Catching up or standing still? National innovative ...people.bu.edu/furman/html/research/files/Furman Hayes - RP04.pdf · leaders among former follower countries. In this paper, we

Research Policy 33 (2004) 1329–1354

Catching up or standing still?National innovative productivity among ‘follower’ countries,

1978–1999

Jeffrey L. Furmana,∗, Richard Hayesb

a Boston University School of Management, 595 Commonwealth Avenue, #653a, Boston, MA 02215, USAb Melbourne Business School and Intellectual Property Research Institute of Australia, University of Melbourne, Australia

Received 18 March 2004; received in revised form 3 September 2004; accepted 3 September 2004Available online 25 November 2004

Abstract

Over the final two decades of the 20th century, a number of formerly industrializing economies and historical imitator countriesachieved levels of innovative capacity commensurate with or greater than those of some economies that were historically moreinnovative. We investigate the factors that enabled suchemerging innovatoreconomies to achieve successful catch-up whilesome historically more innovative countries experienced relative declines in innovative productivity. We focus our analysis onthe estimation of a production function for innovations at the world’s technical frontier. Based on the results of this analysis,

al indicest that the

p, but that

ndvel-wthesaveorldg-ho

ts to

we classify countries into categories reflecting their historical levels of innovative capacities and develop counterfactuthat identify the factors that correspond to long-run improvements in innovative roductivity. These exercises suggesdevelopment of innovation-enhancing policies and infrastructures are necessary for achieving innovative leadershithese are insufficient unless coupled with ever-increasing financial and human capital investments in innovation.© 2004 Published by Elsevier B.V.

Keywords:R&D productivity; Technology catch-up; National innovative capacity; National innovation systems; Patents

1. Introduction

Examining the state of British industrial perfor-mance in 1980, Keith Pavitt cautioned that unless thenation made substantial improvements in its innovative

∗ Corresponding author. Tel.: +1 617 353 4656.E-mail addresses:[email protected] (J.L. Furman),

[email protected] (R. Hayes).

capacity, both through additional industrial R&D aimproved linkages between R&D and product deopment, its prospects for long-run economic growould dim (Pavitt, 1980). This sentiment resonatwith those of economists and policymakers, who hfocused increasing attention in the years since WWar II on the centrality of scientific and technoloical advance in driving economic progress and whave argued that increasing national investmen

0048-7333/$ – see front matter © 2004 Published by Elsevier B.V.doi:10.1016/j.respol.2004.09.006

1330 J.L. Furman, R. Hayes / Research Policy 33 (2004) 1329–1354

innovation are essential to ensure countries’ economicgrowth (Schumpeter, 1942; Bush, 1945; Solow, 1956;Abramovitz, 1956; Romer, 1990; Jones, 1995).

In the near quarter-century since Pavitt’s initialappeal, Great Britain has made investments in its inno-vative capacity; its level of R&D expenditures and itsrealized level of USPTO patenting have increased byapproximately 30% each. At the same time, neighbor-ing Ireland, whose standard of living in the early 1980swas substantially lower than Britain’s, has vastly in-creased its economic and policy commitments to inno-vation, boosting its count of R&D personnel nearly ten-fold and achieving a 350% increase in USPTO patents,thus achieving a rate of per capita patenting comparablewith that of a number of the more innovative countriesin the world. The experience of these countries isillustrative of two striking facts about country-levelinnovative output over the last few decades.

First, among the set of countries that have histori-cally generated significant numbers of innovations atthe world’s technological frontier, the difference in therelative innovative productivity of the most innovativecountries and other innovative countries has declined.While the world’s leading innovator economies, includ-ing the United States, Switzerland, and Japan, havecontinued to increase investments in innovative capac-ity, other members of the group of innovator countrieshave increased their commitments to innovation at aneven greater rate. Thus, although the absolute gap ininnovative productivity between the world’s most in-n s re-m the2

rousn thep stri-a velso to-t riesi hath no-v nd,I g then in in-n ova-t triesw s re-c

The fact that some countries have increased their in-novative capacities so substantially while others havenot presents a puzzle for the study of national systemsof innovation (Freeman, 1987; Dosi, 1988; Lundvall,1992; Nelson, 1993), a literature which does not issuestrong predictions about the emergence of innovativeleaders among former follower countries. In this paper,we investigate developments in national innovative ca-pacities, focusing on the country-level investments, in-stitutional configurations, and national policy decisionsthat shape the success of follower nations in catchingup to the world’s leading innovator countries in termsof per capita innovative output. By studying the emer-gence of innovative capacity in former industrializingand imitator countries and examining the relative lev-eling of investments in innovation in some historicalinnovator countries, we build directly on set of issuescentral to Keith Pavitt’s work (Pavitt, 1979, 1980; Pateland Pavitt, 1987, 1989; Bell and Pavitt, 1992, 1993).Further, in adopting an approach that focuses on statis-tical analysis, we contribute to research that addressesPatel and Pavitt’s (1994)appeal for quantitative analy-sis clarifying the properties of national innovation sys-tems.

We base our analysis on the conceptual frameworkfor understanding national innovative capacity outlinedin Furman et al. (2002), which builds in particular onliterature in macroeconomic growth (Romer, 1990), na-tional industrial competitive advantage (Porter, 1990)and national innovation systems (Nelson, 1993).1 Thec tiono ntt ichw my’si s in

un-t ucea ativec raturei( pa-p mics.S hyl ova-t ri-g ture.T ted inP sc

ovative economies and other innovator countrieains, this gap is relatively smaller at the end of0th century than it was 20 years before.

Second, the set of countries that generate numeew-to-the-world innovations has expanded overast quarter-century, as a number of formerly indulizing countries have sufficiently increased their lef innovative productivity to begin introducing new-

he-world innovations with regularity. These countnclude a number of late industrializing countries tad been primarily imitators (and consumers) of inations at the world’s technological frontier. Irelasrael, Singapore, South Korea, Taiwan are amonations that have achieved remarkable increasesovative output per capita, suggesting that their inn

ive capacities have overtaken those of some counhose economic conditions were more favorable aently as the 1980s.

ore of our empirical analysis involves the estimaf a production function for economically significa

echnological innovations. The framework on whe based our estimation suggests that an econo

nnovative productivity depends on (a) investment

1 We employ the term “innovative capacity” to describe a cory’s potential – as both an economic and political entity – to prodstream of commercially relevant innovations. The term “innovapacity” has been used by a broad range of researchers in lite

n economics, geography and innovation policy. For example,Pavitt1980), employed the term in a manner similar to that in thiser in his broad-based research in innovation policy and econouarez-Villa (1990, 1993)applies the concept within the geograp

iterature, emphasizing the linkage between invention and innion. Neely and Hii (1998)provide a detailed discussion of the oins and definition of innovative capacity in the academic literahe framework presented here builds directly on research repororter and Stern (1999)andFurman et al. (2002)and the referenceited therein.

J.L. Furman, R. Hayes / Research Policy 33 (2004) 1329–1354 1331

broadly available resources for innovation, which werefer to as the common innovation infrastructure, (b)the environment for innovation in its industrial clus-ters, and (c) linkages between these components.

To evaluate this empirically, we employ a paneldataset of 23 countries between 1978 and 1999. Consis-tent with prior research, these regressions show a tightfit between predictors of national innovative capacityand economically significant innovations. These mod-els also bear out the striking result that a number of for-mer follower countries are becoming increasingly pro-ductive in their innovative output. To more fully explorethe factors driving this phenomenon, we categorizecountries into four groups based on historical patternsin their levels of innovative capacity: (1) leading inno-vator countries; (2) middle tier innovator countries; (3)third tier innovator countries; and (4) emerging innova-tor countries. Over the course of the sample, the leadinginnovator countries have the highest levels of innova-tive capacity, followed by the middle tier countries, andthe third tier countries. Average innovative capacity inemerging innovator grows substantially over the courseof the sample, from levels slightly higher than thoseof third tier innovators to levels that exceed those ofthe average middle tier economies. Although not quitecatching up to the world’s most innovative countries,emerging innovator countries as a group do surpass anumber of countries whose historical levels of wealthand innovation had vastly exceeded their own.

The improvements in national innovative capacityi anys est-m iverso ingi ectt nals onala therea al in-s in-n

gingi in-v city,b iesa am-i tingi o in-

novative capacity into components associated with (a)its policies and infrastructure and (b) its investmentsin innovation. This descriptive counterfactual exerciseexposes critical differences between groups of inno-vator countries. It demonstrates that leading innovatorcountries, middle tier innovator countries, and emerg-ing innovator countries have committed in relativelysimilar ways to innovation-enhancing policies. Middletier innovator countries and emerging innovators are,however, distinguished by the extent to which each hasincreased investments in R&D and human capital. Bycontrast, third tier innovator countries have neither sub-stantially increased their investments in R&D expendi-tures and human capital nor have they increased theircommitments to innovation-enhancing policies. We ex-plore both the public policy and theoretical implica-tions of these results in greater detail in our discussion.

The remainder of the paper is structured as follows:Section2 reviews the historical background for thisstudy and discusses prior research on catch-up andthe determinants of national innovative productivity.Section3 introduces the conceptual framework thatdrives our analysis. Section4 outlines our empiricalapproach. Our empirical results appear in Section5.Section6 concludes, discussing the findings of the pa-per in greater generality.

2. Leadership and catch-up in nationalinnovative productivity

2

un-t d’st bero The“ pani anc h re-m cu-r s int om-p vityw not.A re-b pani nt

n emerging innovator countries do not arise fromingle factor alone but rather from increased invent and commitment across a number of the drf national innovative capacity. Moreover, emerg

nnovator countries differ from each other with respo their geographic region of origin and their natioystems of innovation. Just as alternative institutirrangements can support continuous innovation,ppears to be no single dictate prescribing the idetitutional configuration necessary for catch-up inovative productivity and output.

