Effectiveness of Business Innovation and R&D in Emerging Economies: The Evidence from Panel Data Analysis

Transcription

1 Journal of Economics, Business and Managemen, Vol. 3, No. 4, April 2015 Effeciveness of Business Innovaion and R&D in Emerging Economies: The Evidence from Panel Daa Analsis M. Marin Absrac The problems of managemen and opimizaion of business innovaive effor seem o be of vial imporance in he conemporar economics, especiall in he period of global economic slowdown. This paper is he coninuaion of research work aiming o assess, on he basis of esimaion of ransformaion of Cobb-Douglas producion funcion he effeciveness of various pes of business innovaion expendiures of manufacuring enerprises locaed in Poland. The surve is based on daa obained from public saisics (The Cenral Saisical Office). The economeric esimaions presened in he paper are based on he oal sample of maximum 191 firms permanenl (ever ear in he ime series) acive in he field of R&D over he observed 10 ear period beween 2000 and The esimaions are based on he relaive EBITD growh as dependen variable. Independen variables including various caegories of innovaion expendiure plus non innovaive capial expendiure relaive o sales and relaive emplomen growh were lagged from 0 o 6 ears. The sud revealed srong variaion in erms of effeciveness of various caegories of innovaive expendiure and relaivel srong and consisenl posiive lagged random effecs (RE) of boh inernal and exernal R&D expendiure. This, ogeher wih oher findings, migh sugges ha posiive effecs of business R&D are mosl associaed wih specific ime-invarian individual characerisics of business unis. Index Terms Business innovaion, business R&D, effeciveness, emerging economies, fixed effecs, random effecs. I. INTRODUCTION The effecs of innovaion a he macroeconomic level are in general unquesionable in oda s world. In he process of classical, from he curren poin of view research, i has been esimaed ha innovaion, and especiall he commercial applicaions of science and echnological developmens, accoun for up o 75% of economic growh. Social wealh is deermined b echnical progress in up o 90% [1]. These findings are consisen wih more recen sudies and he economic heor [2], [3] ha indicae echnical change as a major source of long-erm producivi growh. The overall rae of reurn o R&D is quie impressive; i is esimaed a 25% for privae reurns and a a oal of 65% in erms of overall social reurns [4]. On he oher hand he issue of ransferring business innovaion expendiure ino sales growh and produc developmen is b no means sraighforward a he microeconomic level. The oupu of business innovaion and R&D are ofen unclear, indirec, and difficul o measure. This problem is even more imporan in he case of emerging and ransiion economies, since research evidence and conclusions based on research work done in developed economies migh no be easil ransferable due o deferen sages of developmen of he wo pes of economies. Innovaion is generall considered as a ool for improving he compeiiveness of firms and heir performance, neverheless his relaionship has no been suppored full b empirical work. This paper is a coninuaion of previous research work ha was based on non-panel daa analsis and economeric esimaions [5], [6] regarding he issue of effeciveness of business innovaion and R&D in he case of emerging and ransiion economies. The general purpose of his research is wofold. In he firs place i is o exen, on he basis of esimaions of panel models, he previous research regarding he issue of effeciveness of business innovaion aciviies via implemenaion of differen mehods and economeric ools applied o exploraion of he issue of fixed effecs and random effecs of innovaive and R&D business expendiure. Secondl, i is o observe he differences beween resuls of previous classical leas square (CLS) esimaions based on he relaive sales and EBITD growh and he presen ones based on esimaion of various panel models. Previous research based on classical (non-panel) esimaions revealed ha relaive EBITD growh as he dependen variable is o he smaller exen explained b various caegories of innovaion expendiure in comparison o he models based on relaive sales growh. The sud showed posiive and consan, regardless of R&D regulari, impac of business R&D effor over relaive EBITD growh. The main aim of he curren research is o examine he issue of effeciveness of business innovaion and R&D effor on he basis of panel mehods of esimaion of regression funcions wih he disincion o fixed effecs (FE) and random effecs (RE). Manuscrip received November 10, 2013; revised Januar 9, This work was suppored b Naional Science Cenre in Poland under Gran number N N wih he funds allocaed o science in he ears as he research projec. Auhor would like o express his graiude for funding and suppor of his research work. M. Marin is wih he Lodz Universi of Technolog, Division of Economics, Wolczanska 215, Lodz, Poland ( marin@p.lodz.pl). II. LITERATURE REVIEW The impac of innovaion on firm s performance has been a maer of significan ineres o economiss and polic makers for decades. I is generall recognized ha effecive and successful DOI: /JOEBM.2015.V

