Determinants of investment in fixed assets and in intangible assets for hightech

Transcription

1 Nunes, P., Serrasqueiro, Z., & Maos, A. (2017). Deerminans of invesmen in fixed asses and in inangible asses for high-ech firms. Journal of Inernaional Sudies, 10(1), doi: / /10-1/12 Deerminans of invesmen in fixed asses and in inangible asses for highech firms Journal of Inernaional Sudies Foundaion of Inernaional Sudies, 2017 CSR, 2017 Scienific Papers Paulo Maçãs Nunes Managemen and Economics Deparmen, Beira Inerior Universiy and CEFAGE Research Cenre Porugal Zélia Serrasqueiro Managemen and Economics Deparmen, Beira Inerior Universiy and CEFAGE Research Cenre Porugal Anónio Fernandes de Maos Managemen and Economics Deparmen, Beira Inerior Universiy Porugal Absrac. Based on a sample of 141 Poruguese high-ech firms for he period and using GMM sysem (1998) and LSDVC (2005) dynamic esimaors, his paper sudies wheher he deerminans of high-ech firms invesmen in fixed asses are idenical o he deerminans of heir invesmen in inangible asses. The muliple empirical evidence obained allows us o conclude ha he deerminans of heir invesmen in fixed asses are considerably differen from hose of heir invesmen in inangible asses. Deb is a deerminan simulaing invesmen in fixed asses, wih age being a deerminan resricing such invesmen. Size, age, inernal finance and GDP are deerminans simulaing invesmen in inangible asses, whereas deb and ineres raes resric such invesmen. These resuls le us make imporan suggesions for he owners/managers of high-ech firms, and also for policy-makers. Received: Ocober, s Revision: February, 2017 Acceped: April, 2017 DOI: / /10-1/12 Keywords: invesmen in fixed asses, invesmen in inangible asses, KIBS, panel daa. JEL Classificaion: C23, G32, L26 173

2 Journal of Inernaional Sudies Vol.10, No.1, INTRODUCTION According o Máñez e al. (2015), a significan number of he European Union policies aim o encourage firms increase heir innovaion aciviies, so as o improve heir produciviy. In his conex, he new European long-erm research policy sraegy Horizon 2020 is especially relevan, and Najjar e al. (2013) and Musceola (2015) sae ha firms innovaion is a deerminan of counries economic growh. Audresch (1995), Henrekson & Johansson (2010) and Colombo e al. (2014) say ha high-ech firms are very imporan for developed counries as he drivers of heir economic growh and employmen. In he conex of high-ech firms aciviies, inangible asses are of grea imporance (Dilling-Hansen & Smih, 2014; Hussinger & Pacher, 2015). However, he grea weigh of inangible asses and lesser weigh of fixed asses, which characerizes high-ech firms compared o firms in oher secors, could conribue decisively o high-ech firms finding i difficul o access exernal finance, when inernal funds are clearly insufficien (Coleman & Robb, 2009). Considering he imporance of high-ech firms for economic growh and employmen paricularly in developed counries, and also considering ha hese firms aciviy is characerized by a significan variey of fixed and inangible asses, his paper is he firs o sudy he differences in he deerminans of invesmen in fixed asses and inangible asses in high-ech firms. Mehodologically, we use he sample of 141 Poruguese high-ech firms, for he period , and using he GMM sysem (1998) and LSDVC (2005) dynamic esimaors as esimaion mehods. The evidence obained in his paper allows us conclude here are significan differences beween he deerminans of invesmen in fixed asses and in inangible asses in high-ech firms. Afer his inroducion, he paper is srucured as follows: i) Secion II presens he mehodology used, namely he daabase, variables and esimaion mehods; ii) Secion III presens he resuls of he paper; and iii) Secion IV presens he conclusions and implicaions. 