Modes of organizing biomedical innovation in the UK and US and the role of integrative and relational capabilities

A

iticc©

K

1

wckaeeant

f

0

Research Policy 36 (2007) 529–547

Modes of organizing biomedical innovation in the UK andUS and the role of integrative and relational capabilities

Jacky Swan a,∗, Anna Goussevskaia a,b, Sue Newell a,c, Maxine Robertson d,Mike Bresnen e, Ademola Obembe a

a Warwick Business School, University of Warwick, Coventry CV4 7AL, UKb Fundacao Dom Cabral, Centro Alfa, Av. Princesa Diana 760, Alphaville Lagoa dos Ingleses, Nova Lima, MG, Brazil

c Bentley College, 175 Forest Street, Waltham, MA 02452, Boston, USAd School of Business and Management, Queen Mary, University of London, Mile End Road, London E1 4NS, UK

e University of Leicester School of Management, University Road, Leicester LE1 7RH, UK

Available online 29 March 2007

bstract

Given that biomedical innovation involves intense collaboration across disciplines, occupations and organizations, a nation’sntegrative capabilities (the ability to move between basic science and clinical development) and relational capabilities (the ability
o collaborate with diverse organizations) have been identified as crucial. This paper deploys qualitative analysis of biomedicalnnovation in the UK and US to identify mechanisms influencing innovation at the project level through which these macro levelapabilities may have effects. From this a propositional framework is developed that helps explain the likely impact of suchapabilities for characteristically different kinds of innovation projects at the micro level.
2007 Elsevier B.V. All rights reserved.

; Organ
eywords: Integrative capabilities; Relational capabilities; Innovation
. Introduction

Biomedical innovation has been defined in variousays but here we see it as a process involving the

reation and application of scientific and technologicalnowledge to improve the delivery of human healthcarend the treatment of disease. This definition is broadnough to include new drugs, diagnostics and drug deliv-ry regimes for human use, but excludes purely animal,
gricultural and natural resource applications of biotech-ology (Rasmussen, 2005). In the biomedical domain,he potential for breakthroughs in science and technol-
∗ Corresponding author. Tel.: +44 2476524271;ax: +44 2476524656.

E-mail address: [email protected] (J. Swan).

048-7333/$ – see front matter © 2007 Elsevier B.V. All rights reserved.doi:10.1016/j.respol.2007.02.014

ization; Biomedical

ogy to radically change (and hopefully improve) medicaltreatments and diagnostic techniques, is high. However,exploitation of scientific breakthroughs for biomedicalinnovation is problematic, as witnessed by the high cost,duration and failure rates in product development (CMRInternational, 2006).

Even where scientific knowledge is validated, manypromising discoveries fail to reach the clinic, with asignificant number of failures occurring in early devel-opment phases of the innovation process (Hilton et al.,2002; Dopson, 2005). This is, in part, because biomed-ical innovations – especially more radical innovations –often cut across established professional, occupational
and organizational boundaries, and threaten to disruptexisting medical practice (Christensen, 2000). This sug-gests that the ability to combine and integrate knowledge(scientific, technological, commercial, clinical, regula-
mailto:[email protected]

dx.doi.org/10.1016/j.respol.2007.02.014

ch Polic
530 J. Swan et al. / Resear
tory) across a distributed array of professional groups,commercial organizations, public research organizations(PROs) and health organizations would be central tobiomedical innovation (Gibbons et al., 1994; Coombset al., 2003). Biomedical innovation processes have thusbeen described as typically non-linear or ‘interactive’,comprising complex, uncertain, high risk and iterativecycles of knowledge integration and networking acrossthese diverse groups (Powell et al., 1996; Dodgson et al.,2004). As Powell et al. (1996), note: “when the knowl-edge base of an industry is both complex and expanding,and sources of expertise are widely dispersed, the locusof innovation will be found in networks of learning,rather than individual firms”.

The structural features of networks (e.g. their den-sity, scope, strength of ties) linking PROs to commercialfirms have been subject to a good deal of analysis (Powellet al., 1996; Owen-Smith et al., 2002). This work hasshown how systematic variation in the composition ofnetworks across nations, and within regions, influencesthe ability to integrate scientific, clinical and commer-cial expertise (Owen-Smith et al., 2002; Owen-Smithand Powell, 2004). Of particular importance for biomed-ical innovation are linkages between scientific researchand commercial and clinical development. Owen-Smithet al. (2002) identify two macro-level capabilities thatinfluence these linkages: ‘integrative’ and ‘relational’capabilities. These refer, respectively, to the ability ofscientists to move back and forth between basic scienceand clinical development; and to the ability of organi-zations within an innovation system to collaborate withother, diverse organizations. In keeping with ‘nationalinnovation systems’ approaches, their work suggests thatthese capabilities stem from macro-level institutionaldifferences in the structure and operation of networksand, thus, differ across nations (Carlsson, 2002; Nelson,1993). In particular, Owen-Smith et al. (2002), throughan analysis of ‘upstream’ R and D linkages, have demon-strated that these capabilities are better developed in theUS institutional context than in Europe (including theUK). This, they suggest, accounts (at least in part) forthe US national advantage in biomedical innovation.

As yet, however, relatively little research has focusedon identifying and explaining the ways in which suchmacro capabilities relate to the process of innovationitself. Thus, whilst there is now evidence linking macrodata on networks to quantitative indicators of innovation(e.g. the development of patents or the diffusion of inno-
vations), there is also scope for qualitative research tocomplement this by investigating the processes throughwhich macro-level capabilities play out in the expe-riences of actual innovation projects, as perceived by
y 36 (2007) 529–547

the various stakeholders involved (Owen-Smith, 2003;Rogers, 1995). For example, how might the ability ofscientists to move back and forth between basic scienceand clinical development (i.e. integrative capabilities)facilitate an innovation process in a particular project set-ting? In answering such questions, we seek to identify thespecific mechanisms through which macro capabilitiesare likely to have effects on innovation projects. Withinnational institutional contexts, there are also clearly widevariations in the ways that innovation projects linkingpublic and private science are actually organized (e.g. asuniversity start ups, as development projects in biotech-nology firms, as Research and Development – R and D– projects in global pharmaceutical firms). Our researchalso seeks to identify different types of organizationalarrangement for biomedical innovation and to explorehow these variations in the organization of innovationmight mediate the impact of a nation’s integrative andrelational capabilities at project level.

Reflecting the absence of prior work, this study isexploratory, inductive and broadly based. This meansthat, whilst we can identify mechanisms at project levelthat link to macro capabilities and play an important role(in our cases) in shaping innovation processes, and haveexplored these across contexts (UK, US and differenttypes of project), our data does not allow us to conductdirect comparative analysis along dimensions defined ‘apriori’. That said our findings are suggestive of compar-ative differences and new theoretical propositions whichfuture, more deductively oriented work could follow-up.

Our argument is structured as follows. We begin byoutlining previous literature on the role of integrative andrelational capabilities across contexts and attempt to linkthis to innovation processes. We focus in our study onthe UK and US since, whilst these both have well devel-oped biomedical industries, previous work has found thatthey are characteristically different in terms of integra-tive and relational capabilities making (to put it crudely)the US context more supportive of biomedical innova-tion (Owen-Smith, 2003). However, it should be notedthat our study is focused on innovation processes—it isnot attempting directly to confirm or refute these earlier,macro findings.

We turn next to the nature of biomedical innova-tion processes and suggest that, as these nearly alwaysentail interactions and interdependencies across spe-cialist knowledge domains and organizations, a moredifferentiated framework is needed to understand how
such processes are organized. Thus, the first phase of ourresearch study involved an interview-based survey of keystakeholders in the US and UK, combined with reviewof existing literature, in order to develop a framework

ch Polic

tistslimafbSbbfrbdf

2c

cainviCipatstfdrtoa

ecchmiae

J. Swan et al. / Resear

o capture characteristically different modes of organiz-ng biomedical innovation projects. This framework isummarized and used as a basis for case selection. Wehen focus our analysis on detailed, longitudinal casetudies in order to identify those mechanisms at projectevel that appeared to play an important role in shap-ng innovation processes, and which could be related to

acro-level capabilities. In keeping with our researchims, the rationale for selecting characteristically dif-erent kinds of project was not direct case comparisonut to capture variation across contexts (Alvesson andkoldberg, 2000). Thus, if a particular mechanism coulde identified as relevant across contexts, then we coulde more confident of its explanatory value as a basis forurther research. In the concluding section, we attempt toelate the ways in which integrative and relational capa-ilities play out in particular kinds of project through theevelopment of a new propositional framework to guideuture research.

. Integrative and relational capabilities acrossontexts

The impact on innovation of national institutionalontexts has clearly been the subject of a significantmount of research (Whitley, 2003). We have chosenn our study to focus on the UK and US because theirational systems of innovation have been found in pre-ious work to be largely supportive of biotechnologynnovation (Casper and Kettler, 2001; Whitley, 2000;asper, 2000)—both have world class research facil-

ties and science bases and internationally recognizedharmaceutical firms, and both support entrepreneurialctivity and have active local markets in the supply ofechnology, scientists and know-how. Recognizing theseimilarities, there are also crucial differences that makehem interesting points of contrast. Clark (1987, 2003),or example, suggests that the UK and US have nationallyistinctive patterns of innovation and distinctive culturalepertoires or ‘styles of thinking’ amongst managers, andhe impact of professional and educational institutionsn the legitimation of knowledge has been found to varycross these contexts (Clark, 2000; Aldrich, 2000).

In the context of biomedical innovation, Owen-Smitht al.’s (2002) delineation of integrative and relationalapabilities – capabilities that link scientific research tolinical and commercial development – is particularlyelpful in understanding the institutional factors that
ay promote innovation. Integrative capabilities facil-
tate the translation of basic research into commercialpplications through the movement of scientists and theirnhanced labour market mobility (cf. Henderson, 1994).

y 36 (2007) 529–547 531

In contrast, relational capabilities facilitate innovationby, for example, supporting collaborative product devel-opment projects between PROs, biotechnology firms andpharmaceutical firms. The key elements in forming thesecapabilities are linkages across life science networks thatstructure national innovation systems. These linkageshave been found to be influenced by the macro institu-tional context, including political (e.g. policy initiatives,regulation), social (e.g. relations between scientists andtechnologists, the mobility of the scientific labour force)and cultural (e.g. values regarding academic participa-tion in commercial activities) factors (Owen-Smith et al.,2002).

