e at SciVerse ScienceDirect

Technology in Society 34 (2012) 227–238

Contents lists availabl

Technology in Society

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

Scientific productivity and the collaboration intensity of Indonesianuniversities and public R&D institutions: Are there dependencieson collaborative R&D with foreign institutions?

Benyamin Lakitan a,b,*, Dudi Hidayat c, Siti Herlinda b,d,e

aMinistry for Research & Technology (RISTEK), Republic of Indonesia, Jakarta, IndonesiabCollege of Agriculture, Sriwijaya University, Palembang, IndonesiacCenter for S&T Development Studies, Indonesian Institute of Sciences (LIPI), Jakarta, IndonesiadResearch Center for Sub-optimal Lands (PUR-PLSO), Sriwijaya University, Palembang, IndonesiaeNational Research Council (DRN), Jakarta, Indonesia

a r t i c l e i n f o

Article history:Received 18 June 2012Received in revised form 26 June 2012Accepted 29 June 2012

Keywords:Scientific productivityCo-authorshipR&D collaborationInstitutionIndonesia

* Corresponding author. Ministry for Research &Republic of Indonesia, Jakarta 10340, Indonesia. Telfax: þ62 213102014.

E-mail address: blakitan@ristek.go.id (B. Lakitan

0160-791X/$ – see front matter � 2012 Elsevier Ltdhttp://dx.doi.org/10.1016/j.techsoc.2012.06.001

a b s t r a c t

Scientific productivity of universities and public R&D institutions in Indonesia is consid-erably low, but a notable rate of increase has been observed since 2008. This may beassociated with the implementation of new regulation to allocate at least 20 percent ofnational budget for education. Despite favorable government policy and financial incen-tives to encourage R&D collaboration among Indonesian universities and R&D institutions,there are still some constraints in managing and maintaining the collaboration. A lowdegree of collaboration among Indonesian researchers from different research organiza-tions, as indicated by the number of co-authored papers, has been observed. On the otherhand, Indonesian universities and public R&D institutions showed strong preference forcollaboration with foreign institutions. Most articles published in peer-reviewed interna-tional journals were results of collaborative research, predominantly with foreign partners.Indonesian researchers at public R&D institutions indicated a higher degree of dependencywith their foreign partners than academicians at universities.

� 2012 Elsevier Ltd. All rights reserved.

1. Introduction

There are twomajor concerns related to universities andpublic research and development (R&D) institutions inIndonesia. These are low scientific productivity andconstraints in establishing productive R&D collaboration.The concern with establishing productive R&D collabora-tion extends beyond universities and publicly funded R&Dinstitutions to include triple helix collaboration with busi-ness enterprises. be it among these two science and tech-nology developer groups or to form triple helix

Technology (RISTEK),.: þ62 81298115560;

).

. All rights reserved.

collaboration with business enterprise and governmentagency.

Recently, scientific productivity at the individual level(researcher or academician) and at the institutional level(university and public R&D institution) has been alarminglylow with only slight annual increases. The primary reasonfor this is a very low national budget allocation for thescience and technology sector. However, there was a majorchange in budget allocation since the Fourth Amendmentof the Indonesian Constitution was approved in 2002which obligates government to allocate at least 20 percentof the total national budget for education. Even though thisobligation is not directly for supporting R&D activities, it isexpected to have a positive influence on universityperformance in R&D. After ten years, it is an appropriatetime to evaluate this government regulation.

B. Lakitan et al. / Technology in Society 34 (2012) 227–238228

R&D collaboration has also been encouraged at least forlast two decades. Some government policies related to thisissue have been deployed, including financial incentive forR&D collaborations. Many different forms and sizes of R&Dcollaboration were established in response to these publicpolicies. However, scientific productivity of public R&D andhigher education institutions has not increased asexpected.

This paper will focus on two issues: scientific produc-tivity and R&D collaboration. One available indicator forevaluating these two issues is scientific publications asa proxy for productivity. This indicator can be further usedas raw data to scrutinize co-authorship between Indone-sian researchers with partners from domestic or foreigninstitutions. Co-authorship is used as proxy for R&Dcollaboration. Problems associated with these proxies[6,17,20] are recognized.

2. Scientific productivity

Scientific productivity is measured based on outputs ofscientific activity. In the broader perspective, the outputsinclude published scientific articles and other texts,prototypes and other physical artifacts, theories andcontent, and a better quality of human resources andinstitutions. Measuring all of these outputs requiresextraordinary efforts and not all of the outputs are quan-tifiable. In many cases they are not available or the infor-mation is not accessible. Therefore, in many studies withthe scientometrics approach, the number of articles pub-lished in peer-reviewed journal is used as proxy to scien-tific productivity.

Many factors could affect scientific productivity forindividual researchers or research organizations. Defazioet al. [8] found the impact of funding on productivity wasgenerally positive. However, the effect might take severalyears before it was detectable as shown in case of fundingfor European Union research collaboration program. Daraioand Moed [7] studied the effect of declining researchfunding during period 1980–2009 in Italy and found thatdespite the fact that the level of funding has been dramat-ically low during the past three decades, Italian science hasbeen able tomaintain its performance up to 2007. However,a recent reduction in the level of scientific production, thelagging behind in international scientific collaboration andthe great heterogeneity of researchers’ productivity maymark the start of a decline of Italian science.

Works of Daraio and Moed [7] and Defazio et al. [8]clearly indicated that the effect of increasing ordecreasing funding for R&D activities was not instantlyrevealed. This delay is due to the fact that some researchesrequire several years before publishable findings can beachieved.

Scientific productivity could also be affected by non-financial factors. Pezzoni et al. [23] related individual R&Dcareer progress with scientific productivity and indicatedthat they were associated with affiliation to importantpublic research organizations, social ties with seniormembers, and commitment toworkwith senior colleagues.Jacob and Lefgren [16] also found that recipients of USNational Institute of Health (NIH) postdoctoral fellowship

produced about one additional publication over the nextfive years, which reflects a 20% increase in researchproductivity.

The phenomenon of entrepreneurial universities overthe last decades has received [28] attention. They statedthat an entrepreneurial orientation by academia might putregions and nations in an advantageous position inemerging knowledge-intensive fields of economic activity.At the same time, such entrepreneurial orientationrequired reconciliation with the scientific missions ofacademia. They also revealed that scientific productivity ispositively associated with entrepreneurial effectiveness.

