# $%&' &!(! $% $ )) ) * )! "

Transcription

1 # $%&'!( $%))*! "

2 INNOVATION AND ECONOMIC GROWTH: WHAT IS THE ACTUAL IMPORTANCE OF R&D? Argenino Pessoa Faculdade de Economia do Poro Rua Dr. Robero Frias Poro, Porugal Absrac This paper deals wih he relaionship beween innovaion and economic growh in he conex of developed world. Afer examining he correlaion beween economic growh and R&D (research and developmen) inensiy, and given ha he impac of R&D on economic growh is mediaed by he rae of growh of echnology, we proceed rying o assess he linkage beween R&D oulays and economic growh, hrough he use of he condiion of free enry ino R&D. Confroned wih daa, our argumenaion shows ha he opimism of he endogenous echnological change models is no confirmed for counries siuaed below he echnological fronier. Nex, based on oher economic and echnological indicaors, a succinc comparison beween he Irish and Swedish cases is made. This comparison reveals he imporance of invesmens no classified as R&D, paricularly he ones ha enhance he exernal compeiiveness of he economy. We conclude ha innovaion policy mus always consider he complexiy of he economic growh process and he oher ways, besides he ones based on formal R&D indicaors, in which echnology has an impac on growh. Keywords: R&D, economic growh, echnological change, innovaion policy. JEL Codes: O30, O32, O33, O38. 1

3 INNOVATION AND ECONOMIC GROWTH: WHAT IS THE ACTUAL IMPORTANCE OF R&D? 1. INTRODUCTION This paper deals wih he conribuion of research and developmen (R&D) o he growh of advanced economies. Boh heoreical and empirical lieraures have shown ha invesmens in R&D are crucial for economic growh. In he heoreical fron, a lo of models (Romer, 1990; Grossman and Helpman, 1991; Aghion and Howi, 1992, o name only he mos quoed) illusrae he funcion of R&D as a growh engine, and demonsrae he reason why governmen mus have a role in achieving an opimum level of R&D. In he empirical fron, several auhors also show he imporance of he R&D reurns. For example, in his survey abou R&D spillovers, Griliches (1992) repored a wide range of esimaes for he social reurn of R&D, wih values ha cluser in he range of 20 o 60 per cen, making R&D a major source of growh, responsible for a leas half of all increases in per capia oupu. Addiionally, Jones and Williams (1998) found ha opimal R&D invesmen is a leas four imes greaer han acual spending. Consequenly, invesmens in R&D seem o have very high social reurns and o be a key componen of economic growh and developmen. Accordingly, hough wih very differen resuls, several governmens have vasly increased heir policy commimen o innovaion wih significan impacs on levels of R&D expendiures of heir counries, relying basically on he impac of science and echnology on economic performance. However, such procedure risks o enlarge R&D inensiy beyond he opimum level since here are some reasons o believe in cumulaiveness in R&D oulays whaever is ype: public or privae (Yoo, 2004). The public suppor o R&D expendiures is raionalized in economic erms by he exisence of posiive exernaliies. These exernaliies can appear in he producion of he final oupu or in he R&D process. Producion exernaliies arise because he developmen of one innovaion has an effec on aggregae labour produciviy beyond is conribuion o he capial sock (i.e. i affecs he Solow residual). When innovaions are embodied, firms enjoy producion exernaliies o he exen ha hey use he new 2

