Per pust ak aan Su l t anah Z an ar i ah

ARTICLESFOR

UTM SENATE MEMBERS

Creativity & InnovationTITLE : SOURCE

From Creativity to Innovation http://www.sciencedirect.com/

The Role of Creative Innovation in Economic Growth : Some International Comparisons

http://www.sciencedirect.com/

Organizational and Institutional Influences on Creativity in Scientific Research

http://www.sciencedirect.com/

Transformational Leardership, Creativity, and Organizational Innovation

http://www.sciencedirect.com/

02 September 2009SOURCE : PERPUSTAKAAN SULTANAH ZANARIAH

Per pust ak aan Su l t anah Z an ar i ah

TITLE : SOURCE

From Creativity to Innovation http://www.sciencedirect.com/

Technology in Society 31 (2009) 1–8

Contents lists available at ScienceDirect

Technology in Society

journal homepage: www.elsevier .com/locate / techsoc

From creativity to innovation

Shahid Yusuf*

Development Economics Research Group, The World Bank, 1818 H Street, NW, Washington, D.C. 20433, USA

Keywords:CreativityInnovationWikicapital

* Tel.: þ1 202 458 2339; fax: þ1 202 522 7533.E-mail address: syusuf@worldbank.org

1 Smil provides an enlightening tour of the innov2 See, for instance, Chapter 3 of OECD [2]; also

commercially successful innovations. Japan has theuniversities, invest in science, and induce researchemove beyond imitation and for young people to de

3 Innovations in financial, retailing, wholesaling,4 Since the mid-1980s, the crafting of new busi

experiment with ways to pare costs, increase flexib

0160-791X/$ – see front matter � 2008 Published bdoi:10.1016/j.techsoc.2008.10.007

a b s t r a c t

Talent is the bedrock of a creative society. Encouraging and developing talent involvesmobilizing culture and tradition, building institutions to increase the stock of humancapital, enhancing its quality, and instilling values that favor achievements and initiative.The productivity that emerges from this talent, in the form of ideas, can be increased bynurturing wikicapitaldthe capital arising from networks. Translating creativity intoinnovation is a function of multiple incentives, and sustaining innovation is inseparablefrom heavy investment in research. Ultimately, the transition from innovation tocommercially viable products requires the midwifery of many service providers and theentrepreneurial skills of firms small and large. � 2007 Elsevier Ltd. All rights reserved.

� 2008 Published by Elsevier Ltd.

1. Introduction

Commercially viable innovations are becoming the linchpin of success in global markets by helping to raise totalproductivity, and they account for a major portion of the growth in advanced and industrializing economies.1,2 Innovation cantake many forms, among which product innovation is but one. Design and incremental process innovations are morecommon, and in recent years myriad innovations have been introduced by providers of services.3 Innovation is changing thestructure and enhancing the capabilities of organizations.4 Moreover, institutional innovations are sharpening marketincentives for entrepreneurial activity and technology trading, which take new ideas, products, and practices into thecommercial domain.

In areas such as genetics, climatology, and the social sciences, innovative uses of computing power are making researchmore productive by automating the framing of multiple hypotheses, and their testing, using advances in data processing andevaluative algorithms [5]. The importance of innovations that improve economic performance and living conditions cannotbe over-emphasized in the face of the opportunities offered by globalization and the multiple challenges arising from scarceresources, dire predictions of accelerating climate change, and the threat of pandemics caused by new and resistantorganisms.

ations that transformed the twentieth century [1].Phelps [3], who suggests that the ‘‘root problem’’ among European economies is the low rate ofhighest ratio of patents per capita, but worries over waning creativity are spurring efforts to reformrs to be more adventurous and take risks. Likewise in China, great emphasis is placed on the need tovelop innovations in industry and art.and IT-based services have stimulated the growth of productivity in the U.S. over the past decade [4].ness models and organizational forms has become a flourishing industry that encourages firms toility, and raise productivity.

y Elsevier Ltd.

Innovation Economic

Growth

Development/

Commercialization

R&D

UrbanEnvironment

Institutions

Incentives

Corporatesystem

Creativity

Talentquality

Wikicapital

Culture

Finance

Fig. 1. Creativity and economic performance.

S. Yusuf / Technology in Society 31 (2009) 1–82

Innovation springs from the creative application of knowledge. Thus, it has two essential ingredients: creativitydartistic,scientific or other5 dand a stock of knowledge. Knowledge and the functionalities it supplies are the essential raw materials,but it is the creative act that is the basis of an innovation. Often an initial invention or deep insight6 is the first of several stagesbefore an innovation is fully realized, a process that often requires the accumulation of new knowledge.

Many innovations, however ingenious, have no market potential.7 Those that appear promising must be refined, tested,and modified before they are commercially ready. This can sometimes be a protracted process requiring additional inno-vations along the way, but it is almost as critical as the initial creative act itself. Repeated commercial success is a function oforganizational capability and the coordinated use of multiple skillsdmanagerial, financial, marketing, and legaldwhichthemselves draw support from a variety of institutions.

What, then, makes a society creative? How does this translate from innovation to superior economic performance? A vastand multi-stranded literature, grounded in several disciplines, yields many clues. But the conditions that induce innovationare complex, many are not easily altered by policy, and some are the result of historical and cultural evolution that is beyondthe influence of policy.

What follows (sketched in Fig. 1), opens an inevitably partial and highly synthetic window onto this literature. This paperis divided into three parts. The first presents some of the conditions that are correlated with creativity. The second partdefines factors that can lead from creativity to innovation. The third part summarizes those conditions that contribute to thecommercialization of innovations.

2. Achieving creativity

It almost goes without saying that culture and traditions strongly influence creative interest, the degree of creativity, andthe forms it can take.8 Not all societies gravitate toward or sustain a culture of systematic scientific enquiry grounded informal rules of logic, proof, and empirical validation of hypotheses. Creativity in some societies might be expressed throughart, music, and crafts, for example, or through institutions that ensure survival in harsh environments. While many forms ofcreativity can be valuable, the economic measure encourages creativity that ultimately leads from innovations to commercialresults. By and large, the scientific approach has proved to be overwhelmingly more fruitful in generating useable knowledge

5 Mokyr provides a historical perspective on creativity and how the competitive market for ideas that emerged in seventeenth-century Europecontributed to it [6].

6 According to Arthur [7], inventors start with a pressing need or a novel phenomenon and ‘‘think in terms of achievable actions and deliverable effects –functionalities – and they combine these in solving problems. Functionalities . are also the currency of standard technological design. But what differ-entiates invention is that the overall problem has not been satisfactorily solved before, that the challenges may run several recursive levels deep, that thesolutions of these may be far from standard, that novel phenomena and unusual effects may have to be used, and that the overall principle is new to thepurpose in question. What are common to originators is not genius or special powers. Rather it is the possession of a very large quiver of functionalities’’[p. 258].

7 This is reflected in the huge number of ‘‘dark patents’’ that lead to no useable outcomes.8 Feinstein notes that ‘‘Our creative interests are a vital, central link connecting our creative endeavors with our culture including our cultural heritage’’

([8], p. 470).

Table 1The 12 most economically competitive countries (2006).

Global competitiveness index World competitiveness scoreboard

Switzerland IcelandFinland DenmarkSweden AustraliaDenmark CanadaSingapore SwitzerlandUnited States LuxembourgJapan FinlandGermany IrelandNetherlands NorwayUnited Kingdom AustriaNorway SwedenIceland Netherlands

Source: IMD World Competitiveness Yearbook 2006; World Economic Forum.

S. Yusuf / Technology in Society 31 (2009) 1–8 3

that serves as the springboard for creative leaps to fruitful innovations.9 That approach is essential for a competitive economyand must be supported by large numbers of creative individuals imbued with the scientific ethos.

The ratio of creative people in a society is likely to be higher if three conditions are fulfilled:

1. The society attaches a high value to learning and promotes talent through cultural reinforcements such as a stable,nurturing family environment from early childhood through the formative years.

An analysis of scores obtained from the Trends in International Math and Science Study (TIMSS)10 suggests that such anupbringing (combined with adequate nutrition) strengthens cognitive and analytic capabilities and motivation, each ofwhich is critical to scholastic performance. The cultural environment of the home needs to be buttressed by qualityschooling provided from primary through tertiary levels.11 Better schooling is becoming a priority as technologicaladvances and computerization increase the demand for workers with superior problem-solving skills, the capacity toconvey complex information, and the ability to work effectively in groups [14,15]. Quality is partially a function of lab andclassroom facilities, and the enlightened use of IT to stimulate learning, pique curiosity, and enhance the ability to solveproblems without recourse to rules. More important, quality depends on the caliber of teachers [16] and on how classroominstruction is reinforced by encouragement and supplementary tutoring at home.12 These factors overshadow class size orthe length of the school year [17].

2. A creative society attaches equal importance to building human capital through better health, beginning from childhoodwhen good nutrition and care have profound and lasting implications for learning capacity in later life [18,19].

Physical health is one part of the picture, but mental well being is an equally important component. The burgeoninginterest in human happiness, while still in its infancy and beset by problems of measurement,13 points to a correlationbetween the perceived happiness of societies and their ranking based on indicators of international economic competi-tiveness.14 Table 1 presents the 12 most economically competitive countries in 2006 as ranked by the Global Competi-tiveness Index and the World Competitiveness Scoreboard. Table 2 presents the top 12 countries ranked by surveys thatmeasure satisfaction with life. The overlap is significant. Happy and healthy people are more likely to be productive, topursue knowledge more avidly, and put it to more ingenious uses. To express this in more conventional terms, it isintuitively plausible that creativity will be affected by the physical, emotional, and intellectual quality of human capital.Thus, one of the uppermost objectives for a creative society is to invest in human capital so as to raise both volume andquality. Lewis Thomas observed that every time he passed a laboratory and heard laughter, he saw it as a sign thatinteresting and surprising findings were being made.

3. With knowledge growing at an exponential rate, students and researchers must become more and more specialized inorder to achieve sufficient mastery over a narrow subfield to be able to advance the frontier of knowledge.

9 Mokyr [9] discusses the scientific revolution in Europedthe encouragement it gave to the sharing of knowledge, to empirical testing and its Baconianorientation, and to improve mankind’s material circumstances. The contrast between the pursuit of scientific knowledge by the West in the eighteenth andnineteenth centuries and the relative absence of such a search and of institutions to support the cumulative process of finding and learning in China areexamined by Landes [10].

10 The details of this analysis and a survey of related literature can be found in Yusuf et al. ([11], Ch. 5).11 The close relationship between the quality of education and economic outcomes is analysed and empirically supported by Hanushek and Kimko [12]

and Hanushek and Woßmann [13].12 Finland, which ranks high on student achievement scores, is noted not for the length of its school year but its emphasis on home tutoring and support.

After-school learning of math, reading, and communication skills is the focus of work by Murnane and Levy [14].13 See Frey and Stutzer [20] for a recent review of the methodological issues and findings from the research on happiness initiated by the ‘‘Easterlin

Paradox,’’ which came to light in the 1970s. See also Layard [21].14 There is also some weak evidence of a casual relationship running from happiness to economic growth [22]. Di Tella and MacCulloch [23] found that

both unemployment and inflation reduce happiness, with unemployment having the greater effect.

Table 2Top 12 countries as ranked by surveys that measure ‘‘satisfaction with life’’.

Satisfaction with life

DenmarkSwitzerlandAustriaIcelandAustraliaFinlandSwedenCanadaIrelandLuxembourgNorwayNetherlands

Source: Veenhoven, R., Average happiness in 95 nations 1995–2005, World Database of happiness.

S. Yusuf / Technology in Society 31 (2009) 1–84

This requirement has two implications: one, that individuals make their first discoveries at later ages than was the casein the past (which also reduces the productivity of innovators, especially if ability in certain fields is greatest at a youngerage) [24]; and two, an increasing proportion of discoveries are made by teams that were built through bridging rela-tionships [25]. Some of the most exciting new findings are the result of multi-disciplinary efforts that aggregate theknowledge, functionalities, and insights of professionals drawn from several fields. As Feinstein observes, ‘‘Creativity ismaking a connection between or combining two elements that have not previously been connected or combined’’ [8]. Infact, as knowledge deepens and becomes more varied, human capital becomes more creative as it is pooled into ‘‘wiki-capital’’15 through the formation of local and global teams, partnerships, associations, and learning societies that facilitatethe deepening and sharing of knowledge and bring together diverse talents with different perspectives, viewpoints, andspheres of knowledge.16 Creative solutions to complex problems become more feasible because wikicapital can harnessa vast array of expertise to tackle a problem from many directions by exploiting the possibilities of heterogeneity.17

Wikicapital is accumulating because so many scientists share a common medium of communication (English and math-ematics), they are increasingly mobile, and they have more opportunities for face-to-face contact. Moreover, collaborativework has been greatly facilitated by information and communication technology and the declining cost of Internet access[25].18 The creativity of wikicapital has been reinforced by advances in techniques for measuring physical and socialphenomena, the sophistication of measuring devices,19 the techniques for assembling and storing vast quantities of data,and by the automation of discovery in certain areas.

Wikicapital is also a function of what is sometimes called ‘‘emotional intelligence,’’20 and is an aspect of personality that isconditioned by culture and can be strengthened by an educational environment and training that attaches importance tocooperative behavior [33,34].

3. Making a creative society innovative

The quality of human capital and its enhanced creativity create preconditions; but catalyzing that innovation requirestriggers and mechanisms that reinforce certain types of productive behavior. One is a culture that is relatively tolerant of risktaking, specifically risks associated with entrepreneurial activity. Such a culture is also more tolerant of entrepreneurialfailure. Not only is business failure not stigmatized, but specific institutions (e.g., bankruptcy and limited liability laws) thatcontain penalties to be imposed on individuals when their business ventures fail, may also encourage fresh initiatives. Thus,

15 As the amount of collaboration has intensified, a subfield called ‘‘wikinomics’’ is emerging. It analyzes the modalities of collaboration and theirimplications for innovation [26]. The globalization of research, which has been gathering momentum since the 1990s, contributes to the accumulation ofwikicapital [27].

16 Arthur ([7], p. 285) observes that the Cavendish Laboratory at Cambridge was a fertile source of inventions in atomic physics because ‘‘it had builta treasury of knowings to do with atomic phenomena’’ and it provided an arena where new ideas could be debated, challenged and tested. The size ofresearch teams has grown steadily as the complexity of problems has increased. See Adams et al. [28].

17 The role of heterogeneity or diversity in helping to solve knotty problems and giving rise to striking innovations deserves emphasis [29]. The literatureon the ‘‘small world’’ phenomenon warns that ‘‘intense connectivity can homogenize the pool of material available to different groups, while at the sametime, high cohesiveness can lead to the sharing of common rather than novel information’’ ([30], p.449).

18 Fleming and Marx note that ‘‘inventor networks are shrinking becoming more ‘‘small world’’ [with] multiple overlapping ties of cohesion [which]engender trust’’ ([25], p. 10).

19 Galison [31] describes how advances in high-energy physics are now dominated by large, often cross-national, teams working with extremelyexpensive equipment that detects and measures physical phenomena.

20 Like IQ, there is now an emotional intelligence quotient (EQ), which measures the ability to manage one’s emotions and to perceive those of a group[21,32].

Table 3R & D expenditures and sectoral distribution in 2004.

Country Share of R&D by sector GERD as percent of GDP

Industry Higher education Government

Finland 70.1 19.8 9.5 3.51Japan 75.2 13.4 9.5 3.13Korea 76.7 9.9 12.1 2.85U.S. 70.1 13.6 12.2 2.68EU-25 63.3 22.1 13.4 1.81China 66.8 10.2 23.0 1.23

Source: OECD Science, Technology and Industry Outlook 2006.

S. Yusuf / Technology in Society 31 (2009) 1–8 5

the social attitude toward certain kinds of risk taking induces willingness on the part of individuals to be ambitious and tosearch for significant and disruptive innovations.

This attitude goes hand in hand with mechanisms for rewarding innovations if they prove to be commercially successful.These can take several forms. One is the assignment and protection of intellectual property rights, which allows individuals orentities to derive benefit from a discovery for a number of years.21 Societies can support creativity by making it possible toacquire intellectual property at an affordable cost (in terms of money and/or time) and ensure that once obtained the rightscan be (affordably) enforced by an effective legal system.22 A second form is the monetizing of intellectual property andknow-how through markets for technology [36,37].

The trading of know-how and commercialization of technology is supported by a variety of mechanisms: venture capital,government funding schemes for small and medium-size enterprises (SMEs) such as the Small Business Innovation Research(SBIR), the floating of initial public offerings (IPOs), and merger and acquisitions (M&As). Income and capital gains taxes thatare not steeply progressive can ensure that a substantial portion of the rewards of entrepreneurship and innovation accrue tothe innovators, while tax credits and generous depreciation allowances can encourage investment in innovative activities.Furthermore, a society that accepts so-called ‘‘good inequality’’ [38] can accommodate wider income differentials therebybuttressing the incentive mechanism that attaches great value to singular achievements. This acceptance of large income andwealth differentials can contribute to social mobility and neutralize the social disapproval of conspicuous consumption by therich.

For the incentive system to deliver sustained results, domestic and foreign competition is needed. Competitive pressuremotivates innovation to pursue commercial rewards and weeds out weak offerings and innovations that have outlived theirusefulness. Competition spurs innovation because many businesses find that it is often a surer means of earning higherreturns than competing solely on the basis of price or quality or servicedalthough those remain important. Thus, the framingand enforcement of rules (including trade policy) governing market competition, complement rules governing intellectualproperty. Together they influence the tempo of innovation, especially through the entry of firms that are a conduit for newproducts and services but also by providing enough latitude for firms large and small to benefit adequately from researchactivities and risk taking, for instance, in the pharmaceutical industry.

Culture, institutions, and incentive mechanisms serve as the matrix within which creativity can flourish and lead toinnovation. There is, however, a geographical locus for innovation, which is no less important than those intangible factors.Much of the creativity that leads to new discoveries and spawns innovation occurs in cities, and the bulk of this is in urbanareas that share certain attributes [39–41]. The hot spots of innovation are more often than not major urban centers23 that areclosely linked with other world cities and open to the circulation of people and ideas facilitated by efficient transport andinformation technology services. In many instances, this openness promotes diversity, which is associated with innovativeapproaches to problem solving [29]. Cities where innovation flourishes are also centers of learning, often home to leadinguniversities and training, and later employing some of the most talented people in the country. Universities and city-basedresearch institutes contribute to the process of discovery through a variety of channels while working with businesses totransform their scientific findings into marketable technologies [43].24 In major urban areas, the path from discovery toinnovation is greatly expedited by the co-location of universities, research institutes, and businesses, each of which perfectand utilize new technologies [45]. Discovery is aided further by the presence of providers of diverse and critical business

21 Jaffe and Lerner [35] observe that, ‘‘based in the Constitution itself, and codified in roughly its modern form in 1836, the patent system was an essentialaspect of the legal framework inwhich inventions, from Edison’s light bulb and the Wright brothers’ airplane to the cellphone and Prozac, were developed’’ [P.1].

22 However, the rules for patents and the process for screening submissions must be designed carefully so as not to award patents to commonplacefindings that give legal leverage to individuals and can stifle innovation. In recent years, many software and design patents have been awarded for triv-ialities that are only deemed new and non-obvious because a new domain and the technological infrastructure to support it have emerged. An easing ofsubmission rules and standards of examination has also occurred in the U.S. The standards for awarding patents must be upheld so as to avoid a flood oflitigation which, beyond a point, can be inimical to innovation [35].

23 Bettencourt et al. [42] show that the larger SMSAs in the U.S. are a more prolific source of patents and that there is evidence of superlinearity effects.24 The emergence of research-oriented, tertiary institutions and their impact on industrial development in the U.S., Europe, and East Asia is described by

Mazzoleni [44].

S. Yusuf / Technology in Society 31 (2009) 1–86

servicesdfinancial, legal, managerial, technical, and othersdwhich serve as handmaidens to the innovation process andassist in forming dynamic high-tech industrial clusters in the vicinity of world-class universities, such as in Cambridge,England, and California’s Silicon Valley [46–49]. Without these intermediaries, ideas would have far greater difficulty gainingcommercial footholds.

While large corporations usually have the resources and expertise to assimilate, develop, and market innovations, andare often actively looking for certain types of technologies,25 SMEs generally encounter much more difficulty in their effortsto access and adapt technologies. Small firms need all the help they can get, and intermediaries (such as business asso-ciations, specialized industrial extension agencies, regional development bodies, university-based technology transferoffices) located in cities are in a position to connect small firms to researchers and assist them in launching new productsor services [45].26 Large, relatively open cities also have more favorable demographic profiles and flexible labor markets.The population of growing cities usually has a lower median age, which is a source of entrepreneurial dynamism, inno-vativeness, and greater savings. These demographic characteristics have contributed substantially to earlier growth in EastAsia [52].

4. Developing and commercializing innovations

While human talent is the source of creativity, innovation also requires financing. Providing quality education is one part ofthe equation, but R&D financing (which enables human and wikicapital to generate innovations) is predicated on the amountof available funding and its distribution across basic and applied research. The various estimates of return on R&D spendingindicate that private returns average 28% while social returns are significantly higher [11,53,54]. Once the human capital baseis developed, and the creative potential deepened, the payoff from outlay on R&D can be handsome. In fact, economies thatrely on innovation to drive growth must be ready to invest between 2% and 3% of gross domestic product annually in R&D inorder to continually augment the stock of knowledge and ensure a sufficient flow of innovations (see Table 3).

Governments have tended to shoulder more of the financing burden for basic research in universities and researchinstitutes, which adds to the pool of scientific knowledge27 although they also support downstream and applied research bycorporations. In most cases, resources from the public sector (including for university-based research) account for between 20and 40% of spending on R&D (refer Table 3). The remainder comes from the private sector, which typically invests more inapplied research and the development of scientific findings.

This division of responsibilities for R&D points to the vital role that the business sector, especially large companies, plays incommercialization and marketing. The economic benefit derived from innovations developed by a creative workforce is onlyfully realized when an innovation acquires a commercial form and value. As noted earlier, the chain extending from thegarnering of knowledge by creative people to the actual marketing of an innovative product or service is a lengthy one, and itgenerally entails a significant expenditure of time and resources.

Researchers of various stripes do much of the upstream work that takes an idea and transmutes it into an innovativetechnology. Developing it to the point where it becomes a marketable product or service is the work of professionals whounderstand the minutiae of commercializing and marketing an innovation backed by the organizational resources of thefirms. This is not only an expensive but also a risky process, as shown by the number of innovations that fail even when theyare developed and launched by companies with vast reserves of expertise and a proven track record. The cost and complexityof commercialization makes large business firms key players in the innovation process. By combining international marketingexperience, human resources, financial management, risk assessment skills, brand names, and intangible organizationalcapabilities, these entities are far better positioned than smaller firms and public agencies to convert an innovation intoa commercial asset. Large firms are not necessarily the most creative in terms of new ideas, nor are their R&D expendituresnecessarily more productive than those of smaller companies in terms of innovations. But the bigger companies stand a betterchance of developing and marketing an innovative product or service on a global scale and reaping large returns with the helpof well-developed research systems and channels of distribution.

Even when an innovation is first launched by a small firm, its eventual success in the global marketplace can depend on analliance with or takeover by a bigger corporation (as is the case with many biotech firms) which brings with it the necessarymarketing muscle, brand name, and manufacturing experience. Cisco is an example of a company that has made innovation-through-takeover into an art form. This is not a minor consideration in a world where the management of manufacturingcosts, product mix, and product customization for different market segments can involve dispersing production acrossa number of units located in several countries and the outsourcing of production to contract manufacturers. Therefore,production for global markets is itself a considerable feat of organization, management, product integration, quality control,and logistics, which can be well beyond the reach of small firms. Even the largest firms struggle to cope with the costs ofsustained innovation, development, and marketing on a worldwide scale. Instead, they are coming to rely increasingly onalliances and/or collaborative arrangements with other firms or specialized intermediaries.

25 The absorptive capacity of firms depends on their own research and preparedness; see Kodama and Suzuki [50].26 See Debackere and Veugelers on the role of university TTOs [51].27 Some of the key research underlying innovation by pharmaceutical industry in the U.S. was done in public labs or with government funding.

S. Yusuf / Technology in Society 31 (2009) 1–8 7

5. Summing up

This analysis of a creative society highlights the many pieces that need to be knit together to bring into existence a systemthat will be a fertile source of new knowledge, that can decant a steady stream of innovations from an accumulating stock ofknowledge, and that has the business capabi1ities to take innovations to the marketplace and achieve the success that is thekey to economic growth.

As indicated earlier, human talent is a precondition. Augmenting it is a function of how culture and tradition are mobilizedby public and private agencies acting in concert to build a stable society, furnish it with institutions that increase the stock ofhuman capital, enhance its quality, and instill values that favor achievement and initiative. This human capital can be mademore creative through the emergence of what I call wikicapital, i.e., the capital arising from networks. The creativity that feedsknowledge in turn needs to be translated into innovation, which is linked to incentives and the attributes of urban envi-ronments (e.g. amenities, services, and labor markets). Sustaining innovation at a high level calls for heavy investment in R&D.But the new ideas, findings, and technological leads that emerge are only the first step. A prolonged and expensive process ofdevelopment and eventual commercialization is required before products and services that pass the market test can emergeon a routine basis.

Development and commercialization calls for expertise, ingenuity, and entrepreneurial creativity in order to achievesuccess. Often breakthroughs are made by small firms but it is the large companies that are responsible for the bulk ofcommercialization. It is their developmental efforts, organizational capabilities, and resources that ultimately ensure that theinnovations generated by a creative society lead to economic growth. Romer predicts that the country that will lead in the21st century will be one that implements innovationsdmeta ideasdsupporting the production of new ideas in the privatesector [55].

Acknowledgements

Kaoru Nabeshima offered insightful comments, and Marinella Yadao provided efficient and cheerful assistance with theproduction of the paper. I am deeply grateful to both.

The findings, interpretations, and conclusions expressed in this study are entirely those of the author and should not beattributed in any manner to the World Bank, to its affiliated organizations, or to members of its Board of Executive Directors orthe countries they represent.

References

[1] Smil V. Transforming the twentieth century. New York: Oxford University Press; 2006.[2] OECD. Competitive cities in the global economy. Paris: OECD; 2006.[3] Phelps ES. Entrepreneurial culture: why European economies lag behind the U.S. Wall Street Journal 2007. Opinion Journal.[4] Solow R. Information technology and the recent productivity boom in the U.S.. Cambridge, MA: Presented at NCN Summit; 2001.[5] Glymour C. The automation of discovery. Daedalus 2004;133(1):69–77.[6] Mokyr J. Mobility, creativity, and technology development. In: Presented at creativity and the economy conference, Berlin; 2005.[7] Arthur WB. The structure of invention. Research Policy 2007;36(2):274–87.[8] Feinstein JS. The nature of creative development. Stanford, CA: Stanford University Press; 2006.[9] Mokyr J. Long-term economic growth and the history of technology. In: Aghion P, Durlauf SN, editors. Handbook of economic growth. Elsevier; 2005.

[10] Landes DS. Why Europe and the West? Why not China? Journal of Econonic Perspectives 2006;20(2):3–22.[11] Yusuf S, Altaf MA, Eichengreen B, Gooptu S, Nabeshima K, Kenny C, et al. Innovative East Asia: the future of growth. New York: Oxford University

Press; 2003.[12] Hanushek EA, Kimko DD. Schooling, labor force quality, and the growth of nations. American Economic Review 2000;90(5):1184–208.[13] Hanushek EA, Woßmann L. The role of education quality in economic growth. World Bank Policy research working paper 4122; 2007.[14] Murnane RJ, Levy F. Teaching the new basic skills: principles for educating children to thrive in a changing economy. New York: Free Press; 1996.[15] Autor DH, Levy F, Murnane RJ. The skill content of recent technological change: an empirical exploration. NBER working paper 8337; 2001.[16] Hanushek EA. Some simple analytics of school quality. NBER working paper 10229; 2004.[17] Hanushek EA. The failure of input-based schooling policies. NBER working paper 9040; 2002.[18] Glewwe P, Jacoby H, King E. Early childhood nutrition and academic achievement: a longitudinal analysis. FCND discussion paper 68; 1999.[19] Bloom DE. Social capitalism and human diversity, The creative society of the 21st century. Paris: OECD Publishing; 2000.[20] Frey BS, Stutzer A. What can economists learn from happiness research? Journal of Economic Literature 2002;XL:402–35.[21] Layard R. Happiness: lessons from a new science. New York: Penguin Press; 2005.[22] Kenny C. Does growth cause happiness, or does happiness cause growth? Kyklos 1999;52(1):3–25.[23] Di Tella R, MacCulloch R. Some uses of happiness data in economics. Journal of Economic Perspectives 2006;20(1):25–46.[24] Jones BF. Age and great invention. NBER working paper 11359; 2005.[25] Fleming L, Marx M. Managing creativity in small worlds. California Management Review 2006;48(1):8–9.[26] Tapscott D, Williams AD. Wikinomics: how mass collaboration changes everything. New York: Portfolio; 2007.[27] Carlsson B. Internationalization of innovation systems: a survey of the literature. Research Policy 2006;35(1):56–67.[28] Adams JD, Black GC, Clemmons RJ, Stephan PE. Scientific teams and institution collaborations: evidence from U.S. universities, 1981–1999. NBER

working paper 10640; 2004.[29] Page SE. The difference: how the power of diversity creates better groups, firms, schools, and societies. Princeton, NJ: Princeton University Press; 2007.[30] Uzzi B, Spiro J. Collaboration and creativity: the small world problem. American Journal of Sociology 2005;111(2):447–504.[31] Galison P. Image and logic: a material culture of microphysics. Chicago: University of Chicago Press; 1997.[32] Emotional intelligence, <http://en.wikipedia.org/wiki/Emotional_intelligence> [Accessed 6.7.2007].[33] Lundvall, Bengt-Ake. Higher education, innovation and economic development. In: Paper presented at the World Bank’s Regional Bank conference on

development economics, Beijing, China; January 16–17, 2007.[34] Mulgan G. The prospects for social renewal. In: The creative society of the 21st century. Paris: OECD Publishing; 2000.