The impact of integrative and relational capabili-ties on biomedical innovation has been demonstratedthrough macro-level data comparing the US and Europe(including the UK). Thus, Owen-Smith et al. (2002)find that the organization of upstream (early stage) Rand D in the US is qualitatively different to Europe,embracing more diverse constituents and knowledgesources, closer linkages between basic science andapplied science, and closer links between public andprivate organizations—in the form of dense region-ally clustered networks (mainly in Boston, Californiaand New York) among hospitals, dedicated biotechnol-ogy firms, large pharmaceutical firms, universities andresearch institutes. Like others, they argue that the UScontext has a stronger history of industry-university col-laboration in (R and D—Rosenberg and Nelson, 1994)with more, and more diverse, interfaces between publicand private organizations. This reflects national differ-ences in educational systems, career development andlabor markets (for example, greater, and more diverse,funding of universities in the US) and the greater move-ment of academics back and forth between PROs, firmsand research hospitals (Whitley, 2003).

With regard to integrative capabilities, these havebeen linked to national institutional differences in edu-cational systems, career development and labor marketmobility (Whitley, 2003). For example, in the US con-text, scientists are more able to move back and forthbetween the public and private sector without detrimentto their scientific careers, as compared to those work-ing in the UK (Owen-Smith et al., 2002; Powell et al.,1996; Mallon et al., 2005). This chimes with work whichsuggests that the US has a stronger ‘knowledge plus’ ori-entation, where joint goals of understanding and use ofbasic science and commercialization are more strongly
established (Stokes, 1997). This is seen to generate a USadvantage in terms of the ability to develop and exploitscientific knowledge for clinical development (Owen-Smith et al., 2002). In contrast, the relationship between

ch Polic
PROs and private firms in the UK is more distant: PROstend to be more specialized and there is a greater dis-tinction between basic research, on the one hand, anddevelopment, on the other. These findings concur withClark (1987) who observes greater ‘conflict with capital’in the UK educational system, with a more pragmatic ori-entation towards applied, or ‘how to’, knowledge in theUS, as opposed to the stronger value placed on knowl-edge and understanding for its own sake in the UK. This,he argues, generates a polarization in the UK between“academic thinking, which is often regarded as unnec-essary and impossible to digest and the rule of thumbempiricism which seems to have a firm grip in manysectors” (p. 223). Whilst Nowotny et al. (2001) notethe increased blurring of boundaries between knowledgetraditionally produced in university, government and pri-vate sector research organizations, this is also deeplypoliticized so that “the university appears simultaneouslyas capturing, but also captured” (p. 79).

With regard to relational capabilities, there is ampleevidence that individual firms, even large pharmaceu-ticals, do not possess all of the resources necessary tosuccessfully develop new therapeutics (Powell et al.,1996). Individuals and firms, therefore, need to collab-orate formally and informally to acquire the necessaryresources. In the biotechnology sector, ‘open’ channelshave been found to be particularly helpful in facilitat-ing opportunities for knowledge creation through theenhanced likelihood of ‘spillover effects’: that is, knowl-edge is more likely to ‘leak’ though more open channels(Owen-Smith and Powell, 2004; Murray, 2002; Kreinerand Schultz, 1993). Similarly, Salman and Saives (2005)show that, as well as being influenced by direct rela-tionships between organizations, innovation outcomesare attributable to informal, unpredictable relationshipsgenerated through indirect ties, which create access toexpertise beyond formal alliance partnerships. Other lit-erature has emphasized the importance of trust-based,informal networks for R and D (e.g. Liebeskind et al.,1996; Kreiner and Schultz, 1993). The US biomedicalsector has also been found to have better established rela-tional capabilities than in Europe reflected, for example,in the scope and density of networks—at least in certainregions.

This prior research is indicative of the ways in whichintegrative and relational capabilities might influenceinnovation processes in broad terms. For example, itsuggests that one important mechanism via which inte-
grative capabilities could influence innovation at projectlevel is through the career identities and values ofindividual scientists involved in project work, in par-ticular the extent to which they see goals of science
y 36 (2007) 529–547

and commerce as mutually acceptable. Similarly, rela-tional capabilities at the macro level (as indicated inprior research by the density and scope of networkties within regions and nations) would appear to berelated, at project level, to the ability of the organiza-tions involved in innovation processes to acquire andcreate relevant expertise, with informal networks beingespecially important in this respect. However, furtherresearch is needed to explore these mechanisms in detail.The value of understanding relationships across dif-ferent levels of analysis has also been observed byGittell and Weiss (2004) who note that, “frameworksfor analyzing organizational phenomena must be respon-sive to the dynamic and complex characteristics andinter-relationships between multiple levels of analy-sis that ‘real life’ situations reflect”. As already seen,an important aspect of these dynamics in the case ofbiomedical innovation concerns the ways in which net-worked relations and the combination of specialist formsof knowledge and expertise are actually coordinatedamongst the different parties involved in innovationprojects. In the analysis below, then, we begin by iden-tifying characteristically different modes of organizingbiomedical innovation projects (focusing on networkrelations and knowledge flows), before looking acrossdifferent kinds of project in the UK and US to draw outthose mechanisms that appear to be important in relatingmacro capabilities to actual innovation processes.

3. Methodology

The findings below are drawn from a 3-yearexploratory study of innovation in the UK and USbiomedical sectors aimed at identifying the factorsfacilitating and impeding innovation projects acrosscontexts. The UK and US contexts were selected inorder to achieve broadly defined variation (Alvessonand Skoldberg, 2000) as earlier work had suggested thatmacro level relational and integrative capabilities weredifferent across these two contexts (Owen-Smith et al.,2002).

The first phase was an interview-based survey with arange of individuals representing key stakeholder groupswho had significant experience of working in early-stagebiomedical innovation projects that could be describedas involving ‘systemic production networks’ (Alter andHage, 1993). Interviewees were initially identified viamembers of our project’s expert UK and US Scientific
Advisory Boards (SAB). Members of the SAB wereall experienced in managing biomedical innovation,and included serial entrepreneurs of biotech compa-nies, venture capitalists, academic scientists and policy

ch Polic

svTal

atUfiwnfOoiobiieWpwlpar

iUiesfie(aacaogtevm

(aLa


pecialists. From these initial contacts, additional inter-iewees were identified using a ‘snowballing’ technique.his kind of non-probability convenience sampling isppropriate when the research is exploratory and popu-ation parameters are unknown (Saunders et al., 2000).

We conducted 97 interviews (44 UK; 53 US). Inddition, 22 meetings were held to discuss further par-icipation as case studies for phase two (17 UK; 5S). Recognizing that there are regional variations,eld work focused on the Boston area in the US,hich has a concentration of biomedical-related orga-izations (including Harvard, MIT and Mass General,or example). In the UK, we concentrated on thexford–Cambridge–London triangle, which is also rec-gnized for its high level of activity and reputation fornnovation in the biomedical area. The primary aimf these first phase interviews was to gather rich, androadly representative, descriptions of experiences ofnnovation in the biomedical field from those directlynvolved. In particular, we wanted to establish the differ-nt ways biomedical innovation projects were organized.

hilst the dimensions we identified (organizational cou-ling and knowledge boundaries, see below) do resonateith earlier work (Alter and Hage, 1993), this ear-

ier work had not specifically considered biomedicalrojects, or the relationship between knowledge flowsnd networks in systemic production networks whichelational and integrative capabilities directly relate to.

The second phase comprised longitudinal case stud-es of innovation projects (six in the US and four in theK), offering exemplars of different ways of organiz-

ng biomedical innovation. The particular focus was onarly development (the move from discovery to the earlytages of commercialization) as the first phase identi-ed this as a critical point at which upstream scientificxpertise needed to interact with other forms of expertiseclinical, commercial, regulatory) and where integrativend relational capabilities would be expected to havesignificant impact (Owen-Smith et al., 2002). A key

riteria for case selection was, not only that the caseppeared to offer a good example of a particular modef organizing as identified by phase 1, but also thatood access to key stakeholders was offered, allowing uso conduct detailed longitudinal research including, forxample, allowing investigators non-participatory obser-ation of strategy and project meetings and access toeeting notes.Our approach was rooted in social constructivism

Kukla, 2000) and therefore a subjective rather thann objective epistemology was assumed (Denzin andincoln, 1998). Hence, we did not aim to compare caseslong fixed dimensions such as ‘success’ of projects,

y 36 (2007) 529–547 533

as we accepted that what counted as ‘success’ wassocially constructed by those involved. For example,in some projects, the acquisition of funding constituted‘success’ when this facilitated future development workand enabled progress to be made on projects. In oth-ers, clinical trial results which secured gateway FDAapproval served as an important proxy for success bythose involved. The length of time available to studythese projects (@30 months) was adequate to trace whatprogress (if any) had been made on projects and what fac-tors may have facilitated or hindered projects. Thus, oneach case visit (held approximately 6 monthly) projectteam members were asked to review whether progresshad matched their expectations in the prior period andalso to indicate their expectations for the forthcomingperiod. However, the time available was not sufficient toassess ‘success’ in definitive terms as the entire devel-opment process would take around 8 years on average(CMR International, 2004). This collective case studyapproach was aimed at facilitating interpretation anddeveloping qualitative insights into the early-stage devel-opment process (Alvesson and Skoldberg, 2000), bycomparing the similarities and differences provided bymultiple settings – in this case – projects (Tsoukas,1989).

Whilst access was negotiated via focal organizationsand/or individuals, the unit of analysis was the innova-tion process, as manifest in particular projects over theperiod of the research, not a specific firm, so intervie-wees spanned the different organizations involved. Thecases were selected on the basis of, first, the choice ofresearch topics and questions being posed (Stake, 1995)and, second, the possibility of capturing both historicand ‘live’ processes to inform the longitudinal analysis(Pettigrew, 1990). Thus, in all cases, activity relating tothe innovation process had been on-going for at least 2years and was projected to continue during the researchperiod. This permitted the collection of data providingcurrent, as well as retrospective, views of the innovationprocess. This case research was interview based, witha minimum of four fieldwork visits per case over the30-month period. On average, 14 interviews were con-ducted per case. Interviews were complemented withaccess to extensive documentary data (including com-panies’ reports, inter-partner correspondence, contractsand meeting minutes) and observational data (includingnon-participant observation of project team and advisoryboard meetings). The interviews in both phases were
recorded and transcribed. Detailed notes were taken andlater written up as a record of any meetings attended.
The first phase data were coded and analyzed usingNVivo software and the ‘memoing’ technique (Glaser,

ch Polic

ment and finances), where discussed, was negotiatedthrough largely informal means. Thus, management insuch projects was decentralized and vertical dependencyon centralized resources was low (Alter and Hage, 1993).


