IMP Journal - IMP Group

THE

IMP Journal INDUSTRIAL MARKETING AND PURCHASING

ISSUE 7 6 ISSUE33 VOLUME VOLUME

2013 2012

Forum for research into business interaction, relationships and networks.

The IMP Journal Volume 7. Issue 3, 2013

Editor

Debbie Harrison BI Norwegian Business School

Håkan Håkansson BI Norwegian Business School

Executive Board

Stefan Henneberg Manchester Business School

Håkan Håkansson BI Norwegian Business School

Lars Huemer BI Norwegian Business School

Annalisa Tunisini Catholic University, Milan

Tibor Mandjak Corvinus University

Peter Naude Manchester Business School

Lars-Gunnar Mattsson Stockholm School of Economics

Editorial Board

Helén Anderson Jönköping International Business School

Stefanos Mouzas Lancaster University

James Anderson Northwestern University

Kristian Möller Helsinki School of Economics

Luis Araujo Lancaster University

Ann-Charlott Pedersen Norwegian University of Science and Technology

Björn Axelsson Stockholm School of Economics

Thomas Ritter Copenhagen Business School

Enrico Baraldi Uppsala University

Robert Salle EM Lyon

Keith Blois Lancaster University

Asta Salmi Helsinki School of Economics

Roberta Bocconcelli Urbino University

Judit Simon Corvinus University

Anna Dubois Chalmers University of Technology

Ivan Snehota University of Lugano

Geoff Easton Lancaster University

Robert Spekman University of Virginia

David Ford University of Bath, School of Management

Alexandra Waluszewski Uppsala University

Lars-Erik Gadde Chalmers University of Technology

Ian Wilkinson University of New South Wales

Simone Guercini Florence University Aino Halinen Turku School of Economics and Business Administration Lars Hallén Mälardalen University, School of Business

Publisher

Louise Young University of Western Sydney Judith Zolkiewski Manchester Business School

IMP Group, ISSN 0809-7259, www.impjournal.org


The IMP Journal Issue 3, Volume 7 Contents

A letter from the editor

Diversity in Homogeneity – A longitudinal bibliometric review of Industrial Marketing and Purchasing (IMP) Group Conferences from 1984 to 2012 Gerhard A. Wuehrer, Angela Elisabeth Smejkal

140

IMP theory in light of process- and system theories Per Ingvar Olsen

159

Approaching (Inter-)Actors in the Business Landscape Antonella La Rocca

171

The Political Embeddedness of Business Networks in a Chinese Context: The Case of a Biopharmaceutical Business Network Åse Linné, Tommy Tsung-Ying Shih

180

The Role of Tie Strength, Relational Capability and Trust in the International Performance of High Tech SMEs. Breda Kenny, John Fahy

188

Produced by IMP Group in cooperation with Manchester Business School Catholic University - Milan BI Norwegian Business School PDF and Layout, made by Kooperaivet Mediagruppen Karlstad. http://mediagruppen-karlstad.se/


A letter from the editor The IMP-Group emerged in the 1970’s as an international cooperation project among researchers dissatisfied with conventional market thinking. The development since then can be described as a journey that has taken us to a large and varied number of physical and theoretical places. The journey has certainly included some consistency in terms of shared empirical and theoretical challenges but it has also brought us some unexpected views. This issue of the IMP Journal is no exception, but is yet another example of the mix of consistency and surprises, expressed in 5 articles: “Diversity in Homogeneity – A longitudinal bibliometric review of Industrial Marketing and Purchasing (IMP) Group Conferences from 1984 to 2012”, by Gerhard A. Wuehrer and Angela Elisabeth Smejkal “IMP theory in light of process- and system theories”, by Per Ingvar Olsen “Approaching (Inter-)Actors in the Business Landscape”, by Antonella La Rocca “The Political Embeddedness of Business Networks in a Chinese Context: The Case of a Biopharmaceutical Business Network”, by Åse Linné and Tommy Tsung-Ying Shih “The Role of Tie Strength, Relational Capability and Trust in the International Performance of High Tech SMEs”, by Breda Kenny and John Fahy In the first article; “Diversity in Homogeneity – A longitudinal bibliometric review of Industrial Marketing and Purchasing (IMP) Group Conferences from 1984 to 2012”, the two authors take us on a historical trip by analysing the content of the publications from the IMP conferences as the main empirical input. The article gives a picture of a journey where certain empirical phenomenon are recurrent but have been investigated in several different types of settings. Thus, it is a journey where empirical results have been accumulated and transferred in a consistent base of theoretical concepts and models. The article can also be seen as an illustration of how joint efforts can create progression in research. In the second article; “IMP theory in light of process- and system theories” the author gives a theoretical argument for the same type of collective research journey. The author brings us into the theoretical world where IMP thinking is systematically related to process and system theories. The article starts out from the fundamental ontological position defining the IMP as an area of theory with the assumption that resources are heterogeneous and only partially knowable, and as a consequence, that relations

and interactions are fundamental constituents of the economy. The article is then arguing that this gives IMP a unique position having specific interfaces with both process and system theories. The three remaining articles are belonging to a series where young IMP researchers have been invited to contribute with papers based on, or related to, their dissertations. (The ambition is to continue to publish these types of articles in future issues.) In the first of these articles; “Approaching (Inter-)Actors in the Business Landscape”, we are faced with the problems of describing actors in an interactive business landscape. How should they be conceptualized when they have to be seen from the viewpoint of counterparts? Two dimensions that have to be dealt with is that they have to be considered as heterogeneous as well as emerging. Neither of these are easy to deal with. In the second article; “The Political Embeddedness of Business Networks in a Chinese Context: The Case of a Biopharmaceutical Business Network”, we are brought into a very special business landscape. The development of new products in China is, at least for high tech areas, dominated by the interface between the Chinese political and commercial spheres. The development and commercialization of a new drug is used to describe and analyse the close interaction between the political and commercial sectors in China. Finally, in the third of these articles; “The Role of Tie Strength, Relational Capability and Trust in the International Performance of High Tech SMEs”, the two authors take us to the border between Ireland and some importing countries. The study is an investigation of how Irish small high-tech companies are acting in the international context and how this acting is related to the strength of ties and the existence of trust and relational capabilities. The study is also as an illustration of the methodological problems that appears when investigating networks as well as important dimensions of the network. The research journeys that this issue sheds light on includes both supporting arguments for earlier established views but also some new more challenging insights. It underlines that there are reasons to continue joint efforts including both consistency and diversity and this both from an empirical as well as theoretical perspective. Sunne November 2013 Håkan Håkansson Editor

The IMP Journal Volume 7. Issue 3, 2013 140

Diversity in Homogeneity – A longitudinal bibliometric review of Industrial Marketing and Purchasing (IMP) Group Conferences from 1984 to 2012 Gerhard A. Wuehrer & Angela Elisabeth Smejkal Johannes Kepler University Linz

Abstract

The work of the Industrial Marketing and Purchasing (IMP) Group has left impressive marks on the research field of industrial marketing. Its influence and development over the last 30 years is well recognized in connection with the thoughts of a general theory of business marketing. This study is devoted to answering the question ‘what is the paradigmatic status of the research group?’ On the basis of research questions, a longitudinal content-based analysis of the last 28 IMP conferences has been conducted by using a semi-automated software analysis tool. Findings indicate that the IMP Group has developed over the years into an ‘academic brand’ (Cova et al. 2009) and, from a Kuhnian sense (De Langhe 2013), represents ordinary science where exploitation of current concepts is the daily business and the exploration of new knowledge could be the next evolutionary step. KEY WORDS: Longitudinal bibliometric analysis, Industrial Marketing and Purchasing (IMP) Group, conferences, content analysis, unsupervised key word extraction, marketing theory, paradigm

The authors want to thank two anonymous reviewers for their helpful comments.

1. Introduction Over the years, the Industrial Marketing and Purchasing (IMP) Group has become an ‘academic brand’ as stated by Cova et al. (2009, see also Ford 2011). Therefore its theoretical and empirical work, via papers presented at IMP conferences since 1984 and subsequent academic discussions, has had an impressive influence on the research of managerial acting in business marketing and purchasing and on the career of many marketing academics in the field (Ford 2011; Baraldi et al. 2012; Lindgreen et al. 2012). Even in the development of a general theory of business marketing, the contribution of IMP is well recognized and mentioned (Hunt 2013). Cova et al. (2009) state that IMP follows a clear picture of conceptual development, dealing with ‘structure and process of the business landscape, a methodological approach and an interest in exploring different aspects of that structure and process’ (Cova et al. 2009, p. 573). Up until now, there has been no clear point of view on the notation of the knowledge body of IMP (Sousa 2010). According to Sousa (2010), the wide range of denomination consists of ‘approach’ (Håkansson 1987), ‘paradigm’ (Easton 1992), ‘European-based research tradition’ (Johanson & Mattsson 1994), ‘school of thought’ (Araujo & Easton 1996) and ‘perspective’ (Turnbull et al. 1996). This leads to the overall research question, what is the paradigmatic status from a scientometric point of view? In general, bibliometric studies are steadily used to evaluate the output of scientific fields within a specific period of time and to map its trends and dynamics (Sanz-Casado et al. 2007). According to Börner et al. (2012), there are the following possibilities to analyze the emergence, growth, and diffusion of scientific activities. Firstly, the textual elements of scientific documents can be used, such as words located in titles, keywords,

and abstracts or using a full-text analysis and doing a word co-occurrence analysis. As such, most prominent keywords can be seen as ‘DNA fingerprints’ (Börner et al. 2003, p. 185) of a research field and its publications. On the other hand, journals, their subjects, and the references could be the basis for such aims of analysis (Börner et al. 2003). So the means of analysis for applying a bibliometric study are different: the most common indicators for evaluation are citation indices and journal impact factors (Wiles et al. 2010), as well as rankings of article publications country-wise for different scientific fields (Robert et al. 2008). Citation indices and journal impact factors exist in nearly every scientific domain, e.g. pain research (Robert et al. 2008), disease or health research (SanzCasado et al. 2007), business and society (De Bakker et al. 2005), consumer behavior research (Muñoz-Leiva et al. 2012), international marketing research (Samiee & Chabowski 2012), information science (Donohue 1972; Huanwen 1996; Payne & Thelwall 2009) and in bibliometric studies itself (Meadows 2005). Today such studies seem to concern almost every scientific field (Glänzel 2003), and a look at the share of proceedings literature in the total publication outputs of different scientific fields supports the relevance of conference papers as subjects for bibliometric studies in addition to journal articles (Glänzel et al. 2006). Specifically, such studies give a bird’s eye view of an overall scientific landscape or of certain disciplines (Robert et al. 2008). The highest interest of the scientific community concerns the contributions of top journals as the outlet of scientific production (Lahiri & Kumar 2012; Lisée et al. 2008). However, publications regarding conference contributions as subjects of bibliometric studies are still scarce (Wuehrer & Smejkal 2012). Several factors may lead to that. Either it is not typical for that field of science (see Sidiropolous & Manolopoulos 2005, Chambers


et al. 2010), publication system is based on journal outlets, or it is difficult to get access to conference papers (Wuehrer & Smejkal 2012), or it is the general neglect of the role and importance of academic conferences. What are the benefits when conducting a bibliometric analysis of conference proceedings? First of all, new perspectives are opened with the potential of identifying evolving topics (Hofer et al. 2010), whilst simultaneously facilitating the detection of stagnancy as the opposite of development. One may expand this thought further. De Langhe (2013) deals with the Kuhnian paradigm and its relevance for a systemic view of philosophy of science. Originally, Kuhn (1962) had ‘the vision of an explanatory and empirically general philosophy of science’ (De Langhe 2013, p. 65), which led him to address issues such as ‘why disciplines tend to cluster in schools, why communities are reluctant to embrace novel frameworks and what drives scientific innovation’ (De Langhe 2013, p. 65). De Langhe (2013) offers explanations as to why the articulation of the Kuhnian paradigm originally was insufficient for the scientific community and presents tools that have been developed since then. As such, he states that the development of network theory and agent-based modeling, the usage of information processing with scientometric data, has been reinforced and allows further analytical procedures to show the dynamics of science throughout certain time periods (De Langhe 2013). Therefore, the following study seeks to clarify the paradigmatic status of the IMP group’s scientific endeavors based on conference proceedings and the co-occurrence of keywords addressed by them. In some scientific domains, conference proceedings are the only outlet of long-term scientific production, an example of which is the IMP Journal, established in 2004 (http://www.impjournal.org). Contributions in scientific journals like Industrial Marketing Management would be available too, but an entire longitudinal (1984-2012) bibliometric study referring to the keyword co-occurrences of IMP conferences is one of the keys to acquiring insight into the scientific status and development of an interesting research domain in the field of business-to-business (b2b) marketing. Furthermore, the IMP Group is seen as being representative for the development of b2b marketing (Cova & Salle 2008). The paper is structured as follows. Firstly, we examine the role of IMP Group conferences in the domain of business research and give a short overview of the development and guiding ideas. This leads to the formulation of the research questions of the article, followed by a description and justification of methodology. The next section presents the main results - on the one side by showing findings of the descriptive analysis and on the other side by giving information via mapping the co-occurrence matrices and the development of the most significant co-occurring keywords over the years. Finally, concluding remarks state the main consequences and implications for further studies.

2. Literature review and research questions 2.1 A short overview on the role of the IMP Group in the industrial marketing domain The IMP Group started as a research project entitled ‘International/Industrial Marketing and Purchasing Project,’ shortened to IMP1 (Håkansson 1982, Håkansson et al. 2009) in 1976 with research participants from five European countries (France, Germany, Italy, Sweden, and the United Kingdom) and universities (www.impgroup.org). The approach is one of the two research

streams recognized by the American Marketing Association (Cova et al. 2009; Harrison 2004). The major reason for founding the IMP Group, according to Wilkinson (2001, p. 35; see also Sousa 2010), was that the participants of the first research project ‘were dissatisfied with the dominant marketing paradigm of the time, which focused on consumer goods and adopted a stimulus response, arm-length approach to the customer with the seller as the active party.’ The initial aim of the first study program was to gain a theory-based empirically deeper insight into buyer-seller relationships as well as to differentiate those insights according to variant markets. As such, theoretical aspects have been considered as well (Håkansson & Snehota 2002; Håkansson 1982) and expanded. Empirically, by mid-1978 researchers had collected the relevant data by interviewing about 900 to 1,000 buyers and sellers (900 in Håkansson (1982) and 1,000 in Håkansson & Snehota (2002); distribution about 50% sellers and 50% buyers) to examine those relationships, where about 80% of them were international (Håkansson & Snehota 2002). The IMP1 is also called the ‘interaction approach’ (Håkansson 1982; Wilson 1995) – a dynamic model of buyerseller relationships (www.impgroup.org). This has already been published in two books, International Marketing and Purchasing of Industrial Goods - An Interaction Approach by Håkansson (1982) and Strategies for International Industrial Marketing by Turnbull & Valla (1986) (www.impgroup.org). Thus, the IMP1 project could be seen as the starting point of the research work and network for scholars for further studies and scientific collaborations (Hunt 2013). In 1986, the second main research project (IMP2) commenced, based on the results of IMP1 (Håkansson & Snehota 2002). ‘Building on the evidence of the existence of strong buyer-seller relationships in industrial markets, the focus of IMP2 was on interdependencies in and between the relationships and the effects of interdependencies on the companies involved. Networks of relationships thus became of main empirical and conceptual interest for IMP2’ (Håkansson & Snehota 2002, p. 37). A new aspect at this point of study was that researchers from other countries (USA, Japan, and Australia), who had not participated in project IMP1, collaborated (Håkansson & Snehota 2002). The result was the ‘ARA’ model (actor bonds-resource ties-activity links) as an extension of the theoretical reasoning to the ‘network approach’ (Håkansson & Snehota 1995). The new ideas were published by Håkansson & Snehota in 1995 in the books ‘Developing Relationships in Business Networks’ and ‘Managing Business Relationships’ by Ford et al. (2003), see also www. impgroup.org/about.php. An additional concept, named the ‘4 resource (4R) model,’ with its aim to shed more light on the interaction of business resources (www.impgroup.org/about.php; Håkansson & Waluszewski 2002), augmented the discussion. In conclusion, the IMP Group and its theoretical and empirical approach consists of three constituent parts: interaction (episodes), relationship (dyads), and network (net and network) at a level perspective, whereas each level is related to in sum about 30 themes, ranging from ‘interaction model’ to ‘network dynamics – change and stability’ (see Harrison 2004). Nowadays, the IMP Group and its scientific work are known as ‘full-fledged Markets-as-Networks Theory, despite reluctantly assumed or promptly denied at early times’ (Sousa 2010, p. 5, referring to Håkansson & Snehota 2000, p. 46). Therefore, the IMP Group has a strong influence on the research field (in viewing, researching, and teaching) of business marketing (Morlacchi et al. 2004; Ford 2011; Backhaus et al. 2011; Hunt 2013) and is well known for its contributions to the field (Gemünden


1998). Furthermore, the IMP consists of the activities of the network itself. Since 1984, an IMP conference takes place (www. impgroup.org/about.php) every year in September (with the exception of 1987) and is visited by researchers from all over the world. At the time of writing, the most recent conference was held in Atlanta, USA in August/September 2013, with the topic ‘Building and Managing Relationships in a Global Network: Challenges and Necessary Capabilities’ (http://robinson. gsu.edu/imp2013). Further activities of the IMP Group are the organization of workshops, doctoral consortia, and seminars. Moreover, the IMP publishes a website and the IMP Journal, which focuses on the investigation of interaction and business relationships too (Morlacchi et al. 2004). The aim is ‘to publish a limited number of outstanding papers that are of major interest and value to the research community’ (http://www.impgroup.org/IMPJournal.php). So it may be stated with De Langhe (2013) that the yearly conference structure of the IMP Group provides a direct and indirect feedback loop augmenting the discussion within the scientific community of researchers in that academic domain. This may lead to an event like a ‘scientific revolution’ (De Langhe 2013, p. 66) and to a consistent group of researchers ‘who share conceptual, theoretical, instrumental and methodological commitments, viz. “paradigms”’ (De Langhe 2013, p. 66). Until now, the following empirical studies (Table 1) can be

Author(s) / Year

mentioned that treat the topics’ development of the IMP Group, insofar as they are dealing with the paradigmatic status of the group’s reasoning and thinking. The contribution of these studies to a deeper understanding of the IMP Group and its influence on academic thinking and b2b marketing is undoubted. However, the focus lies on: describing a state of the art network with thoughts and requirements for future prospects (see Gemünden 1998) analyzing the central key aspects of one years’ abstracts (Easton et al. 2003a; Easton et al. 2003b), respectively comparing the main content of two IMP conferences (Windischhofer et al. 2004) a content and citation analysis of conference papers of the years 1995-1999, 2001-2006 (Barth 2008) So the investigations mentioned above analyze either a limited time period or compare only two conferences with special research interests. Only one of them (Barth, 2008) has a bibliometric approach per se. Therefore, our study makes a contribution to close these gaps by focusing on the content dynamics of all the conference years (1984-2012) so far, considering the dynamics of the central semantic concepts with bibliometric means to analyze and visualize the main findings. Published studies (see Table 1) provide the basis and are reference points for our aims.