S. Yusuf / Technology in Society 31 (2009) 1–88

[35] Jaffe AB, Lerner J. Innovation and its discontents: how our broken patent system is endangering innovation and progress, and what to do about it?.Princeton, NJ: Princeton University Press; 2006.

[36] Pisano GP. Can science be a business? Lessons from biotech. Harvard Business Review 2006;84(10):114–25.[37] Arora A, Fosfuri A, Gambardella A. Markets for technology: the economics of innovation and corporate strategy. Cambridge, MA: MIT Press; 2001.[38] Chaudhuri S, Ravallion M. Partially awakened giants. In: Yusuf S, Nabeshima K, editors. Dancing with giants: China, India, and the global economy.

Washington, DC: The World Bank; 2007.[39] Florida R. The rise of the creative class and how it’s transforming work, leisure and everyday life. New York: Basic Books; 2002.[40] Florida R. Cities and the creative class. London: Routledge; 2005.[41] Chapple K, Markusen A, Yamamoto D, Schorock G, Yu P. Gauging metropolitan ‘‘High-tech’’ and ‘‘I-tech’’ activity. Economic Development Quarterly

2004;18(1):10–24.[42] Bettencourt LMA, Lobo J, Strumsky D. Invention in the city: increasing returns to patenting as a scaling function of metropolitan size. Research Policy

2007;36(1):107–20.[43] Lester R. Universities, innovation, and the competitiveness of local economies. MIT industrial performance center working paper 05–010; 2005.[44] Mazzoleni R. Historical patterns in the co-evolution of higher education, public research, and national industrial capabilities. Vienna: UNIDO; 2005.[45] Yusuf S, Nabeshima K. How universities promote economic growth, Washington DC. The World Bank; 2007.[46] Bresnahan T, Gambardella A, Saxenian A, Wallsten S. ‘‘Old economy’’ inputs for ‘‘new economy’’ outcomes: cluster formation in the new Silicon Valley.

SIEPR discussion paper; 2001.[47] Cooke P. Knowledge economies. London: Routledge; 2002.[48] O’Mara MP. Cities of knowledge: cold war science and the search for the next Silicon Valley. Princeton, NJ: Princeton University Press; 2005.[49] Bresnahan T, Gambardella A. Building high-tech clusters. Cambridge, UK: Cambridge University Press; 2004.[50] Kodama F, Suzuki J. How Japanese companies have used scientific advances to restructure their business: the receiver-active national system of

innovation. World Development 2007;35(6):976–90.[51] Debackere K, Veugelers R. The role of academic technology transfer organizations in improving industry science links. Research Policy 2005;34:

321–42.[52] Bloom DE, Williamson JG. Demographic transition and economic miracles in emerging Asia. World Bank Economic Review 1998;12(3):419–55.[53] Wieser R. Research and development productivity and spillovers: empirical evidence at the firm level. Journal of Economic Surveys 2005;19:593–6.[54] Griliches Z. R&D, education and productivity. Cambridge, MA: Harvard University Press; 2000.[55] Romer PM. Economic growth. In: Henderson DR, editor. The concise encyclopedia of economics. Liberty Fund; 2007.

Shahid Yusuf is Economic Advisor, Development Economics Research Group, The World Bank. He holds a Ph.D. in Economics from Harvard University, anda BA in Economics from Cambridge University. He is the team leader for the World Bank-Japan project on East Asia’s future economy. He was Director ofthe World Development Report 1999/2000, Entering the 21st Century. Prior to that, he was Economic Advisor to the Senior Vice President and Chief Economist,Lead Economist for the East Africa Department and Lead Economist for the China Department. Dr. Yusuf has authored and edited numerous publications onEast Asia.

We follow the conceptual issues in this section with a review of relevant literature on growth, technological innovation,risk, and institutional governance. This is followed by a general formulation of growth and innovation as used in standardtreatments, from which we then define our model of innovation based on a hierarchical set of institutional determinants. Wethen apply our model first to a global sample of countries, followed by regional estimates to gain some measure of differencesin innovation and institutional governance by region. From the global and regional estimates, we then examine the economicvalue of institutional change in expanding the rate of creative innovation, from which we drive some basic conclusions thatmay be helpful in formulating policy guidance.

2. The role of creative innovation in economic growth

Economic growth depends on a variety of factors. Among them are a country’s rate of saving, increases in the stock ofproductive inputs, and technical change. Innovation bears most directly on technical change, and thus is a major determinantof economic growth. In a globalizing world in which rising population places growing pressure on the stock of naturalresources, sustainable growth depends more than ever on how innovation can be nurtured. Innovation is what may beconsidered as knowledge capital, and it stands in distinction to traditional measures of capital, notably physical stocks. Whilethere is some concern about this distinction, because innovation is treated separately from other forms of capital, we use it inthis sense in our analysis put forth here.

Given the importance of innovation to economic growth, it is useful to define the context through which it occurs andhow economic studies explain its role. In the first instance, we can think of innovation as applied knowledge. Invention maybe a necessary pre-requisite to innovation, but not all inventions become innovations. Nor, for that matter, do all innovationssucceed. Taking invention to the market requires agents who are capable and prepared to take on the associated risks overthe time frame through which an innovation moves forward. Such agents typically do not operate in isolation—they reflectinstitutions that provide much of the necessary financial commitment and to distribute the associated risks in ways thatmake continued innovation possible. Thus, if we seek to understand the role of creative innovation in economic growth, it isimportant to include consideration of institutions and risk.

What do we mean by institutions? Institutions refer to the level and depth of financial intermediaries as well as to firmsthat implement an innovation. All countries have some intermediaries and firms, but the quality of institutions can varysignificantly, depending on how countries craft policies to promote economic efficiency. In turn, the quality of governancehas a direct bearing on the level of risk that innovaters confront. It thus is important to examine factors that determine thequality of governance among institutions, as well as how governance bears on the level of risk.

What about risk? Economists tend to think of risks in essentially financial terms, and look at how markets derive relativeprices that reflect the degree of financial risk. However, financial measures represent only one dimension of risk. Otheraspects include political, economic, and environmental risk. Together they constitute the overall context through whichinstitutions must make decisions on launching new innovations.

In some countries, there are proxy measures for various categories of risk. Where markets are more complete, these riskscan be incorporated to some extent into the pricing of resources. For example, political risk might be reflected in the riskpremium on sovereign debt instruments. In turn, economic risk might be translated in terms of the premium returns thatinvestors require in markets where fluctuations are significant. As to environmental risk, whether negative externalitieshave been addressed in public policy will affect the pricing of resources as well.

The problem with many of these dimensions of risk is that in many countries adequate measures do not exist. Somecountries do not have well-developed sovereign debt markets and must rely on proxies such as IMF conditionality, or somealternative measure of financing sustainability. In turn, if equity markets are not well developed or absent, such incompletemarkets make it difficult to measure risk. We take up this question in this paper and suggest an approach to incorporating aproxy measure of risk that can be linked to innovation and economic growth.

3. Studies on growth and innovation

Studies on economic growth (Barro & Sala-I-Martin, 1995; Chenery & Syrquin, 1975; Denison, 1962; Jorgenson, Gollop, &Fraumeni, 1987; Porter, 1990) affirm the central roles of saving and the stock of inputs, but point to several underlyingfactors that may be crucial. Among them are: technology, aid and financial innovation, foreign direct investment, researchand development, and the governance of economic institutions.

Technological change offsets the classical economic problem of diminishing returns. We know how technology affectseconomic growth (David, 1975; Grossman & Helpman, 1991; Jorgenson, 1995; Rosenberg, 1976; Schmookler, 1966; vonHippel, 1988). As Arrow (1962) pointed out, innovation derives from experimentation, and it is a key element in achievingcost efficiencies in production (Leibenstein, 1966). What is less obvious is how to achieve technical change (Burns & Stalker,1966). Does it, for example, depend essentially on markets, as suggested by Rostow (1960), or by Schumpeter’s entrepreneur(1982, 1934, 1913), or does it require some measure of public intervention, as suggested by Aghion and Howitt (1996),Aghion and Tirole (1994), and by Arrow and Kurz (1970).

To the extent that markets alone do not provide a satisfactory rate of technical change can only be determined withreference to some underlying criteria. A benchmark could be sustainable growth, growth of one economy in comparison to

P. LeBel / Journal of Asian Economics 19 (2008) 334–347 335

some previous historical experience, or comparison to another economy with a higher rate of growth (Berthelemy &Varoudakis, 1996; Bordo, Taylor, & Williamson, 2003; Olson, 1982).

If innovation may depend in part on public sector intervention, it also may depend on financial innovation, internationalaid, and governance institutions. Mauro, Sussman, and Yafeh (2006) examine the role of financial innovation in historicalperspective, and note the positive relationship between financial innovation and growth. This supports the findings of Levine(1997) and Berthelemy and Varoudakis (1996). However, financial innovation alone may not explain major differences in percapita income, which suggests that other factors such as foreign direct investment (Aghion & Howitt, 1996; DeMello, 1999;Granstrand, 1999) also are at work.

One factor is the role of international aid. Although Burnside and Dollar (2004) found a positive relationship between aidand growth, this runs contrary to most findings, as summarized in Rajan and Subramanian (2006). The Burnside and Dollarfindings point, however, to the quality of institutional governance, which has been examined in a number of related studies,notably Kaufmann, Kraay, and Mastruzzi (2003), Perotti (1996), and Saint-Paul and Verdier (1993). Because the quality ofgovernance matters, institutions matter, and this forms the focus of the analysis we put forth in this paper. To do so, we firstderive the analytic framework of basic growth models, from which we then apply our institutional variables as they apply tocreative innovation.

4. Modeling economic growth and innovation

Empirical growth models build on the traditional neoclassical approach set forth in Solow (1956, 1957). In this approach,aggregate production function model in which factor accumulation establishes conditions for steady-state growth. Animportant conclusion from this work is that in order to sustain growth, there must be a continuous process of technologicalchange to offset diminishing marginal returns to capital stock accumulation.

In general, we can portray economic growth through a standard neoclassical function:

Y ¼ f ðL;K; TÞ; (1)

where Y is the output, which empirically can be measured in terms of PPP real per capita GDP, L the labor input, K the capitalinput and T is the level of technology.

In empirical studies, this relationship often has taken the form of

Y ¼ AðK; LÞ; (2)

where A is the level of technology.One variant of Eq. (2) is the Harrod–Domar model, in which labor inputs expand in proportion to increases in capital

stocks. Under this balanced factor proportions approach, the warranted rate of growth reduces to the ratio of the nationalsavings ratio to the incremental capital–output ratio. More formally, the warranted rate of growth can be expressed as

r ¼ s

k; (3)

where r is the warranted, or steady-state rate of growth in output, as indicated in (1), s the rate of savings, which in empiricalestimations can be determined as a percentage of GDP, and k is the incremental capital–output ratio, or investment in time t

divided by the change in GDP from t to t + 1.In a closed economy, growth can thus be portrayed as a function of the rate of savings, which encapsulates the allocation

of capital and labor inputs. When we include the role of trade, the economy’s rate of growth thus can be portrayed as

Y ¼ f ðS;TrÞ; (4)

where S is the national saving rate, expressed as a percentage of GDP and TR is the degree of trade dependence, expressed as aweighted share of imports and exports as a percentage of GDP.

Empirical problems abound in these formulations, notably in measuring the level of technology and determining itsimpact on per capita income. In Solow’s early models (1956, 1957), technology was treated in the residual of regressionequations. This approach led to international empirical studies by Denison (1962) as well as work that inspired many of theoriginal lending policies of the World Bank in developing countries (Chenery & Syrquin, 1975). In most of these studies,technology once again was considered to be exogenous to the growth process, taking second place to factor accumulationand required levels of international aid to achieve target levels of growth in real per capita income.

An alternative approach to growth accounting has come to be known as endogenous growth theory (Aghion & Howitt,1992). Inspired originally by work undertaken by Romer (1990), this approach builds on insights put forth in Schumpeter’sTheory of Economic Development (1934, 1911), and in his Capitalism, Socialism, and Democracy (1942). For Schumpeter,growth depends first and foremost on the entrepreneur, as elaborated in his Theory. In his latter work, innovation serves toexplain persistent differences in rates of return across industries, and may, as in Adam Smith’s steady state, cease to occuronce levels of wealth have reached a level that no longer stimulates its production. That very success, Schumpeter suggested,is how capitalism would then be transformed into a socialist economy, in contrast to Marx’s prediction of imminent collapsefrom a rising rate of exploitation. This latter, and now quaint, interpretation seems distant at best, given the collapse of the

P. LeBel / Journal of Asian Economics 19 (2008) 334–347336

Soviet Union and the expansion of market-driven globalization. As emphasized by Romer, nurturing innovation works bestnot by subsidizing physical capital accumulation, but by increasing the incentives for research.

Recent research that draws on Schumpeterian innovation theory utilizes several interrelated measures of growth. Keyamong them are research and development expenditures, patent and trademark applications, scientific citations, and netflows of copyright and trademark revenues. In an ideal setting, one could frame the optimal level of research anddevelopment as that which generates a maximum level of innovation. Thus

U ¼Z 1

0e�rtyðtÞdt ¼

Z 10

e�rtX1t¼0

Pðt; tÞAtxa

!dt; (5)

where U is the level of social welfare, e.g. a welfare adjusted level of per capita income, t the number of innovations, t thetime, and A is the level of technology.

If innovations arrive according to some Poisson style process, we can then portray their rate as

Yðt; tÞ ¼ ðlntÞt

t!e�lnt : (6)

Expected welfare can then be defined as

UðnÞ ¼ A0ðL� nÞa

r � lnðg � 1Þ : (7)

The socially optimal level of research and development expenditures would be where the first derivative of 7 is set to zero,in which case we then derive the reduced expression:

1 ¼ lðg � 1Þð1=aÞðL� n�Þr � ln�ðg � 1Þ ; (8)

where L is the quantity of labor input and g is the factor increase in output from each innovation.Under these conditions, the level of research would lead to an average rate of growth in welfare adjusted per capita

income of

g� ¼ ln� ln g: (9)

Although this framework provides a useful starting point for empirical estimates, there are several limitations that shouldbe noted. One is that an aggregate formulation does not capture the transitional phases of growth in many developingcountries, in particular, the shift of resources from agriculture into industry and services. Another is that knowledge itselfcannot be readily captured in an empirical form. A third is that the implementation of successful innovation requires that onetake into consideration the role of institutions and transactions costs. Thus, while the theoretical framework specified inEqs. (1)–(9), we find it useful and necessary to reformulate the framework when we take up the role of institutions. We takeup these issues in the following section.

5. A model of creative innovation

If economic growth depends partly on factor accumulation and for an open economy, partly on international trade, wecan enrich our growth model through incorporation of two additional factors, namely, risk and innovation. In previous work,we have examined the role of aggregate country risk on economic growth and find that it presents a transactions cost thatcan lower per capita income (LeBel, 2005). Management of risk requires that one take stock of institutional variables, namely,property rights and judicial independence. Increased levels of property rights and judicial independence tend to loweraggregate country risk, and in so doing, raise real per capita incomes. By including aggregate country risk and itsdeterminants in our growth accounting, we thus respond to one of the critiques of endogenous growth theory.

We now turn to creative innovation. Although research and development expenditures provide one measure ofinnovation, data are infrequent and sparse in many instances, thus making it difficult to derive meaningful internationalcomparisons of its impact on economic growth. However, there are other indicators that may serve as proxies for creativeinnovation. From them, we derive an index of creative innovation, which we define below.

We propose an index of creative innovation that contains two key elements: per capita scientific citations and the ratio ofper capita royalty fees to per capita royalty fee payments. Countries that engage in creative innovation do so in part throughthe frequency of scientific citations. In turn, when we consider both scientific and artistic innovation, these changes will havean effect on a country’s royalty revenues and royalty payments. For countries with low levels of scientific and artisticinnovation, royalty payments will exceed royalty revenues. As creative innovation expands the ratio of royalty revenues toroyalty fees will increase. We thus use the per capita net royalty ratio as the second component of our creativity innovationindex.

P. LeBel / Journal of Asian Economics 19 (2008) 334–347 337

Formally, we now define an index of creative innovation as

INNOVINDEX ¼ per capita scientific citationsþ per capita net royalty ratio

2: (10)

It is reasonable to ask why the choice of the innovation index given in (10). In some studies, spending on research anddevelopment, along with the number of patents, are used to measure innovation. Indeed, R&D and patents are linked andthey do reflect innovation activity. However, in many cases, expenditures on R&D are poorly tracked, particularly indeveloping economies, and the same applies to patents. Moreover, an index based on patent applications or patents granted,while appealing, presents several problems in empirical modeling.

First is that even where adequate information on patent applications or patent grants is available, the economic impactmay well be lagged and quite heterogeneous in any specific context. For example, analog television technology wasdemonstrated at the U.S. 1939 World’s Fair but did not have much commercial impact until the late 1940s after the SecondWorld War. Similarly, fax machine technology existed as far back as the late 1930s, but did not acquire more ubiquitous useuntil miniaturization changes were implemented in the 1970s and 1980s. As a third example, Philips electronics marketedlaser disk technology as early as the 1980s, well before the VHS and Betamax contest, and only when compressiontechnology enabled DVD’s to carry full-length films did videotapes begin their long exit from the market.

In some cases, firms may use patents as a barrier to entry by prospective competitors, thus vitiating the expected positiverelationship between patents, innovation, and economic growth. Finally, patent documentation is weak in many countriesthat we wish to examine, reflecting the weak status of property rights in some countries. If we restricted our sample only tothose countries that have strong patent laws, the range of the institutional factors we wish to examine would probably not bepossible.

Given the above observations and because we wish to test for innovation over a broad range of countries, we use scientificcitations as a proxy for research and development, while net royalties provide a measure of the impact of patents. In so doing,we recognize that if property rights are weakly enforced, even royalty payments may not capture the full range of innovation.While we do not claim that this index can capture all of the relevant dimensions of creative innovation, it does enable us toexamine how innovation affects the level of per capita income, and in turn, how institutional factors influence its level.

We now turn to the measurement of risk. As noted above, risk is prevalent in many dimensions and market prices in thepresence of incomplete contracts make it difficult to rely on relative prices to reflect varying risk premia. For this reason, wedecided to use a composite index of aggregate country risk. Our index is based on the ICRG measure reported by the WorldBank for individual countries. It ranges from 0 for the highest level of risk to 100 for the lowest level. As this is counter-intuitive to the expected inverse relationship between risk and income, we have derived the complement of the index, whichwe have labeled RCCRISK in our model.

As will be applied in our model, we also develop determinants of risk, in particular, the level of property rights and thedegree of judicial independence, which then can be used to tie risk to the level of innovation. The intuition behind thisapproach is straightforward: measures to reduce the level of risk produce positive effects on the level of innovation, which inturn, have a positive effect on the level of per capita income. Because markets are incomplete in many of the countries weexamine, we look to our aggregate country risk composite as a way of demonstrating the linkages between institutions,innovation, and economic growth.

We now specify the structure of our model of growth through creative innovation. Instead of an aggregate productionfunction approach as indicated in Eqs. (1) and (2), we use the framework of Eq. (4), namely, the rate of saving and the level ofinternational trade dependence. In turn, we add the role of aggregate country risk, which provides a proxy for the level ofefficiency in institutional governance. We then add to this our index of creative innovation, which we treat as exogenous toper capita income in this analysis. Our first order specification of economic growth thus is

PPPRPCGDP ¼ f ðGNSGDP;TRDEP;RCCRISK; INNOVATIONÞ; (11)

where PPPRPCGDP is the purchasing power parity real per capita GDP, GNSGDP the rate of national saving as a percentage ofGDP, TRDEP the level of trade dependence as a percentage of GDP, RCCRISK an index of aggregate country risk andINNOVINDEX is the index of creative innovation as defined in (10).

We first derive panel regression estimates of Eq. (11), allowing for sequential incorporation of risk and innovation. Resultsof preliminary estimates for our global sample of 103 countries over the 1980–2005 time period are show in Table 1. We findthat while savings and trade dependency are important determinants of real per capita income, aggregate country risk has alarger negative effect than either one alone. Measures to reduce aggregate country risk through institutional reform carryimportant effects for economic growth. When we factor in foreign direct investment, it has a positive, but statisticallyinsignificant effect on growth. This suggests that the choice of institutional regime may have much to do with the positiveeffects of foreign direct investment.

Turning to innovation, we look first at the individual effect of scientific citations on growth and find that it is statisticallyand economically significant. In fact, scientific citations carry a larger economic effect than either savings or tradedependency alone, and they offset the negative effect of aggregate country risk. When we then examine the effect of ourinnovation index on economic growth, it outweighs all other variables by a rough factor of 3–1. In short, innovation is a majordeterminant of per capita income, and measures to expand its level carry important consequences for globalization policies.

P. LeBel / Journal of Asian Economics 19 (2008) 334–347338

In Table 1, we report panel estimates using fixed and pooled samples using cross-section weights. Fixed effects estimatesallow us to take into consideration some of the inter-country characteristics, notably the extent to which the regressioncoefficients do not vary across individual countries or across time. However, for our purposes, ordinary panel regressionestimates are consistent not just in our global sample, but also when applied to the sub-sample regions we examine.

Of the estimates in Table 1, version H reflects the structure of the basic model we develop below. Version H also suggeststhat while a country’s rate of saving and its trade dependency are significant determinants to the level of income, risk andinnovation may be more important. We thus need to examine the determinants of both risk and innovation, as thesedeterminants may clarify the role of institutions and governance in achieving a given level of income.

5.1. Determinants of innovation

Although raw values for our index of risk and our index of innovation are significant determinants of the level of per capitaincome, because we are interested in their determinants, we now proceed to elaborate a nested hierarchy, as is shown inFig. 1. Instead of raw values for our index of risk and for our index of innovation in our model of growth, we apply estimatedvalues of these indices using various institutional determinants.

Fig. 1 indicates the directional causality relationships used in our model of creative innovation. To derive the model inFig. 1, we apply Granger causality tests to a set of institutional variables. In each case, we use Granger causality F-null tests to

Table 1

Global sample basic growth estimates

A. Fixed B. None C. Fixed D. None E. Fixed F. None G. Fixed H. None

C 6061.29 318.27 3536.08 83.87 3536.08 83.87 4461.10 7616.09

GNSGDP (t) 8.61

(4.207)

235.88

(33.915)

11.15

(8.035)

110.57

(30.717)

11.15

(8.035)

110.57

(30.717)

3.45

(2.073)

56.98

(16.182)

TRDEP (t) 28.21

(26.551)

33.76

(16.557)

16.71

(21.613)

28.00

(19.803)

16.71

(21.613)

28.00

(19.803)

14.44

(17.344)

14.15

(9.973)

RCCRISK (t) �17.02

(14.706)

�139.44

(38.512)

PCSCITES (t) 27.22

(54.943)

29.52

(99.535)

INNOVINDEX (t) 54.44

(54.943)

59.05

(99.535)

54.17

(55.021)

49.23

(81.988)

Adjusted R-squared 0.9598 0.6655 0.9873 0.8535 0.9873 0.8535 0.9781 0.8993

F-statistic 615.45 2664.04 1978.23 5198.41 1978.23 5198.41 1129.94 5976.71

Number of cross-sections 103 103 103 103 103 103 103 103

Number of observations 2678 2678 2678 2678 2678 2678 2678 2678

Method PLS PLS PLS PLS PLS PLS PLS PLS

Effects

Cross-section Fixed None Fixed None Fixed None Fixed None

Period None None None None None None None None

GLS weights CS CS CS CS CS CSW CS CS

Granger null values

GNSGDP (pr.) 4.76

(0.009)

4.60

(0.009)

4.76

(0.009)

4.60

(0.009)

4.76

(0.009)

4.60

(0.009)

4.76

(0.009)

4.60

(0.009)

TRDEP (pr.) 40.23

(0.000)

40.23

(0.000)

40.23

(0.000)

40.23

(0.000)

40.23

(0.000)

40.23

(0.000)

40.23

(0.000)

40.23

(0.000)

RCCRISK (pr.) 5.30

(0.005)

5.30

(0.005)

PCSCITES (pr.) 16.30

(0.000)

16.30

(0.000)

INNOVINDEX (pr.) 16.30

(0.000)

16.30

(0.000)

16.30

(0.000)

16.30

(0.000)

Dependent variable: PPPRPCGDP.

Fig. 1. Expanded model of institutional innovation.

P. LeBel / Journal of Asian Economics 19 (2008) 334–347 339

establish a hierarchy for the variables in the model, and in which we can reject possible reverse causality, running fromgrowth to institutions. Below each determinant is the value of the Granger null causality test, followed by the correspondingprobability of the null hypothesis in parentheses. Through nested panel regression estimates, we establish estimates for eachdeterminant. We then use estimated values to re-estimate our growth equation based on the global sample defined inEq. (11). Results from our revised model are summarized in Table 2.

The expanded model estimates point to several findings. First, while predicted savings rates and trade dependencycontinue to exert a positive effect on growth, once risk and innovation are taken into account, their role is reduced. And whilepredicted aggregate country risk has the largest single influence on per capita income, our second most importantdeterminant is the predicted index of creative innovation. Obviously, measures to reduce aggregate country risk as well as toincrease the level of innovation are keys to raising real per capita income. As such, they complement measures to increase therate of savings and the degree of trade dependency.

5.2. Extensions of the expanded model

We now conduct extensions to our expanded model. First we test for the significance of risk and creative innovation usinggeographic sub-samples of the global model. Second, we undertake estimates of the impact of single and multiple measuresto reduce aggregate country risk and expand creative innovation on the predicted level of per capita income. Results of ourgeographic regional estimates are shown in Table 3.

Our regional sub-model estimates validate our global model findings, namely, that risk and innovation work in oppositedirections in terms of their effects on per capita income. However, while risk and innovation are important determinants ofper capita income, strategies to manage them will vary according to underlying conditions in a given economic region. Forexample, innovation has the strongest effect on per capita income among Asian countries, followed by those in East Europe.This may reflect the relative starting positions of these regions in applied innovations, but it also may reflect theestablishment of a more innovative environment in these regions based on recent economic reforms.

To better gauge the relative importance of creative innovation and risk, we now conduct simulations in a two-stepprocess, using comparisons based on 2005 mean values for the respective variables and changes. First, we derive the impacton per capita income from a one-time change in an institutional parameter. Second, we then use the prevailing discount rateto derive present values of the change in per capita income from the one-time changes in institutional parameters. Finally,we derive the ratio of one-time and present-value changes in parameters to per capita income.

Results of changing institutional variables are shown in Table 4. Strengthened political rights increase judicialindependence, which then increases real per capita by the effect on reductions in aggregate country risk and in

Table 2

Global sample expanded model regression estimates

A. Fixed B. None C. Fixed D. None E. Fixed F. None

C 6991.92 27395.21 �516.68 �98.46 739.39 15021.03

GNSGDP (t) 7.44

(3.633)

6.03

(1.297)

7.93

(3.836)

100.57

(28.043)

6.88

(3.291)

16.79

(5.821)

TRDEP (t) 27.37

(25.644)

10.70

(8.007)

26.59

(24.644)

32.38

(22.825)

26.22

(24.238)

4.04

(3.513)

RCCRISKF (t) �23.54

(4.977)

�524.07

(94.525)

�15.39

(3.090)

�282.78

(54.758)

INNOVINDEXF (t) 112.04

(7.691)

60.39

(83.282)

101.10

(6.757)

37.70

(50.008)

Adjusted R-squared 0.9700 0.8054 0.9700 0.8168 0.9703 0.8709

F-statistic 826.91 3695.15 826.47 3979.46 827.29 4517.32

Number of cross-sections 103 103 103 103 103 103

Number of observations 2678 2678 2678 2678 2678 2678

Method PLS PLS PLS PLS PLS PLS

Effects

Cross-section Fixed None Fixed None Fixed None

Period None None None None None None

GLS weights CS CS CS CS CS CS

Granger null values

GNSGDP (pr.) 4.76

(0.009)

4.60

(0.009)

4.76

(0.009)

4.60

(0.009)

4.76

(0.009)

4.60

(0.009)

TRDEP (pr.) 40.23

(0.000)

40.23

(0.000)

40.23

(0000)

40.23

(0.000)

40.23

(0.000)

40.23

(0.000)

RCCRISKF (pr.) 10.10

(0.000)

10.10

(0.000)

10.10

(0.000)

10.10

(0.000)

5.30

(0.005)

5.30

(0.005)

INNOVINDEXF (pr.) 16.50

(0.000)

16.50

(0.000)

16.50

(0.000)

16.50

(0.000)

16.30

(0.000)

16.30

(0.000)

Dependent variable: PPPRPCGDP.

