Pecking Order Theory versus Trade-Off Theory: are service SMEs capital structure decisions different?

Transcription

1 Serv Bus (2011) 5: DOI /s EMPIRICAL ARTICLE Pecking Order Theory versus Trade-Off Theory: are service SMEs capial srucure decisions differen? Zélia Silva Serrasqueiro Manuel Rocha Armada Paulo Maçãs Nunes Received: 28 March 2011 / Acceped: 15 Sepember 2011 / Published online: 29 Sepember 2011 Ó Springer-Verlag 2011 Absrac This paper seeks o analyse if he capial srucure decisions of service small and medium-sized enerprises (SMEs) are differen from hose of oher ypes of firm. To do so, we consider four research samples: (i) 610 service SMEs; (ii) 126 service large firms; (iii) 679 manufacuring and consrucion SMEs; and (iv) 132 manufacuring and consrucion large firms. Using he wo-sep esimaion mehod, he empirical evidence obained in his sudy shows ha he capial srucure decisions of service SMEs are differen from hose of oher ypes of firm. Service SMEs capial srucure decisions are closer o he assumpions of Pecking Order Theory and furher removed from hose of Trade-Off Theory compared wih he case of oher ypes of firm. Keywords Oher firms Pecking Order Theory Service SMEs Trade-Off Theory Two-sep esimaion mehod JEL Classificaion G32 L80 Z. S. Serrasqueiro P. M. Nunes Managemen and Economics Deparmen, Beira Inerior Universiy, Covilhã, Porugal Z. S. Serrasqueiro P. M. Nunes CEFAGE Research Cener, Évora Universiy, Évora, Porugal Z. S. Serrasqueiro (&) Faculdade de Ciências Sociais e Humanas, Universidade da Beira Inerior, Esrada do Sineiro, Polo IV, Erneso Cruz, Covilhã, Porugal zelia@ubi.p M. R. Armada Managemen and Economics School, Minho Universiy, Braga, Porugal M. R. Armada NEGE Research Cener, Minho Universiy, Braga, Porugal 123

2 Pecking Order Theory Versus Trade-Off Theory: Are Service SMEs Capial Srucure Decisions Differen? 1. Inroducion In he conex of firms financing decisions, afer he imporan conribuions by Modigliani and Miller (1958, 1963), various sudies were devoed o invesigaing wha he main deerminans of firms capial srucure were. In his connecion, wo heories are paricularly relevan in explaining firms capial srucure: i) Trade-Off Theory (Kraus and Lizenberger, 1973; Sco, 1977; Kim, 1978); and ii) Pecking Order Theory (Myers, 1984; Myers and Majluf, 1984). According o Trade-Off Theory, firms seek o achieve he arge deb raio ha corresponds o he poin where he marginal benefis equal he marginal coss of deb. According o Pecking Order Theory, variaions in deb do no are no moivaed by he wish o achieve he arge deb raio, bu are he consequence of he accumulaed funding needs ha are no covered by inernal finance, and in his siuaion firms prefer o urn o deb raher han exernal equiy. According o hese wo main heories, capial srucure decisions are influenced by several firm characerisics, namely: profiabiliy, size, age, growh opporuniies, risk, asse srucure, non-deb ax shields, and ohers. Pecking Order and Trade-off heories do no converge regarding he influence of hese deerminans on capial srucure decisions. Small and medium-sized enerprises (SMEs), mainly smaller and younger firms wih growh opporuniies ha are associaed wih high risk, are known o have difficuly in obaining exernal finance. Size and angible asses can be paricularly imporan for SMEs wih insufficien inernal finance o be able o obain deb on more advanageous erms o fund he muliple invesmen opporuniies. 2

3 Service SMEs have peculiar characerisics ha may jusify he differences in heir capial srucure decisions in comparison o hose aken by oher ypes of firm. Indeed, smaller size ogeher wih lower level of angible asses may aggravae he problems of informaion asymmery beween service SME owners/managers and crediors, leading crediors o impose paricularly adverse condiions for hese firms o obain deb. The service indusry secor is especially imporan in he European economy in general, and in peripheral economies such as Porugal in paricular. In his conex, he service indusry secor is especially imporan for economic and employmen growh in Porugal. SMEs accoun for 99.68% of all firms in Porugal (INE, 2010). Among Poruguese SMEs, service SMEs are paricularly imporan. Beween 2000 and 2005, wih respec o SMEs in he service indusry, growh in he number of firms was found o be he 10.1%, employmen growh was 8.6%, and growh of business urnover was 7.3%. In comparison, concerning SMEs in manufacuring indusry, growh in he number of firms was found o be 4.5%, employmen fell by 0.2%, and business urnover fell by 0.3%. Given he considerable imporance of service SMEs in he Poruguese economy, i is paricularly relevan o sudy how informaion asymmery beween hese firms and crediors, influences heir capial srucure decisions, making service SMEs a special case regarding financing sraegies, compared o wha occurs in oher ypes of firms. Therefore, his paper seeks o analyse if he capial srucure decisions of service small and medium-sized enerprises (SMEs) are differen from hose of oher ypes of firm To reach he goal of his sudy, we use four research samples: i) 610 service SMEs; ii) 126 large service firms; iii) 679 SMEs in manufacuring and consrucion indusries; and iv) 132 large firms in manufacuring and consrucion indusries. In his way, we compare he resuls obained for service SMEs wih hose for larger firms wih higher 3