4 goods. In he R&D process wo kinds of exernaliies may be presen. One kind is relaed o he R&D workers and he oher is associaed o he exising sock of economic useful knowledge. As has been argued by several auhors (e.g. Sokey, 1995; Jones and Williams, 2000), alongside posiive exernaliies in he R&D process like sand on he shoulders, here is also he possibiliy of some ohers being negaive ( fishing ou hypohesis, sepping on oes effec, creaive desrucion, ec.), which makes he empirical assessmen of he conribuion of R&D o economic growh very doubful if i relies on he calibraion of parameers ha symbolize such exernaliies. Of course here is always he possibiliy of assessing he dimension of exernaliies hrough regression mehods, like in Pessoa (2005), bu his mehod has also problems (spurious regressions, endogeneiy, ec.). So, he presen paper follows a differen pah: he use of he free enry condiion ino R&D o assess he link beween R&D oulays and economic growh. The purpose of presen paper is wofold. On he one hand, we aim o show ha he increase of R&D inensiy is no an assured means of improving economic growh, paricularly in counries siuaed below he echnological fronier. On he oher hand, we ry o show ha here are oher ways in which echnology impacs on growh besides he ones based on formal R&D indicaors. So, he remainder of he paper is srucured as follows. Secion 2 deals wih he relaionship beween R&D and economic growh. In secion 3, we analyse one implicaion of he free enry ino R&D, in a model of R&D-driven growh. In Secion 4 we compare he Irish case wih he Swedish one in erms of R&D and economic growh relaionship. Secion 5 concludes. 2. R&D AND GROWTH: WHAT DOES DATA TELL US? Innovaion policy has frequenly relied on a linear model of he impac of science and echnology on economic developmen. The main assumpion underlying his model is ha R&D carried ou by researchers/scieniss leads o a new idea, which becomes a new produc, for which a producion process is developed by indusrial engineers, and for which a markeing plan is hen se up, conducing o is increasing demand in he marke. This linear model is ofen empirically suppored on he posiive correlaion 3

5 beween Gross Domesic Expendiure on Research and Developmen (GERD) and he level of developmen measured by GDP per capia. GERD as a % of GDP y = 1E-05x R 2 = Tur Figure 1. GERD and GDP per capia Kor Gre Fin Jap GDP per capia (curren PPPs) Swe Swi Irl US Nor Source: Based on daa from OECD (2006) In figure 1, he R&D inensiy measured by GERD as a percenage of GDP in is ploed agains he GDP per capia in 2005 for 29 OECD counries 2. Alhough here is a posiive relaionship beween he level of developmen and R&D inensiy he correlaion is far from perfec, indicaing ha here are oher facors affecing he level of he counry s developmen. Bu even if here was a srong correlaion we should keep in mind ha correlaion is no causaion. Furhermore, here are many facors omied in he regression ha affec simulaneously TFP (oal facor produciviy) growh and he incenives o inves in R&D, such as all hose ha enhance disembodied produciviy. All hose facors, like he managerial and organizaional pracices, learning by doing, ec., have a clear effec on TFP and, a he same ime, induce firms o inves in R&D 3. So, R&D oulays seem raher a proxy of he level of developmen han a cause of i. Parly because of his, when we search for he acual relaionship beween R&D 1 Or where daa are no available for 2004, he neares year wih availabiliy of daa. 2 All OECD members excep Luxembourg. 3 There is some supporive evidence of he poenial imporance of his bias. For insance, Jones and Williams (1998) verify ha he effec of R&D on TFP growh almos vanishes afer including fixed effecs in heir economeric model. 4

6 inensiy and GDP growh, he picure is no very confirmaive of he posiive effecs of he former on he laer. In fac, as depiced in figure 2, where every poin idenifies one of 29 OECD counries, each being represened by he average annual volume percenage change of GDP and by he GERD inensiy (boh measured as deviaions from he OECD average), here is no posiive correlaion beween average growh rae of GDP beween 1995 and 2005 and GERD as a percen of GDP in GDP growh ( ) Source: Based on daa from OECD (2006). Noes: The raes depiced are deviaions from OECD average. Figure 2. R&D inensiy and GDP growh y = x + 8E-16 Irl R 2 = Kor Ice 0 Can US Fin Swe 3 GERD as a per cen of GDP As i is apparen from he figure, among 14 counries ha experienced R&D inensiy above he OECD average, only 5 (he USA, Canada, Finland, Iceland and Souh Korea) show a GDP growh rae higher han he OECD average. The figure also makes he conrasing posiions of Ireland and Sweden apparen. While he former presens he highes rae of growh, gaining in ha way he qualificaion of Celic Tiger, wih low R&D inensiy, he laer illusraes an example where he highes R&D inensiy coexiss wih a rae of oupu growh below he OECD average, originaing he idea of a Swedish Paradox 5. 4 Some may argue ha here mus be a lag beween increase in R&D and economic growh. So, we have also regressed he economic growh rae of period on he R&D inensiy boh for he middle of 1990s and for Bu he resuls, available from he auhor upon reques, are very similar o he ones found for Alhough he paradox idea has had differen versions, all of hem share he common fundamenals ha high Swedish R&D oulays do no produce sufficien economic resuls (see, Kander and Ejermo, 2006). 5