1978; Miles and Huberman, 1994). Interviewees wereasked to recount stories of their own experiences ofbiomedical innovation projects and to talk about thefactors and critical events that had influenced theseprojects. In the analysis, we focused on identifying theinstitutional level factors that interviewees discussed aseither facilitating or impeding the innovation processdescribed. This included, for example, access to finance,availability of expertise and access to technology. Weused this analysis to consider how institutional-level rela-tional and integrative capabilities were playing out indifferent innovation projects.

In phase two, recognizing the complexity of the cases,the research for each case was jointly conducted by atleast two investigators in order to co-develop interpreta-tions in real time throughout the research period. Datawere managed, coded and recoded, using NVivo, the ini-tial coding structure having been developed from phase1. On completion of fieldwork, detailed case descriptionswere produced by the investigators on each case (on aver-age 10,000 words) containing primary data (quotes frominterviews, inserts from documents, etc.), and structuredaccording to particular themes. All case descriptionswere then content analyzed by the whole team in orderto identify the processes and project dynamics that facil-itated or impeded project progress (however, defined byproject participants), relating each to either the abilityof scientists to move back and forth between basic sci-ence and clinical development (i.e. linked to macro-levelintegrative capabilities), or to the ability of organiza-tions within an innovation system to collaborate withother, diverse organizations (i.e. linked to macro-levelrelational capabilities). We refer to these processes anddynamics for convenience as ‘mechanisms’ and theanalysis below explains why and how we see these mech-anisms as related to macro capabilities.

Initially, 12 mechanisms were identified, whichrelated to integrative capabilities and 17 mechanisms,which related to relational capabilities. Further inductiveanalysis of these 29 mechanisms by each team mem-ber and then by the whole team led to a clustering ofthese into 8 primary mechanisms identified from the 10cases. The major criteria by which interpretive researchis assessed – trustworthiness, credibility, confirmabilityand transferability – were therefore addressed by inde-pendent verification across the 10 case studies and the 6researchers on the project (Denzin and Lincoln, 1998).In addition, over the duration of the research, five SAB
meetings were also held. At each of these meetings,major findings to-date were discussed, including, at thefinal meeting, a detailed discussion to test our analysisof how the mechanisms we identified linked to macro
y 36 (2007) 529–547

capabilities (most SAB members having had experienceof both US and UK contexts). We explicitly sought fromour advisory board members, feedback as to how far thecategories, models and frameworks that we were devel-oping, resonated with their own experiences in the field.Their validation of our analyses provided further supportfor our findings.

4. Modes of organizing biomedical innovation

From the first-phase analysis, coupled with literaturereview, two broad dimensions were found to be help-ful in characterizing innovation projects. We refer tothese here as ‘organizational coupling’ and ‘knowledgeboundaries’. These dimensions are depicted in Fig. 1and described briefly next. Details on the derivation ofthese dimensions from the phase 1 data can be foundelsewhere (Swan et al., 2005).

Organizational coupling refers, then, to the organi-zation and management of collaborations and networkties between partners. Variation along this dimensionranged from networked and loosely coupled modes,to more hierarchical and tightly coupled modes (cf.Alter and Hage, 1993; Hardy et al., 2003; Owen-Smithand Powell, 2004). In the former, innovation projectswere pursued by partners joined in a loosely couplednetwork of organizations, with work being conductedacross several organizations, each of which had signif-icant autonomy in the management of their own work.Allocation of resources to tasks (including people, equip-

Fig. 1. Modes of organizing biomedical innovation.

ch Polic

CtdeicoavlP

tlwchi‘Labds

itftktOcakc(2mipk

tnc(e(ses


ommitment to tasks was based primarily on mutualrust and obligation to a super-ordinate goal (e.g. pro-ucing a breakthrough new treatment) and promises ofxpected return, rather than on detailed or formally bind-ng contracts. Where formal contracts did exist, theseentred broadly on mutual obligations and the allocationf future financial gains (e.g. split of revenues gener-ted). Knowledge flows could be described as occurringia relatively ‘open channels’, characterized by diffuseinkages and ‘knowledge spillovers’ (Owen-Smith andowell, 2004).

In contrast, in tightly coupled modes, most innova-ion activity was carried out and coordinated within aarge focal firm but with clearly identified parts of theork (e.g. manufacturing, clinical trials) being formally

ontracted to other parties. Management was relativelyierarchical and there was high dependency on central-zed resources, knowledge flows occurring via closedconduits’, or pipelines (Owen-Smith and Powell, 2004).egally binding contracts existed to secure deliverablesnd protection of IP and detailed financial responsi-ilities and returns as well as the allocation of tasks,eadlines, roles responsibilities, risk management ando forth.

The knowledge boundaries dimension relates specif-cally to ways in which knowledge was combined acrosshe different specialist domains involved and rangesrom ‘high’ to ‘low’. It is important to emphasize herehat, in keeping with a constructivist approach, we sawnowledge domains not as purely cognitive, but as tiedo boundaries of specialized practice (Carlile, 2004;rlikowski, 2002). Nearly, all biomedical projects dis-

ussed in phase 1 deployed multidisciplinary teamsnd so spanned cognitive domains—different forms ofnowledge (scientific, commercial, clinical, regulatory)learly needed to be brought together in developmentHoward-Grenville and Carlile, 2006; Gay and Dousset,005). The important issue here, then, was not simplyultidisciplinarity, but whether or not the work involved

n innovation projects actually demanded new ways ofracticing across these domains that meant that existingnowledge/practice boundaries had to be overcome.

High knowledge boundaries arose in situations wherehere were high novelty areas involved, where medicaleed was ambiguous and/or contested and where impli-ations for clinical practice were difficult to forecastCarlile, 2004). In such projects, the intensity of knowl-dge sharing between those involved in upstream science
e.g. scientific research) and those involved in down-tream application (e.g. clinical practice) was greater. Anxample of such a situation was the development of tis-ue engineered products, which required the progressive
y 36 (2007) 529–547 535

transformation and blending of knowledge, expertiseand artifacts across existing disciplines and occupations(Bock et al., 2003). High knowledge boundaries alsoarose in situations where the parties involved needed tocombine their expertise, in order to articulate new knowl-edge and practices, but had not worked together beforeor, indeed, with others in those particular disciplines oroccupations—for example, when each organization hada ‘piece’ of technology, or IP, and only the combina-tion of the pieces would allow the development of theproduct. These kinds of situation engendered ‘pragmaticboundaries’ (Carlile, 2004), meaning that alignment ofprofessional interests and development of shared expec-tations among stakeholders was crucial.

The combination of these two dimensions provideda novel framework for understanding characteristi-cally different modes of organizing innovation projects,depicted in the four quadrants of Fig. 1. In the nextsection, these different kinds of innovation project areillustrated through case descriptions. Whilst our analy-sis, as seen, drew from all 10 cases, we have elected,due to pressures on space, to focus here on 6 to illus-trate these (3 UK, 3 US). We also focus on the projectsrepresented by quadrants I, III and IV. Whilst we col-lected ‘vignettes’ of quadrant II-type projects from thefirst phase, our cases studies were not concentrated herebecause these examples were typically centred on moretraditional, incremental innovation processes containedwithin the R and D departments of large global pharma-ceutical firms, with technology being either developedin-house, bought or licensed. Whilst these kinds of inno-vation project are interesting in their own right, they areconsidered less relevant to our main area of interest (i.e.the link between innovation processes at project leveland a nation’s capabilities).

4.1. Quadrant I cases

Quadrant I, characterized by loosely coupled orga-nizational relationships and low knowledge boundaries,was typically populated by small early stage spin-offcompanies started by academics with entrepreneurialinterests and/or commercial experience. There was highdependency on the parent university or PRO (Powell etal., 1996) and multiple sources of funding were soughtfor facilities, equipment, consumables and specialistexpertise (e.g. from research grants, venture capitalinvestors and larger biotechnology or pharmaceutical
firms). However, the development of the science andtechnology required relatively low levels of knowledgeintegration across those groups involved, knowledgebeing produced in sub tasks performed more or less

ch Polic
separately and pooled subsequently (Thompson, 1967).Innovations here did promise significant improvementsin treatment, but were minimally disruptive to existingmodes of treatment delivery and so did not depend cen-trally on combining the production of the science ortechnology with the particular expertise of clinicians thatwould likely deploy it (Christensen, 2000).

NewPharma was a start up based in the US developinga new therapeutic for a neurological disease based on thefounding academic’s discovery in his Hospital lab. Thecompany was established with personal funding fromthe founders, the scientist who made the original discov-ery and an ‘entrepreneurial’ partner who took the roleof CEO. The three had previous experience of launch-ing a company together. The founding scientist did nottake an operational role in the company but served asa member of the management and scientific advisoryboards. Whilst he viewed himself as a ‘risk taker’, therewas an express awareness of the limits to which he couldengage in commercialization activities and still retain hisacademic position.

NewPharma obtained a licensing option from theHospital (who held the patent) with a view to followingone of two options: license their compound to a phar-maceutical company or develop it within NewPharma.They delivered presentations to senior pharmaceuticalexecutives and research scientists with links to pharma-ceuticals. Where individuals appeared to be interested,they were invited to help NewPharma negotiate licens-ing deals with pharmaceutical companies with whichthey had connections, in return for shares in the com-pany. These ‘deal breakers’ were essential to get accessand credibility with potential investors in pharmaceuticalcompanies. Simultaneously, there was a need for valida-tion of the discovery by academic peers. This was to beachieved by submitting a paper to a reputable journal.During the period of the study, NewPharma was unsuc-cessful in getting investment. Based on feedback theyreceived from the presentations that expressed concernthat a single discovery was not a sufficient platform toattract investment, NewPharma continued to search fora partner with additional patents to build a platform.

