Analyzing the functional dynamics of technological

407–429

Available online at www.sciencedirect.com

Analyzing the functional dynamics of technological

innovation systems:A scheme of analysis

Anna Bergek a,∗,Staffan Jacobsson b,Bo Carlsson c,

Sven Lindmark d,Annika Rickne e

a Department of Management and Engineering,Link¨o ping University,SE-58183Link¨o ping,Sweden

b Department of Energy and Environment,IMIT and RIDE,Chalmers University of Technology,SE-41296G¨o teborg,Sweden

c Weatherhea

d School of Management,Cas

e Western Reserve University,11119Bellflower Road,Cleveland,OH44106-7235,USA

d IMIT,RIDE and Department of Innovation Engineering and Management,Chalmers University of Technology,SE-41296G¨o teborg,Sweden

e Centre for Innovation,Research and Competence in the Learning Economy(CIRCLE),Lund University,Box118,SE-22100Lund,Sweden

Received18January2006;received in revised form4May2007;accepted19December2007

Available online11February2008

Abstract

Various researchers and policy analysts have made empirical studies of innovation systems in order to understand their current structure and trace their dynamics.However,policy makers often experience difficulties in extracting practical guidelines from studies of this kind.In this paper,we operationalize our previous work on a functional approach to analyzing innovation system dynamics into a practical scheme of analysis for policy makers.The scheme is based on previous literature and our own experience in developing and applying functional thinking.It can be used by policy makers not only to identify the key policy issues but also to set policy goals.

©2007Elsevier B.V.All rights reserved.

Keywords:Innovation system;Dynamics;Functional analysis;Policy

1.Introduction

Scholars on innovation and technology have almost completely rejected the market failure approach as a basis of policy action.It is argued repeatedly in the literature(e.g.Malerba,1996;Metcalfe,1992,2004; Smith,2000)that the approach isflawed and insuffi-cient.A systems approach to innovation is often seen ∗Corresponding author.Tel.:+4613282573;

fax:+4613281873.

E-mail addresses:anna.bergek@liu.se(A.Bergek),

staffan.jacobsson@chalmers.se(S.Jacobsson),

Bo.Carlsson@case.edu(B.Carlsson),sven.lindmark@chalmers.se (S.Lindmark),annika.rickne@circle.lu.se(A.Rickne).as a more appropriate alternative.In particular,the con-cept of‘innovation system’has won the approval of an increasing number of academic researchers interested in the processes underlying innovation,industrial trans-formation and economic growth.The innovation system approach has also been adopted by regional and national authorities/agencies as well as by international organi-zations(e.g.the OECD,the European Commission and UNIDO)interested in stimulating these processes.

Various researchers and policy analysts have made attempts to study innovation systems empirically in order to describe and understand their structure,dynamics and performance.However,recent surveys of the literature (e.g.Edquist,2004;Liu and White,2001)have acknowl-edged the lack of comparability between these studies

0048-7333/$–see front matter©2007Elsevier B.V.All rights reserved. doi:10.1016/j.respol.2007.12.003

408 A.Bergek et al./Research Policy37(2008)407–429

as well as the conceptual heterogeneity in the innovation system literature.Perhaps as a consequence of this,the innovation system approach has been criticized for not providing practical enough guidelines for policy mak-ers(cf.Edquist,2004;Klein Woolthuis et al.,2005). There is,thus,a need for a practically useful analyti-cal framework that allows for the assessment of system performance as well as the identification of factors influ-encing performance.

This paper presents a scheme of analysis which addresses these issues and may be used by researchers, as well as policy makers,to analyze specific innova-tion systems in order to identify key policy issues and set policy goals.The contribution is twofold.First,the paper describes a systematic step-by-step approach to analyzing innovation systems,describing and assessing performance and identifying key policy issues.1Sec-ond,and most important,the paper presents a framework that not only captures the structural characteristics and dynamics of an innovation system,but also the dynamics of a number of key processes,here labeled‘functions’, that directly influence the development,diffusion and use of new technology and,thus,the performance of the innovation system.The functions have been synthe-sized from a number of different system approaches to innovation and provide a basis for performance assess-ment as well as comparison between different innovation systems in terms of system dynamics.

The paper is organized as follows.In Section2,we position ourselves in the innovation system literature and explain how our approach on functions in innovation sys-tems was developed.In Section3,we outline the scheme of analysis.Section4summarizes our conclusions and presents some recommendations for further research.

2.Positioning and development of the analytical approach

2.1.Innovation system as an analytical construct

A general definition of a system is a group of com-ponents(devices,objects or agents)serving a common purpose,i.e.working towards a common objective or overall function.The components of an innovation sys-tem are the actors,networks and institutions(Carlsson and Stankiewicz,1991)contributing to the overall func-1The scheme of analysis was developed as part of a collaborative project with VINNOV A(the Swedish Agency for Innovation Systems) (see Bergek et al.,2005).In this project we also analyzed three empiri-cal cases(“IT in home care”,“Mobile data”(see Lindmark and Rickne, 2005)and“Biocomposites”)in collaboration with VINNOV A.tion of developing,diffusing and utilizing new products (goods and services)and processes(cf.Bergek,2002; Carlsson and Stankiewicz,1995;Galli and Teubal, 1997).

Although the system concept may suggest collective and coordinated action,an innovation system is primar-ily an analytical construct,i.e.a tool we use to better illustrate and understand system dynamics and perfor-mance.This implies that the system in focus does not have to exist in reality as fully-fledged.Instead,it may be emerging with very weak interaction between com-ponents.

Moreover,interaction between components may be unplanned and unintentional rather than deliberate even in a more developed innovation system.Using the notion of an“overall function”does not imply that all actors in a particular system exist for the purpose of serving that function or are directed by that function.Actors do not necessarily share the same goal,and even if they do,they do not have to be working together consciously towards it(although some may be).Indeed,conflicts and tensions are part and parcel of the dynamics of inno-vation systems.2Clearly,we do not see the system’s components as directed or orchestrated by any specific actors.

A number of different innovation system concepts have been put forward in the literature,including national systems of innovation(Freeman,1987;Lundvall,1992a; Nelson,1992),regional innovation systems(Asheim and Isaksen,1997;Cooke et al.,1997),sectoral systems of innovation and production(Breschi and Malerba,1997; Malerba,2002)and technological systems(Carlsson and Stankiewicz,1991).3There are also other similar socio-technical system concepts(cf.Bijker,1995;Geels,2004; Hughes,1983).In this paper,we focus on technolog-ical innovation systems(TIS)(Bergek et al.,2007a), i.e.socio-technical systems focused on the development, diffusion and use of a particular technology(in terms of knowledge,product or both).4

2Schumpeterian competition is a vital part of vibrant innovation sys-tems,andfirms and other actors also compete in shaping expectations of a technology and in building legitimacy for it.

3For an overview,see Carlsson et al.(2002)and Edquist(1997). 4The concept of technology incorporates(at least)two interrelated meanings.First,technology refers to material and immaterial objects –both hardware(e.g.products,tools and machines)and software(e.g. procedures/processes and digital protocols)–that can be used to solve real-world technical problems.Second,it refers to technical knowl-edge,either in general terms or in terms of knowledge embodied in the physical artifact.In line with Layton(1974)and Das and Van de Ven(2000),we include both of these meanings(i.e.both artifact and knowledge)in our definition of technology.A.Bergek et al./Research Policy37(2008)407–429409

TISs do not only contain components exclusively ded-icated to the technology in focus,but all components that influence the innovation process for that technology.A TIS may be a sub-system of a sectoral system(when the focus is one of the sector’s products or a knowledgefield that is exclusive to the sector)or may cut across several sectors(when the focus is a more“generic”knowledge field that several sectors make use of,e.g.microwave technology(see Holm´e n and Jacobsson,2000)).TISs may have a geographical dimension,but are often inter-national in nature.5

2.2.Previous innovation system approaches to innovation policy6

A central proposition in the systems literature on policy is that just as the nature of actors/markets may obstruct the formation of a TIS,so can institutions and networks(e.g.Carlsson and Jacobsson,1997;Edquist, 1999;Malerba,1996;Metcalfe,2004;Rotmans et al., 2001;Unruh,2000).Eventually,such weaknesses in system structure may lead to“system failure”,i.e.a sys-tem that fails to develop or does so in a stunted fashion (Carlsson and Jacobsson,1997).

Most of the literature discussing innovation system failure tends to focus on perceived weaknesses in the structural composition of a system.For example,all the four types of system failures identified by Klein Woolthuis et al.(2005)in their recent synthesis and re-categorization of previous system failure literature are related to structural components:infrastructural fail-ures(related to actors and artifacts),institutional failures (related to institutions),interaction failures(related to networks)and capabilities failures(related to actors). However,it is difficult,if not impossible,to evaluate the “goodness”or“badness”of a particular structural ele-ment or combination of elements without referring to its effects on the innovation process.For example,how do we know whether the existence of a particular actor network is a strength(e.g.a source of synergy)or a weak-ness(e.g.a source of lock-in or“group-think”)(cf.Klein Woolthuis et al.,2005),without identifying its influence on the innovation process and its key sub-processes?

Thus,in order to be able to identify the central policy issues in a specific innovation system,we need to sup-plement a structural focus with a process focus.In this

5A TIS with a high degree of regional concentration comes close to the definition of a technological cluster or region(cf.Cantner and Graf,2004;Maskell,2001;Porter,2000).

6This section is based on Bergek et al.(2007b).paper,we present a framework outlining seven key pro-cesses–here labeled‘functions’–which have a direct and immediate impact on the development,diffusion and use of new technologies,i.e.the overall function of the TIS as defined above.It is in these processes where pol-icy makers may need to intervene,not necessarily the set-up of the structural components(actors,networks, institutions).The functions approach to innovation sys-tems thus implies a focus on the dynamics of what is actually“achieved”in the system rather than on the dynamics in terms of structural components only.This is,indeed,its main benefit:It allows us to separate struc-ture from content and to formulate both policy goals and policy problems in functional terms.7We will return to this point.

2.3.The development of the“functional dynamics”approach

As noted previously,concerns have been raised with regard to the conceptual heterogeneity of the innova-tion system concept.This was one of the starting points of the functional dynamics approach presented in this paper:ourfirst identification of a number of functions (Johnson,19988)was made in an attempt to see whether there was any agreement between different innovation system approaches with regard to what they described “happened”in the system and,if so,to identify the key processes that they agreed upon.A scrutiny of the received literature revealed that the system approaches indeed shared an understanding of a set of such basic “functions”,defined as the contribution of a component or a set of components to the overall function of the innovation system(Johnson,1998,2001).9

Thefirst list of functions/processes was,thus,identi-fied through a scrutiny of a number of central innovation

7For example,the lack of research institutes has often been identified as a major problem in the Swedish National Innovation System,without much empirical evidence that this structural characteristic influences innovation processes in any important way.By focusing on functions, we could be able to analyze how research institutes in other countries influence the innovation process,and then see if this type of influence is absent in the Swedish system or if the same type of influence is present through another type of actor.