7 The fac ha we don see a posiive correlaion beween R&D inensiy and growh doesn mean per se ha echnology doesn posiively affec economic growh. I only shows ha he link beween he wo variables is complex, wih he efficiency on he use of oulays classified as R&D as a mere aspec of ha complexiy. Moreover, if we add he complemenariy beween R&D and oher inpus (for insance, equipmen, level of schooling of labour force, qualificaion of business managemen) one can suppose ha he absence of correlaion is due raher o hese facors han o he absence of he posiive effecs of echnology. On he oher hand, assessing he echnological effecs hrough indicaors of R&D oulays is no he only possible pah o evaluae he economic impac of he innovaion process. As was already argued (Pessoa e Silva, 2001), paen couns 6, albei wih limiaions 7, consiue he measure of he oupu of research ha beer represens he capaciy for using invenions wih economic purposes. In fac, a paen does correspond o a minimal amoun of invenion ha has passed boh he rial of he invesmen of effor and resources by he invenor and his insiue or firm ino he developmen of his idea, produc or process, and he examinaion of he paen office. Consequen o his examinaion, a paen is only graned if hree condiions are simulaneously fulfilled 8 : i) indusrial applicabiliy he invenion mus be of pracical use; ii) invenive sep he invenion mus no be merely deduced by a person wih average knowledge of he echnical field; iii) novely, ha is, he invenion mus show some new characerisic which is no known in he bulk of exising knowledge in is echnical field. So, he posiive correlaion beween he growh of paen couns, or even beween he number of paens weighed by some indicaor of he counry dimension, and he growh rae of GDP per capia (figure 3) i is much more likely han he correlaion beween R&D inensiy and economic growh. 6 Paen saisics are used wih various funcions. For he funcions of paen couns as economic indicaors, as well as for he difficulies ha arise in heir use and inerpreaion, see Griliches (1990). 7 Among oher limiaions, he mos frequenly cied respec o he fac ha: a) paen couns overlook he real economic imporance of a paricular paen, b) differen echnologies are differenly paenable; and c) because applying o a paen has specific coss, which accrue o he research coss, many firms prefer he secre o he paen as a proecing device. 8 In fac, here is anoher condiion: he subjec of invenion mus be acceped as "paenable" under he counry s law. In many counries, scienific heories, mahemaical mehods, plan or animal varieies, discoveries of naural subsances, commercial mehods, or mehods of medical reamen (raher han medical producs) are no paenable. 6

8 Figure 3. Growh of GDPpc and Growh of US paens (54 counries) Growh of GDPpc, y = 0.218x Source: Based on daa from World Bank (2006) and USPTO (2006). Noes: The raes depiced are deviaions from world average raes. Fin Ir HK Isr Tha Tur Growh of US paens, Ind Mal Sin Chi Kor R 2 = Each one of he poins in Figure 3 idenifies one of 54 counries, each being represened by he deviaion from he world average growh rae of boh GDP pc and number of US paens. As i is apparen from he figure, he counries ha experienced he highes raes of US paens are simulaneously he ones ha show he highes raes of economic growh in per capia erms, for he period. However, in spie of he posiive correlaion expressed in he figure, and in spie of he opimism of he growh models wih endogenous echnology, here are ye many quesions awaiing he righ answer. In he nex secion, we ll use he free enry in R&D o assess he link beween R&D oulays and economic growh. 3. FREE ENTRY INTO R&D In his secion we exploi an aribue of he ideas-driven growh model and calibrae i for he US, Ireland and Sweden. Our main assumpion is ha here is free enry ino R&D. This implies ha, in equilibrium, he value of he resources devoed o R&D equals he value of he newly developed echnologies. 7