At the macro level, Horta and Veloso [15] conducteda comprehensive comparative analysis of the evolution ofthe EU15 and US scientific output and impact throughoutthe 1990s, looking at publications and impact trends byscientific field. Their results showed that changes inscientific production for the two blocks were driven byparticular scientific fields which grew or declined at a fastrate during the decade. Throughout this period, the EU15had eight fields of science, corresponding to 13% of the totalpapers published, growing at a rate faster than 10% inrelation to the world average, while the US had only fourfast growing fields, representing 6% of its total output. Thesituation was exactly reversed for the decline, with the UShaving more than doubled the number of scientific fieldswhen compared to the EU15 declining at a rate faster than10%. Despite this recent trend, the US maintains leadershipin impact across all scientific fields.

Based on results of all studies cited earlier, there isevidence to support arguments that: (1) scientific produc-tivity is directly influenced by the magnitude of fundingavailable for research and scientific activities, however,there is a time lag for several years after increase ordecrease of research funding initiated; (2) individualscientific productivity is also influenced by non-financialfactors such as institutional and social (patronage)factors; (3) it is possible to manage a balance betweenscientific productivity and entrepreneurial effectiveness orin other words, between efforts to achieve academicexcellence and to increase direct contribution for economicdevelopment; and (4) a scientifically productive institutionor country may not necessarily have dominant impactacross all or of any specific scientific field, since impact isassociated more with quality and relevancy rather than thequantity of scientific outputs produced.

3. Diverse definition of collaboration and use of co-authorship as its proxy

R&D collaboration has been encouraged at all levels, i.e.individual, laboratory or expert group, institution, andcountry. They are also encouraged between R&D institu-tions, industries, and government agencies. This three-party collaboration is known as the Triple Helix [9]. At anindividual level, Melin [20] indicated that the main benefitsgenerated from R&D collaboration were increased knowl-edge, higher scientific quality, contacts and connections forfuture work, and the generation of new ideas.

Collaboration involving R&D institutions have manydifferent formats and sizes, bi-lateral or multi-lateral,

B. Lakitan et al. / Technology in Society 34 (2012) 227–238 229

within a specific field or covering multiple fields of science.These varied kinds of R&D collaborations are believed tobenefit the parties involved. Bader [3] interpreted R&Dcollaboration as the integration of the company or someother organization or consortia into the university’s inno-vation process, in order to come up with new innovativeideas starting from the very beginning. Bader’s interpre-tation is a very university-centered type of collaboration.

The most commonly cited reasons for research collab-oration include: accessing expertise, accessing equipmentor resources one does not have, encouraging cross-fertilization across disciplines, improving access to funds,obtaining prestige or visibility, learning tacit knowledgeabout a technique, pooling knowledge for tackling largeand complex problems, enhancing productivity, educatinga student, and increasing the specialization of science [6].Interesting to note, based on their survey, other reasons forindividuals to become involve in R&D collaborationincludes for fun and pleasure [6] or social reasons [20].

From institutional, government, and funding agencyperspectives, encouraging R&D collaboration is a reason-able policy in order to increase funding efficiency, tominimize unnecessary replications of research activities, tomaximize the use of laboratory equipment and other R&Dfacilities, and to intensify communication and interactionamong researchers for maximizing knowledge exchange.Accordingly, Katz and Martin [17] also indicated thatamong factors which motivate collaboration are the fund-ing agencies’ need to save money, growing availability andfalling (real) cost of transport and communication, a desirefor intellectual interactions with other scientists, a need fora division of labor in more specialized or capital-intensiveareas of science, requirements of interdisciplinaryresearch, and following government encouragement forinternational and cross-sector collaboration.

Even though R&D collaboration creates benefits and isencouraged, it is not easy to clearly define collaboration.Melin [17] agreed that R&D collaboration is very difficult todefine. Partly, the notion of research collaboration is largelya matter of social convention among scientists. What somemight deem a collaboration, others may merely regard asa loose grouping or a set of informal links. Consequently,these collaborative contributions can also vary in level fromvery substantial to almost negligible. Sometimesa researcher may be seen as a ‘collaborator’ and listed asa co-author of an article published simply by virtue ofproviding material or performing a routine assay.

Melin [20] also recognized that collaboration can bemeasured and indicated in a number of ways. The amountof phone calls, international flights, or growing usages of e-mail are indicators that have been used. However, a moredirect way of measuring collaboration is through co-authorships, since a published product must exist as anoutcome of the cooperative effort, a more substantialindicator than just communication in one way or another.There has to be a distinction between fruitful chats overcoffee and systematized collaboration with publications asits result. Furthermore, Katz andMartin [17] concluded thatwhat constitutes a collaboration varies across institutions,fields, sectors and countries, and very probably changesover time as well.

It is recognized that co-authorship is unable to captureintensity or type of R&D collaboration and obviously nota perfect measure of the collaboration. Use of co-authorship as proxy of R&D collaboration has been ques-tioned by Katz and Martin [17]. One primary argumentagainst this approach is that many cases of collaborationthat are not ‘consummated’ in a co-authored paper andwhich are consequently undetectable with this approach.Conversely, there are other cases of, at best, only veryperipheral or indirect forms of interaction between scien-tists which nonetheless yield co-authored publications.Therefore, co-authorship is only a rather approximatepartial indicator of collaboration. Such an indicator must betreated with caution [17]. However, at present, there is nocomposite index available to comprehensively measure allpossible intensity and types of R&D collaboration.

Although the assessment of R&D collaboration using co-authorship is by no means perfect, it nevertheless hascertain advantages [26]. Firstly, it is invariant and verifiable,given access to the same data-set, other investigatorsshould be able to reproduce the results. Secondly, it isa relatively inexpensive and practical method for quanti-fying collaboration. Thirdly, some would argue that it isnon-intrusive and indeed non-reactive since themeasurement does not affect the collaboration process.

4. Does collaboration increase scientific productivityand quality?

Ynalvez and Shrum [29] made a clear distinctionbetween professional networks and scientific collabora-tion. Scientific productivity was significantly linked toprofessional networks, but there was no evidence of anyassociation with scientific collaboration. A possible answerto this puzzle appears to be rooted in the practice whereresearchers viewed collaborative research projects notmainly as a means to producing knowledge and gainingrecognition, but for acquiring professional opportunitiesand extrinsic rewards. Professional networks in this casewere seen as a less formal interaction and communicationamong researchers, while scientific collaboration wasperceived as more formal or regulated forum forresearchers to work together.