4 level of angible asses o analyse he possible influence of problems of informaion asymmery in relaionships beween service SME owners/managers and crediors on he financing decisions of service SMEs, compared o wha occurs in oher ypes of firm. Mehodologically, so as o avoid possible resul bias due o he maer of survival, we use he wo-sep esimaion mehod proposed by Heckman (1979). In he firs sep, we use probi regressions, considering all service SMEs and oher ypes of firm. In he second sep, afer calculaing he inverse Mill s raio, and including i in he regressions, considering surviving service SMEs and oher surviving firms, we esimae he regressions wih respec o he applicabiliy of Pecking Order Theory and Trade-Off Theory o he capial srucure decisions of service SMEs and oher ypes of firm. This paper has various original conribuions regarding he sae-of-he-ar of he lieraure on SME capial srucure in general, and he capial srucure of service SMEs in paricular. Firsly, i is pioneering in sudying specifically he applicabiliy of represenaive models of Pecking Order Theory and Trade-Off Theory o he capial srucure decisions of service SMEs. Secondly, i is pioneering in comparing he applicabiliy of represenaive models of Pecking Order Theory and Trade-Off Theory beween he siuaion of service SMEs and ha of oher ypes of firm, namely large service firms, manufacuring and consrucion SMEs and large manufacuring and consrucion firms. Thirdly, he sudy is pioneering in applying he wo-sep esimaion mehod o he capial srucure decisions of service SMEs, and comparing hem wih he capial srucure decisions of oher ypes of firm. The muliple empirical evidence presened in his paper allow us o draw a paricularly imporan conclusion for managers, praciioners and academics: he capial srucure decisions of service SMEs are considerably differen from hose of oher ypes of firm, such as large service firms, and SMEs and large firms in manufacuring and 4

5 consrucion indusries. Decisive for his resul could be he combinaion of smaller size and lower level of angible asses ha paricularly affec service SMEs, compared o oher ypes of firm. Afer his inroducion, his paper has he following srucure: i) Secion 2. Mehodology and Research Hypoheses, presens he models o esimae, he variables used and corresponding measures, he esimaion mehod and research hypoheses; ii) Secion 3. Daabase and Descripive Saisics, presens he daabase used and he descripive saisics of he variables used in he sudy; iii) Secion 4. Resuls, presens he empirical evidence obained in he sudy; iv) Secion 5. Discussion of he Resuls, goes on o discuss he empirical evidence obained; and v) finally, Secion 6. Conclusions and Implicaions, presens he main conclusions and implicaions of he sudy. 2. Mehodology and Research Hypoheses In his secion we presen he models used o es he applicabiliy of Pecking Order and Trade-Off heories o he capial srucure decisions of service SMEs and oher ypes of firm. Iniially, we presen he models o esimae ogeher wih he variables used and heir corresponding measures, and afer his we presen he research hypoheses Pecking Order Theory Relaionship beween Financial Defici and Deb POT Model I Firsly, o es Pecking Order Theory, we use he model proposed by Shyam-Sunder and Myers (1999). This model consiss of esing a regression beween financial defici and variaions in deb. 5

6 To solve he problem of possible bias in he resuls due o he survival issue, we use he wo-sep esimaion mehod proposed by Heckman (1979). In he firs sep, we esimae a probi regression, aking all surviving and non-surviving firms, considering all explanaory variables corresponding o he regressions o esimae in he second sep. Therefore, he regressions o esimae can be presened as follows: Pr(, (1) 1) 0 FD d z where: Pr(, 1) is survival probabiliy; is he parameer o measure he impac of i financial defici on survival probabiliy, FD i, is financial defic d represens emporal dummies, and z i, is he error erm. In he second sep of esimaion, we consider he inverse Mill s raio 2, as an explanaory variable for service SMEs and oher ypes of firm. According o Pecking Order Theory, variaions in deb occur exclusively as a funcion of firms financing needs. We exend he model proposed by Shyam-Sunder and Myers (1999), including he inverse Mill s raio in regressions as an explanaory variable, so as o solve he problem of possible resul bias as a consequence of he survival issue. Therefore, he regressions o esimae can be presened as follows: (2) D 0 BFD d where is he difference beween deb in he curren period and deb in he previous D i, period, B is he parameer o measure he impac of financial defici on deb variaions, 2 The inverse Mill s raio is he raio beween cumulaive densiy funcion and he densiy funcion. The designaion of inverse Mill s raio is due o he fac ha Mill s raio considers he inverse of Hazard raio (also known as force of moraliy). For a deailed descripion of calculaion of he inverse Mill s raio, see Heckman (1979). 6

7 is he inverse Mill s raio, d represens emporal dummies, and is assumed o have normal disribuion. is he error ha To esimae equaion (2), we urn o OLS regressions for wo fundamenal reasons: i) he non-exisence of he lagged dependen variable in he relaionship forecas by equaion (2) makes use of dynamic esimaors impossible; and ii) since he dependen variable is in firs differences, non-observable individual effecs ( v i ) become irrelevan, and i is no impossible o esimae he relaionship forecas in equaion (2) wih panel models considering random or fixed non-observable individual effecs. Given ha heeroskedasiciy is normally a relevan phenomenon in empirical sudies ha use cross-secion daa, sandard deviaions of he parameers are esimaed according o he Whie esimaor. This esimaor allows us o obain sandard deviaions of esimaed parameers consisen wih he possible exisence of heeroskedasiciy. Deerminans of Deb POT Model II Following oher sudies (Michaellas e al., 1999; Aybar e al., ; López-Gracia and Sánchez-Andújar, 2007; López-Gracia and Sogorb-Mira, 2008), we es he Pecking Order Theory, on he basis of a second regression model, considering he deerminans of deb according o wha is prediced by Pecking Order Theory: i) cash flow; ii) age; and iii) ineracion beween cash flow and growh opporuniies. As in Model I, in he firs sep we esimae probi regressions in order o esimae he inverse Mill s raio. The regressions o esimae can be presened as follows: Pr( 1) d z CF 0 1 AGE 2 HGOLCF 3 LGOHHCF 4 (3) 3 Aybar Arias, C., Casino Marínez, A. and López Gracia, J. (2004). Efecos Financieros y Esraégicos sobre la Esrucura de Capial de la Pequeña y Mediana Empresa. Moneda y Crédio, 58,