Commonality does, however, exist across emernnovator countries: They exhibit ever-deepeningestments in the drivers of national innovative capaoth by committing to innovation-enhancing policnd investing in physical and human capital. We ex

ne the drivers of catch-up more precisely by creandices that decompose a country’s commitments t

.1. Historical background

In the years since World War II, the set of cories contributing regularly to innovation at the worlechnological frontier has expanded, raising a numf questions for conceptual and empirical study.economic miracles” of post-war Germany and Janvolved vast improvements in physical and humapital and culminated in the 1970s and 1980s witarkable increases in innovative productivity. It is

ious that, despite the destruction of their economiehe wake of World War II, Germany and Japan acclished such leaps in national innovative productihile countries such as England and France didlthough the United States played a critical role inuilding innovative capabilities in Germany and Ja

n the years after World War II, their most significa


gains in innovative capacity occurred in the 1970s and1980s, when national choices rather than US edictsdrove commitments to innovation.

This experience recurs in a different form in thefinal two decades of the 20th century, as a set of coun-tries nearly joins the group of elite innovator countriesalthough their economic and political circumstancesat the start of the 1980s are similar to or less favor-able than a set of countries whose innovative produc-tivity does not increase substantially over this period.These emerging innovators do not appear to have thesame historical advantages that benefited Germany andJapan. For example, emerging innovator countries suchas South Korea, Singapore, Ireland, and Finland, werenot rebuilding shattered economies that had historicallegacies of innovative leadership. Instead, these coun-tries developed imitator economies and transformedthem into innovative leaders by systematically and con-tinuously increasing their commitments to innovationover time.

We focus our empirical analysis in this paper onthis most recent time period, from 1979 to 1999, forwhich international data availability enables statisticalanalysis on the country-level determinants of innova-tive output. This proves to be an empirically interest-ing time frame: during this period, the set of countrieslisted above, as well as some other Scandinavian andAsian countries, vastly increased their innovative pro-ductivity. At the same time, a number of other countrieswith similar initial economic conditions and similarlyl n-c nds r ca-p ,2 n1 eri-c r ofU ow-e atinAm de-v n,2 vep de-v andI ale s ofo

2.2. Perspectives on innovation in economicgrowth and catch-up

The factors that affect economic progress acrosscountries have been of primary interest to politicalscientists, economic historians, economists, and pol-icymakers – and the role of technology has been prin-cipal in the debates.Veblen (1915)was pioneeringin comparing countries’ relative economic standingand identifying penalties associated with initial in-dustrial advantages.Gerschenkron’s (1962)view ofcatch-up expands on Veblen, suggesting that later-industrializing countries may be able to accelerate theirgrowth rates by adopting technology developed byleader countries and, although considerable obstaclesexist, may be able to leapfrog leader countries by de-veloping institutions that deal with contemporaneouschallenges more effectively than those developed inprevious periods. These authors identify a fundamentalquestion regarding whether laggard countries’ wealthand technological progress increase at a higher rate thanthat of leader countries.

Debate about the factors affecting catch-up andthe extent of convergence in economic conditionsacross countries has intensified since World War II.Since Solow (1956)and Abramovitz (1956)identi-fied the importance of technological progress in eco-nomic growth, questions about the role of innovationhave been a central feature of this debate.2 A num-ber of distinct research traditions have emerged aroundt rpo-r ostf ch-n tal)i th.S as-s t in-c trast,r ion-a ss andf ratea hiles tion

n-t keye ates.

ow initial levels of new-to-the-world innovation, iluding, for example, numerous Latin American aouthern European countries, did not improve theiacities for innovation as substantially (Furman et al.000; Furman and Stern, 2000). For example, betwee976 and 1980 a sample of emerging Latin Aman and Asian countries received similar numbeSPTO patents; by the second half of the 1990s, hver, patenting in the Asian economies dwarfs Lmerican countries’ output (AppendixTable A.1) (forore detailed studies of country-specific innovative

elopment, seeAmsden, 1989; Kim, 1997; O’Sulliva000; Trajtenberg, 2001). In some cases, innovatiroductivity increases concomitant with economicelopment. However, the example of Great Britainreland presented in Section1 demonstrates that initiconomic wealth alone does not fully explain levelr increases in innovative productivity.

hese issues, each of which conceives of and incoates technology in a different way. On one hand, mormal models of economic growth conceive of teology as a key input (along with labor and capi

n determining economic output and long-run growuch modeling efforts often require simplifyingumptions about the nature of technology and do noorporate its more nuanced characteristics. By conesearch in more historical, descriptive, or evolutry (e.g.,Nelson and Winter, 1982) traditions, rejecttrict simplifying assumptions about technologyocuses on more fine-grained factors that affect thend direction of technical change. For example, wome formal models make the simplifying assump

2 Similarly, Vannevar Bush’s report,Science: The Endless Froier(1945) identified scientific and technological progress as alement of national policy debates, particularly in the United St


that technology flows freely across place and time, eco-nomic historians and evolutionary theorists documentthe limitations of such assumptions.3

Within the tradition of formal economic models,this distinction is quite important. In early neoclassicalgrowth models, technology is viewed as spilling overfreely across countries, leading to certain convergencein levels of economic progress, leaving only “transi-tional dynamics” (Fagerberg, 1994, p. 1149) to explaindifferences across countries, subject to constraints as-sociated with capital mobility.4 Follow-on efforts in thein the 1960s incorporate learning-by-doing into formalmodels, but these ideas do not have an immediate im-pact on mainstream economics. The importance of acountry’s stock of knowledge and the parameters af-fecting the mobility of knowledge across borders ismore fully incorporated in the early 1990s, in mod-els of ideas-driven, endogenous growth (Romer, 1990;Grossman and Helpman, 1991). In these models, theability to apply existing technology and generate newinnovations differs systematically across economiesand convergence in economic wealth is not inevitable.

Empirical literature assessing drivers of economicgrowth and the extent of convergence across coun-tries is deep and varied (Barro and Sala-i-Martin, 1992,1995; Islam, 1995; Sala-i-Martin, 1996; Quah, 1997).Several authors in a primary strand of this literatureconclude that conditional convergence has occurredamong industrialized economies (Baumol, 1986), butthat this result does not hold if one selects countriesb athert s inm l.,1 r seto ing

sarilya utionst st e ab-s of theUw wasa h onc , e.g.,p owl-e

F

economies in Asia experience total factor productiv-ity and economic growth.Young (1995)documentsthis experience in Hong Kong, Singapore, South Ko-rea, and Taiwan, concluding that vast improvements inthese countries’ levels of per capita income result fromsubstantial growth in labor and capital over the period.

Complementing formal models and large scale em-pirical analysis, economic historians and technologyscholars have developed a perspective on the role oftechnology in economic advance in which a nuancedunderstanding of innovation is central.5 Fagerberg(1994) describes this perspective as the “technologygap” approach, and identifies a number of its cen-tral tenets. Specifically, he notes that authors in thisview (includingAmes and Rosenberg, 1963; Nelson,1981; Nelson and Winter, 1982; Nelson and Wright,1992) emphasize that technological innovation doesnot flow freely across economic actors or distancesbecause its creation and use are so closely tied to spe-cific firms, networks, and economic institutions. In thisview, the ability of economically lagging countries tocatch-up to leader countries depends on the investmentsin technology, as incorporating advances made else-where is essential to the process of catch-up.Ohkawaand Rosovsky (1973)note this explicitly, and charac-terize the ability to assimilate external technologies as“social capability.” Consistent with the argument thatspecific investments in innovative capabilities are es-sential for assimilating new-to-the-country innovation,Abramovitz (1986)proposes that countries whose eco-n thel ru-e po-r sons,B ini vel-o uip-m icala

er-s ems( el-s ar

hee

son,L

ased on economic leadership in the late 1800s rhan selecting from among the economic leaderore recent periods (DeLong, 1988; Baumol et a989). Convergence appears to apply to a greatef countries in the 1990s, as formerly industrializ

3 It is important to note that these literatures are not necest odds, and that some authors have made important contrib

o both streams. For example,Romer’s (1990)model of endogenouechnical change employs a concept of technology that is mortract than that of his historical essay examining the causesnited States’ technical leadership in manufacturing(1996). Like-ise, Abramovitz’s early growth accounting research (1956)keystone for early formal models, though his later researc

atch-up (1986) adopts a more phenomenon-driven approachroposing “technical congruence” as a notion to explain why kndge flows imperfectly across countries.

4 For additional elaboration, see also,Fagerberg (1987, 1988)andagerberg and Verspagen (2002).

omic environments more closely match that ofeader country will have better “technological congnce” and will, thus, be more successful in incorating advances made elsewhere. For related reaell and Pavitt (1992, 1993)argue that investments

nnovative capacity are essential for catch-up in deping countries, as investments in production eqent alone are insufficient for incorporating techndvances made elsewhere.

The natural progeny of the technology gap ppective, the literature on national innovation systFreeman, 1987; Dosi et al., 1988; Lundvall, 1992; Non, 1993; Edquist, 1997),6 focuses on the particul

5 Keller and Gong (2003)also provide a recent review of tvolution of economic growth and the role of technology.

6 This perspective is first articulated fully in the papers by Nelundvall (1988), andFreeman (1988)in Part V ofDosi et al. (1988).


configurations of firms, networks, and institutions thataffect innovative outcomes in different countries.7 Un-like the technology gap or economic growth literatures,research in the national innovation systems tradition isnot focused explicitly on relative levels of economic ortechnological development. Instead, this research hasemphasized rich, descriptive accounts of the constel-lations of organizations and policies that contribute topatterns of innovative behavior in particular countries,highlighting the institutions and actors whose roles inimportant industries are particularly decisive and em-phasizing the diversity in national approaches to inno-vation. Such actors include private firms, universities,public and quasi-public research organizations, gov-ernmental departments and ministries (e.g., military,aeronautics and health) as well as the institutions, le-gal authorities, budget-setting agencies, and norms thatinfluence the nature and extent of innovative efforts inan economy.8 Consistent with evolutionary theorizing(Nelson and Winter, 1982), this perspective also em-phasizes that processes leading to technical advanceinvolve detailed search efforts, iterative learning, andcomplex interactions among the actors described above(Lundvall, 1992). Understanding the processes operat-ing in a country’s (or region’s) innovative system re-quires far-reaching examinations of the relationshipsamong its actors and technological infrastructure. As aconsequence, research in this tradition has been pre-dominantly qualitative, promptingPatel and Pavitt’s(1994)call for follow-on research quantifying the char-a tions

ionh em-p sys-t ocusr na-t mico om-p roe-

w ointedo turea thesei es.

cese ture.M

conomic growth that model the ideas-generating sector(innovation-generating sector) of the economy as anendogenous determinant of economic growth (Romer,1990; Jones, 1995; Porter and Stern, 2000). In investi-gating the drivers of innovative outputs in the OECD,Furman et al. (2002)is among a number of recent pa-pers that build on both of these research streams toevaluate the determinants of innovation and innovativeproductivity at the country level. For example,Hu andMathews (2004)investigate developments in innova-tive capacity in a sample of five East Asian countries,concluding that public financing played a key role infostering the growth of their innovative capacities (see,alsoGans and Stern, 2003). We design this paper tocontribute to that emerging line of research, focusingon the factors that have allowed a number of formerfollower countries to achieve substantial improve-ments in their ability to generate new-to-the-worldinnovations.