2 Journal of Economics, Business and Managemen, Vol. 3, No. 4, April 2015 innovaion is difficul o accomplish and relaivel rare. The sud [7] relaed o public companies represening around 60% of global R&D expendiures proved ha above cerain minimal level here is in fac no correlaion beween R&D spending and business performance exemplified b i.e. sales and profi. Auhor argues ha i is in fac hard o alk abou effeciveness of innovaion in erms of, so o sa, correlaion or coexisence. I is raher hi-or-miss siuaion, ha srongl deermines he effeciveness issues and dilemmas. The ke issue regarding he effeciveness is herefore no in he firs place how much o spend bu how o spend. The acual innovaion spending of individual compan should be based upon he individual innovaion effeciveness curve ha helps companies o assess he innovaion spending agains he financial reurns from innovaion expendiure. In a recen sud [8] regarding R&D expendiure and profiabili in he pharmaceuical indusr in he US, focusing on relaionships beween size adjused R&D budges and sandardized excess abnormal reurns, i was proved ha he marke does no reac significanl o informaion on innovaions or innovaive aciviies in he pharmaceuical indusr a he same ime here is no relaionships beween excess abnormal reurns and annual R&D budges. In a sud dedicaed o relaions beween innovaion invesmen and performance in indusrial firms [9], auhors sress imporance of angible invesmens and uni labour cos for achieving high producivi levels. R&D expendiure and inangible capial are complemenar variables of earlier menioned aspecs, ha have delaed effecs in erms of producivi and profiabili. The acual exisence of subsanial economies of scale and specific capial or labor inensi in he paricular secor of indusr have variable, according o he size of he prior, effec on relaions beween innovaion invesmen and performance of indusrial firms. Differences in he profiabili of business R&D are ofen aribued o he firm s echnolog conex. Wihin various aspecs of echnolog conex i has been found ha complexi, generali and aciness of a firm s echnolog conex, deermined b indusrial resources, influence he rae of reurn o invesmen in R&D [10]. Sud relaed o he issue of produc innovaion sraegies and performance of new high-ech firms in China [11] proved ha he innovaion effeciveness is o large exen deermined b environmenal facors and insiuional suppor. Anoher sud [12] found ha here is a posiive relaionship beween innovaion aciviies and producivi. In making decisions regarding innovaion acivies firms rel on he knowledge accumulaed from previousl abandoned innovaions and cooperaion wih oher firms and insiuions and oher members of heir group. Resuls of he sud reveal several differences in behavior of firms in wo groups of counries. Wesern Europe and advanced ransiion economies from Cenral and Easern Europe. According o reference [13] he acquisiion of exernal R&D or cooperaion wih exernal parners in R&D projecs exers a posiive impac upon he profiabili of firms. Inernal R&D onl does no significanl impac he profiabili of firms. Exernal innovaion sraegies are, herefore, likel o generae profi gains a he expense of a significanl high risk. R&D (inpu) and innovaion (oupu) are highl persisen a he firm level. Among exernal/environmenal facors, marke dnamism affecs R&D and innovaion. Pas innovaive behavior is clearl more decisive in explaining he curren sae of R&D and innovaion aciviies han exernal facors or firm-level heerogenei [14]. On op of his firms ha have R&D ha is more producive, exhibi higher and less volaile fuure operaing performance [15]. One conribuion of his sud is ha i demonsraes ha he relaion beween R&D expense (i.e., inpus) and fuure operaing performance is beer undersood b incorporaing informaion abou he producivi (i.e., oupus) of a firm s R&D oulas in he form of paen couns and ciaions. Inernal research capabiliies, paricularl hose wih a srong basic research componen, are essenial o enabling a firm o generae creaive oupus [16]. Research also found evidence supporing he hpohesis of increasing reurns o scale o R&D spending which is consisen wih economic argumens for he advanages of