8 in which: CF i, corresponds o cash flow, given by he raio of earnings afer ineres and axes plus depreciaion o oal asses; AGE i,, is age, given by he number of years he firm has been in exisence; HGOLCF i,, are he growh opporuniies of firms (i), a a given momen (), corresponding o siuaions of high growh opporuniies and low cash flow; LGOHCF i,, are he growh opporuniies of firm (i), a a given momen (), corresponding o siuaions of low growh opporuniies and high cash flow. To calculae HGOLCF, we consider iniially a dummy variable wih he value of 1 i corresponding o firms ha, a a given momen, have simulaneously growh opporuniies above he median of growh opporuniies of he oal sample and cash flow under he median of cash flow of he oal sample; and he value of 0 in he remaining siuaions. To calculae LGOHCF i, we consider, iniially, a dummy variable wih he value of 1 when firms, a a given ime, have simulaneously growh opporuniies under he median of growh opporuniies of he oal sample and cash flow above he median of cash flow of he oal sample; and he value of 0 in he remaining siuaions. Finally, o calculae he variables HGOLCF i, and LGOHCF i,, we muliply he previously calculaed dummy variables by sales growh (considered as a measure of growh opporuniies). In he second sep, o esimae he relaionships forecas by Pecking Order Theory beween deerminans and deb, for service SMEs and oher ypes of firm, we use saic panel models. The regressions o esimae can be presened as follows: D CF 0 i 1 u d AGE 2 HGOLCF 3 LGOHCF 4, (4) 8

9 in which: D i, corresponds o deb, given by he raio of oal liabiliies o oal asses and u i are firms specific facors ha are no direcly observable from deb deerminans. We use he LM es in order o verify if non-observable individual effecs (u i ) are relevan in explaining deb. The null hypohesis indicaes he irrelevance of nonobservable individual effecs, agains he alernaive hypohesis of relevance of nonobservable individual effecs. In he case of rejecing he null hypohesis, due o he relevance of non-observable individual effecs in explaining deb, an OLS regression is no he mos appropriae way o esimae he relaionships beween deerminans and deb. If non-observable individual effecs are relevan in explaining deb, we also use he Hausman es o check wheher non-observable individual effecs are correlaed wih he independen variables. The null hypohesis indicaes he non-exisence of correlaion beween non-observable individual effecs and he independen variables, agains he alernaive hypohesis ha non-observable individual effecs are correlaed wih he dependen variable. By no rejecing he null hypohesis, non-observable individual effecs are no correlaed wih he independen variables, and so i makes no difference esimaing relaionships beween deerminans and deb wih a random or fixed effec panel model. In he case of rejecing he null hypohesis, here is correlaion beween non-observable individual effecs and he independen variables, and so he relaionships beween deerminans and deb should be esimaed using a fixed effec model. This paper presens he resuls of he LM and Hausman ess for he differen ypes of firms considered, and he mos suiable model for esimaing resuls wih respec o o relaionships beween deerminans and deb for he differen ypes of firms considered. 9

10 Research Hypohesis According o Pecking Order Theory, variaions in deb occur exclusively as a funcion of firms financing needs, i.e., deb variaions are a funcion of he financial defici found a a given ime (Shyam-Sunder and Myers, 1999). In addiion, according o Pecking Order Theory, we can also expec ha: i) firms wih greaer cash flow resor less o deb (Myers, 1984; Myers and Majluf, 1984); ii) older firms resor less o deb (Diamond, 1989; Ang, 1991); iii) firms wih greaer growh opporuniies and low cash flow resor more o deb (Myers, 1984); and iv) firms wih low growh opporuniies and high cash flow resor less o deb (Myers, 1984). In he SME conex, various sudies poin oward firms preferring o fund hemselves hrough reained profis, or else hrough recourse o he savings of family or friends (Pei and Singer, 1985; Ang 1992; Scherr and Hulbur 2001; López-Gracia and Sánchez-Andújar, 2008; Ramalho and Silva 2009). SME preference for reained profis is relaed o he paricular difficulies faced by SMEs in obaining exernal finance, due o heir greaer likelihood of bankrupcy (Pei and Singer, 1985; Ang, 1991). Indeed, Franco and Haase (2010) conclude ha financing resricions are one of he main problems affecing hese firms survival in heir operaing markes. Cruz-Ros e al. (2010) conclude ha inangible asses are paricularly relevan for increased performance in service firms. However, he auhors conclude ha service firm aciviies based on inangible asses may conribue o hese firms having greaer difficuly in obaining deb. According o Cressy and Olofsson (1997) and Abor (2007), for SMEs wih higher level of angible asses i is easier o obain deb han for SMEs whose aciviies are based predominanly on inangible asses. For Cressy and Olofsson (1997), greaer inangibiliy of asses in service firms in general, and service SMEs in paricular, conribues decisively o hese firms o have greaer difficuly in obaining 10

11 deb han may be he case of firms belonging o oher secors of aciviy where angible asses are more imporan. Service SMEs may be paricularly dependen on inernal finance o fund heir muliple growh opporuniies. Smaller size, associaed wih he imporance of inangible aciviies in hese firms aciviies, and consequenly lower capaciy in providing angible asses securiy on deb may aggravae he problems of informaion asymmery. Therefore, crediors may hinder service SMEs access o deb. When inernal finance is insufficien, service SMEs paricular difficuly in obaining exernal funding may preven hese firms from aking advanage of he muliple opporuniies available o his ype of firm. The majoriy of service SMEs do no fulfill he requiremens o be lised in he sock marke, and when firms have lile radiion of using venure capial, as occurs in peripheral economies like Porugal, SMEs are resrained in using alernaive funding sources o inernal finance. Based on he above argumens, we can expec ha: i) he impac of financial defici on variaions of deb is of a greaer magniude for service SMEs han for oher ypes of firm; ii) as a funcion of cash flow, age, low growh opporuniies and high cash flow, service SMEs urn less o deb han oher ypes of firms,; and iii) in siuaions of high growh opporuniies and low cash flow, service SMEs urn less o deb han oher ypes of firm. Therefore, we formulae he following research hypohesis: H1: The capial srucure decisions of service SMEs are closer o wha is forecas by Pecking Order Theory han he capial srucure decisions of oher ypes of firm. 11