Table 1 Publications on the dynamics of main topics of IMP conferences Title

Methodology

- - Barth (2008)

The history and the development of the IMP Group reasoning. A bibliometric approach

- - - -

Windischhofer et al. (2004)

A Review of IMP Conferences 1998 and 2003 with Special Focus on Dynamics

-

-

Easton et al. (2003a) and Easton et al. (2003b)

An exploratory taxonomic study of IMP Group conference papers

- - -

-

Gemünden (1998)

The Development of IMP – an Analysis of Conference Proceedings 1984-96 -

Bibliometric analysis of co-authorships Analysis of IMP members and egonet analysis of single IMP players Content-based analysis of the conference papers 1995 to 1999, 2001 to 2006, key-worded by the author (based on abstracts) Citation analysis for a picture of the networks of references Analysis of the content of each paper by multidimensional coding of key terms and key notions Categorizing of papers and comparing the results of each year with the others Levels of analysis o General Level o Interaction Level o Empirical Scope Level o Methodological Level o Temporal Level MS Excel was used for analysis by using a quantitative descriptive method Qualitative analysis of abstracts Focus on content of the 16th annual IMP conference The results have been four focused categories: relationships, networks, management, and contingencies (see also Windischhofer et al. 2004) Systematic content analysis with the following variables: o Origin of the authors of the papers o Unit of analysis o Kind of theoretical contribution o Empirical base of the publication o Regional scope of the study o Explanation of success Quantitative descriptive analysis based on abstracts, work-in-progress papers and competitive papers published in the IMP conference proceedings 1984-1996 (see also Windischhofer et al. 2004)


2.2 Research questions When speaking about ‘IMP as fashion - past, present and future,’ Harrison (2004) offered some interesting ideas that could be summarized describing the paradigmatic status of the research approach and the consequences to which it was related. She speaks metaphorically about the group as having a midlife crisis and declares that ‘it can be easy at one of these conferences – especially in recent years – to come away with the rather depressing impression that IMP is not in the mature phase of the lifecycle, but is in fact in decline’ (ibid., p. 5). Four years later, Cunningham (2008) repeats and reiterates Harrison’s ideas when he speaks about the fact that some of the masterpieces of IMP thought might be 15 to 25 years old, somewhat dusty and withered, and that there are anxieties of increasing uniformity, repetition, and stereotyping. Other schools have also emerged with their focus on relationship marketing and services marketing. They build on contributions of organization and business strategy science. Besides this, Cunningham argues a loss of sight of key roles in interfirm relationships and network interdependencies in determining a company’s competitiveness. He also stresses that for the moment there is no clear answer to the question of whether transactional or relational exchanges generate better firm performance. In addition, he complains about the fact that IMP has neglected the developments in b2c markets where consumers are active and interact with the service provider. Finally, he sees a danger that the annual IMP conference contributors and events become too stylized and ritualistic, presenting exclusively upon topics of co-operative relationships and networks. Competitive behavior and independent actions of companies are too often neglected. And further – the conferences resemble tribal meetings (Cunningham 2008). According to those strong arguments we state the following research question: Research question 1: The main source of the further development of the IMP Group since its foundation in 1976 has been the IMP projects (1 and 2) as well as the models discussed there (see also Section 2.1. of the paper). Since IMP1 laid the ground for an interesting development more than 20 years ago, which may lead to the concern that there is a stagnation of ideas, we need to analyze the actual keywords and their connectedness in comparison with the early ones. So we ask: “Which topical keywords and co-occurrences appear in the conferences year wise?” Over the years, the IMP Group has developed to what Cova et al. (2009) call an ‘academic brand’ by showing a consistency and homogeneity or a paradigmatic perspective (Sousa 2010) in the central concepts of the IMP conferences. The diffusion of knowledge is something specific and is strongly related to its communication via scientific publications or scientific meetings, maybe without formal publication (Vitanov & Ausloos 2012). Additionally, Vitanov and Ausloos (2012) speak about the potential of workshops, conferences or meetings to increase the speed of knowledge diffusion and, as such, the growth of knowledge. A pioneer in the field of modeling scientific growth was Derek Price (Vitanov & Ausloos 2012; Fernández-Cano et al. 2004) who provided this field of investigation with measurements on scientific growth by indicators such as number of published papers, number of authors, dissertation production, or citations (Price 1961; Price 1963; Price and Gursey 1975, referred to by Vitanov & Ausloos 2012). Here ‘growth’ is related to an increase of number of papers by time. But growth measured by number of written papers is not exhaustive, per se. How key concepts and keywords grow interrelated is far more interesting, as that illustrates the content connectedness of the scientific body.

However, Price (1956) presented a model of scientific growth consisting of three stages: a) preliminary phase, b) phase of exponential growth, and c) saturation stage, represented by a logistic, S-shaped curve (Fernández-Cano et al. 2004). The saturation stage is reached when new concepts coming into the scientific field as a result of phase a) and b) are exhaustingly researched. Then the growth is in a stage of slowing down until new fields and research opportunities are found (Vitanov & Ausloos 2012) and the next growth cycle starts. This leads to a series of Sshaped growth functions, where the following cycle starts at a higher level than the previous one. The general contribution of Price’s model and investigations on scientific growth are still important regarding the usage of mathematical methods for such research questions and the model of scientific growth is assumed to ‘be reproduced to the point of satiety in a great number of books, papers and languages over the next 25 years’ (FernándezCano et al. 2004, p. 303). As a longitudinal, bibliometric view of a time period of nearly 30 years (exactly 28) has not been published until now, this allows implications on its dynamic aspects (Chabowski et al. 2011, Muñoz-Leiva et al. 2012) that should be considered because the scientific development also brings about rapid changes in our research assumptions (Windischhofer et al. 2004). Therefore, the model of Price (1956) offers a good basis for our investigation on the development of the co-occurring keywords during the investigated period of time. On that ground, we want to analyze the similarity or non-similarity of co-occurring keywords from a bibliometric point of view for providing a view of aggregate knowledge of the content of conference contributions over time. This leads to our second research question. Research question 2: How similar (or not) is the structure of the conferences’ keyword co-occurrences (1984 to 2012) of the IMP Group from a bibliometric point of view?

3. Methodology for longitudinal keyword analysis Although the investigation of scientific fields has increased in recent years, up until now there has been no unique framework on modeling science mathematically (Scharnhorst et al. 2012). Processes for conducting a bibliometric study differ according to the number of steps carried out and the detailed description of doing such an analysis (Ding et al. 2001; Börner et al. 2003; Hofer et al. 2010). For performing the following study, we progress with three adapted stages: data collection, data standardizing, and data mapping (Ding et al. 2001).

3.1 Data Collection Setting up such a research project includes the definition of the unit of analysis. On the basis of literature resources (di Gregorio & Davidson 2008) we have to differ between ‘unit of analysis’ and ‘unit of observation’ (sometimes also called ‘unit of data collection’ (Neuendorf 2002)). According to Wasserman and Faust (1994) ‘the unit of observation is an actor, from whom we elicit information about ties’ (p. 43). ‘Unit of analysis’ and ‘unit of observation’ are central elements within the research design (di Gregorio & Davidson 2008), although they are sometimes the same in some studies (Neuendorf 2002), for example in those which employ questionnaires and where individuals are the target groups of investigation. In that case, the unit of observation is represented by the individual respondent – the variables are measured on each unit. The unit of analysis will also be the individual person, as statistical analyses refer to their


answers and will be done on the basis of the number of people responding. As such, the differentiation is not always simply clear (Neuendorf 2002). In our case, the unit of observation and the unit of analysis are not the same. The units of observation for the current study are the IMP conferences with the papers presented there (for comparison with other bibliometric studies with conferences as units of observation, see e.g. Hofer et al. 2010, Wuehrer & Smejkal 2012). The central aim of our study is to compare and to analyze the co-occurrences of the keywords. So, the unit of analysis is represented by the co-occurrences of the keywords in the different conferences, which will result in 28 different co-occurrence matrices (see also Section 3.2.2 and an example of such a matrix in Table 3). Such ‘a co-occurrence matrix is obtained by counting frequencies of pairwise term cooccurrences’ (Matsuo & Ishizuka 2004). In our case, each matrix will be a result from all the papers’ analyses for each conference. The steps will be described in detail on the following pages. The focus in collecting the raw data lies in the saving of pdf and word-files of IMP conference contributions within the time period 1984-2012. The documents have been stored year- and conference-wise in a database. Papers of the following conferences were not available for the authors as they were neither published on the IMP Group homepage nor stored in the Manchester University archive: Phuket/Thailand, 2005; Phuket/ Thailand, 2007; Kuala Lumpur/Malaysia, 2009. The authors’

attempts in obtaining access to this data were without success. Consequently, contributions available for download are the basis for the study (reference sites for downloads: www.impgroup.org and https://research.mbs.ac.uk/mimp/IMPArchive.aspx)1. In the year 2002, two conferences were held in Perth/Australia and Dijon/France. In order to enable a consequent, year-wise analysis, those papers have been saved and analyzed all together. Whereas in other comparable bibliometric studies the focus lies in analyzing the words and co-words of titles and abstracts, which is sometimes mentioned as a limitation (e.g. De Bakker et al. 2005), the following investigation was concentrated on the analysis of the whole text including the references of each document. Use of the entire text body, and not only the abstract as a summary of the most important aspects of the text, signifies that the additional data (in the form of references) will have an influence on the results by a greater data variance than without the reference positions. Because references are an important part of a research agenda, we work on the ‘big picture’ in any way by including every aspect of a document (title, abstract, text, references). Table 2 shows the number of conference contributions per year from 1984 to 2012. A total of 3,109 IMP papers (competitive 1. We want express our sincere thanks to the colleagues and scholars of the Manchester Business School for the work and endeavors they have made to make that data available to the public.

Table 2 Conference location and number of IMP conference papers per year, number of key-worded papers Year

1984

1985

1986 1987 1988

1989

Conference location

Manchester, United Kingdom

Uppsala, Sweden

Milano, Italy

1991

Uppsala, Sweden

1992

Lyon, France

1994 1995

Bath, United Kingdom Groningen, The Netherlands Manchester, United Kingdom

Number of papers available for analysis on the basis of research on IMP homepage/University of Manchester Archive

Papers not available for analysis1

Number of papers key-worded by author

20

20

0

0

2nd open International IMP Research Seminar

13

13

0

0

39

39

0

0

30

30

0

0

39

39

0

0

46

46

0

0

55

55

0

2

67

67

0

0

36

36

0

0

101

101

0

0

72

64

8

1

3rd International IMP Research Seminar on International Marketing no conference occurred Research DevelopManchester, ments in International United Kingdom Marketing Pennsylvania, USA

Number of papers downloadable on IMP homepage/ University of Manchester Archive

Research Developments in International Marketing

Lyon, France

1990

1993

Title of conference

Research in Marketing: An International Perspective Research Developments in International Industrial Marketing and Purchasing International Business Networks: Evolution, Structure and Management Business Networks In an International Context: Recent Research Developments 9th IMP Annual Conference Meeting the Challenges of New Frontiers Interaction, Relationships and Networks: Past-Present-Future

The IMP Journal Volume 7. Issue 3, 2013 145 1996

Karlsruhe, Germany

1997

Lyon, France

1998

Turku, Finland

1999

Dublin, Ireland

2000

Bath, United Kingdom

Interaction, Relationships and Networks Processes in Business-to-Business Markets Interaction, Relationships and Networks: Visions for the Future Interactions, Relationships and Networks: Towards the New Millennium Not available2 Interaction, Relationships, and Networks: Strategic Dimensions

2001

Oslo, Norway

2002

Dijon, France Perth, Australia

Perth: Culture and Collaboration in Distribution Networks

Lugano, Switzerland Copenhagen, Denmark

Managing in Networks Interacting, Influencing, Strategizing – Where are We Heading?

Rotterdam, Netherlands Phuket, Thailand

Dealing with Dualities

2003 2004

2005

2006

Milan, Italy

2007

Manchester, United Kingdom Phuket, Thailand

2008

Uppsala, Sweden

2009

Marseille, France Kuala Lumpur, Malaysia

2010 2011 2012

Budapest, Hungary Glasgow, Scotland Rome, Italy Sum

Opening the Network-New Perspectives in Industrial Marketing and Purchasing Exploiting the B2B Knowledge Network: New Perspectives and Core Concepts An Interactive Perspective on Business in Practice and Business in Theory Handling Plurality of Relationship Forms in Networks: From Clans to Clubs, from Cliques to Communities. Theoretical and Managerial Perspectives Business Networks – Globality, Regionality, Locality Not available Combining the Social and Technological Aspects of Innovation: Relationships and Networks

83

83

0

0

114

114

0

1

69

69

0

3

215

215

0

7

110

103

7

2

143

143

0

18

179

179

0

23

168

166

2

26

130

130

0

19

130

130

0

113

207

206

1

182

158

158

0

133

165

165

0

105

167

167

0

140

219

218

1

116

195

195

0

174

161

158

3

144

3,131

3,109

22

1,209

Fotnotes 1. Reasons: Papers were damaged or not available on the IMP-homepage but listed as conference papers. 2. The title of the conference and the conference website was not available. 3. The keyword ‘different’ is related to the word ‘difference’ while Leximancer extracted the concept ‘different.’ The authors decided not to change that word extraction manually.

or work-in-progress) were available for analysis. Twenty-two papers had to be excluded because of damage or unavailability for download on the IMP website. A variation in the number of papers is shown (Table 2), but since 1999 more than 100 contributions have been submitted to each conference. The development over time shows that the number of papers of the IMP conference has increased from 21 papers in 1984, 55 in 1991, 70 in 1998, 130 in 2005, and 158 in 2012. The mean of contributions per year is 111.8, rounded up to 112 papers (basis=3,131 as number of total papers found for this analysis). The number of author key-worded papers ranges from 0 to 182; 38.4% (=1,203)

of all downloaded papers (=3,131) have been key-worded over the whole time period.

3.2 Data Standardizing For the purpose of a standardized representation of keywords within the bibliometric analysis, the authors discussed possibilities to overcome the problem of significant difference between key-worded and not key-worded conference contributions. As a result, ‘Leximancer’ was used as the content analysis tool (https://www.leximancer.com/), since it offers a standardized al-


gorithm to extract the main concepts of every single conference article. 3.2.1 Semi-automated vs. automated content coding Today, executing a content analysis focuses not only on extracting the most frequent keywords but concentrates on the cooccurrence of concepts as well (Stockwell et al. 2009; Angus 2013). Hence, some methods have been developed related to this aim like hyperspace analogue to language, latent semantic analysis, and Leximancer. Stockwell et al. (2009) give an insight into those methods by referring to Burgess and Lund (1997) and Landauer et al. (1998). Leximancer is a semi-automated tool for content coding and analysis (Stockwell et al. 2009, referring to Smith 2000) and is used to extract the main concepts of data (see Cretchley et al. 2010; Angus 2013). Content analysis allows going through a text by detecting the presence of words or concepts (Carley 1993). ‘Concepts in Leximancer are collections of words that generally travel together throughout the text’ (Leximancer Manual, 2011, p. 10). Keywords that appear very frequently are called ‘concepts’2 (Cretchley et al. 2010). Additionally, the text segments can be viewed in detail as the relevant passages are marked within each concept (Smith & Humphreys 2006). The system offers different possibilities of examination. One possibility is to analyze the text without prior definition of a specific dictionary and codebook. By this, Leximancer detects automatically, via thesaurus building, the main concepts from the data. Hence, researchers do not have to create their own coding scheme (Leximancer Manual, 2011). Another way (semi-automated) is to set up a dictionary (Smith & Humphreys 2006) with specific keywords, in the form of setting a filter, which the algorithm uses within the analytical process. The main advantage of the software algorithm is that it is able to identify and code the main concepts as a tool for content analysis. It detects words or concepts in a group of text documents. The software analyzes and quantifies the text with categories and their relationships. This can also be done for time-indexed datasets, which allows additional investigation of the evolution and history of texts (Leximancer Manual, 2011). The program uses word frequency and co-occurrence to identify groups of terms that are likely to occur together in the text (Cretchley et al. 2010). Classification and text analysis with the specific software is very consistent as it ends in the same results no matter how many times a data set is coded and recoded (Cretchley et al. 2010; Angus 2013). According to Ding et al. (2001), such software overcomes the problem of using keywords by indexers of specific databases and the problem of time when researchers are constructing the thesauri or classification systems themselves (see also Smith & Humphreys 2006; Angus 2013). So the text of the conference contributions was automatically coded for all years, and no prior keyword list was used in order to eliminate subjective researcher-based influence. 3.2.2 Automated extraction of keywords and co-occurrence matrices For the present study, we processed the text data for standardization in the following way: a) Extraction of keywords In Step 1, the aim was to create a coherent keyword list, extracted from the whole data set (conference papers from 1984 2. Keeping in mind that ‘concepts’ are most frequent ‘keywords,’ we use both expressions in the study for semantically talking about ‘keywords.’

to 2012). The algorithm was set by default to ensure an automatically-based extraction. So it extracted keywords that were used in different variants and classified those meanings in one concept, e.g. concept ‘business’ includes ‘business,’ ‘businesses’ and ‘business’s’ or the concept ‘brand’ integrates ‘brand,’ ‘brands,’ ‘brand’s,’ ‘branded,’ and ‘branding.’ Although most of the semantic classifications were coherent, it was necessary to evaluate the conceptual arrangements. Some of them had to be split due to the difference in their semantic meaning, e.g. ‘communication’ was divided into ‘communicator’ and ‘communication’ as the first one addresses an actor and the second one a function (for the process of vocabulary standardization see Ding et al. 2001, Hofer et al. 2010, Wuehrer & Smejkal 2012). For aims of comparability a keyword list from the entire text database was generated. b) Computation of co-occurrence matrices In Step 2, the final keyword list with 167 extracted keywords served as a filter to compute the co-occurrence matrices for further investigations of the conferences. The 167 keywords are the union set (according to set theory) of all conferences. Each year from 1984 to 2012 was separately analyzed by using the keyword list as filter, extracted and prepared in Step 1. So in total, the computation result led to 28 keyword co-occurrence matrices for each conference. Leximancer quantifies the co-occurrence of keywords that are identified on the basis of the filter applied to all of the texts of the conferences’ contributions. The matrix is the display of the main relationship between the keywords (Leximancer Manual 2011, p. 9). ‘The co-occurrence of words used in text provides useful information for a narrative inquiry on a subject’ (Liesch et al. 2011, p. 24). As such, ‘a word can be defined by its context in usage’ (Liesch et al. 2011, p. 24 referring to Smith & Humphreys 2006, p. 262). One co-occurrence matrix shows the semantic relationships between the concepts within the unit of observation (=each conference with all papers available for analysis) (Liesch et al. 2011). The 28 keyword cooccurrence matrices from 1984-2012 were used for the next step – keyword mapping analysis.

3.3 Keyword mapping - Analysis First of all, the diagonal values of each co-occurrence matrix were set to zero, to guarantee the analysis only with the co-occurrences (Hofer et al. 2010). Furthermore, the study should be based on data with five or more co-occurrence frequencies in order to minimize the probability of including some ‘noise’ in data analysis (Leydesdorff 2001; see also Wuehrer & Smejkal 2012), although there is no unique opinion about cut-off values in scientometric studies (Hofer et al. 2010). Hence, data with frequencies from 0 to 4 were set to 0. An example of a co-occurrence matrix (conference year 1984) is provided in Table 3, where the first 10 keywords are shown. The pattern shows that in this year, high co-occurrences arose of ‘market’ and ‘relationship,’ ‘supplier’ and ‘customer,’ as well as of ‘relationship’ and ‘customer.’ So, interpreting some pieces of the cut-out, the keywords ‘market’ and ‘relationship’ co-occurred 100 times in the conferences, whereas a link connects ‘market’ with ‘customer’ 127 times, and so forth. In order to visualize the networks, the authors utilized the information theoretic clustering method of Rosvall and Bergstrom (2008), which offers the following functions (for the process of implementing the method, see Wuehrer & Smejkal 2012): Identifying change in large networks; showing and visualizing important structures and individual year-wise networks. For

The IMP Journal Volume 7. Issue 3, 2013 147 Table 3 Example of cut-out of co-occurrence matrix, IMP Group conference 1984, number of keywords extracted = 123 Keywords market

market 0

relationship 100

customer 127

marketing 117

supplier 102

product 94

importance 62

company 60

firm 76

usefulness 65

relationship

100

0

131

84

118

63

81

61

43

39

customer

127

131

0

78

206

70

73

40

49

45

marketing

117

84

78

0

63

65

66

34

56

32

supplier

102

118

206

63

0

72

83

88

28

42

product

94

63

70

65

72

0

53

46

37

52

importance

62

81

73

66

83

53

0

59

26

28

company

60

61

40

34

88

46

59

0

12

48

firm

76

43

49

56

28

37

26

12

0

41

usefulness

65

39

45

32

42

52

28

48

41

0

Figure 1 Delineation of methodological process

additional analysis, we used software tools like Gephi (https:// gephi.org/) and ORA (Carley et al. 2012; Carley & Pfeffer 2012) to explore the network parameters over time. The flow of data collection and subsequent tasks are shown in Figure 1, which depicts in detail the steps already mentioned and explained.

4. Results and visualization The aim of data mapping and description of analysis was to provide the reader with knowledge about the IMP Group conferences of each year, to visualize the main topical concepts arranged in certain time periods (1984-1991; 1992-1998; 1999-2005; 2006-2012), and to offer insight into the network parameters by showing figures like density and network diameter. Results are subdivided into two main sections. First of all, we start with a descriptive analysis of bibliometric parameters and a visualization of the whole network. Next, we remain at the perspective of analyzing each year, but divide the whole data into substructures of seven years. The aim is not to lose the general conspectus. Last, we want to show the most co-occurring keywords in each time period via visualization and give insight into the dataset by that presentation.