3. Conceptual approach

3.1. Overview and introduction

Informed by the research traditions described inthe previous section, we pursue a conceptual and em-pirical approach with the aim of acknowledging thesubtleties of the national innovation systems and tech-nology gap literatures and incorporating its lessonsi verso urek ss ab ther er-a ratea ri-a mica men-t d re-s linge

er-s ityi as-d 8i sureo r of

cteristics, inputs, and outputs of national innovaystems.

Although the national innovation systems traditas not yet generated a great deal of large-scaleirical analysis, the nuanced national innovation

ems and technology gap literatures have helped fesearch efforts on exploring the determinants ofional innovative output as well as overall econoutput. This development occurred parallel to and clementary with advances in the literature on mac

7 These authors echoGerschenkron (1962)and North (1990),ho are among the numerous economic historians who have put the importance of national institutions in affecting the strucnd nature of competition across countries and described how

nstitutions have a long-run impact on national economic fortun8 It is important to note that important though subtle differen

xist among authors within the national innovation systems literacKelvey (1991)reviews some of these perspectives.

n a way that also allows us to assess the drif national innovative output. In order to measey constructs in a way that is comparable acroroad range of countries, we trade off some ofich detail of the national innovation systems litture; at the same time, we are able to incorpogreater degree of sensitivity for institutional va

tion than is characteristic of more formal econopproaches. We interpret our approach as comple

ary to, rather than a substitute for, both case-baseearch in innovation studies and more formal modefforts.

Accordingly, the framework we employ for undtanding the drivers of national innovative productivs fairly eclectic. It builds on recent models of ideriven economic growth (Romer, 1990; Jones, 199),

n which economic growth depends in great mean the production of the ideas-generating secto


the economy. The rate at which new ideas are pro-duced depends, in turn, on the stock of knowledge(previously generated ideas) and the extent of efforts(human and financial capital) devoted to the ideas-generating portion of the economy. The notion of anideas production function forms the core of our empir-ical approach to understanding catch-up in innovativeproductivity.

We build, as well, on ideas developed byRosenberg(1963)andPorter (1990)regarding the manner in whichmicroeconomic processes interact with the macroenvi-ronment and national institutions to affect the overalllevel of innovative activity in an economy. We in-corporate this understanding of the importance of themicrostructure of competition in our view of nationalinnovative productivity and catch-up.

The final pillar of our approach to understandingthe drivers of innovative output comes from the na-tional innovation systems literature, which emphasizesthe array of national policies, institutions, and relation-ships that drive the nature and extent of country-specificinnovative output.

3.2. Determinants of national innovative capacity

To explain the sources of differences among coun-tries in the production of innovations at the world’s

technical frontier, we employ the framework intro-duced byFurman et al. (2002). According to this frame-work, national innovative capacity is understood as aneconomy’s potential for producing a stream of com-mercially relevant innovations. In part, this capacitydepends on the technical sophistication and labor forcein a given economy; however, it also reflects the in-vestments, policies, and behaviors of the private sectorand the government that affect the incentives to en-gage in R&D and the productivity of the country’sR&D enterprise. The framework organizes the de-terminants of national innovative capacity into threemain elements (seeFig. 1): (1) a common pool ofinstitutions, resource commitments, and policies thatsupport innovation, referred to as the common innova-tion infrastructure; (2) the particular innovation orien-tation of groups of interconnected national industrialclusters; and (3) the quality of linkages between thetwo.

The innovative performance of a country’s economyultimately depends upon the activities of individualfirms and industrial clusters. Some of the most criti-cal investments that support innovative activity affectall innovation-oriented sectors in an economy. Thesecross-cutting factors comprise thecommon innovationinfrastructure(represented by the left-hand portion ofFig. 1). Consistent with models of ideas-based growth

al inno
Fig. 1. Nation vative capacity.


(Romer, 1990), the framework suggests that a country’sR&D productivity depends upon its historical stockof knowledge (denotedAt) as well as the amount ofscientific and technical talent dedicated to the pro-duction of new technologies (denotedHA,t). Innova-tive productivity also depends on national investmentsand policy choices (denoted asXINF), including factorssuch as expenditures on higher education, intellectualproperty protection, and openness to international com-petition, which will exert an over-arching impact on in-novativeness across the range of a country’s economicsectors (Nelson, 1993).

While the common innovation infrastructure pro-vides resources for innovation throughout an economy,it is the firms in specific industrial clusters that intro-duce and commercialize those innovations. The inno-vative capacity of an economy, then, depends upon theextent to which a county’s industrial clusters supportand compete on the basis of technological innovation.Drawing on the “diamond” framework developed inPorter (1990), we emphasize four key elements of themicroeconomic environment – the presence of high-quality and specialized inputs; a context that encour-ages investment and intense local rivalry; pressure andinsight gleaned from sophisticated local demand; andthe presence of a cluster of related and supporting in-dustries – that have a central influence on the rate ofinnovation in a given national industrial cluster (theseare the diamonds on the right-hand side ofFig. 1). Thepotential also exists for productivity-enhancing knowl-e rep-r n ther

a-t uc-t in an theq thea sci-e erc ex-p theu dyei tishc be-c notB ofs un-

try’s stronger university–industry relationships and thegreater availability of capital for technology-intensiveventures (Arora et al., 1998; Murmann, 2003).

4. Empirical approach and data

4.1. Empirical approach – estimating nationalinnovative productivity

We base our approach to assessing national inno-vative productivity on the ideas production functionarticulated byRomer (1990), Jones (1995), Porter andStern (2000). We use the national innovative capacityframework described above as a guide to direct ourmodel and analysis. Specifically, we describe a pro-duction function for economically significant techno-logical innovations, choosing a specification in whichinnovations are produced as a function of the factorsunderlying national innovative productivity:

Aj,t = δ(XINFj,t , YCLUS

j,t , ZLINKj,t )HAλ

j,t Aφj,t (1)

where for each countryj in yeart, Aj,t represents theflow of new-to-the-world innovations,HA

j,t reflects thetotal level of capital and labor resources devoted tothe ideas sector of the economy, andAj,t symbolizesthe stock of useful knowledge available to drive fu-ture ideas production. In addition,XINF refers to thelevel of cross-cutting resource commitments and pol-i tioni n-m rs,a eent trialc

icalm ta gt atestp tifi-c riodo van-t on ine cog-n achi m-p ce of

dge to spill over across industrial clusters (this isesented by the lines connecting the diamonds oight-hand side ofFig. 1).

Finally, the extent to which the potential for innovion supported by the common innovation infrastrure is translated into specific innovative outputsation’s industrial clusters will be determined byuality of linkages between these two areas. Inbsence of strong linking mechanisms, upstreamntific and technical activity may spill over to othountries more quickly than opportunities can beloited by domestic industries. For example, whilenderlying technology for creating the chemical

ndustry was the result of the discoveries of the Brihemist Perkin, the sector quickly developed andame a major exporting industry for Germany,ritain. At least in part, this migration of the fruitscientific discovery to Germany was due to that co

cy choices which constitute the common innovanfrastructure,YCLUS refers to the particular enviro

ents for innovation in a country’s industrial clustendZLINK captures the strength of linkages betw

he common infrastructure and the nation’s induslusters.

The reasoning we apply to arrive at an empirodel to estimate (1) follows the logic ofFurman el. (2002)and reflects our principal aim of allowin

he data to illustrate the phenomenon to the greossible extent. As the source of statistical idenation, we employ a panel dataset over a time pef more than 20 years. We can, therefore, take ad

age of both cross-sectional and time series variatistimating the parameters associated with (1). Reizing the benefits (and pitfalls) associated with e

dentification strategy, our analysis explicitly coares how estimates vary depending on the sour


identification.9 We are careful in our analysis to includeyear dummies to account for the evolving differencesacross time in the overall level of innovative output. Wealso include either country dummies or other measuresto control for aggregate differences in technological so-phistication (e.g., as reflected in GDP per capita). Bycontrolling for year and country effects in most of ouranalysis, we address some of the principal endogeneityand autocorrelation concerns.10

We base our specification of the innovation produc-tion function on the assumption that each of the termsof (1) are complementary with one another, in the sensethat the marginal productivity associated with increas-ing one factor is increasing in the levels of each ofthe other factors. (More precisely, this simplification isbased on the assumption that the factorsXINF, YCLUS,andZLINK enter (1) exponentially. Thus (1) becomesAj,t = δX

δINFj,t , Y

δCLUSj,t , Z

δLINKj,t HAλ

j,t Aφj,t and simplifies

to (2) after logarithmic transformation.) Denoting thenatural logarithm ofX asLX, our main specificationreduces to the following form:

LAj,t = δ + δINFLXINFj,t + δCLUSLYCLUS

j,t

+ δLINK LZLINKj,t + λLHA

j,t + φLAj,t + εj,t

(2)

The log–log form of this specification allows manyof the variables to be interpreted in a straightforwardway in terms of elasticities, is less sensitive to outliers,a ,1

4a

en-t -the-

thatc inno-v ty. Ont ionalc utput,b angesi raticc

msw tions

world innovation and the concepts underlying nationalinnovative capacity and develop a dataset that tracksthese measures across countries and over time. Con-structing a measure of commercializable innovationsthat is comparable and available across countries overthe course of our dataset and is indicative of nationalinnovative output is a difficult task. Consistent withFurman et al. (2002), we focus our analysis of vis-ible commercializable innovations on “internationalpatents” (PATENTS), which we define as the number ofpatents granted by the U.S. Patent & Trademark Officeto inventors from foreign countries.11

We recognize that no measure is perfect in charac-terizing the precise extent of innovation in an economyand readily acknowledge the well-understood hazardsof using patenting as an indicator of innovative ac-tivity (Schmookler, 1966; Pavitt, 1982, 1985, 1988;Griliches, 1984, 1990; Trajtenberg, 1990). As Grilichesnotes succinctly, “not all inventions are patentable, notall inventions are patented, and the inventions that arepatented differ greatly in ‘quality’, in the magnitude ofinventive output associated with them (1990, p. 1669)”.Such difficulties are exacerbated when comparing in-novation across countries because the propensity topatent also differs across country (Eaton and Kortum,1996, 1999; Kortum and Lerner, 1999).