12 2.2. Trade-Off Theory Adjusmen of Acual Level of Deb oward Targe Deb Raio and Deb Deerminans We presen he parial adjusmen ha allows us o es he degree of adjusmen of acual deb oward arge deb raio, and he relaionships beween deb and deerminans forecas by Trade-Off Theory. Firsly, we esimae probi regressions, considering as explanaory variables he deerminans of deb according o Trade-Off Theory. The regressions o esimae can be presened as follows: Pr( 1) NDTS 5 PROF 0 ETR 6 1 SIZE EVOL D 7 2 S TANG 3 d z GO 4, (5) in which: profiabiliy ( PROF, ) is given by he raio of earnings before ineres and i axes o oal asses; size ( SIZE, ) is he naural logarihm of oal asses; asse i angibiliy ( TANG, ) is he raio of angible asses o oal asses; growh opporuniies i GO, ) is he growh of oal asses; non-deb ax shields ( NDTS, ) is he raio of ( i i depreciaions and amorizaions o oal asses; effecive ax rae ( ETR, ) is he raio of acual income ax paid o ne axable income before axes; level of risk ( EVOL, ) is he absolue value of he firs difference of percenage change of earnings before ineres, axes and depreciaion. In he second sep of esimaion, we esimae he adjusmen of he acual deb level in service SMEs and oher ypes of firm, oward respecive arge deb raios, as well as he relaionships beween deerminans and deb forecas by Trade-Off Theory, using he GMM sysem (1998) dynamic esimaor. To do so, jus as López-Gracia and i i 12

13 Sánchez-Andujar (2007) and López-Gracia and Sogorb-Mira (2008), we use he parial adjusmen model, which is given by: D D D * D ), (6) 1 ( 1 in which: D 1 is he deb of firm i in he period -1; D * is he arge deb raio of firm i in period, and is he speed of adjusmen of acual level of deb oward arge deb raio. To esimae he above equaion i is necessary o find he arge deb raio ha is no direcly observable. In his sudy we consider jus as López-Gracia and Sánchez- Andújar (2007) and López-Gracia and Sogorb-Mira (2008), ha arge deb raio depends on firms specific characerisics. Therefore, firms arge deb raio is given by: n D * i K Z, k, DS d ui v, (7) K 1 in which Z K i,, is he deerminan k of he deb of firm i a ime, K are he coefficiens of each deb deerminan and D v i, is he error erm. Subsiuing (7) in (6) and regrouping he erms, we have: n 0D 1 K Z k, i K 1, (8) in which: 0 (1 ), K K, i ui, d, and i, v i,. To esimae he equaion (8) on he basis of radiional panel mehods, considering fixed or random individual effecs, we obain biased and inconsisen esimaes of he parameers, given ha, asides from he exisence of correlaion beween and D 1, i here is also correlaion beween and D 1, i.e., firms non-observable individual effecs and he error are correlaed wih he lagged deb. In addiion, use of dynamic 13

14 esimaors raher han use of radiional panel mehods has he following exra advanages: i) greaer conrol of endogeneiy; ii) greaer conrol of possible collineariy beween explanaory variables; and iii) greaer effeciveness in conrolling effecs caused by he absence of relevan explanaory variables for he resuls. This sudy uses he Generalized Momens Mehod GMM sysem (1998) esimaor by Blundell and Bond (1998) o esimae he model of parial adjusmen. Blundell and Bond (1998) conclude ha when he dependen variable is persisen, he GMM sysem (1998) esimaor is more robus han he Generalized Momens Mehod GMM (1991) esimaor 4. Blundell and Bond (1998) exend he GMM (1991) esimaor, considering a sysem of variables a level and in firs differences. For he variables a level he insrumens are presened in firs differences, and for he variables in firs differences he insrumens are presened a level. Neverheless, he GMM sysem (1998) esimaor can only be considered valid if: i) he resricions, a consequence of use of he insrumens, are valid; and ii) here is no second-order auocorrelaion. To es he validiy of he resricions, we use he Hansen es. The null hypohesis indicaes ha he resricions, imposed by use of he insrumens, are valid. By rejecing he null hypohesis, we conclude ha he resricions are no valid, and so he resuls are no robus. We es for he exisence of firs and second-order auocorrelaion. The null hypohesis is ha here is no auocorrelaion. Rejecing he null hypohesis of nonexisence of second-order auocorrelaion, we conclude ha he resuls are no robus. For he resuls of he GMM sysem (1998) esimaor o be considered robus, he 4 In his sudy, we find persisence of deb for service SMEs and oher ypes of firms. The correlaion coefficien of presen deb and previous deb is in service SMEs, in large service firms, in oher SMEs, and in oher large firms. Therefore, i is clearly advisable o use he GMM sysem (1998) esimaor, raher han he GMM (1991) esimaor. 14

15 resricions imposed by use of he insrumens have o be valid and here can be no second-order auocorrelaion. To es he robusness of he empirical evidence obained, we use he LSDVC (2005) esimaor, proposed by Bruno (2005). The LSDVC (2005) esimaor is appropriae in siuaions wihou a grea number of observaions. If he number of crosssecions is below 30, i is recommendable o use he LSDVC (2005) esimaor o es he robusness of he resuls obained wih he GMM sysem (1998) esimaor. The resuls are presened in an appendix. Research Hypohesis According o Trade-Off Theory, firms adjus he acual deb level oward a arge deb raio (Lev and Pekelman, 1975; Ang, 1976; Taggar, 1977; Jalilvand and Harris, 1984). Lower ransacion coss borne by firms in obaining deb will correspond o greaer adjusmen of he acual deb level oward he arge deb raio. Also according o Trade- Off Theory, we can expec ha: i) firms wih level of profiabiliy urn more o deb (Kraus and Lizenberger, 1973; Sco, 1977; Kim, 1978); ii) larger firms urn more o deb (Rajan and Zingales, 1995); iii) firms wih more angible asses urn more o deb (Myers, 1977; Sco, 1977; Myers and Majluf, 1984; Harris and Raviv, 1991); iv) firms wih greaer growh opporuniies urn less o deb (Jensen and Meckling, 1976; Myers, 1977; Sulz, 1990; McConnell and Servaes, 1995; Barclay e al., 2006); v) firms wih greaer non-deb ax shields urn less o deb (DeAngelo and Masulis, 1980); vi) firms wih higher effecive ax raes urn more o deb (Haugen and Senbe, 1986; Sco, 1976; DeAngelo and Masulis, 1980); and vii) firms wih greaer risk urn less o deb (Bradley e al., 1984; Mackie-Mason, 1990). 15

