Fagerberg2008-National Innovation Systems, Capabilities and Economic Development

8/2/2019 Fagerberg2008-National Innovation Systems, Capabilities and Economic Development

1/19

Research Policy 37 (2008) 14171435

Contents lists available at ScienceDirect

Research Policy

j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / r e s p o l

National innovation systems, capabilities and economic development

Jan Fagerberg , Martin SrholecCentre for Technology, Innovation and Culture (TIK), University of Oslo and Centre for Advanced Study (CAS) at the Norwegian Academy of Science and

Letters, Moltkte Moesvei 31, N-0317 Oslo, Norway

a r t i c l e i n f o

Article history:Received 24 October 2007Received in revised form 29 May 2008Accepted 6 June 2008Available online 15 August 2008

JEL classification:

E11F43O30

Keywords:

InnovationNational innovation systemCapabilitiesGovernance

Development

a b s t r a c t

This paper focuses on the role of capabilities in economic development. In recent years,the quality and availability of data on different aspects of development have improved, andthis provides new opportunities for investigating the reasons behind the large differencesin economic development. Using factor analysis on data for 25 indicators and 115 countriesbetween 1992 and 2004, we identify four different types of capabilities: thedevelopmentof the innovation system, the quality of governance, the character of the political sys-tem and the degreeof openness of the economy. Innovation systems and governanceareshown to be of particular importance for economic development.

2008 Elsevier B.V. All rights reserved.

1. Introduction

Not long ago most economists believed that differ-ences in development levels across countries were to beexplained by one single factor, namely differences in theamount of accumulated capital per worker (Solow, 1956,see Fagerberg, 1994 for an overview). However, from the1960s onwards the idea that differences in developmentare mainly caused by technological differences received

increasing support (Gerschenkron, 1962). This view was, ofcourse, consistent with the perspective on growth devel-oped by Schumpeter (1934, 1943), and during the 1980sa lot of new work on cross-country differences in levels ofdevelopment and growth performance inspired by thisper-spective emerged (Freeman et al., 1982; Fagerberg, 1987,1988; Dosi et al., 1990; Verspagen, 1991). The focus ontechnology as the driving force of growth and develop-

Corresponding author. Tel.: +47 97 22 84 16 00; fax: +47 22 84 16 01.E-mail address: [email protected] (J. Fagerberg).

ment has also been taken up by advocates of the so-callednew growth theory (Lucas, 1988; Romer, 1990; Aghionand Howitt, 1992).

Authors that emphasize the crucial role of technol-ogy for development tend to stress that catching up intechnology is by no means a free ride. According to thisperspective, countries that do not succeed in developingappropriate technological capabilities and other comple-mentary factors should be expected to continue to lag

behind. Concepts such as social capability (Ohkawa andRosovsky, 1974; Abramovitz, 1986), technological capabil-ity (Kim, 1980, 1997), absorptive capacity (Cohen andLevinthal, 1990) and innovation system (Lundvall, 1992;Nelson, 1993; Edquist, 1997) have been suggested and aburgeoning empirical literature has emerged focusing onthese aspects of development (see Fagerberg and Godinho,2004; Archibugi and Coco, 2005 for overviews). However,as we show in the next section of this paper, there is a bigoverlap between several of these concepts andthe relation-ship between conceptual and empirical work in this area isoften weak.

0048-7333/$ see front matter 2008 Elsevier B.V. All rights reserved.doi:10.1016/j.respol.2008.06.003
http://www.sciencedirect.com/science/journal/00487333mailto:[email protected]://dx.doi.org/10.1016/j.respol.2008.06.003http://dx.doi.org/10.1016/j.respol.2008.06.003mailto:[email protected]://www.sciencedirect.com/science/journal/00487333


2/19

1418 J. Fagerberg, M. Srholec / Research Policy 37 (2008) 14171435

To some extent this reflects that, until recently, therewas a lack of appropriate data that could be used to putnumbers on the various aspects emphasized by the the-oretical literature. But in recent years, the quality andavailability of data on different aspects of developmenthave improved, and this may give researchers a newopportunity for investigating the reasons behind the largedifferences in economic performance. In the third sec-tion of the paper we therefore, following the theoreticalwork in this area, proceed to an empirical analysis of thecapabilities needed to succeed in development. Ratherthan picking individual indicators and combine them in anessentially arbitrary way we follow Adelman and Morris(1965, 1967) and Temple and Johnson (1998) in mappingthe most central elements with the help of factor anal-ysis. The underlying assumption behind this approach isthat indicators reflecting the same dimension of the real-ity will tend to be correlated. The results, presented inthe fourth section of the paper, clearly illustrate the mul-tidimensional character of capabilities, resulting in fourdifferent dimensions, which we label innovation system,governance, political system and openness, respec-tively. Finally, we explorethe extent to which cross-countrydifferences in capabilities may help us understand whysome countries excel economically while other continueto be poor. We show that what matters most for successis a well-functioning innovation system and good gover-nance.

2. Capabilities and development: lessons from the

literature

The first systematic attempts to study the relationshipsbetween technology, capabilities and development weremade by economic historians who wanted to understandwhy some countries managed to catch-up with the richerones while other countries continued to be poor. Morethan40yearsagoAlexanderGerschenkronpointedoutthattechnological catch-up, although potentially highly lucra-tive, is an extremely challenging venture (Gerschenkron,1962). Based on a study of the performance of a number ofEuropean countries relative to the then leading country Great Britain he concluded that to succeed in technolog-ical catch-up less advanced countries had to develop whathe called new institutional instruments, e.g., organiza-tions capable of identifying the most promising optionsahead and muster the necessary resources for exploitingthese opportunities. Gerschenkrons work is often associ-ated with his focus on investment banks, which he saw ascritical in mobilizing resources for development. However,as Shin (1996) points out, it is possible to see his writingsas an attempt to arrive at a more general understanding ofthe conditions for catch-up, focusing on the instruments or capabilities to use a more recent term that needto be in place for successful catch-up to take place. Shinalso emphasizes the historically contingent nature of thecapabilities needed for catch-up. For example, the factorsthat constrained German catch-up towards the end of thenineteenth century are not necessarily the same as those

experienced by Japan in the early post World War Twoperiodor otherAsian countriesmore recently. Hence, while

the need for well-developed capabilities may be quite gen-eral, their precise nature maywell differ between historicaltime periods.

Moses Abramovitz, arguing along similar lines as Ger-schenkron, suggestedthat differences in countries abilitiestoexploitthepotentialforcatch-upmaytoalargeextentbeexplained by differences in what he called social capabil-ity. What he hadin mind was not so much individual skills,important as these may be, but rather what organizationsin the private and public sector are capable of doing andhow this is supported (or hampered) by broader societalfactors. These are some of the aspects of social capabilitythat he emphasized as being particularly important1:

- managerial and technical competence;- a stable and effective government, capable of supporting

economic growth;- financial institutions and markets capable of mobilizing

capital on a large scale;- the spread of honesty and trust in the population.

The concept social capability soon became very popu-lar in applied work but there have not been many attemptsto develop empirical measures reflecting the factors thatAbramovitz considered to be important. In later work, hepointed out that the concept was vaguely defined andexpressed pessimism with respect to the possibilities foradequate measurement (Abramovitz, 1994b, p. 24 and 36).In practice, it has often been assumed to be synonymouswith educational attainment (Baumol et al., 1989) which,although arguably an important element, is a much morenarrow perspective than what Abramovitz had in mind.

The works of Gerschenkron and Abramovitz focused

mainly on evidence from Europe and the United States.However, from the 1970s onwards several studies of catch-up (or lack of such) in other parts of the world emerged.For example, there is by now an ample literature demon-strating that the catch-up of not only Japan (Johnson, 1982)but also other so-called newly industrializing countriesinAsia (Amsden, 1989; Wade, 1990; Kim, 1997) were asso-ciated with conscious capability building as envisaged byGershenkron and Abramovitz. The argument that capa-bility building is a precondition for successful catch-uphas received further backing from a series of empiricalstudies of industrialization processes in Asia and Latin-America undertaken during the 1970s and 1980s (Kim,

1980; Fransman, 1982; Fransman and King, 1984; Dahlmanet al., 1987; Lall, 1987, 1992). The successful catch-upof a number of newly industrializing countries in the1970s and 1980s (the NICs) also served as inspirationfor the development of new perspectives on the dynam-ics of the global economy that put the development ofappropriate technological activities (or capabilities) at thecore of the analysis (Fagerberg, 1987, 1988; Dosi et al.,1990; Verspagen, 1991; for an overview see Fagerberg andGodinho, 2004).

One case which received much attention was the riseof Korea from being one of the poorest countries in the

1 See Abramovitz (1986, pp. 387390; 1994a, pp. 3435; 1994b, p. 88).
http://dx.doi.org/10.1016/j.respol.2008.06.003http://dx.doi.org/10.1016/j.respol.2008.06.003http://dx.doi.org/10.1016/j.respol.2008.06.003http://dx.doi.org/10.1016/j.respol.2008.06.003http://dx.doi.org/10.1016/j.respol.2008.06.003http://dx.doi.org/10.1016/j.respol.2008.06.003http://dx.doi.org/10.1016/j.respol.2008.06.003http://dx.doi.org/10.1016/j.respol.2008.06.003http://dx.doi.org/10.1016/j.respol.2008.06.003http://dx.doi.org/10.1016/j.respol.2008.06.003http://dx.doi.org/10.1016/j.respol.2008.06.003http://dx.doi.org/10.1016/j.respol.2008.06.003http://dx.doi.org/10.1016/j.respol.2008.06.003http://dx.doi.org/10.1016/j.respol.2008.06.003http://dx.doi.org/10.1016/j.respol.2008.06.003


3/19

J. Fagerberg, M. Srholec / Research Policy 37 (2008) 14171435 1419

world to a first world technological powerhouse in justthreedecades.LinsuKim,whomadetheauthoritativestudyon the subject, suggested the concept technological capa-bility (Kim, 1980, 1997) as an analytical device to interpretthe Korean evidence. He defined it as the ability to makeeffectiveuse of technologicalknowledgein efforts to assim-ilate, use, adapt and change existing technologies. (Kim,1997, p.4).2 Hence, theconcept includes not only organizedR&D, which arguably is a small activity in many developingcountries, but also other capabilities needed for the com-mercial exploitationof technology.As has becomecommonin the literature he considered three aspects of it: pro-duction capability, investment capability and innovationcapability. Kims assessment was that the requirementsshould be expected to become more stringent, in particularwith respect to innovation capabilities, as countries climbup the development ladder. Thus, for a firm or country inthe process of catching up, the appropriate level of tech-nological capability would be a moving target, in constantneed of improvement.