2. METHODOLOGY 2.1. Daabase This sudy uses he SABI (Analysis Sysem of Iberian Balance Shees) daabase from Bureau van Dijks for he period As our subjec of analysis is high-ech firms we selec high-ech firms according o ISIC Rev.3. The sub-secors of high-ech firms are: 2423 Pharmaceuicals; 30 Office, Accouning and Compuing Machinery; 32 Radio, TV and Communicaion Equipmen; 33 Medical, Precision and Opical Insrumens; 353 Aircraf and Spacecraf. So as o avoid bias in he esimaed resuls and simulaneously have a more represenaive sample of he siuaion of high-ech firms in Porugal, we consider he following ypes of high-ech firms: i) hose ha are in he marke for he whole period of analysis ( ); ii) hose enering he marke during he period of analysis ( ) and; iii) hose leaving he marke during he period of analysis ( ). Arellano and Bond (1991) conclude ha when using dynamic panel esimaors, i is necessary for unis of analysis o be included in he daabase for a leas four consecuive years, oherwise hey canno be considered in he second order auocorrelaion ess which are fundamenal o validae he resuls obained using dynamic panel esimaors. Therefore, we only include in he sample high-ech firms which are presen for a leas four consecuive years. 174

3 Paulo Maçãs Nunes, Zélia Serrasqueiro, Anónio Fernandes de Maos Deerminans of invesmen in fixed asses and in inangible asses for high-ech firms Based on he crieria presened, he sample of high-ech firms considered in his sudy is made up as follows: i) 87 high-ech firms presen in he marke for he whole period of analysis ( ); ii) 33 highech firms enering he marke during he period of analysis ( ) and iii) 21 high-ech firms leaving he marke during he period of analysis ( ). The final sample is composed of 141 high-ech firms Variables As dependen variables we consider invesmen in fixed asses and invesmen in inangible asses. The independen variables are: i) size; ii) age; iii) cash flow; iv) deb; v) GDP; and vi) ineres rae. The following able presens he variables used in his sudy ogeher wih heir corresponding measures. Table 1 Variables and Measuremen Variables Dependen Variable Invesmen in Fixed Asses (INVFA ) Invesmen in Inangible Asses (INVITA ) Independen Variables Size (SIZE -1 ) Age (AGE -1 ) Cash Flow (CF -1 ) Deb (LEV ) Gross Domesic Produc (GDP ) Ineres Raes (IR ) Measuremen Percenage of change of growh in fixed asses Percenage of change of growh in inangible asses Logarihm of Sales Logarihm of number of exisence firm years Raio beween operaional resuls before ax and ineress plus amorizaions and oal asses Raio beween deb and oal asses Logarihm of Gross Domesic Produc Per capia Marke ineres rae, corresponding o he 3 monh Euribor rae where: 2.3. Esimaion Mehods The regressions o esimae can be represened as follows: INVFA INVFA SIZE AGE i LEV 4 INVITA LEV 4 0 GDP 5 GDP INVITA 0 5 IR 6 IR 6 1 u d e SIZE 1 i i 2 u d e AGE INVFA, is he invesmen in fixed asses of firm i in he period ; 2 CF 3 CF 3 ; (1) ; (2) INVITA i, is he invesmen in inangible asses of firm i in period ; INVFAi, 1 is he invesmen in fixed asses of firm i in period -1; SIZE INVITAi, 1 is he invesmen in inangible asses of firm i in period -1; 1 is he size of firm i in period -1; AGE 1 is he age of firm i in period -1; CFi, 1 is he cash flow of firm i in he period -1; LEV is he deb of firm i in he period -1; GDP 1 is gross domesic produc in he period -1; IR 1 1 is he ineres rae in period -1; ui are effecs ha are non-observable hrough he independen variables; 175