At the same time, we focus on international patent-ing rates as “the only observable manifestation ofinventive activity with a well-grounded claim for uni-versality” (Trajtenberg, 1990, p. 183) and, thus, them ova-t web datas exer-c . Fore m-m ni-c al

tinga

First,o thati esso ional”p ande ech-n hird,w n the

nd is consistent with prior research in this area (Jones998).

.2. Measuring innovative output across countriesnd time

To perform our proposed analysis, we must idify observable measures that characterize new-to

9 Cross-sectional variation allows inter-country comparisonsan reveal the importance of specific determinants of nationalative capacity, yet it may be subject to unobserved heterogeneihe other hand, time series variation yields insight into how nathoices manifest themselves in terms of observed innovative out may be subject to its own sources of endogeneity (e.g., ch

n a country’s fundamental characteristics may reflect idiosynchanges in its environment).10 Porter and Stern (2000)have investigated potential probleith endogeneity in an innovation production function specificaimilar to the one used here.

ost useful measure available for comparing innive output across countries and over time. Thoughelieve that the advantages of international patentuggest it as the best measure for our purposes, weise caution in our use and interpretation of the dataxample, we construct PATENTS to include only coercially significant innovations at the world’s tech

al frontier.12 Moreover, in using realized internation

11 Furman et al. (2002)discusses the use of international patens a proxy for national innovative output in greater detail.12 Focusing on international patents helps satisfy this criteria.btaining a patent in a foreign country is a costly undertaking

s only worthwhile for organizations anticipating a return in excf these substantial costs. Second, USPTO-granted “internatatenting (PATENTS) constitutes a measure of technologicallyconomically significant innovations at the world’s commercial tology frontier that should be consistent across countries. Te are careful to accommodate the potential for differences i


Table 1Variables and definitions

Variable Full variable name Definition Source

Innovative outputPatentsjt+2 International patents granted

in yeart+ 2For non-US countries, patentsgranted by the USPTO. Forthe US, patents granted by theUSPTO to corporations orgovernments. To ensure thisasymmetry does not affect theresults we include a USdummy variable in theregressions

USPTO patent database

Quality of the common innovation infrastructureA GDP PER CAPITAj,t GDP Per Capita Gross Domestic Product per

capita, constant price, chainseries, US$

Penn World Tables, OECDScience & TechnologyIndicators

A GDP78j,t GDP 1978 1978 Gross Domestic Productconstant price, chain series,billions of 2000 US$

Penn World Tables

HA FTE R&D PERSj,t Aggregate PersonnelEmployed in R&D

Full time equivalent R&Dpersonnel in all sectors

OECD Science &Technology Indicators

HA R&D$j,t Aggregate Expenditure onR&D

Total R&D expenditures inYear 2000 millions of US$

OECD Science &Technology Indicators

XINF OPENNESSj,t Openness to internationaltrade and investment

Exports plus imports, inconstant dollar prices, dividedby GDP, expressed as apercent

Penn World Tables

XINF IPj,t Strength of protection forintellectual property

Average survey response byexecutives on a 1–10 scaleregarding relative strength ofIP (available beginning in1989)

IMD WorldCompetitiveness Report

XINF ED SHAREj,t Share of GDP spent onsecondary and tertiaryeducation

Public spending on secondaryand tertiary education dividedby GDP

World Bank, OECDEducation at a Glance

Quality of the cluster-specific innovation environmentYCLUS PRIVATE R&D FUNDINGj,t Percentage of R&D funded

by private industryR&D expenditures funded byindustry divided by totalR&D expenditures

OECD Science andTechnology Indicators

Quality of linkagesZLINK UNIV R&D PERFj,t Percentage of R&D

performed by universitiesR&D expenditures performedby universities divided bytotal R&D expenditures

OECD Science andTechnology Indicators

patents as an indicator of national innovative activity,we draw on a wide-range of research in economicsand innovation studies, includingSoete and Wyatt(1983), Evenson (1984), Patel and Pavitt (1987, 1989),Dosi et al. (1990), Eaton and Kortum (1996, 1999),

propensity to apply for patent protection across countries and overtime (as highlighted byScherer, 1983) by evaluating robustness ofour results to year and country-specific fixed effects.

Kortum (1997), Vertova (1999)and Furman et al.(2002).13

While we acknowledge that the “true” rate oftechnological innovation is unobservable and thatPATENTS is an imperfect proxy, our decision to use

13 For example,Patel and Pavitt (1987, 1989)compare the rela-tive innovativeness of European countries using USPTO-approvedpatents as a benchmark.


Table 2Sample countries

Australia Finland Ireland Norway SwedenAustria France Italy Polanda SwitzerlandBelgium Germanyb Japan Portugala Turkeya

Canada Greecea Mexico Slovak Republica United KingdomCzech Republica Hungary Netherlands South Korea United StatesDenmark Iceland New Zealand Spain

a These countries are included in supplemental analyses, but are omitted from the core regression analyses because of data limitations.b Prior to 1990, data are for West Germany only; after 1990, results include all German Federal states.

Table 3Means and standard deviations

Variable N Mean Standard deviation

Innovation outputPatents 473 3550.20 9193.53

Quality of the common innovation infrastructureA GDP PER CAPITA 473 18324.53 4582.88A GDP78 473 578.57 1000.85HA FTE R&D PERS 473 199797.80 383363.60HA R&D $ 473 15941.54 35650.40XINF OPENNESS 473 63.57 28.63XINF IP 245 6.72 1.09XINF ED SHARE 473 3.23 1.01

Cluster-specific innovation environmentYCLUS Private R&D funding 473 50.64 14.31

Quality of linkagesZLINK UNIV R&D PERF 473 22.25 6.55

this variable rests on the belief that PATENTS is posi-tively correlated with the true level of new-to-the-worldinnovative output in our panel dataset and that it repre-sents the best available indicator that allow us to com-pare national innovative output across a broad set ofcountries over time. We remain aware of the limita-tions of this measure, test it carefully for robustness,and bear these limitations in mind when interpretingour results.14

A list of our variables, definitions, and sources ap-pears inTable 1; the set of countries included in ouranalysis is listed inTable 2; and summary statisticsappear inTable 3. For all countries except the UnitedStates, we define PATENTS as the number of patents

14 In previous work (Furman et al., 2002), we explored severalalternative measures to PATENTS, including the rate of publicationin scientific journals (JOURNALS), the realized market share of acountry in “high-technology” industries (MARKET SHARE), andtotal factor productivity (TFP) and discuss the relative advantagesand disadvantages of using these measures.

granted in yeart+ 2 in the United States. This accountsfor the average lag between patent application and ap-proval. For the United States, we use the number ofpatents granted to government and corporations (non-individuals), in the United States in yeart+ 2.15

Across all years, the average country in our sam-ple obtains approximately 3550 PATENTS. Reflectingthe skewness in the data, the standard deviation in in-ternational patenting is substantially higher than themean (nearly 9200). At the country level, these dataevidence an increase in PATENTS in countries such asJapan, Finland, and South Korea, a solid increase inPATENTS in many western European countries, andonly modest increases in PATENTS in countries suchas Italy, Spain, and New Zealand.

15 To ensure that this asymmetry between US and non-US patentsdoes not affect our results we include a US dummy variable in allregressions that include US data. Note that the key results are alsorobust to the use of PATENTS based on date of application, and arealso robust to the use of alternative lag structures.


4.3. Measuring the drivers of innovative output

Limitations in the quality and extent of availabledata constitute the principal challenge in developinga dataset that allows us to measure the drivers of in-novative productivity in emerging innovator countries.We obtain the majority of our data from series pub-lished by the OECD Science and Technology Indica-tors, the World Bank, the USPTO and the Penn WorldTables. Prior to the 1990s, few countries outside of theOECD kept regular, reliable records on science and en-gineering or R&D-related activities. Thus, our ability tocompile a comprehensive historical dataset for a largesample of countries remains limited.16 As economistsand policy-makers have focused increasing attention oninnovation as a source of economic growth, nationalstatistical agencies and international bodies have un-dertaken more concerted efforts at gathering these data.As a consequence, we are able to expand on previousdata collection efforts to develop a dataset that reflectsinvestments in the drivers of national innovative pro-ductivity for 29 countries between 1978 and 1999. Ourcore dataset, on which we run our regressions, includes23 countries for which consistent data series are avail-able over the course of the sample period. In additionalanalyses, we are able to include six additional countriesfor which consistent data are available for a subset ofyears.17

We measure the strength of the common innova-tion infrastructure using variables that reflect the ex-t (c ital( PP tly,

ubli-c agesa bilitya st le ofE f his-t thath apore,T atlye

frome , sev-e econdy ng andf

reflecting the extent to which ideas are embodied ingoods and services. GDP78 equals the gross domesticproduct in 1978, the initial year of our sample. GDP78is a fixed measure, which reflects the initial stock ofknowledge in the economy, while GDP PER CAPITAconstitutes a variable measure. Measured in year 2000US$, GDP78 averages nearly 580 billion dollars acrosscountries. GDP PER CAPITA averages US$ 18,324over the sample.