The concept technological capability has been usedin a large number of studies at various levels of aggrega-tion (see Romijn, 1999, for an overview). Although initiallydeveloped for analyses of firms, it has also been appliedto industries and countries. Sanjaya Lall, in a survey (Lall,1992), emphasized three aspects of national technologicalcapability as he phrased it; the ability to muster the nec-essary (financial) resources and use them efficiently; skills,including not only general education but also specializedmanagerial and technical competence; and what he callednational technological effort, which he associated withmeasures such as R&D, patents and technical personnel. Healso noted that national technological capability does notonly depend on domestic technological efforts but also onforeign technology acquired through imports of machin-ery or foreign direct investments. This argument is alsoemphasized by advocates of the so-called new growththeory according to which small countries are at a dis-advantage in innovation and depend on free trade and aliberal stance towards international capital flows in orderto overcome this problem (Grossman and Helpman, 1991;Coe and Helpman, 1995).

Lall (1992) also made a distinction between technolog-ical capabilities proper and their economic effects. Theseeffects, henoted, didalsodepend on theincentives that eco-nomicagents facewhether resultingfrom politicaldecisionmaking (e.g., governance) or embedded in more long-lasting institutions (thelegal framework, for example). Thisreasoning is of course very similar to that of Abramovitz(see the discussion above of social capabilities). Thus,technological and social capabilities should be expected tointeract in the development process. Moreover, by closerexamination it becomes evident that there is a consider-able overlap between these two concepts. For example,

2 This definition is, as the reader may observe, quite similar to that ofabsorptive capacity. Wesley Cohen and Daniel Levinthal who suggestedthetermabsorptivecapacity defined it as the ability of a firmto recog-nize the value of new, external information, assimilate it and apply it to

commercial ends (Cohen and Levinthal, 1990, p. 128). Kim (1997) usedthe two concepts interchangeably.

both include aspects related to skill formation and finance.This clearly complicates the analysis. It might be arguedthat there is a need for more leaner, more well-definedconcepts. This points to the need for a methodology thatcan advice on which aspects to combine into a commondimension, an issue that will be explored in the follow-ing.

However, the observation that technological and socialfactors interact in the process of economic developmentmight also be taken as supporting the view that a broader,more systemic approach that take such interactions intoaccount is required. Such concerns led during the 1980s tothedevelopmentofanewsystemicapproachtothestudyofcountries abilities to generate and profit from technology,the so-called national innovation system approach. Theconcept, first used in public by Christopher Freeman in ananalysis of Japan (Freeman, 1987), soon became a popularanalytical tool for researchers who wanted to get a firmergrasp on the interaction processes underlying a countrystechnological and economic development (Lundvall, 1992;Nelson, 1993; see Edquist, 2004 for an overview). But theadoption of the innovation system approach to developingcountries is a relatively recent phenomenon (Viotti, 2002;Muchie et al., 2003; Lundvall et al., 2006) and arguablystill in its infancy. Moreover, there is currently no agree-ment in the literature on how innovation systems shouldbe defined and studied empirically.3 Hence, trying to putnumbers on this may be a challenge, as Archibugi and Coco(2005) point out. Still there have been some attempts inthat direction, see, for example, Furman et al. (2002) andFurman and Hayes (2004), and in the next section we willdiscuss this issue in more detail.

3. Measuring capabilities

A main aim of this paper is to contribute to an improved(and more transparent) relationship between conceptualand applied work in this area. For this purpose, we needa data set that is as comprehensive as possible, both withrespect to measurable aspects and (time and country) cov-erage. Typically, most developed market economies figureprominently amongthose withgood coverage, while devel-oping countries and former socialist economies lack dataon many potentially useful indicators. Based on an ini-tial screening of data for 175 countries and more than100 potentially relevant indicators, we narrowed downthe sample to 115 countries and 25 indicators. To limitthe influence of shocks occurring in specific years, weexpress the indicators as 3-year averages for the initialand final periods (19921994 and 20022004). A longertime period would clearly been desirable but that wouldhave implied that many of the data sources taken intoaccount would not been available and/or that most devel-oping countries would have had to be excluded from theanalysis. Stilltherewere a fewmissingdata pointsfor many

3 Some researchers in thisarea emphasize a need fordeveloping a com-mon methodology, based on the functions and activities of the system,to guide empirical work (Liu and White, 2001; Johnson and Jacobsson,

2003; Edquist, 2004), while others advocate the advantage of keepingtheapproach open and flexible (Lundvall, 2003).


4/19


Table 1

Measuring capabilities

Aspect Measure Capability

Science, research and innovation Scientific publications, patents, R&D (total/business), innovation counts TechnologicalOpenness Openness to trade, foreign direct investment, technology licensing, immigration TechnologicalProduction quality/standards International (ISO) standards TechnologicalICT infrastructure Telecommunications, internet, computers TechnologicalSkills Primary, secondary and tertiary education, managerial and technical skills Technological and social

Finance Access to bank credit, stock-market, venture capital Technological and socialQuality of governance Corruption, law and order, independence of courts, property rights, business friendly

regulationSocial

Social values Civic activities, trust, tolerance SocialType of political system Civil (political) right s; checks and balances; democracy or autocracy ?

Note: Measures in italics were not included in the analysis due to lack of data.

countries/indicators, which we estimated with the helpof information on the other indicators and countries (seeAppendix A for details).

Based on thediscussion in thepreceding section, Table1presents an overview of the aspects that we expect to be ofparticular relevance for development along with examples

of possible empirical indicators.As discussed earlier, the concept technological capabil-ity refers to the ability to develop and exploit knowledgecommercially. An important element of this is the ability toinnovate, what Kim (1997) termed innovation capability.There are several data sources that may capture differentaspects of this. For example, the quality of a countrys sci-ence base, on which invention and innovation activitiesto some extent depend, may be reflected in articles pub-lished in scientific and technical journals.4 Research anddevelopment (R&D) expenditures measure some (but notall) resources that are used for developing new productsor processes, while patents count (patentable) inventions

coming out of that process. However, R&D data is not avail-able for many developing countries, which led us not toinclude it. Patent data, on the other hand, are available forall countries5 but many if not most innovations are neverpatented, since patents are awarded to inventions, notinnovations, and the propensity to patent varies consider-ably acrosssectors andlevels of economic development.So,as for many other indicators, this gives at best a very partialview of what we wish to measure. Firms own judgmentsabout their innovativeness (innovation counts) might havebeen another possible source of information but such dataare unfortunately only available for a relatively small num-ber of countries (see Smith, 2004 for an overview).

Openness (or interaction) across country borders mayfacilitate technology transfer (spillovers) and stimulateinnovation. This issue is as mentioned above particularlyemphasized in work inspired by thenewgrowththeories.Four channels of technology transfer across country bor-ders have been examined in the literature: trade, foreigndirect investment, migration and licensing (for overviews

4 There can be an upward language/regional bias for English-speakingnationsand/orcountrieswitha close links tothe UnitedStates.No attemptwas made to correct for these possible biases.

5 We use only patents granted by the United States Patent and Trade-

markOffice (USPTO) to assure consistency in terms of criteria for novelty,originality, etc.

see Cincera and Van Pottelsberghe de la Potterie, 2001;Keller, 2004). However, due to lack of data for many coun-tries, especially in the developing world, we will only beable to take into account the two former channels, e.g.,trade andforeign direct investments. To avoida bias againstlarge economies (that for natural reasons trade/interact

relatively more internally) both indicators were measuredorthogonal to country size.6

Another important element of technological capabilityemphasized by Kim is production capability. As an indi-cator of this we include the adoption of quality standards(ISO 9000). Although ISO certification is mainly proce-dural in nature, it is increasingly seen as a requirementfor firms supplying high quality markets, and is thereforelikely to reflect a high emphasis on quality in produc-tion. We also include three indicators of ICT use: personalcomputers, Internet users and fixed/mobile phone sub-scribers. Although earlier studies suchas Lall (1992) didnotplace much emphasis on this dimension nowadays a well-

developed ICT infrastructure must be regarded as a mustfor a country that wish to catch-up. Arguably this holds notonly for production and investment capability but for theability to innovate as well.

Theimportant role that a countrysfinancialsystem mayplay in mobilizing resources for catching-up was pointedoutalreadyby Gershenkronand isalsoemphasized bymorerecent research (see, e.g., King and Levine, 1993; Levine,1997; Levine and Zervos, 1998). We capture this aspectby the amount of credit (to the private sector) and bycapitalization of companies listed in domestic capital mar-kets. Another important set of factors emphasized by, forexample, Abramovitz, and for which there is also solid sup-

port in the literature, is skills (Nelson and Phelps, 1966;Barro, 1991; Benhabib and Spiegel, 1994; for an overviewsee Krueger and Lindahl, 2001). We include three indi-cators, the quality of basic education (as reflected in theteacherpupil ratio in primary schools) and the rates ofenrolment in secondary and tertiary education. However,despite detailed scrutiny we have not been able to findinformation on specialized managerial and technical skillsthat could be used in this study.