4 Journal of Inernaional Sudies Vol.10, No.1, 2017 d are annual dummy variables measuring ime effecs no measured by GDP and ineres raes; and i e, is he error, which is assumed o have normal disribuion. Aiming o esimae regressions (1) and (2) we use dynamic panel esimaors. Using dynamic panel esimaors has he following advanages over radiional mehods of panel daa regressions (i.e. OLS regressions, regressions admiing he exisence of random effecs, regressions admiing he exisence of fixed effecs): i) high conrol of endogeneiy; ii) greaer conrol of possible collineariy beween independen variables; iii) more effecive conrol of he effecs of he possible absence of independen variables relevan in explaining he dependen variable. Blundell & Bond (1998) sae ha when he dependen variable is persisen (i.e. when high correlaion is found beween he dependen variable in he curren period and he dependen variable in he previous period), he resuls obained wih he GMM (1991) esimaor may no be robus. We calculae he correlaion coefficiens beween invesmen in fixed asses in he curren and previous periods, obaining a correlaion coefficien of , and beween invesmen in inangible asses in he curren and previous periods, obaining a correlaion coefficien of Therefore, considering he persisence of he invesmen, i is more suiable o use he GMM sysem (1998) esimaor han he GMM (1991) esimaor. Two condiions are necessary for he resuls obained wih he GMM sysem (1998) esimaor o be considered robus. Firsly, he resricions creaed by he insrumens used mus be valid. Secondly, here can be no auocorrelaion of second order errors. To es he validiy of he resricions creaed by use of he insrumens, we use he Hansen es. The null hypohesis is validiy of he resricions, as a consequence of he insrumens used, wih he alernaive hypohesis being non-validiy of he resricions. We also es for he exisence of auocorrelaion of firs and second order errors. The null hypoheses indicae non-exisence of error auocorrelaion, wih he alernaive hypoheses indicaing he exisence of error auocorrelaion. If we do no rejec he null hypoheses of validiy of resricions, as a consequence of he insrumens used, and absence of second order auocorrelaion, he resuls obained wih he GMM sysem (1998) esimaor can be considered robus and consequenly open o inerpreaion. Oherwise, he resuls obained wih he GMM sysem (1998) dynamic esimaor canno be considered robus and are consequenly no open o inerpreaion. Bruno (2005) concludes ha when samples have a number of cross-secions under 30, and consequenly he number of observaions is quie low, resuls obained wih he GMM sysem (1998) esimaor may suffer from some bias. When he number of observaions in no very high, Bruno (2005) proposes a dynamic fixed effec esimaor, namely he LSDVC (Leas Squares Dummy Variable Correced) esimaor. As in his paper, he number of cross-secions is under 30 and he number of observaions is no very high, we use he LSDVC (2005) esimaor o es he robusness of he resuls obained wih he GMM sysem (1998) esimaor. Based on he above, aiming o esimae he regressions expressed by equaions (1) and (2) we resor o he GMM sysem (1998) and LSDVC (2005) dynamic esimaors. 3. RESULTS 3.1. Descripive Saisics and Correlaion Marix Table 2 presens he descripive saisics of he variables used in he paper ogeher wih heir correlaions. 176

5 Paulo Maçãs Nunes, Zélia Serrasqueiro, Anónio Fernandes de Maos Deerminans of invesmen in fixed asses and in inangible asses for high-ech firms Descripive Saisics and Correlaion Marix Table 2 Mean S.D. (1) (2) (3) (4) (5) (6) (7) (8) (1) INVFA (2) INVITA ** 1 (3) SIZE ** 0.08** 1 (4) AGE ** 0.06** 0.19** 1 (5) CF ** 0.18** 0.08** 1 (6) LEV * -0.16** 0.05** ** 1 (7) GDP ** 0.22** 0.11** ** -0.03* 1 (8) IR * -0.17* -0.03* -0.06** -0.04** -0.05** 0.06* 1 Noes: 1. ** saisical significan a 1% level; * saisical significan a 5% level. We can sae ha invesmen in fixed asses is quie similar o invesmen in inangible asses. In addiion, we find ha he variables of: i) invesmen in fixed asses, invesmen in inangible asses and cash flow are considerably volaile, since he sandard deviaions of he variables are above he respecive means; and ii) size, age, deb, Gross Domesic Produc and ineres rae are no very volaile, as he sandard deviaions of he variables are under he