P. LeBel / Journal of Asian Economics 19 (2008) 334–347340

expansions in the level of creative innovation. Strengthened property rights reduces corruption, which in turn reducesaggregate country risk, while expanding economic freedom, thus increasing the level of creative innovation. Increases ina country’s reserve to import ratio, a standard measure often advocated in economic reform programs, has the primaryeffect of reducing aggregate country risk, and thus expanding the level of real per capita income. This also is true ofincreases in the level of foreign direct investment relative to GDP. There may be, in fact, additional positive effects on thelevel of FDIP through changes in institutional variables and their attendant effects on risk, even though we have not madesuch estimates here.

We also estimate the effects of an increase in a region’s national saving rate and its level of trade dependence on real percapita GDP. Although increases in these variables do produce positive effects on the level of real per capita GDP, they areoutweighed in most instances by reductions in aggregate country risk and in increases in creative innovation.

We now turn to the derivation of the relative impact of one-unit one-time changes in independent variables on real percapita income. Table 5 reports the results of these ratios. Our results indicate that in general, for all regions, the relativepositive effect of greater innovation is higher than for reductions in risk in all regions.

Using mean regional rates of interest, we now derive present values for the effects of the respective independent variableson real per capita GDP. These values provide a basis on which to determine the extent to which, on a per capita basis, oneshould consider investing in improvements in institutional governance variables. As long as the costs of improvements in

Table 3

Expanded model regional regression estimates

A. Global B. Africa C. Asia D. CACarib E. WEurope F. EEurope G. MENAf

C 15021.03 1523.83 13400.41 4261.67 20408.98 7067.88 3684.88

GNSGDP (t) 16.79

(5.821)

27.62

(11.822)

13.37

(3.209)

6.18

(9.221)

214.33

(65.437)

31.82

(11.804)

32.85

(I7.765)

TRDEP (t) 4.04

(3.513)

6.25

(8.126)

28.23

(22.468)

6.02

(14.864)

48.74

(116.200)

4.22

(3.617)

25.21

(45.170)

RCCRISKF (t) �282.78

(54.758)

�23.49

(6.826)

�458.38

(31.465)

�32.19

(24.909)

�384.99

(140.401)

�170.11

(18.868)

�63.61

(12.924)

INNOVINDEXF (t) 37.70

(50.008)

268.00

(74.808)

6283.58

(22.700)

426.29

(94.326)

61.85

(22.002)

2994.27

(17.254)

618.46

(46.885)

Adjusted R-squared 0.8709 0.9193 0.9949 0.9952 0.9926 0.8858 0.946404

F-statistic 4517.32 2220.25 826.47 4232.33 14861.17 551.45 1267.981

Number of cross-sections 103 30 13 17 17 11 12

Number of observations 2678 750 338 408 442 285 288

Method PLS PLS PLS PLS PLS PLS PLS

Effects

Cross-section None None Fixed None None None None

Period None None None None None None None

GLS weights CS CS CS SUR CS SUR CS SUR CS SUR CS SUR

Granger null values

GNSGDP (pr.) 4.76

(0.009)

4.60

(0.009)

9.52

(0.000)

4.60

(0.009)

3.52

(0.030)

4.60

(0.009)

0.078

(0.925)

TRDEP (pr.) 40.23

(0.000)

40.23

(0.000)

4.62

(0.011)

40.23

(0.000)

18.40

(0.000)

40.23

(0.000)

4.65459

(0.0103)

RCCRISKF (pr.) 10.10

(0.000)

10.10

(0.000)

10.06

(0.000)

10.10

(0.000)

0.15

(0.857)

5.30

(0.005)

0.84406

(0.4311)

INNOVINDEXF (pr.) 16.50

(0.000)

16.50

(0.000)

9.07

(0.000)

16.50

(0.000)

16.41

(0.000)

16.30

(0.000)

0.55696

(0.5736)

Dependent variable: PPPRPCGDP.

Table 4

Absolute effect of a one-time one-unit change in independent variables on real per capita GDP

Global Africa Asia CACARIB WEurope EEurope MENAf

Absolute change in model variable

POLRTS1 187$ 14$ 379$ 194$ 3,052$ 144$ 95$

PROPRT1 308$ 103$ 319$ 227$ 509$ 1286$ 257$

RESIMPCOVRATIO1 283$ 125$ 231$ 81$ 427$ 279$ 394$

FDIGDP1 14$ 106$ 243$ 22$ 409$ 283$ 93$

GNSGDP1 75$ 30$ 15$ 27$ 622$ 41$ 33$

TRDEP1 0$ 6$ 30$ 29$ 456$ 4$ 25$

1 unit increase in MIGPOPGRATIO 60$ 100$ 59$ 24$ 419$ 36$ 14$

1 unit decrease in RRCRISKF 283$ 126$ 425$ 81$ 679$ 297$ 64$

1 unit increase in INNOVINDEXFA 38$ 267$ 6209$ 223$ 469$ 2368$ 618$

Global mean predicted base PPPRPCGDF 8128$ 1944$ 8025$ 5408$ 23,427$ 7157$ 5096$

P. LeBel / Journal of Asian Economics 19 (2008) 334–347 341

institutional governance variables are less than the values reported in Table 6, the underlying implicit rate of return will becompetitive with existing rates of interest in a given geographic region.

It is reasonable to ask why should we examine one-time, relative, and present-value effects of our institutional variables.Most models of economic growth rely on traded and observable values. However, in our institutional analysis, we find ituseful to separate out how these institutional variables bear on the level of per capita income. An analogy would be if all ofthe countries in our sample had sovereign debt ratings and well-developed capital markets, prices of assets would provide anefficient measure of the risk premia that agents confront in making allocation decisions. However, in their absence, thequestion of governance becomes more crucial, and we emphasize that there is a significant economic value in strengtheningsuch determinants as property and political rights. This acquires particular significance in the context of international effortsto promote greater transparency in governance, but for which there often is little economic value assigned in doing so.

6. Tests on variables and model equations

To examine the robustness of our estimates, we undertake several tests on variables used in the model. First we examinestationarity using unit root tests using the Augmented Dickey Fuller Test and the Phillips Perron Test. Then we look atcointegrating relations used in the model indicated in Fig. 1. Results of these tests are given in Table 7.

With the exception of Eastern Europe, we cannot reject a unit root for PPPRPCGDP, but can do so for the other variables inthe expanded model. However, our Pedroni Engle–Granger and Johansen Fisher tests provide support for cointegration in thevarious geographic samples, which may offset the problem of stationarity within individual variables. For this reason, wekeep our revised model estimates, noting that stationarity cannot be ruled out entirely. In so doing, it is reasonable to askwhy we have organized the panels using only geographic samples rather than around some alternative criterion such as highand low income, weak or strong property rights, or some other combination. In our present analysis, we made this choicesimply to illustrate variations across regions, and not to suggest that regions provide ready alternatives for policy choices.

7. Policy implications and conclusions

Institutions matter in achieving economic growth. Although economic models traditionally have ignored the economicimpact of governance on growth, it is increasingly clear that a failure to do so can produce weak or counterproductive effects.This applies not just to a country’s rate of saving or trade dependence. It also extends to such areas as foreign directinvestment. Where it becomes critical is in terms of the impact of institutional governance on aggregate country risk and in acountry’s rate of creative innovation. Measures to reduce aggregate country risk and expand creative innovation may havesignificant payoffs. We note briefly some of the kinds of policy measures that derive from our model.

In terms of aggregate country risk, efforts to strengthen property rights and judicial independence have significantpositive effects. While greater political rights increase judicial independence, expanded property rights reduce the level of

Table 5

Relative effects of a one-time one-unit change in independent variables on real per capita GDP

Global Africa Asia CACARIB WEurope EEurope MENAf

Relative change (%)

POLRTS1 2.30 0.71 4.72 3.58 13.03 2.01 1.87

PROPRT1 3.79 5.29 3.97 4.20 2.17 17.97 5.04

RESIMPCOVRATIO1 3.49 6.45 2.87 1.50 1.82 3.90 7.73

FDIGDP1 0.18 5.45 3.03 0.40 1.74 3.96 1.82

GNSGDP1 0.92 1.52 0.19 0.51 2.65 0.57 0.64

TRDEP1 0.01 0.32 0.37 0.53 1.95 0.06 0.49

1 unit increase in MIGPOPGRATIO 0.74 5.13 0.74 0.44 1.79 0.51 0.27

1 unit decrease in RRCRISKF 3.49 6.48 5.30 1.49 2.90 4.15 1.25

1 unit increase in INNOVINDEXFA 0.46 13.75 77.36 4.13 2.00 33.08 12.14

Table 6

Present-value effects of one-time one-unit changes in independent variables on real per capita GDP

Relative change Global Africa Asia CACARIB WEurope EEurope MENAf

POLRTS1 2.30% 0.71% 4.72% 3.58% 13.03% 2.01% 1.87%

PROPRT1 3.79% 5.29% 3.97% 4.20% 2.17% 17.97% 5.04%

RESIMPCOVRATIO1 3.49% 6.45% 2.87% 1.50% 1.82% 3.90% 7.73%

FDIGDP1 0.18% 5.45% 3.03% 0.40% 1.74% 3.96% 1.82%

GNSGDP1 0.92% 1.52% 0.19% 0.51% 2.65% 0.57% 0.64%

TRDEP1 0.01% 0.32% 0.37% 0.53% 1.95% 0.06% 0.49%

1 unit increase in MIGPOPGRATIO 0.74% 5.13% 0.74% 0.44% 1.79% 0.51% 0.27%

1 unit decrease in RRCRISKF 3.49% 6.48% 5.30% 1.49% 2.90% 4.15% 1.25%

1 unit increase in INNOVINDEXFA 0.46% 13.75% 77.36% 4.13% 2.00% 33.08% 12.14%

P. LeBel / Journal of Asian Economics 19 (2008) 334–347342

corruption, expand economic freedom, and thus a country’s level of creative innovation. In turn, reductions in corruptionthat are accompanied by expansions in judicial independence also reduce the level of aggregate country risk.

In terms of creative innovation, since our innovation index builds on scientific citations and the ratio of net royalties to percapita GDP, measures to increase their level produce positive effects on per capita income. Scientific innovations reflect acountry’s education and research capacity. Investments in education and research produce obvious effects in scientificcitations. In turn, many scientific achievements are complemented by the level of creativity in other domains, as in music andthe arts. How a country nurtures the environment in which these innovations take place is critical, particularly in that theytranslate into greater royalty producing income relative to royalty payments for a given country. Strengthened propertyrights are a necessary mechanism for this to take place, while a nurturing and open environment also has a role to play.

In conclusion, aggregate country risk and creative innovation typically carry greater economic consequences on acountry’s level of per capita income than such traditional variables as the rate of national saving and trade dependence.Devising suitable policies built around credible models is an important step in raising per capita income. The results of thisanalysis lend support to such efforts.

Appendix A. Data sources and model specification

Preliminary to our analysis, we have gathered time-series and cross-section data from a variety of sources, including the World

Bank Development Indicators, the Heritage Foundation Index of Economic Freedom, and Freedom House political variables.

Table A.1 lists the definitions, scales, and sources of variables used in the present analysis.

See Tables A.2–A.4.

Table 7

Tests for unit roots and cointegration

A. Global B. Africa C. Asia D. CACarib E. WEurope F. EEurope G. MENAf

(A) Panel unit root tests

PPPRPCGDP

ADF–Fisher chi-square (pr.) 119.96

(1.000)

70.89

(0.159)

3.66

(1.000)

16.84

(0.994)

3.41

(1.000)

37.97

(0.018)

7.99

(0.999)

Phillips Perron–Fisher chi-square (pr.) 100.53

(1.000)

62.64

(0.383)

3.81

(1.000)

16.80

(0.994)

2.77

(1.000)

22.81

(0.412)

9.68

(0.996)

GNSGDP

ADF–Fisher chi-square (pr.) 353.63

(0.000)

127.97

(0.000)

52.91

(0.001)

53.36

(0.019)

47.37

(0.064)

38.82

(0.025)

40.95

(0.017)

Phillips Perron–Fisher chi-square (pr.) 329.76

(0.000)

119.60

(0.000)

40.22

(0.037)

56.78

(0.009)

39.51

(0.237)

25.00

(0.297)

47.73

(0.003)

TREDP

ADF–Fisher chi-square (pr.) 256.80

(0.009)

87.88

(0.011)

16.97

(0.910)

49.54

(0.041)

31.01

(0.615)

48.72

(0.001)

34.02

(0.084)

Philips Perron–Fisher chi-square (pr.) 235.79

(0.076)

76.84

(0.070)

15.14

(0.955)

48.18

(0.054)

22.96

(0.925)

43.49

(0.004)

32.96

(0.105)

RCCRISKF

ADF–Fisher chi-square (pr.) 308.87

(0.000)

52.98

(0.728)

47.66

(0.006)

76.71

(0.000)

67.57

(0.001)

20.41

(0.557)

24.39

(0.439)

Phillips Perron–Fisher chi-square (pr.) 295.79

(0.000)

48.00

(0.868)

38.39

(0.056)

88.80

(0.000)

117.13

(0.000)

20.57

(0.548)

24.62

(0.427)

INNOVINDEXF

ADF–Fisher chi-square (pr.) 254.10

(0.013)

35.48

(1.000)

31.60

(0.226)

72.28

(0.000)

46.76

(0.071)

28.34

(0.165)

35.99

(0.056)

Phillips Perron–Fisher chi-square (pr.) 416.80

(0.000)

92.62

(0.004)

27.29

(0.394)

55.75

(0.011)

31.57

(0.587)

32.15

(0.075)

42.41

(0.012)

(B) Panel cointegration tests (unrestricted version: PPPRPCGDP, GNSGDP, TRDEF, RCCRISKF, INNOVINDEXF)

Pedroni Engle–Granger test

Panel ADF-Stat (pr.) 5.44

(0.000)

4.49

(0.000)

1.00

(0.241)

0.67

(0.320)

2.84

(0.007)

1.83

(0.075)

1.17

(0.202)

Panel PP-Stat (pr.) 1.13

(0.212)

3.02

(0.004)

1.22

(0.189)

1.94

(0.061)

1.09

(0.220)

0.92

(0.262)

1.72

(0.091)

Johansen Fisher test

None (pr.) 1567.00

(0.000)

348.00

(0.000)

263.90

(0.000)

356.60

(0.000)

380.00

(0.000)

225.10

(0.000)

256.50

(0.000)

At most 1 (pr.) 708.60

(0.000)

173.10

(0.000)

137.90

(0.000)

182.20

(0.000)

173.20

(0.000)

112.20

(0.000)

118.00

(0.000)

P. LeBel / Journal of Asian Economics 19 (2008) 334–347 343

Table A.1

Variable definitions and sources

Variable Variable symbol Definition Scale

Lowest Highest

Political rights POLRTS A supportive measure of

democratic institutions

1.00 7.00 Freedom House,

Freedom in the World

Property rights PROPRT A measure of the

strength of economic

freedom

1.00 5.00 The Heritage Foundation,

The Index of Economic Freedom

Judicial independence JUDIND A measure of the

strength of economic

freedom

1.00 10.00 The Heritage Foundation,

The Index of Economic Freedom

Economic freedom ECFREE The aggregate index

of economic freedom

1.00 5.00 The Heritage Foundation,

The Index of Economic Freedom

Reserve import

coverage ratio

RESIMPCOVRATIO Ratio of Reserves to imports 0.00 1.00 The World Bank,

World Development Indicators

Foreign direct

investment to

GDP ratio

FDIGDP Ratio of FDI to GDP 0.00 1.00 The World Bank,

World Development Indicators

Net migration to

population ratio

NETMIGPOPRATIO Net official migration

to population

Negative Positive The World Bank,

World Development Indicators

Revised country

composite risk

RCCRISK An index of political,

economic, financial, and

environmental country risk

0.00 100.00 ICRG, as reported

by the World Bank, and re-scaled

National saving rate GNSGDP Ratio of national saving to GDP Negative Positive The World Bank,

World Development Indicators

Trade dependency ratio TRDEP Ratio of exports and imports to GDP 0.00 Positive The World Bank,

World Development Indicators

Per capita scientific

citations

PCSCITES Per capita scientific citations 0.00 Positive The World Bank,

World Development Indicators

Innovation index INNOVINDEX Average of per capita

scientific citations

and net royalty ratio

0.00 Positive The World Bank,

World Development Indicators

PPP per capita

real GDP

PPPRPCGDP Real per capita

GDP at purchasing

power parity rates

Positive Positive The World Bank,

World Development Indicators

Real interest rate REALINRATE Real discount rate

of central bank

in a country

Negative Positive The World Bank,

World Development Indicators

Corruption index CORRUPA Corruption perceptions

index, inverted scale

0.00 10.00 Corruption Perceptions Index, Inc.

Table A.2

Expanded regional model: 2005 mean original and predicted values

Global Africa Asia CACARIB WEurope EEurope MENAf

Panel number 103 30 13 17 17 11 12

Original values

POLRTS 4.8641 3.7667 4.2308 5.7059 7.0000 1.4558 2.3333

PROPRT 2.9107 2.3933 2.7692 2.5294 4.5882 3.3000 2.1667

JUDIND 5.1332 4.6963 5.2170 4.7220 7.6163 4.5296 3.4200

ECFREE 2.0085 1.5532 1.8377 2.0065 2.9135 1.8995 1.5108

RESIMPCOVRATIO 4.6408 4.3266 6.9832 4.8464 1.8222 0.7988 7.3979

FDIGDP 3.1164 3.0106 1.9420 3.0580 4.6752 0.0009 1.9623

NETMIGPOPRATIO 0.0000566 �0.0015440 0.0056350 �0.0059000 0.0146620 0.0024150 �0.0016480

RCCRISK 29.12 38.69 27.94 31.10 15.94 33.46 27.83

GNSGDP 19.7955 12.9247 30.8755 18.2433 23.2020 26.9792 22.0137

TRDEP 77.5171 64.3841 89.8601 65.9278 93.0468 73.2015 70.4656

INNOVINDEX 71.3280 2.5644 1.0440 4.5358 7.8070 2.9488 2.0193

PPPRPCGDP $10652.78 $2377.57 $9270.38 $6313.38 $30405.65 $6568.80 $6126.92

REALINRATE 6.3069 10.3003 4.8831 7.4590 3.1015 �11.7728 4.1405

Predicted values

JUDINDF 5.0928 4.5202 5.0446 4.5043 7.8508 4.1275 3.4695

CORRUPAF 6.0196 7.2585 6.5947 6.9075 2.5777 6.4242 7.2696

ECFREEF 1.9193 1.4461 1.7626 1.9631 2.7434 1.9380 1.2750

RCCRISKF 34.9219 43.5928 32.9869 38.8303 17.0240 34.4199 38.8532

INNOVINDEXF 60.3168 2.5644 1.0640 4.4257 7.6219 1.6622 2.2360

PPPRPCGDPF $8065.11 $1946.76 $8025.12 $5393.44 $23834.80 $7157.50 $5096.00

P. LeBel / Journal of Asian Economics 19 (2008) 334–347344

Table A.3

Estimating equations for revised expanded global model

Dependent variable JUDIND CORRUPA ECFREE RCCRISK INNOVINDEX PPPRPCGDP

C 4.84 9.79 4.50 55.02 48.75 15021.03

POLRTS 0.05

(8.941)

PROPRT �0.13

(5.178)

0.03

(2.018)

JUDINDF �0.67

(9.210)

�9.31

(3.717)

CORRUPAF �0.44

(6.144)

5.29

(2.003)

RESIMPCOVRATIO �0.98

(19.950)

ECFREEF 5.25

(3.721)

FDIGDP 0.48

(12.717)

NETMIGPOPRATIO 11.85

(2.058)

MIGPOPGRATIO

RCCRISKF �219.58

(22.199)

GNSGDP 24.34

(3.307)

TRDEP 10.91

(4.875)

INNOVINDEXF 47.65

(44.050)

INNOVINDEXF1

Adjusted R-squared 0.9954 0.9935 0.9963 0.8843 0.9199 0.8545

F-statistic 5607.85 3923.36 6886.84 195.81 293.84 4053.53

Number of cross-sections 103.00 103.00 103.00 103.00 103.00 103.00

Number of observations 2678 2678 2678 2678 2678 2678

Method PLS PLS PLS PLS PLS TSPLS

Effects

Cross-section Fixed Fixed Fixed Fixed Fixed None

Period None None None None None Fixed

GLS Weights CS CS CS CS CS CS

Granger F-null values

POLRTS (pr.) 7.07

(0.001)

PROPRT (pr.) 30.38

(0.000)

10.41

(0.000)

JUDINDF (pr.) 23.93

(0.000)

6.79

(0.001)

5.27

(0.005)

CORRUPAF (pr.) 12.05

(0.000)

RESIMPCOVRATIO (pr.) 25.47

(0.000)

ECFREEF (pr.) 12.28

(0.000)

FDIGDP (pr.) 2.84

(0.058)

NETMIGPOPRATIO (pr.) 0.90

(0.405)

MIGPOPGRATIO (pr.)

RCCRISKF (pr.) 10.10

(0.000)

GNSGDP (pr.) 7.91

(0.000)

TRDEPF (pr.) 27.51

(0.000)

INNOVINDEXF (pr.) 16.50

(0.000)

P. LeBel / Journal of Asian Economics 19 (2008) 334–347 345

Per pust ak aan Su l t anah Z an ar i ah

TITLE : SOURCE

The Role of Creative Innovation in Economic Growth : Some International Comparisons

http://www.sciencedirect.com/

The role of creative innovation in economic growth: Some internationalcomparisons§

Phillip LeBel *

School of Business, Department of Economics and Finance, Montclair State University, 1 University Avenue, Montclair, NJ 07043, United States

A R T I C L E I N F O

Article history:

Received 6 November 2007

Received in revised form 28 April 2008

Accepted 29 April 2008

JEL classification:

D02

D73

O38

P48

Keywords:

Economic growth

Innovation

Risk

Institutional governance

A B S T R A C T

For many countries, export-driven policies have thus far produced dramatic increases in

real per capita income. At the same time, sustainable growth requires that technological

innovation proceed at comparable rates if mutual gains from globalization are to be

realized. In this paper, we derive a measure of innovation and test the extent to which

institutional policy choices enhance or delay its diffusion. To do so we use a panel

regression model, with data on a sample of 103 countries in different geographic regions

for the 1980–2005 period. Our findings provide empirical evidence of the positive role of

creative innovation in economic growth, and from which we derive several basic policy

conclusions.

� 2008 Elsevier Inc. All rights reserved.

1. Introduction

For many countries in East Asia, export-driven policies have led to significant increases in per capita income over the pastseveral years. This ‘‘Asian’’ model of growth is based in several key elements. It depends in the first instance on favorablerates of exchange, access to primary and intermediate inputs, and finally on relative access to the major industrializedeconomies where these exports have gone. Over the longer term, however, as differences in per capita incomes diminish,sustainable economic growth will depend not just on the above factors, but also on the ability to innovate. In this paper, wedevelop a model of creative innovation to explain relative differences in growth, test for its determinants, and then calibratehow changes in institutional variables produce significant variations in per capita income. To do so, we rely on a globalsample of 103 countries that covers the 1980–2005 period. We develop a nested panel model that is applied to the globalsample as well as to six geographic sub-samples. Our findings point to several policy conclusions.

The focus of this paper is on the role of institutions in economic growth. What motivates this perspective is the growingrecognition that standard models of economic growth capture only a portion of the underlying dynamics that drive savingand investment in general, and risk management and creative innovation in particular. To do so, we proceed in several steps.First we examine the relationship between creative innovation and economic growth in which we underline the importanceof creative innovation and how institutions shape the underlying level of risk that accompanies innovations.

Journal of Asian Economics 19 (2008) 334–347

§ Financial support provided by the Montclair State University Faculty Scholarship Program. In addition, the author expresses appreciation to two

anonymous referees for useful comments in the preparation of this paper.

* Tel.: +1 973 655 7778.

E-mail address: lebelp@mail.montclair.edu.

Contents lists available at ScienceDirect

Journal of Asian Economics

1049-0078/$ – see front matter � 2008 Elsevier Inc. All rights reserved.

doi:10.1016/j.asieco.2008.04.005

References

Aghion, P., & Howitt, P. (1992). A model of growth through creative destruction. Econometrica, 60(2), 323–351.Aghion, P., & Howitt, P. (1996). Research and development in the growth process. Journal of Economic Growth, 1, 49–73.Aghion, P., & Tirole, J. (1994). The management of innovation. Quarterly Journal of Economics, 109, 1185–1209.Arrow, K. J. (1962). The economic implications of learning by doing. Review of Economic Studies, 29(1), 155–173.Arrow, K. J., & Kurz, M. (1970). Public investment, the rate of return, and optimal Fiscal policy. Baltimore: The Johns Hopkins University Press.Barro, R. J., & Sala-I-Martin, X. (1995). Economic growth. New York: McGraw-Hill Publishing.Berthelemy, J.-C., & Varoudakis, A. (1996). Economic growth, convergence clubs, and the role of financial development. Oxford Economic Papers, 48, 300–328.Bordo, M., Taylor, A. M., & Williamson, J. G. (2003). Globalization in historical perspective. Chicago: University of Chicago Press for the NBER.Burns, T., & Stalker, G. (1966). The management of innovation. London: Tavistock Publishing.Burnside, C., & Dollar, D. (2004). Aid, policies, and growth: Revisiting the evidence. Policy Research Working Paper, WPS3251. Washington, DC: The World Bank.Chenery, Hollis, & Moises, Syrquin, with the assistance of H. Elkington (1975). Patterns of Development, 1950-1970. Oxford: Oxford University Press for the World

Bank.David, P. A. (1975). Technical choice, innovation and economic growth. Cambridge, UK: Cambridge University Press.DeMello, L. (1999, January). Foreign direct investment-led growth: Evidence from time series and panel data. Oxford Economic Papers, 51(1), 133–151.Denison, Edward F. (1962). The Sources of Economic Growth in the United States. Washington, DC: Committee on Economic Development.Granstrand, O. (1999). The economics and management of intellectual property: Towards intellectual capitalism. Northampton, MA: Edward Elgar Publishers.Grossman, G. M., & Helpman, E. (1991). Innovation and growth in the global economy. Cambridge, MA: MIT Press.Jorgenson, D. W. (1995). Productivity. Cambridge, MA: MIT Press.Jorgenson, D., Gollop, F. M., & Fraumeni, B. M. (1987). Productivity and U.S. economic growth. Cambridge, MA: Harvard University Press.Kaufmann, D., Kraay, A., & Mastruzzi, M. (2003). Governance matters III: Governance indicators for 1996–2002. Policy Research Working Paper 3106. Washington,

DC: The World Bank.LeBel, P. (2005). Optimal choices for risk management: The economic value of institutional reform in globalizing economies. Global Business and Finance Review,

10(3), 113–128.Leibenstein, H. (1966). Allocative efficiency vs. X-efficiency. American Economic Review, 56, 391–415.Levine, R. (1997). Financial development and economic growth: Views and agenda. Journal of Economic Literature, 35(2), 688–725.Mauro, P., Sussman, N., & Yafeh, Y. (2006). Emerging markets and financial globalization: Sovereign bond spreads in 1870–1913 and today. New York: Oxford

University Press.Olson, M. (1982). The rise and decline of nations: Economic growth, stagflation, and social rigidities. New Haven, CT: Yale University Press.Perotti, R. (1996). Growth, income distribution, and democracy: What the data say. Journal of Economic Growth, 1(2), 149–187.Porter, M. E. (1990). The competitive advantage of nations. New York: The Free Press.Rajan, R., & Subramanian, A. (2006). Aid and growth: What does the cross-country evidence really show? NBER working paper no. 11513. Cambridge, MA: National

Bureau of Economic Research.Romer, P. (1990). Endogenous technological change. Journal of Political Economy, 98(5), S71–S102.Rosenberg, N. (1976). Perspectives on technology. New York: Cambridge University Press.

Table A.4

List of countries used in regional panel analysis

Global Africa Asia South America and

the Caribbean

West Europe East Europe Middle East

and North Africa

Mexico Benin Bangladesh Belize Austria Albania Iran, Islamic Rep.

Canada Botswana China Costa Rica Belgium Bulgaria Lebanon

United States + all others Burkina Faso India El Salvador Denmark Czech Republic Oman

C. Af. Republic Indonesia Guatemala Finland Estonia Qatar

Cameroon Japan Honduras France Hungary Syrian Arab Republic

Chad Korea, Rep. Nicaragua Germany Latvia Turkey

Congo, Dem. Rep. Malaysia Panama Greece Lithuania Yemen, Rep.

Congo, Rep. Pakistan Argentina Ireland Poland Egypt, Arab Rep.