9 Le s denoe by A he level of echnology associaed wih R&D invesmens. In he horizonal differeniaion model (Romer, 1990; Grossman and Helpman, 1991, ch. 3), his is he number of capial varieies. To compue he R&D conribuion o produciviy growh, we can use a producion funcion o relae he growh rae of A o he growh rae of labour produciviy. All over he paper we use X & o designae he ime derivaive of variable X, and Xˆ o denoe he growh rae of variable X. So, le equaion (1) represen he Cobb-Douglas producion funcion of he economy: where produciviy, Y = A Z K L (1) σ α 1 α Y denoes he economy-wide oupu, K and capial share, and σ is he elasiciy of exernaliy). Z is he level of disembodied L symbolize capial and labour respecively and where α is he Y wih respec o A (i.e., he producion From equaion (1), we can calculae he elasiciy of produciviy growh wih respec o R&D-driven echnology growh (i.e., σ +α K A ), and he conribuion of R&D o produciviy, ( σ +α K A)Aˆ, bu his calculaion depends on he value of he produciviy exernaliy, which is unknown. As in he presen noe we inend o avoid he calibraion of boh exernaliies in he producion and in he R&D process 9, we alernaively sar by invesigaing he relaionship beween he amoun of resources devoed o R&D, share of resources devoed o R&D. Le R Y, and he growh rae of echnology Â, where R denoes he P A symbolize he marke value of a design paid by a firm o produce a variey of capial 10. The free enry condiion implies ha innovaors make zero profis in equilibrium; as a resul he cos incurred o develop he paens (Pa) is equal o he marke value of he flow of new echnologies ( P A A& ). Pa = P A (2) A & 9 See Pessoa (2005) for references and for some enaive evidence on he esimaes of R&D exernaliies. 10 For he sake of simpliciy, in he remainder of he paper we suppress he subscrip in parameers and variables. 8

10 Which is he condiion of free enry. This condiion can be rewrien as in equaion (3), where we can see ha he relaionship beween he share of resources devoed o R&D and he growh rae of echnology is arbiraed by he value of innovaions. Y A P A = (3) R Because successful innovaors earn a paen from he blueprin, hey can charge a markup (η ) above he marginal cos of producion. Whaever he value of he mark-up, he insananeous profis earned by an innovaory firm, which produces inermediaes, are: A ˆ η 1 Y π = α (4) η A Bu, wha is he marke price of an innovaion? Suppose for simpliciy, as in Romer (1990) s model, ha paens are infiniely lived and ha innovaors are no overaken by new innovaors wih more sophisicaed capial goods. Then he value of an innovaion, P A, mus saisfy he following equaion: where ρ is he relevan discoun facor. ρ P = π + & (5) A P A In seady sae, all variables grow a consan raes. From equaion (3), his implies ha ˆ P A = Yˆ Aˆ (6) Subsiuing he variables of equaion (5) by heir expressions in equaions (6), (4) and (3) and ransferring  o he lef-hand side we ll obain he following expression for he growh rae of echnology in erms of R : ˆ ˆ ρ Y A = η 1α 1 η R (7) There are wo imporan observaions o be made from his expression. Firs,  in expression (7) does no depend direcly on he size of he exernaliies in R&D or on he degree of he diminishing reurns o aggregae R&D invesmens. This implies he 9