The importance of governance and institutions, fur-nishing economic agents with incentives for creation and

6

The variables were regressed against (the log of) land area and theresiduals from these regressions were then used in the analysis.


5/19


diffusion of knowledge, is generally acknowledged in theliterature. Although such factors often defy hard mea-surement, especially in a broad cross-country comparison,there exist somesurvey-based measures, oftencollected byinternational organizations, that may throw some light ontheseissues. We findit usefulto distinguishbetween, on theone hand, the quality of governance with respect to inno-vation and economic life more generally and the characterof the political system on the other. For the former, we usesurvey data reflecting how easy it is to set up and operatea business, whether property rights exist and are enforced,how widespread corruption is conceived to be, the extentto which law and order prevail and courts are seen as beingindependent. All these aspects, we will argue, fall withinthe realms of politics and may, to some extent at least, beachieved within quite different political systems. However,we find it pertinent to include the latter dimension as well,since it cannotbe excluded that there is a stronginteractionbetween the two, and that some of the aspects for whichwehaveinformationmay reflectboth.To measure thechar-acter of the political system we include, in addition to theindicators mentioned above, variablesreflectingthe degreeof democracy versus autocracy, checks and balances in thepolitical system, degree of competition for posts in theexecutive and legislature and the extent of political rightsand civil liberties. Since Western democracies will tend tohavehigh values on most of thesevariables, a possibleinter-pretationmightbethatwhatismeasuredhereisthedegreeof Westernizationof a countrys institutions.7 Moreinfor-mationon definitions andsourcesof thesevariables is givenin Appendix A.

However, the impact of governments actions may aspointed out by Abramovitz also depend on the prevail-ing social values in society such as, for example, tolerance,honesty and trust and civic engagement. Such values,facilitating socially beneficial, cooperative activities, areoften seen as expressions of so-called social capital (seeWoolcock and Narayan, 2000 for an overview). It wouldclearly have been interesting to include such aspects herebut unfortunately the limited coverage of the existing sur-veys on these issues precluded this.

4. Identifying capabilities: factor analysis

Given the relatively large number of indicators there isobviously a lot of information to exploit. One of the key

challenges confronting us in this study is how to com-bine this information into a smaller number of dimensionswith a clear-cut economic interpretation. The most widelyused approach to construct composite variables is to selectrelevant indicators and weigh them together using prede-termined (usually equal) weights, see Archibugi and Coco(2005) for an overview. The problem with this is thatthe selection of indicators and choice of weights tend to

7 Institutions are commonly referred to as the rules of the game(North, 1981) as opposed to actions carried out with the help of theserules (e.g., politics or governance). In practice, the distinctions between

politics, governance and institutions other are often blurred (Glaeser etal., 2004).

be quite arbitrary. An alternative approach, pioneered byAdelman and Morris (1965, 1967), uses so-called factoranalysis (Basilevsky, 1994) to advise on questions like this.This method is based on the very simple idea that indi-cators referring to the same dimension are likely to bestrongly correlated, and that we may use this insight toreduce the complexity of a large data set (consisting ofmany indicators) into a small number of composite vari-ables, each reflecting a specific dimension of variance inthe data. As is common in factor analysis, the variableswere standardized. We used the mean and standard devi-ation of the pooled data for the standardization, whichimplies that the change of a composite variable over timewill reflect both changes in each countrys position (rel-ative to other countries) and changes in the importanceof the underlying indicators over time (relative to otherindicators). In this way, we take into account as muchinformation as possible. We used the method of principal-component factors and the oblique oblimin rotationprocedure to arrive at the solution (see Appendix B fordetails).

In Table 2, we present results of the factor analysis ondatafortheinitialandfinalperiodforthe115countriescov-ered by the investigation (230 observations). The analysisled to the selection of four principal factors jointly explain-ing 74% of the total variance. The table presents loadingsof the variables on these factors after the rotation. Theseloadings are the correlation coefficients between the indi-cators (rows) and factors (columns) and provide the basisfor interpreting the different factors.

The first factor loads highly on several indicators asso-ciated with different aspects of technological capabilitysuch as patenting, scientific publications, ICT infras-tructure, ISO 9000 certifications and access to finance.However, it also correlates highly with education, so itcuts across the established distinction between techno-logical and social capabilities. We suggest interpretingit as a synthetic measure of the capabilities influencingthe development, diffusion and use of innovations to useEdquist (2004)s definition of an innovation system. Hence,we label this factor innovation system.

Fig. 1 plots the factor score on the innovation sys-tem against GDP per capita (as a measure of the level ofeconomic development). There is very close correlationbetween the two.8 To the extent that there are deviationsfrom the regression line this primarily comes from a groupof resource rich economies (OPEC countries for instance),having slightly higher GDP per capita levels than the qual-ity of their innovation systems would indicate, and some ofthe former centrally planned economies for which it is theother way around.

8 It might be noted that the innovation system variable also corre-lates strongly with R&D and innovation counts for the subset of countriesfor which reliable data exist. For the 74 countries for which R&D datawere available the correlation coefficient between R&D as a percentageof GDP (in logs) and the innovation system variable was 0.80. For the 25EU/EFTA countries that took part in the fourth Community InnovationSurvey (EUROSTAT) the correlation coefficient between the share of firms

withinnovation activities (inlogs) andthe innovationsystem variable was0.68.


6/19


Table 2

Results of factors analysis

Innovation system Governance Political system Openness

USPTO patents (per capita) 0.63 0.33 0.14 0.08Science and engineering articles (per capita) 0.63 0.42 0.06 0.12ISO 9000 certifications (per capita) 0.81 0.04 0.00 0.26Fixed line and mobile phone subscribers (per capita) 0.94 0.00 0.03 0.06Internet users (per capita) 0.81 0.18 0.09 0.31

Personal computers (per capita) 0.79 0.17 0.00 0.17Primary school teacherpupil ratio 0.82 0.17 0.10 0.18Secondary school enrolment (% gross) 0.92 0.07 0.03 0.13Tertiary school enrolment (% gross) 0.95 0.09 0.08 0.18Domestic credit to private sector (% of GDP) 0.47 0.44 0.00 0.09Market capitalization of listed companies (% of GDP) 0.46 0.32 0.05 0.21Merchandise imports (% of GDP) 0.11 0.04 0.12 0.77Foreign direct investment inward stock (% of GDP) 0.05 0.01 0.10 0.85Impartial courts 0.09 0.88 0.04 0.07Law and order 0.21 0.59 0.07 0.00Property rights 0.00 0.87 0.16 0.01Regulation 0.11 0.71 0.04 0.00Informal market (corruption) 0.27 0.67 0.03 0.21Index of democracy and autocracy 0.04 0.03 0.96 0.03Political constraint 0.09 0.02 0.80 0.03Legislative index of political competitiveness 0.02 0.20 0.84 0.00

Executive index of polit ical compet it iveness 0.08 0.23 0.84 0.07Political rights 0.01 0.22 0.89 0.02Civil liberties 0.01 0.27 0.82 0.04

Note: Four factors with eigenvalue >1 were detected, which explain 74.0% of total variance; extraction method: principal-component factors; rotation:oblimin oblique. Number of observations = 230 (pooled data for 115 countries in the initial and final period). Source: See Appendix.

It is interesting to note that the innovation systemvariable developed here differs in important respects fromthe more narrowly defined innovative capacity variable,basedonpatentstatistics,suggestedby Furmanetal.(2002)and Furman and Hayes (2004).9 The problem with measur-ing innovation capability solely by patents is that patentsare given to (globally novel) inventions. Minor innova-tions/adaptations, which arguably make up the bulk ofinnovativeactivity world-wide, will not be counted follow-ing thisapproach sincesuch innovationsare notpatentable.Hence, for developing countries most of the innovativeactivity would get unrecognized by this approach. This lim-itation, is of course, not an argument for excluding patentsfrom the investigation. Rather the point is that there areseveral potentiallyrelevant sources of information on inno-vation, all imperfect and that excluding all but one of theseis likely to lead to a measurement bias.

The second factor loads high on various aspects reflect-ing the quality of governance such as adherence toproperty rights, a well-functioning judicial system, littlecorruption and a favourable environment for business. Asin the previous case this factor score correlates positivelywith the level of economic development (and significantlyso). However, as is evident from Fig. 2, the relationship isnot as strong as in the case of the innovation system. The

9 While they point to large economies such as the US, Japan and Ger-many as being among thegloballeaders, ourstudy indicates thatthe mostadvanced innovation systems are to found in smaller countries (in termsof population)such as Australia, Sweden Denmarkand Norway. Arguably,the reason for this difference may be that the latter countries have welldeveloped capabilities for exploiting knowledge, factors that are impor-

tant for the innovation system measure developed here, but ignored byFurman and associates.

main reason for this is that there is a group of countries,mainly of African origin, that deviates from the main pat-tern by being much poorer than the quality of governanceshould indicate.

The third factor, in contrast, loads particularly high onindicators reflecting the character of the political sys-tem. In short, countries with political systems that areclose to those of the Western world, rank high on thisdimension, while countries with systems that differ fromWestern democratic ideals, get a low mark. In contrast tothe innovation system and governance variables, however,thecharacterofthepoliticalsystemisnotcloselycorrelatedwith levels of development. In fact, Fig. 3 reveals that somecountries with distinctly authoritarian regimes do ratherwell economically. However, most countries cluster to theright in the figure indicating a Western type political sys-tem irrespective of their level of economic development.