SampaTech was a small company developing noveltherapeutics for hepatitis. It was founded by two sci-entists from a leading UK university, who developedthe basic technology in collaboration with a large phar-maceutical firm (that subsequently withdrew from theproject) and another university. By early 2005, the com-
pany had acquired two rounds of seed funding: one fromone of the universities and another from a donation. Theintention was to develop the lead project, out-license itand use the royalties for further developments. SampaT-
y 36 (2007) 529–547

ech had a loosely coupled management structure, relyingon a number of part-time executives and administrativepersonnel, coordinated by one of the two founding sci-entists (who continued their academic activities at theuniversity). Although, the academic Director had a his-tory of interaction with industry, via involvement on theadvisory boards of biotechnology companies, the com-pany still relied heavily on the technology transfer office(TTO) of the university, which provided access to biotechcompanies and the venture capital (VC) community.

SampaTech’s first CEO had formerly worked as amanager in a large pharmaceutical firm but was laterconsidered to lack the scientific background required tosecure further VC funding. A new CEO was appointed,with the help of the technology transfer office, who had‘the right’ profile. She had previously started severalbiotechnology companies thereby developing a reputa-tion as a ‘serial entrepreneurial scientist’. This change ofCEO resulted in a major refocusing of the organizationstrategy, which was initially geared towards the devel-opment of three technology platforms, thereby requiringextensive funding. By reassessing the strategy, the newCEO narrowed the area of development to therapeutics,thereby generating external pharmaceutical interest inlicensing the lead product.

4.2. Quadrant III cases

The projects in quadrant III, like those in quadrant II,were usually led by larger biotechnology or pharmaceu-tical firms. However, whereas quadrant II, projects wereaimed at incremental improvement of currently availabletreatments, quadrant III describes cases where the com-panies ventured into highly innovative areas where thedevelopment of breakthrough technologies placed highdemands on the focal organization to collaborate withbasic researchers and, given their potential to disruptmedical practice, also required constant interaction with(and input from) end users (health professionals) andregulatory bodies. These innovation projects were man-aged centrally, based on formal contractual agreementswith smaller companies and specialist research organiza-tions in ways reminiscent of supply chain interactions.However, whilst inter-organizational relationships andfinancial resources were tightly controlled by the cen-tral organization, knowledge was widely distributed andneeded to be brought together across different domainsof practice.

AmericanBio was a relatively large US biotechnol-ogy company. ELBOW was a product for cartilagerepair based on tissue engineering technology. TheELBOW project was conducted by a multifunctional

ch Polic

cnFsAtBmpsrc

oposetdtwtpsecastaoapcdtr

nAaeAfpasiawd


ore team. Importantly, AmericanBio had a strong inter-al regulatory group responsible for interaction with theDA. This regulatory expertise was crucial for ELBOW,ince initially there was no regulatory framework andmericanBio were able to shape the regulations during

he approval process for their first generation product.esides the challenges involved in regulation, sales andarketing had also proven costly and complex as the

roduct disrupted established ways in which orthopedicurgeons (the main users) practiced. Thus, developmentequired a significant degree of interaction with the userommunity.

AmericanBio was developing a new generationf ELBOW. Setbacks with its internal developmentrompted the decision to search for external technol-gy that could help ‘leapfrog’ the project through earlytage clinical trials. The company had a special inter-st in EU companies, because the lack of regulation ofissue engineered products in Europe meant that patientata on the technology was available that might help easehe progression to clinical trials in the US (interestingly,hile AmericanBio had played a major role in helping

o shape US regulation, which made it difficult for com-etitors to enter the market, they now had to face thoseame regulatory barriers to develop their own new gen-ration product). AmericanBio identified and acquired aompany in Europe, which had the technology needednd initial clinical data. One important criterion in theirelection was a match in terms of organizational cul-ures. AmericanBio had earlier carried out diligence onnother company, but decided not to acquire it becausef ‘significant organizational differences’. Following thecquisition, meeting timelines in product developmentroved difficult, one of the reasons being that the clini-al data was not as ready for FDA approval as expected,espite the fact that AmericanBio had conducted a veryhorough due diligence process and, according to oneespondent, ‘knew where to look for dead bodies’.

Body was based in the UK and developed its busi-ess based on expertise in producing human antibodies.NTIBODY-2 is a project concerning development oftherapeutic antibody for inflammatory disease. Duringarly stage clinical trials, Body decided to out-licenseNTIBODY-2, because they did not have the resources

or later stage trials that would necessitate large patientopulations. The anticipation of the future partneringrrangement had a significant impact on Body’s deci-ions about the design of their early stage trials. A major
ssue was in estimating what level of risk to accept forn expected return from the partnering arrangement thatas yet to be agreed. The more robust the efficacy evi-ence for ANTIBODY-2, the better the financial deal
y 36 (2007) 529–547 537

with a partner; but more trials would entail greater costsand there was always a risk that the results might notprove favourable.

Pharmaceutical companies who expressed an interestin partnering were given an information pack about theproject produced by Body’s project team. Whilst thesefirms were conducting their own due diligence projects,they were also expected to deliver ‘capability presenta-tions’ outlining the sorts of trials they planned to conductto progress the project and the resources they wouldcommit. This was because Body wanted to be confi-dent that the partner would have sufficient knowledgeand resource to move forward with ANTIBODY-2 andtake it to market quickly. Within Body, there was a lot ofinformal personal networking across the various teamsinvolved and with outsiders—as one respondent noted‘everybody in this industry knows everybody’. However,the decision was made to exclude from partnering dis-cussions anyone involved in existing relationships withthe candidate companies. In this way, formal networkswere emphasized and information management duringdue diligence was deemed crucial.

4.3. Quadrant IV cases

Quadrant IV also contains projects in highly novelinnovative areas that could break regulatory grounds andwere likely to disrupt healthcare practice (Christensen,2000). The novelty of the technology, or combina-tion of technologies, typically generated an informalinter-organizational web of smaller companies and col-laborating PROs. One practitioner observed that these‘sexy technologies’ created an ‘aura of attraction’ thatdrove interest and collaboration. These innovationstypically depended on highly networked individualsto orchestrate loosely coupled, decentralized projectsand innovation relied heavily on the co-production ofknowledge across varied domains of specialist practice.Reflecting this, tasks could be described as fully, orreciprocally, interdependent, where the sub-tasks con-tinuously interacted because the knowledge, outputs anddecisions from one had a direct impact on the oth-ers, and rewards were groups-based (Thompson, 1967;Wageman, 1995).

DiagnosticLabs was a small US company special-izing in diagnostic assays that initiated a developmentproject to transform them into a ‘theragnostic’ com-pany, combining diagnostic and therapeutic products.
The logic was that availability of a targeted drug wouldincrease the market for their diagnostic and vice versa.This project was championed by their recently appointedCEO, who had a reputation for managing successful

ch Polic
biotech companies. The choice of a specific disease as aninitial area for development, characterized by a high mor-tality rate and no approved treatment, also reflected med-ical need. The new project built upon an existing diag-nostics kit for this disease, which was being developedby DiagnosticLabs through collaboration with academicpartners. DiagnosticLabs lacked clinical trials and regu-latory expertise and so, via the CEO’s personal networks,formed an alliance with Bioclinical, a company special-izing in clinical trials consulting and services, whichprovided a dedicated team to lead the clinical trials.

The CEO used her personal connections to identifya company TherapeuticCo that held IP for the matchingtherapeutic. She originally believed that this IP was sup-ported by sufficient pre-clinical data to allow the projectto go straight into clinical trials. Bioclinical conducteda due diligence assessment of TherapeuticCo’s IP on a‘good will’ basis and concluded that the preclinical dataavailable would not be sufficient to gain FDA approvalfor clinical trials and to convince VCs to provide theinvestment needed. TherapeuticCo was not interested inmaking additional investments in a non-core area anda newly appointed CEO at the company did not want todedicate further time to the project. In addition, Diagnos-ticLabs’s owner decided to sell the company, so haltingnew investment. On top of this, there was a break downin the relationship between the CEOs of DiagnosticLabsand Bioclinical, as the Bioclinical team realized thatDiagnosticLabs was pushing for VC money for their ownuse and not for the project alliance as a whole. In part asa result, the development project was abandoned.

NewTissueCo was a spin-out company from TEC—aleading tissue engineering research centre in the UKbased at a university hospital. It had preferential rightsto exploit TEC’s technologies to make ‘scaffolds’ onwhich to grow stem cells and, potentially, new organs andbones. This field is extremely novel with potential appli-cations in the long term. NewTissueCo was establishedby two highly regarded and experienced scientists withthe help of the university’s business development officeand seed corn funding. Although, the original intentionhad been to find funding for the commercialization oftissue-engineered products, the lack of VC interest in thislonger-term vision led to a shift in emphasis towards theexploitation of existing IP through licensing deals withbiotech companies that would enable the generation ofrevenues to fund other, riskier projects.

Originally, the company was managed by two former
executives of large pharmas and operated as a virtualcompany employing a small number of part-time con-sultants. By 2004, a new CEO was appointed who hadextensive experience in the relevant scientific and com-
y 36 (2007) 529–547

mercial fields, including business start-ups. Recently,NewTissueCo had established a commercial licensingdeal with a larger company, BiotechCo, and had alsoattracted further VC funding. The company was alsoinvolved in a number of research collaborations withTEC scientists, involving grant applications to variousUK government and international funding bodies. Withthese developments, the new CEO was able to recruitnew full-time staff and start investments in productionfacilities to support the licensing deal.

The continued engagement of the lead ‘star scientists’involved in setting up NewTissueCo had important rep-utational and social capital effects, as well as importantpractical and political effects in keeping the companywell connected to leading scientific teams and attuned tothe politics of the host university. Close internal connec-tions within the university also lent scientific credibilityto dealings with external commercial partners—creatinga symbiosis between scientific and commercial interests.Importantly, networking not only involved developingrelationships with and through the principal scientists,but also with researchers working within their teams,including Ph.D.s, who could potentially assist with gen-erating new IP and commercialization.

5. Mechanisms linking innovation processes tomacro capabilities

Tables 1 and 2 summarise the mechanisms found inour cases to be important in enabling the innovationprocess for different kinds of project (recognising, as dis-cussed, that what counted as ‘success’ was interpreteddifferently across projects), grouped in our analysisterms of how they related to integrative or relationalcapabilities.