Cote d’Ivoire Philippines Bolivia Italy Romania Libya

Ethiopia Singapore Brazil Luxembourg Slovak Republic Tunisia

Gabon Sri Lanka Chile The Netherlands Russian Federation Algeria

Ghana Thailand Colombia Norway Morocco

Guinea Vietnam Ecuador Portugal

Kenya Paraguay Spain

Madagascar Peru Sweden

Malawi Uruguay Switzerland

Mali Venezuela, RB United Kingdom

Mauritania

Mauritius

Mozambique

Niger

Nigeria

Senegal

South Africa

Sudan

Tanzania

Togo

Uganda

Zambia

Zimbabwe

103 30 13 17 17 11 12

P. LeBel / Journal of Asian Economics 19 (2008) 334–347346

Rostow, W. W. (1971, 1960). The stages of economic growth (2nd ed.). New York: Cambridge University Press.Saint-Paul, G., & Verdier, T. (1993). Education, democracy and growth. Journal of Development Economics, 42(2), 399–407.Schmookler, J. (1966). Invention and economic growth. Cambridge, MA: Harvard University Press.Schumpeter, J. A. (1982, 1934, 1913). The theory of economic development. New Brunswick, NJ: Transaction Books. reprint of 1934 translation from German original

by Redvers Opie.Solow, R. M. (1956). A contribution to the theory of economic growth. Quarterly Journal of Economics, 70(1), 65–94.Solow, R. M. (1957). Technical change and the aggregate production function. Review of Economics and Statistics, 39, 312–320.von Hippel, E. (1988). Sources of innovation. New York: Oxford University Press.

P. LeBel / Journal of Asian Economics 19 (2008) 334–347 347

Per pust ak aan Su l t anah Z an ar i ah

TITLE : SOURCE

Organizational and Institutional Influences on Creativity in Scientific Research

http://www.sciencedirect.com/

O

Ta

b

c

d

a

AA

KSORRC

1

cdoirdc1(Ma1stl1AppS

pj

0d

Research Policy 38 (2009) 610–623

Contents lists available at ScienceDirect

Research Policy

journa l homepage: www.e lsev ier .com/ locate / respol

rganizational and institutional influences on creativity in scientific research

homas Heinzea,∗, Philip Shapirab,c, Juan D. Rogersb, Jacqueline M. Senkerd

Faculty for Social and Economic Sciences, Department of Sociology, University of Bamberg, Lichtenhaidestraße 11, 96045 Bamberg, GermanySchool of Public Policy, Georgia Institute of Technology, Atlanta, GA 30332-0345, USAManchester Institute of Innovation Research, Manchester Business School, University of Manchester, M13 9PL, UKScience and Technology Policy Research (SPRU), University of Sussex, Brighton BN1 9QE, UK

r t i c l e i n f o

rticle history:vailable online 28 February 2009

eywords:

a b s t r a c t

This paper explores institutional and organizational influences on creativity in scientific research. Usinga method for identifying creative scientific research accomplishments in two fields of science (nanotech-nology and human genetics) in Europe and the US, the paper summarizes results derived from twenty

cientific creativityrganizational environmentesearch contextesearch managementase study method

case studies of highly creative research accomplishments, focusing on contextual patterns at the group,organizational, and institutional levels. We find that creative accomplishments are associated with smallgroup size, organizational contexts with sufficient access to a complementary variety of technical skills,stable research sponsorship, timely access to extramural skills and resources, and facilitating leadership.A potential institutional threat to creative science is the increase in competitive research council funding

instit

at the expense of flexiblepolicy are considered.

. Introduction

Scientific creativity is a key driver for scientific and technologi-al progress, and also a precondition for advances in other societalomains. Yet, our knowledge and understanding of how researchrganizations and institutional environments, and changes in both,mpinge upon capabilities of research groups to conduct creativeesearch is fragmented. The complex relationships between pro-uctivity, social stratification, reward structures, and organizationalontext in scientific research were frequently studied until the mid-970s within the institutional paradigm of the sociology of sciencesee, for example, Shepard, 1956; Meltzer, 1956; Merton, 1957;

eltzer and Salter, 1962; Stein, 1962; Pelz, 1964; Crane, 1965; Pelznd Andrews, 1966; Cole and Cole, 1967; Reskin, 1977; Zuckermann,977; Andrews, 1979; Long and McGinnis, 1981). Since then sciencetudies have been dominated by a social-constructivist paradigmhat focuses on the micro-conditions of knowledge production inaboratory settings and epistemic cultures (Latour and Woolgar,979; Knorr-Cetina, 1981, 1999; Knorr-Cetina and Mulkay, 1983).

t the same time, the study of creativity has become popular insychology, although organizational and institutional questionslay only a marginal, if any role (Dunbar, 1997; Amabile, 1996;ternberg, 2003; Simonton, 1999, 2004). It was only recently that

∗ Corresponding author.E-mail addresses: thomas.heinze@uni-bamberg.de (T. Heinze),

shapira@mbs.ac.uk (P. Shapira), juan.rogers@pubpolicy.gatech.edu (J.D. Rogers),acky.senker@ntlworld.com (J.M. Senker).

048-7333/$ – see front matter © 2009 Elsevier B.V. All rights reserved.oi:10.1016/j.respol.2009.01.014

utional sponsorship. Implications for research management and research

© 2009 Elsevier B.V. All rights reserved.

new attempts were undertaken to re-establish an organizationaland institutional perspective in the study of scientific accom-plishments. For example, Hollingsworth (2000, 2002) and Hage(2006) have published on organizational structures that fosterbreakthrough research. Hemlin et al. (2004) have explored variousinstitutional factors that are associated with what they call “cre-ative knowledge environments”. Yet, in their book on serendipityin science, Merton and Barber (2004) conclude that the institu-tional analysis of discoveries in science is still in its infancy. Manyimportant questions remain about what creative scientific accom-plishments are, how we can identify them, in which organizationsthey occur most often, and which institutional factors are influentialin shaping cutting-edge research environments.

The desire to know more about the factors that contribute to sci-entific creativity is given further impetus by the substantial changesseen over the last few decades in the institutional and organi-zational conditions under which scientific research is conducted(Jansen, 2007; Laudel, 2006; Schimank, 2005; Etzkowitz, 2003;Owen-Smith, 2003; Langfeldt, 2001; Bourke and Butler, 1999).Public research funding is now increasingly allocated through com-petitive processes, rather than long-term institutional block-grants;increased research collaboration is encouraged through a vari-ety of measures including through organized research centers,networks, centers of excellence, and interdisciplinary teams, to

address diverse challenges of complexity, convergence, knowledgeexchange, scale, scope, and internationalization in contemporaryscience; and evaluation systems for research performance areincreasingly implemented as a supplement to peer review (Münch,2008; Thèves et al., 2007; Lepori et al., 2007; Corley et al., 2006;

ch Poli

SacmCaee2dt

rgatasoqaicaei

eiddaoiosoher

2o

raergd1

oaficeamcrteo

T. Heinze et al. / Resear

hapira and Kuhlmann, 2003; Chompalov et al., 2002; van Leeuwennd Tijssen, 2000; Henkel, 1999). In the context of heightenedompetitive pressures to foster science-driven business develop-ent and the rise of new global locations for research (especially

hina), research policymakers in developed economies hope thatdjustments to research organizations and broader institutionalnvironments for scientific research will not only promote morefficiency but also boost scientific excellence and creativity (Blau,005). There is an increasing need for recommendations about theesign of science policy to support highly creative researchers andheir groups.

This paper explores factors which influence the ways in whichesearch groups conduct their work. Besides features of the researchroup itself, such as size and career stage of group leaders, our mainnalytical and empirical focus is on organizational variables andhe institutional environment in which these groups operate, suchs leadership, funding structures, or competitive pressures. Ourtudy is built on a longitudinal multi-method research design basedn survey, interview, archive and bibliometric data, and uses bothuantitative and qualitative research methods including as networknd regression techniques, and in-depth interviews and case stud-es (Heinze et al., 2007; Heinze and Bauer, 2007). We identifiedreative research accomplishments in two broad fields of science,nalyzed why certain research groups are more creative than oth-rs, and investigated which factors in their work environment werenfluential for their accomplishments.

We begin the paper by reviewing contributions to the lit-rature on scientific creativity and by highlighting selected keyssues important for further research (Section 2). Second, we intro-uce our methodology (Section 3). Third, we discuss in moreetail the results from our case studies of highly creative researchccomplishments, focusing particularly on findings related torganizational and institutional influences on scientific creativ-ty including work group factors, such as size of research groupsr communication patterns, and organizational features, such asponsorship or disciplinary variety (Section 4). Then, we discussur findings in the light of previous results, and we demonstrateow our findings improve our understanding of creative knowledgenvironments (Section 5). Finally, we consider the implications foresearch management and research policy (Section 6).

. Literature review: definitions, approaches and findingsn scientific creativity

The importance of creativity in numerous areas of society hasesulted in studies of creativity from diverse fields, including man-gement and business (Sutton, 2002), arts (Maritain, 1977; Berkat al., 2003), politics (Otten, 2001; Nagel, 2002), and urban andegional development (Florida, 2002). However, there is a conver-ence in characterizing creativity as encompassing capabilities too things that are new and useful (see Ochse, 1990 and Amabile,996, for a summary of definitions).

In the world of science, creativity is similarly defined in termsf knowledge and capabilities that are new, original, surprising,nd useful (Hollingsworth, 2004; Simonton, 2004). As in otherelds, standards and norms are established in science against whichlaims for innovative contributions are assessed, although sci-nce, more than other fields, has evolved procedures, disciplines,nd institutions to accredit new knowledge (Whitley, 2000). Inaking judgments about scientific creativity, scientific peers use

riteria such as plausibility, validity, and originality. There are well-ecognized tensions here, since criteria of plausibility and validityend to encourage conformity, while originality draws upon andncourages dissent. The history of science is replete with examplesf path-breaking research achievements that were initially rejected

cy 38 (2009) 610–623 611

by the scientific establishment because they challenged existingparadigms (Kuhn, 1962; Polanyi, 1969; Hessenbruch, 2004). Inother cases, work that was initially proclaimed publicly to be highlycreative was found, following more considered scientific review, tobe flawed (Lewenstein, 1992). In short, the scientific communitymust be persuaded that novel and unexpected contributions havevalue, and claims that research is highly creative need validationover time and by other scientists.

There are varied approaches to examining and empirically mea-suring creativity. These include examining creative individuals, theproducts or outcomes of creative work, creative processes and cre-ative knowledge environments (Stumpf, 1995; Hemlin et al., 2004).At the individual level, there has been much discussion – not neces-sarily with consensus – about the relationship between intelligenceand creativity (Mansfield and Busse, 1981; Sternberg, 2003). Therehas also been a focus on the behavioral traits of creative individu-als, including their level of curiosity, risk tolerance, motivation, andwillingness to overcome failure, leading to arguments that creativepeople typically tolerate higher levels of contradiction, ambigu-ity, and uncertainty in their work (Sternberg et al., 1997; Weinert,2000). Still, such individual characteristics are neither easily mea-sured nor uniformly correlated with creative accomplishments,leading others to concentrate on tangible scientific publication out-comes and citations to identify highly creative researchers.

A prominent attempt to assess scientific creativity through out-comes is publication and citation analysis within an evolutionary-probability theoretical frame (Simonton, 1999, 2004). Simontonargues that scientists who are highly productive in publishingpapers encounter a greater likelihood that one or more of theirpapers will come to the attention of other scientists, be cited,and recognized as creative. In other words, the more contribu-tions to knowledge that a scientist produces, the higher his orher chances are that one of these contributions resonates well inthe scientific community. This approach is not without criticismbecause, for example, some highly creative scientists publish onlya few papers, while citation counts typically consider only jour-nal publications and not books or other contributions, such as newscientific instrumentation. Another outcome approach is based onstudying prestigious prize winners in science (Zuckermann, 1977;Hollingsworth, 2002). Of course, such prizes are highly selective –and there are surely more creative research accomplishments thanNobel committees can recognize. Hollingsworth (2002) addressesthis problem by obtaining access to short-listed Nobel Prize nomi-nees until the 1940s.

Creative processes, including the selection of problems, meth-ods, partners and knowledge sources, have been another area ofinquiry. Rather than focusing on innate individual traits, work oncreative processes has highlighted the opportunity structures incollaboration networks that facilitate the generation and diffusionof novel ideas. Proponents of network brokerage argue that peoplewho are placed at the intersection of heterogeneous social groupshave an increased likelihood of drawing upon multiple knowledgesources, leading to the generation of new ideas. For example, man-agers who occupy brokerage positions are more often than othersthe source of good ideas (Burt, 2004; Rodan and Galunic, 2004).In contrast, proponents of cohesive collaborative networks arguefor the benefits of trust, shared risk taking and easy mobilization infacilitating information and knowledge transfer. According to thesestudies, individuals with cohesive social ties are more likely to beinvolved in innovations (Uzzi and Spiro, 2005; Obstfeld, 2005). Inreference to this ongoing debate, Fleming et al. (2007) argue that

although brokering inventors are more likely to generate new ideas,the brokered network structure itself is less suited to diffusing theseideas. Therefore, network structures that enhance the generation ofnovel ideas may inherently diminish the likelihood of their diffu-sion.

6 ch Pol

t–icdapoomimsp

ifoaUguaB(tUtacT2rvbnrlsAaosW1ttttsr

secs

pmoz

r

12 T. Heinze et al. / Resear

Considerations of the research environment add a series of fur-her elements – including organizational and institutional featuresto the examination of scientific creativity. Research environments

nfluence opportunities for research collaboration and multidis-iplinarity, which may in turn affect processes of knowledgeiscovery. But once we contemplate the role of organizationalnd institutional aspects, a range of other factors may come intolay in stimulating creative research situations, including auton-my for researchers, adequate facilities and funding, developmentf complementary disciplines and fields, staff selection, manage-ent structures, and leadership. Less conducive factors include:

nsufficient basic funding, limited time for research, bureaucraticanagement, narrow range of disciplinary expertise, and exces-

ive evaluation and accountability pressures (Hemlin et al., 2004,p. 16–17, 195–196).

Perhaps the first comprehensive effort to empirically exam-ne modern research environments was a study of 17 researchacilities in the United States across various types of fields and lab-ratories by Pelz and Andrews (1966).1 Another major effort tonalyze research organizations in a comparative fashion was theNESCO study by Andrews (1979) on more than 1200 researchroups across six countries.2 Again other studies have investigatedniversity departments and their role in setting research goalsnd influencing scientists’ productivity (Long and McGinnis, 1981;aird, 1986; Allison and Long, 1990), More recently, Hollingsworth2002) examined a large number of research breakthroughs inhe biomedical sciences across 128 research organizations in thenited States, of which 28 had two or more major discoveries in

he first half of the twentieth century.3 There are also historicalccounts of exceptionally successful institutes in the biomedi-al sciences, such as Rockefeller University, California Institute ofechnology (Hollingsworth, 2000, 2004) or Institut Pasteur (Hage,006). These studies differ in methodology, for example, withespect to the identification of research groups, by key outcomeariable (productivity, recognition or research breakthroughs), andy level of analysis (group, department, institute, or project). Sig-ificantly, although some are published recently, they all examineesearch organizations and institutional environments from ear-ier periods. For example, Hollingsworth (2002) and Hage (2006)tudy breakthroughs in the first half of the 20th century; Pelz andndrews (1966) report on research organizations of the 1950s;nd Andrews (1979) capture the situation before 1975. Researchrganizations and institutional environments have changed exten-ively in the last three decades following periods of post-Second

orld War expansion (Windolf, 1997), 1970s stabilization (Ziman,994), and more recent restructuring (as noted in Section 1 ofhis paper). While our emphasis is on recent research organiza-

ions and institutional environments, it is insightful to considerhe findings from studies focusing on earlier scientific genera-ions. In particular, this literature raises three important themestill relevant today, namely: specialization, communication, andesearch autonomy; group size and departmental effects; and

1 Pelz and Andrews analyzed industrial, government and university labs whichpanned the following R&D fields: pharmaceuticals, glass and ceramics, electronics,lectrical equipment, weapons guidance, animal diseases, commercial uses of agri-ultural products, basic research in several physical sciences, biological and socialciences (Pelz and Andrews, 1966, p. 2).

2 Andrews and colleagues studied within the fields of mathematics, astronomy,hysics, chemistry, life sciences, earth and space sciences, agricultural sciences,edical sciences, technological sciences, and social sciences the following types

f research organizations: academic organizations, academies, cooperative organi-ations, productive enterprises, and private institutions (Andrews, 1979, pp. 17–52).3 Hollingsworth (2002) examined universities, medical centers, free standing

esearch institutes, and industrial research laboratories in the biomedical sciences.

icy 38 (2009) 610–623

resources, recruitment and leadership, as discussed in the followingsections.4

2.1. Specialization, communication, and research autonomy

Pelz and Andrews (1966, pp. 22–27, 35) find that scientistsare most productive when they both interact vigorously withand involve their colleagues in setting up their research goals.Research productivity is correlated with high frequency of intra-organizational communication. Hollingsworth (2000, 2004) arguesthat research breakthroughs are typical for research organizationswhere scientists communicate across disciplinary and thematicborders, and where research leaders provide strategies for inte-grating scientific diversity with rigorous standards of scientificexcellence. For example, because the Rockefeller University wasorganized around laboratories rather than scientific disciplinesand fields, it had a greater capacity to adapt quickly to researchstrategies and to allow effective communication across cognitiveboundaries (Hollingsworth, 2004, pp. 34–35). Further strategies forintellectual integration within the boundaries of an organizationare mobility of researchers and teamwork between departments(Hage, 2006). However, the way in which the individual andorganizational levels interpenetrate is somewhat contested in theliterature. Hollingsworth (2000) emphasizes scientific excellenceand depth of domain-specific knowledge at the individual levelin combination with intellectual integration at the organizationallevel. In contrast, Pelz and Andrews show that high-performancescientists are often not in agreement with their organization interms of research agenda and strategy. The authors argue that “alaboratory remains vigorous when it encourages a certain tensionbetween what the members want, and what they think the orga-nization wants” (Pelz and Andrews, 1966, p. 139).5 Such tension,however, is only bearable if scientists share the same motivationfor their work: “It seemed helpful if sets of close colleagues shareda common enthusiasm for similar kinds of problems and preferredsocial relations” (Pelz and Andrews, 1966, p. 146). Pelz and Andrewsfound thematic breadth to be most effective when combined withfreedom in goal setting and research strategy, whereas coordinatedresearch settings were better suited to more specialized researchers(Pelz and Andrews, 1966, pp. 29–31, 158–173).

2.2. Group size and departmental effects

There is considerable evidence in the literature that researchperformance initially tends to rise as group size increases, but thatabove a certain group size threshold, this effect tails off or becomesnegative, i.e. either no increase or even a decrease in performance(for a review, see von Tunzelmann et al., 2003). For example, analy-ses based on the large dataset of Andrews (1979) show that above athreshold of 4–6 team members, per capita performance decreasesmarkedly, particularly in academic natural science groups. Also,in order for groups larger than 5–7 scientists to reach the perfor-mance levels of smaller groups, both coherent research programsand group leaders with strong time commitments to research activ-ities are needed. Although the curvilinear relationship betweengroup size and performance is evident both for quantity and qual-

ity of research across various countries and fields, quality seems tobe affected more negatively from large group size than per capitaresearch quantity (Andrews, 1979, pp. 55–94, 192–222). In addition,the department level has been found influential. Long and McGinnis

4 See also Bland and Ruffin (1992) for a more detailed literature review on pro-ductive research environments.

5 On a more general level, March (1991) argues that organizations learn moreeffectively from individuals who are slow (rather than fast) in acquiring what isknown and taken for granted by the organization.

ch Policy 38 (2009) 610–623 613

(cdotptr(

2

h(pasruttsmtstatigeppalea

w4opehtiloepmrsswtfcc

3a

t

Table 1Typology of scientific creativity.

Type of scientific creativity Examples

1 Formulation of new ideas (orset of new ideas) that open upa new cognitive frame or bringstheoretical claims to a newlevel of sophistication.

Theory of specific relativity inphysics (EINSTEIN, 1905)

2 Discovery of new empiricalphenomena that stimulatednew theorizing

Biodiversity → Theory ofevolution (Biology), DARWIN(1859)

3 Development of a newmethodology, by means ofwhich theoretical problemscould be empirically tested.

Factor analysis → Theory onmental abilities (Psychology),SPEARMAN (1904a, 1904b, 1927)

4 Invention of novel instrumentsthat opened up new searchperspectives and researchdomains.

Scanning tunnelingmicroscopy → Nanotechnology(Physics), BINNIG & ROHRER(1982)

5 New synthesis of formerlydispersed existing ideas intogeneral theoretical lawsenabling analyses of diversephenomena within a common

General systems theory (Biology,Cybernetics, Sociology),BERTALANFFY (1949), ASHBY(1956), LUHMANN (1984)

In parallel with the nomination survey, we identified rele-vant prizes in the two fields, drawing on respondent nominations,other expert input, and our own knowledge. It should be notedthat few prizes are specifically dedicated to nanotechnology, an

T. Heinze et al. / Resear

1981) and Allison and Long (1990) find for the fields of physics,hemistry, mathematics and biology that scientists with growingepartmental prestige tend to show an increase in both the numberf publications and the number of citations. Again other studies findhat university departments with clear research goals show higherroductivity levels than those without such goals (Baird, 1986), andhat flat and decentralized structures in research organizations cor-elate with higher productivity at the level of organizational unitsBirnbaum, 1983).

.3. Resources, recruitment and leadership

Other important variables for productive research climate areuman resources, instrumentation and funding. Pelz and Andrews1966) report that actual resources are associated less highly withroductivity than the resources researchers perceived they couldccess. Similarly, Andrews (1979) finds that the perceived acces-ibility of human resources but not the de facto level of humanesources explains the largest amount of variance in a researchnit’s performance. Recruiting outstanding scientists in a researcheam is another important variable. For example, Dill (1985) showshat highly productive research units can be distinguished by theignificance they attach to hiring talent. The importance of recruit-ent also points to the influence of leadership. Hage (2006) argues

hat plural organizational leadership ensures diversity of researchtrategies and richness in ideas. The three directors of Institute Pas-eur operated with diverse recruitment patterns but all three keptlook-out for creative people in their fields and then attempted

o convince them to come to the Institute. However, leaderships crucial not only for recruitment, but also for directing researchroups. Andrews (1979, p. 68, 102, 219) finds that effective lead-rs are involved in ongoing research. Active participation in theraxis of scientific work is important for leaders to understand theroblems of the group, to motivate group members and to organizecoherent research program. This finding is also reflected in the

iterature review by Mumford et al. (2002) who suggest that lead-rship in creative environments requires predominantly technicalnd scientific expertise.

In summary, the available literature provides several ideashich informed the development of our interview protocol (Section

). However, opportunities abound for new work that probes. Muchf the extant literature is based on historical analyses of sciencerior to recent developments in the structure and dynamics of sci-ntific research in the advanced economies. Moreover, while thereas been a significant focus on intra-organizational communica-ion and the balancing of individual and institutional research goals,nter-organizational and institutional aspects have been somewhatess studied. For example, there has been little discussion of the rolef research councils in sponsoring new research fields; there is littlemphasis on how basic research activities are framed differently inublic sector research organizations and private laboratories; andost studies are concerned with understanding productivity and

ecognition, but less with scientific creativity. In these respects, ourtudy of creative scientific events and the research groups respon-ible for them seeks to generate new insights, particularly sincee are embedding our cases in current organizational and institu-

ional contexts. In the next section we introduce the methodologyor identifying highly creative research accomplishments and thease study approach for examining work environments of highlyreative scientists and their groups.

. Methodology: identification of creativeccomplishments and case study design

The exploration of the features of the organizational and institu-ional context that have an effect on scientific creativity was carried

cognitive frame.

Source: Heinze et al. (2007). See source for full references to and discussion of exam-ples.

out by means of a set of case studies anchored around selectedindividuals and their groups identified as highly creative in a newnomination method, previously reported in Heinze et al. (2007).Since the highly creative researchers were identified by indepen-dent experts and prize review panels, our work is based on ex post,external attributions of creativity by others rather than by modelingcreative mechanisms at the individual level.

3.1. Identification of highly creative research accomplishments

First of all, we conducted a survey that obtained more than400 European and US nominations from 185 experts in twofields of research, human genetics and nanoscience/nantechnology(referred to as nanotechnology), across five categories of promi-nence: highly cited researchers, active academic researchers,active industry researchers, journal editors, and research programmanagers.6 The two research fields were chosen to offer a com-parison between a more established, disciplined-embedded field(human genetics) and an emerging interdisciplinary field (nan-otechnology). Furthermore, both fields have undergone substantialgrowth in recent years (see Heinze, 2004 and Youtie et al., 2008for nanotechnology; Sulston and Ferry, 2002 for human genetics),while the science system as a whole is in a steadier state (Ziman,1994). Field growth is an important variable for the developmentof creative ideas, because more novel ideas are produced, and theforces of sorting out original ideas are relatively weak (March, 2007,pp. 16–17). Therefore, it is especially fruitful to study the organi-zational context of creative research in growing research fields. Inorder to account for creative activity of several sorts, we introduceda typology of scientific creativity, with five stipulated categories andone open category for respondents to include types not included inthe original list (Heinze et al., 2007; Table 1).7

6 For details of the sample frame and operationalization of the categories, seeHeinze et al. (2007).

7 The scientific creativity typology is discussed in detail in Heinze et al. (2007).

614 T. Heinze et al. / Research Pol

Table 2Distribution of creative scientists, combining nominations and prize winners.

Nanotechnology Human genetics

Europe US Europe US

Multiple prize winners 9 5 10 1Multiple nominations 7 21 0 3Prize winner and nomination 16 17 5 9Multiple prize winners and

multiple nominations3 4 0 0

Total highly creativescientists

22 29 14 11

Total scientists in database 224 204 150 111

Sco

eapPSrdnfmupit

3

fbcHewgnoi

rthpcpicieeaiptcff

ource: CREA database, 2005 (Heinze et al., 2007). There is overlap between the fourategories, so the total number of highly creative scientists is smaller than the sumf the rows 1–4.

xception being the Feynman Prize. Prizes for nanotechnologyccomplishments are usually associated with a discipline (such ashysics or materials research) or an organization (such as the Max-lanck-Society in Germany or the Centre National de la Recherchecientifique in France). Our approach was therefore to identifyelevant prizes broadly, then to carefully review all awards and lau-ations to explicitly identify relevant prize winners. We merged theomination and prize winner data so as to offer a consolidated basis

or studying creative research accomplishments (Table 2). Supple-ented by additional web-based research, this provides us with a

nique data source of information about creative research accom-lishments in our two target science domains. We are particularly

nterested in scientists with multiple nominations, since recogni-ion of their research is derived from more than one source.

.2. Case study methodology

Our database of highly creative scientists offers a foundationrom which to develop case studies, since we can identify scientistsy the number and type of creativity nominations, by field, and theharacter and timing of their creative research accomplishment.8

ence, drawing on the subset of multiply-nominated creative sci-ntists (i.e. those with most nominations from different sources),e undertook case studies of creative events of twenty research

roups across Europe and the United States in the two fields ofanotechnology and human genetics. These cases explored therganizational and institutional dimensions of work environmentsn which creative research has been conducted.

The theoretical framework for the case studies addresses theelationship between features of the context of research work andhe occurrence of a creative event that has already happened andas been recognized by colleagues and other experts who partici-ated in the nomination process. The theoretical question for thease study design is whether there is a predominant contextualattern for creative events in scientific research. Since creativity

s surely also a matter of individual talent, our approach cannotompletely isolate the contextual factors from the individual abil-ties of the researchers as causes of the creative event. Rather, ourxploration addresses a set of necessary conditions for a creativevent to come to fruition given that the researchers in our casesre already regarded as highly talented. Moreover, the span of anndividual’s research career is generally long enough for the sameerson to have worked in changing contextual conditions and it is

herefore likely that the pursuit of research goals involves strategicalculations for the researcher in which an assessment of contextualactors is inherent in choosing alternative paths of action. There-ore, it is possible to compare the effects of contextual factors at

8 Although not reported here, we also know their names and current institutions.

icy 38 (2009) 610–623

different times in the career of the same researcher. In order toobtain reliable data about the context from our informants, weused information from the nomination process, including publi-cations, citations, prize citations, among other external indicators,to identify a specific creative event so that the features of the con-text could be related to actual conditions of work at a determinedtime and place. This significantly reduced the potential for recallproblems, ambiguous statements and generic opinions in responseto our interview questions because the creative events were promi-nent occurrences in the lives of our respondents, the circumstancesof which they are likely to remember in detail.

The cases were selected using parameters most relevant to fea-tures of the context, such as the research field (nanotechnologyand human genetics), organizational affiliation (universities, med-ical facilities, industry R&D labs), geographical location which alsoprovided diversity in various institutional and cultural dimensions(different regions of the United States and several European coun-tries with their different funding mechanisms, academic styles,promotion rules, among other things), and different sorts of creativeevent as identified by nominators using our proposed typology.By conducting case studies under all these conditions we aimedat establishing whether there were emergent patterns that couldconfirm the presence of essentially the same sort of phenomenonacross cases. In other words, and as a brief reminder, the logicof multiple case studies is not based on a representative samplefor generalizing to a large population, as statistical inference does.Rather, cases are selected to elucidate the mechanism of a phe-nomenon for generalization to theory and concepts (George andBennett, 2005; Eisenhardt, 1989).