4.1 Bibliometric parameters3 – a holistic perspective 3. For a better understanding of bibliometric and technical terms, we provide the reader with a short glossary of terms at the end of the

The keywords and the number of potential keywords stay the same over the whole time period, as we worked in previous steps with the filter list of 167 keywords (see also Section 3.2.2) and to ensure the same set of keywords for every conference. The number of keywords, extracted on the basis of analysis of each conference year varies between 112 (1985) to 149 (2012). Links represent a range from 1,982 (1985) to 16,092 (2011), where the least number of links in 1985 is strongly related to the least number of conference contributions (13) over the whole time period (Table 4). The diameter shows how far most distant topics are separated via others in the network. The highest average distance is 4 and the lowest one is 3. The results indicate that in earlier conference years, topical networks have been of wider thematic structure, which has been leveling-off at the value of 3 since 1992. Comparisons of the diameter development with other research groups and their conference contributions are scarce. Just as a reference point, the topical diameter of the AIB-conferences (Wuehrer & Smejkal 2012) is roughly double the size of the IMP Group in a year-by-year perspective. This indicates that in the later scientific community the heterogeneity is smaller, and therefore less topical ideas are presented, expressed, and discussed during such conferences. This also means semantically oriented ‘focus’ and concentration. The density of the selected keywords in the entire text corpus (= all contributions of the conpaper.


ference per year) describes how ‘dense’ keywords are connected in the overall text network. As a network proportion parameter it varies between ‘0’ and ‘1’ and is oriented on the maximum number of lines with each node in a specific network graph (Otte & Rousseau 2002). The higher the number of lines in relation to the maximum number of lines, the higher the density value. A full representation would reveal a density of 1 (Wasserman & Faust, 1994). An overtime view at the density of each network shows that especially the first topical networks were with sparser density values or less densely knitted of 0.071 (1985) to 0.239 (1989). This means that the interrelationship of keywords during the first years is very loose and as such the contribution texts are diverse, each of them bringing other perspectives and the themes are more or less disconnected. This does not mean they are uninteresting; it has the quality of ‘in statu nascendi.’ Sometimes the connectedness of keywords to each other increases, and in the next conference proceedings it is the opposite. Such phenomena typically occur when the body of knowledge in itself is questioned, research directions are unsecure and each of them struggles for importance. Variation may also be induced when heterogeneous thoughts are discussed and the conferences have something of a laboratory character and no ‘mainstream’ has developed so far, or when new scholars and participants present their ideas and research findings. The absence of variation may be explained by sociological aspects such as homophily (Bramoullé et al. 2012). Afterwards the density increased relatively steadily over time and moved up to values of 0.576 (2010), 0.577 (2011), and 0.549 (2012). The network representing the highest density occurred in 2011 (see Table 4).

results in graphs and tables when it was considered necessary for ease of interpretation. Following the work of Backhaus et al. (2011), who conducted a longitudinal analysis on the structure and evolution of b2b marketing, we followed their method of clustering time frames of a ‘minimum length of approximately 3 years’ (Backhaus et al. 2011) – exactly 7-year periods (19841991, 1992-1998, 1999-2005, 2006-2012; as medium-term business cycles, also called ‘Juglar cycles’ have a length of 7 to 11 years too (Grinin et al. 2010), we followed that approach by using 7-year periods)). Firstly, we applied the information-theoretic clustering method approach by following Rosvall and Bergstrom (2010) and Rosvall and Bergstrom (2008), respectively, in order not only to describe the topical nodes of the networks in a descriptive way, but to show structural changes (significant or not) over time, too. These steps result in a deeper insight consisting of modules (= a significant, meaningful group of nodes; the title of modules is the same as the most prominent keyword in the module; Rosvall & Bergstrom 2010) and of nodes that stand for the keywords of the conference contributions (Wuehrer & Smejkal 2012). ‘We therefore identify the modules that compose the network by finding an efficiently coarse-grained description of how information flows on the network. A group of nodes among which information flows quickly and easily can be aggregated and described as a single well connected module; the links between modules capture the avenues of information flow between those modules’ (Rosvall & Bergstrom 2008, p. 1118). So a module is a higher complex text structure of keywords, which goes beyond single co-occurrences (see Tables 5 to 8). Calculations highlight that each year is represented by one module which has varied since the beginning of Table 4 Bibliometric network parameters per year from 1984 to 2012 the IMP Group conferences between ‘market,’ NumNumber of ber of ‘firm,’ and ‘relationship’ (time period 1984Number DenNetwork Year keywords (potential) 1988) and has stabilized on the module ‘relaof links sity** diameter (=nodes) keywords tionship’ since 1988 (Tables 5 to 8). So from extracted (=nodes) 1984 123 167 4,584 0.164 3 1988 onwards the keyword ‘relationship’ and its 1985 112 167 1,982 0.071 3 specific co-occurrences coin the contributions, 1986 124 167 3,664 0.131 4 1988 127 167 6,076 0.218 3 presentations and most probably the discussions 1989 129 167 6,686 0.239 3 among the participants. 1990 134 167 9,510 0.341 4 Each of the modules is represented by a certain 1991 133 167 9,426 0.338 4 1992 137 167 10,830 0.388 3 number of interlinked keywords. The authors 1993 131 167 6,738 0.242 3 colored the presented structure of nodes (= top 1994 140 167 12,382 0.444 3 1995 135 167 10,444 0.374 3 15 keywords of every conference year) accord1996 135 167 10,004 0.359 3 ing to the reference year 1984 (Tables 5 to 8). So 1997 137 167 11,688 0.419 3 the importance of the keywords can be compared 1998 132 167 9,138 0.328 3 1999 146 167 14,986 0.537 3 by the colors year-wise. The structure of most 2000 142 167 12,672 0.454 3 prominent keywords seems to be very homoge2001 142 167 14,214 0.510 3 2002 144 167 14,506 0.514 3 neous throughout the whole time span (19842003 142 167 14,528 0.521 3 2012). Some of them are not typical for one year, 2004 142 167 13,816 0.495 3 2005 138 167 12,374 0.444 3 but represent the core of concepts independent of 2006 147 167 15,890 0.570 3 period. Keywords co-occurring in the top 15 ev2007 145 167 15,398 0.552 3 ery year are ‘relationship,’ ‘product,’ and ‘firm’. 2008 144 167 15,414 0.553 3 2009 142 167 15,286 0.548 3 The keywords ‘market,’ ‘company,’ ‘usefulness,’ 2010 145 167 16,078 0.576 3 and ‘process’ appeared in 27 of the networks (= 2011 149 167 16,092 0.577 3 2012 149 167 15,336 0.549 3 conference years), followed by ‘customer’ (26), Average figure 137* 11,419* 0.409* 3* ‘supplier’ (24), ‘different’ (24), ‘network’ (24), of parameters ‘business’ (20), ‘importance’ (19) and half of the Legend: *Figure is adapted to a round lot; ** density computed on basis of potential keywords for whole time period ‘study’ with 14 occurrences comparative reasons as demanded by QAP-procedure (Table 5). In conclusion, there is a variance of the networks regarding the top keywords. There are new or other emerging keywords differing between 2 in 1989 and 7 in 2008 4.2 Longitudinal analysis and 2011 (Tables 5 to 8). The overall pattern means homogeFor visualizing longitudinal analysis, the authors clustered the neity with variations of appearing keywords, which vanish in


the consecutive period and may reappear in some conferences years later. For instance, ‘network’ appears as a prominent topic in 1985 (Table 5), but loses that importance until 1989, finally reappearing in 1991 as second ranked within the module ‘relationship.’ These variations can be observed until 2012 (Tables 6 to 8).

The percentages in brackets after the keywords indicate the relative prominence (= page rank) of them within the linkages of the entire module (Rosvall & Bergstrom 2010; Rosvall et al. 2009; Rosvall & Bergstrom 2008). So the keyword ‘market’ in 1984 is the most prominent, it is also densely interlinked as cooccurring keyword with the rest of the network. So the page rank

Table 5 Overview of modules and 15 most prominent and significant co-occurring keywords including percentage of all flow of network (=page rank), 1984-1991


Table fotnotes 3. The keyword ‘different’ is related to the word ‘difference’ while Leximancer extracted the concept ‘different.’ The authors decided not to change that word extraction manually.

The IMP Journal Volume 7. Issue 3, 2013 150 Table 7 Overview of modules and 15 most prominent and significant co-occurring keywords including percentage of all flow of network (=page rank), 1999-2005


of a keyword within a module clarifies the prominent position of a keyword in a complex structured information flow. As such, whenever the discussion in the conference papers deals with markets, the most adjoining and related topics are, in 1984 for example (see Table 5), ‘relationship,’ ‘customer,’ ‘marketing,’ ‘supplier,’ and so on. In total, all of them are in reference to each other and generate 45.5% of all topical linkages. Here, almost half of the whole flow in form of keyword interconnections is made up by those 15 keywords. The 15 keywords coin between 32.8% (1990, 1999) and 57.0% (1985) of all keyword interconnections of each network, with a mean of 36.4%. The following graphs (Figures 2 to 5) illustrate the network structure of the most prominent co-occurring keywords in the predefined time intervals. The size of the keyword nodes and the percentages may be interpreted as such according to Wuehrer and Smejkal (2012): ‘if an avid and impartial reader of the [IMP conference proceeding] would start studying them at a randomly selected paper, and would follow randomly selected keyword links (Rosvall & Bergstrom 2011) to other papers, he/she would spend about [3.96%] of the time just in those dealing with the topic ‘relationship’. The same interpretation applies for the other percentages and sizes of keyword nodes’ (p. 553). The authors

chose to illustrate in every figure the number of keywords that first leads to the extension of the flow of 50% of the whole network. This leads to exactly 30 nodes for the periods of 19841991 and 1992-1998 and to 29 nodes for periods of 1999-2005 and 2006-2012. The period of 1984-1991 (Figure 2) is dominated by the phenomenon ‘relationship’ (3.96%) in the center of the network. Topics like ‘product’ (2.94%), ‘market’ (2.82%), ‘firm’ (2.81%), ‘supplier’ (2.33%), and ‘marketing’ (2.16%) are densely linked to the core. The network illustrates 30 nodes and 50.6% of all keyword interconnections. This means that a reader randomly selecting a conference paper would spend 50.6% of time dealing with those 30 keywords when looking through conference contributions from 1984 to1991. Keywords with less importance for the topics covered by the top 30 nodes include ‘country’ (1.01%), ‘level’ (1.05%), ‘information’ (1.06%), ‘international’ (1.08%), and ‘structure’ (1.08%). Figure 3 represents the time period from 1992 to 1998, where ‘relationship’ (4.9%) is the most prominent network concept again. The most densely linked keywords are ‘product’ (2.74%), ‘network’ (2.49%), ‘market’ (2.33%), ‘firm’ (2.33%), and ‘customer’ (2.17%). ‘Network’ appears as a prominent linked key-


Figure 2 Results of content analysis for 1984-1991; 30 co-occurring nodes represent 50.6% of network importance

word to ‘relationship.’ In general, the network shows 30 nodes representing 50.9% of network connections. The outer circle of the 30 nodes, less dominant in relation to the center, consists of keywords like ‘management’ (1.07%), ‘time’ (1.06%), ‘information’ (1.04%), and ‘approach’ (1.01%). The period of 1999-2005 (Figure 4) shows once again the keyword ‘relationship’ (5.40%) at the center of the network. In this timeframe, ‘network’ (2.89%) remains closely linked to ‘relationship,’ as do ‘business’ (2.30%), ‘firm’ (2.21%), ‘customer’ (2.20%), and ‘product’ (2.18%). The illustrated network consists of 29 nodes representing 50.2% of the total network flow. The outer circle is illustrated by keywords like ‘management’ (1.12%), ‘technology’ (1.08%), ‘information’ (1.08%), ‘system’ (1.06%), and ‘time’ (1.06%). The period from 2006-2012 (Figure 5) is not significantly different from the other networks shown (Figures 2 to 4). Twenty-nine nodes represent 50.2% of the total network flow. The key player is ‘relationship’ (4.87%), mostly linked to ‘network’ (3.09%), ‘business’ (2.74%), ‘customer’ (2.25%), ‘process’



(2.21%), and ‘company’ (2.08%). The concepts less prominent within the flow of IMP conferences are ‘organization’ (1.14%), ‘management’ (1.09%), ‘knowledge’ (1.08%), ‘time’ (1.07%), and ‘role’ (1.02%). Hence, the question arises as to what all these illustrations of certain IMP conference eras tell us about the topical development of the group. One point is that the core throughout the whole time period (1984-2012) is made up by the topic ‘relationship.’ The inner topical linkage structure of the networks varies, but not significantly. It consists of ‘market-,’ ‘product-,’ ‘firm-,’ ‘supplier-,’ ‘customer-,’ ‘business-’, and ‘process-’ related issues. ‘Network’ is not as centered in the first era (1984-1991) as it is in later periods of time (1992-2012), when it develops a stronger link to the concept of ‘relationship’ and other important keywords. Moreover, the six core concepts of each network stand for 16.96% to 17.24% of the whole network flow (exactly 1984-1991 17.02%; 1992-1998 16.96%; 1999-2005 17.18%; 2006-2012 17.24%). Nearly one fifth of the whole network flow is represented by those keywords. The overall findings here indi-


The IMP Journal Volume 7. Issue 3, 2013 152 Table 9 Correlation matrix of networks from 1984-2012 and MRQAP-based R² results for dependent variables 1985-2012

Figure 6 Development of density and MRQAP-based R² of each IMP Group conference network from 1985-2012*

* In 1987 there was no IMP conference, so the year is missing on the diagram.


cate again, as also seen by the year-wise inspection of the topical modular structure of the conferences (Tables 5 to 8), that the interconnection between a set of more or less the same keywords gains weight and importance throughout the time period.

4.3. Conference topic density and self-similarity The second research question deals with the development of the interwoven topical keywords over time. Several measures could be applied to analyze the similarity of co-occurrences of keywords. The Pearson correlation coefficient is one of them (Goshtasby 2012), which measures how the cell frequencies (i.e. the co-occurrences) in the keyword matrices are similar or dissimilar. Therefore we investigated the correlations (Table 9) between the networks by using ORA (Carley et al. 2012; Carley & Pfeffer 2012), a software tool that can be used to evaluate one or more networks (Carley et al. 2012). Kas et al. (2012) refer to Krackhardt (1987, 1992) and explain that the quadratic assignment procedure (QAP) and its regression version multiple regression-quadratic assignment procedure (MRQAP) ‘have been used to detect structural significance and compare networks in terms of their correlation’ (ibid, p. 6). ORA includes the QAP and its regression counterpart MRQAP procedure as analysis tools (Carley et al. 2012). The requirement for the application of this technique is to have the same set of keywords for all networks that are used for analysis (Kas et al. 2012, in our paper we have 28 matrices with 167x167 rows and columns - see Table 4). Firstly, we examined if the 28 networks are significantly correlated and hence, tested for correlations between the matrices (Harrison et al. 2011). The algorithms of QAP calculate Pearson’s correlation coefficient between corresponding cells of pairs of matrices. With random and repeated permutation of rows and columns of the matrices (100 permutations were used), a conclusion is allowed if the random measure is larger than or equal to the observed relations between the matrices (Oloritun et al. 2013; Dekker et al. 2007). The assumption behind this procedure is that the co-occurrences of the concepts between the different years are strongly correlated. Secondly, our aim was to investigate the development of regression models throughout the years. As such, we used MRQAP for our goal by applying ‘Double Semi-Partialing’ (DSP), as suggested by Dekker et al. (2007), which ‘uses a pivotal test statistic and is robust in the presence of skewness and multi-collinearity […]. The DSP method does partial out the effects of each independent variable on the other independent variable’ (Kumar et al. 2008, p. 22). We carried out the regression analysis by MRQAP with each conference year from 1985 to 2012 as dependent variable, where the timely previous conference years have been analyzed as independent networks. Our aim was to observe the development of R-squared (R²) as coefficient of determination. The results indicate that each conference year, and hence the topics treated there, are indeed strongly positively correlated to the previous other ones. The correlation coefficients vary between 0.469 and 0.972 (significance = 0.000). If one interprets year 1984 as the reference point to which subsequent conferences are linked, every subsequent year could be interpreted as content-correlated to a greater or lesser extent compared with the previous contributions and probably discussions. Looking at the content-wise consistence of the different conference matters, written and discussed in the years 1984 to 2012, the topics remain focused on the relationship and networks agenda and the adhering sub-topics, which may change. But when going in

detail there is a clear divide between the topical similarities as measured by keyword co-occurrence correlations, which occur around 1994/1995. In the interval before, the year-wise correlation varies stronger and the average correlations are lower than afterwards. As the correlation coefficients between the consequent years increase from 1995 until 2012, it can be assumed that the contributions by content become more and more similar, which means repetition with small pieces in news. The multiple regression results show that the structure of keywords and their co-occurrences in 2012 can be explained by 96.8% (R-squared (R²)) by the prior conferences from 1984 to 2011. The similarity between 1985 and 1984 is 54.9% or 0.741 as measured by the correlation coefficient. All of the 378 correlation coefficients have a positive sign. The density of network (Wasserman & Faust 1994) describes how keywords are intensively connected to each other. A dense text corpus is a structure of meaningful connected words. If the development of density is not correlated with the development of similarity of key word matrices as measured by correlation, then it would follow an irregular, random behavior. The development of the density of keywords (taken from Table 4) and its explanation by similarity of the structure of keywords and their co-occurrences represented by R-squared (taken from Table 9) can be modeled as a logistic curve (for justification see Vitanov & Ausloos 2012, who refer to Price 1963; Price 1961; Price & Gürsey 1975) (Figure 6). The observed values with year indices and the course of the function demonstrate that in early years the similarity between conferences topic-wise was low. In the later years there was a step increase around 1993, with some outliers in 1990 and 1998. In the consecutive conferences, density of topical keywords increases, but then starts to level off at around 0.57. The content-similarity between the year-wise science domains reaches a level of almost 97%. The logistic model explains 92.0% of variance and the constant is 249.322 (significance=0.000, with assumed upper bound=0.8).

5. Discussion of results Research question 1 asks ‘which topical keywords and co-occurrences appear in the conferences year-wise.’ The results (Tables 5 to 8) show that ‘relationship’ and ‘network’ are amongst the top 15 co-occurring keywords (28 and 24 times, respectively), whereas ‘actor,’ ‘activity,’ and ‘resource’ are included 9, 8, and 2 times, respectively, throughout all conference years. Considering the displayed figures (Figures 2 to 5), the keyword ‘resource’ co-occurs in every of the four periods; ‘activity’ and ‘actor’ emerge for the first time between 1992 and 1998, and since then have been steadily represented in the networks. This could be reasoned by the starting point of IMP2 project in 1986 and its result of ‘ARA,’ which was a source for forwarding research in the domains of ‘actor,’ ‘activities,’ and ‘resources’ in later periods of time (1992-2012). Additionally, the strong position of ‘relationship’ and ‘network’ is confirmed by the illustrated figures. It seems that the IMP projects have been the groundwork for the further progress of the network group. Solely, ‘interaction’ is neither apparent in the top 15 co-occurrence ranking nor in the presented networks of time periods. We assume that ‘interaction’ is identified as the ‘whole ensemble,’ which is composed of the elements of the interaction approach. As Håkansson (1982) points out, buyer-seller relationships and interactions are in focus and certain sub-divided elements are described to analyze industrial marketing and purchasing (Håkansson 1982).


Furthermore, the results indicate some of the concerns expressed by scholars (Cunningham 2008; Harrison 2004) familiar with the status of the scientific development represented at the conferences. They shed some light on the scientific development of topics, which are pursued by the scholarly community. As such, their investigation becomes more interesting when no other publication outlets are institutionalized. Yet the IMP Group has a remarkable position, which can be declared as a deliberate ‘academic brand’ (Cova et al. 2009), which is based on the significant content correlation of co-occurring concepts between the conference years – see also research question 2. The bibliometric parameters show a clear and consistent picture over the entire time span. The core dimensions stay the same. So, each of the projects is associated with a small core of originators. An increasing number of co-research workers came together in the years afterwards presenting and discussing their studies in conferences, exchanging ideas via article contributions. Interestingly, the early initiatives provided focus for a long time. At least there are the research threads from international business, channel theory and conventional marketing which were amalgamated into something new, the IMP approach. Even if IMP claims that the thoughts of the group are a ‘reinterpretation’ of a continuing landscape (Gadde & Araujo 2007). Anyway, a brand is continued self-similarity in a sense of consistency throughout dynamic time periods (da Silveira et al. 2013). Even, ‘[..], if the development of the IMP Group to date is perhaps best interpreted as an unplanned, but consistent quest by a number of different individuals for empirically-based answers to the question of what happens in the complex network within which business takes place’ (Cova et al. 2009, p. 573), the outcome of the endeavors is homogeneity and focus. This has positive and negative implications. On one side, researchers like Hunt (2013), Wilkinson (2001), Cunningham (2008) and Cova et al. (2009) are unique in their estimation of the amazing effort the IMP Group has demonstrated throughout the years with its influence not only on business marketing but on other scientific fields as well. It dealt with concepts considered to be innovative, imaginative, and a defiance of intellectual manner (Cunningham 2008). Furthermore, this homogeneity in concepts can be assumed to be the basis for the development of a content-consistent research stream like the IMP, as it shows that this group does not follow managerial fads and, if so, maybe to a lesser extent than other comparable groups. If we compare our results to, for example, the investigation of the AIB conferences 2006-2011 (Wuehrer & Smejkal 2012), the field of international business shows more heterogeneity and dynamism on the longitudinal view. On the other hand, concerns arise regarding homogeneity and the focus of the network, e.g. the need for new creative inputs is questioned (Cunningham 2008), and the responsibility of those researchers who are in a central position within the network and their influence on the development of the brand is pointed out (Cova et al. 2009). Harrison’s (2004) concerns may demand a network of researchers that enables a broad scope of views and empirical investigations. From our data, we can see that since the year 2006, the number of new keywords of the top 15 ranking (Table 8) has been increased in comparison to the earlier years. This could be a first indicator that new research areas are coming up and spin around without specific structure in the network. The future will show which of them are dominantly focused or combined with the traditional approaches.