5.1. Integrative capabilities

Firstly, our findings echo earlier work by highlight-ing the importance of access in projects to individualswho work ‘at the interstices’ of science and commercein order to acquire relevant knowledge and expertise andto build the skills base (Powell et al., 1996; Murray,2002; Casper and Murray, 2005). For example, in Diag-nosticsLabs, the opportunity to develop a ‘theragnostic’was identified by the CEO—an experienced biomedicalentrepreneur who also had close networks with PROs.Other research suggests that this mechanism is shaped
by a nation’s integrative capabilities, which reflect dif-ferences in labour market institutions. For example, inher study of innovation in tissue-engineered cartilage,Murray (2002) found that the commercialization of sci-

J. Swan et al. / Research Policy 36 (2007) 529–547 539

Table 1Mechanisms linked to integrative capabilities

Mechanisms Examples from cases

(1) Access to people working at interstices to acquire knowledgeand reproduction of skills base

SampaTech: Reliance on TTO network to establish commercial contacts

NewPharma; Reliance on ‘deal breakers’ to help commercialize thepotential productNewTissueCo: Reproducing the scientific/commercial through training andemploying Ph.D. studentsDiagnosticLabs: Linking to the academic community throughcollaborations to develop IP and to conduct clinical trials

(2) Establishing scientific and commercial credibility in order toensure funding through partnering, VC or research funds (asthere are no ‘centralized’ resources available)

SampaTech: Selecting a CEO with the right profile in order to sustainconnections and dialog with pharmaceutical companies

NewPharma: Persistently trying to publish in Science in order to ‘validate’science in the eyes of potential investorsNewTissueCo: Importance of the role of the scientific founders and hostuniversity in providing scientific credibilityDiagnosticLabs: CEO has credibility within the VC community based onprevious entrepreneurial experiences

(3) Symbolic figureheads SampaTech: No figurehead involved and so the company relied heavily onthe TTO to move the company forwardNewPharma: No figurehead scientist involved, which acted as a limitingfactorNewTissueCo: Leading scientist’s vision, commitment and personalexperiences played a powerful role in pushing forward commercializationDiagnosticLabs: No figurehead involved implying that credibility restedwith the CEO

(4) Career perceptions and professional values in relation tomotivation to engage with innovation commercialization activity

SampaTech: Scientists thinking of themselves as scientists, puttingaltruistic reasoning before commercial gain, thus constraining commercialactivityNewPharma: Purely commercial interests are perceived as valid in theirown right; entrepreneurial characteristics apparentNewTissueCo: Blending of scientific, clinical and commercial orientationsin the professional identity and career choices of key individuals workingfor the companyDiagnosticLabs: Medical and commercial objectives are merged together.Proposed ‘theragnostic’ would allow both diagnosis and treatment of aconditcreate

eocstsi(ormcHs

nce into new medical treatments depends crucially onverlapping, but distinctive, scientific and technologi-al networks at the institutional level. Connections tocientific networks, via key individuals who work athe interstices of such networks, provide ‘knowledgepillover’ effects that shape technological progress andnfluence firms’ abilities to develop intellectual capitalZucker et al., 1998). This finding echoes earlier researchn the important role of ‘boundary spanners’ who brokerelationships across networks (e.g. scientific and com-
ercial) and facilitate the transfer of knowledge across
ontexts (Tushman and Scanlan, 1981; Carlile, 2002).aving access to individuals who work at the inter-

tices of science, commerce and the clinic also played

ion that had no previous treatment and would, at the same time,a bigger market by coupling sales of diagnostics with therapeutics

an important role in increasing the ‘absorptive capacity’of our case firms, by improving their ability to recognizenew information, assimilate it and apply it to commer-cial ends (Cohen and Levinthal, 1990). There were avariety of ways in which our case companies gainedaccess to people working at the interstices of networks:through establishing links with academics (as in Diag-nosticLabs), by enlisting ‘deal breakers’ to participate innegotiations with potential partners (as in NewPharma),by training Ph.D.s who could retain links with research
groups and, at the same time, assist in commercializa-tion activities (as in NewTissueCo) and, where PROswere involved, relying on a university technology trans-fer office to access networks (as in SampaTech).

540 J. Swan et al. / Research Policy 36 (2007) 529–547

Table 2Mechanisms linked to relational capabilities

Mechanisms Examples from cases

(5) Alignment of interests and expectations Body: Conducting two-way due diligence with potential partners to assesscapabilities and resources but also to negotiate interests and generate mutualunderstanding about expectationsAmericanBio: In deciding which company to acquire, they looked for a‘cultural’ match, which was decisive in choosing between the two candidates.Thorough due diligence, however, was not sufficient to know beforehand ifclinical data was good enoughNewTissueCo: Sought a licensing deal and investment from companies whoseproducts, knowledge base, experiences and capabilities related closely to, butat the same time complemented, their ownDiagnosticLabs: One way due diligence did not explore alignment of interests.As a result, therapeutics company was not interested in further investments inimproving their IP. Further misalignment of interests occurred with thedecision of the owner to sell the company

(6) Building upon existing networks to generate resources andsustain more risky and long term projects

Body: Company has a large number of diverse collaborations ranging throughR and D alliances to IP licensing, that they drew upon to identify potentialpartners to take the product through to developmentAmericanBio: Use clinicians from their existing network to promote theproduct by publishing results and increasing the body of experience andpatientsNewTissueCo: Using one business deal to leverage others becomes animportant way of enhancing commercial credibility and so building thecompany securing resources for long term ‘core’ projectDiagnosticLabs: Diagnostics company got involved in the project throughpersonal networking of its CEO. Based on the same relationship, free duediligence was conducted on the therapeutics IP

(7) Using networks to shape regulations and ensure approval Body: Regulation expertise was considered important in selection of thepotential partner, as Body would rely on it to ensure approvalAmericanBio: Company has a regulatory group dedicated to interaction withFDA. They were able to shape the regulatory framework for the firstgeneration product. Next generation product approval is also being developedthrough interaction with FDANewTissueCo: Founding scientists play an important role as opinion leadersand advocates for the (emerging) discipline in scientific, business, political andpublic policy arenasDiagnosticLabs: Using its connection to bioclinical to access regulatoryexpertise

(8) Product ‘magnets’ Body: Therapeutic in development is for a well recognized indication meaningrelatively known path to commercializationAmericanBio: ELBOW was the first product of its kind on the marketNewTissueCo: ‘Revolutionary’ nature of the work they are doing provides a

cus foriagnosti

foD

Another crucial mechanism was having members ofprojects with scientific and commercial credibility inorder to attract VC investment and/or major researchfunds. In NewTissueCo and SampaTech, for example,the choice of CEO with the ‘right’ scientific profile and
track record with investors was critical. This credibil-ity appears to be easier to establish in the US contextand particularly in the Boston region (where our casesfocused) where there is a strong history of successful
research, but also hampers commercialization effortscLabs: The ‘theragnostic’ concept was new and developed over time

biotechnology ventures and VC investment (McMillanet al., 2000). Whilst not stressed in research on inno-vation and networks, this dual orientation does resonatewith research on how VC operates. Zider (1998), forexample, found that the reputation of entrepreneurs plus
their business track record and ‘presentability’ to outsideinvestors, is as, if not more, important in attracting fund-ing as their knowledge base or particular ideas; this heargues helps to explain US superiority in VC markets.

ch Polic

Amlf

inNpcftbtfisdgvcvafocfimaiirttcsfi

bmssclsplliTwpb


ccording to Zider (1998, p. 138): “Many entrepreneursake the mistake of thinking that venture capitalists are

ooking for good ideas when, in fact, they are lookingor good managers in particular industry segments”.

Another way of establishing credibility for investments via the scientific publication records of the biotech-ology firm’s scientific team (Deeds et al., 1997). InewPharma, for example, those involved believed thatublication in a reputable scientific journal was cru-ial for public validation. Furthermore, the affiliationsounders had with other organizations were also impor-ant for increasing credibility. This finding echoes worky Higgins and Gulati (2003) who demonstrated thathe affiliations that senior managers of biotechnologyrms held with other organizations had an importantymbolic value for prestigious investment banks wheneciding whether to support initial public offerings. Thereater the prestige of the relationships, the higher thealuation – and hence private funding – a firm couldommand and so the greater the likelihood of an inno-ation reaching development. Across our cases, suchffiliations appeared to be easier to establish in the US,or example, with academic scientists more frequentlyn the Scientific Boards of prestigious pharmaceuti-al or biotech companies. This is consistent with thending that integrative capabilities encouraging move-ent across scientific, clinical and commercial domains

re stronger in the US (Owen-Smith et al., 2002). Thedea of a firm’s reputation-building practices provid-ng legitimacy for investors is also supported in recentesearch by Nicholson et al. (2005), who demonstratedhat biotechnology firms took a substantial discount onheir first out-licensing deals with large pharmaceuti-al firms, which was then quickly recouped throughignificantly higher valuations from VC at subsequentnancing rounds.

The importance of reputation building and the sym-olic value of affiliations relates closely to anotherechanism identified in our cases, which was the

ymbolic role played by ‘Figureheads’—usually leadcientists with international reputations and commer-ial and/or clinical experience. Such individuals, wheninked closely to projects, played a key role in mobilizingupport amongst diverse users and clinicians. For exam-le, the lead scientist in NewTissue embodied (quiteiterally, having been a former patient himself as well as aead clinician) the overall vision of the project, symboliz-ng its clinical and scientific significance to user groups.
he importance of ‘star scientists’ has been noted else-here, but usually in terms of the role such individualslay in linking commercial activity to the knowledgease of academe (Zucker et al., 1998). Whilst this was
y 36 (2007) 529–547 541

important (and discussed above), here we use the term‘Figurehead’ to highlight also their symbolic and motiva-tional effects. We did not observe as many ‘figureheads’in our US cases, nor was their importance emphasized asmuch by interviewees, suggesting that this mechanismmay be particularly crucial in the UK context in bridgingwhat were often more polarized values and perceptionsof those engaged in academic, clinical and commercialpractices (Mallon et al., 2005).