Finally, there is a fourth factor that correlates highlywith two indicators only: imports of goods and servicesand foreign direct investments. Hence, this factor reflectsthe degree of openness to trade and foreign capital. Wetherefore label it openness. As noted this is a factor thatis deemed especially important by followers of the newgrowth theory. However, Fig. 4 shows that the compos-ite variable based on this factor does not correlate witheconomic development. This holds irrespective of whethercountry size is controlled for or not. It might be arguedthough that openness (in an economic sense) should notonly be about flows of goods and money across borders butalso about flowsof people and ideas, for which we have notbeen able to find much relevant information that could beincluded in the analysis. This being said, one indicator thatmight reflect openness to ideas to some extent is accessto internet. It is interesting to note, therefore, that Internet


7/19


Fig. 1. GDP per capita and innovation system(average level over 20022004).Note: For definition of the innovation system variable seeTable 1. Source: SeeAppendix A.

accessis only weakly correlated with openness to trade andforeign direct investment but strongly correlated with theinnovationsystem.Thismayindicatethatopennesstotradeandforeign capital andopennessto ideas do not necessarilygo hand in hand.

Most indicators load highly on one factor only. Butthe indicators of financial development load moderatelyon both innovation system and governance. A well-functioning financial system is obviously important forinnovation,but it is arguably also intimately related to good

governance. Hence, although the quality of the innovationsystem and that of governance represent different aspectsof reality (and it is possible to distinguish between them)there is also a connection between the two.

5. Capabilities and development: evidence on the

relationship

Simple correlations may mask more complex relation-ships, so in the next step we carry out a multivariate

Fig. 2. GDP per capita and governance (average level over 20022004). Note: For definition of the governance variable see Table 1. Source: See Appendix A.


8/19


Fig. 3. GDP per capita and political system (average level over 20022004). Note: For definition of the political system variable see Table 1. Source: SeeAppendix A.

regression of the relationship between the four capabilitiespreviously identified and the level of economic develop-ment as reflected by GDP per capita. To avoid simultaneitybias in the estimates we use data from the initial period(average 19921994) for the four capabilities and the finalperiod (average 20022004) for GDP per capita.

As customary in the literature we include, in addition tothe capabilities mentioned above, a battery of exogenousvariables that are the result of processes in the distant pastand that areextremely hard or impossible to change within

a reasonable time frame (through political action). How-ever, since such exogenous variables may well influencethe formation of capabilities and economic development,it is important to take these into account in the analysisthat follows (to avoid biased inferences). Examples includevariables such as language, religion, ethnic divisions, colo-nial legacy as well as differences in geography and nature.After a screening of the recent literature on the role of suchexogenous variables for growth (Acemoglu et al., 2002;Alesinaetal.,2003;Bloometal.,2003;Fearon,2003;Gallup

Fig. 4. GDP per capita and openness (average level over 20022004). Note: For definition of the openness variable see Table 1. Source: See Appendix A.


9/19


Table 3

Regression resultslevels

Estimation method (1) (2) (3) (4) (5)OLS Iteratively

re-weighted leastsquares

Stepwise regression Stepwise regressionexcl. the poorestquartile

Stepwise regressionexcl. the poorest half

Constant 0.02 (0.78) Innovation system 0.85*** (18.14) 0.86*** (19.73) 0.71*** (14.02) 0.67*** (9.83) 0.76*** (11.00)

Governance 0.16*** (3.35) 0.13*** (3.08) 0.18*** (3.93) 0.30*** (4.91) 0.28*** (4.48)Political system 0.04 (1.08) 0.04 (1.31) Openness 0.04 (1.07) 0.04 (1.25) Geography, nature and history No No Yes Yes YesF 336.83 268.32 218.79 156.37 97.47R2 0.90 0.92 0.90 0.93Observations 115 115 115 86 57

Note: Dependent variable is log of average GDP per capita over 20022004 (PPP, constant 2000 USD). The independent variables of capabilities (factorscores) are their lagged levels from the period 19921994. Absolute value of robust t-statistics in brackets; asterisks (*, **, ***) denote significance at the10%, 5% and 1% levels. Standardized variables used in the estimates (beta coefficients reported).

etal.,1999;MastersandMcMillan,2001;SachsandWarner,2001; Sachs et al., 2004), the following 13 variables wereselected: longitude of country centroid, latitude of country

centroid, log of surface area, access to ocean, land in desertecozone, landin tropicalecozone, log of population density,ethnic fractionalization, religion fractionalization, malariafatal risk, log of oil deposits per capita, log of the num-ber of people killed in natural disasters per capita and logof years since national independence (see Appendix A fordetails). Besides an ordinary least squares estimate, we testfor the robustness of the results with respect to thecompo-sition of the sample by re-estimating this relationship witha robust regression technique, iteratively reweighted leastsquares.10 To identify the specification with the best statis-tical properties we use a (backward) stepwise regression.11

We also provide separate results for samples that only

include high and medium income countries.Table 3 presents the results from the regression anal-

ysis. Note that so-called beta coefficients are reported,which allows for direct comparisons of parameter values(see Wooldridge, 2003, pp. 114115). The main result isthat irrespective of econometric method the developmentof the innovation system and the quality of governanceare positively and significantly associated with economicdevelopment. For the two remaining capabilities there isno significant relationship.

The introduction of indicators reflecting differences innature, geography and history led to a slight increase in theexplanatory power of themodel anda small decrease in the

correlation between GDP per capita and the developmentof the innovation system. Hence, these results confirmthat such differences do matterfor economic development.Moreover, the results indicate that one important reason

10 Iteratively reweighted least squares is a robust regression technique,which assigns a weight to each observation, with lower weights given tooutliers (rregcommand in Stata 9.2).

11 The aim of the stepwise procedure (stepwise reg command in Stata9.2) is to eliminate (insignificant) variables that do not contribute to theexplanatory powerof themodel(givena chosensignificance level). Ateachstep,the stepwise methodalso attempts to reintroducealready eliminatedvariables to control for a possibility that some of them might become

significant later on.We specified the thresholdfor removal at 20% level ofsignificance and the level for reintroducing a variable at 15%.

for this may be that differences in nature, geography andhistory influence the ability of a country to develop a well-functioning innovation system. The results are robust to

changes in the composition of the sample except that theeffect of governance nearly doubled when the least devel-oped countries were excluded.

Can the implications from the above analysis be sus-tained in a dynamic framework? Many contributions to theempirical literature on cross-country differences in growthperformance, despite theoretical differences, share a com-mon empirical framework. So-called Barro-regressions(Barro, 1991) consist of regressing economic growthagainstinitial GDP per capita and a number of other variablesthat may be deemed relevant. In this framework the GDPper capita variable measures the potential for catch-up (orconvergence), while the other variables represent aspects

that are assumed to condition the ability to exploit thispotential.12 As conditional variables, we include the capa-bility measures developed in this paper and the exogenousvariables reflecting differences in geography, nature andhistory.

The popularity of Barro-regressions in applied workprobably owes much to the fact that it is consistent withdifferent theoretical perspectives. As shown by Barro andSala-i-Martin (2003, pp. 274275), the inclusion of the GDPper capita variable may be consistent both with Solowstraditional neoclassical growth model (in which case thelevel of GDP per capita is assumed to reflect the capitalintensity of the economy) and a Schumpeterian perspec-

tive (with a low GDP per capita indicating a high potentialfor diffusion). However, different theoretical perspectivesmay lead to different conclusions with respect to whichconditional variables to include. For instance, until recentlyit was common to assume away level-effects in analyses

12 The first to introduce this techniqueto thestudy of growth and devel-opment was probably Cornwall (1976). Cornwall used GDP per capita tomeasurethe gapin technologybetweenfrontier andthe late-comercoun-tries. The higherthis gap, he pointed out, thehigher thepotentialfor highgrowth in late-comer countries through successful imitation of superiortechnology developed elsewhere. However, in his view, to exploit thisgap countries needed to do additional investments (and that is where

the conditional aspect comes in). See Fagerberg (1994) for an extendeddiscussion.


10/19


of economic growth. With the advent of new growth the-ory this changed because, according to this approach, ahigher level of, say, R&D may induce higher growth indef-initely. But level effects may also result from temporaryshocks, causing variables to deviate from their steady-statevalues.

Table 4 presents the results. In the first model weregressed the potential for catch-up (log of initial GDPper capita) and the changes in the capabilities on growthof GDP per capita. Upgrading of the innovation systemand the quality of governance appear to be highly sig-nificant and positive predictors, while the changes inthe political system and openness were found to beuncorrelated with growth. However, the model predictspoorly.

In the second model, we add the initial levels of thecapabilities along with their changes as possible condi-tional variables. Both the level and change of the quality ofthe innovation system and that of governance were shownto be positive and significant predictors, while as beforethe two remaining capabilities were found not to mattermuch. Moreover, the GDP per capita came out significantand with the expected (negative) sign. Also the explana-tory power of themodel increased significantly. In thethirdmodel we applied, as before, a robust regression technique(iteratively reweighted least squares) to test for the impactof outliers. Not much changed with the exception that wenow found modest support for positive effects of changes(but not levels) in the degree of openness and the characterof the political system.

In the fourth model, we included the 13 exogenousvariables previously introduced, reflecting differences ingeography, nature and history. A (backward) stepwiseregression procedure was used to identify the model withthe best statistical properties. A healthy increase in theexplanatory power of the model was recorded, confirmingthat such exogenous variables do indeed matter for devel-opment. However, with these external variables includedthe estimated impact of the level of the innovation sys-tem is no longer significant. Arguably, these results lendsupport to the previous finding that unfavourable aspectsrelated to geography, nature and history may hamper thedevelopment of a well-working innovation system. As inthe previous case there is some support for an impact ofopenness and political system on growth (although for thelevels this time). Butcontrary to widespread expectations awestern type political system appears to have, if anything,a negative impact on growth.