respecive means. Gujarai and Porer (2010) sae ha when he correlaion coefficiens beween independen variables are under 50%, problems of collineariy beween independen variables are no paricularly relevan. All he correlaion coefficiens beween independen variables are found o be under 50%, and so problems of collineariy beween independen variables will no be relevan in his sudy Regressions Table 3 presens he regressions referring o he deerminans of invesmen in fixed asses 1 and deerminans of invesmen in inangible asses, considering he GMM sysem (1998) and LSDVC (2005) dynamic esimaors as mehods. Wheher considering invesmen in fixed asses or invesmen in inangible asses, he resuls of he Hansen es le us conclude we canno rejec he null hypohesis of validiy of he insrumens used. Furhermore, also wheher considering invesmen in fixed asses or in inangible asses, he resuls of he second order auocorrelaion ess indicae we canno rejec he null hypohesis of absence of second order auocorrelaion. Based on he Hansen and second order auocorrelaion ess, we find ha irrespecive of considering invesmen in fixed asses or invesmen in inangible asses, he resuls are robus and consequenly open o inerpreaion. Concerning he deerminans of invesmen in fixed asses and hose of invesmen in inangible asses, he resuls obained wih he LSDVC (2005) dynamic esimaor are almos idenical o hose obained wih he GMM sysem (1998) dynamic esimaor, confirming he empirical evidence obained here. The muliple empirical evidence obained allows us o conclude ha: i) deb is a deerminan simulaing invesmen in fixed asses; ii) age is a deerminan resricing invesmen in fixed asses; iii) size, age and cash flow are deerminans simulaing invesmen in inangible asses; and iv) deb and he ineres rae are 1 Alernaively, jus as Colombo e al. (2014), we esimae he deerminans of invesmen in fixed asses using he error correcion model (ECM). The resuls are similar o hose presened in Table 3 and can be requesed from he auhors. 177

6 Journal of Inernaional Sudies Vol.10, No.1, 2017 deerminans resricing invesmen in inangible asses. In addiion, posiive relaionships are found beween invesmen in he curren period and invesmen in he previous period, irrespecive of focusing on invesmen in fixed asses or in inangible asses, revealing ha invesmen in fixed and inangible asses is persisen over ime. Table 3 Invesmen Deerminans Dependen Variable: INVFA Dependen Variable: INVITA Independen Variables GMM sysem (1998) LSDVC (2005) GMM sysem (1998) LSDVC (2005) INVFA ** ( ) ** ( ) INVIA ** ( ) ** ( ) SIZE ( ) ( ) ** ( ) ** ( ) AGE * ( ) ** ( ) ** ( ) * ( ) CF ( ) ( ) ** ( ) ** ( ) LEV ** ( ) ** ( ) * ( ) ** ( ) GDP (0.12I30) ( ) ** ( ) ** ( ) IR ( ) ( ) ( ) ** ( ) CONS ** ( ) ** ( ) ( ) ( ) Hansen(χ 2 ) m 1 (N(0.1)) -6.18** -6.62** -4.39** -4.14** m 2 (N((0.1)) Firms Observaions Noes: 1. Sandard deviaions in parenhesis. 2. ** significan a 1% level; * significan a 5% level. 3. The regressions include ime dummy variables bu are no shown. 4. CONCLUSIONS AND IMPLICATIONS Based on a sample of 141 high-ech firms for he period and using he GMM sysem (1998) and LSDVC (2005) dynamic esimaors, his paper sudies he deerminans of invesmen in fixed asses and in inangible asses. The paper makes an imporan conribuion o he lieraure on high-ech firms invesmen, since i is pioneering in comparing he deerminans of invesmen in fixed asses and ha in inangible asses. The muliple empirical evidence obained reveals ha he deerminans of invesmen in fixed asses are quie differen from hose of invesmen in inangible asses. Firsly, deb is a deerminan simulaing invesmen in fixed asses, wih age resricing invesmen in fixed asses. Secondly, size, age, cash flow and GDP simulae invesmen in inangible asses, while deb and ineres raes resric such invesmen. Furhermore, invesmen in fixed asses and invesmen in inangible asses are found o be persisen, wih ha persisence being greaer in he case of invesmen in fixed asses. 178