11 assumpion ha small innovaors waning o undersand he marke price of heir innovaions ( P A ) do no consider he effec of heir invesmen decisions on aggregae variables, such as he ineres rae or he growh rae of oupu. Given ha he exernaliies appear hrough hese aggregae variables, we do no need o calibrae hem once we conrol for Ŷ and ρ. Second, he quaniaive resul of his secion comes from he fac ha, if Ŷ < ρ,  is increasing in R. The link beween hese wo variables does no come from a producion funcion for echnology; i follows from he posiive relaionship ha he free enry condiion (3) idenifies beween  and R. I is useful o give some values o hese parameers in order o access he growh rae of R&D driven echnology (  ). As i is apparen in able 1, we have esimaed  based on he value of he parameers (η, α, ρ ) and in he acual figures of Ŷ and R. We have chosen he figures for he parameers in he following way: i) for α we have used 1/3, which is he capial share usually applied o he US economy; ii) for he discoun rae, ρ, we have considered 0.07, which is he average real reurn of he sock marke in he las hundred years 11 ; and iii) for he mark-up η we have chosen an inermediae value (1.2) of he empirical esimaes which range from 1.05 o 1.4 (see Norrbin, 1993; and Basu, 1996). For Ŷ and R we have used daa from OECD (2006). Table 1. Parameers and growh rae of R&D driven echnology η α ρ Ŷ R  Ireland 1.2 1/ Sweden 1.2 1/ US 1.2 1/ The values presened in able 1 imply ha  is abou for he US, for Sweden and 0,0014 for Ireland. Bu he ideas growh model (Romer, 1990; Grossman and Helpman, 1991) predics ha in he long run, once aained he seady-sae pah, 11 See Mehra and Presco (1985). 10

12  =Ŷ. So, from his poin of view, while he US esimae is very realisic (from 1995 o 2005 GDP increased a a average growh rae, bu ha rae increased for if all he pos-war period is considered), he sory is very differen for he oher wo counries, where he highes growh rae, Ŷ, is associaed o he lowes Â, and vice versa. Why are here such discrepancies in Ireland and Sweden? While a complee explanaion of he reasons is ouside he scope of his noe, our argumen shows ha if Ŷ > ρ,  decreases wih increases in R. Of course we don know if we are using he correc values for he parameers, bu we can see ha esimaes for  ha mach wih Ŷ will imply ha ρ mus be equal o 0.14 for Ireland and equal o 0.04 for Sweden. Bu i is difficul o find a reason for such a discrepancy in discoun raes if we keep in mind ha he repored long-erm ineres rae is for Ireland and for Sweden (OECD, 2006). Growh of produciviy is no only he resul of echnological change, bu depends, among oher facors, on he efficiency wih indusrial firms and oher organizaions do work (Griliches, 2000). Alhough formal R&D aciviies are an imporan inpu for echnological innovaion, hey are no he only one, and so oher deerminans of echnological innovaion have o be invesigaed as well. On he oher hand, besides echnological innovaion here are oher forms of innovaion: markeing, design and engineering capabiliies, raining and learning (e.g., learning by doing), developmen of new producion faciliies, and organizaional invesmen and change (Dosi, 1988; Kline and Rosenberg, 1986; OECD, 1997; Rosenberg, 1976) are examples of non-r&d aciviies ha are acknowledged o play an imporan role in a firm s innovaion effors and performance. This informal innovaion, ha is innovaion no explicily planned and budgeed and consequenly largely hidden in (aggregae) innovaion daa can be conrased o formal R&D aciviies ha are radiionally considered as a sysemaic and organized aciviy by innovaion or R&D surveys (cf. OECD, 1963, 1997, 2002). The informal innovaive aciviies ha ake place during producion could have a significan impac on produciviy growh (Dosi, 1988; Rosenberg, 1982), alhough he sysemaic evidence abou informal R&D is hard o measure (Rosenberg, 1982:

13 122) and very scarce (Griliches, 2000, p. 88). Even if hese ideas are no new, hey are ofen forgoen wih he pressure for formalizaion. 4. INFORMAL R&D AND THE INTERNATIONAL TECHNOLOGY FLOWS Inermediae goods flow inernaionally: in an increasingly globalised world, new echnologies invened in counry A can be acquired by counry B and used o produce a final good in counry C. This has wo implicaions for he growh models sandard analysis. On he one hand, worldwide R&D invesmens can be more relevan han he domesic invesmens for calibraing he R&D of a given counry. On he oher hand, an innovaor can sell his/her innovaion all over he world and so i is necessary o consider he effec of he enlarged marke in he value of innovaions. These consideraions urn he analysis more difficul, and show how simplisic an assessmen of he R&D effecs is, as was he case wih he one of he previous secion. Table 2. The Celic Tiger vs. he Swedish Paradox. Economic and echnological indicaors Ireland Sweden GDP per capia (curren PPPs), 1999, (OECD=100) GDP per capia (curren PPPs), 2005, (OECD =100) Average growh of GDP, in volume ( ) Expendiure on educaional insiuions (public and privae) in percen of GDP, 1998 Expendiure on educaional insiuions (public and privae) in percen of GDP, 2002 Gross Domesic Expendiure on R&D (GERD), in percen of GDP, GERD in percen of GDP, Business enerprise expendiure on R&D (BERD), as a percen of value added in indusry, 1997 Business enerprise expendiure on R&D (BERD), as a percen of value added in indusry, 2004 Business secor capial/oupu raio, Business secor capial/oupu raio, Expors in percen of impors: High-Tech indusries Medium High-Tech indusries Source: OECD (2000, 2006). 12

14 So, using anoher ype of indicaors alongside wih ones ha measure formal R&D, able 2 gives a more accurae picure of he echnological influence on economic growh and illusraes he reason why some alk in he Celic Tiger and he Swedish Paradox. Alhough in economic erms boh Ireland and Sweden have increased heir weigh in he OECD average, i is difficul o suppor hese gains wih he formal R&D. In fac Sweden presened higher han Ireland R&D inensiy in 1997, boh in GERD and BERD, and from 1977 o 2004 Sweden increased hose figures. Ireland, on he conrary, decreased boh indicaors from 1977 o Similar behaviour was visible in educaion expendiures. However, he apparenly reduced impac of R&D doesn mean ha echnology and innovaion don play an imporan role in he Irish Miracle. The new producs and processes as well as new ways of organizaion generally evolve in a gradual pah and in ways lile or no a all relaed o wha we associae o formal R&D 13. In fac, a shor analysis of he rade conen shows ha, in erms of peneraion in foreign markes, Ireland presens a clear advanage in high and medium high echnology. From where did ha advanage come? From a more efficien use of R&D and educaion expendiures? From posiive exernaliies associaed o a higher dynamism of expors combined wih he inernaional echnology ransfer? From oher facors no covered by saisical indicaors and, for ha reason, omied in quaniaive sudies? While a complee answer o hese quesions is ouside he scope of his paper, because i is impossible o carry i ou wihou a hisorical analysis (Griliches, 2000), for now i is worh noing ha Ireland is a Tiger probably because i succeeded in he ariculaion of several posiive facors associaed no only o he quesions of he previous paragraph bu also o ohers ha only an exhausive invesigaion can uncover. Among hem, are some informal R&D (Griliches, 2000), which are no apparen in saisics, and some aci knowledge in he sense used by Hayek (1945, p. 521): knowledge of paricular circumsances of ime and locaion. 12 This does no mean ha he acual level of R&D has decreased. The high growh rae of Irish GDP allows a decrease in relaive erms wih a sabiliy or even wih a sligh increase in absolue erms. Idenical commens can be made concerning educaion expendiures. 13 An example is suficien o ilusrae his senence: The resources used by Henry Ford o inven mass producion and assembley lines didn correspond a he ime, or even nowadays, o he oficial definiion of R&D. 13