Sincethepreviousexercise suggestedthat theresults forthe openness and political system variables may be sensi-tivetothecompositionofthesample,weincludeinthefifthand sixth columns a test for the robustness of the results totheremoval of thepoorest countries (first thepoorest quar-tile and then the poorest half). The most striking result isthat the political system (degree of westernization) has adiametrically opposite effect in poor and rich economies.A significantly negative effect turns to a significantly pos-itive impact when only the richest half is included. Alsoopenness to imports and foreign direct investment seems

to matter more for the richer economies (although there issome evidence of a positive impact, though small, for the Tab

le

4

Reg

ressionresultsgrowth

Model

(1)

(2)

(3)

(4)

(5)

(6)

Basicmodel

Initiallevels

Iterativelyre-weighted

leastsquares

Stepwiseregression

Stepwiseregressionexcl.

thepoorestquartile

S

tepwiseregressionexcl.

thepooresthalf

Con

stant

0.01(0.20)

Log

oftheinitialGDPpercapita

0.18**(2.19)

0.76**(2.16)

0.76***(3.23)

0.51***(2.93)

0.74***(2.80)

0.89***(3.39)

Innovationsystem

0.74**(2.28)

0.71***(3.17

)

0.43*(1.69)

0

.56**(2.07)

Gov

ernance

0.39***(2.68)

0.44***(3.80

)

0.36**(2.45)

0.46**(2.50)

0

.43**(2.34)

Politicalsystem

0.07(0.57)

0.06(0.65)

0.21*(1.94)

0.41***(2.94)

0

.34***(2.71)

Ope

nness

0.07(0.64)

0.04(0.45)

0.18**(2.15)

0.17*(1.79)

0

.37***(3.12)

innovationsystem

0.31***(3

.24)

0.48***(4.67)

0.49***(5.77

)

0.27***(3.20)

0.33**(2.45)

0

.30**(2.55)

g

overnance

0.36***(3

.80)

0.38***(3.87)

0.47***(6.70

)

0.34***(3.14)

0.48***(4.70)

p

oliticalsystem

0.10(1.43

)

0.12(1.32)

0.15*(1.98)

0

.24**(2.08)

o

penness

0.03(0.21

)

0.03(0.20)

0.20**(2.16)

Geo

graphy,natureandhistory

Yes

Yes

Y

es

F

4.70

4.20

10.51

6.80

7.60

2

7.74

R2

0.19

0.30

0.43

0.52

0

.73

Obs

ervations

115

115

115

115

86

5

7

Not

e:DependentvariableisannualgrowthofGD

Ppercapita(PPP,constant2000USD)over1

9922004.Absolutevalueofrobustt-statisticsinbrackets;asterisks(*,**,***)denotesign

ificanceatthe10%,5%and

1%levels.Standardizedvariablesusedintheestimates(betacoefficientsreported).


11/19


larger sample as well). Moreover, the explanatory powerof the model increases with development. One possibleexplanation is that there is more diversity among thepoorer economies in how their economies work. But itmay also have to do with measurement problems, whicharguably may be expected to be more severe in poor coun-tries.

The possibility of a endogeneity bias in the estimates,due to a possible feedback from economic growth oncapability changes, was investigated by the Hausman (orDurbinWuHausman) procedure (for further details seeWooldridge, 2002, pp. 118122).13 The test failed to detectevidence of endogeneity bias.

6. Concluding remarks

In recent years, the quality and availability of dataon different aspects of development have improved, andthis might give researchers in this area new opportunitiesfor investigating the reasons behind the large differences

in economic performance across countries. Based on areview of the existing theoretical and empirical literatureand an extensive search for available data sources thispaper has, with the help of factor analysis, identified a setof capabilities which following (one or more of) thetheoretical perspectives that have been advanced in thisarea should be assumed to be of critical importance forcatch-up:

The development of the innovation system.The quality of governance.The character of the political system.The degree openness to trade and foreign direct invest-

ment.

The empirical analysis suggests that a well-developedinnovation system is essential for countries that wish tosucceed in catch-up. There is a strong, significant androbust statistical relationship between (level and changeof) GDP per capita on the one hand, and (level and changeof) the innovation system on the other. Historical anddescriptive evidence also suggest that countries that havesucceeded in catch-up have given a high priority to thisdimension of development (Wade, 1990; Nelson, 1993;Kim, 1997).

Although a well-functioning innovation system

emerges from the analysis as an essential prerequisitefor development, it is not sufficient. Good governance isalso critical for the ability to realize the desired economicresults. Sometimes it is asserted that this is mainly a ques-tion of successfully westernising the political system,e.g., adapting to the institutional arrangements that have

13 The following exogenous variables of geography, history and naturethatwereexcluded fromthe bestmodelby thestepwiseregression wereused as instruments: latitude of country centroid, ethnic fractionaliza-tion, malaria fatal risk, land in desert ecozone and log of the number ofpeople killed in natural disasters per capita (see Appendix A for details).All of them proved to be valid instruments according to the criteria of

the method, e.g., they were significantly correlated to at least one of thepotentially endogenous variables.

proved to be successful in the United States and otherwestern democracies. This study, confirming previousresearch by Barro (1996) and Glaeser et al. (2004), findsthe support for these assertions to be rather weak. On thecontrary we show that it is among the richer economiesthat a political system of the Western type is shownto be conducive to growth. For the poor countries it is,if anything, the other way around. In fact, among thecountries that over the years have succeeded in catchingup there are several examples of countries with institu-tional arrangements that differ a lot from western ideals(such as the recent performance of China and Vietnam,the Asian Tigers before the 1990s or pre-world-war-twoJapan).

Another result from the study is that there is little sup-port for the argument that differences in openness to tradeand foreign direct investments matter much for devel-opment. This holds even when, as in the present case,country size is controlled for. The result is consistent withprevious research by Rodrik et al. (2004). It also accordswith previous findings by Xu (2000) and Dunning andNarula (2000) that poor countries due to lack of absorp-tive capacity are much less likely than other countries tobenefit from foreign direct investments. Although a posi-tive correlation between openness and growth is reportedin some cases, the degree of correlation is low, and sen-sitive to changes in the composition of the sample. Again,it is among the richer economies that openness to tradeand foreign direct investment seems to matter most forgrowth.

In general the picture that emerges from this study isone of a global knowledge-based economy with strongelements of endogenous growth. Countries that succeedin developing and sustaining strong innovation capabili-ties and well-functioning systems of governance do welleconomically while those that fail tend to fall behind.However, many countries in the poorer part of the globefind it hard to develop the capabilities, a well-functioninginnovation system in particular, necessary for joining thisvirtuous dynamics. An innovation system is something thatis built incrementally over many years. The results reportedhere suggest that many present-day poor countries havebeen hampered in developing an innovation system byunfavourable aspects related to geography, nature and his-tory. This arguably provides an additional argument fordevelopment aid, since some countries are much worseplaced than others for reasons that are beyond the controlof people living today.

Acknowledgements

Earlier versions of this paper were presented theUNIDO Industrial Development Report 2005 Workshop inVienna, 1112 May 2005, the 11th International JosephA. Schumpeter Society Conference in Nice/Sophia Antipo-lis, 2124 June 2006 and the 4th Globelics Conference,Trivandrum, 47 October 2006. We wish to thank theparticipants at these events and the editors and referees

of this journal for useful suggestions. All usual caveatsapply.


12/19


Appendix A. Data and sources

A brief overview of definitions, sources andtime/country coverage of the indicators is given inthe table below. The main source of data is the World Bank(World Development Indicators 2006) complemented bydata from othersources such as UNESCO, UNCTAD,ISO, ITU,Heritage Foundation etc. National sources were only usedfor Taiwan (when necessary). Due to problems of reportingin the early years only data from 1993 onwards were usedfor science and engineering articles from the membercountries of the former Soviet Union and Yugoslavia. Forsome indicators we reversed the scale, while keepingthe original range, to have the indicator in increasingorder (with low value signalling a low level and viceversa).

We originally collected data for all independent states(approximately 175 countries) and a large pool of indica-tors(approximately100 indicators).Thescreeningrevealedthat a group of (mostly least developed) countries sufferedfrom many missing data. Similarly, data for a large numberof relevantindicatorswereavailable only fora groupof high(medium) incomecountries and/oronly forthe most recentperiod. Some indicators suffered from high volatility (pri-marily in developing world), methodological changes overthe period or were merely variations of each other. In orderto strike a balance between breath of indicators and coun-try coverage we limited the analysis to 115 countries (seeAppendix B for the full list) and 25 indicators (plus the 14

exogenous variables reflecting differences in geography,nature and history). We use 3-year averages (19921994and 20022004) to limit influence of shocks and measure-ment errors occurring in specific years.

Although the selected indicators have broad cover-age, in some cases there were missing values that hadto be dealt with. Missing values were most frequent forpersonal computers, the teacherpupil ratio, market cap-italization of listed companies and some of the governanceindicators. A number of developing countries were notincluded in the early surveys of computers and Internetpenetration. We assumed that the initial value was zeroif the first reported value was less than one per millionpopulation (6 observations for computers and 18 obser-vations for Internet). Some missing data on the marketcapitalization were filled by information from Stock Mar-kets (www.escapeartist.com/stock/markets.htm ). Coun-tries without a stock market in a given period were givenzero (a total of 45 observations).

Theremaining missing data were estimated using infor-mation on other indicators in the dataset and the imputeprocedure in Stata 9.2 (see the Stata 9 Manual for details).It should be noted that if the imputed data exceeded theminimum observed value of an indicator elsewhere, whichhappened in a few cases forindicators that areby definitiontruncated at zero, we have replaced the imputed values bythe minimum observed figure. The proportion of data esti-mated by this procedure is given in the last column of thefollowing table.