16 Service SMEs can be paricularly affeced by problems of informaion asymmery in he relaionships beween owners/managers and crediors. Service SME special characerisics, such as higher levels of inangible asses, and consequenly lower level of angible asses may lead lenders o make credi difficul. Furhermore, SMEs smaller size and consequenly higher level of risk and probabiliy of bankrupcy may lead crediors o associae his ype of firm wih high risk, and consequenly make erms of credi difficul. Small firms are known o have a high business risk associaed wih heir aciviies, and high bankrupcy coss, which may imply high ransacion coss in obaining deb (Reves and Sapio, 2010). Due o he high informaion asymmery in relaionships beween he owners/managers of service SMEs and crediors, hese firms may find i paricularly difficul o obain deb, compared o oher ypes of firms. Indeed, firsly he greaer size of large service firms may conribue o reducing he informaion asymmery implici in he relaionships esablished wih crediors, conribuing o deb being obained on more advanageous erms. Secondly, in general, he higher level of angible asses in firms in oher indusries, namely hose belonging o he manufacuring and consrucion indusries implies ha heir risk defaul become lower. Therefore, i is easier for firms belonging o he manufacuring and consrucion indusries o obain deb. For example, large firms in he manufacuring and consrucion indusries migh find much easier o obain deb, due o heir greaer size ogeher wih higher level of angible asses, han service SMEs. Elson and Audresch (2009) sae ha recourse o deb on paricularly adverse erms imposed by crediors can harm firm s profiabiliy, herefore conribuing o diminished 16

17 probabiliy of survival. Therefore, firms ha face paricularly harsh credi erms may choose deb only in cases of exreme fund needs. The greaer informaion asymmery associaed wih relaionships beween service SME owners/managers and crediors, and consequenly higher ransacion coss for his ype of firm, may lead service SMEs o consider he reaching of he arge deb raio as a merely secondary objecive. Firms wih greaer size (Pei and Singer, 1985; Ang, 1991), and/or wih higher level of angible asses (Holmes and Ken, 1991; Abor, 2007), consequenly wih less problems of informaion asymmery, may have lower ransacion coss associaed wih deb. Consequenly, greaer firms and firms wih greaer level of angible asses can choose sraegies more oriened o he long-erm, i.e., more direced oward increasing firm value raher han using exernal finance only o cover insufficiencies of inernal finance. On he basis of he above exposiion, we can expec ha he financing behavior of service SMEs is furher removed from he assumpions of Trade-Off Theory as oher ypes of firm, and so we can expec ha: i) service SMEs have a lower speed of adjusmen of he acual level of deb oward arge deb raio han do he oher ypes of firm; ii) service SMEs resor less o deb as a funcion of profiabiliy, size, angible asses, and effecive ax rae han do he oher ypes of firm; and iii) service SMEs resor more o deb as a funcion of growh opporuniies, non-deb ax shields and risk, compared o he oher ypes of firm. Based on he above, we formulae he following research hypohesis: H2: The capial srucure decisions of service SMEs are furher removed from he forecass of Trade-Off Theory han do he capial srucure decisions of oher ypes of firms. 17

18 3. Daabase and Descripive Saisics 3.1. Daabase This sudy uses he SABI (Analysis Sysem of Iberian Balance Shees) daabase, supplied by Bureau van Dijk for he period beween 1999 and The selecion of firms by indusry secor was done in accordance wih he NACE (Classificaion of Economic Aciviies in he European Union) classificaion 5. We selec SMEs based on he European Union recommendaion L124/36 (2003/361/CE). According o his recommendaion, a firm is considered an SME when i mees wo of he following crieria: i) fewer han 250 employees; ii) oal asses under 43 million euros; and iii) urnover under 50 million euros. Firms ha do no mee hese crieria are considered large firms. To solve he problem of possible resul bias due o he maer of he survival effec, and also seeking o have a more represenaive sample of he Poruguese firms, we consider hree ypes of firm: i) firms belonging o he marke for he whole period of analysis ( ); ii) firms enering he marke during he period of analysis ( ); and iii) firms leaving he marke during he period of analysis ( ). Given he use of dynamic panel esimaors, here are resricions in including firms ha are in he sample for a very limied number of years. According o Arellano and Bond (1991), firms mus be presen in he daabase for a leas four consecuive years o be considered in he economeric analysis and in he second-order auocorrelaion ess 5 Services include: Real Esae Aciviies; Rening of Machinery and Equipmen wihou Operaor and Personal and Household Goods; Compuer and Relaed Aciviies; and Oher Business Aciviies. Oher firms include: Manufacure of food producs; Manufacure of exiles; Manufacure of wood and paperrelaed producs; Manufacure of chemicals and chemical producs; Manufacure of basic meals; Manufacure of machinery and equipmen; and Consrucion. 18