Each case included a fairly complete account, both histori-cally and technically, of the creative event obtained from the keyresearcher (or researchers) associated with the accomplishmentand validated by others familiar with the event (colleagues, col-laborators, competitors or external observers). A comprehensivefile with information on publications, co-authors, and citations,research themes, and the organizational context of the researchgroup was also compiled. Then, in-depth interviews were con-ducted with the group leader, and follow-up interviews withcolleagues and associates, group members, and other scientistswho were knowledgeable about the circumstances under whichthe creative event materialized. In total, we conducted 44 inter-views between November 2005 and February 2007 (see List ofinterviews).

The interview protocol included questions related to the prepa-ration phase prior to the creative accomplishment, the creativeaccomplishment phase, and factors related to research group, orga-nizational and institutional levels. The main variables about whichdata is gathered are derived from the literature review above(Section 2). At the group level, variables include the size andcomposition of research team, communication patterns, qualityof research leadership, and need and access to outside resourcessuch as specialized equipment. At the organizational level, thevariable set comprises (all variables at the time of the accom-plishment): structure and size of the organization, centralizationof decision-making, clarity of research goals, features of the fund-ing arrangement, accountability burden, reputation and visibilityof the organization. Finally, the institutional level variables includejob mobility opportunities, competitiveness within the researchfield, and munificence of the funding environment. The case studymethod enables an in-depth analysis, capturing more of the textureand detail of behavior than is possible in conventional aggregate

data-oriented methods. In addition, the case study method is “non-linear” in the sense that the researcher learns from the case and,if appropriate, adjusts the focus of the research during the courseof the project. Hence, there is no need to hold fast to hypothesesif they are clearly being discredited in favor of more accurate and

T.Heinze

etal./Research

Policy38

(2009)610–623

615

Table 3Creativity case summaries.

Case number

1 2 3 4 5 6 7 8 9 10

Type of multiple expert andprize nominations

MultPrizNom MultNom MultNom MultPrizNom MultNom MultPrizNom PrizNom MultPrizNom PrizNom MultNom

Type(s) of CE 2 2, 3 3 1, 2, 3 3, 5 3, 5 1, 5 2, 3 2, 3 3Multiple CEs Yes Yes Yes Yes Yes Yes Yes YesCE preparation phase 1988–1997 1985–1992 1983–1990 1990–1996 1993–1996 1997–1998 1995–2000 1993–1997 1995–1999 1988–1992CE accomplishment phase 1997–1998 1995–2000 1990–1999 1997–2002 1997–2002 1998–2000 2000–2004 1997–1998 2000–2005 1993–2000Career stage Mid Early Early Mid Early Early Early Mid Mid EarlyField Nano Nano Nano Nano Nano Nano Nano Nano Nano NanoCE accomplishment institution Basic Ind. Lab. Basic Ind. Lab. Basic Ind. Lab. Basic Ind. Lab. Univ. Univ./Gov. Lab. Univ./Gov. Lab. Univ. Univ. Univ.CE accomplishment country JP, US US US US US DE, FR DE NL US USCurrent institution Univ. Univ. Univ. Ind. Lab Univ. Univ./Gov. Lab. Univ./Gov. Lab. Univ. Univ. Univ.Current country D S S S D S S S S S

Case number

11 12 13 14 15 16 17 18 19 20

Type of multiple expert andprize nominations

MultPrizNom MultNom MultNom MultPriz MultNom PrizNom MultPriz PrizNom MultPriz PrizNom

Type(s) of CE 2 3, 5 1, 3, 5 1, 3 3, 4, 5 2 1, 3 3 2 1, 2Multiple CEs Yes Yes Yes Yes Yes YesCE preparation phase 1974–1983 1993–1997 1980–1992 1970–1990s 1990–1996 1986–1991 1988–1994 1975–1990 1981–1985 1985–1992CE accomplishment phase 1983–1988 1998–2002 1993–1999 Mid 1990s 1996–1997 1991–1996 1994–1995 1990–2000 1985–1993 1993–2002Career stage Mid Early Mid Mid Mid Early Early Mid Mid EarlyField Nano Nano Nano Nano Nano HG HG HG HG HGCE accomplishment institution Univ. Univ Univ Univ. Univ. Gov. Lab. Univ./Gov. Lab. Hosp./Gov. Lab. Univ. Univ.CE accomplishment country UK US US US US UK DE FR NL USCurrent institution Univ. Univ. Univ. Univ. Univ. Hosp./Gov. Lab. Univ. Hosp./Gov. Lab. Univ. Univ.Current country S S S S S S S S S S

Notes: MultPriz = multiple prizes, MultNom = multiple nominations, PrizNom = nomination and prize, MultPrizNom = multiple nominations and prizes; CE type abbreviations: 1 = New theoretical concepts, 2 = New empiricaldiscovery, 3 = New methodology, 4 = New instruments, 5 = New synthesis; Country abbreviations: JP = Japan, FR = France, US = United States, DE = Germany, UK = United Kingdom, NL = Netherlands. Current country: D = different;and S = same.

6 ch Pol

vr

3

rlfitbocncai(

aph1gwituoimoap

ifpmgtb(tIcoaEtc

4s

igsacTici

16 T. Heinze et al. / Resear

alid explanations. Thus, the case study provides flexibility for theesearcher to follow the most fruitful path.

.3. Case description

Of the twenty cases, ten were undertaken with researchers cur-ently located in Europe, and ten with creative researchers currentlyocated in the US. Fifteen cases were undertaken in nanotechnology,ve in human genetics. There is some concentration in the creativityypes 3 (new methodology) and 2 (empirical discoveries), followedy type 1 (theoretical concepts) and 5 (new synthesis). Thirteen outf twenty accomplishments can be characterized by more than onereativity type, and 1&3, 2&3 and 3&5 are the most frequent combi-ations in this regard. Five scientists/groups fall in the most selectategory: multiple nominations and multiple prizes. Seven casesre in the multiple nomination category, five in the prize and nom-nation category, and three cases in the multiple prizes categoryTable 3).

Fourteen cases appear as “serial” producers of creativeccomplishments, indicating some institutionalization of effectiveractices for creative research. About half of the accomplishmentsave their roots in the mid-1980s, and three cases even in the late970s, indicating the substantial time necessary to move from ideaeneration to accomplishment. This time lag was due to resistanceithin the research community to accept these novel results and to

ncorporate them in the collective stock of knowledge. In one case,he group leader worked for about one decade solely on the problemntil the research community accepted the novelty and usefulnessf his work. In another case it took about nine years until an exper-mental result, that contradicted an established theory, could be

atched with a new theoretical explanation provided by a collab-ration partner. However, there are also several cases with rapiddvancement from the preparation to the accomplishment phase,articularly when groups were involved in “priority races”.

The creative events were accomplished across a wide range ofnstitutional environments, predominantly in universities (N = 11),ollowed by basic research labs in industry (N = 4), settings in theublic research sector spanning both a university and a govern-ent lab (N = 3), government labs (N = 1), and settings spanning

overnment labs and hospitals (N = 1). The institutions in which ourarget scientists/groups work today are mostly universities (N = 15),ut also mixed settings including hospitals and government labsN = 4). Only one group with a creative accomplishment in an indus-rial basic research lab has remained in this institutional context.t should also be noted that US cases have some geographical con-entration in areas which are already known to have a large sharef R&D, such as the San Francisco Bay area and the greater Bostonrea. In contrast, we are not able to identify such concentration forurope (Table 3). In the next section we report on the organiza-ional and institutional factors that were obtained by systematicross-case analysis.

. Organizational and institutional factors influencingcientific creativity

This section discusses key results from the twenty case stud-es with an emphasis on factors that support creative scientists androups in their research, but also with findings regarding barriers tocientific creativity. The two levels: organizational and institutionalre clearly intertwined, but we discuss them separately for analyti-

al clarity. As noted above, our unit of analysis is the research group.he group is embedded in both an organizational and broadernstitutional environment which contributes to or constrains theapability of group leaders and group members to conduct researchn a way that seems most fruitful to them.

icy 38 (2009) 610–623

The comparison of the dimensions and variables from the casestudy protocol with the emerging dimensions demonstrates thelearning process we underwent as we were exposed to the casestudy material. It shows how some of our initial expectationsabout the basic elements of the context of creativity were enrichedand corrected by it. For example, spatial arrangements and infras-tructure emerged as somewhat more important influences thananticipated. Furthermore, larger institutional developments, suchas the severe budget cuts in (if not breakdown of) the former SovietUnion research system, and the increasing international mobilityof scientists, figured more prominently in our data than initiallyexpected. Finally, the comparatively strong representation of indus-trial R&D laboratories as organizational environments for scientificcreativity, particularly until the early 1990s, corrected our initialexpectation that the academic heartland is exclusively institution-alized in universities and government laboratories. In our policyconclusions, we will return to these and other issues.

4.1. Organizational level

The cases indicate that creative researchers have a numberof distinctive ways in which they manage their research groups.This includes highly effective supervisor–student relationships, thecareful selection of new group members for complementary skillsand attributes, and the flexibility to address new problems or ideasthat arise. We also find that groups in our sample are relatively smallat the time of the main creative event: typically around six to eightresearchers, including juniors and students, and sometimes withonly 2–3 researchers, but they benefit from organizational contextsthat provide sufficient access to a relatively large variety of technicalskills. A frequent factor associated with scientific accomplishmentsis stable research sponsorship, provided either through some kindof basic institutional funding or dedicated funding schemes forjunior scientists.

4.1.1. Research autonomyFirst of all, among the prerequisites for a productive scientific

atmosphere is a context in which there is a set of broadly definedproblems or long term targets and carefully selected individualsare brought in as group members. They are then given freedom topursue a more focused problem within the larger set of problemsas a step toward the broad target. Freedom to define and pursueindividual scientific interests within or beyond a broadly definedthematic area is central to understanding why scientists and theirgroups are highly creative. Individual scientific freedom is madeeven more productive when group members are conducting theirwork in units with many other bright and curious scientific mindswho stimulate each other. Mutual curiosity and interest is a strongnorm in all groups under consideration. For example, one groupleader reported that one of his PhD students got upset becausehe expected the group leader to approach him about the ongoingexperiment more frequently than just twice a week. Subsequently,the group leader communicated more frequently with his students.

4.1.2. Small group sizeSecond, we identified small group size as important organi-

zational dimension for the development of creative work. Thisconfirms findings previously reported (Section 2). In fifteen of ourtwenty case studies, small group size was highly influential for thecreative accomplishment; in another four case studies this variablehad some influence. The analysis of the case studies indicates that

research groups responsible for creative events often start with twopeople, the group leader and a PhD student or a post-doc. Later on,leaders deliberately limited their groups to no more than six toeight researchers (excluding technicians and other support staff).Small group size has a number of advantages. It allows the group

ch Poli

lsmlcawedofrpd

itnqtaiscr

4

tarceyatcfanspstsooic

nitaaSidiotaltaa

T. Heinze et al. / Resear

eader to be actively involved in research and to stimulate effectivecientific exchanges within the group. This is corroborated by theajority of our cases. In contrast, we have reason to believe that

arge research groups are less able than small groups to unleash thereative potential of their group members, because group leadersre forced to spend more time on administration than science whicheakens the crucial link between group members and group lead-

rs. In addition, small groups typically show a lack of hierarchicalecision-making in relation to research activities. The flat structuref communication, with no difference in communication betweenormal hierarchical levels, fuelled the dynamics regarding creativeesearch accomplishments. Furthermore, small group size fostersroductive mentor–student relationships that larger groups haveifficulty to establish and to maintain.

However, several groups (although not all) did grow significantlyn the period following the main creative event. This growth seemso be associated with following up and capitalizing on the opportu-ities that the creative event opened up. This raises an interestinguestion of the value of the research activities in each period, sincehe later ones would be deemed less creative but are critical forctually realizing the potential of the creative event itself. Also,t seems that group growth as a particular reward mechanism incience produces unintended negative effects, such as more hierar-hical decision-making and reduced group leader involvement inesearch activities.

.1.3. Complementary varietyThird, the small groups were typically embedded in an organiza-

ional context that had a complementary variety of scientific skillsnd instrumentation. For example, one scientist reported his expe-ience within an industry lab: “We were going to have lunch and ofourse if you come back from lunch with a thousand ideas becauseverybody is in a slightly different field but not so far from you soou can talk – it’s close enough.” We also found collaboration withinuniversity between theoreticians and experimentalists in physics

hat had enabled a highly recognized creative event in one of ourases. This type of environment provides numerous opportunitiesor stimulation, collaboration, the acquisition of new knowledgend research techniques or access to instrumentation. The combi-ation of small work units in rich research contexts with requisitecientific variety allows for rapid elimination of dead ends whenursuing high-risk ideas. Researchers saw this as a critical factorince it allowed them to quickly test many of the possible pathso a solution for their problem and discard the ones that did noteem promising. Our case studies provided numerous examplesf the importance to scientific creativity of a large, well-endowedrganizational environment, with a good intellectual and technicalnfrastructure and access to a large diversity of skills and interdis-iplinary knowledge across the organization.

However, we found that the scientific diversity of an orga-izational environment alone may not foster creativity unless it

s also linked to organizational arrangements that support mul-idisciplinary contact. These include spatial arrangements, suchs the organization of laboratory facilities or office space, butlso staircases or coffee bars designed to promote interaction.ocial arrangements, such as lunchtime patterns, can also play anmportant role in fostering communication opportunities acrossepartmental borders. In university contexts, for example, these

nteractions take place mostly across department boundaries. Inur cases, some laboratories were more adept than others at facili-ating these exchanges by cultivating a culture of shared resources

nd reduced bureaucratic requirements. For example, one groupeader described the physical infrastructure of the university wherehe creative event materialized, as one in which all disciplinesre united “under one roof”. Walking along the corridors initiatedcquaintanceships and discussion between scientists from vari-

cy 38 (2009) 610–623 617

ous disciplines. “Within three minutes I change from chemistry tophysics to biology. When I walk to the electron microscope, I wentthrough the faculties of physics and biology. These contact pointsare very important.”

4.1.4. Communication with groups in external organizationsFourth, effective communication with other groups that have

complementary knowledge and expertise are an important fac-tor for accomplishing creative events. For example, several theorygroups depended on data from experimental groups in otherresearch institutes, often abroad, or measurement groups neededaccess to sophisticated materials which they could neither producein their own labs nor acquire from specialized companies. Inter-estingly, the emergent communication pattern showed that mostof the in depth communication on matters close to the problemof interest to the group occurred with groups that were outsidethe organization; sometimes they were collaborators and on otherscompetitors. On the other hand, the most important type of com-munication with groups within the organization was of a broadermultidisciplinary nature and related to key skills the group itselfdid not possess. In other words, there is somewhat of an inverserelationship between cognitive distance and physical distance inthe typical patterns of scientific communication.

In addition to our case-study findings on communication acrossorganizational boundaries, we examined the collaboration patternsof creative scientists in a quantitative way. Drawing on our databaseof multiply-nominated highly creative nanotechnology scientists,Heinze and Bauer (2007) find that these scientists collaborate muchmore frequently with other peers than scientists from a compar-ison group of similarly productive researchers; they have largercollaboration networks and more often link disconnected peers.Because of this particular communication pattern, creative scien-tists also publish in a wide range of academic journals, and thusthey are capable of speaking to different audiences and special-ties. Although, for technical reasons, Heinze and Bauer (2007) donot distinguish between intramural and extramural communica-tion, the increasing gap in the volume of collaborations between thetwo types of scientists (i.e. creative researchers versus comparisongroup) demonstrates that getting timely access to complementaryexpertise, skills and instrumentation from other groups is impor-tant to achieving creative events.

4.1.5. Facilitating leadershipFifth, as we had anticipated from the literature review, both

group and organizational leadership are important for the devel-opment of creative work. Effective group leaders perform manyimportant roles. They bridge otherwise disconnected knowledgedomains, carefully select new group members for their comple-mentary skills and attributes, have the flexibility to address newproblems or ideas that arise, help post-docs with a good idea both toattract funds so they can become self-sustaining and develop goodintuition about the right measure of risk to take on with their newidea, and provide a protected area under which group membersconduct their research. We also identified two types of mentor-student relationships. In the first type, mentors provide a researchavenue where their students can develop their particular researchinterests. For instance, mentors provide ideas and directions, butstudents arrange the experimental setting. In the second type, men-tors recruit highly talented students in their group but do not setthem on a particular research track. Their role is more responsiveto the needs of the students.

In addition to group leadership, we also witnessed the impor-tance of directing research organizations through active leadership.In more than half of the cases, the director’s research vision atthe R&D management level of the organization was influential inshaping the creative accomplishment. This vision is not so much

6 ch Pol

amvpfiitjdomdacc

4

rfefrrlubflprisfhgwdTuytfdg

aIeFESnIorgaat

4

ta

18 T. Heinze et al. / Resear

bout goal clarity as it is about goal-fruitfulness in generatingore focused problems that are tractable and significant. Typical

isions are “finding the highest possible storage density for com-uter memory” or “explain the Fragile-X syndrome” which are moreruitful than clear, since the actual initial goal may undergo signif-cant metamorphosis in the process of its pursuit. Perhaps mostmportantly, these leaders gave their junior staff freedom. Half ofhe creative events in our case studies are based on the research ofunior scientists, highlighting the importance of providing indepen-ent research support to outstanding individuals at an early stagef their careers (see career stage in Table 3). As the focus of researchissions is on the solution of a major problem and not on advancing

isciplinary knowledge, this factor also has the attribute of encour-ging interdisciplinary work because research teams were typicallyomposed of people from a variety of disciplinary backgrounds thatontributed to meeting the goals of the mission.

.1.6. Flexible research fundingSixth, flexible research funds were found pivotal in several

esearch breakthroughs in our set of cases. Flexibility means thatunds are not earmarked for specific purposes; that group lead-rs have discretion about when and how to spend them; and thatunds can be used for high risk/high potential investments. In thisegard, flexibility allows scientists to shift research funds in theesearch direction that seems most fruitful to them. One groupeader argued: “When someone has to invest or wants to investnexpectedly much more money in a project, you need the flexi-ility. You need a good funding level in order to be able to affordexibility.“In particular, core institutional funds, which are inde-endent from success in attracting external grant money fromesearch councils, have been found highly important to support-ng scientific accomplishments in eight out of twenty cases, and ineven other cases these funds had some influence. Surprisingly, theour industrial research laboratory groups in our sample receivedigh levels of such institutional core funding. For example, oneroup leader recalled that staff scientists in the industrial laboratoryere not encouraged to write grant proposals, but to communicateirectly with management about their demands for new resources.he interviewee argued that research managers who recruited himsed to say: “I did not hire you to be a manager, I hired you becauseou are a scientist. I want you to do science. I want you to be inhe lab.” In contrast, group leaders in universities were sometimesorced to raise funds for their research with small grants from manyifferent agencies, and progress was achieved only because theseroup leaders used grants from research councils imaginatively.

Another category of flexible funds are large, multi-year researchwards provided to scientists in an ascending stage of their career.n total, six junior scientists were awarded prestigious and well-ndowed individual awards. They were either supported by theörderpreis of the Krupp von Bohlen und Halbach Foundation, theuropean Commission’s Young Investigator Award, the Nationalcience Foundation’s Young Investigator Award, the James McDon-ell Foundation’s Centennial Fellowship, or the Howard Hughes

nstitute’s Investigator Award. An in-depth study of these (andther) award schemes shows that they differ considerably withespect to target group and field, selection process and criteria, bud-et size, and funding duration; and that several of these schemesre powerful tools to support junior scientists (Heinze, 2008). Inddition, we find unanimous consensus among our intervieweeshat too few such awards are currently available.

.2. Institutional level

So far, our findings suggest that there are several organiza-ional factors that support the capabilities of research groups toccomplish creative scientific results. In addition, however, we have

icy 38 (2009) 610–623

identified features in the institutional environment which facili-tate or constrain the creative capabilities of research groups. Inthis section, we will report on such institutional factors. In brief,our cases indicate that job mobility is a necessary condition forcreativity in science, because when scientists accept a new job,they tend to move to research units that offer an opportunity tochange field or to address intrinsically risky research problems.Also, although competition is believed to be an important institu-tional mechanism in science, we observe several cases with little orno influence of competitive pressures in the institutional environ-ment on the preparation or accomplishment phase of a creativeevent. Finally, we find that whilst the conservative proceduresadopted by research funding agencies for allocating grants may beappropriate for “normal science” in established disciplines, theycreate many problems for scientists with original research ideas.

4.2.1. Job mobilityWe did not find that job mobility is as unidirectional a factor

as one might assume. Several group leaders spent many years inthe same place and either had been in one main institution theirentire career or had made one major change in their career. The uni-versity setting has absorbed many researchers who moved awayfrom industry labs when the attractive conditions in the latterdeteriorated. But we did not find evidence of competitive recruit-ing mid-career as a mechanism to encourage creative research.Resources for hosting visitors or spending periods of time withother groups working on the same problem area had a greater effecton the success of the creative pursuits of our interviewees.

We found that when they move, creative scientists tend tomove to research units that offer an opportunity to change fieldor to address intrinsically risky research problems. Fundamentalresearch labs of large, leading industrial companies were a magnetfor such scientists, at least until the early 1990s. Other cases demon-strate that the United States has the most open academic job marketin this regard and offers ample opportunities not only for scientistsfrom Western European countries, but particularly for researchersfrom the former Soviet Union where the public research sectorunderwent severe budget cuts in the early 1990s. In eight of ourcases “immigrant scientists” moved to different countries (includ-ing the United States, France, Japan, and Germany) permanently orfor a long period in order to pursue their research. These researchersreported that they had to work much harder than native scientistsin order to receive recognition, but all described their moves aspivotal for the development of their scientific skills, their successin accomplishing the creative event, and future career options. Weconclude from these observations that job mobility is an impor-tant condition for an institutional environment that is conducive tocreative research.

4.2.2. Reputational competition in the intellectual fieldCompetition for reputation in the intellectual field seems to

work in different ways depending on the phase of the creativework. In seven cases, competition was highly influential in eitherthe creative event preparation or accomplishment phase. Duringthe preparation phase, friendly competition within an organizationis an important motivator. At the cusp, when an important resultseems within grasp, chances are that groups in different organi-zations are also close to a significant achievement and the race tobe first with important results will carry over even into the accom-plishment phase, where important derivative results are pursued ina competitive fashion. For example, several groups were involved

in “priority races” between competing scientists/groups. One groupreceived materials for analysis and characterization from anothergroup, but these materials were given, at the same time, to a rivalgroup. In another case the priority race took place quite fiercelyand without any mutual communication. In both cases, priority was

ch Poli

gat

itonhotatwitga

4

ibbaafiomltwsrbatpa

ogatcarqoa

totosijlbcrmtpc

T. Heinze et al. / Resear

iven informally to the groups in our sample, but their publicationsppeared around the same time and even in the same journal ashose of their rival peers.

Yet, we also observed cases where competitive pressures in thenstitutional environment had little or no influence on the prepara-ion or accomplishment phase. Most of these were found in the fieldf nanotechnology. One reason may be that the field of nanotech-ology is still emerging: the new opportunities in nanotechnologyave made possible many new research paths, such that severalf our creative cases developed these with little initial competi-ion, although still with substantial risk. Some of these paths havettracted considerable attention over time. One interviewee arguedhat in the preparation phase of the accomplishment, this new pathas met with enthusiasm by the research community: “People liked

t, right away. When engineers first saw it, it was immediately a hit,here was ecstatic enthusiasm, it provided understanding, and weot inundated by request for talks and presentations.” Today, therere more than 50 groups worldwide working in this new area.

.2.3. Funding agency behaviorThe manner in which responsibility for a certain field of research

s allocated to a specific division of the funding agency and advisedy experts in the area is a significant barrier for creative researchecause each division tends to award funds to scientists who haverecord of publications in the area. Several group leaders had

chieved their creative event on the basis of moving to a neweld, or integrating new fields with their area of expertise. Butne university group leader said her group “had no chance” to getoney from a funding agency for their “wild ideas”. The group

eader recalled that one always needs preliminary results in ordero compete for external funds. Therefore, getting into a new fieldithout having preliminary results is regarded as “almost impos-

ible”. Another group leader argued that “field-hopping is bad foresearch grant income because it takes five years to build up credi-ility to get research funding”. The current research system does notppear flexible enough to accept that a scientist with an excellentrack record in a given field can have the capability to investigate ahenomenon that involves moving into a new field and that therere synergies in funding such research.

A second problem is connected to funding agencies jumpingnto the bandwagon once the results of breakthroughs in researchenerate attention. Research councils and other funding agenciesllocate funds for a program in the field and often set goals for workhat is either already done, or unrealistic. Although programme offi-ers in funding agencies may have a scientific background, theyre perceived by several of our group leaders to have been out ofesearch for too long to understand how research works. In conse-uence, the guidelines in calls for research proposals, according tone group leader, “tend to be wrong, and they do not present thectual priorities in the field”.

Furthermore, many funding agencies require research proposalso set targets, or give exact details of the likely results, but this isften not possible with exploratory, open-ended research, charac-erized by one group leader as “a meandering path, you’re branchingut, making new things all the time and closing up other things ando you’re moving through a difficult landscape to find your way tonteresting things”. We also found evidence that renewal funding iseopardised if the expected results are not achieved. Several groupeaders argued that a substantial portion of their research had noteen described in any way in the research proposal. So, when itomes to grant renewals, funding agencies might argue that this

esearch did not achieve its goals. Funding agencies also now requireore accountability by scientists, and have increased the adminis-

rative burden on them. They require scientists to provide frequentrogress reports, show they have worked the proposed hours orarried out the working steps according to the original proposal.

cy 38 (2009) 610–623 619

Our case study results suggest that creativity would be promoted byhaving more flexibility in the use of grant income and less demandsfor constant progress reports.

4.3. Factor combinations

The variables at the organizational and institutional level mustbe understood as interrelated contributing influences rather thanas independent, cumulative factors. Set in its context, the creativeresearch process that we were able to detect from interviews andsupporting documents of the cases has mechanisms that combinemany of the influences mentioned above in ways that are morethan their simple aggregation. One of these mechanisms is foundin large R&D laboratories in industry. Several of our highly creativeresearchers were recruited to these labs at an early stage in theircareers, either as post-docs or junior staff researchers. They werethen integrated into a mission-oriented research program whilealso allowed significant freedom to pursue an aspect of the overallprogram that most interested or excited them. This work environ-ment was characterized by organizations that provided significantjob stability for their staff scientists, a base level of funding, andaccess to a large diversity of skills and multidisciplinary knowl-edge across the organization. These research laboratories werewell equipped with instruments and experimental capabilities thatallowed the pursuit of empirical research in any direction the prob-lem might suggest and had in-house, expert technicians to providereliable experimental results in a relatively short period of time.

The second mechanism is the university setting which is ratherdifferent from the industrial labs described above. We found thatscientists that experienced their main creative events in a univer-sity context made efforts to create a setting as close as possible tothe one described above while preserving the broader mission ofacademic work. The central difference between the main modeldescribed above and the academic setting, in the words of oneof our interviewees, is that “industry labs are equipment rich andlabour poor while universities are labour rich and equipment poor.”Therefore, university scientists must devote considerable efforts togain access to the necessary equipment and compensate for thetime demanded by graduate students, who are “in training mode”by selecting problems that are not too time sensitive. Three otherimportant areas in which the academic setting departs from thefundamental research laboratory in industry mechanism are theconspicuous lack of core funding to protect against interruptions ofthe work, the burden of non-research academic obligations such ascommittee work and other service activities of the university, butalso the strong individual freedom associated with a group leader’sposition in the academic setting. In the next section, we review ourfindings in the light of the literature reviewed above, and we pointto future directions for research.

5. Discussion

Previous studies reported that intra-organizational communica-tion across disciplinary or departmental boundaries is associatedwith a productive research climate (Andrews, 1979; Pelz andAndrews, 1966). Although in some cases this view is confirmed,we also found that extramural collaborations have a much greaterbenefit for scientific progress than was previously assumed.The changes in institutional and organizational conditions men-tioned in Section 1, especially the encouragement of research

collaboration, may explain the growing importance of extramuralcollaborations to scientific progress. However, timely access to skillsand partners are not necessarily available within the boundaries ofthe research organization in which the focal group is embedded.As mentioned, there is an inverse relationship between cognitive

6 ch Pol

afiotwsc

a(itfiwgcoll

smpbttss(

dAtottbpptlo

fiir(ofsaAsrarwrchrw

il

development is underway. In the next and final section of this paper,we discuss conclusions for research management and research pol-icy.

20 T. Heinze et al. / Resear

nd physical distance in the typical patterns of communication thatacilitate the accomplishment of creative events. While accessibil-ty to outside skills and resources tends to expand the capabilitiesf research teams to make rapid progress on matters close tohe problem of interest to the group, other teams and resourcesithin organizational boundaries provide a scientific reservoir for

erendipitous observations generated through effective intramuralommunication.

However, the opinion that deep knowledge and specializationt the individual level is integrated at the organizational levelHollingsworth, 2002) is not fully supported. Creative scientistsn our sample typically possess a rather broad scientific profilehat distinguishes them from more specialized normal scientists—anding that is also corroborated by Heinze and Bauer (2007)ith respect to the nanotechnology domain. In addition, several

roup leaders have accomplished creative events because they hadhanged their research field, for example from chemistry to optics,r from semiconductor physics to biophysics. In these cases, intel-ectual variety is integrated at the individual level rather than at theevel of the entire research organization.

