy prior o paymen (Smih 1987; Long, Maliz and Ravid 1993; and Lee and Sowe 1993) and, 4
hence, reduce he informaion asymmeric beween buyer and seller. In addiion, rade credi is an imporan supplier selecion crierion when i is hard o differeniae producs (Shipley and Davis 1991; and Deloof and Jegers 1996), i is used as an effecive price cu (Brennan, Maksimovic, and Zechner 1988; Peersen and Rajan 1997), encourages cusomers o acquire merchandise a imes of low demand (Emery 1987), reduces ransacion coss (Ferris 1981; and Emery 1987) and srenghens long-erm supplier-cusomer relaionships (Ng, Smih, and Smih 1999; Wilner 2000), among oher advanages. Thus, a high invesmen in working capial migh increase a firm s performance. However, his addiional invesmen in working capial may also adversely affec is operaing performance if he coss of a higher invesmen in working capial exceed he benefis of holding more invenories and/or graning more rade credi o cusomers. Deloof (2003), for insance, suggess ha firms migh no assess he qualiy of he producs bough before paying if hey reduce heir received rade credi period. Moreover, according o Soenen (1993), high invesmens in working capial migh also lead companies o bankrupcy and heir suppliers could cu off he supply of he regularly purchased merchandise (Cuña 2007) or, in he case of non-paymen, i could be recovered and sold o anoher cusomer. From he poin of view of invenories, keeping sock available also supposes coss such as warehouse ren, insurance and securiy expenses, which end o rise as he level of invenory increases. Finally, he finance lieraure has demonsraed ha an increase of invesmen in curren asses would increase oal asses wihou a proporional increase in profiabiliy. As a resul of he coss and benefis of a higher invesmen in working capial, here may exis an invered U-shaped relaionship beween he firm s profiabiliy and invesmen in working capial and, hence, firms migh have an opimal working capial level ha 5
balances coss and benefis and maximizes heir profiabiliy. Specifically, we expec firms profiabiliy o rise as working capial increases, unil a cerain working capial level is reached, given ha he increased profiabiliy will no offse he high risk borne. Conversely, beyond his opimum, due o he low reurn of curren asses, we expec ha increases in working capial are relaed wih decreases in he profiabiliy. Tha is, we expec firm profiabiliy and working capial o relae posiively a low levels of working capial and negaively a higher levels. The empirical evidence, however, is no consisen wih he rade-off beween profiabiliy and risk hypohesis commened above (Jose, Lancaser and Sevens 1996; Shin and Soenen 1998; Wang 2002; Deloof 2003; and Garcia-Teruel and Marinez- Solano 2007; among ohers). These sudies have analyzed a linear relaionship beween working capial and profiabiliy and heir resuls sugges ha firms can increase heir performance by reducing heir working capial levels. However, hose findings ignore he risk of loss of sales and inerrupions in he producion process relaed wih low levels of working capial, which also migh be capured wih a non-linear relaion. 3. Model and hypoheses This secion describes he model employed for esing he main hypohesis menioned in he previous secion, ha is, ha here exiss a concave relaionship beween he firm s operaing profiabiliy and invesmen in working capial. This would allow us o confirm ha firms have an opimal working capial level a which heir profiabiliy is maximized. We use he Cash Conversion Cycle (CCC) as measure of working capial managemen because i has been he measure mos used by previous sudies, due o he criicism of 6