Porac et al. [25] compared publication outcomes of twoteams within a multi-university scientific alliance. Thisalliance by Ynalvez and Shrum’s description should beclassified as collaboration. Scientists in one team sharesimilar scholarly backgrounds and work in a well estab-lished paradigm, while scientists in the second team havedifferent backgrounds and work in an emergent discipline.While the alliance has increased the productivity of bothteams, this increase was highest for the more heteroge-neous team. In addition, while the variety of knowledgeconcepts employed in their research was initially higher forthe heterogeneous team, however, this gap narrowed overtime.

University research centers may organize researchersfrom across disciplines which (collectively as a researchunit) to possess scientific and technical capacity relevant toscientific and technical demand of specific current issues orof their potential sponsoring partners. Affiliation with this

B. Lakitan et al. / Technology in Society 34 (2012) 227–238230

multidiscipline center can be effective for enhancing overallproductivity as well as at facilitating cross-discipline, cross-sector, and inter-institutional productivity [24].

Furthermore, collaboration does not necessarily resultin a significantly higher number of citations an articlereceived [19]. Some researchers tend to relate this numberof citations to research quality. However, there are legiti-mate criticisms against using citations as a proxy for qualityof an article, since citations may merely be measuring theimpact rather than the innovative quality of an article. Infact, negative citations are not distinguished from positivecitations. If an article’s impact is the ability to stimulatefurther thought, then citation and quality are positivelycorrelated.

In short, R&D collaboration cannot guarantee anincrease in scientific productivity and quality of R&Doutputs, however, many studies indicated that collabora-tion composed of heterogenous or multi-disciplinarymembers had a better chance to be more productive andto produce better quality outputs.

5. Preferred versus ideal partners for collaboration

R&D collaboration is not always purely based on scien-tific reasons. For instance, Melin [20] found that there weresome academic reasons for the collaboration such as forfinding partners with special competence (41% of 195respondents), for access to special data or equipment (20%),and for developing and testing new methods for (9%) ofrespondents; however, there were also non-academicreasons such as for establishing or maintaining socialrelationship (16%); and the other occurred under theumbrella of a supervisor–student relationship (14%).

The non-academic reason for collaboration is real andcannot be overlooked. Involvement of former graduatestudents is a key determinant of institutional collaboration,especially collaboration with firms and with foreignscientific institutions [1]. Reasons for this type of collabo-ration may also be predominantly social rather thanacademic.

Adams et al. [1] also observed trends in co-authorshipduring period of 1981 to 1999. They found that team sizehad nearly doubled over the 19-year period. It was alsofrequently found that articles published during the lastdecade were co-authored by more than 20 persons, espe-cially in health/medical sciences.

The trend in increasing team size may be related to thebelief that scientific output and influence also increaseswith team size and that influence rises along with institu-tional collaboration. Since increasing team size implies anincrease in the division of labor, these results suggest thatscientific productivity increases with the scientific divisionof labor [1]. This argumentmay be valid if increases in teamsize are directly correlated with the diversity of expertiserequired in conducting the research.

One of the benefits expected from R&D collaboration isan exchange of knowledge during research activities.Fritsch and Franke [11] found that R&D cooperation wasonly of relatively minor importance as a medium forknowledge spillover on innovation activities in threeGerman regions, i.e. Baden (western part of the state of

Baden-Wuerttemberg), region Hanover–Brunswick–Goet-tingen (in the state of Lower Saxony), and Saxony (one ofthe new German states which was until 1990 undera socialist regime). Amazingly, knowledge spilloversstemming from other actors within the respective regionwere observed. However, Fritsch and Franke [11] could notexplain how the majority of innovation-relevant knowl-edge spillovers occur within a region, not due to collabo-ration between regions.

In our perspective, knowledge spillover or exchange isnot associated with geographical or territorial distance. Theexchange is more likely driven by the knowledge gapbetween collaborated parties. Therefore, it should not besurprising that only minor spillover was observed if threeregions studied by Fritsch and Franke were at almost equalknowledge level. In contrast, an outstanding individual ora highly reputable institution within the region coulddrives knowledge flow to others within the region, be it ina formal collaboration or not, as long as there is mode ofcommunication available among related parties.

Using data on all co-publications between 313 regions in33 European countries for the period 2000–2007, Hoekmanet al. [14] found that there was considerable heterogeneitybetween regions and countries in their propensity tocollaborate which they attribute to differences in size,quality and accessibility.

Practical implications for the management of R&Dcollaboration emerge. First, trust is an indispensableingredient for any cooperative effort. Second, a certaindegree of interdependency between partners is needed tooffer reciprocal incentives for cooperating. Third, scientistsshould be aware that the frequency of communication andthe closeness of partners seem to be important in cooper-ation projects, regardless of the type of knowledgeinvolved. However, when tacit knowledge is involved, thecloseness of partners seems to be of major relevance [21].

The propensity in selecting partner for R&D collabora-tion could be based on pure academic or socially practicalreasons. Core of academic reason is exchange or transfer ofknowledge which could be in form of equal partnership orleader–follower relationship. For the later form, moredeveloped countries or institutions receive compensationusually in the form of access to research subjects thatbelong to recipients of knowledge transferred. Non-academic reasons may stretch over a very wide spectrumof possibilities.

Objectives for collaboration might be different betweenindividuals or organizations in developed and developingcountries. Bozeman and Corley [6] reported that mostresearchers in developed countries are not particularlycosmopolitan in their selection of collaborators, they tendto work with the people in their own work group or withwhom they previously met, i.e. alumni of the sameuniversity or had participated at the same training, or somekinds of other social connection.

R&D units in universities in developing countries thathave limited in-house resources and wish to strengthentheir research capability should implement strategies thataim at extending their potential through collaborationwitha variety of external partners [22]; especially with univer-sities and R&D institutions in developed countries.

B. Lakitan et al. / Technology in Society 34 (2012) 227–238 231

6. Role of regulations and public policies

In the US, university alliances have been spurred, in part,by the availability of government funds from the NationalScience Foundation (NSF) and other public agencies thathave been earmarked for collaborative inter-universityresearch. Good examples of these collaborative endeavorsare NSF’s Science and Technology Centers focusing ontopics such as nanotechnology, adaptive optics, andbehavioral neuroscience. Other forms of inter-universitycollaboration are evident in the NSF sponsoredEngineering Research Centers on systems engineering,optoelectronics, and advanced electronic materials pro-cessing. Each of these endeavors represents a multimilliondollar program of collaboration, usually organized aroundone or two leading universities, and is focused on thetransfer and combination of specialized domain knowledgeacross university boundaries. As in other types of alliances,the goal of inter-university collaboration is to spur newinsights in the domains covered by a particular alliance bybringing together researchers at different universities whootherwise would not, or could not, collaborate in theirresearch [25].