However, our results do confirm previous findings regarding thetrong correlation between small group size and per capita perfor-ance (von Tunzelmann et al., 2003). This is noteworthy because

revious studies usually examined productivity and recognition,ut not creative scientific accomplishments or research break-hroughs. Compared to the entire populations of researchers in thewo fields, our twenty cases represent a very small fraction only,o the clear impact of small team size can be interpreted both astrong confirmation but also as an extension of previous evidencebecause our dependent variable is creativity and not productivity).

Interestingly, we obtained mixed findings with respect toepartmental effects as identified by Long and McGinnis (1981) orllison and Long (1990). Whereas the authors argue that once scien-

ists have joined departments, their productivity level matches thatf the department, in several cases of our immigrant group leadershe causality seemed to point in the opposite direction. Typically,hese immigrant scientists had to prove their scientific capabilitiesy working much harder and by contributing substantially highererformance than native scientists before they were invited to joinrestigious departments. On the other hand, group leaders in indus-ry reported that the dynamics and pace within these fundamentalaboratories was an important driver and an inspiration for theirwn creativity.

Perhaps the two (strongly interconnected) variables where ourndings diverge most sharply from previous evidence are fund-

ng and organizational leadership. Most importantly, in the publicesearch sector the predominance of institutional block-grant funds1st stream funding) has been replaced by a new regime basedn competitive research council grants (2nd stream) and privateoundation or industry sponsorship (3rd stream). Whereas previoustudies were concerned with the relationship between perceived orctual resource levels and performance (Andrews, 1979; Pelz andndrews, 1966), our findings suggest that the continued expan-ion of peer-reviewed funding, in particular at early stages of theesearch process, may eliminate ideas that are judged by peerss speculative, unorthodox, or transdisciplinary. Peer-review crite-ia, such as plausibility and validity tend to encourage conformity,hile originality draws upon and encourages dissent. For this

eason, funding arrangements based on peer review tend to dis-riminate against the early stages of exploratory research, as theyave an inherent tendency to support conventional mainstream

esearch and scientific work that follows established research lineshile ignoring visionary and high-risk approaches.

Apart from a conservative bias, the double trend of decreasingnstitutional funding and increasing external sponsorship has ateast two other consequences. First of all, winning funding com-

icy 38 (2009) 610–623

petitions and reviewing increasing amounts of research proposalsrequires substantial time investment by scientists, time that theycan neither spend on laboratory work and group interaction, norfor reading and contemplation. Since style and content of researchproposals are different from presenting arguments and evidencein journal articles, these activities have reduced the precious timeof the group leaders studied. Second, increasing extramural spon-sorship requires a new type of organizational leadership. Whileresearch directors are expected to articulate a research vision, torecruit outstanding personnel, and to motivate scientists (as arguedin previous literature), a new type of expectation has emerged: theyneed the capability to equip research organizations with appropri-ate funding from diverse sponsors and balance research budgets.Organizational leaders need to be successful in acquiring newgrants and opening up additional funding channels. They must becompetent in continuously monitoring the complex landscape offunding agencies and sponsorship programs. These new leadershiprole requirements are non-voluntary because organizations usuallycannot afford to neglect their funding environment.9 Our case stud-ies demonstrate, however, that not only university provosts andinstitute directors but increasingly group leaders are confrontedwith meeting these new roles. The consequences were describedby one group leader, who had formerly worked in a fundamentalindustry laboratory, as follows: “When I came [to this university],I thought I would still do research. I haven’t done just one experi-ment in the seven years since I’ve been here, in the lab myself. Ofcourse, I direct experiments but I don’t carry them out myself. (. . .)People do most of my ideas but I’m a manager”.

The discussion shows that while the institutional literature inscience studies (mostly from the 1950s to the 1970s) offers usefulstarting point with respect to group and organizational variables,new themes have emerged that reflect the broader institutionalchanges in the research system since the 1970s. There are sev-eral examples where we believe our exploratory results couldbe fruitfully extended: (1) to understand the institutional forcesthat led to the marked decline of industrial companies fundingexploratory and basic research (for an introduction see, for exam-ple, Chesbrough, 2003); (2) to explore the changed relationshipbetween industry and public sector research in generating anddiffusing knowledge (for an introduction see, for example, Evans,2004); (3) to analyze quantitatively career trajectories of creative(entrepreneurial) group leaders (for an introduction see, for exam-ple, Levin and Stephan, 1991). More generally, it would be highlydesirable to learn more about the differences between creativityin the natural and technical sciences on the one hand, and in thehumanities and the social sciences on the other hand (for an intro-duction see, for example, Guetzkow et al., 2004). Also, we needmore general theoretical propositions that serve as a frameworkfor generalization and for stating cumulative hypotheses. Clearly,the renewed interest in organizational and institutional questionsabout the governance of research is an opportunity for more sys-tematic theorizing. Several colleagues have started to investigatecontingent variables at the group, organization and institutionallevel with organizational research outcomes (see, for example,Jordan, 2006; and the contributions in Jansen, 2007). Although agovernance theory of research organizations does not yet exist, its

9 There are exceptions, such as the German Max-Planck-Society whose institutesreceive permanent institutional funding. However, many Max-Planck institutes areactively seeking external funding in the 2nd and 3rd stream (for an overview of theGerman research landscape see Heinze and Arnold, 2008 and Heinze and Kuhlmann,2008).

ch Poli

6

aUgtsmccamtr

aeiafoepitcMcbatssanaaa

msnatzriTrimlstmtasasdb

as

Howard Hughes Foundation US 8.1.2007*

T. Heinze et al. / Resear

. Policy conclusions

This paper uses the results of a new nomination method to selectset of highly creative scientific accomplishments in Europe and thenited States in two scientific fields, nanotechnology and humanenetics, in order to explore organizational and institutional fac-ors associated with creative research. We employ a multiple-casetudy approach of twenty highly creative research accomplish-ents encompassing diverse types of scientific creativity. In these

ases, we did not find remarkable differences by scientific field orreativity type; rather, the principal findings relate to mechanismst the group, organizational, and institutional level. Several of theseechanisms are relevant for current policy making. In this sec-

ion, we explore some of the lessons for research management andesearch policy.

We collected evidence that a stimulating work context offersmple opportunities for fruitful scientific exchange, often acrossstablished cognitive domains. In contrast, the exploration modes weakened when research groups are large and organized hier-rchically. Despite this finding, large hierarchical groups can beound almost in every university and in every government lab-ratory. Large groups are either valued by research cultures (forxample in Germany), or they emerge systematically because theredominant funding mechanism produces large structure after

nitial scientific successes (for example in the United Kingdom andhe United States). Several of our case studies demonstrate that suc-essful groups can grow substantially in a short time period (theatthew effect, see Merton, 1957). Clearly, this mechanism is in

onflict with the fact that breakthroughs are typically accomplishedy small groups. Therefore, senior research management should beware that highly creative research can be more difficult to under-ake in large research groups, and that path-opening solutions incience seem to emerge more readily from small research units. Theize of research groups should be considered an important man-gement objective for effective research governance, particularly inew and frontier research areas. Policy makers also need to thinkbout new mechanisms that relieve successful scientists from man-ging too many projects and too large groups, because large groupsnd hierarchical structures are barriers for creative research.

Well-endowed research institutes with a good intellectual andaterial facilities infrastructure and access to a large diversity of

kills and multidisciplinary knowledge provide numerous inter-al and external opportunities for stimulation, collaboration, thecquisition of new knowledge and research techniques, or accesso instrumentation. Interactions within and across such organi-ational contexts are particularly fruitful when groups work inelated and complementary fields of expertise and when researchnstitutes have the requisite variety of skills and knowledge.hose involved in the planning of periodic reorganizations ofesearch institutions should ensure that changes maintain andncrease the breadth of disciplinary expertise available. Further-

ore, the scientific diversity of research organizations may be moreikely to support creativity if linked to organizational, social andpatial arrangements that support planned and unplanned mul-idisciplinary contact. Organizational opportunities may include

ulti-disciplinary or cross-unit seed research awards, lab staff rota-ions, cross-training and inter-unit seminars and exchanges. Spatialrrangements, such as the allocation of offices, junior researchpace, hallways, coffee bars or laboratory facilities, and socialrrangements, such as lunchtime patterns, may also be organizedo as to encourage the opportunities for communication across

epartmental borders, between staff, regardless of their status, andetween disciplines.

For more than three decades, the science system has been oper-ting under “steady state” conditions (Ziman, 1994). Steady-statecience has been accompanied by a decreasing lifespan for research

cy 38 (2009) 610–623 621

projects, and this trend has given strength to the forces that elimi-nate unorthodox and original ideas. In several of our cases a certainkind of funding helped to reap the fruits of novel ideas: fundingbased on trust that scientists will do their work as well as they can.However, for many years, scientific research activity has been con-fronted with a high level of distrust, and this distrust is visible in thewidespread use of performance indicators and by growth in mea-sures such as evaluation, progress reports, management reports,audit certificates and the like. Permissive, trust-based funding doesnot appear to play the central role that it should play in the budgetsof funding agencies. Therefore, policy makers should now be readyto accept that there is a clear need to provide appropriate fundingfor exploratory and high-risk research, even under the regime ofsteady-state science.

When considering the institutional landscape that fosters cre-ative research, it should be noted that a conspicuously favorableenvironment was found in a set of fundamental R&D labs in industrythat, due to changes in market conditions and industry strategies,has been drastically reduced in size over the last decade. Many ofthe industrial labs mentioned by our interviewees no longer existand some that do no longer allow the sort of work that resulted inthe creative events that earned them the recognition reflected inour nomination process. This raises another sort of policy questionthat is not limited to lessons for R&D management and grant mech-anisms aimed at stimulating individual choices by researchers.Rather, it points to the overall direction of the innovation systemand whether this change in research arrangements will have aneffect on its capacity to be as creative as it has been to date.

Acknowledgements

This paper is based on research undertaken by the Project onCreativity Capabilities and the Promotion of Highly InnovativeResearch in Europe and the United States (CREA), sponsored bythe European Union Newly Emerging Science and Technologiesprogram (see: http://www.cherry.gatech.edu/crea/). Researchassistance for the expert nomination survey and the developmentof data on scientific prizes was provided by Gerrit Bauer, AjayBhaskarabhatla, Taehyun Jung, Li Tang, Jue Wang, and JingjingZhang. We are grateful for survey responses supplied by numerousexperts in Europe and the United States. An earlier version ofthis paper was presented at the 40th Anniversary Conferenceof Science and Technology Policy Research (SRPU), University ofSussex, Brighton, UK, 11–13 September, 2006. We appreciate themany helpful comments and suggestions provided by conferenceparticipants and by subsequent reviewers.

Appendix A. List of interviews

Country Interview dates

Cambridge University UK 5.4.2006*, 16.7.2006Columbia University US 22.1.2007*, 29.1.2007Eidgenössische Technische

Hochschule ZürichCH 7.3.2006

Emory University US 22.11.2005Erasmus-MC, Rotterdam University NL 14.2.2006Forschungszentrum

Karlsruhe/Universität KarlsruheDE 14.7.2006

Georgia Institute of Technology US 19.1.2007, 30.1.2007*,6.3.2006, 13.7.2006

Harvard University US 6.5.2006Hopital Necker, Paris FR 17.3.2006

IBM Watson Lab US 5.12.2005Instituut voor Atoom-en

Molecuulfysica, AmsterdamNL 11.7.2006

Massachusetts Institute ofTechnology

US 4.4.2006, 6.4.2006*

6 ch Pol

A

MMMNPR

RSTTUUUUU

U

UUVWW

W

NM

R

AA

A

B

B

B

B

B

B

B

C

C

C

C

C

D

D

E

E

E

F

22 T. Heinze et al. / Resear

ppendix A (Continued )

Country Interview dates

ax-Planck-Institut, Dresden DE 5.5.2006ax-Planck-Institut, Stuttgart DE 27.2.2006cDonnell Foundation US 20.11.2006*ew York University US 8.8.2006*urdue University US 9.2.2007*heinisch-WestfälischeHochschule Aachen

DE 20.4.2006

ice University US 31.1.2007*tanford University US 22.3.2006echnische Universität München DE 11.1.2007echnische Universiteit Delft NL 1.6.2006C Santa Barbara US 28.2.2007*niversität Bayreuth DE 22.5.2006*, 10.11.2006niversität Heidelberg DE 6.2.2006niversität Karlsruhe DE 11.12.2005, 9.2.2006niversité Louis Pasteur,Strasbourg

FR 2.6.2006

niversity Illinois,Urbana-Champaign

US 20.3.2006

niversity of Connecticut US 23.2.2007niversity of Iowa US 11.9.2006olkswagen-Stiftung DE 14.11.2006*ellcome Trust UK 2.5.2006*estern General Hospital,Edinburgh

UK 28.11.2005, 24.8.2006*

estern Michigan University US 9.1.2007*

otes: *Interview by telephone, all other interviews in-person. Interview date in Day,onth, Year format.

eferences

mabile, T.M., 1996. Creativity in Context. Westview Press, Boulder, Colorado.llison, P.D., Long, S., 1990. Departmental effects on scientific productivity. American

Sociological Review 55, 469–478.ndrews, R. (Ed.), 1979. Scientific Productivity: The Effectiveness of Research Groups

in Six Countries. Cambridge University Press, Cambridge.aird, L., 1986. What characterizes a productive research department? Research in

Higher Education 25, 211–225.irnbaum, P.H., 1983. Predictors of long-term research performance. In: Epton, S.R.,

Payne, R.L., Pearson, A.W. (Eds.), Managing Interdisciplinary Research. Wiley andSons, New York, pp. 47–59.

erka, W., Brix, E., Smekal, C. (Eds.), 2003. Woher Kommt das Neue? Kreativität inWissenschaft und Kunst. Böhlau, Wien.

land, C.J., Ruffin, M.T., 1992. Characteristics of a productive research environment:a literature review. Academic Medicine 67, 385–397.

lau, J., 2005. Europe seeks greater creativity in basic research. Research TechnologyManagement May–June, 2–3.

ourke, P., Butler, L., 1999. The efficacy of different modes of funding research: per-spectives from Australian data on the biological sciences. Research Policy 28,489–499.

urt, R.S., 2004. Structural holes and good ideas. American Journal of Sociology 110(2), 349–399.

hesbrough, H., 2003. Open Innovation. The New Imperative for Creating and Prof-iting from Technology. Harvard Business School Press, Boston.

hompalov, I., Genuth, J., Shrum, W., 2002. The organization of scientific collabora-tions. Research Policy 31 (5), 749–767.

ole, S., Cole, J., 1967. Scientific output and recognition: a study in the operation ofthe reward system in science. American Sociological Review 32, 377–390.

orley, E.A., Boardman, P.C., Bozeman, B., 2006. Design and the management ofmulti-institutional research collaborations: theoretical implications from twocase studies. Research Policy 35 (7), 975–993.

rane, D., 1965. Scientists at major and minor universities: a study of productivityand recognition. American Sociological Review 30, 699–714.

ill, D.D., 1985. Theory versus practice in the staffing of R&D laboratories. R&DManagement 15, 227–241.

unbar, K., 1997. How scientists think: online creativity and conceptual change inscience. In: Ward, T.B., Smith, S.M., Vaid, S. (Eds.), Conceptual Structures andProcesses: Emergence, Discovery and Change. APA Press, Washington, DC, pp.461–493.

isenhardt, K.M., 1989. Building theories from case study research. The Academy ofManagement Review 14, 532–550.

tzkowitz, H., 2003. Research groups as ‘quasi-firms’: the invention of the

entrepreneurial university. Research Policy 32 (1), 109–121.

vans, J.A., 2004. Sharing the Harvest: The Uncertain Fruits of Public/Private Collab-oration in Plant Biotechnology. Doctoral Dissertation, Department of Sociology,Stanford University.

leming, L., Mingo, S., Chen, D., 2007. Collaborative brokerage, generative creativity,and creative success. Administrative Science Quarterly 52, 443–475.

icy 38 (2009) 610–623

Florida, R., 2002. The Rise of the Creative Class: And How It’s Transforming Work,Leisure, Community and Everyday Life. Basic Books, New York.

George, A., Bennett, A., 2005. Case Studies and Theory Development in the SocialSciences. MIT Press, Cambridge.

Guetzkow, J., Lamont, M., Mallard, G., 2004. What is originality in the humanitiesand the social sciences? American Sociological Review 69, 190–212.

Hage, J., 2006. Radical innovation and institutional change: French biomedicine,1888–1919. In: Annual Meeting of the American Association for the Advance-ment of Science, 17 February.

Heinze, T., 2004. Nanoscience and nanotechnology in Europe: analysis of publica-tions and patent applications including comparisons with the United States.Nanotechnology Law & Business 1, 427–445.

Heinze, T., 2008. How to sponsor ground-breaking research: a comparison of fundingschemes. Science & Public Policy 35, 302–318.

Heinze, T., Arnold, N., 2008. Governanceregimes im Wandel. Eine Analyse desaußeruniversitären, staatlich finanzierten Forschungssektors in Deutschland.Kölner Zeitschrift für Soziologie und Sozialpsychologie 60 (4), 686–722.

Heinze, T., Bauer, G., 2007. Characterizing creative scientists in nano S&T: produc-tivity, multidisciplinarity, and network brokerage in longitudinal perspective.Scientometrics 70 (3), 811–830.

Heinze, T., Kuhlmann, S., 2008. Across institutional boundaries? Research collabora-tion in German public sector nanoscience. Research Policy 37, 888–899.

Heinze, T., Shapira, P., Senker, J., Kuhlmann, S., 2007. Identifying creative researchaccomplishments: methodology and results for nanotechnology and humangenetics. Scientometrics 70 (1), 125–152.

Hemlin, S., Allwood, C.M., Martin, B.R., 2004. Creative Knowledge Environments. TheInfluences on Creativity in Research and Innovation. Edward Elgar, Cheltenham,UK.

Henkel, M., 1999. The modernization of research evaluation: the case of the UK.Higher Education 38 (1), 105–122.

Hessenbruch, A., 2004. Nanotechnology and the negotiation of novelty. In: Baird,D., Nordmann, A., Schummer, J. (Eds.), Discovering the Nanoscale. IOS Press,Amsterdam, pp. 135–144.

Hollingsworth, R., 2000. Major Discoveries and Biomedical Research Orga-nizations: Perspectives on Interdisciplinarity, Nurturing Leadership, andIntegrated Structure and Cultures. Paper Presented at a Research Symposium,University of Saskatchewan. <http://www.usask.ca/research/communications/symposium/hollingsworth.php>.

Hollingsworth, R., 2002. Research organizations and major discoveries in twentieth-century science: a case of excellence in biomedical research. WZB DiscussionPaper P02-003, Berlin.

Hollingsworth, R., 2004. Institutionalizing excellence in biomedical research: thecase of Rockefeller University. In: Stapleton, D.H. (Ed.), Creating a Tradition ofBiomedical Research. Contributions to the History of the Rockefeller University.Rockefeller University Press, New York, pp. 17–63.

Jansen, D. (Ed.), 2007. New Forms of Governance in Research Organizations. FromDisciplinary Theories towards Interfaces and Integration. Springer, Dordrecht.

Jordan, G.B., 2006. Factors influencing advances in basic and applied research:variation due to diversity in research profiles. In: Hage, J., Meeus, M. (Eds.),Innovation, Science, and Institutional Change. Oxford University Press, Oxford,pp. 173–195.

Knorr-Cetina, K., 1981. The Manufacture of Knowledge. Oxford University Press,Oxford.

Knorr-Cetina, K., 1999. Epistemic Cultures. How The Sciences Make Knowledge. Har-vard University Press, Cambridge, MA.

Knorr-Cetina, K., Mulkay, M. (Eds.), 1983. Science Observed: Perspectives on theSocial Study of Science. SAGE, Beverly Hills, CA.

Kuhn, T., 1962. The Structure of Scientific Revolutions. University of Chicago Press,Chicago.

Langfeldt, L., 2001. The decision-making constraints and processes of grant peerreview, and their effects on the review outcome. Social Studies of Science 31,820–841.

Latour, B., Woolgar, S., 1979. Laboratory Life. The Construction of Scientific Facts.Princeton, New Jersey.

Laudel, G., 2006. The art of getting funded: how scientists adapt to their fundingconditions. Science and Public Policy 33, 489–504.

Lepori, B., van den Besselaar, P., Dinges, M., Potì, B., Reale, E., Slipersæter, S., Thèves, J.,van der Meulen, B., 2007. Comparing the evolution of national research policies:what patterns of change? Science and Public Policy 34, 372–388.

Levin, S.G., Stephan, P.E., 1991. Research Productivity Over the Life Cycle: Evidencefor Academic Scientists. The American Economic Review 81, 114–132.

Lewenstein, B.V., 1992. Cold fusion and hot history. Osiris 7, 135–163.Long, J.S., McGinnis, R., 1981. Organizational context and scientific productivity.

American Sociological Review 46, 422–442.Mansfield, R.S., Busse, T.V., 1981. The Psychology of Creativity and Discovery. Nelson-

Hall, Chicago. IL.March, J.G., 1991. Exploration and exploitation in organizational learning. Organiza-

tion Science 2, 71–87.March, J.G., 2007. The study of organizations and organizing since 1945. Organization

Studies 28, 9–19.

Maritain, J., 1977. Creative Intuition in Art and Poetry. Princeton University Press,

Princeton, NJ.Meltzer, L., 1956. Productivity in organizational settings. Journal of Social Issues 12,

32–40.Meltzer, L., Salter, J., 1962. Organization structure and performance and job satisfac-

tion of scientists. American Sociological Review 27, 351–362.

ch Poli

M

M

M

M

N

O

O

O

O

P

P

PR

R

S

S

S

T. Heinze et al. / Resear

erton, R.K., 1957. Social Theory and Social Structure. Free Press of Glencoe, NewYork.

erton, R.K., Barber, E., 2004. The Travels and Adventures of Serendipity. A Study inSociological Semantics and the Sociology of Science. Princeton University Press,Princeton.

umford, M.D., Scott, G., Gaddis, B., Strange, J.M., 2002. Leading creative people:orchestrating expertise and relationships. The Leadership Quarterly 13, 705–750.

ünch, R., 2008. Stratifikation durch Evaluation. Mechanismen der Konstruktionvon Statushierarchien in der Forschung. Zeitschrift für Soziologie 37, 60–80.

agel, S.S. (Ed.), 2002. Policy Creativity: New Perspectives. Nova Science Publishers,Hauppauge.

bstfeld, D., 2005. Social networks, the tertius iungens orientation, and involvementin innovation. Administrative Science Quarterly 50, 100–130.

chse, R., 1990. Before the Gates of Excellence. The Determination of Creative Genius.Cambridge University Press, Cambridge, UK.

tten, H.R., 2001. Wie kreativ ist der homo politicus? Überlegungen zu Max Weber.In: Bluhm, H., Gebhardt, J. (Eds.), Konzepte Politischen Handelns. Kreativität –Innovation – Praxen. Nomos, Baden-Baden.

wen-Smith, J., 2003. From separate systems to a hybrid order: accumulative advan-tage across public and private science at research one universities. ResearchPolicy 32 (6), 1081–1104.

elz, D.C., 1964. The ‘creative years’ and the research environment. IEEE Transactionsin Engineering Management EM-11, 23–29.

elz, D.C., Andrews, F.M., 1966. Scientists in Organizations. Productive Climates forResearch and Development. John Wiley and Sons, New York.

olanyi, M., 1969. Knowing and Being. Chicago University Press, Chicago.eskin, B.F., 1977. Scientific productivity and the reward structure of science. Amer-

ican Sociological Review 42, 491–504.odan, S., Galunic, C., 2004. More than network structure: how knowledge het-

erogeneity influences managerial performance and innovativeness. StrategicManagement Journal 25, 541–562.

chimank, U., 2005. “New public management” and the academic professions: reflec-tions on the German situation. Minerva 43, 361–376.

hapira, P., Kuhlmann, S. (Eds.), 2003. Learning from Science and Technology PolicyEvaluation. Edward Elgar, Cheltenham, UK.

hepard, H.A., 1956. Creativity in R&D teams. Research and Engineering, October,10–13.

cy 38 (2009) 610–623 623

Simonton, D.K., 1999. Origins of Genius: Darwinian Perspectives on Creativity.Oxford University Press, New York.

Simonton, D.K., 2004. Creativity in Science: Chance, Logic, Genius, and Zeitgeist.Cambridge University Press, Cambridge, UK.

Stein, M.I., 1962. Creativity and the scientist. In: Barber, B., Hirsch, W. (Eds.), TheSociology of Science. The Free Press of Clencoe, New York, pp. 329–343.

Sternberg, R.J., 2003. Wisdom, Intelligence, and Creativity Synthesized. CambridgeUniversity Press, Cambridge, UK.

Sternberg, R.J., Ohara, L.A., Lubart, T.I., 1997. Creativity as investment. CaliforniaManagement Review 40, 8–32.

Stumpf, H., 1995. Scientific creativity: a short overview. Educational PsychologyReview 7 (3), 225–241.

Sulston, J., Ferry, G., 2002. The Common Thread: A Story of Science, Politics, Ethicsand the Human Genome. Joseph Henry Press Books, Washington, DC.

Sutton, R.I., 2002. Weird ideas that spark innovation. MIT Sloan Management Review,83–87.

Thèves, J., Lepori, B., Larédo, P., 2007. Changing patterns of public research fundingin France. Science and Public Policy 34, 389–399.

Uzzi, B., Spiro, J., 2005. Collaboration and creativity: the small world problem. Amer-ican Journal of Sociology 111, 447–504.

van Leeuwen, T., Tijssen, R., 2000. Interdisciplinary dynamics of modern sci-ence: analysis of cross-disciplinary citation flows. Research Evaluation 9 (3),183–187.

von Tunzelmann, N., Ranga, M., Martin, B., Geuna, A., 2003. The Effects of Size onResearch Performance: A SPRU Review. University of Sussex, Brighton.

Weinert, F.E., 2000. Individuelle Kreativität und kollektives Ergebnis. Der Architekt.Zeitschrift des Bundes deutscher Architekten 48 (3), 24–31.

Whitley, R., 2000. The Intellectual and Social Organization of the Sciences, 2nd ed.Oxford University Press, Oxford.

Windolf, P., 1997. Expansion and Structural Change: Higher Education in Germany,the United States, and Japan, 1870–1990. Westview Press, Boulder.

Youtie, J., Shapira, P., Porter, A., 2008. National nanotechnology publications andcitations. Journal of Nanoparticle Research 10 (6), 981–986.

Ziman, J., 1994. Promotheus Bound. Science in a Dynamic Steady State. CambridgeUniversity Press, Cambridge.

Zuckermann, H., 1977. Scientific Elite: Nobel Laureates in the United States. FreePress, New York.

Per pust ak aan Su l t anah Z an ar i ah

TITLE : SOURCE

Transformational Leardership, Creativity, and Organizational Innovation

http://www.sciencedirect.com/

Available online at www.sciencedirect.com

h 62 (2009) 461–473

Journal of Business Researc

Transformational leadership, creativity, and organizational innovation

Lale Gumusluoglu a,⁎, Arzu Ilsev b,1

a Bilkent University, Faculty of Business Administration, Department of Management, Bilkent, Ankara, Turkeyb Hacettepe University, Faculty of Economics and Administrative Sciences, Department of Business Administration, Beytepe, Ankara, Turkey

Received 1 December 2006; received in revised form 1 June 2007; accepted 1 July 2007

Abstract

This study proposes a model of the impact of transformational leadership both on followers' creativity at the individual level and on innovationat the organizational level. The model is tested on 163 R&D personnel and managers at 43 micro- and small-sized Turkish software developmentcompanies. The results suggest that transformational leadership has important effects on creativity at both the individual and organizational levels.At the individual level, the results of hierarchical linear modeling show that there is a positive relationship between transformational leadershipand employees' creativity. In addition, transformational leadership influences employees' creativity through psychological empowerment. At theorganizational level, the results of regression analysis reveal that transformational leadership positively associates with organizational innovation,which is measured with a market-oriented criterion developed specifically for developing countries and newly developing industries. Theimplications of the findings along with some potential practical applications are discussed.© 2008 Elsevier Inc. All rights reserved.

Keywords: Transformational leadership; Creativity; Organizational innovation; Turkey

Innovation through creativity is an important factor in thesuccess and competitive advantage of organizations (Woodmanet al., 1993) as well as for a strong economy (Drucker, 1985).Today, almost all organizations face a dynamic environmentcharacterized by rapid technological change, shortening productlife cycles, and globalization. Organizations, especially techno-logically-driven ones, need to be more creative and innovativethan before to survive, to compete, to grow, and to lead (Junget al., 2003; Tierney et al., 1999).

The literature includes several definitions of creativity andinnovation. A widely accepted definition states that creativity isthe production of novel and useful ideas, and innovation is thesuccessful implementation of creative ideas within an organiza-tion (Amabile, 1983, 1998; Amabile et al., 1996). Thus, creativityis at the individual level, while innovation is at the organizationallevel (Oldham and Cummings, 1996).

⁎ Corresponding author. Tel.: +90 312 290 2319; fax: +90 312 266 4958.E-mail addresses: tomruk@bilkent.edu.tr (L. Gumusluoglu),

ailsev@hotmail.com (A. Ilsev).1 Tel.: +90 312 299 2064; fax: +90 312 299 2055.