6. Concluding remarks and prospects for further

studies As De Langhe (2013) states, ‘the components of a complex system cannot be understood exhaustively without taking into account their past and their relation to each other’ (p. 69). In this paper, we investigated the longitudinal way of IMP conference topics´ development (from 1984 to 2012) by conducting a content-based bibliometric review. We presented the co-occurrence and relations of most prominent keywords throughout the whole time period. What we did not want to do was to operate as carping critics of a research community, which has its outstanding merits. The contributions of the IMP Group to the field of business marketing and purchasing are remarkable (Cunningham 2008; Hunt 2013). Nevertheless, the results from this study indicate that the content of contributions at IMP Group conferences concentrates on certain topics related to the main projects of IMP (1 and 2) and could therefore be interpreted as in a stable and static development level. According to the consistent main research content, we agree with Cova et al.’s (2009) statement that IMP is an ‘academic brand.’ According to them, the question arises as to what will happen with the network and its thoughts and development if initial researchers drop out from the network ‘because of their age’ (ibid, p. 573). The knowledge embodied in the conference contributions leads to temporal and structural re-arrangements. The question is whether that affects and changes the status of ‘normal science,’ ‘paradigm shift,’ or ‘crisis’ (Bettencourt et al. 2009). An indicator for ‘normal science’ is when giant components emerge either consisting of collaborating scientists or topical keywords that are densely linked (Leskovec et al. 2005). Here, we refer to Table 4, which shows density values levelling off at around 0.5 since 1999. This also relates to the Kuhnian paradigm view, where science is seen as a process (= complex system) comprised of interactions (De Langhe 2013). But if science is a process, the IMP Group may be at an intellectual crossroads of exploiting ‘old’ knowledge or integrating new knowledge (= exploration) (De Langhe 2013; see also Kuhn 1977). If a systemic philosophy of science would be in the foreground, the IMP would look to both sides (exploiting current knowledge as well as exploring future research opportunities), including the works which have been done so far as well as to theoretical future developments. Moreover, it would then be necessary to consider further factors generally explained by De Langhe (2013) – virtues, division of labor, independence, and progress. Hunt (2013) argues that resource-advantage ‘(R-A)-theory is toward a general theory of marketing’ (p. 283) and relates it to other approaches, also to the theoretical structure of IMP. He comes to the conclusion that the investigated approaches (RA-theory, IMP perspective, Alderson’s theory, Institute for the Study of Business Markets Value Delivery Framework) have some aspects in common (Hunt 2013). Perhaps the fusion of major streams could be a source for innovative standpoints of this science domain. Relating the findings to the discussion started by Cunningham (2008), Cova et al. (2009), or Hunt (2013), we keep in mind that the results refer to a content-based analysis of the conferences of the group. Therefore, our arguments are focused on the development of the topics of IMP conferences. It is possible that the conference of 2013 was a starting point for innovative and explorative inputs, and could signal new directions. ‘Globalization’ and ‘internationalization’ would give a new impetus and lead the field explicitly toward a new and interesting domain. Against the background that IMP 1 and 2 projects have been significant drivers for the research group itself and as


A short glossary of terms Term

Description

Density

The sum of the links divided by the number of possible links (i.e. the ratio of the total link strength that is actually present to the total number of possible links). For Carley et al. (2012), a reduction of density means less connected keywords in a text network and vice versa

Flow (=page rank)

The system-wide network flow shows the interactions between keywords as nodes. This enables the weights of the links of the network to be visible (Rosvall & Bergstrom 2008, p. 1121).

Link

The representation of the tie, connection, relation, link between two keywords (=nodes) (Carley et al. 2012).

Network diameter

The diameter is the longest graph distance between any two keywords in the network (i.e. how far apart are the two most distant nodes) (https://gephi.org/ users/download/ provides a definition extracted from description within the usage of the software), or how ‘compact’ a network consisting of keywords is (http://faculty.ucr.edu/~hanneman/nettext/C7_Connection.html#geodesic).

Node

A representation of a real-world node (like people, agents, organizations (Carley et al. 2012) or keywords))

QAP Correlation

Calculates measures of nominal, ordinal, and interval association between the relations in two matrices, and uses quadratic assignment procedures to develop standard errors to test for the significance of association (Carley et al. 2012)

such for the main topics on which scientists have concentrated, the question arises whether a new IMP project could light up and initiate new inputs for the groups further away. A new label of ‘IMP 3.0’ could coin the efforts.

References Angus, D., Rintel, S., and Wiles, J. (2013). Making sense of big text: A visual-first approach for analysing text data using Leximancer and Discursis. International Journal of Social Research Methodology, 16 (3), pp. 261-267. Araujo, L., and Easton, G. (1996). Networks in socio-economic systems: A critical review. In Iacobucci, D. (ed.), Networks in Marketing. London: Sage, pp. 63-107. Backhaus, K., Lügger, K., and Koch, M. (2011). The structure and evolution of business-to-business marketing: A citation and co-citation analysis. Industrial Marketing Management, 40 (6), pp. 940–951. Baraldi, E., Gressetvold, E., and Harrison, D. (2012). Resource interaction in inter-organizational networks: Foundations, comparison, and a research agenda. Journal of Business Research, 65 (2), pp. 266-276. Barth, P. (2008). The history and the development of the IMP Group reasoning. A bibliometric approach. Unpublished dissertation. Linz: Johannes Kepler University. Bettencourt, L.M., Kaiser, D.I., and Kaur, J. (2009). Scientific discovery and topological transitions in collaboration networks. Journal of Informetrics, 3(3), pp. 210–221. Börner, K., Chen, C., and Boyack, K.W. (2003). Visualizing knowledge domains. Annual Review of Information Science and Technology, 37 (1), pp. 179–255. Börner, K., Boyack, K.W., Milojević, S., and Morris, S. (2012). An introduction to modeling science: Basic model types, key

definitions, and a general framework for the comparison of process models. In Scharnhorst, A., Börner, K., and van den Besselaar, P. (eds.), Models of Science Dynamics. Encounters between Complexity Theory and Information Sciences. Heidelberg: Springer, pp. 3–22. Bramoullé, Y., Currarini, S., Jackson, M.O., Pin, P., and Rogers, B.W. (2012). Homophily and long-run integration in social networks. Journal of Economic Theory, 147 (5), pp. 1754– 1786. Burgess, C., and Lund, K. (1997). Modelling parsing constraints with high-dimensional context space. Language and Cognitive Processes, 12 (2/3), pp. 177–210. Carley, K.M., and Pfeffer, J. (2012). Dynamic Network Analysis (DNA) and ORA. In Schmorrow, D.D., and Nicholson, D.M. (eds.), Advances in Design for Cross-Cultural Activities Part I, CRC Press, pp. 265–274. Reprinted from: Carley, K.M., and Pfeffer, J. (2012). Dynamic Network Analysis (DNA) and ORA. 2nd International Conference on Cross-Cultural Decision Making: Focus 2012, San Francisco, CA, July 21-25, 2012. Carley, K.M., Pfeffer, J., Reminga, J., Storrick, J., and Columbus, D. (2012). ORA User’s Guide 2012. Carnegie Mellon University, School of Computer Science, Institute for Software Research, Technical Report, CMU-ISR-12-105. Carley, K.M. (1993). Coding choices for textual analysis: A comparison of content analysis and map analysis. Sociological Methodology, 23 (1993), pp. 75–126. Chabowski, B.R., Mena, J.A., and Gonzalez-Padron, T.L. (2011). The structure of sustainability research in marketing, 1958-2008: A basis for future research opportunities. Journal of the Academy of Marketing Science, 39 (1), pp. 55–70. Chambers, D., Roberts, S., and Lewis-Cameron, A. (2010). Papers from conference proceedings of the 1st international tourism conference - ‘Beyond the boundary: Creating new episte-


mologies in tourism’. Tourism and Hospitality Research, 10 (4), pp. 311–314. Cova, B., and Salle, R. (2008). The industrial/consumer marketing dichotomy revisited: A case of outdated justification? Journal of Business & Industrial Marketing, 23 (1), pp. 3-11. Cova, B., Ford, D., and Salle, R. (2009). Academic brands and their impact on scientific endeavour: The case of business market research and researchers. Industrial Marketing Management, 38 (6), pp. 570–576. Cretchley, J., Gallois, C., Chenery, H., and Smith, A. (2010). Conversations between careers and people with schizophrenia: A qualitative analysis using Leximancer. Qualitative Health Research, 20 (12), pp. 1611–1628. Cunningham, M.T. (2008). Pictures at an exhibition of business markets: Is there a case for competition? The IMP Journal, 2 (1), pp. 46–59. Da Silveira, C., Lages, C., and Simões, C. (2013). Reconceptualizing brand identity in a dynamic environment. Journal of Business Research, 66 (1), pp. 28–36. De Bakker, F.G.A., Groenewegen, P., and Den Hond, F. (2005). A bibliometric analysis of 30 years of research and theory on corporate social responsibility and corporate social performance. Business & Society, 44 (3), pp. 283–317. De Langhe, R. (2013). The Kuhnian Paradigm. Topoi, 32 (1), pp. 65–73. Dekker, D., Krackhardt, D., and Snijders, T.A.B (2007). Sensitivity of MRQAP tests to collinearity and autocorrelation conditions. Psychometrika, 72, pp. 563-581. Di Gregorio, S., and Davidson, J. (2008). Qualitative Research Design for Software Users. Glasgow: Bell and Bain Ltd. Ding, Y., Chowdhury, G.B., and Foo, S. (2001). Bibliometric cartography of information retrieval research by using coword analysis. Information Processing & Management, 37 (6), pp. 817–842. Donohue, J.C. (1972). A bibliometric analysis of certain information science literature. Journal of the American Society for Information Science, 23 (5), pp. 313–317. Easton, G. (1992). Industrial networks. A review. In: Axelsson, B., and Easton, G. (eds.), Industrial Networks. A New View of Reality. London: Routledge, pp. 1–27. Easton, G., Zolkiewski, J., and Bettany, S. (2003a). An Exploratory Taxonomic Study of IMP Group Conference Papers. Paper published at the 18th IMP Conference. Dijon. Easton, G., Zolkiewski, J., and Bettany, S. (2003b). Mapping industrial marketing knowledge: A study of an IMP conference. Journal of Business & Industrial Marketing, 18 (6/7), pp. 529–544. Fernández-Cano, A., Torralbo, M., and Vallejo, M. (2004). Reconsidering Price’s model of scientific growth: An overview. Scientometrics, 61 (3), pp. 301-321. Ford, D., Gadde, L.E., Håkansson, H., and Snehota, T. (2003). Managing Business Relationships. 2nd edition. Chichester: Wiley. Ford, D. (2011). IMP and service-dominant logic: Divergence, convergence and development. Industrial Marketing Management, 40 (2), pp. 231–239. Gadde, L.E., and Araujo, L. (2007). Business recipes, historical narratives and the discovery of networks. The IMP Journal, 1 (3), pp. 2–25. Gemünden, H.G. (1998). The development of IMP - An analysis of the conference proceedings 1984-1996. In Gemünden, H.G., Ritter, T., and Walter, A. (eds.), Relationships and Networks in International Markets. Oxford, UK: Elsevier Science

Ltd., pp. 3–12. Glänzel, W. (2003). Bibliometrics as a research field - A course on theory and application of bibliometric indicators. Available from http://www.cin.ufpe.br/~ajhol/futuro/references/01%23_Bibliometrics_Module_KUL_BIBLIOMETRICS%20AS%20A%20RESEARCH%20FIELD.pdf. Accessed 8 February 2013. Glänzel, W., Schlemmer, B., Schubert, A., and Thijs, B. (2006). Proceedings literature as additional data source for bibliometric analysis. Scientometrics, 68 (3), pp. 457–473. Goshtasby, A.A. (2012). Image Registration. Principles, Tools and Methods. London: Springer. Grinin, L., Korotayev, A., and Malkov, S. (2010). The mathematical model of Juglar cycles and the current global crisis. In Grinin, L., Herrmann, P., Korotayev, A., and Tausch, A. (eds.), History & Mathematics: Processes and Models of Global Dynamics. Volgograd: ‘Uchitel’ Publishing House, pp. 138–187. Available from http://www.socionauki.ru/authors/grinin_l_e/ files/grinin_malkov_korotayev_mathematical_model_of_juglar_cycle.pdf. Accessed 1st October 2013. Håkansson, H. ed. (1982). International Marketing and Purchasing of Industrial Goods: An Interaction Approach. Chichester, New York: Wiley. Håkansson, H. ed. (1987). Industrial Technological Development: A Network Approach. London: Routledge. Håkansson, H., Ford, D., Gadde, L.E., Snehota, I., and Waluszewski, A. (2009). Business in Networks. Chichester: Wiley. Håkansson, H., and Snehota, I. (2002). The IMP perspective, assets and liabilities of relationships. In Ford, D. (ed.), Understanding Business Marketing and Purchasing - An Interaction Approach. London: Thomson Learning, pp. 35–50. Håkansson, H., and Snehota, I. (2000). The IMP perspective, assets and liabilities of relationships. In Sheth, J.N., and Parvatiyar, A. (eds.), Handbook of Relationship Marketing. Thousand Oaks, CA: Sage, pp. 69–94. Reprinted from: Ford D. ed. (2002): Understanding Business Marketing and Purchasing An Interaction Approach. London: Thomson Learning. Håkansson, H., and Snehota, I. (1995). Developing Relationships in Business Networks. London, New York: Routledge. Håkansson, H., and Waluszewski, A. (2002). Managing Technological Development. London: Routledge. Harrison, F., Sciberras, J., and James, R. (2011). Strength of social tie predicts cooperative investment in a human social network. PlosOne, 6(3), pp. 1–5. Harrison, D. (2004). IMP as fashion: Past, present and future. Available from http://www.impgroup.org/paper_view. php?viewPaper=4482. Accessed 15 March 2013. Hofer, K.M., Smejkal, A.E., Bilgin, F.Z., and Wuehrer, G.A. (2010). Conference proceedings as a matter of bibliometric studies: The Academy of International Business 2006–2008. Scientometrics, 84 (3), pp. 845–862. Huanwen, C (1996). A bibliometric study of library and information research in China. Asian Libraries, 5 (2), pp. 30–45. Hunt, S.D. (2013). A general theory of business marketing: R-A theory, Alderson, the ISBM framework, and the IMP theoretical structure. Industrial Marketing Management, 42 (3), pp. 283–293. IMP Group Homepage (2013). About the IMP Group. Available from http://www.impgroup.org/about.php. Accessed 12 March 2013. IMP Group Homepage (2013). About the IMP Journal. Available from http://www.impgroup.org/IMPJournal.php. Accessed 12 March 2013.


IMP Journal Homepage (2013). About the IMP Journal. Available from http://www.impjournal.org/. Accessed 4 April 2013. Johanson, J., and Mattsson, L.G. (1994). The markets-as-networks tradition in Sweden. In Laurend, G., Lilien, G., and Pras, B. (eds.), Research Traditions in Marketing. Boston: Kluwer, pp. 321–342 Kas, M., Carley, K.M., and Carley, L.R. (2012). Trends in science networks: Understanding structures and statistics of scientific networks. Social Network Analysis and Mining, 2 (2), pp. 169–187. Krackhardt, D. (1987). QAP partialling as a test of spuriousness. Social Networks, 9 (2), pp. 171–186. Krackhardt, D. (1992). A caveat on the use of the Quadratic Assignment Procedure. Journal of Quantitative Anthropology, 3 (4), pp. 279-296. Kuhn, T. (1962). The Structure of Scientific Revolutions. Chicago: Chicago University Press. Kuhn, T. (1977). The Essential Tension. Chicago: University of Chicago Press. Kumar, V., Krishnan, R., and Krackhardt, D. (2008). Dynamics of Network Structure and Content in an Organizational Forum. Working Paper, Carnegie Mellon University. Available from http://qbox.wharton.upenn.edu/documents/opim/seminars/ForumPaper_Nov01_2008.pdf. Accessed 4 April 2013. Lahiri, S., and Kumar, V. (2012). Ranking international business institutions and faculty members using research publication as the measure. Management International Review, 52 (3), pp. 317–340. Landauer, T., Foltz, P., and Laham, D. (1998). An introduction to latent semantic analysis. Discourse Processes. 25 (2-3), pp. 259–284. Leskovec, J., Kleinberg, J., and Faloutsos, C. (2005). Graphs over time: Densification laws, shrinking diameters and possible explanations. KDD’05 Proceedings of the eleventh ACM SIGKDD international conference on knowledge discovery data mining. doi:10.1145/1081870.1081893. Leximancer Manual (2011). Leximancer Pty Ltd. Available from https://www.leximancer.com/site-media/lm/science/ Leximancer_Manual_Version_4_0.pdf. Accessed 18 December 2012. Leydesdorff, L. (2001). The Challenge of Scientometrics. The Development, Measurement, and Self-Organization of Scientific Communications. Parkland, III: Universal Publication. Liesch, P.W., Håkansson, H., McGaughey, S.L., Middleton, S., and Cretchley, J. (2011). The evolution of the international business field: A scientometric investigation of articles published in its premier journal. Scientometrics, 88 (1), pp. 17–42. Lindgreen, A., Hingley, M.K., Grant, D.B., and Morgan, R.E. (2012). Value in business and industrial marketing: Past, present, and future. Industrial Marketing Management, 41 (1), pp. 207–214. Lisée, C., Larivière, V., and Archambault, É. (2008). Conference proceedings as a source of scientific information: A bibliometric analysis. Journal of the American Society for Information Science, 59 (11), pp. 1776–1784. Matsuo, Y., and Ishizuka, M. (2004). Keyword extraction from a single document using word co-occurrence statistical information. International Journal on Artificial Intelligence Tools, 13 (1), pp. 157-169. Meadows, J. (2005). A practical line in bibliometrics. Interlending & Document Supply, 33 (2), pp. 90–94. Morlacchi, P., Wilkinson, I.F., and Young, L. (2004). Social Networks of Researchers in Business to Business Marketing:

A Case Study of the IMP Group 1984-1999. University of Sussex. Brighton. Available from https://www.sussex.ac.uk/ webteam/gateway/file.php?name=sewp116.pdf&site=25. Accessed 16 January 2013. Muñoz-Leiva, F., Viedma-del-Jesús, M.I., Sánchez-Fernández, J., and López-Herrera, A.G. (2012). An application of coword analysis and bibliometric maps for detecting the most highlighting themes in the consumer behaviour research from a longitudinal perspective. Quality & Quantity, 46 (4), pp. 1077–1095. Neuendorf, K.A. (2002). The Content Analysis Guidebook. Thousand Oaks California: Sage Publications. Oloritun, R.O, Madan, A., Pentland, A., and Khayal, I. (2013). Identifying close friendships in a sensed social network. Procedia – Social and Behavioral Sciences, 79 (2013), pp. 18–26. Otte, E., and Rousseau, R. (2002). Social network analysis: A powerful strategy, also for the information sciences. Journal of Information Science, 28 (6), pp. 441-453. Payne, N., and Thelwall, M. (2009). A longitudinal analysis of alternative document models. Aslib Proceedings, 61 (1), pp. 101–116. Price, D.J. de Solla, and Gürsey, S. (1975). Some statistical results for the numbers of authors in the states of the United States and the nations of the world. In Institute for Scientific Information (ISI) (eds.), Who is Publishing in Science 1975 Annual. Philadelphia: Institute for Scientific Information, pp. 26–34. Price, D.J. de Solla (1963). Little Science, Big Science. New York: Columbia University Press. Price, D.J. de Solla (1961). Science since Babylon. New Haven: Yale University Press. Price, D.J. de Solla (1956). The exponential curve of science. Discovery, 17(1956), pp. 240–243. Robert, C., Wilson, C.S., Donnadieu, St., Gaudy, J.F., and Arreto, Ch.-D. (2008). Bibliometric analysis of the scientific literature on pain research: A 2006 study. PAIN, 138 (2), pp. 250–254. Rosvall, M., Axelsson, D., and Bergstrom, C.T. (2009). The map equation. The European Physical Journal Special Topics, 178 (1), pp. 13–23. Rosvall, M., and Bergstrom, C.T. (2008). Maps of random walks on complex networks reveal community structure. PNAS, 105 (4), pp. 1118–1123. Rosvall, M., and Bergstrom, C.T. (2010). Mapping change in large networks. PLoS ONE, 5 (1), pp. 1–7. Rosvall, M., and Bergstrom, C.T. (2011): Multilevel compression of random walks on networks reveals hierarchical organization in large integrated systems. PLoS ONE, 6 (4), pp. 1–10. Samiee, S., and Chabowski, B.R. (2012). Knowledge structure in international marketing: A multi-method bibliometric analysis. Journal of the Academy of Marketing Science, 40 (2), pp. 364–386. Sanz-Casado, E., Suárez-Balseiro, C., Iribarren-Maestro, I., Ramírez-de Santa Pau, M., and Pedro-Cuesta, J. de (2007). Bibliometric mapping of scientific research on prior diseases, 1973–2002. Information Processing & Management, 43 (1), pp. 273–284. Scharnhorst, A., Börner, K., and van den Besselaar, P. (2012). Foreword. In Scharnhorst, A., Börner, K., and van den Besselaar, P. (eds.), Models of Science Dynamics. Encounters between Complexity Theory and Information Sciences. Heidelberg: Springer, pp. vii–x. Sidiropolous, A., and Manolopoulos, Y. (2005): A new perspective to automatically rank scientific conferences using digital


libraries. Information Processing and Management, 41 (2), pp. 289–312. Smith, A.E. (2000). Machine mapping of document collections: The Leximancer system. In Vercoustre, A.M., and Hawking, D. (eds.), Proceedings of the Fifth Australasian Document Computing Symposium. Australia: Sunshine Coast, pp. 39–43. Smith, A.E., and Humphreys, M.S. (2006). Evaluation of unsupervised semantic mapping of natural language with Leximancer concept mapping. Behavior Research Methods, 38 (2), pp. 262–279. Sousa, F.J. (2010). Business Relationships and Corporate Nature and Scope: A Critical Realist Analysis. Saarbrücken: LAP LAMBERT Academic Publishing AG & Co. KG. Stockwell, P., Colomb, R.M., Smith, A.E., and Wiles, J. (2009). Use of an automatic content analysis tool: A technique for seeing both local and global scope. International Journal of Human-Computer Studies, 67 (5), pp. 424–436. Turnbull, P., Ford, D., and Cunningham, M. (1996). Interaction, relationships and networks in business markets: An evolving perspective. Journal of Business and Industrial Marketing, 11 (3/4), pp. 44–62. Turnbull, P.W., and Valla, J.-P. eds. (1986). Strategies for International Industrial Marketing. The Management of Customer Relationships in European Industrial Markets. London, Dover, N.H: Croom Helm.