scale in innovaion. Profiing from innovaion requires srong complemenar capabiliies beween R&D and oher business aciviies wihin he firm, i.e. markeing and manufacuring [17]. Reference [18] emphasizes he imporance of so called behind innovaion. In ha respec effeciveness of commercializaion sraeg is viall imporan. The cos of innovaion mus include creaion of novel, developmen and commercializaion. The overall effec of innovaion process depends on proper produc posiioning, well-ailored pricing polic along produc life ccle. Therefore idenificaion of effecive pahs for maximizing reurn on business innovaion seems o be relaivel imporan issue. On op of heir impac on he firm s marke value, innovaion and R&D ma have an influence on he firm s financial performance in erms of income and sales growh. Anoher research indicaes he relaionships R&D and firm s financial performance. The firms which have higher R&D spending obain he higher average sales growh han he marke average [19]. Reference [20] suggess imporan differences in erms of innovaion-relaed facors deermining he producivi growh of European companies in general (represened in he surve b Ausrian, French, Duch, and Briish firms) and of Ialian companies in paricular. This ma sugges imporan variaion regarding facors deermining he effecs of innovaive aciviies even among advanced economies represening a similar level of economic developmen. The problem of wo wa relaionships and effeciveness of inernal R&D expendiures and exernal echnolog sourcing, boh in he form of business R&D conracing from exernal sources and R&D cooperaion was examined on a basis of a sample of Flemish R&D acive companies [21]. The research was based on he basic principle ha availabili of exernal echnolog ma limi invesmen in he in house research and he issue of complemenari. Auhor found ha exernal echnolog sourcing have a significan posiive effec on inernal R&D onl in he siuaion when companies (echnolog recipiens) have sufficien absorpive capaci in he form of a full-ime saffed R&D deparmen. A he same ime, firms are found o be more frequenl engaged in R&D cooperaion, he more he spend on inernal R&D. 441

3 Journal of Economics, Business and Managemen, Vol. 3, No. 4, April 2015 The sud relaed o he issue of decisions beween inernal and exernal R&D [22] proved ha i is efficien for companies o engage in boh pes of R&D. According o auhors inernal R&D enables he firm o ranslae exernal knowledge ino innovaion opporuniies for he firm. Exernal R&D faciliaes spillovers from he ouside informaion nework o he firm s specific knowledge sock. In order for he firm o engage in boh pes of R&D aciviies he need for a criical mass exiss (in erms of he amoun of firms R&D effor or in erms of he quali of he firm s exernal environmen providing sufficien echnological opporuniies). Auhors also found ha in high-ech he complemenar effec beween inernal and exernal R&D akes place whereas in low-ech indusries inernal and exernal R&D end o be subsiues. Among various facors deermining effeciveness of business innovaion including reurn on invesmen-like measures, innovaion managemen seems o be of paramoun imporance. Alhough i has been found ha he overall quali of innovaion managemen has weakened recenl, good innovaion managemen ma increase firms profiabili (expressed b EBIT) b up o 13% in comparisons wih average performers [23]. III. THE METHODOLOGY AND MODEL The surve presened in his paper, regarding he panel daa analsis he effeciveness of business innovaive aciviies wih special emphasis on he issue of effeciveness of business R&D aciviies in he case of emerging economies, is based on he esimaion of regression funcion ha is based on he ransformaion of Cobb-Douglas producion funcion. The dependen variable in his case was defined as he relaive growh of earnings before ineres, ax and depreciaion (EBITD), ha is considered as he mos suiable measure of profiabili of innovaive aciviies a he firm level, since i akes under consideraion onl shor erm coss and revenues ha reflec he core of business and innovaive operaions a he firm level. The daa uilized in he esimaion of he regression funcions is obained from public saisics (The Cenral Saisical Office) and represens he ime series of innovaion and R&D expendiures and oupu measures exemplified b EBIDT and is relaive growh. The economeric esimaions are based on he oal sample of 191 firms permanenl (in ever ear of he ime series) acive in he field of R&D over he observed 10 ear period. The subsample of firms represening he highes regulari in erms of R&D effor was aken under consideraion since he previous esimaions carried ou b auhor on he basis of subsample of firms represening