saic measures such as curren raio and quick raio (Emery 1984b; Soenen 1993). This variable is calculaed as (accouns receivables/sales)*365 + (invenories/purchases)*365 - (accouns payable/purchases)*365. Thus, CCC concerns he managemen of receivables accouns, he managemen of invenories and he rade credi received, wih a shorer CCC meaning a more aggressive working capial policy. Previous lieraure indicaes he imporance of considering hese hree componens a he same ime because hey influence one anoher and firm s profiabiliy and value. Schiff and Lieber (1974), for insance, indicae he imporance of aking ino accoun he inerrelaionship beween invenory and accouns receivable policies. To validae our hypohesis, we regress he firm s operaing profiabiliy agains cash conversion cycle and is square. The inclusion of hese wo variables allows us o es boh he profiabiliy and risk effecs commened above. In addiion, following Deloof (2003) and Garcia-Teruel and Marinez-Solano (2007) among ohers, we conrol for firm size, growh opporuniies and leverage. Thus, our profiabiliy model is as follows: PRO 2 = β 0 + β 1CCC + β 2CCC + β 3 SIZE + β 4GROWTH + β 5 LEV + λ + η i + ε (1) where PRO is he profiabiliy of firm i a ime ; CCC is he Cash Conversion Cycle of firm i a ime, and CCC 2 is square. SIZE is he size of he firms, GROWTH he growh opporuniies, and LEV he leverage. The parameer λ is a ime dummy variable, η i is he unobservable heerogeneiy or he firm s unobservable individual effecs, and ε is he random disurbance. Like Deloof (2003), we use wo proxies o measure ha profiabiliy. PRO 1 is calculaed by he gross operaing income ((sales cos of sales)/oal asses); while PRO 2 represens he ne operaing income ((sales cos of sales - depreciaion & amorisaion)/oal asses). We use hese variables because hey migh reflec he operaing aciviies of he firm beer han he overall reurn on asses 7
and we relae hem o cash conversion cycle, which is oher operaing variable. The size (SIZE) is measured as he naural logarihm of asses; growh opporuniies (GROWTH) by he raio (sales 1 -sales 0 )/sales 0 ; and leverage (LEV) as he raio of deb o oal asses. The parameer λ is a ime dummy variable ha changes in ime bu is equal for all firms in each of he ime periods considered. This parameer is designed o capure he influence of economic facors ha may also affec firm s profiabiliy, bu which companies canno conrol. Finally, capures he paricular characerisics of each firm. η i is he firm s unobservable heerogeneiy and The quadraic relaion proposed in equaion (1) presens one breakpoin, which can be derived by differeniaing he firm profiabiliy variable wih respec o he CCC variable and making his derivaive equal o 0. On solving for he CCC variable, we obain ha his breakpoin is CCC = ( -β 1 / 2β 2 ). To verify our main hypohesis, his should be a maximum, since his would indicae ha here is an invered U-shaped relaionship beween working capial and profiabiliy and, hence, firms have an opimal working capial level a which hey maximize heir operaing performance. Since his will be a maximum only if he second parial derivae of he profiabiliy wih respec o he Cash Conversion Cycle variable ( 2β 2) is negaive, β 2 should be negaive. 4. Mehodology and Daa Mehodology The esimaion mehod was seleced in order o avoid unobservable heerogeneiy and he problem of possible endogeneiy. Firms are heerogeneous and here are always characerisics ha migh influence heir profiabiliy ha are difficul o measure or 8