8Johnson was Anna Bergek’s maiden name.

9It should be noted here that we use the concept of“functions”without any reference to the sociological concepts of“functionalism”and“functional analysis”.Our analogy is instead technical systems, with“hard”system componentsfilling different technical functions, thereby contributing to the system’s overall(technical)function.As noted previously,the“overall function”is analytically defined and does not imply that actors exist for the purpose of serving that function or are directed by it.410 A.Bergek et al./Research Policy37(2008)407–429

system references,including work by Christopher Free-man,Richard Nelson,Charles Edquist,Bengt-˚Ake Lundvall,Bo Carlsson and Rikard Stankiewicz,com-plemented with literature on related concepts such as socio-technical systems(e.g.Wiebe Bijker and Thomas P.Hughes),development blocs(e.g.Eric Dahm´e n)and industrial networks and clusters(e.g.H˚a kan H˚a kansson and Michael Porter).The processes described in this lit-erature were categorized into a list of eight functions (Johnson,1998).10A similar list of functions was later developed through an empirical study of the biomaterials industry(Rickne,2000).

Through empirical application,11additional literature studies and discussions amongst ourselves and with other researchers pursuing similar approaches,the list has been revised and refined several times.We have added insights from political science(e.g.Sabatier,1998),sociology of technology(e.g.Kemp et al.,1998)and organization theory(e.g.Van de Ven,1993),which in particular have highlighted the political nature of the innovation process and the importance of legitimation.The current frame-work,which we present in detail later on in this paper, includes seven functions on which there is quite large agreement between different functions approaches(see Appendix A for a description and comparison of various contributions).12

Since the framework presented here is based on pre-vious literature,it is perhaps reasonable to assume that a “conventional”innovation system analysis would iden-tify the same processes(although termed differently). However,this does not seem to be the case.First,Edquist (2004)identifies a number of“activities”,defined as “those factors that influence the development,diffusion, and use of innovation”(p.190).Some of these activi-ties are,however,structural in nature(e.g.“creation of organizations”).Moreover,since several of the activities are much more specific in nature than our functions(e.g.“incubator support”vs.“resource mobilization”),they do not cover all aspects of our functions(see Appendix A).Second,in a recent paper(Bergek et al.,2007b)we concluded that although most of our functions were men-tioned in the“Policy and innovation system”literature of the1990s(largely),none of the literature we reviewed 10We searched for references of things that“happened”,took place or“were done”(by any component)in the innovation system.All these were typed into categories that were assigned a label.These labels were used as names for thefirst functions.

11See Bergek and Jacobsson(2003),Jacobsson and Bergek(2004), Jacobsson et al.(2004),Bergek et al.(2005),and Hekkert et al.(2007). 12We do not claim this to be a complete andfinal set of functions. Additional studies will have to refine and possibly add to the list.mentioned all functions at the same time.Instead,policy researchers seemed to focus on a few functions or on general policy problems to be solved,often in an unsys-tematic way and without stating any clear reason for that particular focus.13In addition,two of our functions (‘development of positive externalities’and‘legitima-tion’)were either mentioned in passing or completely left out,which is surprising since their importance has been noted in several other strands of literature that we have reviewed.Here lies the main difference between the functions approach and“conventional”innovation sys-tem analyses with respect to key processes:explicitly stating and including all functions,which allows for the systematic identification of policy problems.It should be noted,though,that this list of functions may require further revisions as and when the research on innovation system dynamics provides new insights.

3.The scheme of analysis

A scheme of analysis is a description of a number of sub-analyses–in the following referred to as“steps”–that need to be taken by the analyst.Our approach implies that the analyst needs to go through six such steps (Fig.1).Thefirst step involves setting the starting point for the analysis,i.e.defining the technological innovation system(TIS)in focus.In the second step,we identify the structural components of the TIS(actors,networks and institutions).In the third step,we move from structure to functions.With an analysis of functions,wefirst desire to describe what is actually going on in the TIS in terms of the seven key processes where we come up with a picture of an“achieved”functional pattern,i.e.a description of how each function is currentlyfilled in the system.The subsequent fourth step is normative;we assess how well the functions are fulfilled and set process goals in terms of a“desired”functional pattern.In thefifth step,we identify mechanisms that either induce(drive)or block a development towards the desirable functional pattern. We can then specify key policy issues related to these inducement and blocking mechanisms,and this is the sixth andfinal step.

It should be noted that the analysis will most often not proceed in a linear fashion(as the focus on“steps”might suggest).In contrast,the analyst has to expect a great number of iterations between the steps in the process of the analysis.For reasons of simplicity,however,we will discuss the six steps sequentially.

13Malerba(1996)and Carlsson and Jacobsson(1997)are the main exception,covering most of our functions.A.Bergek et al./Research Policy37(2008)407–429

411

Fig.1.The scheme of analysis(adapted from Oltander and Perez Vico,2005).

In the following,we will articulate the analytical con-tent as well as some methodological opportunities and problems for each step.We use a number of empirical examples to illustrate our reasoning.These should be seen as“light”illustrations rather than empirical evi-dence of the validity of the framework.14

4.Step1:the starting point for the analysis:

defining the TIS in focus

The empirical operationalization of the TIS concept is not always as straightforward as it may seem at a first glance.Indeed,analysts face several choices when it comes to deciding the precise unit of analysis–or focus–of the study.The outcome of these choices deter-mines what particular TIS is captured,with respect to both structure and functions,and it is therefore crucial to make a deliberate choice,to re-evaluate this through-out the analysis,to draw conclusions as to how the choice of starting point has affected the picture painted, and to communicate the unit of analysis clearly to the recipients of the analysis,be they policy makers or other researchers.Nevertheless,this is often neglected in empirical analyses,and the failure to make explicit 14Most examples draw from longer texts of ours that are avail-able:Holm´e n and Jacobsson(2000),Rickne(2000),Bergek and Jacobsson(2003),Jacobsson and Bergek(2004),Jacobsson et al. (2004),Lindmark and Rickne(2005)and Jacobsson and Lauber (2006).the precise unit of analysis seems to be one reason why it is difficult to compare the results of different studies.

We will outline three types of choices that analysts need to consider:(1)the choice between knowledgefield or product as a focusing device,(2)the choice between breadth and depth,and(3)the choice of spatial domain. In this,there is no one correct choice–the starting point depends on the aim of the study and the interests of the involved stakeholders(e.g.researchers or policy mak-ers).

It follows from our definition of technology(see also footnote4)that the focus of attention may either be a knowledgefield or a product/artifact,and the anal-ysisfirst involves choosing between these two as the starting point.One common–and straightforward–starting point for the analysis is in terms of a product or product group,for instance a wind turbine(Bergek and Jacobsson,2003)or a machine tool(Carlsson and Jacobsson,1993).Another option is to start the anal-ysis in a technological knowledgefield(Holm´e n and Jacobsson,2000).A researcher will presumably choose a focus that reflects the nature of the question raised, whereas policy makers will choose a definition that suits their area of responsibility,which for example may be a knowledgefield,a particular product or a product group.

Having decided on product vs.knowledgefield as a practical way of proceeding,we need to choose breadth of the study.Afirst choice concerns the level of aggre-gation of the study.This is relevant for both alternatives but is most prominent when dealing with a particular412 A.Bergek et al./Research Policy37(2008)407–429

knowledgefield.In addition,focusing on a knowledge field involves determining the range of applications in which the technology is relevant.Let us elaborate on these two points.

First,a decision on the level of aggregation of the study means that we choose between including much,in order to get a broad picture,or being more specific,in order to be able to go more into detail.Certainly,the defi-nition of the knowledgefield to study may be very narrow (e.g.stem cells)or much broader(e.g.IT).It may also be defined as one specific knowledgefield(e.g.microwave technology;see Holm´e n and Jacobsson,2000)or as a set of related knowledgefields(e.g.biocompatible materi-als;see Rickne,2000).15

Second,there is a choice of the range of applications of the technology in question that should be included in the study.The analysis may be limited to its use in specific applications,products or industries.Take the emerging application of“IT in home care”,where a TIS may be defined by the use of a generic technology(IT) in a particular application:care of elderly and ill people in their homes instead of in a hospital.Here,a certain application dictates what actors,networks and institu-tions will be included in an analysis.In other cases,the study may include all possible applications.This was, for instance,done by Holm´e n and Jacobsson(2000)for microwave technology.

To illustrate further the need to make deliberate choices regarding the focus of the TIS,we may take the case of an analyst interested in the emergingfield of biocompatible materials and the associated products of bio-implants,drug delivery and artificial organs.Such a TIS may be defined in terms of the products or by the underlying knowledgefields.If the second alterna-tive is chosen,the analysis could be focused on some of the underlying technologies(e.g.some types of biopoly-mers)or on all of them.Furthermore,the boundaries of the system could be set to some specific applications (e.g.medical applications)or all(e.g.include also envi-ronmental applications;see Rickne,2000).Depending on the choices made,different sets of actors,networks and institutions will be incorporated,and thus we cap-ture different TISs or see different parts of the overall picture.

Finding the appropriate focus may not always be straightforward.When the analyst is new to a case,it

15Technical change often involves the combination of many technolo-gies and complementary products/services,which all need to evolve for the value of an initial innovation to materialize.This implies that no matter how narrow or broad the starting point of the analysis is,the analyst needs to be aware of and include related dynamics.may be necessary to have a broad starting point,and nar-row it down as the understanding of the TIS increases and narrower potential foci are identified.For example, in an earlier study two of us analyzed the larger,product group based,Swedish TIS for renewable energy tech-nology(including,e.g.wind turbines,solar cells,solar collectors and bioenergy)(see Johnson and Jacobsson, 2001),after which we narrowed our focus and analyzed the TIS for wind turbines(see Bergek and Jacobsson, 2003).Thefirst step was necessary for us to begin to understand the features of thefield of renewable energy in general,without which we would not have been able to continue with our in-depth case study of wind turbines.

In addition,given the large uncertainties involved when the analysis concerns an emerging TIS,a defi-nite focus may be difficult to choose and may have to be changed over time.Sometimes,the initial expecta-tions may prove to be quite wrong.For instance,the early development of laser technology was expected to find its main application in space warfare,while later the main application proved to be in CD players.Any early focus should therefore be seen as a“snapshot”valid only at a particular point in time.As the analysis unfolds,and as time passes,we may learn that the initial focus needs to be altered.