19 ha are essenial o validae he robusness of he resuls. Thus, in selecing service SMEs and oher ypes of firms we consider, besides he crieria saed above, heir inclusion in he samples for a leas four consecuive years. Based on he above crieria, we obain four research samples wih he following composiion: i) sample of 610 service SMEs, of which 468 are in he marke for he whole period of analysis, 66 ener he marke during he period of analysis, and 76 leave he marke during he period of analysis; ii) sample of 126 large service firms, of which 112 are in he marke for he whole period of analysis, 4 ener he marke during he period of analysis, and 10 leave he marke during he period of analysis; iii) sample of 679 SMEs in manufacuring and consrucion indusry secor, of which 506 are in he marke for he whole period of analysis, 61 ener he marke during he period of analysis, and 112 leave he marke during he period of analysis; and iv) sample of 132 large manufacuring and consrucion firms, of which 107 are in he marke for he whole period of analysis, 6 leave he marke during he period of analysis, and 19 ener he marke during he period of analysis. The srucure of he samples used in he paper is presened in Table 1. (Inser Table 1 Abou Here) 3.2. Descripive Saisics Table 2 presens he descripive saisics of he variables used in his sudy concerning service SMEs and oher ypes of firm 6. (Inser Table 2 Abou Here) 6 For service SMEs, he median of cash flow is , being for large service firms, for oher SMEs, and for oher large firms. The median of growh opporuniies is for service SMEs, being for large service firms, for oher SMEs, and for oher large firms. The median values of he cash flow and growh opporuniies variables used for calculaing he HGOLCF and LGOHCF variables are calculaed according o wha is presened in secion 3. Mehodology and Research Hypoheses. 19

20 I sands ou ha service SMEs, compared o wha occurs in oher ypes of firm, have on average: i) greaer growh opporuniies and greaer risk; and ii) less variaion of deb, less deb, lower age, lower level of angible asses and lower non-deb ax shields. 4. Resuls 4.1. Pecking Order Theory Relaionship beween Financial Defici and Deb POT Model I Table 3 presens he resuls of he survival analysis referring o Model I of Pecking Order Theory. (Inser Table 3 Abou Here) We find ha financial defici conribues o diminished probabiliy of survival in service SMEs, which does no occur wih respec o oher ypes of firm. This resul reveals he paricular imporance ha financing resricions may have for service SMEs, paricularly for heir survival. The following ables presen he resuls of he ess of Pecking Order Theory, regarding Model I. (Inser Table 4 Abou Here) (Inser Table 5 Abou Here) 20

21 A posiive relaionship is found beween financial defici and deb variaions in service SMEs and in he oher ypes of firm considered. However, we find ha he coefficien esimaed ha measures he relaionship beween financial defici and deb variaions is considerably higher in service SMEs han in oher ypes of firm. The resuls of he Chow es confirm his, rejecing in all circumsances he null hypohesis of equaliy of he esimaed parameers measuring relaionships beween financial defici and deb variaions in service SMEs and oher ypes of firms. Addiionally, we find ha he regression consan is no significan when he subjec of analysis is service SMEs, bu i is significan in he case of oher ypes of firm. This resul reinforces he idea of he paricular imporance of financial defici in explaining variaions of he deb of service SMEs, compared o he siuaion of oher ypes of firm. In all regressions, he saisically significan relaionship beween he inverse Mill s raio and variaions in deb allows us o conclude ha he inclusion of he inverse Mill s raio in regressions was effecive in conrolling for possible bias in he resuls esimaed. Deerminans of Deb POT Model II Table 6 presens he resuls of he survival analysis regarding Model II of Pecking Order Theory. (Inser Table 6 Abou Here) The resuls sugges ha cash flow is imporan for he survival of service SMEs and oher ypes of firm. However, he magniude of he esimaed parameers indicaes ha he imporance of cash flows greaer for service SMEs han for oher ypes of firm. In addiion, age seems o be imporan for he survival of service SMEs, and large service firms as well as for he survival of manufacuring and consrucion SMEs. Considering 21

22 he magniude of he esimaed parameers, he imporance of age seems o be greaer for he survival of service SMEs. The following ables presen he resuls of he ess of Pecking Order Theory, concerning Model II. (Inser Table 7 Abou Here) (Inser Table 8 Abou Here) In all of he regressions esimaed, he resuls of he LM es indicae he rejecion of he null hypohesis of irrelevance of non-observable individual effecs. Therefore, an OLS regression is no he mos appropriae way o esimae he relaionships beween deerminans and deb in service SMEs and oher ypes of firm. Addiionally, he resuls of he Hausman es indicae he rejecion of he null hypohesis of absence of correlaion beween non-observable individual effecs and explanaory variables, and so nor is a regression considering random non-observable individual effecs a suiable way o esimae he relaionships beween deerminans and deb in service SMEs and oher ypes of firm. This being so, he mos correc way o esimae he relaionships beween deerminans and deb in service SMEs and oher ypes of firm is o consider fixed nonobservable individual effecs in he regressions. For service SMEs, we find ha: i) greaer cash flow, greaer age and siuaions of low growh opporuniies and high cash flow conribue o reduced deb; and ii) siuaions of high growh opporuniies and low cash flow conribue o increased deb. For he remaining ypes of firm, cash flow is also found o conribue o reduced deb, he same is verified wih respec o age for large service firms as well as for manufacuring and consrucion SMEs 7. We also find ha siuaions of high growh opporuniies and low 7 Alhough in his case he relaionship is only saisically significan a 10% level. 22

23 cash flow conribue o increased deb in large service firms as well as in manufacuring and consrucion SMEs, whereas siuaions of low growh opporuniies and high cash flow conribue o reduced deb in manufacuring and consrucion SMEs. The resuls of he Chow es indicae ha for all deerminans, he null hypohesis of equaliy of esimaed parameers is rejeced. The resul of he Chow es of parameers as a whole confirms hose differences. In all regressions esimaed, a saisically significan relaionship is found beween he inverse Mill s raio and deb. We can herefore conclude ha he inclusion of he inverse Mill s raio in regressions was shown o be effecive in solving possible bias in he resuls obained Trade-Off Theory Table 9 presens he resuls of he survival analysis regarding he Trade-Off Theory model. (Inser Table 9 Abou Here) From he resuls presened in Table 9 and considering he magniude of he esimaed parameers, we highligh he following: i) profiabiliy is paricularly imporan for he survival of service SMEs, risk being paricularly imporan for diminished survival of his ype of firm; and ii) size and angible asses are paricularly imporan for he survival of SMEs, and large firms belonging o he manufacuring and consrucion indusries. Tables 10 and 11 presen he resuls of he ess of Trade-Off Theory. (Inser Table 10 Abou Here) (Inser Table 11 Abou Here) 23