The government of Indonesia also developed similarprograms to encourage R&D collaboration among univer-sity, public R&D institution, and industry. One of thecollaborative programs (2000–2008) was the NationalStrategic Research (known as RUSNAS), focused on selectedissues, i.e. food diversification, tropical fruits, vehicleengines, palm oil processing industries, coral reef fishculture, information technology and microelectronics,renewable energy, and cattle breeding. Even though theRUSNAS program had been discontinued, basic policy onencouraging R&D collaboration is still on the priority list.

Similar program are also deployed by the EuropeanUnion. EU-funded research networks require researchers tocollaborate as a condition for securing research funding.Defazio et al. [8] observed that during the period of fund-ing, collaboration facilitated by EU program did not lead toan increase in research production. However, in the post-funding period they find that although the number ofcollaborations decreases within the network, the impact ofcollaboration on productivity is positive and significant.Their findings suggest that collaborations formed to capi-talize on funding opportunities, while not effective inenhancing researcher productivity in the short run, may bean important promoter of effective collaborations in thelonger run.

Despite favorable government policy for R&D collabo-ration in many countries, Heinze and Kuhlmann [12] stillargued that governance structures may support or hinderscientists’ efforts to engage in collaborative work relationsacross institutional boundaries within the highlysegmented public research system, such as in Germany. Theissue of governance structures as hindering factor is morecritical in the context of developing country such asIndonesia. By analyzing innovation policies in Brazil andEstonia, [18] have shown that the lack of governing capacityis a major cause of innovation policy implementationfailure. This may also explain the unsatisfactory result ofthe RUSNAS program in promoting research collaboration.

Other favorable public policies include subsidies orfinancial incentives for R&D activities in industry. Feldmanand Kelley [10] recognized that R&D subsidies are aneffective public policy instrument when knowledge spill-overs exist, yet ex ante, it is difficult to identify projects thathave the greatest potential to increase innovation andeconomic growth. Policy makers should include in thedesign instruments aimed at fostering scientific collabora-tion, especially on how to evaluate their impact. Most of thelimitations of studies on impact of collaboration are due tothe lack of an ex-ante evaluation design [27]. Despite thisissue on ex ante evaluation and more work to be done indeveloping the tool, policies for encouraging researchcollaboration should be continued since the receipt ofa government R&D subsidy (to a firm) increased fundingavailability from other sources.

Government subsidies should be wisely treated aspublic investment. However, as the ‘Swedish Paradox’ andsubsequently the ‘European Paradox’ have suggested,investments in new knowledge do not automaticallytranslate into competitiveness and growth [2]. Both ofthese euphemistically describe a more general paradoxassociated with knowledge, that high levels of investmentin new knowledge do not necessarily and automaticallygenerate the anticipated levels of competitiveness oreconomic growth.

Lack of positive correlations between investment inR&D and economic growth of a specific sector, region, orcountry may associate with the condition that focus of R&Dis not relevant to real problems or needs of industries orother technology users. Therefore, adoption of the tech-nologies produced by R&D institutions is very low whichconsequently only contribute insignificantly to economy ofthe region or country.

Audretsch and Keilbach [2] suggested that an impor-tant function for public policy is to facilitate the spilloverand commercialization of knowledge by encouragingentrepreneurship. This supply–push approach, however,will only be successful if the technology created is suitablefor producing goods or services according to marketdemand.

7. Scope and data sources

In Indonesia, there are several types of institutionsresponsible for science, technology development, andinnovation (STI). Some conduct research and development(R&D) activities for creating new technologies or improvingunderstanding on scientific issues or real problems; whileother institutions play supporting roles that are crucial foradvancement of STI.

Institutions conducting R&D activities are classified intofive types: (1) universities, besides their main responsi-bility in improving the quality of human resources throughformal education, they are also responsible to develop STI;(2) public R&D institutions, funded by government but notaffiliated directly with any ministry, known as ‘LPNK’; (3)R&D agencies within a ministry, known as ‘Balitbang’, witha clear mandate to provide knowledge and technology asrequired by the ministry in formulating public policies andfor providing technical or technological assessments; (4)

B. Lakitan et al. / Technology in Society 34 (2012) 227–238232

non-government R&D institutions; and (5) an R&D unitwithin industrial or business firms.

The scope of this study is limited to university, LPNK,and Balitbang, since there are a very limited number ofnon-government R&D institutions in Indonesia and thereis limited access to R&D units in industry. In addition, theR&D survey across manufacturing industry conducted byPAPPIPTEK-LIPI in 2001 has shown that R&D expenditurein this sector is indeed quite low. It accounts for less than15% of the total national R&D expenditure. Further esti-mation amounts at most that this might be 25% of thetotal.

Publication in peer-reviewed journals is used as proxyfor scientific productivity and collaboration intensityamong R&D institutions. Raw data sources for scientificarticles published in peer-reviewed journal are SciVerseScopus and Thompson Reuters databases. In addition tothese two databases, data for publication in national jour-nals were from survey conducted in 2011 by Ministry forResearch and Technology, Republic of Indonesia. Data fromSciVerse Scopus and Thompson Reuters were downloadedduring January to February 2012. Data on the number ofresearchers in selected Southeast Asia countries wasdownloaded on 14 February 2012 from UNESCO Institute ofStatistics.

For identifying similarities and differences amonguniversities, LPNKs, and Balitbangs, eight most productiveIndonesian universities were selected, i.e. University ofIndonesia (UI), Institut Teknologi Bandung (ITB), Gadjah-mada University (UGM), Institut Pertanian Bogor (IPB),Padjadjaran University (UNPAD), Airlangga University(UNAIR), Diponegoro University (UNDIP), and HasanuddinUniversity (UNHAS). For more in-depth study, we focusedon the top-three universities, i.e. UI, ITB, and UGM.

Three LPNKs selected were Indonesian Academy ofSciences (LIPI), Agency for Assessment and Application ofTechnology (BPPT), and National Nuclear Energy Agency(BATAN). Three Balitbangs selected were of the Ministry ofAgriculture (MoA), the Ministry of Health (MoH), and theMinistry of Maritime and Fishery (MoMF).