As noted in Section2.1,the analysis does not require the focal TIS to exist in reality as a system:An emerg-ing TIS may be analyzed as well and it may,indeed, even be possible and fruitful to analyze a TIS that only exists as an idea.For example,the Swedish TIS for biocomposites16today only exists in the form of a num-ber of separate sub-systems,each closely related to one application(e.g.packaging or furniture).From a policy perspective,however,it seems to make sense to work towards integrating these into one overall TIS,since this may increase learning,knowledge development and, thereby,the rate of development of the system as a whole.

A TIS may therefore be defined as an analytical con-struct incorporating hitherto disconnected sub-systems and guide policy makers in their decisions.

Having made the choices specified above,the study may also–as a complement–have a spatial focus.While TISs are generally global in character,there may be rea-sons to focus on a spatially limited part of a particular

16A composite is a combination of two or more distinct materi-als,usually some type offibers and a resin matrix.The concept of ‘biocomposites’refers to composite materials,where all input materi-als are renewable in contrast to conventional composites,which are petroleum-based.There are also hybrid forms,for example“wood plastic composites”that consist of woodfibers in a petroleum-based matrix.

system in order to capture other aspects,perhaps those most relevant for a particular set of actors in a national or regional context.However,a geographical delimita-tion should not be used alone.Moreover,an analysis always needs to have a strong international component simply because a spatially limited part of a global TIS can neither be understood,nor assessed,without a thorough understanding of the global context.

5.Step2:identifying the structural components

of the TIS

Having decided on the focus of the TIS(in a prelim-inary way),the next step is to identify and analyze the structural components of the system.First,the actors of the TIS have to be identified.These may include not only firms along the whole value chain(including those up-and downstream),universities and research institutes, but also public bodies,influential interest organizations (e.g.industry associations and non-commercial organi-zations),venture capitalists,organizations deciding on standards,etc.

To identify actors in a specific industry,there are a number of available methods.Several of these normally need to be used:

•Industry associations are a good source,as are exhi-bitions,company directories and catalogues.

•A patent analysis may reveal the volume and direc-tion of technological activity in different organizations and among individuals and may thus be a useful tool to identifyfirms,research organizations or indi-viduals with a specific technological profile(see e.g.Andersson and Jacobsson,2000;Holm´e n and Jacobsson,2000;Rickne,2000).Identification may take place even if the various organizations are not linked in any form(by markets or networks).17•Bibliometric analysis(volume of publications,cita-tion analysis,etc.)will provide a list of the most active organizations in terms of published papers,etc.,and these organizations will include not only universities but also institutes andfirms.

17Patent analysis is,however,far from unproblematic.The link between patent classes and products is unreliable(Bergek et al.,2004) and a patent analysis is probably more useful if we choose a knowl-edgefield as the starting point.Yet even here,we cannot conclude that afirm with patents in a particular class necessarily masters a tech-nology generally associated with that class.For instance,Holm´e n and Jacobsson(2000)carefully scrutinized patents referring to microwave antennas and found that some patents certainly did not reveal any deep knowledge in the knowledgefield in question.•Interviews and discussions with technology or indus-try experts(“gurus”)as well as withfirms,research organizations,financiers,etc.,is a good way to iden-tify further actors.This may be called a“snowballing”method to identify actors,where each actor may point to additional participants(see Rickne,2000).

The second structural component of interest is that of networks,informal as well as formal.A number of different types of networks are relevant.Some are orchestrated to solve a specific task,such as stan-dardization networks,technology platform consortia, public–private partnerships or supplier groups hav-ing a common customer.Other networks evolve in a less orchestrated fashion and include buyer–seller rela-tionships and university–industry links.In this,some networks are oriented around technological tasks or mar-ket formation and others have a political agenda of influencing the institutional set-up(see e.g.Rao,2004; Sabatier,1998;Suchman,1995).Social communities, such as professional networks and associations or cus-tomer interest groups,may also be important to map.

Formal networks are often easily recognized,whereas the identification of informal networks may require dis-cussion with industry experts or other actors,or analysis of co-patenting,co-publishing or collaboration(e.g.joint ventures and joint university–industry projects).In the case of“mobile data”,18the history of the Swedish TIS shows that networks between the two leadingfirms(Eric-sson and Telia)and academic research groups have been prominent and have contributed to knowledge formation and diffusion(Lindmark and Rickne,2005).Sometimes analysts have to look for subtle signs pointing to the existence or non-existence of networks.For example, given that academia and industry failed to communicate on a specific technical solution to an urgent industrial problem,we could conclude that learning networks were weak in the pellet burner industry in Sweden(Johnson and Jacobsson,2001).

Third,institutions such as culture,norms,laws,regu-lations and routines need to be identified(North,1994). Generally,institutions need to be adjusted,or“aligned”, to a new technology,if it is to diffuse(Freeman and Louc¸˜a,2002).Institutional alignment is,however,not an automatic and certain process but rather the oppo-site.Firms compete not only in the market but also

18“Mobile data”are here defined as non-voice communications, where at least one terminal is connected to the system via radio,pro-viding mobility to the user.It is thus a generic technology which can be used in a large number of applications,ranging from e.g.simple SMS to advanced logistics applications.over the nature of the institutional set-up(Davies,1996; Jacobsson and Lauber,2006;Van de Ven,1993).

Institutions may come in a variety of forms and may influence the TIS in different ways.For example,in the case of the emerging TIS“IT in home care”,a key insti-tution is the procurement policies of the county councils, which discriminate against smaller suppliers.In the case of“biocomposites”,the emerging TIS is influenced by a number of EU regulations and directives concerning broad areas such as chemical substances and recycling. This implies that analysts need to have a broad perspec-tive when mapping relevant institutions.19Sometimes it is the very lack of institutions that is of interest.Again,in the case of the emerging TIS“IT in homecare”,a lack of standardization has led to fragmented markets and poor incentives forfirms to innovate.

For TISs that are only just emerging there are inherent uncertainties,implying that the identification of struc-tural components is thorny and many of the sources mentioned above may be difficult to use.It may prove hard to recognize the relevant actors when directories are scarce,no industry associations exist or if the actors themselves are not aware of belonging to a certain TIS. This was the case for early studies of the emerging sys-tem for biomaterials(Rickne,2000)and is,of course, an ever bigger problem in cases where the TIS concept is only an analytical tool for the researcher.Moreover, in early phases networks are usually undeveloped and/or informal and TIS-specific institutions may not yet exist. In instances like this,the structural mapping must be an iterative process,in which additional pieces of informa-tion are added as the analysis proceeds.

Identifying the structural components of the system provides a basis for the following step,which constitutes the core of the analysis:analyzing the TIS in functional terms.

6.Step3:mapping the functional pattern of the TIS

Thefirst step of a TIS analysis in functional terms is to describe the“functional pattern”of the TIS.This anal-ysis aims at ascertaining to what extent the functions are currentlyfilled in that TIS,i.e.to analyze how the TIS is behaving in terms of a set of key processes.This step has no normative features;assessing the“goodness”

19Institutional factors may be even more distant from the focal TIS. Geels(2004)uses the concept of“technological landscapes”,which influences many different TISs.Examples include the greenhouse effect discussion.of the current functional pattern will be dealt with later in the paper.The functional pattern of a TIS is likely to differ from that of other TISs and is also likely to change over time.Thus,the concept should not be inter-preted as implying that the pattern is either repeated or optimal.

In the following,we will explain each of these func-tions.As described above,they have been synthesized from a number of different system approaches to inno-vation and have been applied and further developed by ourselves and other researchers.We begin by explaining the content of the function.We will then give a brief illus-trative example from various case studies that we have undertaken and examples of indicators that may reflect the extent to which the function is fulfilled.Of course, it is not possible to come up with an exactfigure but the analyst has to make a composite judgment based on both qualitative and quantitative data.Exactly how that is done should be made explicit.

6.1.Knowledge development and diffusion

This is the function that is normally placed at the heart of a TIS in that it is concerned with the knowledge base of the TIS(globally)and how well the local TIS performs in terms of its knowledge base and,of course,its evolu-tion.The function captures the breadth and depth of the current knowledge base of the TIS,and how that changes over time,including how that knowledge is diffused and combined in the system.

We can distinguish between different types of knowl-edge(e.g.scientific,technological,production,market, logistics and design knowledge)and between different sources of knowledge development,for example R&D (Bijker,1995;Edquist and Johnson,1997;Hughes,1983; Nelson,1992),learning from new applications,produc-tion,etc.(Bijker,1995;Edquist and Johnson,1997; Hughes,1990;Lundvall,1992b)and imitation(Edquist and Johnson,1997;Nelson,1992).20

An illustrative example is that of the emerging TIS for solar cells in Germany(Jacobsson et al.,2004). Initially,the type of knowledge development was lim-ited to the scientific/technologicalfield and the source was R&D on various competing designs for solar cells. The knowledge base was subsequently broadened as the system expanded along the entire value chain.First, 20It should be noted that part of this knowledge development takes place in the form of entrepreneurial experimentation.In the function ‘entrepreneurial experimentation’,however,we focus on the uncer-tainty reducing effects of these experiments rather than their results in terms of knowledge development.application-specific knowledge was developed down-stream asfirms experimented with solar cells as a building element.Part of the knowledge development took place in schools of architecture where“solar archi-tects”developed new design concepts.Second,upstream technological knowledge was enhanced through R&D performed by the capital goods industry.A significant aspect of that knowledge development was,however, also a very practical and problematic learning process to build automated production lines for the manufacturing of solar cells.

The current level and dynamics of the function could be measured by a range of indicators,includ-ing for instance bibliometrics(citations,volume of publications,orientation);number,size and orientation of R&D projects;number of professors;number of patents;assessments by managers and others;and learn-ing curves.

6.2.Influence on the direction of search

If a TIS is to develop,a whole range offirms and other organizations have to choose to enter it.There must then be sufficient incentives and/or pressures for the orga-nizations to be induced to do so.The second function is the combined strength of such factors.It also covers the mechanisms having an influence on the direction of search within the TIS,in terms of different competing technologies,applications,markets,business models, etc.These factors are not,of course,controlled by one organization–and definitely not by the state(apart from the case of regulations,etc.)–but their strength is the combined effect of,for example:

•visions,expectations(van Lente,1993)and beliefs in growth potential:

◦incentives from changing factor and product prices (Dosi et al.,1990);

◦growth occurring in TISs in other countries;

◦changes in the“landscape”(Geels,2004),e.g.

demographic trends and climate change debates;

and

◦development of complementary resources (Dahm´e n,1988),

•actors’perceptions of the relevance of different types and sources of knowledge,21

21For example,actors are more likely to look for new knowledge within their current technological frame(McLoughlin et al.,2000)or paradigm(Dosi,1982).•actors’assessments of the present and future techno-logical opportunities and appropriability conditions (Breschi et al.,2000),

•regulations and policy(Lundvall,1992b;Porter, 1990),

•articulation of demand from leading customers(e.g. Dosi et al.,1990;von Hippel,1988;Carlsson and Jacobsson,1993),

•technical bottlenecks or“reverse salients”(Rosenberg,1976;Bijker,1995;Hughes,1983; Lundvall,1992b),and

•crises in current business.