Indicator and definition Scaling Source Average over period % of dataestimated

Gross Domestic Product (GDP): GDP

converted to (constant 2000)international USD using purchasingpower parity rates (PPP)

Per capita World Bank (World Development

Indicators 2006)

19921994 and

20022004

0

USPTO patents: Number of utilitypatents (patents for invention)granted by the U.S. Patent andTrademark Office (USPTO). The originof a patent is determined by theresidence of the first-named inventor

Per capita USPTO 19921994 and20022004

0

Science and engineering articles:Counts of articles published injournals classified and covered byScience Citation Index (SCI) andSocial Sciences Citation Index (SSCI).The article counts are based onfractional assignments

Per capita U.S. National Science Foundation(Science and EngineeringIndicators 2006)

19921994 and20022003

0

ISO 9000 certifications: A family ofstandards approved by theInternational Standards Organization(ISO) that define a qualitymanagement and quality assuranceprogram

Per capita International Organization forStandardization (The ISO Surveysof ISO 9000 Certificates)

19921994 and20022003 0

Fixed line and mobile phonesubscribers: Telephone mainlinesconnecting a customers equipmentto the public switched telephonenetwork (PSTN) and users of portabletelephones that subscribe to anautomatic public mobile telephoneservice by cellular technologyproviding access to the PSTN

Per capita World Bank (World DevelopmentIndicators 2006), ITU (WorldTelecommunication IndicatorsDatabase 2005)

19921994 and20022004

0
http://www.escapeartist.com/stock/markets.htmhttp://www.escapeartist.com/stock/markets.htm


13/19



Internet users: Internet users arepeople with access to the worldwidenetwork

Per capita World Bank (World DevelopmentIndicators 2006), ITU (WorldTelecommunication IndicatorsDatabase 2005)

1994 and20022004

1

Personal computers: Computersdesigned to be used by a single

individual

Per capita World Bank (World DevelopmentIndicators 2006), ITU (World

Telecommunication IndicatorsDatabase 2005)

1994 and20022004

9

Primary school teacherpupil ratio:The number of primary schoolteachers (regardless of their teachingassignment) divided by the numberof pupils enrolled in primary school

Ratio World Bank (World DevelopmentIndicators 2006), UNESCO (GlobalEducation Digest 2005)

1991 and20022004

11

Secondary school enrolment: The ratioof the number of secondary studentsof all ages (gross) expressed as apercentage of the secondaryschool-age population

% gross World Bank (World DevelopmentIndicators 2006), UNESCO (GlobalEducation Digest 2005)

1991 and20022004

3

Tertiary school enrolment: The ratio ofthe number of tertiary students of allages (gross) expressed as apercentage of the tertiary school-age

population

% Gross World Bank (World DevelopmentIndicators 2006), UNESCO (GlobalEducation Digest 2005)

1991 and20022004

3

Domestic credit to private sector:Financial resources provided to theprivate sector such as through loans,purchases of non-equity securities,and trade credits and other accountsreceivable, that establish a claim forrepayment

% of GDP World Bank (World DevelopmentIndicators 2006)

19921994 and20022004

0

Market capitalization of listedcompanies: The share price times thenumber of shares outstanding (alsoknown as market value) ofdomestically incorporatedcompanies listed on the countrysstock exchanges at the end of theyear


19921994 and20022004

4

Merchandise imports: The c.i.f. value ofgoods received from the rest of theworld. Goods simply beingtransported through a country (goodin transit) or temporarily admitted(except for goods for inwardprocessing) are not included


19921994 and20022004

0

Foreign direct investment inwardstock: A received investment thatinvolves a long-term relationshipand reflects a lasting interest in andcontrol by a resident entity in oneeconomy of an enterprise resident ina different economy

% of GDP UNCTAD (FDI Database 2005) 19921994 and20022003

1

Impartial courts: The degree to which atrusted legal framework exists for

private businesses to challenge thelegality of government actions orregulation

Index(010)

Gwartney and Lawson(2004)based on the WEF (Global

Competitiveness Report, variousissues)

1995 and20022003

8

Law and order: The degree to whichthe citizens of a country are willingto accept the established institutions,to make and implement laws andadjudicate disputes

Index(010)

PRS Group (International CountryRisk Guide, various issues)dataretrieved from Henisz (2005)

19921994 and20022004

8

Property rights: The degree to which acountrys laws protect privateproperty rights and the degree towhich its government enforces thoselaws. The scale of the indicator hasbeen reversed into increasing order,while keeping its original range

Iindex(15)

Heritage Foundation (Index ofEconomic Freedom, variousissues)based on the EconomistIntelligence Unit (CountryCommerce and Country Reports)

1995 and20022004

1


14/19



Regulation: How easy or difficult it isto open and operate a business. Thescale of the indicator has beenreversed into increasing order, whilekeeping its original range

Index (15) Heritage Foundation (Index ofEconomic Freedom, variousissues)based on the EconomistIntelligence Unit (CountryCommerce and Country Reports)

1995 and20022004

2

Informal market: The perceptions of

people with regard to the extent ofcorruption, defined as the misuse ofpublic power for private benefit. Thescale of the indicator has beenreversed into increasing order, whilekeeping its original range

Index (15) Heritage Foundation (Index of

Economic Freedom, variousissues)based on the TransparencyInternational (CorruptionPerceptions Index)

1995 and

20022004

1

Index of democracy and autocracy: Thedegree of democracy versusautocracy (POLITY2 variable inincreasing order from autocracy todemocracy)

Index (10to 10)

Marshall and Jaggers (2003) 19921994 and20022003

0

Political constraint: The extent towhich a change in the preferences ofany one actor may lead to a change ingovernment policy (POLCONIIIvariable)

Index (01) Henisz (2002, 2005) 19921994 and20022004

0

Legislative index of politicalcompetitiveness: Competitiveness ofelections into legislative branches(LIEC variable)

Index (17) Becket al.(2001) and Keefer (2005) 19921994 and20022004

0

Executive index of politicalcompetitiveness: Competitivenessfor post in executive branches ingovernment (EIEC variable)

Index (17) Becket al.(2001) and Keefer (2005) 19921994 and20022004

0

Political rights: The degree to whichpeople participate freely in thepolitical process derived fromstandards by the UniversalDeclaration of Human Rights. Thescale of the indicator has beenreversed into increasing order, whilekeeping its original range

Index (17) Freedom House (Index of Freedomin the World, various issues)

19921994 and20022004

0

Civil liberties: The degree of thefreedoms of expression and belief,associational and organizationalrights, rule of law, and personalautonomy without interference fromthe state derived from standards bythe Universal Declaration of HumanRights. The scale of the indicator hasbeen reversed into increasing order,while keeping its original range

Index (17) Freedom House (Index of Freedomin the World, various issues)

19921994 and20022004

0

Longitude of country centroid:Longitude is measured from thePrime Meridian with positive valuesgoing east and negative values goingwest

Degrees Gallup et al. (1999)-CID GeographyDatasets

Fixed 0

Latitude of country centroid: Latitude

is measured from the equator, withpositive values going north andnegative values going south

Degrees Gallup et al. (1999)-CID Geography

Datasets

Fixed 0

Surface area: Countrys total area,including areas under inland bodiesof water and some coastal waterways

log of km2 World Bank (World DevelopmentIndicators 2006)

Fixed 0

Access to ocean: Proportion of landwithin 100km of the ocean coastline,excluding coastline in the arctic andsub-arctic region above the winterextent of sea ice

% Gallup et al. (1999)-CID GeographyDatasets

Fixed 0

Land in desert ecozone: Proportion ofland in (temperate or tropical) desertecozone

% Gallup et al. (1999)-CID GeographyDatasetsbased on the UNEP

Fixed 0

Land in tropical ecozone: Proportion ofland in tropical ecozone

% Gallup et al. (1999)-CID GeographyDatasetsbased on the UNEP

Fixed 0


15/19



Population density: Midyearpopulation divided by land area

log ofpeople perkm2

World Bank (World DevelopmentIndicators 2006); missing datafilled from UNEP (The GEO DataPortal)

19922004 0

Malaria fatal risk: Proportion ofpopulation at risk of contracting

falciparum malaria

% Earth Institute (Jeffrey D. SachsMalaria Dataset)

1996 0

Ethnic fractionalization: Theprobability that two randomlyselected people from a given countrywill not belong to the same ethicgroup

Index (01) Alesina et al. (2003) The latest available 0

Religion fractionalization: Theprobability that two randomlyselected people from a given countrywill not belong to the same religion

Index (01) Alesina et al. (2003) The latest available 0

Oil deposits: Proven crude oil reservesin billion barrels (bbl)

log of(bbl+1)per capita

The CIA World Factbook 2005 The latest available 0

Killed in natural disasters: Number ofpersons killed (confirmed as dead,missing and presumed dead) in

disasters of natural origin (droughts,earthquakes, extreme temperatures,floods, slides, waves, wind storms,etc.)

log of killedper capita

UNEP(The GEO Data Portal)basedon the OFDA/CRED InternationalDisaster Database 2004

19922004 0

National independence: Number ofyears since gaining nationalindependence over the period18162004 (maximum truncated at188 years)

log of years Fearon (2003); missing data filledfrom the CIA World Factbook

18162004 0

Appendix B. Factor analysis

Factor analysis is carried out in two steps. First the fac-

tors (or dimensions) are identified and a solution is foundin terms of number of factors to be retained. Principal-component factors estimation, which has been shown tobe robust to different assumption on the distribution of thedata, waschosen. The so-called Kaiser criterion (eigenvalueabove unity) was used to determine the number of factorsto be retained. The solution was also found to be consistentwith a scree test. Second, loadings of the various indica-tors on the retained factors are adjusted through so-calledrotation to maximize the differences between them. Thecommonly used orthogonal rotations suchas varimaxnor-malized rotation, assume that the underlying factors are(totally) uncorrelated. However, in the real world, there is

a good deal of correlation between different factors. Forthis reason, we use the more flexible oblique rotation(see Fabrigar et al., 1999 for a detailed discussion). To inter-pret the results both the pattern and the structure matrixwere analyzed, but since the interpretations do not differqualitatively, we report only the former. Note that if the

orthogonality assumption holds, oblique and orthogonalrotations generate very similar solutions.