This study focused on scientific productivity, collabo-ration intensity, and field of research priority. Scientificproductivity of an R&D institution is based on cumulativenumber of article published per year, involving researchersaffiliated with the institution.

Subject categories used for classifying research focus orpriorities both for R&D institution and countries of interestwere adopted from those of SciVerse Scopus combinedwith those of Thompson Reuters, and then grouped intoeight major categories including: basic sciences, medicineand medical sciences, biological sciences and genetics,agricultural sciences, engineering and industrial relatedsciences, environmental sciences, economics and business,and social sciences.

Collaboration intensity among R&D institutions isassociated with co-authorship in scientific articles pub-lished involving two ormore institutions. Since this study isfocused on Indonesian R&D institutions and limited toselected universities, LPNKs, and Balitbangs as mentionedearlier, therefore, only collaboration involving one or moreof these R&D institutions were recorded.

Concerns and notes from Bozeman and Corley [6] on theusage of the co-authorship parameter, based on data forstudying research collaboration as well as Katz and Martin[17]; questioned the adequacy of co-authorship inattempting tomeasure R&D collaboration are recognized. Itwas argued that inter-institutional and internationalcollaboration need not necessarily involve inter-individualcollaboration and concluded that co-authorship is nomore than a partial indicator of collaboration. Otherparameters that might be used as a proxy for R&D collab-oration are patents with multiple inventors from differentinstitutions or countries as suggested by Bergek andBruzelius [4]. Another proxy mentioned in this investiga-tion includes sharing funding on a specific R&D activity.However, each of these parameters may also havelimitations.

8. Results and discussion

8.1. Scientific productivity

Academic productivity of Indonesian universities wasconsiderably low, as indicated by the total number ofannual publication in peer reviewed journals; however,there was significant increase observed since 2008. Articlespublished annually by the top-eight universities duringperiod of 2008–2011 (total of 130.25 articles or 16.28 arti-cles/university) more than doubled compared to thosepublished annually during period of 2001–2007 (total61.29 articles or 7.66 articles/university) (Fig. 1). Thispromising trend is presumably associated with approval ofthe Fourth Amendment of Indonesian Constitution by thePeople’s Consultative Assembly in 2002 which orderedgovernment to allocated at least 20 percent of nationalbudget (APBN) for education, as clearly stated in Article 31of the amended version of the constitution.

In the following year of 2003, the new law on thenational education system (Law No. 20/2003) was passedby the People’s Representative Council. However, therewere challenges by opposing groups on the article 31.Nonetheless, in 2005 the Constitutional Court decided tokeep the regulation in effect. Since then, budget allocationfor R&D activities has also gradually increased. However,positive impact was only visible since 2008 (Fig. 1), aftersome additional supporting policies and guidelines atinstitutional level were implemented. The national highereducation policy and internal university policy haveencouraged academicians to publish their research find-ings. The policy includes financial incentives rewarded forany published articles in peer-reviewed internationaljournals by Indonesian scientists.

The regulatory framework conditions have been iden-tified as important factors influencing R&D activities,including publishing research results in scientific journal,but the impacts of regulation can be ambivalent. Differenttypes of regulations may generate different impacts andeven a single regulation can influence R&D activities invarious ways depending on consistency of how the regu-lation is implemented in Blind [5].

Naturally, the impact of a new regulation or policy willnot be instant. Blind [5] distinguished between short-term

0

10

20

30

40

50

60

2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011

Publ

ishe

d Pe

er-re

view

ed A

rticl

es-

Year

UI

ITB

UGM

IPB

UNPAD

UNAIR

UNDIP

UNHAS

AVERAGE

Amendmentof the Constitution: 20 percent of government budget

7.66 articles/year 16.28 articles/year

Fig. 1. Scientific articles published in international journals by top-eight Indonesian universities, 2001–2011.

B. Lakitan et al. / Technology in Society 34 (2012) 227–238 233

and long-term effects in his study on the influence ofdifferent regulations on innovation. Defazio et al. [8] alsoreported that the impact of new policy or regulation nor-mally takes some years before it is clearly detected. Forinstance, EU-funded research collaboration also took manyyears before it gave significant impact on scientificproductivity. When delineated between pre-, during- andpost-funding periods, there were some important differ-ences. During the period of funding, collaboration did notlead to an increase in research production. In the postfunding period, it was found that although the number ofcollaborations decreases within the network, the impact ofcollaboration on productivity is positive and significant.This finding suggested that collaboration was formed tocapitalize on funding opportunities, while not effective inenhancing researcher productivity in the short run, it maybe an important promoter of effective collaboration in thelonger run. In the Indonesia case, an increase in scientificpublication was detected after six years of amendment ofthe constitution.

The general trend for scientific productivity of Indo-nesian R&D institution after the amendment of theconstitution has been fairly promising (Fig. 2). However,at this point, it was far from satisfactory when comparedto progress shown by other Southeast Asia (ASEAN)countries, especially Singapore, Thailand, and Malaysia(Fig. 3). Lao, Cambodia, Myanmar, and Timor Leste are notshown in Fig. 3, they are at bottom group with less than50 international peer-reviewed articles published annu-ally. Malaysia has shown a clear exponential increaseduring period of 2002–2011 and may extend beyond thisperiod.

As a reminder for policy-makers in Indonesia, it is wiseto consider that increasing R&D funding must go hand-in-hand with reforming governance of R&D institutions,establishing effective internal control system, and devel-oping workable scenarios for performance enhancement,including incentive systems.

At present, most of the research findings at Indonesianuniversities, LPNKs, and Balitbangs are published innational journals, written in the Indonesian language withonly a few exceptions in English. However, almost all of thejournals provide English abstracts. Unfortunately, most ofthem are published only in a printed version and are notaccessible electronically. For LPNKs, ratio of national/international publication varied from 3.0 to 11.0 withaverage of 4.1 articles; and for Balitbangs, the ratio variedfrom 11.0 to 62.0 with average of 15.1 articles (Table 1).

There are different mandates between the two R&Dinstitutions. LPNKs are expected to develop new technol-ogies and strengthen the R&D capacity, while Balitbangshave a mandate to provide solutions for more technicalissues within its core responsibility and/or comprehensiveinformation to be used for formulating public policies.

If these publications in national journals are considered,scientific productivity of LPNKs should be multiplied by 5for its publication in international journals, andmultiply by16 for Balitbangs. Obviously, the scientific impact of thepublication is still considerably low, since articles written inIndonesian language are very rarely quoted by foreignresearchers.