ver high regulari in erms of business R&D aciviies gave promising resuls. Anoher raionale for his approach is he paper limiaions. The daa cover he period beween ear 2000 and Esimaions include he 0 o 6 ears lag of independen variables which resuls wih he maximum of 764 observaions for he sample of firms aken under consideraion in esimaions presened in he paper. The firms covered b he surve represen medium and large (b EU sandards) manufacuring companies locaed in all pars of Poland. Onl medium and large companies, ha emplo 50+ persons are aken ino consideraion in he sud, since smaller (emploing less han 50 persons) are no covered b he earl surve of innovaive aciviies (PNT-02 surve) carried ou b The Cenral Saisical Office in Poland. The iniial version of he regression funcion is specified underneah. * EBITD EBITD * EBITD 1 R & DIn 2 IETech 5 IEMDom 8 IEMark 11 * 1 R & DEx 3 ICESof 6 IEM Im p 9 IE Rem 12 EMP EMP 0 1 EMP IEBuild 7 1 NIE 4 1 IETrai 10 Descripion of variables: EBITD Earnings before ineres ax and depreciaion EMP oal emplomen R&DIn inernal research and developmen expendiure R&DEx exernal research and developmen expendiure NIE non-innovaion capial expendiure IETech innovaion expendiure on new echnologies IESof innovaion expendiure on sofware IEBuild innovaion expendiure on buildings (associaed wih innovaive aciviies or invesmen) IEMDom innovaion expendiure on domesicall made machiner IEMImp innovaion expendiure on impored machiner IETrai innovaion expendiure on raining (associaed wih innovaive aciviies or invesmen) IEMark innovaion expendiure on markeing IERem remaining innovaion expendiure On he basis of available daa and common sandards, business R&D expendiure is divided ino wo broad caegories: Inernal business research and developmen: Aciviies carried ou b a paricular business eni regardless of he sources of funds uilized o finance hem. Inernal R&D expendiure includes boh running coss and capial expendiure. Exernal business research and developmen: Includes R&D aciviies performed ouside of a paricular business eni b boh domesic and foreign conracors. The issue of inerpreaion of he accrual meaning of fixed effecs and variable effecs of panel regression models requires careful insigh and aenion. Fixed effecs are being used when one is onl ineresed in analzing he impac of variables ha var over he ime. FE help o search for he relaionship beween predicor and oucome variables wihin an eni (counr, person, compan, ec.). Each eni has is own individual characerisics ha ma or ma no influence he predicor variables (for example being a male or female could influence he opinion oward cerain issue or he poliical ssem of a paricular counr could have some effec on rade or GDP or he business pracices of a compan (1) 442

4 Journal of Economics, Business and Managemen, Vol. 3, No. 4, April 2015 ma influence is sock price). When using FE we assume ha somehing wihin he individual ma impac or bias he predicor or oucome variables and we need o conrol for his. This is he raionale behind he assumpion of he correlaion beween eni s error erm and predicor variables. FE remove he effec of hose ime-invarian characerisics from he predicor variables so we can assess he predicors ne effec. Anoher imporan assumpion of he FE model is ha hose ime-invarian characerisics are unique o he individual and should no be correlaed wih oher individual characerisics. Each eni is differen herefore he eni s error erm and he consan (which capures individual characerisics) should no be correlaed wih he ohers. If he error erms are correlaed hen FE is no suiable since inferences ma no be correc and ou need o model ha relaionship (probabl using random-effecs), his is he main raionale for he Hausman es [24]. IV. THE FINDINGS TABLE I: CLASSICAL LEAST SQUARE ESTIMATION (CLS) OF PANEL DATA, 764 OBSERVATIONS, MAXIMUM LAG APPLIED 6 YEARS, DEPENDENT VARIABLE RELATIVE EBITD GROWTH Value of coefficien Sandard error saisics p level cons -0, , ,0410 0,96733 R&D In no lag 20, ,3229 1,9995 0,04591 ** NIE no lag -4, ,2334-2,0813 0,03775 ** R&D In lagged 1-41, ,8173-3,2627 0,00115 *** ear NIE lagged 1 ear 7, , ,0141 0,00266 *** R&D In lagged 3 ears R&D Ex lagged 5 ears IES of lagged 5 ears R&D Ex lagged 6 ears 16,6591 9, ,7616 0,07854 * -33, ,7688-1,6700 0,09533 * -174,351 90,6367-1,9236 0,05478 * 29, ,4948 1,7779 0,07581 * ***, **, and * indicae significance a he 1, 5 and 10 percen levels respecivel. Dependen variable average -0, Sandard deviaion 4, of dependen variable Sum of squares of 16213,51 Sandard error of 4, Deerminaion coefficien R2 0, Adjused R2 0, F (Fisher-Snedecor) saisic (8, 755) Log likelihood raio Schwarz-Baesian informaion