Measures of R&D human capital and country-levelinvestments in R&D (FTE R&D PERS and R&D$)reflect the extent of R&D effort in the economy. Coun-tries in the dataset employ an average of nearly 200,000full-time equivalent R&D workers and invest nearly16 billion dollars annually on R&D over the sampleperiod. Fig. 2A depicts the substantial dispersion inper capita R&D investment in 1999 andFig. 2B thegrowth of R&D expenditures over the sample period.While leading innovator countries like Japan, Sweden,and Switzerland invest more than US$ 900 in R&Dper capita, countries with lower levels of innovativecapacity, such as Mexico, Poland, and Portugal reportfewer than US$ 100 in per capita R&D expenditures in1999. Consistent with the observation that countries’levels of visible innovative output become more sim-ilar over time, many of the countries with the lowestlevels of R&D investment are among those with thegreatest relative increases in R&D investment over theperiod. For example, although South Korea invests lessthan the median amount of R&D per capita in 1999,i easeo ke-w area &Di and1

oni -t rade( ro-t ex-p (EDS ureo mt qual

-N rt, an

ent of a country’s accumulated knowledge stockA),ountry-level investments in R&D and human capHA), and national policies (XINF). GDP78 and GDER CAPITA measure the knowledge stock indirec

16 Some additional data are available from country-specific pations and offices. These are often available only in local langund for recent year and questions exist about their comparacross countries and over time.Hu and Mathews (2004)addres

hese issues in compiling innovation statistics for their sampast Asian economics. The ability to analyze a complete set o

orical data for a wider array of countries – including both thoseave achieved apparent innovative success (e.g., Israel, Singaiwan) as well as currently industrializing countries – would grenhance research in this area.17 For the countries in the core dataset, we interpolated dataxisting years to obtain occasional missing values. For exampleral countries only report educational expenditure data every sear. For these we used an average of the immediately precediollowing years.

ts level of investment represents a staggering incrf 5570% relative to its expenditures in 1978. Liise, Portugal, whose per capita R&D expendituresmong the lowest in the sample, had increased its R

nvestment by more than 1600% between 1978999.

We measure the final component of the commnnovation infrastructureXINF, using indicators of naional policies regarding openness to international tOPENNESS), the strength of intellectual property pection (IP), and the share of GDP allocated toenditures for secondary and tertiary educationHARE). In this paper, we employ a direct measf the OPENNESS.18 Specifically, we use data fro

he Penn World Tables to compute total trade (e

18 Note that this differs fromFurman et al. (2002), in which OPENESS is based on data from the World Competitiveness Repo


Fig. 2.

to exports plus imports) as a proportion of GDP. Thismeasure correlates with the ability of firms in an econ-omy to target larger international markets and with theability of foreign firms to exploit their innovations inthe local economy. Across the sample, the mean levelof trade openness is 63.6%; not surprisingly, this figure

annual survey in which leading executives ranked their perceptionsof their country’s openness to trade. Although the measure we usehere differs, the results are qualitatively similar.

is higher in EU countries. IP is measured using execu-tives’ responses in the World Competitiveness Report.On a Likert scale between 1 and 10 (where 10 representthe strongest degree of protection), sample countriesearn an IP average of 6.7. The average country in thesample devotes 3.2% of GDP to secondary and tertiaryeducation.

To gauge the innovation orientation of industrialclusters and the strength of linkages, we employ com-positional variables that reflect the relative sources of


R&D funding between the public and private sector(PRIVATE R&D FUNDING) and the degree to whichR&D performance takes place in the university sec-tor (UNIV R&D PERFORMANCE).19 For our sam-ple countries, industry sources fund slightly more than50% of all R&D expenditures. As demonstrated inFig. 3A, this measure varies substantially across coun-tries. In 1999, private sources contribute less than 30%of R&D funds in countries such as Portugal, Mexico,and Greece, although they account for approximately70% of funding in South Korea and Japan. There isalso substantial variation in changes in PRIVATE R&DFUNDING over the sample period (Fig. 3B). Whileprivate sources in Iceland and Ireland increased theirfraction of R&D funding by more than 30%, PRI-VATE R&D FUNDING declined in Austria, Portugal,and Switzerland. Note that declines in PRIVATE R&DFUNDING in Austria and Portugal are, in a sense,more meaningful than those in Switzerland, as pri-vate sources fund a substantially higher fraction of na-tional R&D in Switzerland. UNIV R&D PERF aver-ages 22.2% across the sample and evidences similarvariation across countries.

5. Empirical results

5.1. Econometric analysis

Our econometric analysis applies the specificationi e re-s sa nter-p ca-t ntsh blese g ust ef-fi zes

ounda ea-s hesev sig-n (4-4).I ctorr corer IZA-T

the leverage of outliers on the results. In all models,R-squared is greater than 0.94; for models including yeardummy variables, it is greater than 0.997.

Table 4reports the primary national innovative ca-pacity results. Eqs. (4-1) estimates a specification thatreproduces the Romer–Jones ideas production func-tion model. The results show that GDP PER CAPITAand FTE R&D PERS have a significant and econom-ically important impact on PATENTS. The coefficienton FTE R&D PERS implies that a 10% increase in sci-ence and engineering employment is associated withan 11.6% increase in PATENTS. Eqs. (4-2) and (4-3) incorporate the elements of the common innovationinfrastructure and the complete national innovative ca-pacity model, respectively. Consistent with prior work,the key measures of the common innovation infrastruc-ture, the environment for innovation in national clus-ters, and the extent of linkages between the two enter ina statistically and economically significant manner. Co-efficients on variables expressed as a share (includingas ED SHARE and OPENNESS) can be interpreted asthe percentage increase in PATENTS resulting from a1% point increase in those variables. Eq. (4-4) presentsthe preferred national innovative capacity regression.In this model, elements of the common innovation in-frastructure, the environment for innovation in indus-trial clusters, and the linkages between the two enter ina statistically and economically significant manner.

Table 5explores the robustness of the model to anumber of modifications. In order to isolate the extentt thanc ectst n-t el;s or-p s oft entt anta y oft e ca-p andsi ndU e,s rivess mesn est-i their

n (2) to the core dataset of 473 observations. Thults appear inTables 4 and 5. This specification yieldnumber of advantages from the perspective of iretation. First, most of the variables in the specifi

ion enter in log form; consequently, their coefficieave a natural interpretation as elasticities. Variaxpressed as ratios are included as levels, allowino also use an elasticity interpretation for their cocients. Second, the log–log specification minimi

19 We have also examined alternative drivers in our backgrnalysis, including policy variables such as ANTITRUST and mures of the extent to which venture funding is available (VC). Tariables do not enter our models in a consistently statisticallyificant manner, and thus do not appear in the preferred model

n prior work, we have also modeled SPECIALIZATION as a faeflecting the cluster-specific environment for innovation. Theesults in this paper are also robust to the inclusion of SPECIALION.

o which the results are driven by time-series ratherross-sectional variation, we add country fixed effo the model in (5-1), with all country fixed effects eering significantly. (We omit GDP78 from this modince its value is fixed over time, it is effectively incorated into the country fixed effect.) Key measure

he extent of ideas in the economy and the commitmo R&D financial and human capital remain significnd of the expected valence in this equation; man

he more nuanced measures of national innovativacity become insignificant, however. The positiveignificant coefficient on PRIVATE R&D FUNDING

s robust to this modification, although ED SHARE aNIV R&D PERFORMANCE lose their significancuggesting that cross-sectional variation is what dignificance in these variables. OPENNESS becoegative and significant in this formulation, sugg

ng that countries that have, over time, increased


Fig. 3.

openness to international trade have generated fewerPATENTS. This may be an artifact of EU integration.The magnitude of the coefficient on GDP PER CAPITAchanges substantially when country fixed effects areadded, suggesting that the impact of within-country

changes in GDP PER CAPITA over time is differentfrom their impact across the cross-section.

Eqs. (5-2) and (5-3) reproduce key results fromTable 4, substituting PATENT STOCK for GDP PERCAPITA as a measure of the stock of knowledge in


Table 4Determinants of the production of new-to-the-world technologies

Dependent Variable = ln(PATENTS)j,t+2

(4-1) Baselineideasproductionfunction

(4-2) Commoninnovationinfrastructure

(4-3) Nationalinnovativecapacity:including allvariables

(4-4) Nationalinnovativecapacity:preferredmodel

Quality of the common innovation infrastructureA L GDP PER CAPITA 1.584(0.069) 0.697 (0.130) 0.870 (0.138) 0.836 (0.134)HA L FT R&D PERS 1.161(0.013) 0.737 (0.101) 0.865 (0.097) 0.850 (0.099)A L GDP78 −0.355 (0.055) −0.299 (0.062) −0.289 0.063HA L R&D $ 0.757 (0.076) 0.556 (0.077) 0.556 (0.078)XINF ED SHARE 0.069 (0.018) 0.091 (0.018) 0.089 (0.018)XINF IP 0.027 (0.042) 0.018 (0.040)XINF OPENNESS 0.004 (0.001) 0.0023 (0.0008) 0.0018 (0.0008)XINF ANTI-TRUST −0.044 (0.033) 0.002 (0.031)

Cluster-specific innovation environmentYCLUS PRIVATE R&D FUNDING 0.011 (0.002) 0.012 (0.002)

Quality of the linkagesZLINK UNIV R&D PERFORMANCE 0.011 (0.004) 0.011 (0.004)ZLINK VENTURE CAPITAL −0.047 (0.018)

ControlsYear fixed effects Significant Significant SignificantUS dummy 0.136 (0.086) 0.246 (0.085) 0.185 (0.076)Constant −21.931 (0.687)

R-squared 0.9470 0.9971 0.9974 0.9973Observations 473 473 473 473

the economy. The results of these equations echo thoseof the core national innovative capacity equations pre-sented inTable 4.

In addition to the robustness of the results, station-arity is a potential concern given the way we modelinternational patenting. To test for stationarity in ourln(PATENTS) data series, we followed the panel dataunit root tests ofLevin et al. (2002)andIm et al. (2003),which rely on pooled Augmented Dickey Fuller testingfor unit roots. In the case of ln(PATENTS), each testrejected the null hypothesis of nonstationarity, demon-strating that the series is notI(1).

5.2. Categorizing innovator countries

To more deeply understand differences and changesin the level of innovative productivity across countries,we undertake a counterfactual analysis in which wepredict per capita international patenting as a functionof countries’ realized levels of investments in innova-

tion and their policies towards innovation. Essentially,this exercise consists of predicting a country’s expectedinternational patenting rate by applying its observedlevels of the drivers of national innovative capacity tothe regression coefficients obtained in (4-4) and thennormalizing by national population. InFig. 4A, we plotthe results of this exercise for all of the countries in ourcore and expanded samples.