hard o obain, and which are no in our model. Hence, we use panel daa in order o eliminae he risk of obaining biased resuls because of his heerogeneiy (Hsiao 1985). We eliminae he individual effec by aking firs differences. Moreover, we use he insrumenal variable esimaion mehod o avoid he problem of endogeneiy, which migh be presen in our analysis. We use he wo-sep generalized mehod of momens (GMM) esimaor proposed by Arellano and Bond (1991) because, alhough he esimaor of insrumenal variables in one sage is always consisen, if he disurbances show heeroskedasiciy, he esimaion in wo sages increases efficiency. Finally, we should menion ha we also conrol for indusry effecs by inroducing eigh indusry dummies. Daa This sudy uilises a daa panel of non-financial Spanish SMEs. The daa were obained from SABI (Iberian Balance Shees Analysis Sysem) daabase, developed by Bureau Van Dijk. This daabase conains accouning and financial informaion for Spanish firms. The sample comprises small and medium-sized firms from Spain for he period 2002-2007. The selecion of SMEs was carried ou according o he requiremens esablished by he European Commission recommendaion 2003/361/EC of 6 May, 2003, i.e. hey had fewer han 250 employees, urned over less han 50 million euros and possessed less han 43 million euros worh of oal asses. Finally, we eliminaed firms whose informaion is no available for a leas five consecuive years, firms wih los values, cases wih errors in he accouning daa, and exreme values presened by all variables. We obained an unbalanced panel of 1008 Spanish SMEs (5862 firm-year observaions). 9
In Table 1, we repor he mean values of Cash Conversion Cycle by secor and year. In addiion, in he final column we also presen a -saisic on he difference of means o deermine if he mean lengh of CCC held by firms in 2002 differs significanly from hose held in 2007. We conduced his es under he null hypohesis of equal means. Since saisic akes he value 3.09, he null hypohesis is rejeced and, hence, i indicaes ha Spanish SMEs have increased heir invesmen in working capial during his period. Table 2 provides summary saisics of he variables used in he esimaions. Finally, in Table 3 we presen he correlaions of he variables used in our model. In addiion, we also used a formal es o ensure ha he mulicollineariy problem is no presen in our analyses. Thus, we calculaed he Variance Inflaion Facor (VIF) for each independen variable included in our models. The larges VIF value is 1.50, so here is no mulicollinariy problem in our sample because i is far from 5 (Sudenmund 1997). 5. Resuls The resuls obained from model (1) are presened in Table 4. They indicae ha β 1 is posiive and β 2 is negaive, and boh coefficiens are significan. Thus, hey confirm ha firms have an opimal Cash Conversion Cycle ha balances coss and benefis and maximizes operaing performance. In addiion, i indicaes, unlike previous sudies, ha profiabiliy increases wih he invesmen in working capial a low levels of working capial, and decreases a high levels. Thus, our resuls show he imporance of also aking ino accoun he risk of loss of business and inerrupions in he producion process in he working capial managemen-profiabiliy relaion hrough a non-linear relaionship. Given ha his opimal cycle can be calculaed by he expression CCC = 10
(-β 1 / 2β 2 ), our findings indicae ha his is 111.17 days and 89.19 days for he gross operaing income and he ne operaing income, respecively. Hence, he operaing performance rises when firms increase heir CCC up o his opimum, bu once his opimum is reached, a marginal increase in working capial will negaively affec firm s performance. 