Vitanov, N.K., and Ausloos, M.R. (2012). Knowledge epidemics and population dynamics models for describing idea diffusion. In Scharnhorst, A., Börner, K., and van den Besselaar, P. (eds.) Models of Science Dynamics. Encounters between Complexity Theory and Information Sciences. Heidelberg: Springer, pp. 69–125. Wasserman, S., and Faust, K. (1994). Social Network Analysis. Methods and Applications. Cambridge: University Press. Wiles, L., Olds, T., and Williams, M. (2010). Evidence base, quantitation and collaboration: Three novel indices for bibliometric content analysis. Scientometrics, 85 (1), pp. 317–328. Wilkinson, I. (2001). A history of network and channels thinking in marketing in the 20th century. Australasian Marketing Journal, 9 (2), pp. 23–52. Wilson, T. (1995). An integrated model of buyer-seller relationships. Journal of the Academy of Marketing Science, 23 (4), pp. 335–345. Windischhofer, R., Järvelin, A.M., and Mittilä, T. (2004) A Review of IMP Conferences 1998 and 2003 with Special Focus on Dynamics. Paper published at the 20th IMP Conference. Copenhagen. Wuehrer, G.A., and Smejkal, A.E. (2012). The knowledge domain of the academy of international business studies (AIB) conferences: A longitudinal scientometric perspective for the years 2006–2011. Scientometrics, 95 (2), pp. 541–561.

Gerhard A. Wuehrer, Univ.-Prof. Dkfm. Dr, Johannes Kepler University Linz, Austria, Department of Retailing, Sales and Marketing, Altenbergerstrasse 69, 4040 Linz, Austria. Phone: +43-732-2468-9401, Fax: +43-732-2468-9404. E-mail: gerhard.wuehrer@ jku.at Angela Elisabeth Smejkal, Mag. Dr., Johannes Kepler University Linz, Austria Department of Retailing, Sales and Marketing, Altenbergerstrasse 69 4040 Linz, Austria


IMP theory in light of process- and system theories Per Ingvar Olsen BI Norwegian Business School

Abstract This paper compares and discusses IMP theory in relation to Process Theory and Systems Theory, and also relates it to the ANT. It contrasts and illuminates ontological and epistemological underpinnings, methodologies and theories while also throwing some light on their historical roots and mutual influences over time. It argues that IMP is representing a process based economic theory of value creation that is rooted in a fundamental critique of main stream economic and marketing theory as well as in a deviation from the basic conceptions of the systems theories by which it was influenced in an early phase. From there the IMP developed more flexible analytical constructs that show more kinship with process thinking than with systems thinking. The IMP is also found to distinguish itself from both ANT and Systems Theory in maintaining a symmetrical and reciprocal conception of interaction where the others in different ways focus on the unification of an acting unit in relation to an external environment. This is a particular strength of IMP, but also a challenge that calls for further clarification of analytical concepts. Key words: Keywords: IMP theory, industrial networks, process, systems, ANT, ontology, epistemology

1. Introduction The purpose of this article is to compare and discuss essential aspects of the IMP tradition in relation to two areas of theorising that may be seen as relatively close to it, or even as part of it: process theories and systems theories. I will do this with a particular emphasis on discussing how these three areas differ in their fundamental approaches and conceptualizations, and what implications this may have for their modes of research and theorizing. In addition to comparing their fundamental underpinnings, I will address and discuss a few selected topics, such as the different conceptions of agency in IMP and Actor-Network Theory (ANT) and particular characteristics of systems thinking such as its conceptual understanding of internal hierarchy and complexity. One way to portray the relationship between these three areas of theory is by seeing IMP in relation to the two others, with process theory leaning more towards emergent phenomena and systems theories leaning more towards unity of analytical entities, their internal characteristics and the internal/external distinction than does the IMP.

1.1 Roots of process and systems thinking in IMP As a point of departure, it should be noted that also the IMP can be seen as fundamentally rooted in a process view of the world - and of economic activities in particular. It departs from the assumption that resources are heterogeneous and cannot be fully known by anyone. Hence, to relate and to interact is a necessary requirement for both knowledge and value creation. In this per-

spective, the economy must be constituted by interactional processes of learning and value creation that includes the mental, the social as well as the natural/material elements of the world. In essence, this process view seems to be fundamental to how the human experience of creation of society is understood by all three areas of theory. IMP can also be seen as having roots in early systems thinking, in particular in early economic systems theory (Boulding, 1956), in systems approaches to the early developments of the management sciences and operation research (Churchman, 1955, 1994), in Nordic industrial systems analysis (Johanson and Mattsson, 1986) and in the early understanding of distribution systems (Gadde, 2000; 2011, Gadde & Ford, 2008). The latter links back to Breyer’s (1937) system oriented approach to a theory of distribution channels, and to Alderson (1965) and his functional and system based holistic theory of marketing and distribution. In their view, the interpretations of marketing and distribution operations concern how these contribute to service the system as a whole. The distribution system was seen as a functional unity with need for internal order and systemic management, with boundaries that separate it from an external environment that it is acting upon. The IMP network approach can in part be be seen as a critique of this claim for unity and internal/external distinction in systems thinking – and as a search for a different approach to conceptually permit more open and more divergent interactions such as represented by dyads, triads and nets, while it continued to build on the functional and structural insights of the early system theories. IMP does indeed have a diverse set of origins, and has continued to explore contributions from both process thinking and systems theory over the years. Thus, a comparative


analysis of distinctions between IMP, process and systems thinking should be of relevance to many IMP researchers that are already engaging across the boundaries of these theory traditions. The IMP approach is leaning towards what we may denote as “material economic analysis” focusing on the actual, observable economic activities, as opposed, for instance, to focusing on the importance of particular meaning or intent on the side of the participating actors, dyads or networks. As such, the IMP is an empirical, materially oriented real-economy domain of research and theorising. As a consequence, the networks that typically make up the subjects of the IMP narratives are those that are related through their actual economic value creating and -distributing interactions, not through their sharing of ideas about what they want to achieve and how they may be going to get there - which would typically be the subjects in, for instance, a business venturing narrative. In IMP conceptions and analytical constructs, those that are related, interacted and interdependent are not ex ante identified as representatives of similar or co-ordinated intentions. Relationships, interactions and interdependencies are generalized beyond specific meaning and are – in this sense - primarily social-materially represented. As a consequence, struggles over intents and objectives are just as easily found and analysed within dyadic or network relationships as in between rival dyads or networks. In this sense, IMP sees ideas as heterogeneous “entities”– just like resources. And, just like resources, they must interact with others if they are to generate impact and value.

1.2 Representation of “meaning” in IMP analysis However, the role of “the mental” keeps challenging IMP researchers. For instance, the introduction of mental “network pictures” by some may be seen as an attempt to incorporate the roles of particularly influential ideas or “idea-nets” into IMP analysis, by arguing that such phenomena have a formatting impact on how certain networked economies emerge and change (Henneberg, Mouzas and Naude, 2006). One question then is whether this, for instance, should be seen as the introduction of “ideas” as a fourth category in the ARA-model, in addition to resources, activities and actors. Does it add anything to introduce a mental dimension to the entire approach of the general IMP theory? Or could the idea of a mental “network picture” imply some kind of unified, intentional IMP networked agency? The first approach has been discussed, for example, by Håkansson and Waluszewski (2002), who see interacted network structures as associated with idea structures that are more or less divergent and typically more extended than the activated structures. These idea structures can be seen as representing “interacted heterogeneous ideas” that in abstract terms are analytically quite similar to “interacted heterogeneous resources” and hence in congruence with fundamental IMP conceptions. The second approach, however, points in a different direction, by introducing a sort of “one-sided intentionality” on the side of a networked collective, which does not have an immediate relational counterpart, but rather opposes some “separate opponent” seen as a different interrelated network representing a different “network picture”: an external competitor. And both of them are seen as intentional actors in an environment that appears as a competitive battleground where intentional actions are focused on rather than interactional reactions. Hence, the introduction of homogeneous intentionality on the side of a network implies a deviation from the fundamental approach of the IMP theory and turns it over to a more traditional emergent, organizational

and market oriented mode of thinking. As such, these efforts are quite illustrative of what IMP is and what it apparently is not.

1.3 Outline and ambition of article I would like to start this exploration into the three areas of theorising by briefly introducing and discussing a few basic conceptions from within early 20th century process philosophy1 – in relation to the noted IMP fundamentals. Process philosophy is also where Actor Network Theory has obtained a substantial share of its inspiration (Harman, 2009; Latour, 1993), and hopefully this introduction of some common roots will contribute to a clarification of important similarities as well as differences. General Systems Theory appears to have had its major roots in thinking related to biological systems rather than business systems (von Bertalanffy, 1945, 1950,1951,1968; Huchington, 2011; Maturana and Varela, 1980; Quastler, 1953). However, it also emerged in relation to industrial thinking and the early management sciences, not at least through the systems philosopher and early initiator of operations research, West Churchman (1955, 1994), and the American economist Kenneth Boulding (1956). General Systems Theory has also been included in sociology, not least by Luhmann (1984), and has made substantial advances into the area of information and communication science and technology (Simon, 1965). The ambition of this paper is primarily to contribute to the understanding of analytical differences at a fairly abstract level of discussion, and to discuss and exemplify how these differences may be productively used in empirically based studies. This will typically imply some degree of simplification in terms of fully representing the true breath and internal complexities of the three domains of theories addressed. Hence, obviously, more rigorous work should be done to explore the actual richness of what in this case is sacrificed for the purpose of simplicity.

2. Methodological considerations The industrial network-, process- and systems theories that I will present and discuss represent three different modes of thinking about relatedness, interaction and interdependencies. To address this broad topic will require positioning in relation to some general philosophy of knowledge. In addition to this, I have also used this opportunity to develop and present a simple, general model of scientific theorizing, in order to provide a schema for a structured discussion of topics that are complex and that move between perspectives, theories and analytical conceptions at multiple levels of analysis.

2.1 A relational philosophy of knowledge To take the knowledge philosophy first, I position this discussion within a fundamentally relational, realist perspective on knowledge, which presupposes that we can only have knowledge of phenomena that we somehow relate to. Everything else is beyond what we – as human beings – may know of (Whitehead, 1978; Bergson, 1975, 1983; Felt, 2002). This view is fully consistent with the IMP perspective on knowledge, which argues that the world is made of heterogeneous entities that are “unknowable” in the total sense (Håkansson et al, 2009). To relate and to interact is a fundamental requirement, a necessity in order for us to have knowledge of these heterogeneous resources, and 1. For a discussion of process philosophy, see Olsen 2011, Rescher 1996, Leclerc 1990


for us to exploit them for economic objectives. Secondly, I assume that perspectivism is fundamental to any knowledge we may have about the empirical universe (Felt, 2002). We may only make sense of the empirical if the brain is capable of ordering some conceptual framing by which to obtain focus and interpret meaning from the streams of complex sensing of whatever is empirical. Any given conceptual framing represents a disciplining of the mind to comprehend a perspective that is thereby fundamentally limited. Finally, I maintain that every perspective is also constrained by some horizon beyond which we are no longer able to observe. We may move the horizon by introducing technologies to see further, but we cannot fully remove it. Hence, this discussion is rooted in a relational and perspective-horizontalist philosophy of knowledge that I believe is common to both psychologists and phenomenologists alike (Dreyfus and Dreyfus, 1986, 1988). Perspectivism implies that, in order to maintain the intellectual freedom to evaluate what we know from within a particular perspective, we need to revert to a different point of observation from where to apply a different perspective from which to “see things differently”. That is, we may use opportunities for triangulation of perspectives to escape from “logical conceptual traps”.

2.2 The structure of scientific investigation and theorising Based on this general epistemological positioning, I maintain that the different areas of theorising we typically associate with paradigms fundamentally separate at the level of mental, conceptual framing, not at the level of empirically based theories. A new paradigm results from the successful creation of a different perspective, with a new conceptual framing that turns out to be gradually acknowledged as knowledge-productive. It does not seem to result from rejection caused by falsification of theory concerning the empirical as advocated by Popper. Paradigmatic shifts may result from declining returns to old paradigms and increasing or higher returns to new. Only theory seems to change as a result of new empirical evidence, not paradigms. The well-known Lakatosian version of this argument is that the conceptual core of a paradigm is based on some tautological, irrefutable logical construct that is protected by “a belt” of supporting assumptions so that the core is well protected (Lakatos, 1977). It follows from this that what we call “scientific theories” result from the interactions of analytical core constructs with empirical observations, interpretations, measuring etc. to form disciplined explanations about the empirical in the perspective essentially ensured by the given conceptual core. Theory (propositions about the true nature of something) depends on empirical content, where the more general theories cover a broader set of empirical phenomena than medium range theories or theoretical propositions. This internal structure of scientific theorising causes the problem that the process philosopher Alfred N. Whitehead denoted as “the fallacy of misplaced concreteness”. It results from the possibility that the thinker may confuse a simplistic deduction from a purely mentally generated analytical construct in order to claim an explanation about something of empirical matter, rather than from actually investigating into the empirical matter – which Whitehead argues is a necessary requirement for obtaining actual knowledge about it (Whitehead, 1978 (1929)). Hence, the knowledge claim with respect to the concrete empirical object is misplaced. To Whitehead, this was central to his cri-

Figure 1: Model of scientific theorising

tique of historical rationalism within philosophy of knowledge at the time - in the late 1920s (Olsen, 2011). Nowadays, this is referred to as the tendency to mix up the context of justification of analytical constructs with the context of empirical investigation (Bromley, 1990). An example of this would be to claim that “all humans are maximizing their individual utility function” is a true statement about the empirical if the proof of this claim is that this must be so because behaving differently would indicate that the individual has a different preference function, which, when properly accounted for, will perfectly explain his or her choices in life. This conclusion is simply derived from the conceptual, logical combination of individual preferences, individual utility and individual choice to explain any individual decision whatsoever. Theory about the matters of the world – as opposed to mental analytical constructs and speculative metaphysics - must accordingly be based on empirical investigations into something that exists in time and space. This implies that theories about something empirical typically call for more than one paradigm/ perspective-horizon due to the influences of the empirical on the thinker. Hence, induction from grounded empirical studies calls for a plurality of perspectives and theories, whereas deduction typically ensures in-depth and more focused modes of investigation and theory development. Induction is, accordingly, a divergent process, whereas deduction is typically a lot more convergent. As a result, we may easily think of science disciplines as containing more than one paradigm. To somehow visualise these processes of scientific knowledge production, I have outlined the following simple model of the hierarchy of scientific theorising in normal sciences. The model will – hopefully – also serve as a useful point of reference in the following discussion of the three different domains of theory. The model places “general theory” at level 4 in between two different kinds of processes: the mental creative processes of mind at level 1 and 2 that generate propositional analytical framings (fundamental ontological assumptions), and the processes of the real world at level 7 that represent the empirical realities to be studied (Leclerc, 1990; Olsen, 2011). General theories result from the iterative interactions between deductive reasoning (top down) and inductive observations, interpretations and propositions about the empirical (bottom up). The combining is required for empirical data to be identified, selected and processed in a meaningful way. It furthermore illustrates the existence of various levels of knowledge generating work at which there may be inductive-deductive loops and logical implications. Note that deduction typi-


cally connects from level 2 to level 7, whereas induction from the empirical – within a given paradigm – typically runs from level 7 to level 3 only, due to the said protection of the analytical core of the paradigm. The three domains of theory, which I will now present in brief, discuss and compare, fundamentally differ at level 2 in the above model, whereas they of course may all be applied to study the same empirical phenomena at level 7. The result of the combining will be different. One important implication of the discussion above is that triangulating from points of observation within different theoretical domains of this kind may offer sound opportunity to avoid “the fallacy of misplaced concreteness”. To be able to move freely between different perspectives is also a way that researchers may avoid the particular limitations represented by the perspective-horizon constraints of any given theory.

3. The IMP theory The fundamental ontological position defining the IMP as an area of theory is the assumption that resources are heterogeneous and only partially knowable, and as a consequence, that relations and interactions are fundamental constituents of the economy. In the most simplistic representation this concerns a dyadic relationship between two (business) entities. From there, the analytical constructs have extended in multiple dimensions to represent complex and dynamic network models. The tradition emerged as a primarily European research tradition in the early 1980s, not at least in opposition to the completely dominant competitive market based view of marketing and purchasing at the time. It has come to represent the possibly most influential European contribution to the otherwise US dominated area of economic marketing theory. Major contributions to the tradition include the three seminal books by Håkansson 1982, Håkansson and Snehota 1995 and Håkansson et al 2009. These may be said to represent a broad representation of a dominant strand in the emergence of the tradition through nearly three decades. Other important early contributions have been presented in Axelsson & Easton 1992, by Ford (1980, 1990) and by Johanson & Mattsson (1985, 1986). Over these years - and as many scholars have expanded the range of issues addressed, such as to innovation and supply chain management -- the tradition has become quite diverse, as for instance illustrated by the efforts to incorporate “mental pictures” as another basic category into IMP analytical modelling (Henneberg, Mouzas & Naude, 2006).