It is also interesting to note that more productive R&Dinstitutions for each category also have a lower ratio ofnational/international publication, inferring that produc-tivity in terms of publication in national journals may bea necessary step for establishing a scientific base beforesubmitting articles to international peer-reviewed journal.

In general, the number of peer-reviewed articles pub-lished in international journal was generally associatedwith the number of PhD-holder researchers at the institu-tion, as shown in case of LIPI (Table 1). Compared to thescientific productivity of researchers in ASEAN countries,productivity per head count or per full time equivalence ofIndonesian researchers was lower than those ofresearchers in Singapore, Malaysia, Thailand, Philippines,and Vietnam (Fig. 4).

y = 1E-112e0.1312x

R² = 0.9752

50

100

150

200

250

300

350

400

450

2000 2002 2004 2006 2008 2010 2012

Publ

ishe

d Ar

ticle

Year

Fig. 2. Trend of scientific article published by Indonesian scientists, 2002–2011.

B. Lakitan et al. / Technology in Society 34 (2012) 227–238234

8.2. Collaboration intensity

Most of the publications by Indonesian researchers inpeer-reviewed journals were based on collaborative workswith other institutions. Of 5,088 publications by Indonesianuniversities evaluated, 76.3 percent were based on the

SINGAPORE

THAILAND

0

500

1000

1500

2000

2500

3000

2002 2003 2004 2005 2006 20

Artic

le P

ublis

hed

Year

Fig. 3. Scientific Articles Published by Sel

results of collaboration with national and/or foreign insti-tutions. The percentages were even higher for non-university R&D institutions, 90.6 percent and 94.3 percentfor LPNKs and Balitbangs, respectively (Table 2). Ubfal andMaffioli [27] reported similar trend in Argentina thatgovernment support for the funding of R&D collaboration

MALAYSIA

PHILIPPINES

VIETNAMINDONESIA

07 2008 2009 2010 2011

ected ASEAN countries, 2002–2011.

Table 1Total articles published and individual productivity of selected public R&D institutions in Indonesia, 2008–2010.

R&D Institution Total article published2008–2010

Ratio National/International

Average Individual Productivity per Year

Nationaljournal

Internationaljournal

National publicationper researcher

National publicationper graduatedegree-holder

International publicationper PhD holder

Ministry of Agriculture 927 84 11.0 0.13 0.24 0.07Ministry of Industry 248 4 62.0 0.15 0.45 0.10Ministry of Maritime

& Fishery261 7 37.3 0.19 0.28 0.02

Total or Averagefor Balitbang

1436 95 15.1 0.14 0.27 0.06

BPPT 233 44 5.3 0.05 0.11 0.08LIPI 704 236 3.0 0.18 0.34 0.31BATAN 320 29 11.0 0.15 0.39 0.12Total or Average

for LPNK1257 309 4.1 0.12 0.25 0.20

B. Lakitan et al. / Technology in Society 34 (2012) 227–238 235

also demonstrated a positive and significant impact on thenumber of co-authored publications in peer-reviewedjournals.

The high percentage of articles published based on R&Dcollaborations, in the Indonesian case may be translatedinto two very different perceptions: either R&D collabora-tions have been proven to be an effective way to encourageR&D institutions in Indonesia to publish results of theirR&D activities, or researchers and R&D institutions inIndonesia are very dependent on their partners forpublishing in international journals. These two perceptionscould be further evaluated based on the role of the Indo-nesian researchers in publishing the articles as first orcorresponding author or as co-author. It is fair to assumethat researchers who lead or dominantly contribute inplanning and conducting the R&D activities and writing thearticle would have the privilege to be the first author.

It is interesting to note that there are differences andsimilarities among top-three Indonesian universities. ITBand UGM showed that more than half (57.5 percent for ITBand 63.0 percent for UGM) of their collaborated publicationwere authored by their researchers with co-authors fromother institutions, mostly foreign R&D institutions. In thecase of UGM, a significant portion of these collaborativepublications was written by its academic staff based on

0.33

2.96

0.88

0.17

1.51

0.53

0

1

2

3

Indonesia Malaysia Philippines

Artic

le

P

ub

lis

he

d

Per Full Time Equivalen

Fig. 4. Average article published per research

results of thesis research conducted at foreign universities.In accordance with this case, [1] found that former grad-uate students were key determinants of institutionalcollaborations, especially collaboration with foreign scien-tific institutions.

The case was different for UI, where only 21.9 percent ofthe collaborative publications were authored by itsresearchers. A larger portion (37.0 percent) of publicationsinvolving UI researchers were co-authored articles auth-ored by foreign partners. Most of these were in the form ofregional papers on public issues associated with health. Forsuch articles, it seems that anybody involved in thisregional collective work was included as co-author of thepapers. There was a trend of team size increase in collab-orative R&D similar to that observed by [1]. Team size hadincreased by about 50 percent over 19-year period of 1981–1999. Adam’s observed this increase for papers publishedby selected 110 top US research universities.

More than one-fourth (27.7 percent) of the total scien-tific articles of UI were published without co-authors fromother institution. Similarly, ITB also showed that almost onethird (32.4 percent) of its publications were without co-author from other institution (Table 2), indicating thatthese two universities are not dependent on their partnersfor publishing their scientific articles.

2.57

1.66

0.84

2.14

0.91

0.19

Singapore Thailand Vietnamce Per Head Count

er in six ASEAN countries, 2001–2011.

Table 2Co-authorship in scientific publications in Indonesia, 2001–2011.

R&D Institutions Number ofarticleevaluated

Number ofresearcherinvolved

Withoutcollaboration(%)

With collaboration (%) Total collaboratedarticle published(%)

Leading rolein publishing(%)

As first orcorrespondingauthor

As co-author with firstauthor from

ForeignInstitutions

IndonesianInstitutions

A. UniversityITB 1.991 1.466 32.4 38.9 18.5 10.2 67.6 57.5UI 1.921 1.458 27.7 21.0 37.0 14.3 72.3 29.1UGM 1.176 830 9.8 56.9 23.5 9.8 90.2 63.0Average for

University5,088 23.7 38.0 26.7 11.6 76.3 49.8

B. LPNKLIPI 727 326 3.0 30.0 54.0 13.0 97.0 30.9BPPT 224 125 9.0 36.0 52.0 3.0 91.0 39.6BATAN 197 153 21.4 48.2 19.6 10.7 78.6 61.4Average for LPNK 1,148 9.4 36.3 45.7 8.6 90.6 40.1C. BalitbangMoH 420 268 8.7 39.1 41.3 10.9 91.3 42.9MoA 173 157 1.5 15.4 64.6 18.5 98.5 15.6Average for

Balitbang593 5.7 29.3 51.0 14.0 94.3 31.1

ITB

1991

UI

1921

UGM

1179

41 16

16

[1,05]

[0,52]

[0,50]

Fig. 5. Limited R&D collaboration among top-three Indonesia universities,2001–2011 (Numbers in parenthesis are percentage of co-authorshipbetween each couple).