6. Robusness check The model developed in secion 3 is he mos common empirical approach in esing he quadraic relaion beween wo variables. The resuls obained indicae ha here is an invered U-shaped relaionship beween invesmen in working capial and profiabiliy, ha is, firms have an opimal working capial level ha maximizes heir profiabiliy and, hence, heir profiabiliy should decrease when hey move away from his opimal level. Our main goal in his secion is o give robusness o he resuls obained from he firs model by sudying he relaion beween deviaions on boh sides of opimal working capial level and firm profiabiliy. If here exiss an opimum, boh below-opimal and above-opimal deviaions from his should reduce firm profiabiliy. To do i, we use a wo-sage mehodology moivaed by Tong s (2008) sudy, which allows us o verify he exisence of a concave relaion beween working capial and firm profiabiliy. In a firs sage we obain deviaions from opimal CCC, while in he second sage we regress firm profiabiliy agains hose deviaions. If our hypohesis is verified, ha is, deviaions negaively affec profiabiliy, hen his would allow us o give robusness o he resuls obained in he firs model. 11
Sage 1: Following Baños-Caballero e al. (2009), we use equaion (2) as he benchmark regression for he deerminans of Cash Conversion Cycle lengh in SMEs: CCC * = δ 0 + δ CFLOW 1 + δ LEV 2 + δ GROWTH 3 + δ SIZE 4 + δ AGE 5 + δ FA 6 + δ ROA 7 + ε (2) where CCC* represens he opimal Cash Conversion Cycle of firm i a ime ; CFLOW cash flow; LEV he leverage; GROWTH growh opporuniies; SIZE he size; AGE he age; FA invesmen in fixed asses; ROA reurn on asses; and random disurbance. We calculae he CCC as (accouns receivables/sales)*365 + (invenories/purchases)*365 - (accouns payable/purchases)*365; CFLOW is he raio of ne profi plus depreciaion o oal asses; LEV he raio of deb o oal asses; GROWTH he raio (sales 1 -sales 0 )/sales 0 ; SIZE he naural logarihm of asses; AGE he naural logarihm of age; FA he raio (Tangible fixed asses/oal asses); and ROA he raio Earnings Before Ineres and Taxes over oal asses. ε Firms curren Cash Conversion Cycle, however, may no always equal heir desired cycle for several reasons. Firsly, Nadiri (1969) suggess ha firms canno always esimae heir sales accuraely and wih cerainy, and, hence, neiher heir purchases; hey do no accuraely anicipae changes in he opporuniy cos of rade credi or in he raes of defaul and bad debs on heir rade credi; he discovery and collecion of delinquen accouns ake ime and involve coss which may be disribued over ime; and finally disequilibrium in oher asses of he firms, such as invenories, may also reflec ha discrepancy. In his line, Saroris and Hill (1983) indicae ha when firms change heir credi policy hey can also have sources of uncerainy such as he fracion of sales paid wih a discoun, iming of paymens, volume of sales, and he fracions of 12
sales ha are never paid by cusomers. Secondly, he difficulies firms have in order o access capial markes or heir low bargaining power wih cusomers and suppliers migh lead firms o inves below or above heir opimal working capial levels, respecively. Finally, he conflics of ineress beween he main sakeholders (shareholders, managers and crediors) could also give rise o curren working capial level no being equal o desired level. Based on his idea ha firms curren CCC migh no always equal heir opimum, as in Tong (2008), we obain he residuals from he regression (2) and we use hem as a proxy for he deviaions from opimal Cash Conversion Cycle. Thus, once we have idenified he deviaions from opimal cycle in Sage 1, hen in Sage 2 we analyze how hese deviaions affec he firm s profiabiliy. Sage 2: Following Tong (2008), since he residuals can be eiher posiive or negaive, we define he variable Deviaion as he absolue value of he residuals obained from equaion (2), so his measures he deviaions from opimal CCC. Moreover, o es our hypohesis, we also define a dummy variable, AOD, which is equal o 1 for posiive residuals and 0 oherwise. Thus, AOD is equal o 1 if acual CCC is greaer han opimal CCC, and is equal o 0 if oherwise. We hen allowed his dummy o inerac wih he Deviaion variable. In order o es he effec of deviaions from opimum we use he following profiabiliy equaions: PRO = α + α Deviaion + α SIZE + α GROWTH + α LEV + λ + η + ε 0 1 2 3 4 i (3) PRO = γ + γ Deviaion 0 1 + γ 2 5 ( Deviaion* AOD) + γ LEV + λ + η + ε i + γ SIZE 3 + γ GROWTH 4 (4) 13