Wind turbines in Germany(in the early phase of sys-tem evolution)is an illustrative case in point,wherefirms experienced a range of incentives to enter the indus-try(Bergek and Jacobsson,2003).In several cases,the firms’existing markets were in recession at the same time as there was a Californian wind turbine boom and an associated expansion of the Danish wind turbine industry.These latter developments gave clear signals about the attractiveness of the future wind turbine mar-ket(i.e.expectations of future markets).Locally,there was a“green”demand from some utilities and environ-mentally concerned farmers(articulation of demand). Federal R&D policy subsidized not only R&D in many competing designs but also investment in wind turbines in a number of demonstration programs(regulation).

We suggest that this function can be measured,or at least indicated,by qualitative factors of the following types:

•beliefs in growth potential,

•incentives from factor/product prices,e.g.taxes and prices in the energy sector,

•the extent of regulatory pressures,e.g.regulations on minimum level of adoption(“green”electricity cer-tificates,etc.)and tax regimes,and

•the articulation of interest by leading customers.

6.3.Entrepreneurial experimentation

A TIS evolves under considerable uncertainty in terms of technologies,applications and markets.This uncertainty is a fundamental feature of technological and industrial development and is not limited to early phases in the evolution of a TIS but is a characteristic of later phases as well(Rosenberg,1996).From a social perspective,the main source of uncertainty reduction is entrepreneurial experimentation,which implies a prob-ing into new technologies and applications,where many will fail,some will succeed and a social learning processwill unfold(Kemp et al.,1998).22A TIS without vibrant experimentation will stagnate.

An analyst needs to map the number and variety of experiments taking place in terms of,for example:•number of new entrants,including diversifying estab-lishedfirms,

•number of different types of applications,and •the breadth of technologies used and the character of the complementary technologies employed.

To continue with the German wind turbine case in the early phase of its evolution,it is clear that the diversity in experiments undertaken was its main characteristic (Bergek and Jacobsson,2003).In the period1977–1991, a large number of industrialfirms and a range of aca-demic organizations received federal R&D funding for the development or testing of a variety of turbine sizes and designs.As a result of some of these experiments,at least14firms entered wind turbine production,including academic spin-offs,diversifying medium-sized mechan-ical engineeringfirms and large aerospacefirms,all of which brought different knowledge and perspectives into the industry.

6.4.Market formation

For an emerging TIS,or one in a period of transforma-tion,markets may not exist,or be greatly underdeveloped (Carlsson and Stankiewicz,1995;Dahm´e n,1988;Galli and Teubal,1997;Nelson,1992;Porter,1990).Market places may not exist,potential customers may not have articulated their demand,or have the capability to do so, price/performance of the new technology may be poor, and uncertainties may prevail in many dimensions.Insti-tutional change,e.g.the formation of standards,is often a prerequisite for markets to evolve(Hughes,1983).

Market formation normally goes through three phases with quite distinct features.In the very early phase,“nursing markets”need to evolve(Erickson and Maitland,19)so that a“learning space”is opened up, in which the TIS canfind a place to form(Kemp et al., 1998).The size of the market is often very limited.This nursing market may give way to a“bridging market”(Andersson and Jacobsson,2000),which allows for vol-

22It should be noted the word“entrepreneurial”does not refer only to new or smallfirms,but to the more general Schumpeterian notion of an“entrepreneurial function”(i.e.making new combinations).This function may befilled by any type of actor,including large,established firms diversifying into the new technology.umes to increase and for an enlargement in the TIS in terms of number of actors.Finally,in a successful TIS, mass markets(in terms of volume)may evolve,often several decades after the formation of the initial market.

To understand the sequence of the formation of mar-kets,we need to analyze both actual market development and what drives market formation.The timing,size and type of markets that have actually formed,are normally quite easy to measure.For example,we could describe a market for wind turbines in terms of the number of turbines and/or the wind power capacity installed in a particular year and in terms of the distribution between different customer groups(e.g.farmers and energy com-panies).

It is more difficult to analyze what drives that forma-tion,and the analyst needs to have in-depth knowledge of the TIS to do so.We will illustrate the multitude of factors that may drive or hinder market formation with the case of the Swedish“mobile data”TIS(see Lindmark and Rickne,2005).In this case,markets are often global,but the home market is still strategically important to test new concepts and products,to learn, and to obtain early revenues.Swift market formation is,therefore,of essence to any national TIS.However, in the Swedish market,corporate and governmental use is slow.Sluggish procurement procedures and unartic-ulated demand cause great uncertainty about current or future user needs.In addition,Sweden lags behind coun-tries such as Japan and Korea with regard to the dominant consumer market.Indeed,as of2006Sweden had a low rate of adoption of mobile data services,much due to unwillingness of operators to cannibalize current cash cows within mobile telecommunication,inflexible pric-ing systems,lack of standards for platforms,problems with complementary technologies and proprietary solu-tions.

The analyst needs to assess what phase the mar-ket is in(nursing,bridging,mature),who the users are and what their purchasing processes look like,whether the demand profile has been clearly articulated and by whom,if there are institutional stimuli for market for-mation or if institutional change is needed.Indicators to trace these developments include readily available facts (as indicated above)on market size and customer groups as representing what has been achieved,but also quali-tative data on e.g.actors’strategies,the role of standards and purchasing processes.

6.5.Legitimation

Legitimacy is a matter of social acceptance and com-pliance with relevant institutions:the new technologyand its proponents need to be considered appropriate and desirable by relevant actors in order for resources to be mobilized,for demand to form and for actors in the new TIS to acquire political strength.Legitimacy also influences expectations among managers and,by impli-cation,their strategy(and thus the function‘influence on the direction of search’).

As is widely acknowledged in organization theory, legitimacy is a prerequisite for the formation of new industries(Rao,2004)and,we would add,new TISs(cf. Bijker,1995;Carlsson and Stankiewicz,1995;Edquist and Johnson,1997;Hughes,1983).Legitimacy is not given,however,but is formed through conscious actions by various organizations and individuals in a dynamic process of legitimation,which eventually may help the new TIS to overcome its“liability of newness”(Zimmerman and Zeitz,2002).However,this process may take considerable time and is often complicated by competition from adversaries defending existing TISs and the institutional frameworks associated with them.

Although the process of legitimation is often closely associated with institutional alignment,“manipulation”of the rules of the game is only one of several possi-ble alternative legitimation strategies;other alternatives include“conformance”(following the rules of the exist-ing institutional framework,e.g.choosing to follow an established product standard)and“creation”(develop-ing a new institutional framework)(Suchman,1995; Zimmerman and Zeitz,2002).With respect to the lat-ter,however,a new TIS seldom emerges in a vacuum, but instead is often subjected to competition from one or more established TIS.In such cases,some type of manipulation strategy is usually needed.

Mapping the functional dynamics of‘legitimation’includes analyzing both the legitimacy of the TIS in the eyes of various relevant actors and stakeholders(not least the ones that could be expected to engage in the develop-ment of the new technology,e.g.potential capital goods suppliers and buyers),and the activities within the sys-tem that may increase this legitimacy.So,we need to understand:

•the strength of the legitimacy of the TIS,in particu-lar whether there is alignment between the TIS and current legislation and the value base in industry and society;

•how legitimacy influences demand,legislation and firm behavior;and

•what(or who)influences legitimacy,and how.

An interesting illustration of the process of legitima-tion is provided by the case of solar cells in Germany (see Jacobsson and Lauber,2006).After unsuccessful efforts to convince the federal government to launch a nationwide regulatory change in favor of the diffusion of solar cells in the early1990s,a number of activists and interest organizations began lobbying work at the L¨a nder and local levels.After much effort,most L¨a nder expressly allowed cost-covering contracts between sup-pliers of(very expensive)solar power and local utilities. Several dozen cities subsequently opted for this model, which revealed a wide public interest in increasing the rate of diffusion–the legitimacy of solar power was made apparent.Various organizations could later point to this interest when they lobbied for a program to develop yet larger markets for solar cells,now at the federal level.

6.6.Resource mobilization

As a TIS evolves,a range of different resources needs to be mobilized(Carlsson and Stankiewicz,1995; Dahm´e n,1988;Edquist and Johnson,1997;Hughes, 1983;Lundvall,1992b;Nelson,1992;Porter,1990; Rickne,2000).Hence,we need to understand the extent to which the TIS is able to mobilize competence/human capital through education in specific scientific and technologicalfields as well as in entrepreneurship,man-agement andfinance,financial capital(seed and venture capital,diversifyingfirms,etc.),and complementary assets such as complementary products,services,net-work infrastructure,etc.23

As an illustration of this function,we will use a recent analysis of the Swedish security sensor TIS(Oltander and Perez Vico,2005).The mobilization of human resources was found to be strong,partly following a recent reduction of personnel at the Swedish telecom-munication company Ericsson.However,in specific knowledgefields,such as radar and sonar technology, there was a resource shortage,explained by an absence of university education in thesefields.The mobilization offinancial resources was more troublesome.In addi-tion to a generally weak Swedish seed capital market, there were also difficulties in attracting venture capi-tal,resulting from(a)a cautious VC market in general, and(b)a belief that Swedish start-ups will have prob-lems competing internationally with USfirms.In larger organizations(e.g.Saab Bofors Dynamics and Ericsson Microwave)there were some perceived difficulties in raising funding for internal R&D projects,because of an 23Here,we follow the reviewed literature on innovation systems, where resource mobilization is treated as a process separate from the other functions.absence of strong customers and the ongoing transition from military to civilian markets.

There are thus various ways for analysts to measure resource mobilization:

•rising volume of capital,

•increasing volume of seed and venture capital,•changing volume and quality of human resources(e.g. number of university degrees),and

•changes in complementary assets.

In the analysis referred to above,Oltander and Perez Vico(2005)used quantitative measures,such as the num-ber of graduates from sensor-related education(absolute and per capita)in comparison to Germany and Israel and the number of venture capitalfirms with holdings in the security sensor sector,together with qualitative data based on interviews,such as perceptions about the sup-ply of human resources and the VCfirms’interest in the Swedish security sensor sector.