3.1 A brief extract of core IMP concepts In the following I will omit most of these diversities by staying close to what I see as the essentials of the analytical and theoretical contributions of the domain. One way to do so is to obtain a brief extract from IMP theory essentially pulled from a recent and rather dense presentation of the essence of IMP Group theorising by Ford et al. (2011: 82-103). Their presentation argues that to IMP the idea has been central that substantive and lasting interactions between individual (organised) economic agents is a core characteristic of business and economic landscapes. IMP research describes and explains how such interactions provide the means of economic, industrial and marketing coordination across firms. Interaction is the major means by which companies systematically combine their resources, activities and actors with each other in order to harvest collective gains from such combining. These collective gains are the values created in the

economy. It is a ubiquitous process as well as a working structure for the network (ibid: 83). This ontological idea of “business interaction of heterogeneous resources, activities and actors” accordingly constitutes IMP as a research program (Figure 1: level 2) (ibid: 85-86). Based on this, the analytical apparatus is expanded to networks by connecting more economic actors to the original dyadic model – which then analytically defines “an industrial network” (ibid: 87). These models provide a basis for analysing both structural and dynamic outcomes and issues. It is clear that there is a process view within IMP associated with “network interaction”, seen for instance as network changing, problem solving or interface adjusting activities aimed, for example, at creating more effective interfaces across organisational or other kinds of borders. Hence, the IMP is based on a process view as well as structural analysis of the empirical, with the analytical modelling approach focusing primarily on the structural side - to structure analysis of dynamic phenomena. The content of the interactions as well as the substances that are being connected are seen as multidimensional and multiple in kind. To deal with these, the IMP has developed a number of analytical categories and models, of which the ARA model (activities-resources-actors) is particularly important for the entire bulk of methodological constructs (Figure 1: level 3). It quite simply suggests that a given business relationship may be seen as separated into different “layers”, such as “actor layer, activity layer and resource layer”, corresponding to the typology included in the ARA model (ibid: 88-89). Given this approach, the process dimension is in part addressed through a concept of “friction”, which is only partly similar to how friction is understood in mechanical physics. In IMP it is seen more prominently as a “ubiquitous force within network relationships” that results from the impacts of different intentions and interests on each side of a relationship. “Friction” is a succession of mutual reactions – or mutually influencing forces across relationships. Hence, the dynamic – or “creational” perspective in IMP is not focussed on growth or expansion of business entities but on the dynamics that are being created through reciprocal interactions. It also works on human minds – as something that causes more intensified interaction processes to innovate, to adjust interfaces or otherwise to alter business relations or aspects thereof. The IMP concept of friction is rather an attempt at expanding from a structural theory to a theory that also incorporates what we may call “creative processes” or “creations” in direct association with the IMP core (business relationship/interdependence). Rather than establishing a theory that is explicitly based on a separate or genuine idea about “creative processes of becoming of new entities”, the IMP maintains a disciplined connecting of what is creative to the structural core of its domain. It accordingly argues that there is no such thing as a new network, only deviations from already (historically) existing networks caused by sufficiently strong friction forces (ibid :88). There is no need for a theory of how business networks emerge in the first place, because there can be no such processes. By deductive inference, no business can emerge from something not networked, an argument which of course may not result from an empirically grounded theory, but also from a reasonable, as well as logical, justification of the particular analytical construct applied to the context. The IMP has also moved to explicitly incorporate the dimensions of both time and space, where time of course represents an elaboration into the dynamics of interactions and network-


ing over time, whereas space rather relates to what characterises the outcomes of the dynamics observed across different business landscapes. When explaining the particular dynamics and outcomes of interactions across resource and activity layers and across time and space dimensions - based on a substantial number of detailed research projects - the IMP seems to approach what we may denote as a “general theory of business interaction” (ibid:94) (Figure 1: level 4). In this theory a variety of theoretical propositions are put into a theory that explains how business networks over time develop to form particular characteristics, such as increased path dependency of resource constellations, more co-evolution across actor webs and more specialisations across activity sets over time. Furthermore, it also aims at explaining the observed increased heterogeneity of resource constellations, the observed expansion of interactions across actors and the observed higher degrees of interdependency across activities in the spatial dimension. This “General IMP Theory” is essentially contained within the domain of industry, logistics, marketing and purchasing research, but also incorporates elements of technology development and innovation theory, as a number of the empirical research projects either focus on or touch upon such issues.

3.2 The concept of value creation in IMP There are also some attempts made at pulling into more explicit economic theorising. These are efforts aimed at developing a theory rooted in “interaction” and “interdependency” as the core logic of value creation. The value of an entity is argued to depend on what it is being connected to – a basic understanding rather similar to the “value creation as re-combination of resources” argument in Schumpeterian innovation/creative destruction theory (Schumpeter, 1939). Furthermore, the economic value will depend on how this interconnecting is performed within the interrelated networks of activities, resources and actors. The IMP in this way provides the important understanding that to any business entity, relations, interactions and some degree of interdependency represent necessary links to others who thereby “decide” on the value of the given business activities. The underlying logic of this is quite similar to the understanding in semiotic thinking that the meaning (and value) of, for instance a symbol, fundamentally depends on how the symbol is related to other symbols such as when a word is changed by taking away a letter, when letters in a word are ordered differently, or when the positions of words in a sentence are changed, or when another word is brought into the sentence. A structural change immediately results in a change in the meaning of the word or the sentence. Hence, in relation to “the meaning of value”, the IMP applies an analytical approach in which specific network structure and specific (interpretations of) value are simultaneously defined. A change in network structure corresponds to an immediate change in economic value, and vice versa. Below the general IMP theory (Figure 1: level 4), there is a successive emergence of additional or adjusted medium range IMP theories – typically addressing particular settings, such as the relationship between the development, the manufacturing and the marketing context of a certain innovation project and -process, the variety of networked structures across or within various industries and markets, the particularities of technology development networks, etc., that represent contributions at level 5. Most of these are based on detailed, almost anthropological case studies that contain streams of analytical propositions suggested by the researchers as creative reflections of their obser-

vations, typically interpreted within the perspective-horizon of the general IMP theory (Figure 1: level 6). These propositions are the building blocks for the continuous expansion of the IMP theory to deal more adequately with the economic phenomena it addresses.

3.3 The relationship between networks and their “meaning” I find that - in relation to the IMP conceptual framing of relationship, interaction and interdependency and networks thereof as outlined above -- the role of particular representations of meaning, purpose and intent among business actors represents an important analytical challenge to the IMP. This seems to follow as a consequence of the emphasis on processes of reciprocal reactions in relationships and interactions rather than on unity of interacted entities. It represents the difference between traditional economic and marketing approaches based on the idea of intentional business unities and IMP’s reciprocal reaction based approach. As a consequence, the IMP becomes somewhat “neutral” or “distant” to issues associated with intentionality by downplaying the roles of subjective agency, of target oriented mobilization processes, of power strategies and dominance with respect to relational and network control. Potentially important dimensions of rivalry and competition appear to be difficult or at least complex to address. Because the specific meaning of a relationship between two or more parties is not seen as unified or co-ordinated, it is also somewhat complex to make sense of how exactly a change in a relationship crosses over the borders into the entities at the two sides of a relationship to change what they are. It is thereby – conceptually - hard to grasp how the two parties actually move or change in relation to each other. In this perspective, the IMP approach to conceptualizing “value creation” is interesting, as in this case the meaning category “value” is co-defined with its associated network structure, by which we may directly analyse how a change in network structure has value effects on the calculations and accounts within the participating entities (businesses). Value calculations within each of the participating units thereby represent agency forces with substantial effects on network coordination and interaction. We may also note that in this case economic value is seen as directly represented by the given network, while at the same time we may also have different views of the actual value represented within the network. The specific economic value for each participant as well as the overall assessments of the united value of the network may still be contested. In a similar way, we may argue that any given social-material network structure represents a particular idea (or meaning) structure that is partly representing unity, partly contested representations. However, it may also be that these issues have been addressed elsewhere and that pulling from other research domains may at least offer comparative benchmarks for a discussion of how the notion of “specific meaning” may relate to “specific network” in terms of how we understand relationships and interaction as well as the dynamics between inter-relational and inter-organizational change in an overall network based theory such as the IMP. Through the processes of IMP research and theorising, there are typically a number of such empirical observations that are well described but may be quite difficult to conceptualize. This calls for explorative approaches (such as “the roles of mental pictures”) for expanding from the paradigm’s dominant core (such as adding a mental dimension) or for further efforts to de-


velop additional analytical constructs that are genuinely based on the IMP heterogeneous resources and relational interdependency approach. A number of IMP descriptive and discussant texts address these kinds of issues – leading to various efforts to create additional analytical constructs (Medlin, 2002; Leek et al., 2003; Dubois and Araujo, 2004; Waluszewski, 2004; Håkansson and Waluszewski, 2007; Håkansson et al., 2009; Axelsson, 2011, Gadde, 2011; Hulthén & Mattsson, 2011). Others suggest combining IMP theorising with analytical approaches that are more specifically oriented towards analysis of emergent processes – such as Actor-Network Theory (ANT) (Araujo, 1998; Mattsson, 2003; Hoholm, 2011).

4. Process philosophy - and the ANT approach In the following I will briefly present some aspects of process philosophy before moving to a discussion of how the ANT version of process thinking has addressed and conceptualized issues related to this discussion of relationships, interactions and interdependencies. Philosophical debate took a rather dramatic turn as a result of Darwin’s evolutionary theory, which at the time fundamentally challenged the prevalent understanding of world creation with respect to God. For instance, the new process philosophical, evolutionary perspective stimulated a new interpretation of the role of human creativity and knowledge in relation to all aspects of human society and nature. The biological idea of evolutionary process and progress translated into a process theory of societal progress. This occurred in the last decades of the 19th and the first part of the 20th century, a time when science, new technologies, mass-product industrialisation and speedy communication really impressed the world. At the level of metaphysical philosophy, philosophers like Charles Sanders Pierce (1839-1914), William James (1842-1910), Henry Bergson (1859-1941), Alfred N. Whitehead (1861-1947) and William H. Sheldon (1875-1952) came up with a variety of analytical approaches to formulate an overall interpretation of this new world. Pierce and Bergson based their approaches on constructs derived primarily from biology, Whitehead and Sheldon from different areas of physics (quantum theory and electromagnetism respectively), whereas James based his approach on core conceptions from within the new area of psychology (Rescher, 1996). Process philosophy also had a substantial impact on epistemology. In particular the British philosopher Alfred N. Whitehead came to represent a radical rejection of traditional philosophy of knowledge by formulating, in strict scientific terms, the modern interpretation of science as we now generally perceive it: as propositional theories about the world – not necessarily as true statements rooted in some kind of direct intuition between the thoughtful mind and the empirical world as represented, for instance, by mathematics. Thus, he diverged from the idea of science as based on “true and certain mathematical laws of nature”, such as within the historical philosophies that came to dominate in the 17th, 18th and 19th centuries (Leclerc, 1990; Olsen, 2011). Whitehead’s new philosophy of creation is rooted in the understanding that creation results from “processes of interaction”. To put it simply, there are purely mental processes of mind and there are actual processes in nature and society, which he denoted as “physical processes” (physical prehension in Whitehead’s wording). These physical processes act upon others. Hence, not only humans (souls) were seen as having the capacity to act (to impact, to influence) as opposed to “matter”, which by historical rationalism and empiricism was seen only as a “passive receiver”. Also the material, the social and any combinations thereof,

were now seen as “acting” (impacting/influencing). Creation – according to Whitehead – results from the internal interactions of such mental processes of mind with the physical processes of the world. Hence, man is part of creation through “influencing and being influenced by the social-material world he interacts with” (Leclerc, 1990; Dibben, 2008; Olsen, 2011). This conceptual understanding of creation processes represents the analytical core of Whiteheadian process thinking (level 1 in the model above), which, for instance, is carried over to ANT (Latour, 1996) and which is also similar to the fundamentals of IMP thinking. One of the criticisms of Whitehead’s process philosophy regards his theory of time, which is said to exclude a conception of the human experience of continuity, and thereby of entities that we clearly perceive as unified (Felt, 1985). Continuity in theories of time has generally been associated with reflexivity: the human capacity to mentally integrate elements of both past and future in a moving present by means of immediate experience or reflection. This ability is fundamental to our understanding of wholeness, of self as an integrated unit and of life as a continuous, reflexive experience. This understanding of unity is essential in systems theory, and we may at least implicitly interpret system philosophy as one source of this critique of Whitehead’s physical conception of time as being simply a succession of epochal micro-moments of time, because it implies a conception of unity (wholeness) in his philosophy that is rather loose. Contrary to process philosophers’ focus on creation processes, systems thinkers turned their attention to complex unities (systems) that maintain their unity over time, and the relationships between such units and their environments.

4.1 Actor Network Theory (ANT) ANT is a process oriented methodological theory that has major roots in Whitehead’s philosophy (Harman, 2009). It combines the essence of this theory with later contributions - in particular from philosophy of language (semiotics) - to build a unified methodological apparatus (Figure 1: level 3) to study processes of becoming of new actualities in the human-socialmaterial world (Law and Hassard, 1999; Latour, 1993; Callon, 1986). Over the last three decades ANT has primarily prevailed within the area of science and technology studies, but has more recently also expanded into other fields, such as market creation studies (Callon, 1999), innovation studies (Akrich, Callon and Latour, 2002a,b, Hoholm, 2011) and, to some degree, also business network studies (Mattsson, 2003; Araujo, 2007). Somewhat differently from IMP, it is explicitly rooted in various traditions in philosophy – in particular due to contributions of the French philosopher Bruno Latour. As a result, ANT is quite explicit about the status of ANT as essentially a conceptual and methodological theory, rather than an empirically based theory about certain particularities of the world (Laursen and Olesen, 1996). It constitutes a paradigm (level 2) based on a conceptualisation of “creative interaction” (interacting “actor-network” of human-social-material content). On this basis it provides a methodology (level 3) said to be useful to conduct empirical studies where all insights about the empirical are said “to be handed over to empirical studies” (levels 4-7). Many ANT empirical studies have then gradually generated what we might regard as “more or less general ANT theories” at level 4 about various parts of the actual world – for instance in the area of “market making theory” as represented by Callon (1999), Olsen (2005) or innovation theory (Akrich, Callon and Latour, 2002a, b; Oudshoorn and Pinch, 2008; Garud and


Karnøe, 2001; Hoholm, 2011). The actor-network, as defined by ANT thinkers, is, a priori, a circulating, dynamic processing entity that incorporates a representation of internal formatting/structuring as well as of agency, which may “speak for” or otherwise represent the given actornetwork. The acting of this agency, however, cannot be separated from “the network”, and hence the notion of acting includes both the mental and the physical kinds of processes as elaborated by Whitehead. Contrary to Whitehead, however, ANT does not really maintain a clear distinction between the mental and the physical sides of such creative processes, a possible “over-simplification” – or loss of distinction - which has been a source of confusion to many. The structuring of an actor-network analytically starts from something very loose and unstable – such as a new business idea or business plan, gradually becoming more stable as a function of its inclusion-transformation (translation) of more entities into itself. ANT accordingly has a conception of “variable stabilities”, which is also closely associated with its conception of power as constituted by the linking of formatted elements with some persuasive capacity vis-á-vis others (Latour, 1991). The stabilised actor-network may then gradually form more persuasive, linked power structures, such as the capacity to persuade potentially ignorant or disinterested consumers to buy a company’s products. Power is the effect of this linked network of more or less persuasive elements that may develop into even more extended powers, or break apart and thereby lose its powers, say if the product of this particular business venture is later shown to be dangerous to the environment. This dynamic notion of power is core to ANT’s perception of variable relational dependencies and of variable predictability. On this basis, we may observe that there are some striking differences between the IMP and the ANT analytical concepts to describe and represent the networked subjects of their narratives. What constitutes interaction, and hence also relationships and interdependencies, are accordingly also somewhat different. One such difference is that in the IMP, actors are represented as one category out of several that make up the empirical content of a given network, whereas in ANT the network itself is to be identified with a particular agency. The actor-network has acting and/or impacting capacity associated with particular objectives. This has the consequence that the actor-networks that typically appear in ANT analysis are acting networks with particular purpose and meaning to persuade and include others - to impact the world around them, act upon it or expand within it. With such a network construct at its core, ANT has emerged to conceptualize processes and construct analytical models that exploit this acting/impacting analytical construct, for instance by analytically specifying how an actor-network may expand or decline in the world, how it may add to or lose its powers to influence others or how somebody on the outside of the network may be enrolled into it, such as a new investor into a start-up company. In all of these it is presupposed that the actor-network carries particular purpose that is represented in these processes, which, for instance, has to do with the particular interpretations and distributions of complementary roles within the actor-network. Relationships and interactions are accordingly also held together by the particular meaning-dimension of things (Callon & Latour 1981, Callon 1986, Latour, 1991, 1993, Olsen 2000).

4.2 Important differences between ANT and IMP As a result of these analytical differences, the networks typically

identified and analysed by IMP researchers would usually not be the same as those addressed by ANT – in one and the same empirical setting. The IMP networks would be those that are actually interacted in the real-economic, material and practical sense, which appears to be very similar to Whitehead’s understanding of “physical prehension” (interaction). It is about successive, reciprocal reaction processes. Creation is the processes of mutual interaction that are actually going on in between whatever is actually there. To IMP, the economy – or economic value - is the outcome of such interactions between heterogeneous entities – be they resources, activities, actors or ideas. ANT is rather about persuasion. To the ANT, the (actor-) networks perceived of would typically be those that stand out as purposeful entities that are in the process of expanding their impact, domination and extendedness in the world – in rivalry with other similar kinds of (actor-) networks. ANT is rather representing the interacting of “mental and physical prehensions” in Whitehead’s wording, where the mental dimension - at least implicitly- seems to imply a role for unity and intentionality on the side of the actor-network. The result of this is a very different approach to economic theorizing, where the ANT sees the economy as the outcome of formatting processes driven by particular actor-networks that are there to shape and to transform the economy (Callon, 1998). It is about the power of actor-networks to shape or influence the forms of interaction within an economic domain – which implies a representation of intentionality and unity of action. As a result, ANT theory of markets is not so much about the economy, as such, as it is about the shaping of the particular forms that define and shape how the economy works: the economy is a creation, and particular historical actornetworks are the creators of it. If we carry these differences over to the industrial marketing area of research, there seems to be no particular reason why the IMP should adopt a conceptual understanding of agency similar to that of the ANT in order to deal more effectively with the kind of issues I have raised in the introduction of this paper and in my discussion of challenges to the IMP. It seems to me these issues can just as well be addressed and analysed by using the already established ANT approach. One should rather recognize their distinctive ontological and epistemological differences and their associated different advantages for different research objectives and different kinds of analysis. A further development of an IMP conception of agency should, perhaps, rather focus on specifying how we can better make sense of the roles and functions of industrial network management – an interactional, networked conception of managerial agency? “Interacted agency”?

5. Systems philosophy and systems theories As stated in the introduction of this paper, the IMP industrial network approach emerged from a critique of traditional economic and marketing theory and its assumptions about resource homogeneity and independency of actors. It appears to have been substantially influenced and inspired by some of the early systems theories, such as the work of Kenneth Boulding and that of West Churchman, who can be seen as early contributors to the area of management sciences in the 1950s and 60s. Other influential sources were the works of Breyer (1934) and of Alderson (1965) associated with industrial systems and distribution systems thinking (Gadde 2000; 2011). However, we may also interpret the emerging IMP tradition as an implicit critique of the focus on unity and environment in systems thinking, with its necessary concern with borders, in-


ternal consistency and with some kind of coherent action seen as the joint, systemic outcome of coordinated actions within the “unit” labelled “system”. To the IMP, processes of interaction between heterogeneous entities are what constitute the economy, not some systemic order that can be seen as a separable unity as opposed to its environment. Hence, the IMP approach needed a more open analytical framework that permitted more process and less unity requirements. However, process and systems philosophies share the same fundamental interest in dynamic interactions. It is also quite clear that systems philosophy has been influenced by the earlier process philosophers. Rather than emerging out of mainstream philosophy, systems philosophy developed primarily in relation to the life sciences (Quastler, 1953), to cybernetic technologies (Ross Ashby, 1961) and to information technologies (Simon, 1964) where dynamic interaction in relation to unity – or wholeness – emerged as a core analytical issue. There are, accordingly, a wide range of systems theories across the sciences, of which only a few are addressed in this paper. One of their main contributors was the biologist, Ludwig von Bertalanffy, who published his “Algemeinen Systemlehre” in 1945 and, in English,“An Outline of General Systems Theory” in 1950. In his theory, he applied holistic concepts in the study of organisms in opposition to the purely analytical and reductionist paradigm dominant within biology at the time. He also suggested that his system-framework to understand complex processes could be applied to other domains, such as society and technology. The theory focuses particularly on the ordering and formal patterning of complexities, and theorises powers and capabilities of organised systems as contained in their richness and the internal hierarchical ordering of their interrelated parts. His concept of power is different from that of process theory, as it is primarily associated with the internal characteristics of some unit in terms of its complexity and hierarchical order. Other important contributors were Erwin Laszlo (1972), Joël de Rosney (1979) and Ilya Prigogine (1977)2, of whom at least Laszlo and Prigogine were substantially influenced by Whitehead’s process philosophy (Huchingson, 2011; Prigogine, 1980). In addition to this, a number of contributions came from cybernetics and the rapidly growing area of information and control technology – for instance as represented by Herbert Simon and John von Neumann, who also made important contributions to economic theory. Still, systems theory is dominantly rooted in science and technology rather than philosophy, where it is pragmatically applied in different versions. Within sociology, the German sociologist Niklas Luhmann (1984) has been a major contributor to systems theory, pulling from biology (Maturana & Varela, 1980), phenomenology, anthropology, and from mathematics and logics to build a sociological theory with particular emphasis on communications as the fundamental units in social systems. Indeed, it appears to be the case that General System Theory in the 1970s partly emerged as an attempt at unifying a quite diverse development of system theorising at the time – by reverting to a higher level of analytical abstraction. The associated unification program obviously contributed to the highly abstract and philosophically grounded analytical conceptions applied by the systems philosophers at the time. Both Whitehead and Laszlo departed from the same fundamental supposition that the principal entities of human reality are “natural-cognitive” entities. But whereas Whitehead, in2. Prigogine received the Nobel Prize in chemistry in 1977 for his work on thermodynamic systems.

fluenced by quantum physics, based his theory on the idea of self-organising processes of micro natural-cognitive occasions forming “societies of actual occasions/entities” through interaction, systems theory developed in relation to the idea of a living unit, made of complex internal interactions that at the same time interacted with a complex environment. Hence, to Whitehead, unity is merely a clustering of interactions that are outcomes of complex processes. These are also seen to interact with other entities in even more extended creative processes. Ontological priority is clearly given to process over unity. To system philosophers, the priority is the opposite: unity over process.