All dependen and independen variables are he same as hose specified in equaion (1). We have only eliminaed he CCC variable and is square, and we have inroduced he Deviaion variable and he ineracion erm. Therefore, in equaion (3), he sign of α1 indicaes he effec of he deviaions from opimum on firm performance, so we expec ha α1 0, because his would indicae ha he firm s profiabiliy decreases when a firm moves away from is opimal CCC. In equaion (4), γ 1 and ( γ 1+ γ 2 ) represen he influence of below-opimal deviaions (i.e. when AOD akes he value 0) and above-opimal deviaions (i.e. when AOD akes he value 1), respecively, on he firm s profiabiliy. We expec ha γ 1 0 and ( γ 1 + γ 2 ) 0, since his would indicae ha boh below-opimal and above-opimal deviaions reduce he firm s profiabiliy and, hence, ha he firm s operaing performance will increase unil a cerain working capial level is reached, and afer ha opimum he performance will sar o decrease. Thus, firm managers should aim a keeping as close o he opimal cycle as possible and ry o avoid any deviaion (eiher posiive or negaive). The resuls, which are presened in Table 5, are consisen wih hose obained in Secion 5. We find ha he firm s profiabiliy decreases when a firm moves away from is opimal CCC, since he coefficien of he Deviaion variable ( α 1 ) is negaive and significan in equaions (3a) and (3b). In equaions (4a) and (4b), as we commened above, γ 1 indicaes he effec of below-opimal deviaions on operaing performance, while ( γ 1+ γ 2 ) represens he influence of above-opimal deviaions on his performance. We obain ha γ 1 is negaive and significan in boh equaions. Wih regard o he coefficien ( γ 1+ γ 2 ), we obain ha i is also negaive and significan in boh equaion (4a) and equaion (4b). Therefore, he resuls show, as we expeced, ha boh below-opimal and above-opimal deviaions reduce he firm s profiabiliy and 14
firm managers should aim a keeping as close o he opimal cycle as possible, and ry o avoid any deviaion (eiher posiive or negaive) from i. In addiion, he difference in he negaive impacs on firm profiabiliy is no saisically significan beween aboveopimal and below-opimal deviaions, since he coefficien of he ineracion erm ( γ 2 ) is no significan. 7. Conclusions This sudy offers new evidence on he relaionship beween working capial managemen and profiabiliy by conrolling for unobservable heerogeneiy and possible endogeneiy and, unlike previous sudies, given he compeing hypoheses of he effec of an increase in working capial on firm s profiabiliy, we analyze a possible quadraic relaion beween hese variables. In conras o previous findings, which indicae ha he lower invesmen in working capial he more profiabiliy, our resuls show ha here is a concave relaionship beween working capial level and profiabiliy, ha is, companies have an opimal working capial level ha balances coss and benefis and maximizes heir profiabiliy. I allows us o confirm no only he greaer profiabiliy effec, bu also he greaer risk effec for firms wih low levels of working capial. In addiion, our resuls show ha he firms profiabiliy decreases when hey move away from heir opimal working capial. Overall, his paper highlighs he imporance of a good working capial managemen for firms due o he cos of over-invesmen and under-invesmen in working capial 15