6.7.Development of positive externalities

The systemic nature of the innovation and diffusion process strongly suggests that the generation of positive external economies is a key process in the formation and growth of a TIS.24These external economies,or free utilities,may be both pecuniary and non-pecuniary (Scitovsky,1954).

Entry of newfirms into the emerging TIS is central to the development of positive externalities.First,new entrants may resolve at least some of the initial uncertain-ties with respect to technologies and markets(Lieberman and Montgomery,1988),thereby strengthening the func-tions‘influence on the direction of search’and‘market formation’.Second,they may,by their very entry,legit-imate the new TIS(Carroll,1997).New entrants may also strengthen the political power of advocacy coalitions that,in turn,enhance the opportunities for a success-ful legitimation process.An improved legitimacy may, in turn,positively influence changes in four functions:‘resource mobilization’,‘influence on the direction of search’,‘market formation’and‘entrepreneurial exper-imentation’.Third,the greater the number and variety of actors in the system,the greater are the chances for new combinations to arise,often in a way which is unpredictable(Carlsson,2003).An enlargement of the actor base in the TIS therefore enhances not only the opportunities for each participatingfirm in the system

24See in particular Marshall(1920)and Porter(1990).to contribute to‘knowledge development and diffusion’but also for thefirms to participate in‘entrepreneurial experimentation’.

Hence,new entrants may contribute to a process whereby the functional dynamics of the TIS are strength-ened,benefiting other members of the TIS through the generation of positive externalities.This function is thus not independent but works through strengthening the other six functions.It may,therefore,be seen as an indi-cator of the overall dynamics of the system.25 These dynamics may be enhanced by the co-location offirms.Marshall(1920)discussed economies that were external tofirms but internal to location,and outlined three sources of such economies:

•Emergence of pooled labor markets,which strengthen the‘knowledge development and diffusion’function, in that subsequent entrants can access the knowledge of early entrants by recruiting their staff(and vice versa as time goes by).

•Emergence of specialized intermediate goods and ser-vice providers;as a division of labor unfolds,costs are reduced and further‘knowledge development and diffusion’is stimulated by specialization and accumu-lated experience.26

•Informationflows and knowledge spill-overs,con-tributing to the dynamics of‘knowledge development and diffusion’.

In sum,the analyst needs to capture the strength of these functional dynamics by searching for external economies in the form of resolution of uncertainties, political power,legitimacy,combinatorial opportunities, pooled labor markets,specialized intermediates,as well as information and knowledgeflows.

To refer again to the German wind turbine case,we will mention two forms of positive externalities.First, new entrants into the wind turbine industry,as well as into wind power production,increased the political power of the advocates of wind energy so that they could win against opposing utilities in several courts and defend a favorable institutional framework(Jacobsson and Lauber,2006).Second,as the market increased,spe-

25We are grateful to Professors Ruud Smits and Marko Hekkert on this point.The dynamics are,of course,enhanced by the interdepen-dencies of the functions,as was pointed out above.As the system moves into a growth phase characterized by positive feedback loops, these interdependencies are clearly seen.

26See Smith(1776),Young(1928),Stigler(1947),Rosenberg(1976), and Maskell(2001).For a case study of mobile data in Western Sweden, see Holm´e n(2001).

cialized suppliers emerged,with the consequence that barriers to entry for yet morefirms were lowered(Bergek and Jacobsson,2003).

7.Step4:assessing the functionality of the TIS

and setting process goals

The analyst now has a description of the dynamics of these seven key processes,or functions,in the evo-lution of a TIS,as well as a tentative assessment of the strengths and weaknesses of these processes.However, the functional pattern does not in itself tell us whether the TIS is well functioning or not;that a particular func-tion is weak does not always constitute a problem,nor is a strong function always an important asset.In order to assess system functionality–i.e.not how,but how well the system is functioning–we need ways to evaluate the relative“goodness”of a particular functional pattern. This is,of course,the same problem as we alluded to in Section2.2,i.e.evaluating the“goodness”of a particular structure.The advantage with a functional analysis is that we can systematically address the issue of“goodness”in terms of the seven clearly specified key processes.

Although this is a step forward,we face here one of the major challenges for analysts and policy makers,a challenge that needs to be dealt with further in research and in learning processes among practitioners.So far,we have identified two bases for an assessment:(1)the phase of development of the TIS,and(2)system comparisons. Both are associated with different types of problem and in order to balance each other’s weaknesses,they should probably be used in combination.

7.1.The phase of development

We have earlier suggested that it is useful to distin-guish between a formative phase and a growth phase in the development of a TIS and that it is plausible that the definition of“functionality”differs between these phases (Bergek and Jacobsson,2003;Jacobsson and Bergek, 2004).The analyst can then raise the question whether functionality matches the needs of that particular phase or the need of the next phase(if it is judged to be desirable that the TIS is to move in that direction).In other words, the functional pattern,i.e.how the functions,or key pro-cesses,are performed and improved,can be analyzed with respect to the requirements of each phase.

Although it is not always a straightforward exercise, the analyst can use a number of indicators to know whether or not a TIS is in a formative phase.In this phase the constituent elements of the new TIS begin to be put into place,involving entry of somefirms and other orga-nizations,the beginning of an institutional alignment and formation of networks.A rudimentary structure is formed.Apart from exhibiting rudimentary structural components,the formative phase may be indicated by, for instance:

•the time dimension,where we rarely escape formative periods that are shorter than a decade(yet they can last for many decades,as in the case of solar cells);•large uncertainties prevailing as regards technologies, markets and applications;

•price/performance of the products being not well developed;

•a volume of diffusion and economic activities that is but a fraction of the estimated potential;•demand being unarticulated;and

•absence of powerful self-reinforcing features(positive feedbacks)and weak positive externalities.

A common error made by analysts is to judge a TIS that is in a formative phase by using criteria that are more suitable for evaluating a system which is in a growth phase.For example,the formative phase is not char-acterized by a rapid rate of diffusion or rapid growth in economic activities.On the contrary,the volume of activities is small and many experiments take place–the TIS is in a process of formation.Yet,in several cases we know of–renewable energy technologies and wood manufacturing,for instance–emerging TISs were eval-uated,by policy makers and others,by the volume(level) of economic activities.Of course,this led to a great deal of frustration and a feeling of disappointment and failure. By applying other criteria,more suited for a formative phase,a quite different interpretation would be made.

In particular,the formative phase is characterized by high uncertainty in terms of technologies and mar-kets(Kemp et al.,1998;Van de Ven,1993),and the key words are therefore experimentation and vari-ety creation.This requires extensive‘entrepreneurial experimentation’in such a way that‘knowledge develop-ment’occurs within a number of different technological approaches and applications.For this to take place,‘influence on the direction of search’and‘resource mobi-lization’must stimulate not only entry offirms but also ventures embarked upon in many directions.Moreover, a process of‘legitimation’must start,helping to over-come the“liability of newness”associated with new actors and technologies and eventually leading to insti-tutional change.Finally,‘knowledge development’is to a large extent dependent on cooperation between actors(in networks),especially between suppliers and buyers,which require‘market formation’.Thus,eitherestablished markets need to be open to new technolo-gies/products,or new niches need to be identified and stimulated.

At some point in time,the TIS may be able to“change gear”and begin to develop in a self-sustaining way as it moves into a growth phase.In this phase,the focus shifts to system expansion and large-scale technology diffusion through the formation of bridging markets and subsequently mass markets;hence the need for‘resource mobilization’increases by orders of magnitude.Yet, it is normally not self-evident which applications will generate such markets,so a breadth of‘entrepreneurial experimentation’must be kept up.As and when the growing TIS catches the attention of actors in com-peting TISs,‘legitimation’may become even more important.

Although the phase thus presumably matters for how we assess functionality,we want to emphasize that this does not imply that all TISs follow exactly the same development pattern.Indeed,the whole point of the func-tional dynamics approach is that TISs differ so much that there are no“one sizefits all”policy implications.Hence, although some features in TIS development are arguably common to many innovation systems,we fully acknowl-edge that the determining factors,time frames,etc.,differ between cases.We also acknowledge that more research is needed to establish the nature of the different phases. This implies that we must be careful not to specify a “desired”functional pattern too rigidly,and need to be open for reformulation and iteration in the process of analysis.

7.2.Comparisons between TISs

Comparing the focal TIS with other TISs,across regions or nations,is a powerful way of improving the understanding for decision makers(see e.g.Rickne (2000)for biomaterials in Sweden,Massachusetts and Ohio,and Bergek and Jacobsson(2003)for wind turbines in Sweden,the Netherlands and Germany). Researchers and policy makers involved with a partic-ular innovation system thus ought to perform analyses of similar systems being developed elsewhere or of sys-tems in related areas.Most importantly,they need to address the question of how these other systems are performing in order to gauge correctly not only what development it is reasonable to expect of their focal TIS but also in identifying the critical functions.For instance, in our earlier work on the TIS centered on wind tur-bines(Bergek and Jacobsson,2003),Dutch researchers pointed to weak market formation in the Netherlands as an explanation of the perceived unsatisfactory develop-ment of the TIS.Yet,in its formative phase,the Dutch home market was larger(even in absolute terms)than the German and much larger than the Swedish one.A search for an explanation would then shift to how an initially favorable position was not leveraged to propel the sys-tem into a growth phase ahead of the competing German TIS.

Based on the phase analysis and/or one or more comparative analyses,a tentative conclusion regarding functionality of the TIS may be drawn,that is,in rela-tion to what it is reasonable to expect taking the phase of development and/or the comparison with other sys-tems into consideration.It is then also possible to specify policy goals in terms of how the functional pattern should develop in order to reach higher functionality, i.e.towards a“targeted”functional pattern.Such goals (e.g.broaden the knowledge base or widen the range of experiments)can be seen as process goals.Hence, policy goals may be expressed in terms of the seven key processes in contrast tofinal goals(such as growth).Pro-cess goals have the advantage for policy makers in that they are“closer”to the various instruments that can be used,and they also make it easier to evaluate how well a specific policy works.In particular,in early phases of developmentfinal goals may be close to impossible to define,since the uncertainty regarding what the TIS may be able to achieve in the long term,also regarding and what it is desirable to achieve,is very high.