24 Regardless of aking service SMEs or oher ypes of firm as he subjec of analysis, he resuls of he Hausman es indicae ha we canno rejec he null hypohesis of validiy of he insrumens used. In addiion, he resuls of he second-order auocorrelaion ess indicae ha we canno rejec he null hypohesis of absence of auocorrelaion. Based on he resuls of he Hansen and second-order auocorrelaion ess, we can consider he resuls obained wih he GMM sysem (1998) esimaor as valid, and consequenly open o inerpreaion. We find ha service SMEs: i) adjus acual deb oward arge deb raio; ii) reduce he level of deb as a funcion of greaer profiabiliy; and iii) reduce he level of deb as a funcion of greaer size and higher levels of angible asses. As for oher ypes of firm, we find ha: i) hey adjus acual deb oward arge deb raio, ii) reduce he level of deb as a funcion of greaer profiabiliy as well as of greaer non-deb ax shields, while hey increase he level of deb as a funcion of higher level of angible asses deb; iii) manufacuring and consrucion SMEs increase he level of deb as a funcion of greaer size iv) manufacuring and consrucion SMEs and large firms reduce he level of deb as a funcion of greaer growh opporuniies; v) large service firms as well as manufacuring and consrucion large firms increase he level of deb as a funcion of higher effecive ax raes; and vi) manufacuring and consrucion SMEs and large firms diminish he level of deb as a funcion of greaer risk. Regarding he regressions presened in Table 10, bu using he LSDVC (2005) esimaor, he resuls presened in appendix are relaively similar o hose presened in Table 10, wih respec o sign, magniude and saisical significance of he esimaed parameers, which confirms he robusness of he resuls presened. The resuls of he Chow es indicae ha we can rejec he null hypohesis of equaliy of esimaed parameers, excep for he deerminans of growh opporuniies and risk in comparing 24

25 service SMEs and large service firms, and hose of growh opporuniies and effecive ax rae in comparing service SMEs, and manufacuring and consrucion SMEs. The resul of he Chow es of esimaed parameers as a whole confirms ha here are significan differences in he deerminans of deb in service SMEs and oher ypes of firm. Inclusion of he inverse Mill s raio in he regressions was also found o be effecive in conrolling for possible bias in he esimaed resuls, since a saisically significan relaionship is found beween he inverse Mill s raio and deb. 5. Discussion of Resuls The impac of financial defici on deb variaions in service SMEs (B= ) is greaer han in he conex of large service firms (B= ), manufacuring and consrucion SMEs (B= ) as well as for manufacuring and consrucion large firms (B= ). The resuls of he Chow es confirm hese differences. In addiion, he regression consan regarding he relaionship beween financial defici and deb variaions in service SMEs is no saisically significan, bu i is no he case in any oher ype of firm. This empirical evidence indicaes ha recourse o deb by service SMEs is considerably more influenced by insufficien inernal finance han wha occurs in oher ypes of firm. The relaionships idenified beween deerminans and deb in service SMEs and oher ypes of firm allow us o draw various conclusions. Firsly, we find ha service SMEs reduce recourse o deb in siuaions of low growh opporuniies and high cash flow, which agrees wih he forecass of Pecking Order Theory. This resul is no found for any oher ype of firm. Secondly, greaer cash flow implies less recourse o deb in service SMEs and oher ypes of firm, corroboraing wha is forecas by Pecking Order 25

26 Theory. However, he magniude of ha reducion in recourse o deb is greaer in service SMEs han in oher ypes of firm. The resuls of he Chow es confirm hose differences. Thirdly, greaer age conribues o reduced recourse o deb in service SMEs and large firms, and in manufacuring and consrucion SMEs, corroboraing Pecking Order Theory. The magniude of reducion in deb, as a consequence of greaer age, is greaer in service SMEs han in oher ypes of firm. This resul is also confirmed by he Chow es. Therefore, age is no a deerminan of deb for large manufacuring and consrucion firms. Fourhly, siuaions of high growh opporuniies and low cash flow conribue o increased recourse o deb in service SMEs and large firms, and in manufacuring and consrucion SMEs, corroboraing wha is forecas by Pecking Order Theory. However, he magniude of he increase in deb is greaer in service SMEs han in large service firms and manufacuring and consrucion SMEs, a siuaion ha is confirmed by he resuls of he Chow es. As for large manufacuring and consrucion firms, we find ha siuaions of high growh opporuniies and low cash flow imply neiher more nor less recourse o deb. The muliple empirical evidence obained allows us o conclude ha service SMEs urn more o deb as a funcion of financial defici and high growh opporuniies and low cash flow, and urn less o deb as a funcion of cash flow, age and siuaions of low growh opporuniies and high cash flow han is he case in oher ypes of firm. We can herefore conclude ha capial srucure decisions in service SMEs are closer o he forecass of Pecking Order Theory han he capial srucure decisions of oher ypes of firm, and so we can accep he previously formulaed hypohesis H1 as valid. The imporance of inernal finance in SME aciviy (Pei and Singer, 1985; Ang 1992; Scherr and Hulbur 2001; López-Gracia and Sánchez-Andújar, 2008; Ramalho and Silva 2009) seems o be paricularly srong in service SMEs. This imporance is 26