Outliers in the data set can impact correlations signif-icantly and thus distort factor analysis. Simply excludingthe outliers from the sample may not be the best solution,as we then may lose important evidence. Instead, we useall of the variables in logs. After the log-transformation ofthe dataset Mahalanobis distancecriteriondid notidentifyany outliers (Mahalanobis, 1936).

A characteristic of the factoring procedure is that thevariables are standardized by deducting the mean anddividing by the standard deviation. We used the meanand standard deviation of the pooled data (from the ini-tial and the final period), which means that the change of acomposite variable over time reflects both changes in each

countrys position (relative to other countries) and changesin the importance of the underlying indicators (over time)in the composite. In this way, we take into account as muchinformation as possible.

See below for the factor scores by country.


16/19



19921994 20002004 19921994 20002004 19921994 20002004 19921994 20002004

Developed countriesAustralia 0.90 1.62 1.38 1.43 0.82 0.78 0.37 0.67Austria 0.98 1.31 1.39 1.29 0.73 0.79 0.85 0.35Belgium 0.86 1.43 1.28 0.96 0.85 0.84 0.62 1.38Canada 1.01 1.26 1.63 1.40 0.76 0.78 0.67 1.48Denmark 0.97 1.57 1.60 1.67 0.81 0.81 0.89 0.10

Finland 1.11 1.42 1.23 1.49 0.79 0.83 0.82 0.35France 1.06 1.30 1.19 0.86 0.66 0.82 0.60 0.43Germany 0.93 1.34 1.62 1.28 0.71 0.79 0.96 0.11Greece 0.53 1.26 0.60 0.15 0.65 0.68 0.83 0.34Ireland 0.55 1.24 1.33 1.28 0.80 0.82 0.96 1.25Israel 0.89 1.47 1.00 0.90 0.67 0.69 1.13 0.05Italy 0.85 1.50 1.00 0.44 0.71 0.75 1.34 0.28 Japan 0.94 1.44 1.53 0.78 0.77 0.82 3.23 1.77Netherlands 0.98 1.45 1.63 1.38 0.83 0.75 0.17 1.20New Zealand 0.90 1.33 1.63 1.46 0.74 0.83 0.19 0.53Norway 1.01 1.51 1.41 1.24 0.80 0.81 0.41 0.24Portugal 0.44 1.30 0.91 0.69 0.72 0.74 0.13 0.46Spain 0.60 1.38 0.89 0.61 0.78 0.81 0.55 0.55Sweden 1.17 1.61 1.41 1.34 0.74 0.80 0.56 0.70Switzerland 1.05 1.48 1.38 1.40 0.85 0.86 0.35 0.46United Kingdom 0.88 1.50 1.83 1.49 0.70 0.77 0.11 0.26United States 1.09 1.47 1.70 1.49 0.80 0.79 0.74 0.03

Asian TigersKorea 0.46 1.23 0.88 0.53 0.62 0.70 1.13 0.01Singapore 0.59 1.22 1.51 1.51 0.56 0.53 1.45 1.85Taiwan 0.61 1.31 1.59 0.79 0.16 0.69 0.64 0.17

New EU MembersCzech Republic 0.31 1.08 0.73 0.16 0.74 0.73 0.06 1.28Estonia 0.15 1.12 0.65 0.70 0.07 0.67 0.51 1.59Hungary 0.34 1.21 0.97 0.40 0.69 0.69 0.49 1.16Latvia 0.03 1.03 0.06 0.02 0.26 0.69 1.14 0.59Lithuania 0.03 0.99 0.11 0.17 0.17 0.70 1.04 0.78Poland 0.17 1.10 0.42 0.10 0.48 0.73 1.26 0.48Slovakia 0.18 0.94 0.92 0.11 0.01 0.74 0.64 1.19Slovenia 0.54 1.36 0.02 0.47 0.69 0.77 0.49 0.16

West AsiaBahrain 0.28 0.84 1.68 1.19 3.05 2.43 0.21 0.42Iran 0.48 0.61 1.31 1.99 1.69 0.75 0.81 0.62Jordan 0.32 0.66 0.84 0.39 0.60 0.94 0.01 1.02Kuwait 0.06 1.03 1.46 0.97 1.33 1.11 2.41 1.82Oman 0.69 0.35 1.42 0.74 2.63 2.43 0.06 0.49Saudi Arabia 0.03 1.00 1.44 0.39 3.75 3.60 0.02 0.05Syria 0.73 0.17 0.65 1.37 2.47 1.79 0.93 0.06Turkey 0.32 0.64 0.26 0.54 0.39 0.38 0.78 0.47United Arab Emirates 0.18 0.98 1.60 1.07 2.82 2.87 0.09 0.22

Latin AmericaArgentina 0.20 1.06 0.43 1.09 0.66 0.69 1.45 0.15Bolivia 0.63 0.44 0.79 1.53 0.57 0.61 0.28 1.07Brazil 0.12 0.67 0.28 0.28 0.58 0.71 0.95 0.24Chile 0.01 0.63 1.15 0.92 0.69 0.81 0.22 1.31

Colombia 0.26 0.47 0.68 1.10 0.58 0.26 0.70 0.41Costa Rica 0.20 0.52 0.33 0.12 0.75 0.65 0.07 0.67Dominican Republic 0.77 0.08 0.58 1.27 0.38 0.58 0.72 0.36Ecuador 0.50 0.39 0.58 1.52 0.42 0.48 0.69 0.67El Salvador 0.95 0.06 0.07 0.35 0.48 0.46 1.06 0.32Guatemala 1.01 0.01 0.53 1.63 0.05 0.31 0.43 0.25Guyana 1.11 0.03 0.56 0.52 0.47 0.55 1.08 2.43Honduras 1.06 0.12 0.42 1.33 0.51 0.50 0.12 0.89Jamaica 0.51 0.45 0.09 0.49 0.58 0.57 0.05 0.79Mexico 0.16 0.49 0.11 0.48 0.18 0.66 0.08 1.06Nicaragua 0.77 0.08 0.92 1.40 0.21 0.47 0.17 1.32Panama 0.26 0.68 0.13 0.72 0.55 0.78 0.06 0.58Paraguay 1.00 0.23 0.26 2.06 0.43 0.48 0.69 0.61Peru 0.35 0.56 0.99 1.27 0.01 0.65 1.38 0.16Trinidad and Tobago 0.21 0.60 0.90 0.25 0.88 0.55 0.64 0.64


17/19



19921994 20 0020 04 19921994 20 0020 04 19921994 20 0020 04 1992199 4 20 0020 04

Uruguay 0.02 0.75 0.18 0.04 0.76 0.88 1.55 0.35Venezuela 0.04 0.89 0.32 2.52 0.53 0.50 0.42 0.23

East Europe and CISAlbania 0.91 0.06 0.74 1.46 0.27 0.47 1.43 0.52Armenia 0.45 0.39 0.24 0.75 0.15 0.05 1.23 0.22

Belarus 0.08 0.94 0.18 1.49 0.65 2.42 3.20 0.64Bulgaria 0.11 0.93 0.12 0.73 0.51 0.68 1.18 0.89Croatia 0.21 1.13 0.84 0.65 0.33 0.48 1.14 0.85Georgia 0.29 0.60 1.30 1.41 0.60 0.24 3.34 0.31Moldova 0.55 0.42 0.66 0.42 0.48 0.32 1.80 1.00Romania 0.19 0.80 0.76 1.07 0.22 0.58 1.99 0.70Russia 0.01 1.03 0.37 1.19 0.10 0.19 1.85 0.62Ukraine 0.11 0.76 1.28 1.15 0.33 0.18 2.09 0.67

North AfricaAlgeria 0.85 0.03 0.48 1.12 2.49 0.36 1.04 0.04Egypt 0.20 0.58 0.94 0.45 0.86 0.90 0.19 0.03Morocco 0.77 0.10 0.88 0.10 1.03 0.95 0.11 1.07Tunisia 0.57 0.47 0.77 0.40 1.43 1.39 0.88 1.28

South-East AsiaChina 0.65 0.64 0.22 0.42 2.69 2.39 0.19 1.17

Indonesia 0.68 0.10 0.37 1.08 1.98 0.30 0.49 0.24Malaysia 0.03 0.82 0.94 0.28 0.08 0.06 1.67 2.14Philippines 0.48 0.31 0.48 1.13 0.42 0.54 0.33 1.09Thailand 0.33 0.61 0.79 0.26 0.00 0.55 0.48 1.46Vietnam 1.45 0.09 2.18 1.92 2.46 1.88 0.06 1.81

Sub-Sahara AfricaBenin 1.99 1.09 0.22 0.79 0.55 0.58 0.17 0.51Botswana 1.02 0.09 0.73 0.69 0.23 0.49 0.92 0.93Burkina Faso 2.24 1.46 0.88 0.51 1.35 0.38 1.29 1.15Cameroon 1.55 0.69 0.94 1.48 0.88 1.09 1.36 0.36Congo 1.55 1.16 1.53 1.60 0.12 0.89 0.08 1.11Cote dIvoire 1.16 0.46 0.05 1.25 0.93 0.49 0.29 0.62Ethiopia 2.01 1.41 0.42 0.76 1.83 0.11 1.49 1.08Ghana 1.48 0.77 0.21 0.24 0.78 0.56 0.15 1.16Guinea 2.18 1.31 0.06 1.29 2.12 0.31 1.01 0.42