8.Step5:identify inducement and blocking mechanisms

There are many reasons for expecting that the envi-ronment is biased,and will remain biased,in favor of established TISs.27New TISs may consequently exhibit weak functional dynamics and develop slowly,or in a stunted way.The functional dynamics may be weak for a number of reasons.These may be found in features of the structural components of the emerging TIS and in the larger context surrounding it.28This larger context includes the sector in which the new TIS operates,e.g. the electric power sector for the emerging TIS centered on solar cells,but also factors that go beyond that sector. For instance,the reaction,or lack of it,to global warming 27Jacobsson and Johnson(2000)and Johnson and Jacobsson(2001) elaborate on various types of“blocking mechanisms”.See also Unruh (2000)for an extensive review of mechanisms locking us into a carbon economy and Walker(2000)for a case study on entrapment in a large technological innovation system.

28See e.g.Geels(2004),who distinguishes between“regime”and “landscape”levels,where“regime”is broadly equivalent to the sector.acts either as an inducement mechanism29or as a block-ing mechanism in many sectors,and this influences many emerging TISs.

What is being achieved in the TIS is therefore only in part a result of the internal dynamics of the TIS.Exoge-nous factors also come into play,influencing the internal dynamics.Myrdal(1957,p.18)showed a keen under-standing of the interplay between internal and external sources of dynamics and even suggested that“the main scientific task is...to analyze the causal inter-relations within the system itself as it moves under the influence of outside pushes and pulls and the momentum of its own internal processes”.

From a policy perspective,it is particularly important to understand the blocking mechanisms that shape the nature of the dynamics.These could,for instance,be of the following types:

•The proponents of the new technology may be orga-nizationally too weak to contribute to a‘legitimation’process;they may,for example,lose in a“battle over institutions”as they attempt to achieve institutional alignment to the new technology.Unaligned institu-tions may then lead to poor‘market formation’that,in turn,limits the strength of the‘influence on the direc-tion of search’and‘entrepreneurial experimentation’functions.

•Underdeveloped capabilities among potential cus-tomers may lead to an absence,or poor articulation, of demand which results in a poor development of the dynamics of‘market formation’,‘influence on the direction of search’and‘entrepreneurial experimen-tation’.

•Networks may fail to aid new technology simply because of poor connectivity between actors.Tight networks may also through a“lock-in”effect have an‘influence on the direction of search’among potential suppliers and customers away from the new TIS.

As is evident from these examples,there may be quite different things that block the development of functions. The path to achieving a higher functionality may,there-fore,be littered by a range of such blocking mechanisms. These may operate in a formative stage,but they may also obstruct a transition towards a more self-sustained TIS, i.e.one which is to an increasing extent driven by its own

29A case in point is the transnational legislation concerning tradable emission permits which may influence investment decisions in many TISs.momentum rather than by outside pushes or pulls in the form of policy.

It is empirically possible,and very useful,to map the relationship between inducement/blocking mechanisms and functional patterns.We will illustrate this with the example of the emerging TIS for“IT in home care”, which,as was noted above,is defined by the application of a generic technology(IT)to a particular application: care of elderly and ill people in their homes instead of in a hospital.For a number of reasons(demographic,public sector funding restrictions,technological opportunities, etc.),this is a TIS which is thought of as having a large growth potential.30However,it is still in a formative phase,as judged by,for instance,the following features:•There are no software standards and the technical uncertainty is high.

•The number offirms supplying IT solutions is small.•Markets are small,and characterized by high uncer-tainty,e.g.with respect to applications and choice of software.

•The advocacy coalition for the TIS is weak.•The demand is poorly articulated by customers with poorly developed capabilities.

In this formative phase,the functional pattern can be summarized as follows:

•‘Knowledge development and diffusion’:pilot projects in some of the290counties and21county councils,

•‘Market formation’:local pilot projects constitute “nursing markets”,albeit fragmented,

•‘Influence on the direction of search’:government R&D funding,opportunities tofind new markets, awards,

•‘Entrepreneurial experimentation’:a few ITfirms have developed solutions,

•‘Resource mobilization’:EU and government R&D funding,some co-funding byfirms,poor adjustment by the higher educational sector,

•‘Legitimation’:partly underdeveloped legitimacy, especially among care providers,and

•‘Development of positive externalities’:early stage of cluster formation in three cities.31

30This is the judgment of VINNOV A(the Swedish Agency for Inno-vation Systems).

31Since we have not been able to determine how this function influ-ences the other functions,we will not include it in the following discussion.Fig.2.Inducement and blocking mechanisms as well as policy issues in the case of“IT in home care”.N.B.:The function“Development of positive externalities”has been excluded from the discussion(see footnote30).

The current functional pattern is shaped by both inducement and blocking mechanisms(see Fig.2). There are two significant inducement mechanisms:a belief in growth potential and government R&D pol-icy.The former is driven by a range of factors, as was mentioned above.32This inducement mecha-nism has a bearing on the function‘influence on the direction of search’among both care providers(e.g. county councils)and suppliers(ITfirms),as well as on the dynamics of‘market formation’(nursing mar-kets)and‘entrepreneurial experimentation’.The latter inducement mechanism both signals attractiveness and provides resources for research and experiments.Hence, it strengthens the functional dynamics of‘influence on the direction of search’and‘legitimation’,as well as ‘resource mobilization’and‘knowledge development and diffusion’.33

The blocking mechanisms are,however,strong and manifold.‘Market formation’is blocked by an absence of standards(which leads to a fragmented market), two factors that reflect poor awareness and capabilities

32These include demographic changes with a larger share of elderly people in the population,public sector funding restrictions and emerg-ing technological opportunities.

33Indirectly,it also strengthens‘entrepreneurial experimentation’as a consequence of its positive influence on the direction of search.among potential customers(leading to poor articula-tion of demand)and an associated lack of knowledge among suppliers of IT solutions of customer needs. Additionally,‘entrepreneurial experimentation’,‘influ-ence on the direction of search’and‘legitimation’, are each blocked by two factors.These three have a common blocking mechanism in the form of a lack of capability and a poor articulation of demand.This is strengthened by an additional but different factor in each case(uncertainties of customer needs,lack of standard software solutions and a weak advocacy coali-tion).

Some mechanisms block several functions.In par-ticular,a poor articulation of demand(due to lack of capability)blocks not only the three functions mentioned above but also‘market formation’.Moreover,functions are not independent,but rather tend to reinforce one another.A poor‘market formation’affects negatively both‘entrepreneurial experimentation’and‘influence on the direction of search’,whereas little‘entrepreneurial experimentation’negatively influences‘resource mobi-lization’and‘knowledge development and diffusion’. This means that the impact of blocking mechanisms is magnified by such interdependencies.Clearly,it could be argued that policy must focus on reducing the strength of the blocking mechanisms that have such a pervasive effect.

9.Step6:specify key policy issues

Process goals were defined in the fourth step above. Having made explicit both the reasons for setting these specific process goals and how to measure whether the goals are reached,we can now begin to spec-ify the key policy issues related to the mechanisms that block or induce a development of a desir-able functional pattern.We argue that policy should aim at remedying poor functionality in relevant TISs by strengthening/adding inducement mechanisms and weakening/removing blocking mechanisms.In doing so,we take a step away from the traditional“market failure”rationale for policy interventions into inno-vation processes and focus on“system failure”in terms of functional weaknesses rather than structural deficiencies.

We will continue to use“IT in home care”as an illustrative case and refer to the fourth column in Fig.2,where we list six specific policy issues con-nected to removing or reducing the strength of the many blocking mechanisms.Thefirst three of these focus on the potential customers(care providers)and are aimed at removing the most pervasive blocking mechanisms:

•how to raise user capability so that demand is articulated and uncertainties reduced for potential suppliers;

•how to support users in order to(a)increase their knowledge of the benefits of IT in home care and of ways to distribute the costs and benefits over orga-nizational boundaries and(b)diffuse knowledge of the outcome of early experiments in order to reduce uncertainties further;and

•how to support experimentation with new applica-tions in order to reduce the level of uncertainty of needs.

In addition to these three issues,we can also deduce three additional ones,relating to one blocking mecha-nism each:

•how to develop standards in order to move from a frag-mented market of290local councils and21county councils;

•how to alter research and education at universities in order to allow for‘resource mobilization’in terms of staff with relevant background;and

•how to support a weak advocacy coalition so that it can improve the process of‘legitimation’.

Hence,by analyzing weaknesses in the functional pat-tern of the TIS(i.e.“what is actually going on”),we can identify the key blocking mechanisms that,in turn,lead us to a specification of the relevant policy issues.

10.Summary and discussion

The objective of this paper has been to make the inno-vation system approach more useful to innovation system researchers and policy makers by presenting a practical scheme of analysis that can be used to identify the key policy issues and set goals in any given TIS.We have outlined six“steps”in such a scheme.The core of this operationalization of the innovation system perspective referred to the description and evaluation of seven key processes,here labeled functions,in the evolution of a TIS.The main benefit of this framework is that it focuses on what is actually achieved in the system,rather than on the structure of the system(the goodness of which is difficult to evaluate without referring in a systematic way to these processes).

The main application of the framework is the iden-tification of“system failures”or weaknesses,expressed in functional terms.Policy makers can also define pro-cess goals of their intervention in terms of an altered functional pattern,i.e.an altered way in which the seven key processes are operating.By explaining the nature of these processes in terms of the outcome of a balance between various inducement and blocking mechanisms,the functional dynamics approach can then be used as a focusing device for policy makers that seek to identify the key policy challenges for moving a specific TIS towards these process goals.34As the functions approach includes a systematic mapping of a larger number of key processes than in most previous research,its use implies that a wider range of possi-ble policy challenges may be identified.In consequence, the functions approach further strengthens the tendency of an innovation system perspective to open up richer 34Here we need to remind ourselves of what Charles E.Lindblom wrote many years ago:“[p]olicy-making is a process of successive approximation to some desired objectives in which what is desired itself continues to change under reconsideration...Making policy is at best a very rough process.Neither social scientists,nor politicians,nor public administrators yet know enough about the social world to avoid repeated error in predicting the consequences of policy moves.A wise policy-maker consequently expects that his policies will achieve only part of what he hopes and at the same time will produce unanticipated consequences he would have preferred to avoid.If he proceeds through a succession of incremental changes,he avoids serious lasting mistakes in several ways”(Lindblom,1959,p.86).See also Smits and Kuhlmann (2002)on this point.and more difficult innovation policies(Bergek et al., 2007b).35

In the course of this scheme of analysis,we have emphasized the many sources of uncertainties,not only those that are inherent in the process of indus-trial development but also those additional sources facing the analyst in search of useful methods and tools.We are still at an early stage in our understand-ing of how TISs emerge and develop and we need to learn a lot more about methods such as indicators and,most importantly,about how to assess functional-ity.In relation to the latter issue,three points can be made.