This leads to the final important mechanism, whichconcerned career perceptions and professional values ofthose working in PROs. There were two related issueshere. The first concerned the extent to which scien-tists saw their careers in both scientific and commercialterms—that is, as ‘entrepreneurs’ or, in Mallon et al.’s(2005) terms, as ‘strategic opportunists’. In Sampat-ech, for example, scientists pursued commercializationfor ‘altruistic’ reasons as a means to develop science,whereas in New Pharma the lead scientist described him-self as ‘a risk taker’ and ‘entrepreneur’. Other researchhas indicated that entrepreneurial values may be lesswidespread in UK academe. In a qualitative survey ofUK scientists in PROs, Mallon et al. (2005) found thatthe majority had an over-riding sense that obvious com-mercial ambition was not quite acceptable within publicsector science and so did not incorporate this into theircareer planning. Only a third – described as ‘strategicopportunists’ – were prepared to consider a move from‘the bench’ to a commercial career, but most of this grouphad become aware of the opportunities because they hadprevious experience of working outside the public sector.

The second issue concerned the extent to which sci-entists perceived their professional values as scientistsand clinicians to be aligned with, or potentially com-promised by, the pursuit of commercial activity. In ourcases, the actual movement of scientists from academeto commerce was rare. In all the cases, where PROswere involved, the lead scientists remained as scientists,and recruited other people to push forward commer-cialization. However, there were important differences,reflected in the professional values of scientists acrossUK and US contexts. So, in NewPharma and Diagnos-ticLabs, for example, commercialization was seen asmeaningful in its own right, both as a means to clini-cal improvement and as a way of generating personalrevenue for the scientists. In contrast, in SampaTechand NewTissueCo, the major motivation was to ‘makea real difference to patients’ lives’ and to use this activ-
ity to help fund continued scientific development, not tomake their personal fortune. This suggests that, at leastin our cases, the mobility of individuals’ careers acrossboundaries of science and commerce emphasized else-

ch Polic
where (Casper and Murray, 2005) was less significantin incentivizing (or de-incentivizing) commercializationthan the professional values that individuals attachedto their work. One of the problems in the UK is thatindividuals’ career perceptions are still quite polarizedand scientific entrepreneurship less strongly advocated(Clark, 2000; Turpin and Deville, 1995; Mallon et al.,2005).

Finally, our analysis suggested that these mechanismsrelating to integrative capabilities were more crucial inprojects characterised by loose, rather than tight, orga-nizational coupling. This can be explained in terms oftwo major impacts of a nation’s integrative capabilities.First, such capabilities have an impact on the motiva-tion of scientists to engage in commercialization, movingbetween academia and industry and connecting basic Rand D. Thus, mechanisms 1 and 4 were important waysin which this motivation was facilitated. They helpedto orchestrate loose networks characterising projects inquadrants I and IV and made possible different kinds oftransactions (e.g. between key scientists, entrepreneursand investors). Second, such capabilities have an impacton the ability to secure resources (e.g. from researchfunding, VC and through partnerships) needed to moveinnovation forward in situations where these resourcesare not available from and cannot be centrally controlledby one organization. Mechanisms 2 and 3 were impor-tant in this respect. In tightly coupled innovation projects,there was less need to rely on such mechanisms becausethe R and D was controlled from within larger biotech-nology or pharmaceutical firms, which deployed internalresources and were populated by scientists who, giventhat they worked in a commercial environment, weredirectly incentivized to commercialize basic research.In this way, the analysis indicates that a nation’s inte-grative capabilities might be expected to have relativelystronger effects on projects in quadrants I and IV. Inter-estingly, these mechanisms were not simply focused onknowledge integration but, rather, were most importantin terms of building the relationships across the aca-demic/commercial communities in order to establish thereputation and claims to knowledge required to attractinvestment.

5.2. Relational capabilities

Turning to relational capabilities, these concern howeasy/difficult it is in the macro institutional context
to establish and maintain collaborations across diverseorganizations. At project level, alignment of interests andexpectations was a crucial mechanism in our cases andcan be linked to relational capabilities. Key to achieving
y 36 (2007) 529–547

this was the due diligence process. Where due diligencewas two-way (conducted by each party involved), as inthe Body case, projects were able to balance partnerinterests and commitments. Moreover, once a collabo-ration agreement was in place, it was important to havemechanisms to ensure continuous monitoring and allowearly identification of potential misalignment of inter-ests or expectations. This was achieved, for example inthe Body and AmericanBio cases, by matching projectstructures and processes to partners so that issues andconcerns could quickly be identified. Where this wasnot achieved, as in the DiagnosticLabs case, the col-laboration quickly broke down. Other research has alsoidentified that the alignment of interests and expecta-tions between partners is crucial for innovation involvingcollaboration between biotechnology firms and pharma-ceutical firms (Rhodes et al., 2003). As Rhodes et al. putit: “the process of partner identification should not beundertaken opportunistically” (p. 300).

A second mechanism linked to relational capabilitiesthat was important at the project level was the abilityto build upon existing inter-organizational networks togenerate resources and buy-in from users (i.e. clinicians).This links to recent literature on innovation ecosystems(Adner, 2006) that identifies the importance of build-ing markets as well as ‘brilliant products’. Moreover,it is clear that users can contribute to the developmentof the product itself (Lettl et al., 2006; Von Hippel,2005). Enrolment (of users and resources) in our casesinvolved building from existing networks which thenacted as a ‘centre of gravity’ for further networking andenrolment (Kreiner and Schultz, 1993). For example,AmericanBio’s acquisition of a European company gaveit access, not only to that company’ technology, but alsoto the network of influential clinicians and clinical datait had established. AmericanBio also committed signif-icant resources to training clinicians in the new surgicaltechniques so that they could act as ‘opinion leaders’ fortheir product in the clinical community (Rogers, 1995).NewTissueCo also capitalized on existing relationshipsby closing licensing deals and using these to leveragecommercial viability and secure resources for their morelong term and riskier core project. Furthermore, Diag-nosticLabs used their CEO’s existing personal networksto conduct due diligence for free from Bioclinical, withthe promise of future gains should the collaborationprove fruitful.

A third important mechanism was interaction with
regulators, either directly to shape regulation and ensureapproval (as in AmericanBio), or with other firmsthat had regulatory expertise and could help developrequired documentation and interactions with the FDA

ch Polic

(tbi‘M(c

hnivawAvgncpfdspoasc

mkatvtlpltuseoii(iaiWt


as in Body and DiagnosticsLabs). Given the uncer-ainty associated with treatments being developed iniotech companies, such collaborations were extremelymportant, with regulators (and regulations) acting asobligatory passage points’ in networks (Callon, 1986).

any small biotech firms do not have this expertiseHawthorne, 2005) and so are very dependent on spe-ialized consultants to provide it.

Finally, our cases demonstrated the importance ofaving in projects a product ‘magnet’ to bridge orga-izations in the market place (Doz et al., 2001). Thus,n NewTissueCo, the ‘revolutionary’ nature of the inno-ation mobilized different stakeholders in the networknd helped entice other organizations to want to workith the company on this ‘leading edge’ technology. InmericanBio, ELBOW was a well developed but inno-ative product, where desired characteristics for the nexteneration were known in the user and business commu-ities. Similarly, in DiagnosticLabs, the ‘theragnostic’oncept helped bring together different ‘pieces of theuzzle’. When there was no clear product ‘magnet’ –or example, in cases with multiple possible indicationserived from the same compound – the project coulduffer from a lack of focus. Whilst product magnets areerhaps more closely linked to markets than networks,ur data imply that they may help to generate activitynd networking around particular areas of innovation,o potentially shaping (and being shaped by) relationalapabilities.

Our analysis suggested that these mechanisms wereore crucial in those projects characterized by high

nowledge boundaries (quadrants III and IV). Thisppeared to be because of the highly novel nature ofhe innovations in these projects and the very long inno-ation life cycles. This meant that there was a needo constantly leverage resources in order to supportong term commercial developments. Moreover, for aroject to be ‘successful’, it needed to deal with prob-ems associated with the disruption to existing practiceshat these innovations could introduce. Our case projectssed and built upon existing networks to generate andustain resources and ‘buy-in’ (mechanism 6) and tonsure approval (mechanism 7). Furthermore, the naturef these innovations meant that knowledge integrationn the context of complex inter-organizational dynam-cs was crucial. Such projects encountered, in Carlile’s2004) terms, political or ‘pragmatic’ boundaries. Align-ng interests and expectations (mechanism 5) and having

strong product magnet (mechanism 8) was helpfuln terms of dealing with such pragmatic boundaries.

hile projects in quadrants I also involved collabora-ion with other organizations (e.g. technology transfer

y 36 (2007) 529–547 543

offices or investors) their main challenge was to developtheir credibility across the scientific/commercial divide;knowledge integration per se was less important. Inter-estingly, then, the mechanisms we identified that weargue link to relational capabilities were more focusedon ways in which relationships could foster knowledgeintegration by overcoming pragmatic boundaries, ratherthan on building the relationships per se (Carlile, 2004).

6. General discussion and conclusions

This study has explored how macro-level capabilitiesrelate to micro biomedical innovation processes, focus-ing on integrative and relational capabilities found tobe important in the biomedical domain (Owen-Smithet al., 2002). This builds from earlier research that hasshown how institutions governing labor, finance andproduct markets affect innovation activities and the per-formance of sectors and nations (Nelson, 1993; Hall andSoskice, 2001; Clark, 1987). A main contribution hasbeen to identify and unpack mechanisms (depicted inTables 1 and 2) that appear crucial in helping to explainhow macro-level capabilities play out at the level ofmicro-level innovation projects. These mechanisms arenot exhaustive but provide a useful starting point in termsof understanding the processes through which macrocapabilities may come to influence innovation processes.

It is worth noting that the mechanisms identifiedplayed a role in shaping innovation processes but mayalso be shaped by them. For example, having scientificand commercial credibility with investors both influ-enced innovation processes (by attracting funding) butwas also influenced by them (past ‘successes’ generat-ing credibility, for example). Moreover, whilst we did notaddress this in our study, these mechanisms might influ-ence, as well as be influenced by institutional arrange-ments and capabilities. For example, prior research hasfound that labor markets and integrative capabilities atthe macro level influence individuals’ career perceptionsand mobility, but that this relationship is recursive sincethey are also influenced by them (Casper and Murray,2005). Moreover, these mechanisms work together ininnovation processes, not in isolation. For example, inNewTissueCo a strong figurehead role was coupled withcomplementary commercial expertise and a relativelystrong product magnet, which allowed the project tosecure further funding, despite the overriding suspicionof commercially oriented activity by the UK founder.