Acknowledgemens: This research is par of he Projec ECO2008-06179/ECON financed by he Research Agency of he Spanish governmen. We acknowledge financial suppor from Fundación Séneca Science and Technology Agency of he Region of Murcia (Spain)- (Program: PCRTRM 07-10). Research projec 08822/PHCS/08. The auhors also acknowledge financial suppor from Fundación CajaMurcia. REFERENCES Audresch, D. B., & Elson, J. A. (1997). Financing he German mielsand. Small Business Economics, 9, 97-110. Arellano, M., & Bond, S. (1991). Some Tes of Specificaion for Panel Daa: Mone Carlo Evidence and An Applicaion o Employmen Equaions. Review of Economics Sudies, 58, 277-297. Baños, S., Garcia, P. J., & Marinez, P. (2009). Working Capial Managemen in SMEs. Accouning and Finance, Early view. Blinder, A. S., & Maccin L. J. (1991). The Resurgence of Invenory Research: Wha Have We Learned?. Journal of Economic Survey, 5, 291-328. Brennan, M., Maksimovic, V., & Zechner, J. (1988). Vendor financing. Journal of Finance, 43, 1127-1141. Cuña, V. (2007). Trade Credi: Suppliers as Deb Collecors and Insurance Providers. Review of Financial Sudies, 20, 491-527. Deloof, M. (2003). Does Working Capial Managemen Affec Profiabiliy of Belgian Firms?. Journal of Business, Finance and Accouning, 30, 573-587. Deloof, M., & Jegers, M. (1996). Trade Credi, Produc Qualiy, and Inragroup Trade: Some European Evidence. Financial Managemen, 25, 33-43. 16
Demigurc-Kun, A., & Maksimovic, V. (2002). Firms as Financial Inermediaries: Evidence from Trade Credi Daa. World Bank Working Paper. Emery, G. W. (1984a). A pure financial explanaion for rade credi. Journal of Financial and Quaniaive Analysis, 9, 271-285. Emery, G. W. (1984b). Measuring shor-erm liquidiy. Journal of Cash Managemen, 4, 25-32. Emery, G. W. (1987). An Opimal Financial Response o Variable Demand. Journal of Financial and Quaniaive Analysis, 22, 209-225. Fazzar S. M., & Peersen, B. (1993). Working Capial and Fixed Invesmen: New Evidence on Financing Consrains. Rand Journal of Economics, 24, 328-342. Ferris, J. S. (1981). A ransacions heory of rade credi use. Quarerly Journal of Economics, 94, 243-270. Garcia-Teruel, P. J., & Marinez-Solano, P. (2007). Effecs of Working Capial Managemen on SME Profiabiliy. Inernaional Journal of Managerial Finance, 3, 164-177. Grablowsky, B. J. (1984). Financial Managemen of Invenory. Journal of Small Business Managemen, 22, 59-65. Hsiao, C. (1985). Benefis and Limiaions of Panel Daa. Economerics Review, 4, 121-174. Jose, M. L., Lancaser, C., & Sevens, J.L. (1996). Corporae Reurn and Cash Conversion Cycle. Journal of Economics and Finance, 20, 33-46. Kargar, J., & Blumenal, R. A. (1994). Leverage impac on Working Capial in Small Business. TMA Journal, 14, 46-53. 17
Lee, Y. W., & Sowe, J. D. (1993). Produc risk, asymmeric informaion, and rade credi. Journal of Financial and Quaniaive Analysis, 28, 285-300. Long, M. S., Maliz, I. B., & Ravid, S. A. (1993). Trade credi, qualiy guaranees, and produc markeabiliy. Financial Managemen, 22, 117-127. Nadir M. I. (1969). The deerminans of Trade Credi in he U.S. Toal Manufacuring Secor. Economerica, 37, 408-423. Ng, C. K., Smih, J. K., & Smih, R. L. (1999). Evidence on he Deerminans of Credi Terms Used in Inerfirm Trade. Journal of Finance, 54, 1109-1129. Ouyang, L. Y., Teng, J. T., Chuang, K. W., & Chuang, B. R. (2005). Opimal invenory policy wih noninsananeous receip under rade credi. Inernaional Journal of Producion Economics, 98, 290-300. Peel, M., & Wilson, N. (1996). Working Capial and Financial managemen pracices in he small firm secor. Inernaional Small Business Journal, 14, 52-68. Peel, M. J., Wilson, N., & Howorh, C. (2000). Lae Paymen and Credi Managemen in he Small Firm Secor: Some Empirical Evidence. Inernaional Small Business Journal, 18, 17-37. Peersen, M., & Rajan, R. (1997). Trade Credi: Theories and Evidence. Review of Financial Sudies, 10, 661-691. Saroris, W., & Hill, N. (1983). Cash and working capial managemen. The Journal of Finance, 38, 349-360. Schiff, M., & Lieber, Z. (1974). A model for he inegraion of credi and invenory managemen. The Journal of Finance, 29, 133-140. 18