5.1 Core analytical concepts in System Theory Systems theory accordingly introduced a clearer concept of unity by drawing particular focus to what constitutes a distinction, a border or a separation between entities. By differentiating one system from another, the idea of dynamics is separated into two sides: the internal and the external. A system may also be internally composed of a number of other systems that can be further divided into other subsystems. It may, at the same time, be part of more extended systems that might again be seen as included in even more extended systems, etc. This conceptual understanding is an essential part of the core of systems thinking (Figure 1: level 1). Hence, to analyse a system necessitates a dual internal-external perspective at any level of analysis. This understanding of the ordering of the world in layers of more extended interactions was similar to that of Whitehead, but the introduction of distinct unities by systems theory obviously made this view a lot more operational and practical in use. Systems theory, accordingly, needed to deductively theorise what constitutes a dynamic system across all kinds of natural and artificial systems (Figure 1: level 2). Systems as distinctly separated entities required some theory of how autonomy, durability and change could be conceptualised at the ontological level. They did so by outlining certain characteristics, such as adaptive self-stabilisation, self-organisation, hierarchical ordering and non-summativity of systems. These defined the basic identity, stability and dynamic mechanisms of systems that permit us to talk about dynamic systems as such. It also constitutes the basis for developing the methodological tools needed for empirical analysis and theorising within the system – or “dynamic unity”–based approach (Figure 1: level 3). The introduction of a clearer conception of unity within an environment of other units provided a systematic approach to dealing with relations as potentially adaptive connections between units/systems that could be manipulated and influenced. Connectivity and relations were no longer just there, but were problematic, needed to be accounted for and could be acted upon to improve the functioning of the system. Hence, this ontological distinction between systems and process thinking is critical and points at the attractiveness of systems thinking, for instance, for some of those working with information technologies in relation to all kinds of areas of modern economies. Systems theory in this broad sense has a fairly dense, simplified core and a broad and pluralistic range of practical applications. Given this understanding of a relational – or interdependent character of systems that can be integrated into other systems or can be separable into internal subsystems, one may ask how the idea of “freedom” may apply to such systems. The answer from systems philosophers is that freedom primarily is a function of complexity and organisation. Entities of higher-order have hierarchical internal structures. They include rich and varied subsys-


tems of various types with complex connections between them. This organised complexity provides a greater capacity for spontaneity as well as for strategic flexibility in relation to changes in the environment. The greater the complexity, the greater will be the freedom to manoeuvre. Contrary to process philosophy and ANT, systems theory departs from an interdependency assumption in which freedom and variable dependencies have to be accounted for and explained in relation to empirical evidence. The complexity explanation of what constitutes freedom is part of a different conception of power than the constructivist version offered by ANT, namely that power-relations are associated with variable dependencies and flexibilities that favour the complex system over the simple. Hence, process and systems thinking represent different conceptual approaches to dealing with “freedom” and “variable dependencies” which are interesting and potentially applicable also to industrial marketing and management researchers. Systems theory is distinctly paradigmatic. Everything can in principle be looked upon and defined as a system or a subsystem, such as everything that is typically studied within the industrial marketing management domain. However, its concerns are rather different as it typically addresses the relationship between particular units (entities) and their environment in a broad and complex sense, whereas the IMP approach is to simplify “environment” into “relevant relationships to others”, in which case what remains of “the environment” is rather regarded as part of the background. Systems theory does not focus on the activities in between two or more systems, but turns its attention to the borderline between the inside and the outside of the given entity (system). In relation to such a phenomenon as industrial networks, it would either see the industrial network as a system and focus on the interactions between the entire network and its environments, or turns its focus to the various parts (sub-systems), such as represented by individual firms in the network, to analyse how they might interact with, manipulate or adopt to the other parts of the network. There is no analytical space outside or in between systems where interaction is conceptually possible, such as in IMP thinking. In this sense, the concepts of interaction, relationship and interdependency are also rather different from what is represented by the IMP (as well as the ANT). The idea of “a relationship” is particularly hard to grasp through systems thinking, as “a relationship” tends to expand into rather complex images of what constitutes “the other” as something clearly separated from all the other parts of the environment. Or it is being treated as something internal to the system, as relationships between its parts, where the logic of the relationship is to support the overall functioning of the system as a whole. Systems theory has this more holistic three-dimensional spatial analytical form that forces us to think of systems as internally coherent and functional, where network theories simplify by using one-dimensional geometric “node-line” models where what is outside of the nodes and the lines are excluded from attention and handed over to the background (context). However, it is also clear that systems theory has its distinct advantages also in the context of industrial marketing management research, for instance by offering complementary approaches to analyse the relationship between the internal structures and capacities of firms and the way they interact with, seek to dominate or change others (their environment). It may also be productively applied to analyse the role of hierarchies and complexities in business to business relationships with perspectives and analytical tools that are complementary to those of IMP, to deal more

explicitly with issues such as power, domination and relative freedom/flexibility.

6. Concluding comments The aim of this paper has been to initiate a discussion of the IMP theory by focussing its relationships to process theory and to systems theory. It has aimed at illuminating and contrasting epistemological and ontological underpinnings, methodological approaches and theories, while also throwing some light on their historical roots. As such, it has been an attempt at clarifying essential differences between these three traditions as well as some of their mutual influences, overlaps and similarities. This is also an occasion to underline the appreciation of pluralism in science. Triangulation of perspectives is a necessary part of the intellectual exercises of the disciplined creativity needed to develop theory about the world around us – and also to check for possible logical traps of the “fallacy of misplaced concreteness” kind. Where as process- and systems theories represent very broad areas of application across the sciences and can be traced also to very general metaphysical perspectives in the history of philosophy, IMP belongs to an applied area of academic research that we associate with business economics and management studies. However, what I have argued is the core of IMP thinking also reflects a close similarity with the fundamentals of process thinking in seeing interactions as the essential characteristic of the human experience of social-material creation that we associate with “the economy”. I have also argued that the IMP has received important inspirations from early systems thinking – in particular the early operations, logistics and management traditions of the 1950s, -60s and -70s. Dissatisfaction with systems theory concepts and constraints contributed to the emerging of more open, interactional conceptions within the early IMP tradition, concepts that I think reflect more kinship with process thinking than with systems thinking. However, it should also be noted that IMP is not a completely coherent area of theory, and that there are of course strings of more system oriented research represented. In analytical terms, the IMP can more precisely be characterized as rooted in a critique of main stream economics as well as main stream marketing and competitive market thought. In particular it argues that resources must be seen as heterogeneous as opposed to the homogeneity assumption in economics, and that knowledge should not be seen as given and distributed, but as relationally dependent. We may not have knowledge about something that we are not related to and have never somehow been related to. Hence, ignorance – not full information – must be the adequate analytical point of departure for understanding what economic development is, how it emerges and how it becomes more productive, more specialized, more interdependent, etc. Relating and interacting across heterogeneous resources, actors and activities are necessary for any learning and any economic activity to emerge. This is completely consistent, for instance, with Whitehead’s process philosophy of creation. The economic theory of IMP is not primarily about resource allocation, such as is main stream economics. It is rather about value creation and as such we may, at the very general level, see it as a representative of economic process thinking where interactions and interdependencies over time, across space and within and across minds are seen as fundamental constituents of economic value creation in society. The principle of interaction in IMP is reciprocal. The two


sides of the interaction are treated symmetrically, which analytically leads to a succession of actions, reactions, counter-reactions, etc. In contrast to this, I have argued that Actor-Network Theory (ANT), which is substantially influenced by process philosophy as well, is rather interpreting interaction as an act of persuasion and an attempt at transformation of “the other” to become included into the acting entity. Hence, ANT is about emergent phenomena that are somehow unified and semi-stabilized, represented by “spokespersons” or others that are there to expand their actor-networks, to convince, persuade or force different actor-networks to dissolve and/or to be absorbed into the expanding actor-network. The ANT kind of networks that are studied are represented by their meaning, their intentions, ambitions, etc. in ways that are quite similar to how economic and management theories see firms and organizations as acting subjects. Hence, their relationship to other actor-networks is primarily one of rivalry for domination and control, or inclusion and controlled transformation. The conclusion from this discussion is that it does not make much sense for the IMP to incorporate analytical conceptions into IMP analysis that seek to associate networks with some kind of unified meaning-based representation of it – such as attempted by the “mental network picture” approach. This approach contradicts the IMP concept of interaction – which is at the core of what the IMP really is about. It is better, then, just to use the ANT analytical apparatus for such analysis – if adequate to the research problem at hand. In contrast to IMP, the ANT has a rather clear analytical concept of agency, seen as a representation of the acting, intentional capacity – or impact or influence – of a given actor-network. The IMP certainly deals with actors of multiple kinds as empirical objects, but there is no clear and unified conception of how the kind of “interacted agency” of the reciprocal, interacting and at least partially interdependent “agency” can be analytically represented. Systems theory, on the other hand, is conceptually based on the identification of “unity” and on the analytical act of drawing a distinction between the inside and the outside. Similar to Process Philosophy, IMP and ANT, interaction is still fundamental to systems thinking, but the analytical framing is very different. Interaction is now something that occurs in the inside, on the outside and/or across the inside-outside distinction/border. “The inside” tends to be perceived of as a three dimensional space where whatever is there is assumed to play some role in producing a complex functional output to be delivered across to something at the outside. In this sense, the inside tends to have a kind of functional homogenous purpose where all the parts are, or should be, effectively coordinated. On the outside, there is the environment, which is complex, dynamic and demanding with respect to the functional requirements of “the inside”. Hence, the most critical interactions in systems analysis are typically those that occur across the inside-outside distinction. Despite the fact that IMP shares the overall holistic - and also to some degree the functional view of systems theory, the IMP seems to have rejected the implicit call for internal, functional consistency as well as the asymmetric relationship between the inside and the outside represented by systems theory. By shifting from conceptions that in geometrical terms are representations of three dimensional spaces (rooms) where whatever inside the room must be analytically represented and characterized, the IMP (as well as ANT) has moved to one-dimensional representations such as “nodes”, “lines” and “networks” of nodes and lines. This has the effect that one does not have to characterize what is not included in nodes and lines, whatever is in between

or near by, by only including what is actually in focus. Everything else is moved back into a complex context that is seen as important, but analytically, so mostly as a source for identifying things that can be brought into focus -- that should be represented as a relevant/important node in a network analytical model. The analytical simplicity and flexibility that results from this is a tremendous advantage to grounded empirical research of the kind that IMP and ANT are generally doing. However, there are of course also many advantages to applying system analysis when addressing research problems that call for analysis of complex functional phenomena, and quite often IMP research fields contain phenomena that are highly coordinated over substantial periods of time and across geographical space to a degree that one may productively see them as “systems” rather than “networks” and thereby apply more of the analytical apparatus of systems theory. For instance, I have pointed at the potential usefulness of system theories of power and flexibility (variable dependency) in relation to system internal hierarchies and complexities in analysis of networked power and interdependency. Can a system theory conceptual framing of these phenomena be exploited to advance IMP network theory based understanding of interacted agencies in business analysis? I may finally also conclude that the IMP is distinguished from both ANT and Systems Theory in that IMP insists on maintaining an a priory symmetrical and reciprocal conception of interaction, where the two others have given priority to representing a unification of an acting unit in relation to an external environment of “others”. The IMP approach could perhaps be seen as a more radical opposition to the idea of basing analysis on units that can be perceived of as independent. This is a unique strength of the IMP, but also its challenge: to go further to clarify its analytical constructs and to make them more applicable to different kinds of analysis within the broader area of economic theorizing. My hope is that this limited attempt at providing some clarifications and background thinking may be of some use to such efforts. However, it is also clear that, in the broad approach to this discussion that I have taken here, there are numerous topics to which I have not been able to give scope for the significance they probably represent. Those will be left for later work – or for others to explore and argue.

References Akrich, M., M. Callon and B. Latour (2002a): “The Keys to Success in Innovation, Part 1: The Art of Interessement”, International Journal of Innovation Management, vol. 6(2) Akrich, M, M. Callon and B. Latour (2002b): “The Keys to Success in Innovation, Part 1: The Art of Choosing Good Spokespersons”, International Journal of Innovation Management, vol. 6(2) Alderson, W. (1965): Dynamic Marketing Behaviour and Executive Action. Homewood, ILL. Richard D. Irwin Araujo, L. (1998): “Knowing and Learning in Networks”, Management Learning, vol. 29(3) Araujo, L. (2007): “Markets, Market Making and Marketing”, Marketing Theory, vol. 7(3) Axelsson, B. (2011): “Business Relationships and Networks: Reflections on the IMP Tradition”, The IMP Journal, Vol. 4(1): 3-30 Axelsson, B. and G. Easton (eds.) (1992): Industrial Networks: A New View of Reality, London: Routledge Barile, S. and F. Polese (2010): “Smart Service Systems and Viable Service Systems. Applying Systems Theory to Service


Science”, Service Science, vol. 2: pp 21-40 Bergson, H. (1983): Creative Evolution, Lanham, Md: University Press of America Bergson, H. (1960): Time and Free Will, New York, Harper and Row von Bertalanffy, L. (1945): Zu einer allgemeinen Systemlehre, Blätter für Deutsche Philosophie, 3(4) von Bertalanffy, L. (1950): An Outline of General System Theory, British Journal for the Philosophy of Science 1, p. 139-164 von Bertalanffy, L. (1951): General system theory - A new approach to unity of science (Symposium), Human Biology, Dec 1951, Vol. 23, p. 303-361. von Bertalanffy, L. (1968): General Systems Theory: Foundations, Development, Applications, New York: George Braziller Boulding, K. (1956): “General Systems Theory. The Skeleton of Science”, Management Science, 2, 3 (Apr. 1956) pp.197-208 Breyer, R. (1934): The Marketing Institution. New York: McGraw-Hill Bromley, D. (1990): “The Ideology of Efficiency: Searching for a Theory of Political Analysis, Journal of Environmental Economics and Management, vol. 19(1), pp 86-107 Callon, M. and B. Latour (1981): “Unscrewing the big leviathan: how actors macro-structure reality and how sociologists help them do so”, in K. Knorr-Certina & A. Cicourel (eds): Advances in Social theory and methodology towards an Integration of Micro and macro Sociologies. Boston: Routledge and Kegan Poul, pp. 277-303 Callon, M. (1986): “Some elements of a sociology of translation: Domestication of the scallops and the fishermen of St. Brieux Bay”, in John Law (ed.) Power, action and belief, Keele: Sociological Review Monograph No. 32, pp 196-229 Callon, M. (1998): The Laws of the Markets, Blackwell Publishers/The Sociological Review, Oxford UK and Malden, MA, USA Callon, M. (1999): “Actor-Network Theory – The Market Test”, in J. Law and J. Hassard (eds.) Actor Network Theory and After, Blackwell Publishers/Sociological Review, Oxford UK and Malden, MA, USA Churchman, C.W. (1955): “Management Science, the Journal. Management Science 1(2), 187-188 Churchman, C.W. (1994): “Management Science: Science of Managing and Managing of Science, Interfaces, 24(4), 99-110 Dibben, M. (2008): “Organizations and Organizing. Understanding and Applying Whitehead’s Processual Account”, Philosophy of Management, vol. 7(3) Dreyfus, H. and S. Dreyfus (1988): “Making a Mind vs. Modeling the Brain: AI back at a Branchpoint,” The Artificial Intelligence Debate, M.I.T. Press Dreyfus, H. and S. Dreyfus (1986): Mind Over Machine: The Power of Human Intuition and Expertise in the Era of the Computer, New York: Free Press Dubois, A. and L. Araujo (2004): “Research methods in industrial marketing studies”, in Håkansson, H. D. Harrison and A. Waluszewski (eds.): Rethinking Marketing: Developing a New Understanding of Markets, Wiley Felt, J.W. (1985): “Whitehead’s Misconception of “Substance” in Aristotle”, Process Studies, 14(4): pp 224-236 Felt, J. W. (2002): “Epochal Time and the Continuity of Experience”, The Review of Metaphysics, vol. 56: pp 19-36 Ford, D. (1980): “The development of buyer-seller relationships in industrial markets. European Journal of Marketing, 14(5/6), pp 339-53 Ford, D. (1990): Understanding Business Markets. San Diego:

Academic Press. Ford, D., L-E- Gadde, H. Håkansson, I. Snehota and A. Waluszewski (2011): ”Analyzing Business Interaction”, The IMP Journal, Vol. 4(1): pp 82- 105 Gadde, L-E. (2000): “From marketing channels to differentiated networks – distribution dynamics in historical perspective. In: Dahiya, S. (ed.) The Current State of Business Disciplines, 2641 – 2662. Rohtak: Spellbound Publications Gadde, L-E. (2011): “Distribution network evolution – challenges for practice and theory, The IMP Journal, vol. 4(3): pp 160 - 169 Gadde, L-E. and Ford D. (2008): Distribution research and the industrial network approach, IMP Journal, 2 (3): 36-52 Garud, R. and P. Karnøe (eds.) (2001): Path Dependence and Creation, Lawrence Erlbaum Associates Gilson, Etienne (1999): The Unity of Philosophical Experience, San Francisco: Ignatius Press Harman, G. (2009): Prince of Networks. Bruno Latour and Metaphysics, Re.press , Melbourne Henneberg, S.C., S. Mouzas and P. Naude (2006): “network pictures: concepts and representations. European Journal of marketing, 40(3/4), pp 408-29 Hernes, T., T. Bakken and P. I. Olsen (2006): “Spaces as Process. Developing a Recursive Perspective of Organizational Space”, in S. Clegg and M. Kornberger (eds.): Space, Organization and Management Theory, Liber & Copenhagen Business School Press Hoholm, T. (2011): The Contrary Forces of Innovation: An Ethnography of Innovation Processes in the Food Industry, London: Palgrave MacMillan Huchington, J. E. (2011): “Organization and Process: Systems Philosophy and Whiteheadian Metaphysics”, research paper at www.religion-online.org/showarticle.asp?title Hulthén, K. and L-G. Mattson (2011): “Distribution Networks Dynamics: Evolution in the PC Distribution Network, The IMP Journal, Vol. 4(3): pp 170-193 Håkansson, H. (ed.) (1982): International marketing and purchasing of industrial goods: An interaction approach, New York: John Wiley & Sons Håkansson, H., D. Ford, L-E. Gadde, I. Snehota & A. Waluszewski (2009): Business in Networks, Chichester: John Wiley Håkansson, H. and J. Johanson (1992): “A Model of Industrial Networks”, in Axelsson, B. & Easton, G. (Eds.): Industrial Networks: A New View of Reality, London, Routledge, 28-34 Håkansson, H. and I. Snehota (1995): Developing Relationships in Business Networks. London: International Thomson. Håkansson, H. and A. Waluszewski (eds.) (2007): Knowledge and Innovation in Business and Industry: The Importance of Using Others, London: Routledge Johanson, J. and L-G. Mattsson (1985): “Marketing, investments and marketing investments in industrial networks. International Journal of Research in Marketing, 2(3), pp 185-95 Johanson, J. and L-G. Mattsson (1986): “Interorganizational relations in industrial systems: a network approach compared with a transaction cost approach”. International Studies of Management Organisation, 1781), pp . 34-48 Lakatos, I. (1977): The Methodology of Scientific Research Programmes: Philosophical Papers Volume 1. Cambridge: Cambridge University Press Laszlo, E. (1972): Introduction to Systems Philosophy, New York: Gordon and Breach Latour, B. (1991): “Technology is Society Made Durable”, in J. Law (ed.) A Sociology of Monsters: Essays on Power, Tech-


nology and Domination, London/New York. Routledge Latour, B. (1993): We Have Never Been Modern”, Cambridge, Massachusetts, Harvard University Press Latour, B. (1996): “Do Scientific Objects Have a History?”, Common Knolwedge, 5(1): 76-91 Laursen, H. H. and F. Olesen (1996): “Interview med Bruno Latour”, Philosophia, vol.1: pp 76-91 Law, J. And J. Hassard (eds.) (1999): Actor Network Theory and After, Blackwell Publishers/Sociological Review, Oxford, UK Leclerc, I. (1990): “Whitehead and the Dichotomy of Rationalism and Empiricism”, in Rapp, F. And R. Wiehl (eds): Whitehead’s Metaphysics of Creativity, State University of New York Press Leek, S., P. Turnball and P. Naude (2003): “Interactions, relationships and Networks: Past, Present and Future”, paper presented at IMP Group Conference, www.impgroup.org Luhmann, N. (1984): Soziale Systeme: Grundriß einer allgemeinen Theorie, Frankfurt: Suhrkamp. (English translation: Social Systems, Stanford: Stanford University Press, 1995) Mattsson, L-G. (2003): “Understanding market dynamics – Potential contributions to market(ing) studies from ActorNetwork Theory”, paper presented at The IMP Group Conference, www.impgroup.org Maturana, H.R. and F.J. Varela (1980): Autopoiesis and Cognition. The Realization of the Living, Dordrecht, Reidel Medlin, C. (2002): “Interaction: A Time Perspective”, paper presented at The IMP Group Conference, www.impgroup.org Olsen, P. I. (2000): Transforming Electricity. The Case of the Norwegian Electricity Market Reform, Norwegian School of Management, Series of Dissertations, no 1 Olsen, P. I. (2005): “The formatting of electricity, and the making of a market”, in B. Czarniawska & T. Hernes (eds) ActorNetwork Theory and Organizing, Liber & Copenhagen Business School Press Olsen, P. I. (2011): “The Relevancy and Applicability of Process Metaphysics to Organizational Research”, Philosophy of Management, vol. 10(2), pp 53:74