0148-2963/$ - see front matter © 2008 Elsevier Inc. All rights reserved.doi:10.1016/j.jbusres.2007.07.032

Interest is growing in the influence of transformationalleadership on creativity and innovation. Transformational leadersraise the performance expectations of their followers (Bass, 1995)and “seek to transform followers' personal values and self-concepts, andmove them to higher level of needs and aspirations”(Jung, 2001: 187). Researchers have studied the effects oftransformational leadership on the performance of followers andorganizations in the past decade (e.g., Dvir et al., 2002; HowellandAvolio, 1993; Lowe et al., 1996), but only a handful of studieshave examined the effects of this type of leadership on followers'creativity. The conflicting findings as well as the experimentalnature of these studies prompt the present research whichprimarily aims to understand the effects of transformationalleadership on followers' creativity in a real setting.

The intrinsic motivation perspective dominates the creativityliterature. This perspective argues that people are most creativeprimarily via intrinsic motivation (e.g., Amabile, 1983, 1998;Tierney et al., 1999). Amabile et al. (1996) further suggest thatan individual's perception of the work environment is a keydeterminant of his or her creativity. According to their model,the perceived work environment influences the creative work

462 L. Gumusluoglu, A. Ilsev / Journal of Business Research 62 (2009) 461–473

carried out in organizations; that is, the psychological meaningemployees attach to events in their organizations affect theirmotivation to generate new ideas. Previous literature hasexamined several psychological work environment perceptionsthat can influence creative work in organizations. For example,studies show perceptions of support for innovation (Scott andBruce, 1994) and psychological empowerment (Deci et al.,1989) to be important sources of creativity.

Several studies report that transformational leaders empowertheir followers (e.g., Jung and Sosik, 2002) and establish aninnovative climate (Jung et al., 2003). However, availableresearch does not examine the mediating roles of empowermentand innovative climate in the relationship between transforma-tional leadership and followers' creativity. This study proposesthat employees' intrinsic motivation and perceptions of thework environment, specifically perceptions of support forinnovation and empowerment, are the mechanisms underlyingthe effects of transformational leadership on creativity.

Along with the relationship between transformational leader-ship and followers' individual-level creativity and the under-lying potential mediating processes, this study also investigatesthe relationship between transformational leadership andinnovation at the organizational level. Extending the model tothis level of analysis should be a significant contribution to theliterature because only a handful of empirical studies havelooked at the effect of transformational leadership on organiza-tional innovation (e.g., Jung et al., 2003). More importantly,since innovation at the organizational level is the result ofcreative efforts and achievements in commercial organizations,gaining an understanding of the effect of this form of leadershipon organizational innovation is as important as understanding itseffect on employees' creativity. This study aims to examine theeffects of transformational leadership on creativity at the indi-vidual level and innovation at the organizational level. Fig. 1shows the multilevel model developed for this purpose.

According to the proposed model, transformational leader-ship positively relates to followers' creativity. Followers'intrinsic motivation, psychological empowerment, and percep-tion of support for innovation mediate this effect. At theorganizational level, transformational leadership positivelyrelates to organizational innovation. Furthermore, individual-level creativity influences innovation at the organizational level.

Fig. 1. The prop

1. Theoretical background and hypotheses

1.1. Transformational leadership and individual creativity

Burns (1978) introduces the transformational leadershiptheory. Bass and Avolio (1995) further developed the theory.According to them, transformational leadership has fourcomponents; charismatic role modeling, individualized con-sideration, inspirational motivation, and intellectual stimulation.Using charisma, the leader inspires admiration, respect, andloyalty, and emphasizes the importance of having a collectivesense of mission. By individualized consideration, the leaderbuilds a one-to-one relationship with his or her followers, andunderstands and considers their differing needs, skills, andaspirations. By inspirational motivation, the leader articulates anexciting vision of the future, shows the followers how to achievethe goals, and expresses his or her belief that they can do it. Byintellectual stimulation, the leader broadens and elevates theinterests of his or her employees (Bass, 1990b), and stimulatesfollowers to think about old problems in new ways (Bass, 1985).

Transformational leadership behaviors closely match thedeterminants of innovation and creativity at the workplace,some of which are vision, support for innovation, autonomy,encouragement, recognition, and challenge (Elkins and Keller,2003). This leader's behaviors are likely to act as “creativity-enhancing forces”: individualized consideration “serves as areward” for the followers by providing recognition andencouragement; intellectual stimulation “enhances exploratorythinking” by providing support for innovation, autonomy, andchallenge; and inspirational motivation “provides encouragementinto the idea generation process” by energizing followers to worktowards the organization's vision (Bass and Avolio, 1995; Sosiket al., 1998: 113). The resulting intrinsic motivation felt by thefollowers is an important source of creativity (Tierney et al.,1999).

Moreover, since feelings of self-efficacy lead to higher creativeperformance (Mumford and Gustafson, 1988; Redmond et al.,1993), transformational leaders who develop their followers' self-efficacy (Bass, 1990b) can positively affect their followers'creativity. Employees with enhanced self-efficacy are more likelyto be motivated to generate novel ideas and solutions.Furthermore, the emotional relationships a transformational

osed model.

463L. Gumusluoglu, A. Ilsev / Journal of Business Research 62 (2009) 461–473

leader builds with his or her followers (Bass, 1990b) might beanother creativity-enhancing force as emotional attachment islikely to lead to higher levels of creativity (Hunt et al., 2004). Thatis, employees are more likely to respond to this leader's challengeand support for innovation by exhibiting more creativity in theirtasks, given their emotional ties with their leader.

Although transformational leadership seems to be relevant inenhancing followers' creativity, only a few studies investigatethis relationship empirically. With the exception of the researchof Shin and Zhou (2003), these studies (Sosik et al., 1998, 1999;Jung, 2001; Kahai et al., 2003; Jaussi and Dionne, 2003), usedata from the U.S., in experimental settings, and using studentsamples; they report inconsistent findings about this leader'seffect on his followers' creativity at both the individual andgroup level.

The conflicting findings of two of the above studies re-garding creativity at the individual level are of particular interestto the present research. Jaussi and Dionne (2003) report thattransformational leadership does not relate to individualcreative performance of the participating students. This con-trasts with the findings of Shin and Zhou (2003), who inves-tigate the effects of transformational leadership on employees'individual-level creativity in a real business setting using asample of 260 R&D employees and their supervisors from 46companies; they found that Korean employees exhibit morecreativity under transformational leadership. This inconsistencymight readily stem from the different designs of the two studies(experimental vs. real workplace), the different contexts theywere conducted in (the U.S. vs. South Korea), and the differentsamples used (students vs. employees). Therefore, as Mumfordand Licuanan (2004) suggest, more studies should investigate inreal settings whether transformational leadership positivelyaffects followers' creativity.

The present field study proposes a positive relationshipbetween transformational leadership and followers' individual-level creativity primarily due to the creativity-enhancingbehaviors displayed by this leadership. Second, the fit betweenthis leadership style and the collectivist orientation of Turkishpeople is likely to strengthen this proposition on the positivedirection. In collectivist societies, followers expect their leadersto take care of them while followers are ready to identify withtheir leaders' vision and demonstrate their loyalty (Bass, 1990a).Bass (1995) argues that transformational leadership is morelikely to emerge in collectivist cultures than in the individualisticcultures of the West. Jung and Yammarino (2001) report that theeffects of this kind of leadership are stronger among collectiviststhan among individualists. H1: Transformational leadershiprelates positively to followers' creativity.

1.2. Transformational leadership and intrinsic motivation

Intrinsic motivation refers to the motivational state in whichemployees are interested in a task for its own sake, rather thanfor the external outcomes or rewards related to the task (Deciand Ryan, 1985). Intrinsic motivation is one of the mostimportant sources of creativity (Amabile 1983, 1998; Amabileet al., 1996); when an employee is intrinsically attracted to a

task, he or she is more likely to focus on it and explore andexperiment with it, hence exhibit more creative behavior.Empirical studies have also shown that when employees areintrinsically motivated, they exhibit more creative performance(e.g., Tierney et al., 1999; Jaussi and Dionne, 2003).

Oldham and Cummings (1996) report that supportive super-vision is an important determinant of intrinsic motivation andcreativity at work. In line with this, transformational leaders whocare for their employees' feelings and needs, facilitate their skilldevelopment, show them ways to achieve the goals and expressconfidence in them (Bass, 1990b) are likely to enhance theiremployees' interest in their tasks. This study expects thatemployees under this kind of supportive leadership will beintrinsically motivated and ultimately more creative. Thechallenging vision by this leader's inspirational motivation islikely to enhance the excitement and meaning that employeesattribute to their work. The recognition and encouragement thatindividual consideration by a transformational leader offers arelikely to increase the willingness of the employees to focus moreand do better in their tasks; and the challenge from this leader'sintellectual stimulation is likely to energize the employees toexplore and be more attracted to different dimensions of theirtasks. According to Amabile (1983), these all lead to anenhancement of interest in the task itself and higher creativeachievements.

A few studies test the mediating role of intrinsic motivation.Shin and Zhou (2003) find that intrinsic motivation partiallymediated the influence of transformational leadership onfollowers' creativity. For employees high on conservation (i.e.,employees who value conformity, security, and tradition) intrinsicmotivation fully mediates this relationship (Shin and Zhou 2003).

Based on the discussion above and the high level ofuncertainty avoidance and conservation among Turkish people(Hofstede, 1980), the study proposes that transformationalleadership affects followers' creativity through intrinsic motiva-tion. Therefore, H2: Intrinsic motivation mediates the relation-ship between transformational leadership and followers'creativity.

1.3. Transformational leadership and psychologicalempowerment

Psychological empowerment is another source of creativity(Deci et al., 1989). People who are empowered are more likelyto exhibit creative behavior (Jung et al., 2003; Zhou, 1998;).Sheldon (1995) demonstrates that personal autonomy is a corecharacteristic of creative people, and Mumford and Gustafson(1988) suggest that innovative achievement might increasewhen organizations support autonomy.

Transformational leadership may increase the psychologicalempowerment of followers. The transformational leader, byindividualized consideration, builds follower self-confidenceand heightens personal development, which, in turn, leads to theempowerment of followers (Conger, 1999). Transformationalleaders also enhance followers' empowerment by providingmeaning and challenge to their work (Avolio et al., 2004); anumber of empirical studies confirm this effect (Jung and Sosik,

464 L. Gumusluoglu, A. Ilsev / Journal of Business Research 62 (2009) 461–473

2002; Jung et al., 2003). Dvir et al. (2002), in a longitudinalfield experiment with Israeli soldiers, also report a positiveimpact of transformational leadership on followers' empower-ment. Consequently, employees under transformational leader-ship feel empowered and they are likely to seek creativeapproaches in their jobs. Therefore, H3: Followers' psycholo-gical empowerment mediates the relationship between trans-formational leadership and followers' creativity.

1.4. Transformational leadership and perception of support forinnovation

The characteristics of their organization largely affectfollowers' creativity (Siegel and Kaemmerer, 1978; Scott andBruce, 1994; Amabile et al., 1996). According to Scott and Bruce(1994), organizational climate is an important factor for creativity;employees' perceptions of the extent to which creativity isencouraged at the workplace, and the extent to which organiza-tional resources are allocated to supporting creativity influencecreative performance. An employee's perception of an innovativeclimate encourages risk taking, and the challenge to use creativeapproaches at work. The present study includes empiricallyvalidating this proposition in an R&D center, where followers'perceptions of an innovative climate enhance their creativity.

Leadership can affect creative behavior through its influenceon the followers' perceptions of a climate supportive of in-novation. The leader can establish a work environment encoura-ging creativity (Amabile et al., 1996; Amabile et al., 2004), andcreate an organizational climate that serves as a guiding principlefor more creative work processes (Scott and Bruce, 1994).Transformational leaders, by intellectually stimulating theirfollowers, championing innovation, and articulating a compellingvision throughout their organizations, help establish an organiza-tional climate where employees feel challenged and energized toseek innovative approaches in their jobs. Koene et al. (2002) findthat charismatic leadership and consideration have substantialeffects on organizational climate. Similarly, Jung et al. (2003)report a significant positive relationship between transformationalleadership and innovative organizational climate. Building fromthese observations, the study proposes the following hypothesis.H4: Followers' perception of support for innovation mediates therelationship between transformational leadership and followers'creativity.

1.5. Transformational leadership and organizational innovation

Organizational innovation is the creation of valuable anduseful new products/services within an organizational context(Woodman et al., 1993). Since most organizations engage ininnovative activity as a competitive weapon, the present studyadopts a market-oriented approach and expands this definitionto include the returns due to innovation. Accordingly,organizational innovation is the tendency of the organizationto develop new or improved products/services and its success inbringing those products/services to the market. This approach isconsistent with Damanpour's (1991: 561) definition of productinnovations as, “new products/services introduced to meet an

external user or market need,” and the description provided bythe OECD (2004: 64) as, “the successful bringing of the newproduct or service to the market.”

Transformational leaders enhance innovation within theorganization; the tendency of organizations to innovate. Leaders'use of inspirational motivation and intellectual stimulation iscritical for organizational innovation (Elkins and Keller, 2003).Transformational leaders promote creative ideas within theirorganizations; this behavior reflects the “championing role” oftransformational leaders (Howell and Higgins, 1990). Theseleaders have a vision that motivates their followers, increases theirwillingness to perform beyond expectations, and challenges themto adopt innovative approaches in their work. The resultingheightened level of motivation is likely to enhance organizationalinnovation (Mumford et al., 2002). A number of empirical studiessupport such leaders' positive impact on innovation (e.g., Keller,1992; Waldman and Atwater, 1994). These studies examine therelationship between transformational leadership and innovationmostly in R&D units and at the project level. The proposal of aneffect of transformational leadership on innovation at theorganizational level has become a topic of empirical researchonly recently. Jung et al. (2003), in a study of 32 Taiwanesecompanies, find that transformational leadership significantly andpositively relates to organizational innovation as measured byR&D expenditures and number of patents obtained over thepreceding 3 years.

Transformational leaders may also have a positive influenceon the market success of the innovations. Leaders who articulatea strong vision of innovation and display a sense of power andconfidence will strive to ensure the market success of theinnovation. These leaders mobilize their followers to ensure theinnovations' success (Jung et al., 2003). Keller (1992) suggeststhat leading professional employees might require more thantraditional leader behaviors especially in R&D settings wherequality rather than quantity is the primary performance criterion.Furthermore, in addition to the internal roles, the transforma-tional leader may be effective in playing external roles such asboundary spanning and entrepreneuring/championing (Howelland Higgins, 1990); these might be important both for under-standing the needs of the market and for successful marketing ofthe innovation. Therefore, this study proposes a positiverelationship between transformational leadership and organiza-tional innovation which is conceptualized in this paper asincluding both the tendency of the organization to innovate andthe success of innovations. H5: Transformational leadershiprelates positively to organizational innovation.

1.6. Individual creativity and organizational innovation

The individual is the ultimate source of any new idea (Redmondet al., 1993) and provides the foundation for organizationalinnovation (Shalley and Gilson, 2004). Hence, theoretically, thecreative performance of employees provides the raw materialneeded for organizational innovation (Oldham and Cummings,1996). Creative employees are those who tend to identifyopportunities for new products. They may find new uses forexisting methods or equipments, or generate novel but operable

465L. Gumusluoglu, A. Ilsev / Journal of Business Research 62 (2009) 461–473

work-related ideas. These people not only are more likely to comeup with creative solutions to problems and champion ideas toothers, but also develop adequate plans for the implementation ofnew ideas. As Shalley and Gilson (2004) suggest, creativeemployees produce novel and useful ideas about organizationalproducts, practices, or procedures. Besides, these people mightcreate a spillover effect by serving as role models to the rest of theorganization. Shalley et al. (2004) state that creative employees'new ideas are transferable to other employees in the organizationfor their own use and development. Consequently, such creativityat the individual level, through idea generation and implementa-tion, is likely to lead to the development of innovative products atthe organizational level. Creativity of employees positivelyinfluences organizational innovation. H6: Individual creativityrelates positively to organizational innovation.

2. Method

2.1. Sample

A total of 163 employees and their leaders in 43 Turkishentrepreneurial software development companies participated inthis research. Of the 90 micro- and small-sized informationtechnology companies most of which were located in technoparks, 49 satisfied the two criteria of this study:minimum firm ageof 3 years and in-house software development. The leaders of43 entrepreneurial companies agreed to participate in the study.They were both the owner–managers and immediate supervisorsof the R&D personnel. The leaders provided the names of theR&D employees engaged in the problem definition and designstages of software development. Of 168 employees identified asexplained above, five did not fill out the questionnaire.

The sample is a highly homogeneous one in terms of size offirms and type of task performed. All companies are small with3 to 17 employees and all are engaged in the development ofnew products and the improvement of existing productsdescribed as development work by Keller (1992). The firstreason for selecting such a sample is that it is adequate toinvestigate both individual-level creativity and organizationalinnovation. Although the development tasks these companiesare engaged in do require creativity (Couger et al., 1993),empirical researchers have neglected the topic of creativity inthis industry. Since development work produces incrementalinnovations (Elkins and Keller, 2003), and software develop-ment has an increasing share in industrial innovations (OECD,1996), the sample is adequate for measuring organizationalinnovation as well. Second, small entrepreneurial companieswhen compared with large ones may be more innovative due totheir “greater flexibility”, and may have “younger and moregrowth-oriented personnel” (Ettlie, 1983: 29). Moreover,entrepreneurship orientation has been suggested (Kitchell,1995) and empirically found (Salavou and Lioukas, 2003) tobe a driver of innovation.

The sample consists of 130 men (80%) and 33 women(20%). The average age of the followers is 27.6 years. 4.3%have high-school diplomas, 71% have bachelor's degrees, 22%have master's degrees and 3% have PhD's. The employees have

2.25 years of average company tenure and 4.71 years of averagejob tenure in the sector. All participants are Turkish. Theaverage life of the companies is 5.9 years and the average size is9.4 employees.

2.2. Procedure

The fieldwork included interviewing six company owners inthe software development industry three times. The aim of thefirst interview was to understand the specific nature of thedevelopment work the companies were engaged in. In thesecond interview a month later, all participants were providedwith items to measure employees' creativity and were asked toidentify the ones most relevant to their employees' work. Then,the definition of innovation and the specific descriptions of atechnologically new product and improved product adopted inthis study were explained. They unanimously agreed that thestatements reflected the development work they were engagedin. Finally, participants were provided with the measures oforganizational innovation commonly used in empirical research(such as number of patents and R&D intensity) and were askedto recommend measures for their industry. The authors tookthese comments and recommendations into consideration whendeveloping the measure of organizational innovation which wasthen presented to the leaders in the third interview. Theparticipants agreed with the measure without exception.

Data were collected by two separate questionnaires: one forthe employees and the other for their leaders. The ques-tionnaires included company and employee identification codesso that data collected from the leaders and employees could bematched and grouped for analysis. All respondents wereguaranteed confidentiality. The questionnaires were given inenvelopes and employees were told to seal their completedforms. They were collected immediately after completion. Allof the questionnaires were completed during regular workinghours and the authors were present to answer questions andcollect the completed surveys. Since all the participants wereTurkish, the questionnaire items (except the MLQ for which thecopyright had been obtained for the Turkish version) werecarefully translated and back-translated to ensure conceptualequivalence and comparability (Brislin, 1986).

Employees' questionnaires included measures of transfor-mational leadership, perception of support for innovation,psychological empowerment, and intrinsic motivation. Onaverage, 4 employees rated each leader. Employees were alsoasked their age, gender, educational level, job tenure, andcompany tenure.

Leaders' questionnaires were administered in two separatevisits, within a one-month interval. In the first visit, leaders wereasked to provide data on company innovations. They were alsoasked for the age of their firms. In the second visit, theyevaluated their employees' creativity. The reason for conductingthe leaders' questionnaire at two separate times was to preventany bias or inflated results that might have arisen if the leadershad answered the questions about organizational innovation andcreativity of their subordinates at the same time. The averagenumber of employees evaluated by each leader was 4.

466 L. Gumusluoglu, A. Ilsev / Journal of Business Research 62 (2009) 461–473

2.3. Individual-level measures

2.3.1. Transformational leadershipThis study measured transformational leadership using

twenty items from the Turkish version of the Multi-FactorLeadership Questionnaire (MLQ-Form 5X) (Bass and Avolio,1995). Avolio et al. (1999) provide support for the convergentand discriminant validity of the instrument. If subordinatesprovided both the transformational leadership ratings and thecriterion ratings, the results could have been potentially biasedby same-source (MLQ) data. Therefore, only the transforma-tional leadership items were used from the questionnaire.Participants judged how frequently their immediate leaderengaged in transformational leadership behaviors. Ratingswere completed on a 5-point scale with 1 representing “Not atall” and 5 representing “Frequently, if not always”. Sample itemsincluded: “Articulates a compelling vision of the future,” “Treatsme as an individual rather than as a member of the group,” and“Gets me to look at problems from many different angles.”

Exploratory factor analysis using the principal componentsmethod and varimax rotation was conducted on the twentyitems in order to determine their factor structure. After twoitems with factor loadings less than 0.50 were removed, theresulting eighteen items loaded on one factor, which accountedfor 47% of the variance. These items were averaged to form ascale. Reliability (i.e., Cronbach's alpha) of the scale was 0.93.Bycio et al. (1995) show that the dimensions of transformationalleadership fail to exhibit discriminant validity in predictingoutcomes. Furthermore, since this study did not have any apriori expectation that individual dimensions of transforma-tional leadership would differentially affect creativity andinnovation, a single index was used to measure transformationalleadership. Prior research (Judge and Bono, 2000) validates theuse of a single scale to represent transformational leadership.

2.3.2. Intrinsic motivationIntrinsic motivation was measured by five items adapted

from Tierney et al. (1999). On a five point scale ranging from 1(“Corresponds not at all”) to 5 (“Corresponds exactly”),employees indicated the extent to which each of the statementsapplied to them in terms of their current tasks. Sample itemswere “I enjoy coming up with new ideas for products” and “Ienjoy improving existing processes or products.” These fiveitems loaded on one factor and explained 55.24% of thevariance. They were averaged to form a scale with a reliabilityof 0.77.

2.3.3. Psychological empowermentPsychological empowerment was measured by the 12-item

scale developed by Spreitzer (1995). All items were rated usinga 5-point scale ranging from 1 (“Very strongly disagree”) to 5(“Very strongly agree”). Sample items were “I have significantinfluence on what happens in my department” and “I havesignificant autonomy in determining how I do my job.”Exploratory factor analysis revealed that six items had factorloadings less than 0.50. After they were removed, the resultingsix items loaded on one factor, which accounted for 52.59% of

the variance. These items were averaged to form a scale, whichhad a reliability of 0.82.

2.3.4. Perception of support for innovationThis variable was measured by 12 items adapted from Scott

and Bruce (1994). On a 5-point scale ranging from 1 (“Stronglydisagree”) to 5 (“Strongly agree”), employees indicated theextent to which their companies supported creativity. Sampleitems were “This organization can be described as flexible andcontinually adapting to change” and “There are adequateresources devoted to innovation in this organization.” Based onthe factor analysis results, three items with loadings less than0.50 were removed. The remaining 9 items loaded on one factorthat accounted for 55.40% of the variance. These items wereaveraged to form a scale with a reliability of 0.88.

2.3.5. CreativityFollowers' creativity is the dependent variable of the first

part of the study. The subject of investigation in the presentresearch is the creativity of the employees who are working inR&D departments and are expected to turn creative ideas intoinnovative products; thus, both idea generation and implemen-tation by these employees should be considered in measuringcreativity (Mumford et al., 2002). We adapted 13 items thatcapture these two concepts from Tierney et al. (1999) and Zhouand George's (2001) creativity measures.

Leaders evaluated the creativity of their employees onemonth after the employees rated their leadership behavior. On afive point scale ranging from 1 (“Not at all characteristic”) to 5(“Very characteristic”), leaders were asked to report how ofteneach of their employees could be described according to theitems. Sample items were “Promotes and champions ideas toothers” and “Serves as a good role model for creativity.” All ofthe items loaded on one factor, which accounted for 62.99% ofthe variance. The items were averaged to form a scale with areliability of 0.95.

2.3.6. Control variablesFollowers' educational level and job tenure are the control

variables of this study since they are related to creativity.Creativity is the outcome of an individual's accumulatedcreative thinking skills and expertise based on formal educationand past experience (Amabile, 1998). Furthermore, experienceprovides a level of familiarity which might be needed forcreative performance (Shalley and Gilson, 2004). Therefore, jobtenure was used as an indicator of experience.

2.4. Organizational-level measures

2.4.1. Transformational leadership and creativity at theorganizational level

Consistent with Shamir et al. (1998), this study treatstransformational leadership as an organizational-level variable;in other words, as leadership behaviors exhibited to the organi-zation, a micro- or a small-sized company here, as a whole.Therefore, transformational leadership ratings by the subordinateswere aggregated to the organizational level by averaging

467L. Gumusluoglu, A. Ilsev / Journal of Business Research 62 (2009) 461–473

their values for each organization. In addition, in order to test thehypotheses regarding organizational innovation, creativity ratingsof the subordinates by their leaders were aggregated to the orga-nizational level by averaging their values for each organization.

One-way ANOVA provided empirical justification foraggregating subordinate ratings of transformational leadership.The results showed that between-group differences weresignificantly higher than within-group differences (F=3.06,pb0.001). Intra-class correlation ICC1 was 0.52 and ICC2 was0.67. The inter-rater reliability (rwg(j)) (James et al., 1984) forsubordinates' rating the same leader was also examined. Themean rwg(j) value for the transformational leadership ratings was0.96 and the median was 0.97. These results showed thataggregation was appropriate for this variable.

2.4.2. Organizational innovationThis study defines organizational innovation as the tendency

of the organization to develop new or improved products/servicesand its success in bringing those products/services to the market.Consistent with this definition and taking into consideration thecomments the leaders made during the interview, a new criterionfor measuring organizational innovation was developed. Theleaders' common concern was that quantifiable measures such ascopyrights or quality certificates commonly employed to studyestablished companies in developed industries and countriesmight not be applicable either to the growing software develop-ment industry or to the nature of competition among small-sizedentrepreneurial companies in Turkey, because the rules ofcompetition and the legal structure are poorly established.Therefore, a market-oriented approach rather than such quantifi-able input measures was adopted for developing the measurementof organizational innovation.

The measure of organizational innovation in this study is theproduct of two ratios, namely, the coefficient of innovativenesstendency and the success of product innovations. Coefficient ofinnovativeness tendency is the ratio of sales generated byproduct innovations to total sales. This coefficient quantifies theinnovativeness orientation of companies engaged in other workapart from software development such as marketing computerhardware. This measure of innovative activity was also used byCzarnitzki and Kraft (2004), who investigated the innovativeperformance of European firms. In order to operationalize thedefinition of organizational innovation in this study, thismeasure was employed as a coefficient to modify the successof product innovations.

Success of product innovations is the ratio of sales generatedby product innovations to the expenditures in producing thoseproduct innovations. This ratio shows the success of theorganization in both satisfying market needs and utilizing theorganization's resources in producing the innovations. This is abetter measure of outcomes than the R&D expendituresmeasured in absolute numbers. As stated by Jung et al. (2003:540), expenditures for innovation itself do not reflect thesuccess of the company in generating “outcomes,” but rather its“willingness” to support innovation.

New products developed and existing products improved(Keller and Holland, 1983; Woodman et al., 1993) as well as

custom-made projects (OECD, 1996) by the companies areregarded as product innovations in this research. The ques-tionnaire administered to the leaders included the definition ofinnovation (Keller and Holland, 1983) and descriptions by theOECD (1996) of new and improved products along withexamples of innovation in the software development industry(provided in the Appendix). The leaders analyzed every productand custom-made project of their company to determinewhether it would be considered an innovation according tothe guidelines. They answered three questions: total salesgenerated by product innovations during the previous threeyears, total sales of the company during the previous threeyears, and total expenditures in producing those productinnovations during the same time period. The output questionscovered the last three years to take into account the newlyemerging nature of this market in Turkey where softwaredevelopment and sales might take longer.

2.4.3. Control variableFirm age is the control variable in this part of the study, since

prior studies report its positive relationship with organizationinnovation (Hitt et al., 1997; Jung et al., 2003).

3. Results

3.1. Individual-level analysis

3.1.1. Descriptive statisticsTable 1 includes means, standard deviations, alpha coeffi-

cients, and correlations among all individual-level variables.Intercorrelations show that creativity significantly and posi-tively correlates with transformational leadership (r=0.17,pb0.05), intrinsic motivation (r=0.24, pb0.01), and psycho-logical empowerment (r=0.24, pb0.01), but not with percep-tion of support for innovation (r=0.10, n.s.). Transformationalleadership has significant positive correlations with intrinsicmotivation (r=0.31, pb0.001), psychological empowerment(r=0.27, pb0.001), and perception of support for innovation(p=0.71, pb0.001).

3.1.2. Tests of individual-level hypothesesHypotheses 1 through 4 relate to the direct effect of trans-

formational leadership on employee creativity and the mediatorsof this relationship. These hypotheses are tested using Hierarch-ical Linear Modeling (HLM) because the data of this study arenested within organizations, and the model includes cross-levelrelationships between transformational leadership (organiza-tional-level), mediators (individual-level), and employee creativ-ity (individual-level). HLM accounts for dependence among thescores for individuals within the same group and accommodatesvariables at multiple levels (Bryck and Raudenbush, 1992).Building from suggestion by Hofmann and Gavin (1998) allHLM analyses use grand-mean centering.