Schmid, R., & Tyrell, M. (1997). Financial Sysems, Corporae Finance and Corporae Governance. European Financial Managemen, 3, 333-361. Scholens, B. (1999). Analyical issues in exernal financing alernaives for SBEs. Small Business Economics, 12, 137-148. Shin, H. H., & Soenen, L. (1998). Efficiency of Working Capial and Corporae Profiabiliy. Financial Pracice and Educaion, 8, 37-45. Shipley, D., & Davis, L. (1991). The role and burden-allocaion of credi in disribuion channels. Journal of Markeing Channels, 1, 3-22. Smih, K. (1980). Profiabiliy versus liquidiy radeoffs in working capial managemen. In K. V. Smih (Ed.), Readings on he Managemen of Working Capial (pp. 549-562). S Paul, MN: Wes Publishing Company. Smih, J. K. (1987). Trade Credi and Informaional Asymmery. Journal of Finance, 42, 863-872. Soenen, L. (1993). Cash Conversion Cycle and corporae profiabiliy. Journal of Cash Managemen, 13 (4), 53-57. Sudenmund, A. H. (1997). Using Economerics: A Pracical Guide. New York, Addison-Wesley. Tong, Z. (2008). Deviaions from opimal CEO ownership and firm value. Journal of Banking and Finance, 32, 2462-2470. Walker, D. (1989). Financing he small firm. Small Business Economics, 1, 285-296. Wang, Y. J. (2002). Liquidiy managemen, operaing performance, and corporae value: evidence from Japan and Taiwan. Journal of Mulinaional Financial Managemen, 12, 159-169. 19
Whied, T. M. (1992). Deb, Liquidiy Consrains, and Corporae Invesmen: Evidence From Panel Daa. Journal of Finance, 47, 1425-1460. Wilner, B. S. (2000). The exploiaion of relaionship in financial disress: he case of rade credi. Journal of Finance, 55, 153-178. 20
Table 1 Cash Conversion Cycle by year and secor Indusry 2002 2003 2004 2005 2006 2007 2002-2007 Agriculure and 42.65 20.57 33.46 40.23 63.45 68.56 44.84 0.5060 Mining Manufacuring 86.87 85.96 89.05 96.65 99.98 100.83 93.28 2.1586 Consrucion 38.11 37.41 35.05 40.85 47.53 65.31 44.01 2.0418 Wholesale rade 94.48 93.25 94.11 98.69 101.38 105.95 98.01 1.4721 Reail rade 64.54 58.65 66.08 72.10 71.77 78.95 68.70 1.4039 Services -54.10-35.64-36.51-34.44-21.93-43.70-37.08 0.4009 Transpor -2.78-20.51-12.84-9.94-7.42-1.29-9.29 0.0777 Toal 70.57 69.15 71.71 77.72 81.63 84.84 75.97 3.0915 This able shows he mean Cash Conversion Cycle by year and secor. The Cash Conversion Cycle is calculaed as (accouns receivables/sales)*365 + (invenories/purchases)*365 - (accouns payable/purchases)*365. is he -saisic in order o es wheher mean lengh of CCC held by firms in 2002 differs significanly from ha held in 2007, under he null hypohesis of equal means. 21
Table 2 Summary saisics Mean Sandard Perc. 10 Median Perc. 90 deviaion PRO 1 0.5020 0.2207 0.2493 0.4710 0.7957 PRO 2 0.4644 0.2119 0.2260 0.4315 0.7464 CCC 75.97 98.71-25.12 69.86 197.79 ASSETS 10,886.11 6,947.28 4,104.9 8,930 20,427.3 GROWTH 0.0746 0.1665-0.0938 0.0573 0.2589 LEV 0.6325 0.1839 0.3635 0.6536 0.8586 PRO 1 y PRO 2 denoes he gross operaing income and he ne operaing income, respecively. CCC is he Cash Conversion Cycle; ASSETS are he oal asses in housands of euros; GROWTH he growh opporuniies; and LEV he leverage. 22