B. Lakitan et al. / Technology in Society 34 (2012) 227–238236

Based on a combination of an independent ability topublish and their role in collaborative publication, the top-three Indonesian universities do not appear to be depen-dent on collaboration with foreign institutions to publishresults of their R&D activities in peer-reviewed journals.However, R&D collaboration does play a significant role inincreasing publication productivity at these three univer-sities, especially for UGM.

However, LPNK in Indonesia indicated more depen-dence on collaboration with foreign institutions for publi-cation in peer-reviewed journal, except for BATAN. Asa national strategic facility, there is selective access forconducting research for researchers from outside of thisinstitution. Most of R&D activities at BATAN either con-ducted internally by its researchers (21.4 percent) or inform of collaborative researches lead by its researchers(48.2 percent) (Table 2).

In contrast, the majority of scientific publicationsinvolving researchers at LIPI and BPPT were based oncollaborative researches lead by foreign institutions, i.e.,54.0 percent for LIPI and52.0percent for BPPT. R&Dagenciesat Indonesian ministries fall into the same category as LIPIand BPPT. Dependency on foreignpartners was 41.3 percentfor Balitbang at MoH and 64.6 percent for that at MoA.

Another interesting finding is that all Indonesian R&Dinstitutions prefer to collaborate with foreign R&D insti-tutions more than with national institutions. This prefer-ence may be interpreted as a signal that R&D collaborationfrom Indonesian R&D institution’s perspective is more anopportunity to improve R&D capacity and quality ratherthan a way to work together on issues of common interestin spirit of equal partnership. Factors driving IndonesianR&D institution towards this preferencemay include accessto more sophisticated research facilities, knowledge spill-over, opportunities to pursue post graduate degree offeredby foreign partner, and sources of fund provided in the R&Dcollaboration schemes.

Similar reasons were indicated by [22] for universityresearch units in Thailand. Due to limited in-houseresources and in order to strengthen their research capa-bility, R&D units at universities in Thailand prefer tocollaborate with a variety of external partners, especiallywith universities and R&D institutions in developedcountries.

In contrast, for different reasons, R&D institutions anduniversities in the US were not particularly cosmopolitan intheir selection of collaborators. They tend to work with thepeople in their own work group [6]. These two contrastingpreferences may strongly be associated with degree ofbenefits received by each party from R&D collaboration.Most collaboration between R&D institutions or universi-ties of developed and developing countries fall intoscenarios where research objects are located in the devel-oping countries while fund and R&D facilities are providedby developed countries.

SINGAPORE

THAILAND

MALAYSIA

VIETNAM

INDONESIA

PHILIPPINES

y = 0,5695ln(x) + 1,5622R² = 0,9514

1.5

1.7

1.9

2.1

2.3

2.5

2.7

2.9

3.1

3.3

3.5

0 5 10 15 20 25

Dom

estic

Res

earc

herI

nvol

ved

perA

rticl

e

Published Article (x1000)

Fig. 6. Correlation between average domestic researcher involved in publishing an article and total article published by ASEAN countries, 2001–2011.

B. Lakitan et al. / Technology in Society 34 (2012) 227–238 237

There are three universities in Indonesia that consis-tently published more scientific articles than the rest ofuniversities, i.e. UI, ITB, and UGM. Fig. 1 shows their posi-tion during last 11 years (2001–2011). However, co-authorship in peer-reviewed international journalsamong these top-three Indonesian universities was verylow (Fig. 5). Onemay argue that the low co-authorship doesnot necessarily reflect lack of R&D collaboration amongthese universities. However, since there are significantconstraints in many government programs to encourageR&D collaborations among universities in Indonesia. Inmost cases, such government-initiated collaboration wasdiminished as soon as the program was terminated.Therefore, the use of low co-authorship as an indicator oflow R&D collaboration becomes reasonable, at least forIndonesia case. There are some indications, however, thatthese top-three universities published relatively morearticles with staff of other Indonesian universities, ina scenario included in the context of the student-advisorrelationship, where the graduate students are academicstaffs from the other Indonesian universities.

Top universities in Indonesia tend to compete with eachother rather than to collaborate. This might be stimulatedby competitive research grant provided by Indonesiangovernment. There are mixed general R&D policies inIndonesia, one is directed towards academic excellencesupported by purely competitive research grants and theother is more focused on improving R&D capacity

supported by block grants allocated directly to targetedless-developed institutions.

Striving for academic excellence, as remained by Hicks[13]; may compromise other important values such asequity or diversity; and will not serve the goal of enhancingthe economic relevance of research.

Research teams in Indonesia also tend to be smallercompared to teams of more productive countries. ForASEAN countries, team size was also following similarpatterns as those found in productive countries. Singapore,Malaysia, and Thailand as more productive countries, hadlarger team size as opposed to Vietnam, Philippines, andIndonesia which were scientifically less productive coun-tries had smaller team size (Fig. 6). On average, for theIndonesia case, less than two researchers/academicianswere involved in publishing a scientific article. This factreasonably indicated that collaboration between Indone-sian scientists was still weak.

9. Conclusions

Scientific productivity of universities and public R&Dinstitutions in Indonesia is considerably low, but there hasbeen a notable rate of increase since 2008. The annualnumber of article published since 2008 more than doubledthe yearly average for the period before 2008. Directly orindirectly this increase was associated with the imple-mentation of the Fourth Amendment of the Constitution

B. Lakitan et al. / Technology in Society 34 (2012) 227–238238

which required government to allocate at least 20 percentof national budget for education sector, including researchin higher education institution. Incentive policies forpublishing articles in peer-reviewed international journalhave also been part of the higher education program.