crierion Auocorrelaion of - rho1 3, p value for F es 0, ,091 Akaika informaion crierion 4561,930 Hannan-Quinn informaion crierion 0, Durbina-Wasona Sa. 4520, ,255 1, Esimaions of regression funcion based on panel model were carried ou on he basis of daa obained from 191 firms observed over 10 ears period of ime. Firms included in he research sample repored R&D expendiure in ever ear of he ime series. In order o examine he more delaed effecs, he 0 o 6 ear lag beween dependen and independen variable was applied. In he firs place he panel daa esimaions based on classical leas square (CLS) esimaion mehod were execued, hen subjec o he resuls of seleced saisical ess he generalized leas square (GLS) esimaion mehod was aken advanage of. The resuls of iniial esimaion of he panel model based on he CLS mehod afer he removal of insignifican variables are presened in he Table I. On he basis of Breusch-Pagan es (LM = 0, wih p = prob(chi-square(1) > 0, ) = 0,986564) here is no reason o rejec he H0 hpohesis ha esimaions based on CLS mehod are correc. The same conclusion was formulaed on he basis of F es (F(190, 565) = 1,02214 wih p = 0,418982). The esimaions presened in Table I show mixed resuls. Five ou of eigh saisicall significan variables ale sricl relaed o business research and developmen aciviies (inernal or exernal). This migh sugges ha he overall impac of business R&D over he relaive EBIDT growh is ver srong, if no predominan among oher caegories of innovaive expendiures. The model showed ha onl hree oher han business R&D caegories of expendiures - independen variables exer saisicall significan impac over dependen variable: non innovaive expendiure lagged 0 ears and lagged one ear, ogeher wih innovaive expendiure on sofware lagged five ears. In regards o he par of independen variables one can observe cerain regulari, which means ha posiive effecs end o occur as he lag applied increases, his is clearl he fac in he case of exernal business R&D, non-innovaive expendiure and o some exen in he case of inernal business R&D. The model shows ha one can expec less delaed effecs of inernal business R&D expendiure in comparison wih he resuls of exernal business expendiure. Since companies spend less on inernal business R&D ha on exernal R&D, one basic raionale behind his finding migh be ha simpl supposedl larger and more formalized exernal R&D underakings are associaed wih longer period of ime o execue R&D conracs in comparison wih smaller inernal R&D projec ha migh ake less ime o execue and obain visible (posiive or negaive resuls), bu his is onl an assumpion since no daa is available o verif his hpohesis. This issue is supposed o be deepened and verified in he fuure research. The resuls of panel daa esimaions based on classical leas square (CLS) mehod on carried ou on a sample of 191 companies gave ineresing bu o cerain exen inconsisen and difficul o inerpre resuls, herefore second aemp was made o perform esimaions on he basis of narrower sample of firms. The iniial sample of 191 firms acive in he field of business R&D was narrowed down o he sample of 90 firm ha me he following crieria relaive growh of EBITD <100%. An firm ha a leas once in he ime series showed greaer han 100% relaive EBITD growh was excluded from he final sample. The raionale behind his aemp was o avoid disconinuiies and possible daa errors regarding dependen variable, since large porion of firms showed unexpeced and hard o explain and consider credible values of relaive EBITD growh. Despie he issue of daa credibili, greaer han 100% variaion of profi relaed 443

5 Journal of Economics, Business and Managemen, Vol. 3, No. 4, April 2015 dependen variable migh sugges significan disconinuiies and oher exraordinar facors ha cause complicaions o he esimaion process and make final conclusions hard o formulae. The panel daa esimaions on he basis of narrowed sample of 90 firms, based on he classical leas square (CLS) mehod were rejeced on he basis of Breusch-Pagan es (LM = 4,20562 wih p = prob(chi-square(1) > 4,20562) = 0, ). As he resul of his generalized leas square (GLS) esimaion mehod was applied and esimaions of fixed effecs and random effecs models were carried ou. The esimaions of fixed effecs and random effecs based on generalized leas square (GLS) mehod on he basis of Hausman Tes proved ha random effecs (RE) model is more efficien (H = 31,0965 wih p = prob(chi-square(34) > 31,0965) = 0,610686). The F es for variaion for consan proved ha here is no re reason o rejec he H0 hpohesis ha groups have common consan (F(89, 236) = 0, wih p = P(F(89, 236) > 0,739935) = 0,949746). Afer rejecion of insignifican variables he final version of