15 5. CONCLUSION Economic progress occurs because we learn new forms of combining capial and labour in a more efficien way. This learning may be eiher he resul of invesmen effors direced owards he improvemen of produciviy or a side-effec of oher aciviies no inenionally focussed on increasing produciviy. In his paper our purpose has been he assessmen of R&D conribuion o produciviy growh. However, he measured impac of R&D o growh is a odds wih he poenial conribuion ha resuls from heoreical lieraure, because eiher heoreical models are poorly formulaed or he concep of R&D is poorly ranslaed in he daa. In his paper we have addressed boh hypohesis. Respecing o he firs hypohesis, aking in accoun ha he effec of increasing he R&D oulays on he growh rae of GDP (he impac of R on Ŷ ) is mediaed by he growh rae of R&D driven echnology ( Â ), we begin by invesigaing he dimension of ( Â ). From he free enry condiion ino R&D and from he fac ha R&D innovaions are embodied, we have obained an esimae for ( Â ) ha is very differen from he record of economic growh boh in Ireland and in Sweden. This finding allows us o conclude ha he link beween measured R&D and economic growh is no so srong as some sudies and policy commimens declare. The failure in finding a close relaionship beween R&D inensiy and economic growh increases he probabiliy ha oher invesmens direced a improving produciviy can be more imporan. For insance, invesmens in organizaion, managemen, financial engineering, and raining, should usually be in larger scale and, because he innovaions ha resul from hem are disembodied, no paenable, and subjec o a quicker diffusion, he exernaliies associaed wih hem are probably much larger and, consequenly, may have more impac on economic growh han he effec of he R&D oulays. If hese commens were associaed o an increasing preoccupaion in he enrance in world marke, where he decisive paern is one of compeiiviy, hen i is probable ha he posiive exernaliies can come in from ouside, wih posiive repercussions in oupu per capia and produciviy. So, he innovaion policy doesn have o rely only on 14

16 measures ha increase he R&D formal indicaors. Furhermore a policy ha provides R&D incenives can have he perverse effec of reclassifying as R&D invesmen he expendiures ha previously had oher designaions. REFERENCES Aghion, P. and P. Howi (1992). A Model of Growh Trough Creaive Desrucion. Economerica, 60(2), Basu, S. (1996). Procyclical Produciviy: Increasing Reurns or Cyclical Uilizaion? Quarerly Journal of Economics 111, Dosi, G. (1988). "Sources, procedures, and microeconomic effecs of innovaion." Journal of Economic Lieraure 26(3): Griliches, Zvi (1990), "Paen Saisics as Economic Indicaors: A Survey", Journal of Economic Lieraure, Vol. 28, pp Griliches, Z. (1992). The Search for R&D Spillovers. Scandinavian Journal of Economics 94, Griliches, Z. (2000). R&D, Educaion and Produciviy: A Rerospecive, Harvard Universiy Press, Cambridge, Mass. Grossman, G. and E. Helpman (1991). Innovaion and Growh in he Global Economy, Cambridge, MA: MIT Press. Hayek, F. A. (1945). The use of knowledge in sociey. American Economic Review, 35(4), Jones, Charles I. and J. C. Williams (1998). Measuring The Social Reurn To R&D, The Quarerly Journal of Economics, 113, Jones, C. I. and J. C. Williams (2000). To Much of a Good Thing? The Economics of Invesmen in R&D, Journal of Economic Growh, 5,