7 Paulo Maçãs Nunes, Zélia Serrasqueiro, Anónio Fernandes de Maos Deerminans of invesmen in fixed asses and in inangible asses for high-ech firms The empirical evidence obained in his paper les us make imporan suggesions for policy-makers and for he owners/managers of high-ech firms. For he former: i) given he imporance of inangible asses for high-ech firms, and considering ha inernal finance is a deerminan simulaing invesmen in his ype of asse, wih deb being a resricive deerminan, we sugges creaing specific lines of credi o suppor invesmen in inangible asses when inernal financing is clearly insufficien for he purpose; ii) considering ha invesmen in inangible asses is very sensiive o changes in he economic siuaion, increasing as a funcion of GDP and diminishing as a funcion of ineres raes, we sugges creaing specific programmes o suppor he aciviy of high-ech firms in periods of recession characerized by falling GDP and increased ineres raes; and iii) given ha greaer size and age are deerminans simulaing invesmen in inangible asses, we sugges providing specific suppor for small and young high-ech firms wih insufficien inernal finance o ake advanage of all heir good invesmen opporuniies hrough increasing heir inangible asses. For he owners/managers of high-ech firms: i) we sugges forming rusing relaionships wih crediors so ha when inernal finance is insufficien, high-ech SMEs can urn o deb on advanageous erms o finance heir invesmen in inangible asses, hese rusing relaionships being all he more imporan, he smaller and younger he high-ech firm; and ii) considering ha high-ech firms finance heir invesmen in fixed asses hrough deb, wih inernal funding apparenly no being relevan for he purpose, we sugges a rade-off is found beween inernal finance and deb, so ha siuaions of excessive deb do no conribue o considerable sress in managing financial resources, due o paying high charges on he deb conraced, which could mean no making efficien use of good invesmen opporuniies in he fuure. REFERENCES Al-Najjar, B., & Elgammal, M. M. (2013). Innovaion and credi raings, does i maer? UK evidence. Applied Economics Leers, 20(5), Arellano, M., & Bond, S. (1991). Some ess of specificaion for panel daa: Mone Carlo evidence and an applicaion o employmen equaions. The review of economic sudies, 58(2), Audresch, D. B. (1995). Innovaion and indusry evoluion. Mi Press. Blundell, R., & Bond, S. (1998). Iniial condiions and momen resricions in dynamic panel daa models. Journal of economerics, 87(1), Bruno, G. S. (2005). Approximaing he bias of he LSDV esimaor for dynamic unbalanced panel daa models. Economics Leers, 87(3), Coleman, S., & Robb, A. (2009). A comparison of new firm financing by gender: evidence from he Kauffman Firm Survey daa. Small Business Economics, 33(4), Colombo, M. G., Croce, A., & Guerin M. (2014). Does informal risk capial relax he financial consrains of highech enrepreneurial venures?. Applied Economics Leers, 21(5), Dilling-Hansen, M., & Smih, V. (2014). R&D, expor and produciviy: esing he Busos model on Danish daa. Applied Economics Leers, 21(11), Gujara D. N., & Porer, D. C. (2010). Essenials of Economerics. 4h ed. New York`: McGraw Hill Inernaional. Henrekson, M., & Johansson, D. (2010). Gazelles as job creaors: a survey and inerpreaion of he evidence. Small Business Economics, 35(2), Hussinger, K. & Pacher, S. (2015). Informaion Ambiguiy and Firm Value. Applied Economics Leers, 22, Máñez, J. A., Rochina-Barrachina, M. E., Sanchis-Llopis, A., & Sanchis-Llopis, J. A. (2015). The deerminans of R&D persisence in SMEs. Small Business Economics, 44(3), Musceola, M. (2015). Difficulies for small firms o inves in research prerogaives. An empirical analysis of a sample of Ialian firms. Applied Economics, 47(15),