Several notable results emerge from this counter-factual analysis. The first is consistent with a catch-upphenomenon. While the United States and Switzerlandhave the highest levels of predicted per capita inter-national patenting across the period, the relative dif-ference in predicted per capita international patentingbetween these countries and others has declined overtime as other countries have begun to invest more sub-stantially in the drivers of national innovative capacity.For example, Japan dramatically improved its predictedinternational patenting between the early 1970s andthe present and a number of Scandinavian economies,


Table 5Exploring robustness

Dependent variable = ln(PATENTS)j,t+2

(5-1) CoreNIC model,with countryfixed effects

(5-2) Baseline ideasproduction function(w/PAT STOCK)

(5-3) Nationalinnovative capacitymodel (w/PATSTOCK)

Quality of the common innovation infrastructureA L GDP PER CAPITA 2.121 (0.307)A L PATENT STOCK 0.807(0.049) 0.557 (0.034)HA L FT R&D PERS 0.954 (0.129) 0.616(0.118) 0.422 (0.040)HA L R&D$ 0.218 (0.093) 0.289 (0.057)XINF ED SHARE 0.024 (0.023) 0.047 (0.018)XINF OPENNESS −0.006 (0.002) 0.0003 (0.0007)

Cluster-specific innovation environmentYCLUS PRIVATE R&D FUNDING 0.006 (0.003) 0.006 (0.002)

Quality of the linkagesZLINK UNIV R&D PERFORMANCE 0.002 (0.003) 0.001 (0.003)

ControlsCountry fixed effects Significant SignificantYear fixed effects Significant SignificantYEAR −0.045 (0.003)L GDP 1978 −0.255 (0.045)US dummy 0.213 (0.071)

R-squared 0.9991 0.9994 0.9979Observations 473 473 473

including Denmark and Finland, have made invest-ments that led to increased expected internationalpatenting since the mid-1980s. This catch-up phe-nomenon has not, however, occurred uniformly acrossall countries. For example, the estimates associatedwith several western European economies, includingthe UK, France, and Italy do not evidence marked in-creases in relative levels of innovative capacity.

While initial levels of innovative productivity are, atleast in part, the legacies of historical conditions, differ-ential rates of catch-up constitute a separate empiricalpuzzle. As a descriptive exercise that helps us to under-stand the factors driving differential rates of catch-upin innovative productivity, we classify the countries inour dataset into categories based on the historical evo-lution of their innovative capacity (Table 6). We des-ignate four categories: leading innovators, middle tierinnovators, third tier innovator, and emerging innovatorcountries.

Leading innovatorcountries, including the UnitedStates, Switzerland, Germany, Japan, and Sweden,

maintain high levels of innovative capacity through-out the sample period, with Germany, Japan, andSweden experiencing particular growth in their in-novative capacities in the early 1980s. Average ex-pected per capita patenting rates for this group rangefrom a minimum of 80 PATENTS per million per-sons to a maximum of over 170 over the sampleperiod.

Middle tier innovator countries, which include anumber of the western European countries, as well asAustralia, Canada, and Norway, maintain relatively sta-ble levels or slowly increasing level of innovative ca-pacity for much of the sample period. From 1978 to1995, middle tier innovator countries average patent-ing rates increase from approximately 38 to slightlymore than 50 PATENTS per million persons. Predictedpatenting rates rose somewhat more rapidly starting in1995, and reached nearly 80 PATENTS per million by1999. Even by 1999, however, middle tier countrieshave not reached an average level of innovative capac-ity equal to that of the leading innovators in 1978.


Fig. 4.

Third tier innovator countries, which includeItaly, New Zealand and Spain, experience rela-tively low levels of innovative capacity through-out the sample period, although their investmentsin innovation drivers also increase in the final fewyears of the 1990s. Expected international patentingrates in these countries are generally less than 30PATENTS per million persons throughout the sampleperiod.

Emerging innovatorcountries, Denmark, Finland,Iceland, Ireland, and South Korea, evidence a dra-matically different pattern. Beginning with compar-atively low expected patenting rates, these countrieshave increased their commitments to innovation by arelatively greater fraction than other countries in thesample. As a consequence, by 1999, their average ex-pected patenting rates have exceeded those of the mid-dle tier countries. Denmark and South Korea constitute


Table 6Categorizing innovator countries

Leadinginnovator

Middle tier Third tier Emerginginnovator

Germany Australia Hungary DenmarkJapan Austria Italy FinlandSweden Belgium Mexico IcelandSwitzerland Canada New Zealand IrelandUSA France Spain South Korea

NetherlandsNorwayUnited Kingdom

striking examples. In 1978, the predicted internationalpatenting rate was approximately 26 for Denmark andless than 1 for South Korea. Over the subsequent twodecades, these countries invested substantially their fi-nancial and human capital in innovation and raised theircommitments to innovation-supporting policies. By theend of the 1990s, Denmark and South Korea had sur-passed many countries whose historical levels of inno-vative capacity had exceeded their own.

Fig. 4B traces the historical average innovative ca-pacity levels of these groups.20 The figure also reportsaverages for countries in theexpanded dataset.21 Datafor these countries are not sufficiently complete to en-able including these countries in our historical regres-sions, but do enables us to compute predicted patentingrates in the late 1990s. Levels of innovative capacity inthese countries are less than half those of third tier in-novators over the course of the few years for whichreliable data are available.

5.3. Examining the drivers of catch-up

The unpredictable pattern according to which somecountries increased their innovative productivity dra-matically over the past quarter century while otherslagged behind constitutes an empirical puzzle. Ourcounterfactual exercise in the previous section demon-strates that part of the resolution to this puzzle lies in

d by( -g acity.F s mayb

lic,G

the extent to which these countries increased their com-mitments to the drivers of innovative capacity. In thissection we decompose the factors that drive innovativecapacity, in order to shed more light on the extent towhich investments and policy commitments have con-tributed to the rapid and substantial increases in innova-tive capacity in emerging innovator countries and themore modest increases achieved by middle and thirdtier countries. In order to do this, we develop two in-dices, which incorporate realized levels of the driversof innovative capacity along with the weights derivedin model (4-4 in the econometric analysis).

The first index, which we call the Investment In-dex, reflects the contribution of country-level invest-ments in R&D and human capital, and growth in thestock of ideas. Essentially, it is a population-adjustedmeasure ofA andHA. Specifically, it is calculated asthe linear combination of the realized levels of FTER&D PERS, R&D EXPENDITURES, and GDP PERCAPITA, multiplied by their matching coefficient es-timates from (4-4), exponentiated, and normalized bypopulation. The second, index, which we call the PolicyIndex for the purposes of discussion, is based on valuesof XINF, YCLUS, ZLINK . It is constructed in the samemanner as the Investment Index, using ED SHARE,OPENNESS, PRIVATE R&D FUNDING, and UNIVR&D PERFORMANCE. As these measures are notsubject to scaling, we do not adjust this index for pop-ulation.

We plot the historical Investment and Policy indicesfB irst,i ap-p t In-d icet tely1 ub-s t thee taged theP tieri arec f thel thes

ver-a triesi ird

20 Preliminary analysis applying the techniques elaborateIslam, 1995, 2003) to the data underlyingFig. 4B provides sugestive evidence of statistical convergence in innovative capurther research on statistical convergence in innovative outpute a promising avenue for future work.21 Expanded datasetcountries include the Czech Repubreece, Poland, Portugal, the Slovak Republic, and Turkey.

or each group of innovator countries inFig. 5A and. A number of interesting observations emerge. F

mportant differences in Investment Index levels arearent across the categories. The initial Investmenex for leading innovator countries is more than tw

hat of the middle tier innovators and approxima0 times that of third tier innovator countries and stantial differences among these groups remain and of the sample period. By contrast, the percenifference among innovator country categories inolicy Index is substantially smaller, and even third

nnovator countries have Policy Index values thatomparable to (though nonetheless below) those oeading innovator and middle tier countries overample period.

Over the course of the 1980s and 1990s, the age Investment Index for emerging innovator coun

ncreases from an initial level similar to that of th


Fig. 5.

tier innovator countries to a level that exceeds that ofmiddle tier countries. This steady increase reflects, inpart, rising levels of per capita GDP in emerging in-novator countries, but derives as well from increasedcommitments to R&D expenditure and human capital.Fig. 6 plots levels of R&D$ PER CAPITA and FTER&D PERS PER CAPITA by innovator category overtime. In each category, emerging innovator countriesincrease their investments at a rate greater than that ofother categories of innovator countries. The increase in

FTE R&D PERS PER CAPITA among emerging in-novator countries is so significant that per capita R&Demployment in these countries is nearly equal to thatof leading innovator countries by the end of the 1990s.

Differences in the timing of catch-up in the Invest-ment and Policy indices constitute a second observationof interest inFig. 5. Beginning in the 1980s, leading in-novator and middle tier innovator countries have nearlyequal policy index values and the differences betweenthese countries’ index values and those of the third tier


Fig. 6.

innovators is less than 25%. Again the emerging in-novator countries have made startling progress, withpolicy progress taking them to a comparable level tothe leading innovator countries. It is also interestingto note that, with the exception of a few years in thelate 1980s and early 1990s, the Policy Index does notincrease significantly in third tier innovator countriesover the sample period.

The coefficient of variation decreases over timefor both the Investment and Policy indices acrossthe sample of countries. Although both coefficientsof variation decrease over time, the investment in-dex exhibits greater variation across countries thanthe policy index. Similarly, the coefficients of varia-tion decrease over time for both the predicted and ac-tual international patenting rates. These results provide


preliminary evidence of a “narrowing of the gap” innational innovation rates, driven by decreases in thevariation of both investment levels and policy effects.

Taken together, the elements of this descriptive ex-ercise suggest that both changes in the policy environ-ment (i.e., changes inXINF, YCLUS, andZLINK ) andchanges in investment levels (HA) have affected catch-up dynamics in innovative productivity. The largersource of variation and catch-up across countries andinnovator categories appears to derive from the invest-ment index, suggesting that its elements have had themost impact in affecting relative levels of innovativecapacity.