he fixed effecs model is presened in he Table II. TABLE II: GENERALIZED LEAST SQUARE ESTIMATION (GLS) OF PANEL DATA, FIXED EFFECTS (FE), 360 OBSERVATIONS, 90 COMPANIES, MAXIMUM LAG APPLIED 6 YEARS, DEPENDENT VARIABLE RELATIVE EBITD GROWTH. Value of Sandard T saisics p level coefficien error cons 0, , ,2788 0,00003 *** IEM Imp no -1, , ,7219 0,00693 *** lag IE Mark no lag -0, , ,1446 0,03291 ** NIE lagged 3 ears IE Sof lagged 3 ears IE Build lagged 5 ears NIE lagged 6 ears -2, , ,9519 0,00010 *** -16,0759 6,7959-2,3655 0,01873 ** -2, , ,0696 0,03947 ** -1, , ,6920 0,00756 *** ***, **, and * indicae significance a he 1, 5 and 10 percen levels respecivel. Dependen variable average Sum of squares of Deerminaion coefficien R2 F (Fisher-Snedecor) saisic (97, 262) Log likelihood raio Schwarz-Baesian informaion crierion Auocorrelaion of - rho1-0, Sandard deviaion 0, of dependen variable 38,32865 Sandard error of 0, , Adjused R2 0, , p value for F es 0, ,6347 Akaika informaion 411,2694 crierion 792,1076 Hannan-Quinn 562,6979 informaion crierion -0, Durbina-Wasona 1, Sa. The resuls of esimaion of fixed effecs model (included as a reference) show negaive values/effecs of all saisicall significan parameers. This iself is an ineresing finding since i suggess ha all (if an) posiive effec of innovaive expendiure migh come from oher - individual, mosl ime invarian individual characerisics of companies in he sample. The removal he effec of hose ime-invarian characerisics from he predicor variables in he case of FE esimaion, allows o assess he predicors ne effec which in his case seems o be negaive. One can argue ha he expendiure iself is in a sense a prerequisie, and he acual efficienc of innovaive expendiure is deermined no b innovaive expendiure iself bu b oher unique characerisics of he given compan. The ne effec of innovaive expendiure iself (separaed from hose unique individual characerisic is (in he ligh of he above) negaive since i onl increases coss and herefore reduces EBITD relaive growh. The b far greaes value of he coefficien esimaed for innovaive expendiure on sofware lagged hree ears (equals -16, 0759) is iself an ineresing (and hard o inerpre) finding since i ma sugges ha he posiive effec of innovaive expendiure on sofware heavil depends in unique ime invarian characerisics of individual firms and has o be especiall carefull managed and reflec hese specific characerisics in order o ensure posiive reurns. This migh also be he case as far as oher saisicall significan parameers are concerned bu o far smaller exen. The smalles value of esimaed coefficien was esimaed in case of innovaive expendiure on markeing (he value of he coefficien equals -0, ). This migh sugges ha he effec of his caegor of innovaive expendiure is o he smalles exen suscepible o he unique ime invarian characerisics of he firm. The posiive effec migh be ensured o far greaer exen b he expendiure iself. The resuls of GLS esimaions of regression funcion presened in he Table III prove in he firs place consisen and posiive impac of business R&D expendiure (boh inernal and exernal) on relaive EBIDT growh. Srong posiive and saisicall significan hree ears lagged impac (value of esimaed coefficien = 11, 3) of exernal business R&D expendiure was observed. Subsequenl on op of his relaivel srong and posiive five ears lagged impac of boh inernal (value of esimaed coefficien = 3, 2) and exernal (value of esimaed coefficien = 3, 3) business R&D was idenified. All he oher saisicall significan independen variables presened in Table III have eiher negaive or mixed (boh posiive and negaive subjec o lagged applied) impac over dependen variable. In he ligh if he research evidence of random effecs regression model specified in Table III business research and developmen expendiure has he sronges posiive impac over relaive EBIDT growh among all he oher componens of innovaion expendiure specified in he model. On op of his he research evidence sugges ha exernal business R&D migh be more effecive and bring less delaed effecs in comparison o he inernal business R&D. This finding confirms he resuls of previous esimaions and furher suppors he assumpion ha exernal business R&D represens more effecive opion ha inernal business R&D in he case of emerging and ransiion economies. Exernal performers of business R&D migh have greaer experience and usuall well-defined conraced exernal business R&D ma provide noiceabl higher probabili of success and sronger suppors profi growh. In he case of some oher independen variables (innovaive expendiure on new echnologies and remaining innovaive expendiures) one can observe he buildup of posiive reurns wih ime, as he 444