Where previous work has focused on benefits ofmacro capabilities in terms of knowledge flows, ourdata suggest that issues of credibility, values and per-ceptions, and alignments of interests are equally, if not

ch Policy 36 (2007) 529–547
more, important in terms of mobilizing innovation atthe project level. Yet, most policy initiatives aimed atencouraging innovation in life sciences (e.g. technologytransfer, network grants, joint patents and so forth) focuson knowledge flows/knowledge transfer between publicand private organizations. Our study suggests that theeffects of these initiatives may be limited if attentionis not also given to the normative mechanisms high-lighted here. In the UK context, where scientific, clinicaland commercial interests are more clearly demarcated(Clark, 2000), these latter, normative, concerns becomeeven more central than in the US, where the hybridizationof scientific, clinical and commercial values is generallymore acceptable (Clark, 2000). To extend a constructionmetaphor, in the UK, building bridges to allow knowl-edge to flow between PROs, commercial and clinicalorganizations may be problematic because the ends ofthe bridge may be substantively different.

A second contribution is to consider the relativeimportance of macro capabilities at the institutional levelfor characteristically kinds of innovation projects. Thus,we can propose that the influence of institutionalizedcapabilities on innovation process at the micro level issystematically related to different modes of organizinginnovation. Whilst our inductive methodology allows usto generate this proposition, future research would beneeded to confirm (or refute) it. However, our analysissuggests, in line with other theorists, that institutionalarrangements do not determine innovation processes butmay, as Clark (2000) puts it, generate ‘zones of manoeu-vre’ that allow some kinds of activities to occur moreeasily than others. Moreover, it suggests that the relativedisadvantages of being in an institutional context (theUK) that is less supportive of integrative and relationalcapabilities can be overcome where mechanisms of thekinds we have identified can nevertheless be developed(in NewTissue, for example). Similarly, whilst Mallonet al.’s (2005) study of UK scientists in PROs foundthe majority to be uncomfortable with coupling scien-tific and commercial ambition, a significant minority (the‘strategic opportunists’) were prepared to move between‘the bench’ and a commercial career.

This suggests that generic statements about relativenational advantage in biomedical innovation need to betempered by a consideration of the kinds of project andthe combination of mechanisms deployed at project, firmor sector levels (Casper and van Waarden, 2005). Theframework depicted in Fig. 2 attempts, then, to provide
further insights into the relative advantage/disadvantageof particular national contexts for innovation. Thus, ouranalysis suggests that innovation processes in lower leftpart of Fig. 2 (quadrant II) are less affected by macro
Fig. 2. The importance of integrative and relational capabilities formodes of organizing biomedical innovation.

capabilities and so we may not expect significant dif-ferences across US and UK contexts for this mode. Inquadrant II, innovation projects are dominated by largepharmaceutical firms, which operate on a global basis.Arguably, then, national institutional contexts might playa less important role here, as compared with the featuresof the particular lead organizations, which have a rel-atively high degree of resource, autonomy and control(Hardy and Phillips, 1998; Hardy et al., 2003).

On the other hand, projects characterized by the upperright part of Fig. 2 (quadrant IV) are likely to be mostaffected by macro capabilities of the particular institu-tional context in which they operate. Here, mechanismsrelating to both integrative and relational capabilitiesappear to be important in order to orchestrate looserelationships and bridge knowledge boundaries, mean-ing that stronger differences in innovation performancewould be expected between the UK and US for thesekinds of projects. While we do not have comparativedata per se that would allow us to confirm that thesetypes of project were less frequent and less successfulin the UK as compared to the US, our 1st phase sur-vey data did indicate that a greater instance of quadrantIV type innovation projects in the US than in the UK,and more problems associated with such projects in theUK context. Whilst, the case of NewTissue showed thatquadrant IV type projects in the UK could survive, wewould predict that it would take considerably more effortto develop appropriate mechanisms at project level tocope with the challenges of networking across the aca-
demic/commercial divide and building organizationalcollaborations in this context. We would anticipate, then,that quadrant IV type projects would be more problem-atic in the UK context and might move more rapidly

ch Polic

tmAc

acdUtcabaiehwbi

ttaemspMkmua

aeptwmpaUfannIcadc


han their US counterparts, to another quadrant where theacro-institutional context has less of a divisive impact.gain, this is something that future empirical research

ould more directly evaluate.In the middle diagonal of Fig. 2 (quadrants I

nd III), we expect moderate effects of the macroapabilities—they matter but perhaps not enough to pro-uce systematically large differences across the US andK contexts. For example, for quadrant I, where integra-

ive capability has more importance, specific initiativesan be built to cope with the more rigid divide betweencademia and industry in the UK context, for exampley introducing policies that direct TTO’s activities tossist spin-off companies and scientists in their partic-pation within loose networks (as described in the casexamples). In quadrant III, where relational capabilityas more importance, innovation projects are conductedithin larger organizations, and of course some UK-ased firms can be as experienced as US-based firmsn developing the required collaborations.

This kind of analysis extends existing research thatreats the impact of integrative and relational capabili-ies on innovative performance in the biomedical sectors, in broad terms, uniformly positive (Owen-Smitht al., 2002), by suggesting that their impact may beore acute for some kinds of project than others. Our

tudy also helps to address a central critique of com-arative institutional studies concerning, as Casper andurray (2005; 56) put it (in relation to labor mar-

et institutions), “the limited connection made betweenacro-institutions and the micro-dynamics (of individ-

al careers) through which these institutional differencesre manifested”.

Before discussing the implications for policy, therere limitations that should be dealt with. First, this is anxploratory study using a limited number of firms androjects; therefore, we cannot generalize our findingso the larger population of firms and projects. Whilste can suggest contingencies between the impact ofacro capabilities and different kinds of projects, the

ropositions developed here require further research onbroader sample of development projects in the UK andS. Second, whilst our framework is useful in identi-

ying patterns of innovation, it is clearly broadly-basednd more fine-tuned analysis of particular modes wouldo doubt be able to identify further variation in the orga-ization of biomedical innovation (Lockett et al., 2005).t should also be noted that innovation processes are not
onfined to particular ‘boxes’—they may shift modescross their lifecycle. For example, if tissue engineeringoes eventually establish itself as a professionally demar-ated ‘discipline’, then innovation projects could shift
y 36 (2007) 529–547 545

to quadrants with characterised as ‘low’ in knowledgeboundaries. Third, whilst our classification of mecha-nisms in terms of whether they link to integrative orrelational capabilities is useful for analytical purposes,clearly integrative and relational capabilities are them-selves related concepts (both being linked to networkties, for example). Thus, feasibly, some mechanismsmight relate to both capabilities to a greater or lesserextent. Further work would be required to tease our inter-relationships and test further our initial propositions.

Turning finally to policy implications, this study sug-gests a need for national policy aimed at improvingbiomedical innovation to be sensitive to the differentways of organizing innovation identified here. Takingthese implications further, it suggests potential perverseeffects of, supposedly supportive, policy initiatives forknowledge transfer. For example, in the UK, governmentpolicies aimed at helping academic scientists to switch toindustry careers (e.g. by starting up businesses) or uni-versity policies which allow academics to engage in aspecified number of days’ consultancy ‘outside’ of theiracademic work, may actually serve to reinforce the fun-damental gap between academic and commercial valuesand career interests. More important in the UK context,might be to develop more normatively-based initiativesand incentives that encourage, for want of a better term,‘strategic opportunism’ to become a more legitimate partof academic practice (Mallon et al., 2005). Such initia-tives might include, for example, industry secondmentsof doctoral students or incentives for academic scientiststo participate in commercial scientific advisory boards.Finally, our results suggest that attempts to replicate USpolicy are unlikely to be fully effective in the UK context.For example, the UK Department of Trade and Indus-try (DTI), together with UK research councils, are keento replicate ‘MIT-type’ institutional mechanisms to sup-port innovation research (e.g. through initiatives such asthe Cambridge-MIT Institute and the ‘Innovation Chal-lenge’). However, as seen, the UK context may not sup-port the hybridization of professional and occupationalpractices seen in the US, or the development of scientificentrepreneurs, required to make such a model work.

Acknowledgements

This research was co-funded by the Economic andSocial Research Council (UK) and the Warwick Inno-vative Manufacturing Research Centre (UK) and also
co-sponsored by Bentley College, Boston (US). Wewould like to acknowledge the contribution of Dr MarkusPerkmann and Miriam Mendes to the earlier data collec-tion phases of this research project.

ch Polic
References

Adner, R., 2006. Match your innovation strategy to your innovationecosystem. Harvard Business Review 84 (4), 98–107.

Aldrich, H., 2000. Organizations Evolving. Sage, London.Alter, C., Hage, J., 1993. Organizations Working Together. Sage, New-

bury Park, CA.Alvesson, M., Skoldberg, K., 2000. Reflexive Methodology: New Vis-

tas for Qualitative Research. Sage Publications, London.Bock, A.K., Ibarreta, D., Rodriguez-Cerezo, E., 2003. Human

tissue-engineered products: today’s markets and future prospects.Report published by the Institute for Prospective TechnologicalStudies.

Callon, M., 1986. Some elements in a sociology of translation: domes-tication of the scallops and fisherman of St. Brieuc Bay. In: Law,J. (Ed.), Power, Action and Belief. Routledge, London.

Carlile, P.R., 2004. Transferring, translating, and transforming: an inte-grative framework for managing knowledge across boundaries.Organization Science 5 (5), 555–568.

Carlile, P., 2002. A pragmatic view of knowledge and boundaries:boundary objects in new product development. Organization Sci-ence 13, 442–455.

Carlsson, B. (Ed.), 2002. Technological Systems in the Bio Industries:An International Study. Kluwer Academic Publishers, Boston.

Casper, S., 2000. Institutional adaptiveness, technology policy, and thediffusion of new business models: the case of German biotechnol-ogy. Organization Studies 21, 887–914.

Casper, S., Kettler, H., 2001. National institutional frameworks andthe hybridization of entrepreneurial business models: The Germanand UK biotechnology sectors. Industry and Innovation 8, 5–30.