6. Discussion

In this paper we contribute on a line of researchthat investigates the factors affecting national innova-tive output. We pursue an approach at an intermediatelevel of abstraction that combines elements from for-mal economic modeling and the more qualitative andappreciative traditions of research in innovation stud-ies. In this effort, we aim to build on one of the coreareas of Keith Pavitt’s research, and to do so in a mannerconsistent with his principled, eclectic approach to un-derstanding the phenomena associated with innovation.Specifically, we examine the factors that drive levels ofinnovative output in a sample of 29 countries, includingthe majority of the world’s most innovative economiesd ser-v findt nno-v llyp l in-n oure asuret in ac wellw re-s riven ingc

rpin-n ex-p ity.S erea te a

stream of leading-edge innovations while other coun-tries – including a number of countries with historicallyhigher levels of innovation – did not. The results sug-gest that there is no ‘magic bullet’ that explains thesechanges. Rather, innovative leadership arises from arange of sustained investments and policy commit-ments. Our analysis suggests that innovation-orientedpolicies and an appropriate composition of innovation-oriented investments are pre-requites for innovativeleadership, in the sense these characterize each of theworld’s leading innovator countries. Further, our resultsprovide evidence that, though necessary, these choicesalone are not sufficient to ensure innovative leadership.In fact, countries that we classify as middle tier inno-vators – countries whose levels of innovative capac-ity have been fairly stable, and consistently less thanthose of leading innovators – evidence policy commit-ments quite similar to the leading countries throughoutthe sample period, although their levels of innovativecapacity remain substantially lower than those of theleading innovators. While no country achieves a rela-tively high level of innovative capacity without suchinnovation-oriented policy commitments, policy com-mitments appear to be insufficient in the absence ofvastly increased investments in the drivers of innova-tive capacity.

The data do suggest that continuously increasinginvestments in innovation is, ultimately, essential forachieving innovative leadership. Japan, consistentlyamong the world’s most innovative countries duringt &Dp nce,w Dw dlet asei tuali entsc hosei Fore 14-f ex-p 999t canti parto es ino

ourr the

uring the years 1978–1999. A number of key obations emerge. Consistent with prior research, wehat a parsimonious model based on the national iative capacity framework performs in a statisticarecise way in predicting our measure of nationaovative output, international patents. Althoughconometric models are based on a particular me

hat does not capture the universe of innovationountry, we believe that this measure correlatesith our underlying concept of interest and that ourults provide useful evidence that the factors that dational innovative capacity are important in affectountry-level innovation.

This econometric analysis serves as the undeing for our efforts to understand the factors thatlain recent developments in innovative productivpecifically, we investigate why some countries wble to dramatically increase their ability to genera

he sample period, experienced an increase in Rersonnel of approximately 85%; by contrast, Frahich begins the period with a lower fraction of R&orkers in its labor force and is consistently a mid

ier innovator country, only experiences a 32% incren R&D personnel. The results of our counterfacndex suggest that substantial increases in investmharacterize the emerging innovator countries wnnovative capacity develops most over the period.xample, Finland’s real R&D expenditures increaseold over the period and South Korea’s real R&Denditures were more than 450 times greater in 1

han they were in 1978. These increases are signifin both relative terms and absolute terms, and aref the factors that enable their substantial increasverall innovative capacity.

In some sense, a straightforward implication ofesults is that greater inputs into innovation at


country level are associated with greater outputs; how-ever, a simplistic interpretation of this result would beincomplete. Whereas all countries that achieve substan-tial increases in innovative capacity increased invest-ments in key drivers of innovation over the period,each of these countries also maintained innovation-enhancing policies. A number of countries achievedpolicies supportive of innovation, but did not investin R&D to the same degree. By contrast, no countryappears to achieve high levels of investment in inno-vation without innovation-oriented policies. Together,these findings suggest to us that a well-functioning in-novation infrastructure is necessary but not sufficientto create the environment required to achieve sustainedinnovation at the world’s technological frontier.

Moreover, the finding that a number of wealthyeconomies with well-functioning science and inno-vation systems do not increase their investments ininnovation sufficiently to enter the ‘club’ of leadinginnovator countries suggests either that national inno-vative leadership is a strategy that countries choose notto pursue or that continuously deepening commitmentsto innovation are difficult to achieve, even for middletier innovator economies. Although it is important toexercise caution in interpreting the results for policy,the results suggest that, for such middle tier countries,the greatest challenge in enhancing innovative produc-tivity lies in increasing their levels of investment ininnovation rather than in adjusting their national poli-cies or innovation systems. For third tier innovators, ita h di-m gingi alsoq his-t heiri e in-n d’si

horsi e of-t ffectt of-t ionp ing

rn-m exces-s

this research agenda to focus as well on the determi-nants of thelevelof innovative inputs achieved by aneconomy. In the tradition of Keith Pavitt’s research,both qualitative and quantitative approaches could bearfruit. Large-scale empirical analyses that jointly esti-mate the determinants of innovative inputs and outputscould tackle endogeneity issues not addressed by a re-duced form approach; detailed, cased-based researchwould, however, be a necessary complement that elu-cidates the underlying mechanisms by which these pro-cesses occur.

The complementarity between detailed qualitativework and large-scale empirical analysis is also impor-tant for understanding the phenomenon associated withemerging innovator countries. While emerging innova-tor countries share the fact that they have managed toincrease their commitments to innovative capacity in away not matched by middle tier or third tier innovatorcountries, these countries – e.g., Iceland, Ireland, andSouth Korea – vary substantially in the institutionalconfigurations that characterize their national innova-tion systems. That they are all able to achieve relativelyhigh levels of innovative capacity underscores the pointthat there is no universal recipe that enables followercountries to catch-up to leading innovator economies.Understanding both the unique circumstances that haveled these countries to make substantial commitments toinnovative capacity and the common relationships thatallow these countries’ investments to bear fruit requiresthe interplay between qualitative and quantitative re-s

A

ntlyw nalr ctionw hersf ersf a,R fuls ul toM ance.A ouss e-s y &F titute

ppears as if there is room for improvement on botensions. While the success of a number of emer

nnovator economies allows for hope, the resultsuantify the magnitude of the challenges faced by

orically less-innovative economies in increasing tnnovative capacity substantially enough to achievovative productivity commensurate with the worl

nnovator countries.Researchers in innovation studies, including aut

n the national innovation systems approach, haven asked about the country-specific factors that ahe composition of inputs into innovation – anden report on the levels of inputs into the innovatrocess.22 The results of this study suggest deepen

22 Romer (2000), for example, investigates whether US goveent subsidies for scientist and engineering are essential or

ive.

earch methods.

cknowledgements

This paper builds on research conducted joiith Michael E. Porter and Scott Stern and additio

esearch conducted by Joshua Gans in conjunith Scott Stern. We thank each of these researc

or thoughtful discussion, four anonymous reviewor their insightful suggestions, and Virginia Achichard Nelson, and Orietta Marsili for their caretewardship of this Special Issue. We are gratefercedes Delgado for excellent research assistll errors are our own. We acknowledge the graciupport of the Boston University Junior Faculty Rearch Fund, the Victorian Department of Treasurinance and the Intellectual Property Research Ins


Table A.1

Country 1976–1980 1995–1999 Growth rate

Emerging Latin American economiesArgentina 115 228 0.98Brazil 136 492 2.62Chile 12 60 4.00Colombia 28 42 0.50Costa Rica 22 48 1.18Mexico 124 431 2.48Venezuela 50 182 2.64

Emerging Asian economiesChina 3 577 191.33Hong Kong 176 1694 8.63India 89 485 4.45Malaysia 13 175 12.46Singapore 17 725 41.65South Korea 23 12062 523.43Taiwan 135 15871 116.56

Source: Porter et al. (2000).

of Australia. We are grateful for the comments of par-ticipants in the conferences at SPRU, University of Sus-sex and the University of Catania Faculty of PoliticalScience.

Appendix A

SeeTable A.1.

References

Abramovitz, M., 1956. Resource and output trends in the UnitedStates since 1870. American Economic Review 46 (2), 5–23.

Abramovitz, M., 1986. Catching up, forging ahead and falling be-hind. Journal of Economic History 46, 385–406.

Ames, E., Rosenberg, N., 1963. Changing technological leadershipand industrial growth. Economic Journal 74, 13–31.

Amsden Alice, H., 1989. Asia’s Next Giant. Oxford University Press,New York.

Arora, A., Landau, R., Rosenberg, N., 1998. Chemicals and Long-Term Economic Growth: Insights from the Chemical Industry.Wiley, New York, NY.

Barro, R., Sala-i-Martin, X., 1992. Convergence. Journal of PoliticalEconomy 100 (2), 223–251.

Barro, R., Sala-i-Martin, X., 1995. Economic Growth. McGraw-Hill.Baumol, W.J., 1986. Productivity growth, convergence, and welfare:

what the long-run data show. American Economic Review 76 (5),1072–1085.

Baumol, W.J., Batey, S.A., Wolff, E.N., 1989. Productivity andAmerican leadership: The Long View. MIT Press, Cambridge,MA.

Bell, M., Pavitt, K., 1992. Accumulating technological capabilityin developing countries. The World Bank Research Observer,257–272.

Bell, M., Pavitt, K., 1993. Technological accumulation and industrialgrowth: contrasts between developed and developing countries.Industrial and Corporate Change 2, 157–210.

Bush, V., 1945. Science, the Endless Frontier. A Report to the Pres-ident on a Program for Postwar Scientific Research. NationalScience Foundation, Washington, DC.

DeLong, J.B., 1988. Productivity Growth, Convergence, and Wel-fare: Comment. American Economic Review 78 (5), 1138–1154.

Dosi, G. (Ed.), 1988. Technical Change and Economic Theory. PinterPublishers, London, UK.

Dosi, G., Freeman, C., Silverberg, G., Soete, L., 1988. TechnicalChange and Economic Theory. Pinter Publishers, London, UK.

Dosi, G., Pavitt, K., Soete, L., 1990. The Economics of TechnicalChange and International Trade. Columbia University Press, NewYork, NY.

Eaton, J., Kortum, S., 1996. Trade in ideas: patenting and productiv-ity in the OECD. Journal of International Economics 40 (3), 2–8.

Eaton, J., Kortum, S., 1999. International technology diffusion: the-ory and measurement. International Economic Review 40 (3),537–570.

Edquist, C. (Ed.), 1997. Systems of Innovation: Technologies, Insti-tutions, and Organizations. Pinter Publishers, London, UK.