6 Journal of Economics, Business and Managemen, Vol. 3, No. 4, April 2015 lagged increases. One should noe ha in he case of innovaive expendiure on sofware he opposie rend was observed. TABLE III: GENERALIZED LEAST SQUARE ESTIMATION (GLS) OF PANEL DATA, RANDOM EFFECTS (RE), 360 OBSERVATIONS, MAXIMUM LAG APPLIED 6 YEARS, DEPENDENT VARIABLE RELATIVE EBITD GROWTH. Value of coefficien Sandard error saisics p level cons 0, , ,9878 0,00008 *** IEMImp no lag -1, , ,6922 0,00747 *** IEMark no lag -0, , ,0703 0,03921 ** IESof lagged one ear 11,722 6, ,7767 0,07655 * IETrain lagged wo ears -97, ,6835-2,5302 0,01187 ** IERem lagged 2 ears 14,0002 4,5752 3,0600 0,00240 *** R&DEx lagged 3 ears 11,3588 4, ,3954 0,01717 ** NIE lagged 3 ears -1, , ,5700 0,00041 *** IESof lagged 3 ears -12,5258 4, ,5092 0,01259 ** IEMark lagged 3 ears -5,0372 2, ,8763 0,06151 * IERem lagged 3 ears -10,0296 3, ,2676 0,00120 *** IEBuild lagged 4 ears -2,1834 1, ,8906 0,05957 * IERem lagged 4 ears 5, , ,7325 0,00663 *** R&DIn lagged 5 ears 3, , ,3880 0,00079 *** R&DEx lagged 5 ears 3, , ,6621 0,09746 * IETech lagged 5 ears -7,2142 3, ,8351 0,06741 * IEBuild lagged 5 ears -2, , ,6262 0,00904 *** NIE lagged 6 ears -0, , ,1634 0,03124 ** IETech lagged 6 ears 8, , ,5438 0,01143 ** IERem lagged 6 ears -3,942 1, ,5317 0,01182 ** ***, **, and * indicae significance a he 1, 5 and 10 percen levels respecivel. Dependen variable average Sum of squares of Log likelihood raio Schwarz-Baesian informaion crierion -0, Sandard deviaion of dependen variable 41,67047 Sandard error of -122,6818 Akaika informaion crierion 451,3773 Hannan-Quinn informaion crierion V. CONCLUSION 0, , , ,4453 The resuls of panel daa esimaions based boh on classical leas square and generalized leas square mehod revealed srong variaion of he impac of various caegories of innovaive expendiure aken under consideraion in he iniial version of he model and high significance associaed wih mosl posiive impac of boh inernal and exernal business R&D effor over relaive EBITD growh. The posiive impac was found mainl in he case of random effecs. The analsis of random effecs of business R&D acivaes indicae significan, posiive and srong o moderae impac of his caegor of innovaive effor. Random effec daa analsis revealed ha he resuls of exernal R&D are less lagged (hree) and sronger han he effecs of inernal business R&D (five ears lag). These conclusions formulaed on he basis of he narrowed subsample and generalized leas square esimaion mehod of panel daa analsis are in ha paricular respec opposie o he resuls of esimaions based on classical leas square mehod presened earlier in he paper. I seems o be worhwhile o furher invesigae his issue and find ou if here is in fac cerain regulari in erns he iming of effecs of inernal and exernal business R&D, or such regulari in fac does no exis. The more in deph invesigaion of he complexi and rue facors deermining he effeciveness of business innovaive effor wih he special emphasis on business R&D effor is vial for he overall developmen of he compeiiveness of business unis and via opimizaion of innovaion and R&D effor migh encourage companies o on one hand spend more on innovaion and on he oher gain healhier reurns for boh privae and social good. ACKNOWLEDGMENT Auhor would like o express special hanks o Prof. Jan Jacek Szaudnger from Facul of Economics and Sociolog of Universi of Lodz and Prof. Marek Szaj from Facul of Managemen of Czesochowa Universi of Technolog for heir suppor in he field of model specificaion and refinemen. Wihou his suppor and suggesions he work ha has led o his and oher research papers would have been impossible. Auhor would like also o hank he Lodz Saisical Office for he suppor and provision of daa regarding he inpu and oupu measures of manufacuring enerprises and especiall Dr Arur Mikulec for his imporan suppor in he field of daa handling and processing. REFERENCES [1] E. F. Denison, The sources of economic growh in he unied saes and he alernaives before US, Commiee for Economic Developmen, New York, [2] R. M. Sollow, Technical change and he aggregae producion funcion, Review of Economics and Saisics, vol. 39, [3] P. M. Romer, Endogenous echnical change, Journal of Poliical Econom, vol. 98, [4] L Sveikauskas, R&D and producivi growh: A review of he lieraure, US Bureau of Labor Saisics, Working Paper, no. 408, [5] M. Marin, The effeciveness of business innovaive aciviies in he case of emerging economies; he exploraion of relaive sales growh model, Inernaional Journal of Applied Social Science Research, vol. 2, [6] M. Marin, The Effeciveness of Business Innovaion Aciviies in he Case of Emerging Economies; in