In order to test the direct and mediated effects, the studyuses the multilevel mediational modeling method (Krulland MacKinnon, 2001). This method incorporates Baronand Kenny's (1986) mediational analysis procedure into

Table 2Results of the multilevel mediational analysis

Creativity Intrinsicmotivation

Psychologicalempowerment

Perception ofsupport forinnovation

Education 0.19 (0.10)Job tenure 0.02 (0.01)Transformationalleadership

0.25 ⁎ (0.10)

Education 0.09 (0.06)Job tenure 0.01 (0.01)Transformationalleadership

0.22 ⁎ (0.10)

Education 0.16 (0.10)Job tenure 0.02 (0.01)Transformationalleadership

0.19 (0.10)

Intrinsicmotivation

0.28 ⁎ (0.11)

Education 0.25 ⁎ (0.12)Job tenure 0.04 ⁎⁎⁎ (0.01)Transformationalleadership

0.31 ⁎⁎ (0.09)

Education 0.19 (0.12)Job tenure 0.00 (0.01)Transformationalleadership

0.16 (0.08)

Psychologicalempowerment

0.29 ⁎⁎ (0.11)

Education −0.10 (0.08)Job tenure 0.00 (0.01)Transformationalleadership

0.95 ⁎⁎⁎ (0.09)

Education 0.20 (0.11)Job tenure 0.02 (0.01)Transformationalleadership

0.21 (0.12)

Perception ofsupport forinnovation

0.03 (0.08)

Results are Hierarchical Linear Modeling-derived parameters.Values in parentheses are the standard errors of the coefficients.⁎ pb0.05.⁎⁎ pb0.01.

⁎⁎⁎ pb0.001.

Table 1Descriptive statistics, alpha coefficients, and correlations: individual-level scales

Variable Mean S.D. 1 2 3 4 5 6

1. Transformational leadership 3.9 0.68 (0.93)2. Intrinsic motivation 4.4 0.65 0.31 ⁎⁎⁎ (0.77)3. Psychological empowerment 3.4 0.73 0.27 ⁎⁎⁎ 0.31 ⁎⁎⁎ (0.82)4. Perception of support for innovation 3.7 0.80 0.71 ⁎⁎⁎ 0.35 ⁎⁎⁎ 0.31 ⁎⁎⁎ (0.88)5. Creativity 3.7 0.78 0.17 ⁎ 0.24 ⁎⁎ 0.24 ⁎⁎ 0.10 (0.95)6. Education – – −0.10 0.06 0.19 ⁎ −0.10 0.12 –7. Job tenure 4.7 4.18 −0.12 0.03 0.19 ⁎ −0.05 0.06 −0.04

n=163.Alpha coefficients are on the diagonal, in parentheses.⁎ pb0.05.⁎⁎ pb0.01.

⁎⁎⁎ pb0.001.

468 L. Gumusluoglu, A. Ilsev / Journal of Business Research 62 (2009) 461–473

hierarchical linear models. The method also includes the test ofthe direct effects in addition to the mediated effects. Accordingto the method, three conditions are necessary to establishmediation. First, the independent variable (transformationalleadership) should significantly relate to the dependentvariable (creativity). Second, the independent variable shouldsignificantly relate to the mediator. Third, when the dependentvariable is regressed on both the independent variable and themediator, the mediator should significantly relate to thedependent variable and the independent variable should notsignificantly relate to the dependent variable. Full mediationoccurs when the direct effect of the independent variable in thislast condition is reduced to zero, otherwise the mediating effectis partial. To determine whether the mediated effect issignificant, Sobel test for multilevel mediational modelingmethod (Krull and MacKinnon, 2001) is used. Educationallevel and job tenure of the employees are controlled for in allthe hierarchical models. Table 2 summarizes the results of thisanalysis.

Hypothesis 1 states that there is a positive relationshipbetween transformational leadership and individual creativity.As the table shows, there is a significant positive relationshipbetween transformational leadership and creativity (γ01=0.25,pb0.05), after controlling for education and job tenure. There-fore, the findings support Hypothesis 1. This significant rela-tionship also satisfies the first condition of the mediation tests forall three mediators.

Hypothesis 2 suggests that intrinsic motivation mediates therelationship between transformational leadership and indivi-dual creativity. According to the results in Table 2, transforma-tional leadership has a significant association with intrinsicmotivation (γ01=0.22, pb0.05). In addition, intrinsic motiva-tion significantly relates to creativity (γ10=0.28, pb0.05)when entered together with transformational leadership into theequation predicting creativity, where transformational leader-ship has no significant effect (γ01=0.19, n.s.). These resultssuggest a partial mediating effect of intrinsic motivation.However, the result of the Sobel test indicates that intrinsicmotivation does not significantly reduce the effect oftransformational leadership on creativity (t=1.66, n.s.). Sinceintrinsic motivation does not significantly mediate the relation-ship between transformational leadership and creativity, thefindings do not support Hypothesis 2.

Hypothesis 3 predicts a mediating effect of psychologicalempowerment for the relationship between transformationalleadership and individual creativity. The results show that

Table 4Results of the regression analysis for organizational innovation

Step 1 Step 2

Firm age 0.04 0.05 ⁎

Transformational leadership 0.40 ⁎

Creativity −0.17F 3.68 3.52 ⁎

Df 1 3R2 0.08 0.21ΔR2 0.13 ⁎

⁎ pb0.05.

469L. Gumusluoglu, A. Ilsev / Journal of Business Research 62 (2009) 461–473

the conditions required for a mediator effect are satisfied.Transformational leadership significantly relates to psychologicalempowerment (γ01=0.31, pb0.01). Psychological empowermentsignificantly associates with creativity (γ10=0.29, pb0.01)when entered together with transformational leadership intothe equation predicting creativity, where transformational leader-ship has no significant effect (γ01=0.16, n.s.). The results of theSobel test also indicate a significant mediated relationship(t=2.09, pb0.05). These suggest that psychological empower-ment partially mediates the relationship between transforma-tional leadership and creativity. Therefore, the findings supportHypothesis 3.

Hypothesis 4 states that perceived support for innovationmediates the relationship between transformational leadershipand individual creativity. As shown in Table 2, there is asignificant association between transformational leadership andperception of support for innovation (γ01=0.95, pb0.001).However, perception of support for innovation does notsignificantly relate to creativity (γ10=0.03, n.s.) when enteredtogether with transformational leadership into the equationpredicting creativity. Since the third condition is not satisfied,perception of support for innovation does not mediate therelationship between transformational leadership and creativity.Therefore, Hypothesis 4 is not supported.

3.2. Organizational-level analysis

3.2.1. Descriptive statisticsTable 3 presents means, standard deviations, and correlations

among organizational-level variables. Organizational innova-tion has a significant correlation with transformational leader-ship (r=0.30, pb0.05), but not with creativity.

3.2.2. Tests of organizational-level hypothesesHypotheses 5 and 6 relate to the direct effects of

transformational leadership and creativity on organizationalinnovation. These hypotheses are tested by regression analysis.The control variable (firm age) is entered first as a predictor ofinnovation. Then, the main effects predictor variables (trans-formational leadership and creativity) are entered into theregression equation. Table 4 presents the results of this analysis.

Hypothesis 5 predicts a positive relationship between trans-formational leadership and organizational innovation. Resultsof the analysis reveal that, after controlling for firm age, trans-formational leadership has a significant positive effect on orga-nizational innovation (b=0.40, pb0.05). Therefore, Hypothesis 5is supported.

Table 3Descriptive statistics, and correlations: organizational-level variables

Variables Mean S.D. 1 2 3 4

1. Firm age 5.9 3.732. Transformational leadership 3.9 0.53 −0.113. Creativity 3.7 0.50 −0.13 0.294. Organizational innovation 1.6 0.56 0.29 0.30 ⁎ −0.08

⁎ pb0.05.

Hypothesis 6 states that creativity positively relates toorganizational innovation. Since creativity does not have asignificant relationshipwith organizational innovation (b=−0.17,n.s.), Hypothesis 6 is not supported.

4. Discussion

This paper has both theoretical and methodological con-tributions to the literature. This study is the first to investigatethe effects of transformational leadership on creativity-relatedoutcomes at multiple levels within organizations. The findingssuggest that transformational leadership has important effects atboth individual and organizational levels. At the individuallevel, transformational leadership positively relates to fol-lowers' creativity. This finding is valuable for two reasons.First, previous findings were inconsistent and further research inreal settings was needed to support the positive proposition infavor of this leadership (Mumford and Licuanan, 2004). In linewith the findings of Shin and Zhou (2003), this research,conducted in real-work settings, finds a positive relationshipbetween transformational leadership and followers' individualcreativity. Second, this positive relationship exists in collectivistTurkey (Hofstede, 1980), supporting the arguments by Bass(1990a) that transformational leadership is more likely toemerge in collectivist cultures than in the individualist culturesof the West and that collectivists perform better under transfor-mational leadership. A number of studies report a strongerpositive effect of transformational leadership on the creativeperformance of collectivists as compared to individualists (e.g.,Jung and Avolio, 1999; Jung and Yammarino, 2001).

Analysis of the mediators reveals partial mediating effects forintrinsic motivation and psychological empowerment based onBaron and Kenny's (1986) criteria. Yet, the test of mediationshows that intrinsic motivation is not a significant mediator of thetransformational leadership–creativity relationship. This findingseems to contradict Shin and Zhou's (2003) study that shows apartial mediating effect of intrinsic motivation. However, theirresearch does not involve a formal test of the significance of thismediated effect, making it difficult to comment on these incon-sistent results.

The mediating effect of psychological empowerment, on theother hand, is significant. This finding is an important contributionto the literature in that it shows psychological empowerment as acrucial psychological mechanism through which transformationalleadership influences employees' creativity. A reason for

470 L. Gumusluoglu, A. Ilsev / Journal of Business Research 62 (2009) 461–473

psychological empowerment to be a stronger mediator thanintrinsic motivation might be that R&D employees are alreadyintrinsically motivated, which may act as a substitute for theinfluence of a transformational leader on their creative perfor-mance. This leader's effect through enabling them to make theirown decisions and take initiatives might be a more powerfulcreativity-enhancing force for these employees than his or hereffect through influencing their intrinsic motivation.

Contrary to the expectation of this study, the mediating role ofperception of support for innovation is not significant. Thisfinding might have resulted from the high correlation betweentransformational leadership and support for innovation (r=0.71).In addition, the transformational leader's direct behavior onemployees, such as individualized consideration and intellectualstimulation, might affect their emotional well-being and providedirect and clear cues that creative behavior is expected; whereas,employees might not take an innovation supporting climate, anorganization-wide contextual factor, as personally. These argu-ments might apply more to a high power-distance culture, wherethe workers put more value on their leader's building one-to-onerelationships with them. The increased enthusiasm might makethem seek more innovative approaches in their work. PerhapsTurkish people, who rank high on power-distance (Hofstede,1980), respond readily to the transformational leader whoempower them, but they may not see a climate supportinginnovation as important to them personally.

At the organizational level of analysis, in line with the findingsof Jung et al. (2003), this study reports that transformationalleadership has a significant positive association with organiza-tional innovation. Moreover, as stated before, previous researchfocused on this leader's effect on the tendency of organizations toinnovate. The definition of organizational innovation in this studyincludes the success of innovations as well as the tendency toinnovate. The findings suggest that transformational leadersmight not only promote innovative activity within the organiza-tion but also ensure the market success of the innovations.Furthermore, since the innovations under investigation here arerelated to development work, the positive influence of this form ofleadership is identified on incremental innovation. This findingsomewhat contradicts Keller's (1992) suggestion that develop-mental projects which use existing knowledge to produceincremental innovations might need more of a transactionalleader to allocate and coordinate tasks, while research projectswhich need originality and importation of technical information inorder to produce radical innovation might be better led bytransformational leaders. Transactional leadership is not underinvestigation here, but this study suggests that as the transforma-tional character of the leader increases, innovation in develop-mental work increases. This contrary result might have stemmedfrom the collectivist character of the Turkish participants whowould expect their leaders to exhibit transformational leaderbehaviors (Bass, 1995) and would readily respond to transforma-tional leadership.

The proposed relationship between individual-level creativ-ity and organizational innovation is not significant. Severalreasons might explain this finding. First, as Mumford andGustafson (1988) suggest, employee creativity may be

necessary but not sufficient for organizational innovationgiven that creative ideas or solutions might not be considereduseful or might not be successfully implemented. In this case,they will not be converted into actual innovations in theorganization. Furthermore, Perry-Smith and Shalley (2003)argue that novel information is less likely to be communicatedthrough stronger ties (good friends or close relationships) thanweaker ties (more distant relationships or distant colleagues).Given that R&D groups under investigation in this study arecomposed of a handful of employees stronger ties might havebeen in effect leading to less communication of novelinformation. Moreover, creative output of a collective may bea function of not only the creativity of individuals but also groupprocesses such as group cohesion (Woodman et al., 1993),effective communication by group members (Taggar, 2002),and team integration skills such as conflict resolution andcollaborative problem-solving skills (Janssen et al., 2004). Thelack of these factors or even a low level of them might hinderthe effects of individual creativity (Taggar, 2002). This mighthave been the case in the present study. Finally, themethodology employed here might have been a reason for thefailure to find this relationship. In this study, there is a mismatchbetween the 3-year period for which innovation data weremeasured and the company tenures of the participants, whichaverage 2.25 years. Therefore, participant employees might nothave contributed to the innovative projects of the last 3 years.

The methodological contributions of this study are twofold.First, this study investigates transformational leadership,creativity, and innovation in Turkey, a developing country; itshows the external validity of these theories which weredeveloped and tested in Western developed countries. Second,the market-oriented measure developed and used as a proxy fororganizational innovation in this study qualifies as a methodo-logical contribution. It can be used as a measure of innovation innewly developing industries and in entrepreneurial companies,especially in underdeveloped or developing countries wherequantifiable measures such as patents or copyrights are notrelevant. Furthermore, this measure differs from other measuresof organizational innovation in that it reflects not only the firms'propensity to innovate but also the returns on innovations, animportant indicator of competitive advantage.

This study is not without its limitations. Employees' creativitywas evaluated only by their leaders and this might have led toartificially inflated ratings. Another limitation is the cross-sectional design employed; which makes it difficult to infercausality between the variables in such studies. The significantrelationships reported in this study are associative and correla-tional, and may not be causal. For example, the positiverelationship between transformational leadership and creativitymight have been a spurious one due to some contextual factorsthat influence these variables. Thus, longitudinal studies in real-work settings can better analyze the significant relationshipsfound here. In addition, the sample of this study might be anotherlimitation. First, the sample is primarily comprised of males.Second, it includes small-sized entrepreneurial software devel-opment companies operating in Turkey. The findingsmight not begeneralizable to other software development companies or to

471L. Gumusluoglu, A. Ilsev / Journal of Business Research 62 (2009) 461–473

other types of organizations in different industries and/orcountries. Finally, this research does not include group character-istics or processes such as group composition, cohesiveness, andcommunication while investigating individual creativity–organi-zational innovation relationships, which might have preventedcapturing the complexity of this relationship.

4.1. Directions for future research

This study focuses on the mediating processes underlying therelationship between transformational leadership and followers'creativity. Future research might examine the processes thatmediate the relationship between this leadership and organiza-tional innovation. In addition, studies should investigate whetherpsychological empowerment is a significant mediator of thetransformational leadership–employee creativity relationship indifferent countries or industries. Moreover, this study does notsupport the effect of individual creativity on organizationalinnovation. Future research should examine whether themediating and moderating influences of group processes suchas cohesiveness, diversity, and conflict are the determinants oforganizational innovation rather than employees' creativity.

The measure of organizational innovation that this studydevelops and uses might be useful for studies in industries otherthan software development, or in industries which produceradical innovation. Studies in different countries can also usethis measure in order to evaluate its external validity.

4.2. Implications for managerial practice

This research is the first to investigate transformationalleadership and its effects on creativity and organizationalinnovation in Turkey. Equally important, it is conducted inentrepreneurial companies in the software development indus-try. This sector is particularly important for Turkey, because ofits low standing in the world development average (DPT, 2001).All stakeholders, especially managers, should encourage thedevelopment and competitiveness of this industry.

The findings of this study should encourage managers tostimulate their followers by empowering them. They shouldunderstand that this mechanism significantly enhances theiremployees' creative performance. The findings should alsoencourage them to engage in transformational leadershipbehaviors in order to boost the creative performance of theiremployees and to bring about organizational innovation.Findings of this study also provide evidence that transforma-tional leadership should be the subject of management trainingand development in Turkey to improve the innovationperformance of the country.

Appendix A. Descriptions and examples of innovationprovided to the leaders

Innovation. Innovation is an important product, process, or idealeading to a new or improved product that is new to the organi-zation. According to this definition, new products developed,existing products improved, and custom-made projects which

display significantly different attributes from the firms' previousproducts are considered as product innovations in this study.

Technological product innovation

The term “product” covers both goods and services.Technological product innovation can take two broad forms:A technologically new product is a product whose technologicalcharacteristics or intended uses differ significantly from those ofpreviously produced products. Such innovations can involveradically new technologies, can be based on combining existingtechnologies in new uses, or can be derived from the use of newknowledge. A technologically improved product is an existingproduct whose performance has been significantly enhanced orupgraded. A simple product may be improved (in terms of betterperformance or lower cost) through the use of higher-performance components or materials; or a complex productwhich consists of a number of integrated sub-systems may beimproved by partial changes to one of the sub-systems.

Examples of technological innovations in software develop-ment companies. The introduction of new multimedia softwareapplications that can be used for educational purposes, thuseliminating the need for a live human instructor. The develop-ment of a whole range of different customer packages in whichclients are offered varying degrees of assistance/support.

References

Amabile TM. The social psychology of creativity. New York: Springer-Verlag;1983.

Amabile TM. How to kill creativity. Harvard Business Review 1998;76(5):77–87.Amabile TM, Conti R, Coon H, Lazenby J, Herron M. Assessing the work

environment for creativity. Academy of Management Journal 1996;39(5):1154–84.

Amabile TM, Schatzel EA, Moneta GB, Kramer SJ. Leader behaviors and thework environment for creativity: perceived leader support. LeadershipQuarterly 2004;15(1):5–32.

Avolio BJ, Bass BM, Jung DI. Re-examining the components of transforma-tional and transactional leadership using the Multifactor LeadershipQuestionnaire. Journal of Occupational and Organizational Psychology1999;72(4):441–63.

Avolio BJ, Zhu W, Koh W, Bhatia P. Transformational leadership andorganizational commitment: mediating role of psychological empowermentand moderating role of structural distance. Journal of OrganizationalBehavior 2004;25:951–68.

Baron RM, Kenny DA. The moderator–mediator variable distinction in socialpsychological research: conceptual, strategic, and statistical considerations.Journal of Personality and Social Psychology 1986;53:561–71.

Bass BM. Leadership and performance beyond expectations. New York: TheFree Press; 1985.

Bass BM. Bass and Stogdill's handbook of leadership: theory, research, andmanagerial applications. 3rd ed. New York: The Free Press; 1990a.

Bass BM. From transactional to transformational leadership: learning to sharethe vision. Organizational Dynamics 1990b;18(3):19–32.

Bass BM. Transformational leadership. Journal ofManagement Inquiry 1995;4(3):293–8.

Bass BM, Avolio BJ. MLQ, Multifactor Leadership Questionnaire. 2nd ed.Redwood City, CA: Mind Garden; 1995.

Brislin R. The wording and translation of research instruments. In: Lonner WJ,Berry JW, editors. Field methods in cross-cultural research. Beverly Hills,CA: Sage; 1986. p. 137–64.

Bryck AS, Raudenbush SW. Hierarchical linear models: application and dataanalysis methods. London: Sage; 1992.

472 L. Gumusluoglu, A. Ilsev / Journal of Business Research 62 (2009) 461–473

Burns JM. Leadership. New York: Harper & Row; 1978.Bycio P, Hackett RD, Allen JS. Further assessment of Bass's (1985)

conceptualization of transactional and transformational leadership. Journalof Applied Psychology 1995;80:468–78.

Conger JA. Charismatic and transformational leadership in organizations: aninsider's perspective on these developing streams of research. LeadershipQuarterly 1999;10(2):145–79.

Couger JD, Higgins LF, Mclntyre SC. Unstructured creativity in informationsystems organizations. MIS Quarterly 1993;17(4):375–97.

Czarnitzki D, Kraft K. Firm leadership and innovative performance: evidencefrom seven EU countries. Small Business Economics 2004;22:325–32.

Damanpour F. Organizational innovation: a meta-analysis of effects ofdeterminants and moderators. Academy of Management Journal 1991;34:555–90.

Deci EL, Ryan RM. Intrinsic motivation and self-determination in humanbehavior. New York: Plenum Press; 1985.

Deci EL, Connell JP, Ryan RM. Self-determination in a work organization.Journal of Applied Psychology 1989;74(4):580–90.

DPT (The State Planning Organization of Turkey). Ad hoc committee report oninformation technologies and policies; 2001. http://www.ekutup.dpt.gov.tr/bilisim/oik576.pdf.

Drucker PF. Innovation and entrepreneurship: practice and principles.New York: Harper and Row, Publishers; 1985.

Dvir T, Eden D, Avolio BJ, Shamir B. Impact of transformational leadership onfollower development and performance: a field experiment. Academy ofManagement Journal 2002;45(4):735–44.

Elkins T, Keller RT. Leadership in research and development organizations: aliterature review and conceptual framework. Leadership Quarterly 2003;14:587–606.

Ettlie JE. Organizational policy and innovation among suppliers to foodprocessing sector. Academy of Management Journal 1983;26(1):27–44.

Hitt MA, Hoskisson RE, Kim H. International diversification: effects oninnovation and firm performance in product-diversified firms. Academy ofManagement Journal 1997;40(4):767–98.

Hofmann DA, Gavin MB. Centering decisions in hierarchical linear models:implications for research in organizations. Journal of Management 1998;24:623–41.

Hofstede G. Culture's consequences: international differences in work-relatedvalues. Beverly Hills, CA: Sage; 1980.

Howell JM, Higgins CA. Champions of technological innovation. Adminis-trative Science Quarterly 1990;35:317–41.

Howell JM,AvolioBJ. Transformational leadership, transactional leadership, locus ofcontrol, and support for innovation. Journal of Applied Psychology 1993;78(6):891–902.

Hunt JG, Stelluto GE, Hooijberg R. Toward new-wave organizationcreativity: beyond romance and analogy in the relationship betweenorchestra-conductor leadership and musician creativity. LeadershipQuarterly 2004;15:145–62.

James LR, Demaree RG,Wolf G. Estimating within-group inter-rater reliability withand without response bias. Journal of Applied Psychology 1984;69: 85–98.

Janssen O, Van der Vliert E, West M. The bright and dark sides of individual andgroup innovation: a special issue introduction. Journal of OrganizationalBehavior 2004;25:129–45.

Jaussi KS, Dionne SD. Leading for creativity: the role of unconventional leaderbehavior. Leadership Quarterly 2003;14:475–98.

Judge TA, Bono JE. Five-factor model of personality and transformationalleadership. Journal of Applied Psychology 2000;85:751–65.

Jung DI. Transformational and transactional leadership and their effects oncreativity in groups. Creativity Research Journal 2001;13(2):185–95.

Jung DI, Avolio BJ. Effects of leadership style and followers' culturalorientation on performance in group and individual task conditions.Academy of Management Journal 1999;42(2):208–18.

Jung DI, Yammarino FJ. Perceptions of transformational leadership amongAsian Americans and Caucasian Americans: a level of analysis perspective.The Journal of Leadership Studies 2001;8(1):3–21.

Jung DI, Sosik J. Transformational leadership in work groups: the role ofempowerment, cohesiveness and collective efficacy on perceived groupperformance. Small Group Research 2002;33:313–36.

Jung DI, Chow C, Wu A. The role of transformational leadership in enhancingorganizational innovation: hypotheses and some preliminary findings.Leadership Quarterly 2003;14:525–44.

Kahai SS, Sosik JJ, Avolio BJ. Effects of leadership style, anonymity, andrewards on creativity-relevant processes and outcomes in an electronicmeeting system context. Leadership Quarterly 2003;14:499–524.

Keller RT. Transformational leadership and the performance of research anddevelopment project groups. Journal of Management 1992;18(3):489–501.

Keller RT, Holland WE. Communicators and innovators in research anddevelopment organizations. Academy of Management Journal 1983;26(4):742–9.

Kitchell S. Corporate culture, environmental adaptation, and innovationadoption: a qualitative/quantitative approach. Journal of the Academy ofMarketing Science 1995;23(3):195–205.

Koene BAS, Vogelaar ALW, Soeters JL. Leadership effects on organizationalclimate and financial performance: local leadership effect in chainorganizations. Leadership Quarterly 2002;13:193–215.

Krull JL, MacKinnon DP. Multilevel modeling of individual and group-levelmediated effects. Multivariate Behavioral Research 2001;36:249–77.

Lowe KB, Kroeck GK, Sivasubramaniam N. Effectiveness correlates oftransformational and transactional leadership: a meta-analytic review of theMLQ Literature. Leadership Quarterly 1996;7(3):385–425.

Mumford MD, Gustafson SB. Creativity syndrome: integration, application, andinnovation. Psychological Bulletin 1988;103(1):27–43.

Mumford MD, Scott GM, Gaddis B, Strange JM. Leading creative people:orchestrating expertise and relationships. Leadership Quarterly 2002;13(6):705–50.

Mumford MD, Licuanan B. Leading for innovation: conclusions, issues, anddirections. Leadership Quarterly 2004;15(1):163–71.

OECD. The measurement of scientific and technological activities: proposedguidelines and interpreting technological innovation data (Oslo Manual);1996. http://www.oecd.org/dataoecd/35/61/2367580.pdf.

OECD. Small and medium-sized enterprises in Turkey: issues and policies.Paris: OECD; 2004.

OldhamGR, Cummings A. Employee creativity: personal and contextual factorsat work. Academy of Management Journal 1996;39(3):607–34.

Perry-Smith JE, Shalley CE. The social side of creativity: a static and dynamiccausal network perspective. Academy of Management Review 2003;28:89–106.

Redmond MR, Mumford MD, Teach R. Putting creativity to work: effects ofleader behavior on subordinate creativity. Organizational Behavior andHuman Decision Processes 1993;55:120–51.

Salavou H, Lioukas S. Radical product innovations in SMEs: the dominance ofentrepreneurial orientation. Creativity and InnovationManagement 2003;12(2):94–108.

Scott SG, Bruce RA. Determinants of innovative behavior: a path model ofindividual innovation in the workplace. Academy of Management Journal1994;37(3):580–607.

Shalley CE, Gilson LL. What leaders need to know: a review of social andcontextual factors that can foster or hinder creativity. Leadership Quarterly2004;15(1):33–53.

Shalley CE, Zhou J, Oldham GR. The effects of personal and contextualcharacteristics on creativity: where should we go from here? Journal ofManagement 2004;30(6):933–58.

Shamir B, Zakay E, Breinin E, Popper M. Correlates of charismatic leaderbehavior in military units: subordinates' attitudes, unit characteristics, andsuperiors' appraisals of leader performance. Academy of ManagementJournal 1998;41:387–409.

Sheldon KM. Creativity and self-determination in personality. CreativityResearch Journal 1995;8(1):25–36.

Shin SJ, Zhou J. Transformational leadership, conservation and creativity:evidence from Korea. Academy of Management Journal 2003;46(6):703–14.

Siegel S, Kaemmerer W. Measuring the perceived support for innovation inorganizations. Journal of Applied Psychology 1978;63:553–62.

Sosik JJ, Kahai SS, Avolio BJ. Transformational leadership and dimensions ofcreativity: motivating idea generation in computer-mediated groups.Creativity Research Journal 1998;11(2):11–121.

473L. Gumusluoglu, A. Ilsev / Journal of Business Research 62 (2009) 461–473

Sosik JJ, Kahai SS, Avolio BJ. Leadership style, anonymity and creativity ingroup decision support systems. Journal of Creative Behavior 1999;33:227–57.

Spreitzer GM. Psychological empowerment in the workplace: dimensions,measurement, and validation. Academy of Management Journal 1995;38(5):1442–65.

Taggar S. Individual creativity and group ability to utilize individual creativeresources: a multilevel model. Academy ofManagement Journal 2002;45(2):315–30.

Tierney P, Farmer SM, Graen GB. An examination of leadership and employeecreativity: the relevance of traits and relationships. Personnel Psychology1999;52:591–620.

Waldman DA, Atwater LE. The nature of effective leadership and championingprocesses at different levels in an R&D hierarchy. The Journal of HighTechnology Management Research 1994;5(2):233–45.

Woodman RW, Sawyer JE, Griffin RW. Toward a theory of organizationalcreativity. Academy of Management Review 1993;18(2):293–321.

Zhou J. Feedback valence, feedback style, task autonomy, and achievementorientation: interactive effects on creative performance. Journal of AppliedPsychology 1998;83(2):261–77.

Zhou J, George JM.When job dissatisfaction leads to creativity: encouraging theexpression of voice. Academy of Management Journal 2001;44(4):682–96.