First,there is,indeed,a need for more research on how to assess TIS functionality,i.e.the“goodness”of different functional patterns.In this paper,we have presented two ways forward:assessment based on the requirements of particular phases of development and assessment based on comparisons between systems.It would,however,be of benefit for the area of innova-tion system analysis if generally applicable assessment models could be developed.

Second,a promising way forward towards this aim seems to be an assessment based on the phase of devel-opment of the system.In particular,we need to better understand the formative phase and establish to what extent,and in what ways,the functional requirements of that phase differ from those of later phases.Although we acknowledge that systems are different and develop in different ways,we do not think that the variation is infinite.It ought to be possible–and fruitful–to develop a taxonomy of“archetypal”development paths with associated functional patterns by empirical investi-gation.

Third,such taxonomy may also be needed in order to better inform policy makers under what conditions a transition between the formative phase and a growth phase may occur and how the foundation for such a tran-sition can be laid.A transition would involve leveraging the investment made in the formative phase by inducing a“change of gear”in the development of the TIS.As shown in the case of wind power in the Netherlands and solar heating in Sweden,a successful formative phase does not necessarily lead to a successful growth phase (Jacobsson and Bergek,2004).

Finally,a scheme of analysis of this kind builds on present knowledge and it is therefore by no means a

35It also allows for comparisons between different TISs,which are based not on structural characteristics but on the underlying mecha-nisms of the innovation process and their changes in each system.finished product.Only by a systematic learning process can we improve our understanding of the opportunities and limitations of innovation system analysis and policy making.Therefore,we expect further empirical stud-ies–in combination with the research outlined above –to induce several revisions of the framework in the future.

Acknowledgements

We are grateful to the editor and three reviewers for constructive comments.Thanks also to Sven-Gunnar Edlund at VINNOV A for inspiring us to do this work and to Cecilia Sj¨o berg who contributed to the case of “IT in home care”.

Appendix A.Matching functions in the

literature36

As far as we know,there have been nine attempts to identify functions that need to befilled for an innovation system to evolve and perform well:Galli and Teubal (1997),Johnson(1998),Johnson(2001),Rickne(2000), Johnson and Jacobsson(2001),Bergek(2002),Bergek and Jacobsson(2003),Carlsson et al.(2005),Liu and White(2001),and Hekkert et al.(2007).In addition, Edquist(2004)lists a number of“activities”,defined as “those factors that influence the development,diffusion, and use of innovation”(p.190),which is based on a similar comparison as ours.

In the table below,we have matched the functions suggested by these authors,excluding Liu and White (2001)since their framework is included in Edquist’s (2004)synthesis.37Most of these original functions are formulated as verbs,in contrast to the functions in this paper,which are formulated as nouns.This reflects a conscious choice on our part;as described above we want to emphasise the process nature of the functions and remove any notion of a particular actorfilling them. From the table,we can make the following observa-tions.

First,three of the functions are more or less identical in all lists:‘supply resources’,‘create knowledge’and ‘stimulate/create market’(except for Galli and Teubal, 1997),although the authors differ in the degree of 36This section is based on Bergek et al.(2005).

37We have also excluded one of the functions identified by Galli and Teubal(1997)–“policy-making”–since it refers to the activities by one particular type of actor(i.e.policy makers)and can be directed towards all functions.

A.Bergek et al./Research Policy37(2008)407–429425

detail provided for each function.38In this paper,these functions are labeled‘knowledge development and diffu-sion’,‘resource mobilization’,and‘market formation’, respectively.

Second,Bergek and Jacobsson’s(various)function ‘guide the direction of search’is an aggregate of four of Johnson/Bergek’s(1998,2001,2002)functions(by design)and corresponds,at least in part,to Hekkert et al.’s(2007)‘articulation of demand’(which is much broader than the demand articulation that others include in‘market formation’),to Carlsson and Jacobsson’s (2004)‘incentives’and to Rickne’s(2000)‘direct tech-nology,market and partner search’.Edquist’s(2004) function‘creating or changing institutions that provide incentives or obstacles to innovation’is broader and applies also to another function in this paper(see below). In this paper,all these functions are gathered under the label‘influence on the direction of search’.

Third,Bergek and Jacobsson’s(various)as well as Carlsson and Jacobsson’s(2004)function‘promoting positive externalities’is much broader than Galli and Teubal’s(1997),Johnson/Bergek’s(1998,2001,2002), Rickne’s(2000)and Hekkert et al.’s(2007),that focus on one source of external economies—diffusion of infor-mation/knowledge.Indeed,this function was developed a great deal in Jacobsson and Bergek(2004)and built yet further on in Carlsson and Jacobsson(2004).In this paper,we use the label‘development of external economies’for the broader concept.Knowledge diffu-

38For example,in Rickne(2000)the function‘supply resources’corresponds to four different functions.sion is included in the function‘knowledge development and diffusion’.

Fourth,Johnson/Bergek’s(1998,2001,2002)‘coun-teract resistance to change’,which refers primarily to the extremely important process of legitimation,may be linked to Hekkert et al.’s(2007)‘development of advocacy coalitions’,to Rickne’s(2000)‘legitimize technology andfirms’,to Edquist’s(2004)‘creating or changing institutions that provide incentives or obsta-cles to innovation’,and to Galli and Teubal’s(1997)‘design and implementation of institutions’.Carlsson and Jacobsson(2004)discuss this aspect under the head-ing of‘incentives’,and Bergek and Jacobsson(various) included it into‘guide the direction of search’.We are, however,hesitant to include advocacy coalitions in a function,since they are a kind of network,i.e.a struc-tural component.In this paper,this function is labeled ‘legitimation’.

Finally,the function‘promoting entrepreneurial experiments’mentioned by Carlsson and Jacobsson (2004)is not explicitly mentioned by any of the other authors,with the exception of Edquist(2004)who includes“enhancing entrepreneurship”in his function ‘creating and changing organizations needed’.In this paper,we use the label‘entrepreneurial experimentation’in order to emphasize that it is the creation of new combi-nations and variety that is in focus and that many different types of actors–not only new ones–may contribute to this function.

References

Andersson,B.A.,Jacobsson,S.,2000.Monitoring and assessing technology choice:the case of solar cells.Energy Policy28, 1037–1049.

Asheim,B.T.,Isaksen,A.,1997.Localisation agglomeration and inno-vation:towards regional innovation systems in Norway?European Planning Studies5,299–330.

Bergek,A.,2002.Shaping and exploiting technological opportuni-ties:the case of renewable energy technology in Sweden.PhD thesis.Department of Industrial Dynamics,Chalmers University of Technology,G¨o teborg.

Bergek,A.,Jacobsson,S.,2003.The emergence of a growth industry:

a comparative analysis of the German,Dutch and Swedish wind

turbine industries.In:Metcalfe,S.,Cantner,U.(Eds.),Change, Transformation and Development.Physica-Verlag,Heidelberg,pp.

197–227.

Bergek, A.,Berggren, C.,Tell, F.,2004.Do innovation strate-gies matter?A comparison of two electro-technical corporations 1988–1998.In:Proceedings of the Schumpeter Conference, Milano.

Bergek,A.,Jacobsson,S.,Carlsson,B.,Lindmark,S.,Rickne,A., 2005.Analyzing the dynamics and functionality of sectoral innova-tion systems—a manual,report delivered to VINNOV A,30March 2005.In:The Proceedings of the DRUID Tenth Anniversary Sum-mer Conference2005.June25–27,Copenhagen.

Bergek,A.,Jacobsson,S.,Hekkert,M.,2007a.Functions in inno-vation systems:a framework for analysing energy system dynamics and identifying goals for system-building activities by entrepreneurs and policy makers.In:Foxon,T.,K¨o hler,J., Oughton,C.(Eds.),Innovations for a Low Carbon Economy:Eco-nomic,Institutional and Management Approaches.Edward Elgar, Cheltenham.

Bergek,A.,Jacobsson,S.,Hekkert,M.,Smith,K.,2007b.Functionality of innovation systems as a rationale for,and guide to innovation policy.In:Smits,R.,Kuhlmann,S.,Shapira,P.(Eds.),Innovation Policy,Theory and Practice.An International Handbook.Elgar Publishers.

Bijker,W.E.,1995.Of Bicycles,Bakelites and Bulbs:Toward a Theory of Sociotechnical Change.The MIT Press,Cambridge. Breschi,S.,Malerba, F.,1997.Sectoral Innovation Systems: Technological Regimes,Schumpeterian Dynamics,and Spatial Boundaries.In:Edquist,C.(Ed.),Systems of Innovation.Tech-nologies,Institutions and Organizations.Pinter/Cassell Academic, London and Washington.

Breschi,S.,Malerba,F.,Orsenigo,L.,2000.Technological regimes and Schumpeterian patterns of innovation.The Economic Journal 110,388–410.

Cantner,U.,Graf,H.,2004.Cooperation and specialization in Ger-man technology regions.Journal of Evolutionary Economics14, 543–562.

Carlsson,B.,2003.The new economy:what is new and what is not?

In:Christensen,J.F.,Maskell,P.(Eds.),The Industrial Dynamics of the New Digital Economy.Edward Elgar,Cheltenham. Carlsson,B.,Jacobsson,S.,1993.Technological systems and eco-nomic performance:the diffusion of factory automation in Sweden.

In:Foray,D.,Freeman,C.(Eds.),Technology and the Wealth of Nations.Pinter Publishers,London,pp.77–94.

Carlsson,B.,Jacobsson,S.,1997.In search of a useful technology policy—general lessons and key issues for policy makers.In:Carls-son,B.(Ed.),Technological Systems and Industrial Dynamics.

Kluwer Press,Boston,pp.299–315.Carlsson,B.,Stankiewicz,R.,1991.On the nature,function,and composition of technological systems.Journal of Evolutionary Economics1,93–118.

Carlsson,B.,Stankiewicz,R.,1995.On the nature.Function and composition of technological systems.In:Carlsson,B.(Ed.), Technological Systems and Economic Performance:The Case of Factory Automation.Kluwer Academic Publishers,Dordrecht,pp.

21–56.

Carlsson,B.,Jacobsson,S.,Holm´e n,M.,Rickne,A.,2002.Innovation systems:analytical and methodological issues.Research Policy21, 233–245.

Carlsson,B.,Jacobsson,S.,Bergek,A.,2005.Dynamics of innova-tion systems—policy-making in a complex and non-deterministic world,under review.

Carroll,G.,1997.Long-term evolutionary changes in organizational populations:theory,models and empiricalfindings in industrial demography.Industrial and Corporate Change6,119–143. Cooke,P.,Uranga,M.G.,Etxebarria,G.,1997.Regional innovation systems:institutional and organisational dimensions.Research Policy26,475–491.