Evenson, R., 1984. International invention: implications for tech-nology market analysis. In: Zvi, Griliches (Ed.), R&D, Patents,and Productivity. University of Chicago Press, Chicago, IL, pp.89–126.

Fagerberg, J., 1987. A technology gap approach to why growth ratesdiffer. Research Policy 16 (2/4), 87–99.

Fagerberg, J., 1988. Innovation, catching up, and growth. In: Tech-nology and Productivity: The Challenge for Economic Policy.

F es in175.

F ation-Re-

F ance:

F si, G.lish-

F of na-

F riversBest

F f na-ean

G pacityte of

OECD, Paris, pp. 37–46.agerberg, J., 1994. Technology and international differenc

growth rates. Journal of Economic Literature 32 (3), 1147–1agerberg, J., Verspagen, B., 2002. Technology-gaps, innov

diffusion, and transformation: an evolutionary interpretation.search Policy 31, 1291–1304.

reeman, C., 1987. Technology Policy and Economic PerformLessons from Japan. Pinter, London, UK.

reeman, C., 1988. Japan: a new system of innovation. In: Do(Ed.), Technical Change and Economic Theory. Pinter Pubers, London, UK, pp. 330–348.

urman, J.L., Porter, M.E., Stern, S., 2002. The determinantstional innovative capacity. Research Policy 31, 899–933.

urman, J.L., Porter, M.E., Stern, S., 2000. Understanding the dof national innovative capacity. Academy of ManagementPapers in Proceedings.

urman, J.L., Stern, S., 2000. Understanding the drivers otional innovative capacity – implications for central europeconomies. Wirtschaftspolitische Blatter 47 (2).

ans, J.S., Stern, S., 2003. Assessing Australia’s Innovative Cain the 21st Century. Intellectual Property Research InstituAustralia Working Paper.


Gerschenkron, A., 1962. Economic Backwardness in Historical Per-spective. Belknap Press of Harvard University Press, Cambridge,MA.

Griliches, Z. (Ed.), 1984. R&D, Patent and Productivity. ChicagoUniversity Press, Chicago, IL.

Griliches, Z., 1990. Patent statistics as economic indicators: a survey.Journal of Economic Literature 92, 630–653.

Grossman, G., Helpman, E., 1991. Innovation and Growth in theGlobal Economy. MIT Press, Cambridge, MA.

Hu, M.-C., Mathews, J.A., 2004. National Innovative Capacity inEast Asia. Macquarie University Working Paper.

Im, K., Pesaran, M., Shin, Y., 2003. Testing for unit roots in hetero-geneous panels. Journal of Econometrics 115, 53–74.

Islam, N., 1995. Growth empirics: a panel data approach. QuarterlyJournal of Economics 110, 1127–1170.

Islam, N., 2003. What have we learnt from the convergence debate?Journal of Economic Surveys 17 (3), 309–362.

Jones, C., 1995. R&D based models of economic growth. Journal ofPolitical Economy 103, 739–784.

Jones, C., 1998. Introduction to Economic Growth. W.W. Norton &Company, New York, NY.

Keller, W., Gong, G., 2003. Convergence and polarization in globalincome levels: a review of recent results on the role of interna-tional technology diffusion. Research Policy 32, 55–79.

Kim, L., 1997. Imitation to Innovation: The Dynamics of Korea’sTechnological Learning. Harvard Business School Press, Cam-bridge, MA.

Kortum, S., 1997. Research, patenting, and technological change.Econometrica 65 (6), 1389–1419.

Kortum, S., Lerner, J., 1999. What is behind the recent surge inpatenting. Research Policy 28 (1), 1–22.

Levin, A., Lin, C., Chu, C., 2002. Unit root tests in panel data: asymp-totic and finite-sample properties. Journal of Econometrics 108,1–24.

Lundvall, B.-A., 1988. Innovation as an interactive process: fromn. In:inter

L s ain-

M er?. In:cs –lgar,

M . Thens.

N Lit-dies,

N duc-Eco-

N par-

Nelson, R.R., Winter, S., 1982. An Evolutionary Theory of EconomicChange. Harvard University Press, Cambridge, MA.

Nelson, R.R., Wright, G., 1992. The rise and fall of American tech-nological leadership: the postwar era in historical perspective.Journal of Economic Literature 30 (4), 1931–1964.

North, D.C., 1990. Institutions, Institutional Change, and EconomicPerformance. Cambridge University Press, Cambridge, UK.

Ohkawa, K., Rosovsky, H., 1973. Japanese Economic Growth. Stan-ford University Press, Palo Alto, CA.

O’Sullivan, M., 2000. The sustainability of industrial developmentin Ireland. Regional Studies 34 (3), 277–290.

Patel, P., Pavitt, K., 1987. Is Western Europe losing the technologicalrace? Research Policy 16 (2/4), 59–86.

Patel, P., Pavitt, K., 1989. A comparison of technological activities inWest Germany and the United Kingdom, National WestminsterBank. Quarterly Review, 27–42.

Patel, P., Pavitt, K., 1994. National innovation systems: why theyare important, and how they might be measured and compared.Economics of Innovation and New Technology 3 (1), 77–95.

Pavitt, K., 1979. Technical innovation and industrial development.Futures 11 (6), 458–470.

Pavitt, K., 1980. Industrial R&D and the British economic problem.R&D Management 10, 149.

Pavitt, K., 1982. R&D, patenting, and innovative activities: a statis-tical exploration. Research Policy 11 (1), 33–51.

Pavitt, K., 1985. Patent statistics as indicators of innovative activities:possibilities and problems. Scientometrics 7, 77–99.

Pavitt, K., 1988. Uses and abuses of patent statistics. In: van Raan,A. (Ed.), Handbook of Quantitative Studies of Science Policy.North Holland, Amsterdam.

Porter, M.E., 1990. The Competitive Advantage of Nations. The FreePress, New York, NY.

Porter, M.E., Stern, S., 1999. The New Challenge to America’s Pros-perity: Findings from the Innovation Index. Council on Compet-itiveness, Washington, DC.

P unc-ing

P res deana yives de

Q ca-omic

R of Po-

R omic

R r De-king

R ol in-43.

S urnal

user-producer interaction to the national system of innovatioDosi, G. (Ed.), Technical Change and Economic Theory. PPublishers, London, UK, pp. 349–369.

undvall, B.-A., 1992. National Systems of Innovation: TowardTheory of Innovation and Interactive Learning. London, UK, Pter.

cKelvey, M., 1991. How do national systems of innovation diffA critical analysis of Porter, Freeman, Lundvall, and NelsonHodgson, G.M., Screpanti, E. (Eds.), Rethinking EconomiMarkets, Technology, and Economic Evolution. Edward EAldershot, UK.

urmann, J.P., 2003. Knowledge and Competitive AdvantageCoevolution of Firms, Technology, and National InstitutioCambridge University Press.

eely, A., Hii, J., 1998. Innovation and Business Performance: Aerature Review, mimeo. Judge Institute of Management StuUniversity of Cambridge.

elson, R.R., 1981. Research on productivity growth and protivity differences: dead ends and new directions. Journal ofnomic Literature 19 (3), 1029–1064.

elson, R.R. (Ed.), 1993. National Innovation Systems: A Comative Analysis. Oxford University Press, New York, NY.

orter, M.E, Stern, S., 2000. Measuring the ‘Ideas’ Production Ftion: Evidence from International Patent Output, NBER WorkPaper 7891.

orter, M.E., Furman J.L., Stern S., 2000. Los Factores Impulsola Capacidad Innovadora Nacional: Implicaciones para EspAmerica Latina (English title: The Drivers of National InnovatCapacity: Implications for Spain and Latin America), Clavela Economia Mundial. ICEX, Madrid, Spain, pp. 78–88.

uah, D.T., 1997. Empirics for growth and distribution: stratifition, polarization, and convergence clubs. Journal of EconGrowth 2, 27–59.

omer, P., 1990. Endogenous technological change. Journallitical Economy 98, S71–S102.

omer, P., 1996. Why, indeed, in America? American EconReview 86 (2), 202–206.

omer, P., 2000. Should the Government Subsidize Supply omand in the Market for Scientists and Engineers? NBER WorPaper No. w7723.

osenberg, N., 1963. Technological change in the machine todustry, 1840–1910. Journal of Economic History 23, 414–4

cherer, F.M., 1983. The propensity to patent. International Joof Industrial Organization 1, 107–128.


Sala-i-Martin, X., 1996. The classical approach to convergence anal-ysis. The Economic Journal 106, 1019–1036.

Schumpeter, J., 1942. Capitalism, Socialism and Democracy, 2nd ed.George Allen & Unwin, Ltd., London.

Schmookler, J., 1966. Invention and Economic Growth. Harvard Uni-versity Press, Cambridge, MA.

Soete, L.G., Wyatt, 1983. The use of foreign patenting as an inter-nationally comparable science and technology output indicator.Scientometrics 5 (1), 31–34.

Solow, R.M., 1956. A contribution to the theory of economic growth.Quarterly Journal of Economics 70, 65–94.

Suarez-Villa, L., 1990. Invention, inventive learning, and innovativecapacity. Behavioral Science 35, 290–310.

Suarez-Villa, L., 1993. The dynamics of regional inventionand innovation: innovative capacity and regional change in

the twentieth century. Geographical Analysis 25 (2), 147–164.

Trajtenberg, M., 1990. Patents as Indicators of Innovation, EconomicAnalysis of Product Innovation. Harvard University Press, Cam-bridge, MA.

Trajtenberg, M., 2001. Innovation in Israel 1968–1997: a compara-tive analysis using pate data. Research Policy 30 (3), 363.

Veblen, T., 1915. Imperial Germany and the Industrial Revolution.Macmillan, New York.

Vertova, G., 1999. Stability in national patterns of technological spe-cialisation: some historical evidence from patent data. Economicsof Innovation and New Technology 8 (4), 331–354.

Young, A., 1995. The tyranny of numbers: confronting the statisticalrealities of the East Asian growth experience. Quarterly Journalof Economics 110, 641–680.

Documents

Catching up or standing still? National innovative ...people.bu.edu/furman/html/research/files/Furman Hayes - RP04.pdf · leaders among former follower countries. In this paper, we