Casper, S., Murray, F., 2005. Exchange under conflicting institutionallogics: commerce and science in the world of population genetics.Unpublished manuscript.

Casper, S., van Waarden, F. (Eds.), 2005. Innovation and Institutions:A Multidisciplinary Review of the Study of Innovation Systems.New Horizons of the Economics of Innovation series. Edward ElgarPublishing, Cheltenham.

Christensen, C.M., 2000. The Innovator’s Dilemma. Harper Business,New York.

Clark, P., 1987. Anglo-American Innovation. De Gruyter, New York.Clark, P., 2000. Organizations in Action: Competition Between Con-

texts. Routledge, London.Clark, P., 2003. Organizational Innovations. Sage, London.CMR International, 2006. Centre for Medicines Research International

Pharmaceutical R and D Factbook. CMR International, Surrey.Cohen, W.M., Levinthal, D.A., 1990. Absorptive capacity: a new

perspective on learning and innovation. Administrative ScienceQuarterly 35 (1), 128–152.

Coombs, R., Harvey, M., Tether, B.S., 2003. Analyzing distributedprocesses of provision and innovation. Industrial and CorporateChange 12 (6), 1125–1155.

Deeds, D.L., Decarolis, D., Coombs, J., 1997. The impact of firm-specific capabilities on the amount of capital raised in an initialpublic offering: evidence from the biotechnology industry. Journalof Business Venturing 12 (1), 31–46.

Denzin, N., Lincoln, Y., 1998. Entering the field of qualitative research.In: Denzin, N., Lincoln, Y. (Eds.), Strategies of Qualitative Inquiry.
Sage, London, pp. 1–34.
Dodgson, M., Gann, D., Salter, A., 2004. Innovation technology:exploring the impact of simulations on the innovation process.Paper presented at DRUID summer conference. Industrial Dynam-ics, Innovation and Development, Helsingore, Denmark.

y 36 (2007) 529–547

Dopson, S., 2005. The diffusion of medical innovations: can con-figurational sociology contribute? Organization Studies 26 (8),1125–1144.

Doz, Y., Santos, J., Williamson, P., 2001. From Global to Metana-tional: How Companies Win in the Knowledge Economy. HarvardBusiness School Press, Boston.

Gay, B., Dousset, B., 2005. Innovation and network structural dynam-ics: study of the alliance network of a major sector of thebiotechnology industry. Research Policy 34, 1457–1475.

Gibbons, M., Limoges, C., Nowotny, H., Schwartzman, S., Scott, P.,Trow, M., 1994. The New Production of Knowledge: The Dynam-ics of Science and Research in Contemporary Societies. Sage,London.

Gittell, J.H., Weiss, L., 2004. Coordination networks within and acrossorganizations: a multi-level framework. Journal of ManagementStudies 41 (1), 127–153.

Glaser, B., 1978. Theoretical Sensitivity: Advances in the Methodol-ogy of Grounded Theory. Sociology Press.

Hall, P.A., Soskice, D. (Eds.), 2001. Varieties of Capitalism: TheInstitutional Foundations of Comparative Advantage. Oxford Uni-versity Press, Oxford.

Hardy, C., Phillips, N., 1998. Strategies of engagement: lessons fromthe critical examination of collaboration and conflict in an interor-ganizational domain. Organization Science 9 (2), 217–230.

Hardy, C., Phillips, N., Lawrence, T., 2003. Resources, knowledge andinfluence: the organizational effects of interorganizational collab-oration. Journal of Management Studies 40 (2), 321–346.

Hawthorne, F., 2005. Inside the FDA: The Business and Politics Behindthe Drugs that We Take and the Food that We Eat. Wiley, Hoboken,NJ.

Henderson, R., 1994. The evolution of integrative capability: inno-vation in cardiovascular drug discovery. Industrial and CorporateChange 3 (3), 607–630.

Higgins, M., Gulati, R., 2003. The effects of upper echelon affilia-tions on underwriter prestige. Organization Science 14 (3), 244–263.

Hilton, T., Flanzer, J., Cartwright, W., Fletcher, B., 2002. Resistanceto innovation among US drug abuse treatment providers: Whenorganizational knowledge interferes with organizational learning.In: Paper presented at the Organizational Knowledge, Learning andCapabilities Conference, Athens.

Howard-Grenville, J., Carlile, P., 2006. The incompatibility of knowl-edge regimes: consequences of the material world for cross-domainwork. European Journal of Information Systems 15 (5), 473–485.

Kreiner, K., Schultz, M., 1993. Informal collaboration in R and D: theformation of networks across organizations. Organization Studies14 (2), 189–209.

Kukla, A., 2000. Social Constructivism and the Philosophy of Science.Routledge, New York.

Lettl, C., Herstatt, C., Gemuenden, H.G., 2006. Users’ contribution toradical innovation: evidence from four cases in the field of medicalequipment technology. R and D Management 36 (3), 251–272.

Liebeskind, J.P., Oliver, A.L., Zucker, L., Brewer, M., 1996. Socialnetworks, learning and flexibility: souring scientific knowledge innew biotechnology firms. Organization Science 7 (4), 428–443.

Lockett, A., Siegel, D., Wright, M., Ensley, M.D., 2005. The creation
of spin-off firms at public research institutions: managerial andpolicy implications. Research Policy 34, 981–993.
Mallon, M., Duberley, J., Cohen, L., 2005. Careers in public sectorscience: orientations and implications. R and D Management 35(4), 395–407.

ch Polic

M

M

M

N

N

N

O

O

O

O

P

P

R

R

R


cMillan, G.S., Narin, F., Deeds, D., 2000. An analysis of the criticalrole of public science in innovation: the case of biotechnology.Research Policy 29, 1–8.

iles, M.B., Huberman, A.M., 1994. Qualitative Data Analysis. Cam-bridge University Press, Cambridge.

urray, F., 2002. Innovation as co-evolution of scientific and tech-nological networks: exploring tissue engineering. Research Policy31, 1389–1403.

elson, R. (Ed.), 1993. National Innovation Systems: A ComparativeAnalysis. Oxford University Press, New York/Oxford.

icholson, S., Danzon, P., McCullough, J., 2005. Biotech-pharmaceutical alliances as a signal of asset and firm quality.Journal of Business 78 (4), 1433–1464.

owotny, H., Scott, P., Gibbons, M., 2001. Re-Thinking Science:Knowledge and the Public in an Age of Uncertainty. Polity, Cam-bridge.

rlikowski, W.J., 2002. Knowing in practice: enacting a collectivecapability in distributed organizing. Organization Science 13 (3),249–273.

wen-Smith, J., 2003. From separate systems to a hybrid order: accu-mulative advantage across public and private science at ResearchOne universities. Research Policy 32, 1081–1104.

wen-Smith, J., Riccaboni, M., Pammolli, F., Powell, W., 2002. Acomparison of US and European university–industry relations inthe life sciences. Management Science 48 (1), 24–43.

wen-Smith, J., Powell, W.W., 2004. Knowledge networks as channelsand conduits: the effects of spillovers in the Boston BiotechnologyCommunity. Organization Science 15 (1), 5–22.

ettigrew, A.M., 1990. Longitudinal field research on change: theoryand practice. Organization Science 1 (3), 267–292.

owell, W., Koput, W., Smith-Doerr, L., 1996. Interorganizationalcollaboration and the locus of innovation: networks of learningin biotechnology. Administrative Science Quarterly 41 (1), 116–130.

asmussen, B., February 2005. Pharmaceutical industry project.Working Paper Series. Centre for Strategic Economic Studies,Victoria. University of Technology.

hodes, I., Nelson, C., Berman, G., 2003. The key to successful col-laborations: rigorous and independent due diligence. Journal ofCommercial Biotechnology 9 (4), 297–305.

ogers, E.M., 1995. Diffusion of Innovations, fourth ed. Free Press,New York.

y 36 (2007) 529–547 547

Rosenberg, N., Nelson, R., 1994. American universities and technicaladvance in industry. Research Policy 23, 323–348.

Salman, N., Saives, A.-L., 2005. Indirect networks: an intangibleresource for biotechnology innovation. R and D Management 35(2), 203–215.

Saunders, M., Lewis, P., Thornhill, A., 2000. Research Methods forBusiness Students. Prentice Hall, FT.

Stake, R.E., 1995. Case studies. In: Denzin, N.K., Lincoln, Y.S. (Eds.),Handbook of Qualitative Research. Sage, Thousand Oaks, CA, pp.236–247.

Stokes, D.E., 1997. Pasteurs Quadrant: Basic Science and Technolog-ical Innovation. Brookings.

Swan, J., Bresnen, M., Mendes, M., Newell, S., Perkmann, M., Rober-ston, M., 2005. Exploring interactivity in biomedical innovation:a framework and case study analysis. In: Proceedings of the 5thEuropean Conference on Organizational Knowledge, Learning andCapabilities, Boston, March.

Thompson, J., 1967. Organizations in Action. McGraw-Hill, NewYork.

Turpin, T., Deville, A., 1995. Occupational roles and expectations ofresearch scientists and research managers in scientific researchinstitutions. R and D Management 25 (2), 141–157.

Tushman, M., Scanlan, T., 1981. Boundary spanning individuals: theirrole in information transfer and their antecedents. Academy ofManagement Journal 24, 289–305.

Tsoukas, H., 1989. The validity of idiographic research explanations.Academy of Management Review 14 (4), 551–561.

Von Hippel, E., 2005. Democritizing Innovation. The MIT Press,Boston.

Wageman, R., 1995. ‘Interdependence and group effectiveness’.Administrative Science Quarterly, 145–180.

Whitley, R., 2000. The institutional structuring of innovation strategies:business systems, firm types and patterns of technical change indifferent market economies. Organization Studies 21 (5), 855–886.

Whitley, R., 2003. Competition and pluralism in the public sciences:the impact of institutional frameworks on the organization of aca-demic science. Research Policy 32 (6), 1015–1029.

Zider, B., 1998. How venture capital works. Harvard Business Review76 (6), 131–139.

Zucker, L., Darby, M., Brewer, M., 1998. Intellectual capital andthe birth of US biotechnology enterprises. American EconomicReview 88, 290–306.

Modes of organizing biomedical innovation in the UK and US and the role of integrative and relational capabilities

Documents