Despite favorable government policies and financialincentives to encourage R&D collaboration among Indo-nesian universities, public R&D institutions, and industries,there still exists hesitancy and constraints in managing andmaintaining the collaboration. Negligible positive impactson scientific productivity associated with collaborationamong Indonesian institutions were observed. Instead,Indonesian universities and public R&D institutionsshowed a strong preference to collaborate with foreigninstitutions.

Most of articles published in peer-reviewed interna-tional journals were results of collaborative research,predominantly with foreign partners. Researchers at LPNKand Balitbang indicated a higher degree of dependencywith their foreign partners for publishing the articles thanwas observed by academicians at universities.

In general, R&D collaboration has to be continuouslyencouraged in Indonesia. This research points to two majorbenefits for Indonesia as well as other developing countriesin encouraging this “dependency”. Firstly, the collaborationwith R&D institutions of developed countries will provideopportunities for Indonesian R&D institutions to acquireadvanced S&T and to access more sophisticated R&D facil-ities. Secondly, the collaboration among Indonesian R&Dinstitutions will be necessary for improving national R&Dcapacity and for increasing R&D budget efficiency.

Dependencies on foreign partners in publishing scien-tific articles of some Indonesian public R&D institutions areexpected to be gradually transformed into a more equalpartnership. This transformation will be in line with prog-ress in raising R&D capacity of the Indonesian institutions.The progress can be boosted if the Government ofIndonesia decides to allocate at least one percent of its GDPfor funding R&D activities as recommended by NationalInnovation Committee in 2011.

Acknowledgements

Part of the data used in this paper was from survey onpublic R&D institutions funded by theMinistry for Researchand Technology, Republic of Indonesia in 2011. Thecomments and feedback contributed by colleagues atSriwijaya University, Indonesian Institute of Sciences, andtwo anonymous reviewers are deeply appreciated.

References

[1] Adams JD, Black GJ, Clemmons JR, Stephan PE. Scientific teams andinstitutional collaborations: evidence from U.S. universities, 1981–1999. Research Policy 2005;34:259–85.

[2] Audretsch DB, Keilbach M. Resolving the knowledge paradox:knowledge-spillover entrepreneurship and economic growth.Research Policy 2008;37:1697–705.

[3] Bader, M.A. Managing IP in R&D collaborations. Regional workshopon searching of IP information for university IP coordinators. Skopje,Macedonia; 7–9 June 2006.

[4] Bergek A, Bruzelius M. Are patents with multiple inventors fromdifferent countries a good indicator of international R&D collabo-ration? The case of ABB. Research Policy 2010;39:1321–34.

[5] Blind K. The influence of regulations on innovation: A quantitativeassessment for OECD countries. Research Policy 2012;41:391–400.

[6] Bozeman B, Corley E. Scientists’ collaboration strategies: implica-tions for scientific and technical human capital. Research Policy2004;33:599–616.

[7] Daraio C, Moed HF. Is Italian science declining? Research Policy2011;40:1380–92.

[8] Defazio D, Lockett A, Wright M. Funding incentives, collaborativedynamics and scientific productivity: evidence from the EU frame-work program. Research Policy 2009;38:293–305.

[9] Etzkowitz H, Leydesdorff L. The dynamics of innovation: fromnational systems and ‘‘Mode 2’’ to a triple helix of university–industry–government relations. Research Policy 2000;29:109–23.

[10] Feldman MP, Kelley MR. The ex ante assessment of knowledgespillovers: government R&D policy, economic incentives and privatefirm behavior. Research Policy 2006;35:1509–21.

[11] Fritsch M, Franke G. Innovation, regional knowledge spillovers andR&D cooperation. Research Policy 2004;33:245–55.

[12] Heinze T, Kuhlmann S. Across institutional boundaries? Researchcollaboration in German public sector nanoscience. Research Policy2008;37:888–99.

[13] Hicks D. Performance-based university research funding systems.Research Policy 2012;41:251–61.

[14] Hoekman J, Frenken K, Tijssen RJW. Research collaboration ata distance: changing spatial patterns of scientific collaborationwithin Europe. Research Policy 2010;39:662–73.

[15] Horta H, Veloso FM. Opening the box: comparing EU and USscientific output by scientific field. Technological Forecasting andSocial Change 2007;74:1334–56.

[16] Jacob BA, Lefgren L. The impact of NIH postdoctoral training grantson scientific productivity. Research Policy 2011;40:864–74.

[17] Katz JS, Martin BR. What is research collaboration? Research Policy1997;26:1–18.

[18] Karro E, Kattel R. Coordination of innovating policies in thecatching-up context: a historical perspective on Estonia and Brazil.International Journal of Technological Learning, Innovation andDevelopment 2010;3:293–329.

[19] Medoff MH. Collaboration and the quality of economics research.Labour Economics 2003;10:597–608.

[20] Melin G. Pragmatism and self-organization research collaborationon the individual level. Research Policy 2000;29:31–40.

[21] Niedergassel B, Leker J. Different dimensions of knowledge incooperative R&D projects of university scientists. Technovation2011;31:142–50.

[22] Numprasertchaia S, Igel B. Managing knowledge through collabo-ration: multiple case studies of managing research in universitylaboratories in Thailand. Technovation 2005;25:1173–82.

[23] Pezzoni M, Sterzi V, Lissoni F. Career progress in centralizedacademic systems: social capital and institutions in France and Italy.Research Policy 2012;41:704–19.

[24] Ponomariov BL, Boardman PC. Influencing scientists’ collaborationand productivity patterns through new institutions: universityresearch centers and scientific and technical human capital.Research Policy 2010;39:613–24.

[25] Porac JF, Wade JB, Fischer HM, Brown J, Kanfer A, Bowker G. Humancapital heterogeneity, collaborative relationships, and publicationpatterns in a multidisciplinary scientific alliance: a comparative casestudy of two scientific teams. Research Policy 2004;33:661–78.

[26] Subramanyam K. Bibliometric studies of research collaboration:a review. Journal of Information Science 1983;6:35.

[27] Ubfal D, Maffioli A. The impact of funding on research collaboration:evidence from a developing country. Research Policy 2011;40:1269–79.

[28] Van Looy B, Landoni P, Callaert J, van Pottelsberghe B, Sapsalis E,Debackere K. Entrepreneurial effectiveness of European universi-ties: an empirical assessment of antecedents and trade-offs.Research Policy 2011;40:553–64.

[29] YnalvezMA,ShrumWM.Professional networks, scientific collaboration,and publication productivity in resource-constrained research institu-tions in a developing country. Research Policy 2011;40:204–16.