Dahm´e n,E.,1988.‘Development blocks’in industrial economics.

Scandinavian Economic History Review36,3–14.

Das,S.S.,Van de Ven,A.H.,2000.Competing with new product tech-nologies:a process model of strategy.Management Science46, 1300–1316.

Davies,A.,1996.Innovation in large technical systems:the case of telecommunications.Industrial and Corporate Change5, 1143–1180.

Dosi,G.,1982.Technological paradigms and technological trajecto-ries:a suggested interpretation of the determinants and directions of technical change.Research Policy11,147–162.

Dosi,G.,Pavitt,K.,Soete,L.(Eds.),1990.The Economics of Technical Change and International Trade.Harvester/Wheatsheaf,New York. Edquist,C.,1997.Systems of innovation approaches—their emer-gence and characteristics.In:Edquist, C.(Ed.),Systems of Innovation:Technologies,Institutions and Organizations.Pinter Publishers,London.

Edquist,C.,1999.Innovation policy—a systemic approach.TEMA-T Working paper.Link¨o ping University,Link¨o ping.

Edquist,C.,2004.Systems of innovation:perspectives and challenges.

In:Fagerberg,J.,Mowery,D.C.,Nelson,R.R.(Eds.),The Oxford Handbook of Innovation.Oxford University Press,Oxford. Edquist,C.,Johnson,B.,1997.Institutions and organizations in sys-tems of innovation.In:Edquist,C.(Ed.),Systems of Innovation: Technologies,Institutions and Organizations.Pinter Publishers, London,pp.41–63.

Erickson,W.B.,Maitland,I.,19.Healthy industries and public pol-icy.In:Dutton,M.E.(Ed.),Industry Vitalization.Pergamon Press, New York.

Freeman,C.,1987.Technology Policy and Economic Performance.

Pinter Publishers,London.

Freeman,C.,Louc¸˜a,F.,2002.As Time Goes by.From the Indus-trial Revolutions to the Information Revolution.Oxford University Press,Oxford.

Galli,R.,Teubal,M.,1997.Paradigmatic shifts in national innovation systems.In:Edquist,C.(Ed.),Systems of Innovation:Technolo-gies,Institutions and Organizations.Pinter Publishers,London,pp.

342–370.

Geels,F.W.,2004.From sectoral systems of innovation to socio-technical systems:insights about dynamics and change from sociology and institutional theory.Research Policy33,7–920.428 A.Bergek et al./Research Policy37(2008)407–429

Hekkert,M.P.,Suurs,R.A.A.,Negro,S.O.,Smits,R.E.H.M., Kuhlmann,S.,2007.Functions of innovation systems:a new approach for analyzing technological change.Technological Fore-casting and Social Change74,413–432.

Holm´e n,M.,2001.Emergence of regional actor systems—generic technologies and the search for useful and saleable applications.

PhD Thesis.Department of Industrial Dynamics,Chalmers Uni-versity of Technology,G¨o teborg.

Holm´e n,M.,Jacobsson,S.,2000.A method for identifying actors in a knowledge based cluster.Economics of Innovation and New Technology9,331–351.

Hughes,T.P.,1983.Networks of Power—Electrification in Western Society1880–1930.The Johns Hopkins University Press,Balti-more.

Hughes,T.P.,1990.The evolution of large technological systems.In: Bijker,W.E.,Hughes,T.P.,Pinch,T.J.(Eds.),The Social Construc-tion of Technological Systems:New Directions in the Sociology and History of Technology.The MIT Press,Cambridge,pp.51–82. Jacobsson,S.,Bergek,A.,2004.Transforming the energy sector:The evolution of technological systems in renewable energy technol-ogy.Industrial and Corporate Change13,815–849. Jacobsson,S.,Johnson,A.,2000.The diffusion of renewable energy technology:an analytical framework and key issues for research.

Energy Policy28,625–0.

Jacobsson,S.,Lauber,V.,2006.The politics and policy of energy system transformation—explaining the German diffusion of renewable energy technology.Energy Policy34,256–276. Jacobsson,S.,Sand´e n,B.,B˚a ngens,L.,2004.Transforming the energy system—the evolution of the German technological system for solar cells.Technology Analysis&Strategic Management16, 3–30.

Johnson, A.,1998.Functions in innovation system approaches.

Unpublished Working Paper.Department of Industrial Dynamics, Chalmers University of Technology,G¨o teborg.

Johnson,A.,2001.Functions in innovation system approaches.In: Electronic Paper at the Proceedings of the Nelson and Winter Conference,Aalborg.

Johnson,A.,Jacobsson,S.,2001.Inducement and blocking mecha-nisms in the development of a new industry:the case of renewable energy technology in Sweden.In:Coombs,R.,Green,K.,Walsh, V.,Richards,A.(Eds.),Technology and the Market:Demand,Users and Innovation.Edward Elgar,Cheltenham.

Kemp,R.,Schot,J.,Hoogma,R.,1998.Regime shifts to sustainabil-ity through processes of niche formation:the approach of strategic niche management.Technology Analysis and Strategic Manage-ment10,175–195.

Klein Woolthuis,R.,Lankhuizen,M.,Gilsing,V.,2005.A system failure framework for innovation policy design.Technovation25, 609–619.

Layton,E.T.,1974.Technology as knowledge.Technology and Culture 15,31–41.

Lieberman,M.,Montgomery, C.,1988.First-mover advantages.

Strategic Management Journal9,41–58.

Lindblom,C.E.,1959.The science of“muddling through”.Public Administration Quarterly19,79–88.

Lindmark,S.,Rickne,A.,2005.Dynamics and functionality of the Swedish mobile Internet innovation system.In:Proceedings of the 16th European Regional Conference of The International Telecom-munications Society(ITS),Porto.

Liu,X.,White,S.,2001.Comparing innovation systems:a framework and application to China’s transitional context.Research Policy30, 1091–1114.Lundvall,B.˚A.(Ed.),1992a.National Systems of Innovation—Toward

a Theory of Innovation and Interactive Learning.Pinter Publishers,

London.

Lundvall,B.˚A.,1992b.Introduction.In:Lundvall,B.˚A.(Ed.),National Systems of Innovation—Toward a Theory of Innovation and Inter-active Learning.Pinter Publishers,London,pp.1–19. Malerba, F.,1996.Public Policy and Industrial Dynamics:an Evolutionary Perspective.Report submitted to the European Com-mission.

Malerba,F.,2002.Sectoral systems of innovation and production.

Research Policy31,247–2.

Marshall,A.,1920.Principles of Economics,8th ed.Macmillan and Company Ltd.,London.

Maskell,P.,2001.Towards a knowledge-based theory of the geograph-ical cluster.Industrial and Corporate Change10,921–943. McLoughlin,I.,Badham,R.,Couchman,P.,2000.Rethinking politi-cal process in technological change:socio-technical configurations and frames.Technology Analysis&Strategic Management12, 17–37.

Metcalfe,S.,1992.The Economic Foundation of Technology Policy.

Equilibrium and Evolutionary Perspectives.University of Manch-ester,Manchester.

Metcalfe,S.,2004.Policy for Innovation.ESRC Centre for Research on Innovation and Competition,University of Manchester,Manch-ester.

Myrdal,G.,1957.Economic Theory and Underdeveloped Regions.

Ducksworth Pubs,London.

Nelson,R.R.,1992.National innovation systems:a retrospective on a study.Industrial and Corporate Change2,347–374.

North,D.C.,1994.Economic performance through time.The Ameri-can Economic Review84,359–368.

Oltander,G.,Perez Vico,E.,2005.A survey of the Swedish security industry and an innovation system analysis of the Swedish security sensor industry.Master Thesis Report No.2005:1.Department of Innovation Engineering and Management,Chalmers University of Technology,G¨o teborg.

Porter,M.,1990.The competitive advantage of nations.Harvard Busi-ness Review68,73–93.

Porter,M.E.,2000.Location.Competition and economic develop-ment:local clusters in a global economy.Economic Development Quarterly14,15–34.

Rao,H.,2004.Institutional activism in the early American automobile industry.Journal of Business Venturing19,359–384.

Rickne, A.,2000.New Technology-Based Firms and Industrial Dynamics.Evidence from the Technological System of Biomate-rials in Sweden,Ohio and Massachusetts.PhD Thesis.Department of Industrial Dynamics.Chalmers University of Technology, G¨o teborg.

Rosenberg,N.,1976.Perspectives on Technology.Cambridge Univer-sity Press,Cambridge.

Rosenberg,N.,1996.Uncertainty and technological change.In:Lan-dau,R.,Taylor,T.,Wright,G.(Eds.),The Mosaic of Economic Growth.Stanford University Press,Stanford,pp.334–355. Rotmans,J.,Kemp,R.,van Asselt,M.,2001.More evolution than revolution.Transition management in public policy.Foresight3, 15–31.

Sabatier,P.A.,1998.The advocacy coalition framework:revisions and relevance for Europe.Journal of European Public Policy8,98–130. Scitovsky,T.,1954.Two concepts of external economies.Journal of Political Economy62,143–151.

Smith,A.,1776.An Inquiry into the Nature and Causes of the Wealth of Nations.Strahan and Cadell,London.A.Bergek et al./Research Policy37(2008)407–429429

Smith,K.,2000.Innovation as a systemic phenomenon:rethinking the role of policy.Enterprise and Innovation Management Studies1, 73–102.

Smits,R.,Kuhlmann,S.,2002.Strengthening interfaces in innovation systems:rationale,concepts and(new)instruments.In:Proceed-ings of the EC STRATA Workshop on New challenges and new responses for S&T policies in Europe,Brussels.

Stigler,G.,1947.The Theory of Price.Macmillan,New York. Suchman,M.C.,1995.Managing legitimacy:strategic and institutional approaches.Academy of Management Review20,571–610. Unruh,G.C.,2000.Understanding carbon lock-in.Energy Policy28, 817–830.

Van de Ven,A.H.,1993.The development of an infrastructure for entrepreneurship.Journal of Business Venturing8,211–230.van Lente,H.,1993.Promising technology.In:Rip,A.(Ed.),The Dynamics of Expectations in Technological Development.Uni-versiteit Twente,Twente.

von Hippel,E.,1988.The Sources of Innovation.Oxford University Press,Oxford.

Walker,W.,2000.Entrapment in large technology systems:insti-tutional commitment and power relations.Research Policy29, 833–846.

Young,A.,1928.Increasing returns and economic progress.Economic Journal38,527–542.

Zimmerman,M.A.,Zeitz,G.J.F.,2002.Beyond survival:achieving new venture growth by building legitimacy.Academy of Manage-ment Review27,414–431.