27 reinforced by he paricular relevance of cash flow for he survival of service SMEs, and also by he fac ha financial defici conribues o diminishing he survival of service SMEs ha does no occur in oher ypes of firm. The conclusions of Franco and Haase (2010) ha financing resricions may affec SME survival seem o be paricularly relevan in service SMEs. The smaller size of SMEs, and consequenly greaer probabiliy of bankrupcy (Pei and Singer, 1985; Ang, 1991), associaed wih higher level of inangible asses (Cruz-Ros e al., 2010), may decisively conribue o service SMEs o have paricular difficuly in obaining deb. Therefore, service SMEs may urn o deb only in cases of need, i.e., when inernal funds are clearly insufficien o finance heir aciviies. The empirical evidence obained corroboraes he conclusions of Cressy and Olofsson (1997) and Abor (2007), since service SMEs lower level of angible asses compared o manufacuring and consrucion 8, SMEs level of angible asses may decisively conribue o increasing he problems of informaion asymmery faced by service SMEs in he relaionships wih crediors, conribuing o inernal finance being heir main source of funding. Regarding he applicabiliy of he assumpions of Trade-Off Theory o he capial srucure decisions of service SMEs and hose of oher ypes of firm, we find ha heir level of adjusmen of acual deb level oward arge deb raio is lower (α= ) han he level of adjusmen verified in large service firms (α= ), in manufacuring and consrucion SMEs (α= ) as well as in manufacuring and consrucion large firms (α= ). The resuls of he Chow es confirm he differences in he adjusmens of acual deb level oward he arge deb raio beween service SMEs and oher ypes of firm. 8 From he resuls presened in Table 2, we see ha on average, he angible asses of service SMEs are , while he angible asses of manufacuring and consrucion SMEs are

28 Wih respec o he relaionships beween deerminans and deb, he empirical evidence allows us o draw imporan conclusions regarding he applicabiliy of he assumpions of Trade-Off Theory o he capial srucure decisions of service SMEs and oher ypes of firm. Firsly, greaer profiabiliy does no conribue o increased recourse o deb in service SMEs and oher ypes of firm, which does no corroborae wha is forecas by Trade-Off Theory, because firms do no ake advanage of greaer profiabiliy o increase deb ax-shields, arising from greaer level of deb. Secondly, greaer growh opporuniies, greaer non-deb ax shields, and greaer risk imply less recourse o deb by SMEs and large manufacuring and consrucion firms, resuls which are in accordance wih Trade-Off Theory. In service SMEs, growh opporuniies, nondeb ax shields, and risk are no deerminans of heir level of deb. In large service firms, only non-deb ax shields are a deerminan of deb, bu growh opporuniies and risk are no deerminans of deb. Thirdly, higher effecive ax raes conribue o greaer recourse o deb by large service firms and large manufacuring and consrucion firms, hese resuls corroboraing he forecass of Trade-Off Theory. Higher effecive ax raes are no a deerminan of deb in service SMEs and large service firms. Fourhly, greaer size conribues o increased recourse o deb by service and manufacuring/consrucion SMEs, corroboraing he forecass of Trade-Off Theory, bu i is no a deerminan of deb in large service firms and large manufacuring/consrucion firms. In addiion, greaer angibiliy of asses conribues o service SMEs, and large service as well as manufacuring/consrucion firms urning more o deb, hese resuls agreeing wih Trade-Off Theory. I sands ou ha size and asse angibiliy are of greaer relaive imporance for increased recourse o deb in service SMEs han in oher ypes of firm, a siuaion confirmed by he resuls of he Chow es. 28

29 Conrary o our expecaions, greaer size and higher level of angible asses are of greaer relaive imporance for service SMEs urning o deb, compared o he case in oher ypes of firm. Considering only relaionships beween size and deb, and beween angible asses and deb, we would conclude ha he capial srucure decisions of service SMEs are closer o he forecass of Trade-Off Theory han he capial srucure decisions of oher ypes of firm. These resuls may arise from he imporance of greaer size and level of angible asses o reduce he problems of informaion asymmery, herefore crediors gran deb in more favourable erms o service SMEs. The marginal imporance of size and angible asses o obain deb may be greaer for service SMEs han for oher ypes of firm. Service SMEs have a lower level adjusmen of acual deb oward arge deb raio han is found in oher ypes of firms. Addiionally, greaer growh opporuniies, greaer non-deb ax shields and greaer risk do no imply less recourse o deb in service SMEs. Higher effecive ax raes do no imply more recourse o deb in service SMEs, unlike wha occurs in he majoriy of he siuaions in oher ypes of firm. Alhough no for all deerminans, he empirical evidence suggess ha he capial srucure decisions of service SMEs are furher from he assumpions of Trade-Off Theory han he capial srucure decisions of oher ypes of firm, and so we can accep he previously formulaed hypohesis H2 as parially valid. I sands ou ha service SMEs have lower speed of adjusmen of he acual deb level oward arge deb raio han oher ypes of firm. This resul suggess he lower imporance ha he goal o reaching he arge deb raio has for service SMEs compared o he case in oher ypes of firm. Conribuing o hese resuls are he smaller size and he lower levels of angible asses of service SMEs ha imply high ransacion coss, 29

30 which may explain he lower speed of adjusmen of acual level of deb oward he arge deb raio. Värlander and Julien (2010) conclude ha personal relaionships esablished wih cliens can be paricularly imporan o increase performance in service firms. This imporan conclusion by he auhors can be exended o he relaionships beween he owners/managers of service SMEs and crediors. Indeed, esablishing rusing and longlasing relaionships may be paricularly imporan for reducing he informaion asymmery beween service SME owners/managers and crediors, conribuing o lower ransacion coss in obaining deb. Therefore, service SMEs may increase he speed of adjusmen of he acual level of deb oward arge deb raio. On he one hand, he fac ha service SMEs do no reduce he level of deb as a funcion of greaer growh opporuniies, level of risk and non-deb ax shields, and do no increase he level of deb as a funcion of ax rae, shows ha hese firms are less concerned wih a rade-off beween he benefis and coss of deb compared o oher ypes of firm. On he oher hand, he fac ha size and angible asses are paricularly relevan for service SMEs o obain deb indicaes ha greaer size and high level of angible asses may be deerminan for service SMEs o consider more effecively a rade-off beween he benefis and coss of deb. The muliple empirical evidence obained in his sudy clearly indicaes ha problems of informaion asymmery beween firm owners/managers and crediors can be paricularly relevan in he case of service SMEs. Indeed, in hese circumsances, as Elson and Audresch (2009) conclude ha firms wih high informaion asymmery associaed wih heir aciviies may only resor o deb when inernal finance is clearly insufficien. 30