Kenya 1.19 0.41 0.08 0.81 0.67 0.50 0.04 0.34Madagascar 1.91 1.29 0.80 0.84 0.38 0.53 1.27 0.15Malawi 2.42 1.47 0.03 0.19 1.41 0.32 0.03 0.92Mali 2.41 1.48 0.53 0.59 0.12 0.59 0.04 0.46Mozambique 2.46 1.52 0.77 1.17 1.25 0.38 0.39 1.82Namibia 1.02 0.19 0.79 0.71 0.50 0.30 1.42 1.64Niger 2.38 1.78 0.80 1.25 0.39 0.38 0.02 0.08Nigeria 1.37 0.47 0.02 1.34 2.03 0.30 0.67 0.76Senegal 1.55 0.85 0.09 0.50 0.04 0.52 0.75 0.70South Africa 0.21 0.51 0.71 0.34 0.31 0.76 0.32 1.05Tanzania 2.10 1.47 0.31 0.43 1.27 0.05 0.48 1.26Togo 1.67 0.76 0.70 1.79 1.41 0.85 0.66 0.72Uganda 1.90 0.99 0.39 0.44 1.70 0.72 1.49 0.32Zambia 1.38 1.14 0.39 0.47 0.26 0.21 0.61 1.59Zimbabwe 0.98 0.02 0.23 2.50 0.82 1.29 0.70 1.34

Central AsiaBangladesh 1.42 0.60 1.44 2.03 0.52 0.33 2.63 0.63India 1.01 0.15 0.03 0.10 0.43 0.62 2.16 0.41Mongolia 0.85 0.01 0.46 0.41 0.16 0.27 0.03 1.94Nepal 1.40 0.66 0.74 1.07 0.36 0.71 1.97 0.69Pakistan 1.28 0.57 0.44 0.74 0.36 1.63 0.80 0.14Sri Lanka 0.91 0.21 0.35 0.28 0.17 0.42 0.34 0.07

OceaniaFiji 0.56 0.40 0.13 0.39 0.08 0.33 0.12 0.21Papua New Guinea 1.47 0.82 0.43 0.07 0.58 0.65 0.63 1.96

Note: For definition of the variables see Table 1.


18/19


References

Abramovitz, M.,1986. Catchingup, forgingahead, andfalling behind.Jour-nal of Economic History 46 (386), 406.

Abramovitz, M., 1994a. The origins of the post-war catch-up and con-vergence boom. In: Fagerberg, J., Verspagen, B., von Tunzelman, N.(Eds.), The Dynamics of Technology, Trade and Growth. Edward Elgar,Aldershot, pp. 2152.

Abramovitz, M., 1994b. Catch-up and convergence in the postwar growth

boomand after.In: Baumol, W.J.,Nelson, R.R., Wolf,E.N. (Eds.), Conver-genceof ProductivityCross-NationalStudies and Historical Evidence.Oxford University Press, Oxford, pp. 86125.

Acemoglu, D., Johnson, S., Robinson, A., 2002. Reversal of fortune: geog-raphy and institutions in the making of the modern world incomedistribution. Quarterly Journal of Economic 117, 12311294.

Adelman, I., Morris, C.T., 1965. A factor analysis of the interrelationshipbetween social and political variables and per capita gross nationalproduct. Quarterly Journal of Economics 79, 555578.

Adelman, I., Morris, C.T., 1967. Society. Politics and Economic Develop-ment. The Johns Hopkins Press, Baltimore.

Aghion, P., Howitt, P., 1992. A model of growth through creative destruc-tion. Econometrica 60, 323351.

Alesina, A., Devleeschauwer, A., Easterly, W., Kurlat, S., Wacziarg, R., 2003.Fractionalization. Journal of Economic Growth 8, 155194.

Amsden,A.H., 1989. Asias Next Giant: South Korea and Late Industrializa-tion. Oxford University Press, New York.

Archibugi, D., Coco, A., 2005. Measuring technological capabilities at thecountry level: a survey and a menu for choice. Research Policy 34,175194.

Barro, R.J., 1991. Economic growthin a cross sectionof countries.QuarterlyJournal of Economics 106, 407443.

Barro, R.J., 1996. Democracy and growth. Journal of Economic Growth 1,127.

Barro, R.J., Sala-i-Martin, X., 2003. Economic Growth, second ed. The MITPress Cambridge, MA.

Basilevsky, A., 1994. Statistical Factor Analysis and RelatedMethods: The-ory and Applications. John Wiley & Sons Inc., London.

Baumol, W.J., Batey Blackman, S.A., Wolff, E.N., 1989. Productiv-ity and American Leadership: The Long View. The MIT Press,Cambridge, MA.

Beck, T., Clarke, G., Groff, A., Keefer, P., Walsh, P., 2001. New tools in com-parative political economy: the database of political institutions. TheWorld Bank Economic Review 15, 165176.

Benhabib, J., Spiegel, M.M., 1994. The role of human capital in economicdevelopment: evidence from aggregate cross-country data. Journal ofMonetary Economics 34, 143173.

Bloom, D.E., Canning, D., Sevilla, J., 2003. Geography and poverty traps.Journal of Economic Growth 8, 355378.

Cincera, M., Van Pottelsberghe de la Potterie, B., 2001. International R&Dspillovers: a survey. Cahiers Economiques de Bruxelles 169, 332.

Coe, D., Helpman, E., 1995. International R&D spillovers. European Eco-nomic Review 39, 859887.

Cohen, W.M., Levinthal, D.A., 1990. Absorptive capacity: a new perspec-tive on learning and innovation. Administrative Science Quarterly 35,128152.

Cornwall, J., 1976.Diffusion, convergence and Kaldors law. EconomicJour-nal 85, 307314.

Dahlman,C.J., Ross-Larson, B., Westphal, L., 1987. Managing technologicaldevelopment.Lessons from the newlyindustrialized countries. WorldDevelopment 15, 759775.

Dosi, G., Pavitt, K., Soete, L.G., 1990. The Economics of Technical Changeand International Trade. Harvester Whetsheaf, London.

Dunning, J.H., Narula, R., 2000. Industrial development, globalizationand multinational enterprises: new realities for developing countries.Oxford Development Studies 28, 141167.

Edquist, C., 1997. Systems of Innovation: Technologies. Institutions andOrganizations, Pinter, London.

Edquist, C., 2004. Systems of innovation: perspectives and challenges. In:Fagerberg, J., Mowery, D., Nelson, R. (Eds.), The Oxford Handbook ofInnovation. Oxford University Press, Oxford, pp. 181208.

Fabrigar, L.R., Wegener, D.T., MacCallum, R.C., Strahan, E.J., 1999. Evaluat-ing the use of exploratory factor analysis in psychological research.Psychological Methods 4, 272299.

Fagerberg, J., 1987. A technology gap approach to why growth rates dif fer.Research Policy 16, 8799.

Fagerberg, J., 1988. International competitiveness. Economic Journal 98,355374.

Fagerberg, J., 1994. Technology and international differences in growthrates. Journal of Economic Literature 32, 11471175.

Fagerberg, J., Godinho, M.M., 2004. Innovation and catching-up. In: Fager-berg, J., Mowery, D., Nelson, R. (Eds.), The Oxford Handbook ofInnovation. Oxford University Press, Oxford, pp. 514544.

Fearon, J.D., 2003. Ethnic and cultural diversity by country. Journalof Eco-nomic Growth 8, 195222.

Fransman,M.,1982.Learningandthecapitalgoodssectorunderfreetrade:the case of Hong Kong. World Development 10, 9911014.

Fransman, M., King, K., 1984. Technological Capability in the Third World.Macmillan Press, London.

Freeman, C., Clark,J., Soete, L.G.,1982. Unemployment and TechnicalInno-vation: A Study of Long Waves and Economic Development. Pinter,London.

Freeman, C., 1987. Technology Policy and Economic Performance: Lessonsfrom Japan. Pinter, London.

Furman, J.L., Porter, M.E., Stern, S., 2002. The determinants of nationalinnovative capacity. Research Policy 31, 899933.

Furman, J.L., Hayes, R., 2004. Catching up or standing still? National inno-vative productivity among follower countries. Research Policy 33,13291354.

Gallup,J.L., Sachs,J.D., Mellinger,A., 1999.Geographyand EconomicDevel-opment. Harvard University. CID Working Paper no. 1/1999.

Gerschenkron,A., 1962. Economic Backwardnessin Historical Perspective.The Belknap Press, Cambridge, MA.

Glaeser, E.L., La Porta, R., Lopez-de-Silanes, F., Schleifer, A., 2004. Do insti-tutions cause growth? Journal of Economic Growth 9, 271303.

Grossman, G.M., Helpman, E., 1991. Innovation and Growth in the GlobalEconomy. The MIT Press, Cambridge, MA.

Gwartney, J., Lawson, R., 2004. Economic Freedom of the World: 2004Annual Report. The Fraser Institute, Vancouver. Data retrieved fromwww.freetheworld.com.

Henisz,W.J.,2002. The institutional environmentfor infrastructureinvest-ment. Industrial and Corporate Change 11, 355389.

Henisz, W.J., 2005. POLCON Database 2005. Wharton School of the Uni-versity of Pennsylvania, Philadelphia.

Johnson, C.A., 1982. MITI and the Japanese Miracle: The Growth Of Indus-trial Policy, 19251975. Stanford University Press, Stanford.

Johnson, A., Jacobsson, S., 2003. The emergence of a growth industry: acomparative analysis of the German, Dutch and Swedish Wind Tur-bineIndustries.In: Metcalfe, S.,Cantner,U. (Eds.), Transformations andDevelopment: Schumpeterian Perspectives. Physical/Springer, Heidel-berg.

Keefer, P., 2005. Database of Political Institutions: Changes and VariableDefinitions. World Bank, Development Research Group, New York.

Keller, W., 2004. International technology diffusion. Journal of Economic

Literature 42,

Date post:	06-Apr-2018
Category:	Documents
Upload:	johan-camilo-rondon-amaya
View:	223 times
Download:	0 times

Fagerberg2008-National Innovation Systems, Capabilities and Economic Development

Documents