Search for he Lagged Effecs, Inernaional Journal of Business, Managemen and Social Sciences, vol. II, issue 10(II), [7] A. Kandbin, Which innovaion effors will pa, MIT Sloan Managemen Review, vol. 51, issue 1, [8] P. S. Koku, R&D expendiure and profiabili in he pharmaceuical indusr in he unied saes, Journal of Applied Managemen Accouning Research, vol. 8, issue 1, [9] A. M. Bobillo, J. A. R Sanz, and F. T Gaie, Innovaion invesmen, compeiiveness and performance in indusrial firms, Thunderbird Inernaional Business Review, vol. 48, issue 6, [10] S. K. Evil and B. Chakravarh, Resource conex and he reurns o invesmens in R&D, Academ of Managemen Proceedings and Membership Direcor, [11] H. Li and K. A. Gima, Produc innovaion sraeg and he performance of he new echnolog venures in china, Academ of Managemen Journal, vol. 44, issue 6, [12] I. Hashia and N. Sojčićc, The impac of innovaion aciviies on firm rerformance using a muli-sage model: evidence from he communi innovaion surve 4, Research Polic, vol. 42, issue 2, [13] J. Maaa and M. Woerer, Risk innovaion: The impac of inernal and exernal R&D sraegies upon he disribuion of reurns, Research Polic, vol. 42, issue 2,

7 Journal of Economics, Business and Managemen, Vol. 3, No. 4, April 2015 [14] A. Triguero and D. Corcoles, Undersanding innovaion: An analsis of persisence for Spanish manufacuring firms, Research Polic, vol. 42, issue 2, [15] S. Pandi, C. E. Wasle, and T. Zach, The effec of research and developmen (R&D) inpus and oupus on he relaion beween he uncerain of fuure operaing performance and R&D expendiures, Journal of Accouning, Audiing and Finance, [16] K. W. Arz, P. M. Norman, D. E. Hafield, and L. B. Cardinal, Longiudinal sud of he impac of R&D, paens, and produc innovaion on firm performance, Journal of Produc Innovaion Managemen, vol. 27, Issue 5, [17] A. Leiponen, Compeencies, innovaion and profiabili of firms, Economics of Innovaion and New Technolog, vol. 9, issue 1, [18] M. P. Drake, N. Sakkab, and R. Jonash, Maximizing reurn on innovaion invesmen: Spending more on innovaion does no necessaril ranslae ino acceleraing sales, marke share or profi, Research-Technolog Managemen, Nov.-Dec., [19] M. S. Brenner and B. M. Rushon, Sales growh and R&D in he chemical indusr, Research-Technolog Managemen, vol. 32, no. 2, [20] M. Piana and A. Vaona, Innovaion and producivi in european indusries, Economics of Innovaion and New Technolog, vol. 16, no. 7, [21] R. Veugelers, Inernal R&D expendiures and exernal echnolog sourcing, Research Polic, vol. 26, no. 3, [22] D. B. Audresch, A. J. Menkveld, and A. R. Thurik, The decisions beween inernall and exernal R&D, Journal of Insiuional and Theoreical Economics, vol. 152, issues 3, [23] A. D. Lile, Good innovaion managemen increases profiabili b 13%, Business Wire (English), Februar, [24] O. T. Rena, Panel daa analsis, fixed & random effecs, Daa and Saisical Services, Princeon Universi. Marek Marin was born in Lodz, Poland on 23 Augus He graduaed in 1992 from Technical Universi of Lodz, Poland. In 2002 he gained Ph.D. in economics a Universi of Lodz, Poland. He sudied as Tempus suden a Universi of Srahclde in Glasgow, UK and Universidade Nova de Lisboa in Lisbon, Porugal. He did research projec a Universi of Cenral England in Birmingham wih conjuncion wih Ason Science Park Birmingham, UK during Tempus funded placemen. The major fields of sud include microeconomic aspec of innovaive aciviies, managemen of echnologies and economic aspecs of developmen of seleced fields of echnolog oriened firms i.e. bioechnolog and energ echnologies. As a suden and shorl afer, he worked as diver and lifeguard. He worked as consulan for several business consuling firms and direcor of Lodz branch of finance and leasing associaion, Wroclaw, Poland. A he presen momen he works as he assisan professor a Division of Economics, Insiue of social science, Facul of organizaion and managemen, Lodz Universi of Technolog, Lodz, Poland. Previous research ineress include problems of developmen and barriers o growh of new echnolog based firms, financing echnolog based venures and academic enrepreneurship. Curren ineress include problems relaed o managemen of innovaion, effeciveness of innovaive and research and developmen aciviies, developmen of bioechnolog companies in emerging markes and seleced economic aspecs of energ issues. Dr. Marin is a member of Polish Economic Socie. He obained Ousanding Research Award from The Insiue for Business and Finance Research, Hilo, USA. He worked as he reviver for inernaional scienific journal and edior of several monographs summarizing research projecs and inernaional conference monographs. 446