17 Kander, Asrid and Ejermo, Olof (2006) The Swedish Paradox, WP 2006/01. Circle Publicaions, Lund Universiy. Kline, S. J. and N. Rosenberg (1986). An overview of innovaion. In: R. Landau and N. Rosenberg. The Posiive Sum Sraegy: Harnessing Technology for Economic Growh. Washingon, Naional Academy Press: Mehra, R. and E. C. Presco (1985). The Equiy Premium: A Puzzle, Journal of Moneary Economics 15, Norrbin, S.C. (1993). The Relaionship Beween Price and Marginal Cos in U.S. Indusry: A Conradicion, Journal of Poliical Economy 101, OECD (1997). The Measuremen of Scienific and Technological Aciviies: Proposed Guidelines for Collecing and Inerpreing Technological Innovaion Daa (Oslo Manual). Paris: Organisaion for Economic Co-operaion and Developmen. OECD (2002). Proposed Sandard Pracice for Surveys of Research and Experimenal Developmen: Frascai Manual. The Measuremen of Scienific and Technical Aciviies Series. Paris: Organisaion for Economic Co-operaion and Developmen. OECD (2000). OECD in Figures, Paris: Organisaion for Economic Co-operaion and Developmen. OECD (2006). OECD in Figures, Ediion, Paris: Organisaion for Economic Co-operaion and Developmen. Pessoa, A. (2005). Ideas driven growh: he OECD evidence. Poruguese Economic Journal 4, Pessoa, A. and M. R. Silva (2001). "Padrões de I&D: Porugal no conexo da OCDE" in Como Esá a Economia Poruguesa? IV Conferência sobre a Economia Poruguesa, CISEP Cenro de Invesigação sobre a Economia Poruguesa, Lisboa, pp Romer, P. (1990). Endogenous Technological Change. Journal of Poliical Economy 98, S71-S

18 Rosenberg, N. (1976). Perspecives on Technology. Cambridge: Cambridge Universiy Press. Rosenberg, N. (1982). Inside he Black Box: Technology and Economics. Cambridge: Cambridge Universiy Press. Sokey, N.L. (1995). R&D and Economic Growh, Review of Economic Sudies 62, USPTO (2006), Paens by Counry/Sae and year, Unied Saes Paen and Trademark Office, hp:// Word Bank (2006), World Developmen Indicaors, CD-ROM, World Bank. Yoo, S-H. (2004). Public R&D expendiure and privae R&D expendiure: a causaliy analysis, Applied Economics Leers 11,

19 Recen FEP Working Papers! "!# $%& '()))*+,,-. /01%234!"#"$#& & 4 $ '"(("$""$"$")("*" %& $ 5 2 %*06,-(""#3&347#8 %+ 64 &02361,(.(/0(1(9 : % 2303#2 /*"""456(("5"5" %3 &8;2 %0!7 ("5"5"(/*""" % &&8;<!7 89:(; )/ / =2 %% 4/3!7 /*""" 8 % & / 454;! <(=>/(# #; 28 % % # 3! #('"#!"#"$("5"'-3 % > 4<? &&&2?(#8;11 % & 2 4 '-2(@ 8(3& & @+ 13@) A!"#"$3& &4 + $ 7 <(8 8% % 98; 7!"#"$(8""#"& 3 & 4$ 7 A.() 3% 3 /011 1!7!"#"$& & 4 % $ / 3 (5$("5"'-3/?8&? 8;&3/ 33 # A / 3 B(#("5"'-& &2 3 5(5* 0 4 & 3 /((= (A 9& : %&?#2 3

20 & + 3 % & + 3 % #('# ( " 5"'- & 2?#2 3 % *% % 0 & 2 %04;?#2 3 #('# ( " 5"'- 3 > &?3 "5"'-(C? &* B & & %&?()C,*+,,(?3 5 ( " 5" '- 9 &B3%& 4?3 2!"# ( 4 "( 5 &?!73 ( 5( 5 $ 5 &* 2 4 & & ;?!73 # " 5"'- ( D* A( D* 7* & E0 < & &&?!73 " " 0 F "!2 G?!73 D* A( ( " $ & & 6?!73 ; ) 8 ( 9: / 03&?!73 8 9: (; ) 54 64%2 8/?9?!73! E 5FG1G ( 8 9: / 3 = &?!73! 5D(D# ( 5 9CD* % $ 4=0?!3 8,(84A**%& 3&?!3, 8 A ( "5"'- 3 * &2 D 4 7?!3 8,(805&1 A* ;?!3 Edior: Sandra Silva (sandras@fep.up.p) Download available a: hp:// also in hp://ideas.repec.org/paperseries.hml

21