Table 3 Correlaion marix PRO 1 PRO 2 CCC SIZE GROWTH LEV PRO 1 1.0000 PRO 2 0.9916*** 1.0000 CCC -0.2166*** -0.1976*** 1.0000 SIZE -0.3404*** -0.3593*** 0.1181*** 1.0000 GROWTH 0.0016 0.0082-0.0649*** 0.1374*** 1.0000 LEV -0.2201*** -0.2035*** -0.0896*** -0.0222* 0.1387*** 1.0000 PRO 1 y PRO 2 denoes he gross operaing income and he ne operaing income, respecively. CCC is he Cash Conversion Cycle; SIZE he size; GROWTH he growh opporuniies; and LEV he leverage. *Indicaes significance a 10% level;**indicaes significance a 5%level; ***indicaes significance a 1% level. 23
Table 4 Esimaion resuls of Cash Conversion Cycle-profiabiliy relaion 2 ( PRO ), 0 1, 2, 3, 4, 5,, (1a ) 1 i = β + β CCC i + β CCC i + β SIZE i + β GROWTH i + β LEV i + λ + η i + ε i 2 ( PRO 2 ) = β 0 + β 1CCC + β 2 CCC + β 3 SIZE + β 4 GROWTH + β 5 LEV i, + λ + η i + ε (1b) Equaion (1a) Equaion (1b) CCC 0.02359** (2.24) 0.01791* (1.76) CCC 2-0.01061*** (-2.94) -0.01004*** (-2.88) SIZE -0.27873*** (-8.08) -0.26192*** (-7.77) GROWTH -0.01552 (-0.47) -0.01766 (-0.55) LEV -0.19814* (-1.65) -0.17159 (-1.45) m 2-0.74-0.87 Hansen Tes 46.82(45) 45.47(45) Observaions 4854 4854 The dependen variable is he gross operaing income in equaion (1a) and he ne operaing income in equaion (1b). CCC is he Cash Conversion Cycle divided by 100 and CCC 2 is square; SIZE he size; GROWTH he growh opporuniies; and LEV he leverage. Time and indusry dummies are included in he esimaions, bu no repored. Z saisic in brackes. *Indicaes significance a 10% level;**indicaes significance a 5%level; ***indicaes significance a 1% level. m 2 is a serial correlaion es of second-order using residuals of firs differences, asympoically disribued as N(0,1) under null hypohesis of no serial correlaion. Hansen es is a es of over-idenifying resricions disribued asympoically under null hypohesis of validiy of insrumens as Chi-squared. Degrees of freedom in brackes. 24
Table 5 Esimaion resuls of deviaions from opimal Cash Conversion Cycle-profiabiliy relaion ( PRO ), 0 1, 2, 3, 4,, ( 3 a ) 1 i = α + α Deviaion i + α SIZE i + α GROWTH i + α LEV i + λ + η i + ε i ( PRO ), 0 1, 2, 3, 4,, ( 3 b ) 2 i = α + α Deviaion i + α SIZE i + α GROWTH i + α LEV i + λ + η i + ε i ( Deviaion* AOD) + γ + γ + γ + λ + η ε (4 ) ( PRO ) 1 = γ 0 + γ 1Deviaion i, + γ 2 3SIZE i, 4GROWTH i, 5LEV i, i + a ( Deviaion* AOD) + γ + γ + γ + λ + η ε (4 ) ( PRO ) 2 = γ 0 + γ 1Deviaion i, + γ 2 3SIZE i, 4GROWTH i, 5 LEV i, i + b PRO 1 PRO 2 Equaion (3a) Equaion (4a) Equaion (3b) Equaion (4b) Deviaion -0.0347** (-2.02) -0.0424** (-2.09) -0.0285* (-1.73) -0.0329* (-1.67) (Deviaion*AOD) 0.0051 (0.19) -0.0029 (-0.11) SIZE -0.2588*** (-6.36) -0.2803*** (-7.27) -0.2441*** (-6.14) -0.2674*** (-7.09) GROWTH -0.0107 (-0.26) 0.0053 (0.15) -0.0081 (-0.20) 0.0051 (0.15) LEV -0.1936 (-1.56) -0.1752 (-1.48) -0.1853 (-1.52) -0.1635 (-1.41) F 3.16 3.09 m 2-0.90-0.83-1.13-1.01 Hansen Tes 38.75(36) 49.04 (45) 38.42 (36) 47.71 (45) Observaions 4854 4854 4854 4854 The dependen variable in equaions (3a) and (4a) is he gross operaing income. The dependen variable in equaions (3b) and (4b) is he ne operaing income. Deviaion denoes he deviaions from opimal CCC; (Deviaion*AOD) he ineracion erm; SIZE he size; GROWTH he growh opporuniies; and LEV he leverage. Time and indusry dummies are included in he esimaions, bu no repored. Z saisic in brackes. *Indicaes significance a 10% level;**indicaes significance a 5%level; ***indicaes significance a 1% level. F is he F-es for he linear resricion es under he null hypohesis Ho: γ 1 +γ 2 = 0 in equaions (4a) and (4b). m 2 is a serial correlaion es of second-order using residuals of firs differences, asympoically disribued as N(0,1) under null hypohesis of no serial correlaion. Hansen es is a es of over-idenifying resricions disribued asympoically under null hypohesis of validiy of insrumens as Chi-squared. Degrees of freedom in brackes. 25