Oudshoorn, N. and T. Pinch (2008): “User-Technology Relationships: Some Recent Developments”, in The Handbook of Science and Technology Studies”, MIT Press Prigogine, I.; Nicolis, G. (1977). Self-Organization in Non-Equilibrium Systems, Wiley. Prigogine, I. (1980). From Being To Becoming, Freeman. Quastler, H. (1953): Information Theory in Biology, Urbana, Illinois, University of Illinois Press Rescher, N. (1996): Process Metaphysics: An Introduction to Process Philosophy, Albany, New York: SUNY Press de Rosney, J. (1979): The Macroscope: A New World Scientific System, New York: Harper and Row Ross Ashby, W. (1961): An Introduction to Cybernetics, New York: John Wiley and Sons Schumpeter, J.A. (1939): Business Cycles. A Theoretical, Historical and Statistical Analysis of the Capitalist Process. Now York/Toronto/London: McGraw-Hill Book Company Simon, H. A. (1965): “The Architecture of Complexity”, General Systems, vol. 10, pp 63-76 Sphorer, J., S. L. Vargo, N. Caswell and P. P. Maglio (2008): “The Service System is the basic Abstraction of Service Science”, procedings to the 41th Hawaii International Conference on Service Science Sphorer, J., P.P. Maglio, J. Bailey and D. Guhl (2007): “Towards a Science of Service Systems”, Computer, January Vargo, S.L. and R.F. Lusch (2004): Evolving a new Dominant Logic of Marketing, Journal of Marketing, vol. 68: pp 1-17 Waluszewski, A. (2004): “A competing or co-operating cluster of seven decades of combinatory resources? What’s behind a prospering biotech valley?” Scandinavian Journal of Management, vol. 20: pp 125-150 Whitehead, A. N. (1978): Process and Reality, New York: The Free Press

Per Ingvar Olsen, Associate Professor, Department of Innovation and Ecomomic Organisation, Nydalsveien 37, 0484 Oslo, BI Norwegian Business School, Oslo, Email: [email protected]


Approaching (Inter-)Actors in the Business Landscape Antonella La Rocca

University of Lugano – USI

Abstract

Diffused interdependences and the centrality of interaction processes evidenced in research on business markets imply the need to re-examine some of the central concepts in marketing and to reflect on how research can address them. This particularly applies to the concept of actor that from an interaction perspective appears rather different from how it is generally intended in management and marketing. Two research traditions in particular − symbolic interactionism and the actor-network theory − can be related to the discussion of the approach to actors from the IMP perspective. In this paper, we first examine the traits that characterize actors in business relationships, in particular the variable entity of actors, their relational specificity, and their emergent nature. We then discuss the implications of these traits regarding how actors can be approached in empirical research. Key words: business actors, interaction, business relationships, research methodology

1. Introduction Over the last few decades, a considerable body of research has focused on customer-supplier relationships in business markets and has been concerned with explaining the existence and the consequences of the relational processes for business markets dynamics and for the businesses that operate in such contexts (for an overview, see for instance Håkansson, Ford, Gadde, Snehota & Waluszewski, 2009). Much of the research in this stream investigated how business relationships emerge between companies and examined the consequences of business relationships for the businesses involved and for the dynamics of business markets when they take on the form of business networks. Studies in this research stream have the common trait of considering relational processes, in particular the interaction in customer supplier relationships, as the central process in marketing. Focus on customer-supplier relationships has grown increasingly popular, providing a different perspective on several issues central to theorizing about marketing (Achrol & Kotler, 2011; Johanson & Vahlne, 2011; Waluszewski, Hadjikhani, & Baraldi, 2009). Taking the relational rather than the transactional perspective implies the need to revisit some of the central concepts in marketing in general (Morgan & Hunt, 1994) and in business marketing in particular (Ford, Gadde, Håkansson, Snehota & Waluszewski, 2010). Among these is the concept of the actor. The notion of the actor has been increasingly used in management context and, more recently, in the marketing literature. Most of the research in management and marketing actually takes the actor perspective and starts from a given actor, but discussion about what and who is an actor is practically absent; it is somehow considered to be obvious. Research in marketing is certainly a case in question; it assumes more or less explicitly that, in marketing, actors are given entities − individuals and organizations. This contrasts with the findings suggesting that, in business relationships, actors can be rather variable entities (e.g. Håkans-

son & Snehota, 1995). In the initial IMP research, the actor dimension of business relationships has been approached with a certain ambiguity, and the concept of actor proposed remained ambivalent; it has been used to designate the supplier and the customer organizations as actors in business relationships, but at the same time, the individuals who act and interact between the two businesses have been considered actors as well (Håkansson & Snehota, 1995). While the concept of actor has been problematic, it has never been much problematized, and only recently have studies focusing on the interaction processes in business relationships started to examine the actor dimension in business relationships more systematically from an interaction perspective and pointed out that the concept of actor is not a given-forgranted entity (Håkansson et al., 2009; La Rocca, 2011). There are at least two other research traditions that dealt with the concept of actors from an interaction perspective: symbolic interactionism and the so-called Actor-Network Theory (ANT) both converge on the idea of actor as anything but given. Rather, both of these research traditions suggest that actors in interaction are an emergent and multiform entity and appear to point out several features of actors that also emerged in the IMP research on the interaction in business relationships. Defined in relation to interaction in business relationships, actors appear as a variable, multiform, and emergent entity not coinciding with the formal organizations of the customer or supplier. The aim of this paper is to review the traits characterizing actors from an interaction perspective and to examine the methodological challenges in approaching actors in the business landscape in empirical research. In particular, we will discuss the implications of the actor concept for choosing the unit of analysis in approaching actors in business relationships and for capturing the way actors emerge and vary with the context. In the next part of the paper, we will briefly review how the concept of actor has been approached in the research traditions of symbolic interactionism (e.g. Blumer, 1969) and ANT (Callon, 1998; Latour, 1987). In the following section, we will revisit the distinctive traits of actors in a business relationship as they


emerge in the IMP research on business markets (e.g. Håkansson et al., 2009). In section four, methodological challenges of exploring actors from an interaction perspective are discussed with regard to methodological dilemmas related to empirical research on the entity of actors in business relationships that I treated as perceived identity in my research. Section five contains some general reflections on methodological issues in researching the interactive business landscape.

2. The Idea of Actor in Interaction

Despite the growing use of the concept of actor in management

and, more recently, in the marketing literature, discussion on the idea of what and who is an actor is largely absent. The entity of an actor is not an issue that is discussed; it is somehow taken for granted. Most of the research in management and marketing actually assumes the actor perspective and starts from a given actor, which tends to be either the firm or the individual manager. Such an approach is linked to the scope of conceptualizing and theorizing in management, which is to understand how individuals and companies as actors cope with various circumstances and situations, and the main purpose is to provide actionable results and to arrive at some kind of normative implications for how actors should cope with the various situations they face. Two assumptions characterize the traditional approach to actors in a management context: First, it does not need to be discussed what an individual or an organization is because it is obvious. Discussions regarding where an organization starts and ends, or what an individual is, are rare in management research. A related second taken-for-granted assumption is that the actor is an entity that is characterized (if not defined) by being capable of intent and purposeful behaviour. This means that an actor is an actor because she is capable of acting on purpose and the intents that, in a management context, are derived from an assessment of the situation context. Not only in management but also in most of the social sciences and in economic theory, individuals and organizations are both traditionally considered as actors because they are assumed to be capable of rational and purposeful behaviour. Such an assumption persists even if there has been a rather long debate whether such an assumption holds for organizations. It has been vigorously questioned (Cyert & March, 1963; Mintzberg, 1973), and the idea of bounded rationality (Simon, 1957) has been acknowledged and made a significant inroad in management studies in general and organisational research in particular. Although it admits the bounds on calculative rationality, management research generally takes the actor perspective and aims at providing guidance for how to enhance the capacity of the actor, be it individual or organization, to cope with various situations and contexts. The question that inspires much of the research in management is normative: how can one improve the performance of the company or of the individual manager? The rationale for approaching actors from the interaction perspective is substantially different from such a normative approach. The turn to the relational perspective is consequent to addressing phenomena that can be readily observed but are not explained in the existing theory; in particular, when collective phenomena and outcomes cannot be explained from single actors’ intentionality. The question has been discussed as a counter position of micro-motives and macro-behaviours (Schelling, 1978). The main interest and aim of research along these lines is not the performance of the actor; rather it is to explain how

certain relationships develop and how, as a consequence, various social structures emerge. Research in this direction is focused on investigating the role of actors in organizing and forming social structures and is not primarily concerned with the success and performance of the individual actors. Two research traditions in particular have dealt extensively with actors from interaction perspective. The first is the symbolic interactionism in sociology (e.g. Blumer, 1969; Goffman, 1974), and the second is the ANT approach in science studies (e.g. Callon, 1998; Latour, 1987; Law & Hassard, 1999) Both of these have been proliferating in various directions and are notoriously difficult to summarize, but they share several insights with respect to the role and concept of actors and their traits when approached from an interaction perspective. Approaching the concept of the actor from the interaction perspective opens the issue of the entity of an actor and requires rethinking of the concept of actor. It presents various challenges. Goffman (1967), an important scholar of interactions, observed that, to consider actors from the interaction perspective is to consider “moments and their men” rather than “men and their moments” (p. 3). That sheds a different light on the entity of an actor. It implies that the entity of an actor is “in the eye of the beholder” and that, in interaction, an actor is an entity that is producing an effect on the counterpart. Since routine or ritual behaviours can have effects on others, generate interactions, and produce tangible outcomes for others and the actor as well, an entity can be conceived as an actor without being capable of rational and purposeful behaviour. Studies of interactive relationships have shown that relationships develop as a consequence of how interaction between two parties unfolds, regardless of the knowledge, rationality, and motivations of the interacting parties (Thibaut & Kelley, 1959). The ANT in particular goes as far as to emphasize that material objects are part of social networks and can be seen as “actants” that produce important effects in formation of relationships (Latour, 1987) and can only be distinguished from actors as the latter are “actants endowed with character” (Akrich & Latour, 1992, p. 259). The conception of agency for the interactionists and ANT proponents does not presuppose intentionality and assumes that actors make the difference as they mediate and have some discretion and their outputs cannot be predicted by their inputs. As any action undertaken by an actor is always embedded in the flow of other actions going on among other related actors (Abott, 1995; Emirbayer & Mische, 1998), the actual discretion of actors (degrees of freedom in selecting a course of action) is actually rather limited. Both the symbolic interactionists and the ANT proponents argue strongly that creating meaning, or sense-making in a related tradition (e.g. Weick, 1995), plays an important role in the formation of relationships. ANT in particular stresses also that the clusters of actors involved in creating meaning are both material and semiotic. The entity and identity of the actor in interaction is then defined by what the actor represents for specific others in a given relationship and in a given situation. What actors represent in a relationship is defined by the expectations others have about the role of the actor. When identity is ascribed to an agent (actor) by the other actor, it always involves what the counterpart represents in relation to what the other actor is and what the actor is concerned about. In this regard, the interaction perspective is related to role theory; indeed, the two concepts are so intertwined that “to use the term role is necessarily to refer to interaction” (Stryker & Statham, 1985, p. 323). Roles are relational concepts and, moreover, the purpose and meaning of a role depend on the


network of complementary relationships and roles in which an actor is embedded (Biddle, 1979; Katz & Kahn, 1978). Studies of interaction between roles in formal organizations have led to question whether from the interaction perspective the distinction between individual vs. organizational actors is meaningful. It has been argued that, in interaction, individuals represent (and tend to be interpreted as representing) organized entities. It has been argued that “when we look at individual behavior in organizations, we are actually seeing two entities: the individual as himself and the individual as representative of his collectivity… the individual not only acts on behalf of the organization in the usual agency sense, but it also acts, more subtly, ‘as the organization’…. as a result, individual behavior is more ‘macro’ than we usually recognize” (Chatman, Bell & Staw, 1986, p. 211). Similar themes to those discussed in these two research traditions have also emerged in relation to the idea of actor in buyerseller relationship in business markets.

3. Actors in business relationships Research on business markets in the IMP research tradition has reached the conclusion that “a basic feature of the business landscape is the intricate interdependencies between the companies that populate it” (Håkansson et al., 2009, p.1). Interdependences have important implications for the conceptualization and analysis of the formation and development of business networks, play an important role in the analysis of business relationships, and have implications also for how to approach the concept of actor. Acknowledging the diffused interdependences and development of close customer supplier relationships as important phenomena in business markets raises the question of why and how business relationships arise and develop. That in turn leads to attention for the role of actors in developing business relationships and creating interdependences. Existing marketing theory offers little, if any, help in explaining why business relationships emerge in business markets in first place. The metaphor of a rainforest has been proposed to capture the nature and significance of interdependences in the landscape in which business activity takes place. The rainforest metaphor can be juxtaposed onto the more traditional and common metaphor of the (competitive) jungle that has been and still is commonly used in most of the disciplines that deal with business behaviours in a market (Håkansson et al., 2009). The idea of the jungle evokes a market as consisting of autonomous actors who fight each other (compete) and only the best survives and becomes the winner. It also evokes the idea that survival of a business depends on its ability to identify competitors and to create and defend its competitive advantage in relation to these (Marglin, 2008). Such an idea of the business world led researchers to focus on antagonistic competitive behaviours (Waluszewski & Håkansson, 2006). The rainforest metaphor, unlike that of the jungle, evokes as typical characteristics of the business landscape variety, motion, and relatedness. It suggests that companies (and actors in general) are not only rivals but are involved in many different forms of interactions such as, for instance, cooperation. The rainforest is not defined by its population, but by the interactions between those that populate it, and rivalry is not the only, or the most important, way to relate to each other; while companies can fight and compete, there are always other forms of relations such as complementary or cooperative relations. This is characterized by interdependencies among businesses that cannot be avoided, since “each is vital to and dependent

on others that it borders and overlaps” (Håkansson et al., 2009, p. 6). Taking such a perspective, the market does not consist of isolated and solitary entities but of mutually dependent entities and “tangible and intangible resources of many types, stemming from many different organizational units, are related, confronted, and adapted in ways which are beneficial for those involved in the doing of business” (Håkansson et al., 2009, p. 7). Focusing the attention on interactions in general rather than on one particular type of interaction, namely the antagonistic one, allows for consideration of the process of both cooperation and conflict and suggests that those that prosper in the rainforest metaphor are those who are most needed by others and those that make the best “use” of connections with others. Against this background of diffused interdependences, the concept and role of actors emerges with several distinct features that converge with those proposed regarding actors in research inspired by ANT and symbolic interactionism. This convergence reflects the common angle from which the actors are approached; as for ANT and interactionism, the primary interest is in explaining the emergence of social structures − in the IMP case, the emergence of business networks. The rainforest metaphor highlights as central the role of actors in mutual relating in the business landscape. In the IMP research on customer supplier relationships in the business market, the actor dimension has been treated less systematically, and early approaches to the actor concept have been somewhat ambiguous, conceding that actors in business networks are individuals and firms are formal organizations (Håkansson & Snehota, 1995). However, more recently, the need to approach actors in business relationships from an interaction perspective and to rethink the idea of actors in business relationships has been emphasized (Håkansson et al., 2009). It has been acknowledged early in the IMP research that agency in business relationships rather than defined in terms of intentionality and purposeful behaviour is related to the consequences for the counterpart in the relationships (e.g. Håkansson & Snehota, 1995). The interaction-based perspective espoused by the IMP is one that sees an actor’s identity as acquired in interaction with others “because its behaviour, regardless of the motives or underlying reasons for it, is a matter of concern to or affects another” (Håkansson et al., 2009, p. 138). Agency is given by discretionality in the sense that there is no direct link to output from a given input for an actor. The actual freedom in selecting a course of action is very limited, in particular because of the importance of the material component in business relationships. Indeed, the resource and activity dimensions have a “heaviness” that confines the discretion of the actors in defining workable solutions in business relationships. Interdependences in the business landscape, in particular in the material resource dimension, limit not only the discretion of single specific actors but also their control over the outcomes of the own actions. However, even when the discretion of actors is limited, various degrees of undeterminedness remain for actors in business relationships. While interdependences limit the autonomy of the single actors, it is at the same time interdependences that make the actors important, as business actors connect and interact selectively. The economic and financial outcomes for a business depend on whom a business is connected to, because business relationships permit a business to access and make use of other resources, activities, and actors. The material component in business relationships is highly significant. Indeed, individual actors in business relationships appear to represent an “organized entity” which counterparts can use. Counterparts are concerned with resources and capabilities


that can be mobilized and accessed through the actor. Such an organized entity is more than the individual per se but also different from the formal organization for which the individual acts as an agent. That makes the set of resources and others that can be accessed through an actor in interaction part of the entity of an actor in interaction. Thus, the idea of relational identity surfaces also in business marketing research. From an interaction perspective, actors represent organized entities that include features of the actors but also its associations with others. The entity of the actor is not given because the organized entity attributed to the actor is defined by the needs and expectations of the counterpart, who actually defines its content and boundaries of the actor. Therefore, the entity that an actor represents is not simply given by the properties of the actor as an individual or of the formal organization that the actor represents. The entity of an actor is always relationship and context specific; others define it in relation to their expectations and situational conditions, and that makes the boundaries of the actor’s entity fuzzy and continuously re-enacted. Considering the entity of an actor in business relationships, it becomes evident then that it reflects how those who interact with the actor see things and make sense of the interaction situations rather than simply reflecting its features. There is a link to semiotics and interpretations emphasized by both the interactionists and the proponents of the ANT. In sum, the concept of actor in interactive business relationships converges with the one that comes to the forefront in symbolic interactionism and in the ANT, and compared to how the actor concept is generally used in management and marketing, it appears to differ in at least three respects. Firstly, actors are not a given entity, such as individuals or formal organizations. Rather, the entity of an actor is a variable set of material and immaterial resources defined as the set of features that matter for the counterpart in a certain context. The entity of the actor corresponds to the identity attributed to the actor by others, which tends to reflect a role that is broader than the individual characteristics but also different from the formal organization the individuals represent. Secondly, the entity of actor in business relationships is always relationship specific, and as a consequence, it varies with the relationships in which an individual or organizations are involved. The perceived identity of an actor varies thus across relationships, and because of multiple relational involvement, actors tend to represent more or less simultaneously different entities (identities) for different counterparts and these multiple identities can interfere with each other. Thirdly, compared to the received concept of actors in management and marketing, the entity of actor is not given and static; rather it is continuously re-enacted and keeps changing over time as the context and counterparts conditions and expectations change. The concept of actor from an interaction perspective is thus continuously emergent. These features of actors from an interaction perspective have some implications for how empirical research on actors in business relationships should be carried out.

4. Methodological challenges of researching actors in an interactive landscape Because it is so different from the received view on actors in management, the concept of actor from an interaction perspective poses a number of severe challenges when we want to investigate the actor dimension of business relationships empirically. While some issues to be addressed have been identified, others still keep emerging. In light of the emergent feature of

actors in business relationships, we can now examine what research issues, methods, and tools are involved in capturing and explaining these features and analysing their implications for the businesses involved. Coping with the variable nature, the relationship specificity and transient nature of the actors from an interaction perspective implies addressing methodological issues. These issues will be discussed as three methodological dilemmas that I have faced in my research on how actors mutually perceive each other (attribute each other identities) when they interact in business relationships and how that affects their behaviors in interaction and thus the development of business relationships (La Rocca, 2011). The aim of that study was to show the interactive nature of actors and the multifacetedness and emergent nature of actors in business relationships. Approaching actors from the perspective of interaction in business relationships requires in particular consideration of the interdependences that distinguish the business landscape, and overall, the experience of approaching actors from such a perspective is so far rather limited, while the need to explore the implications of such an actor concept is urgent.

4.1 Taking an outside-in (bilateral) view on actors Considering actors from the interaction perspective, their entity (perceived identity) can only be identified and defined in relation to a certain action context in which they appear. Choosing interaction in business relationships as the context of acting means that actors are identified as those who are perceived as actors by interacting counterparts. Usually it is related to roles that individuals have, as these in various capacities represent the customer or supplier in the specific relationship. As actors are defined by the perceptions and interpretations of those they relate to, researching their role in the development of inter-organizational relationships (and thus in economic organizing) must take the counterparts’ perspective and cannot be limited to self-reported accounts of goals and actions. This implies that, when we are set to investigate actors, the unit of analysis must be relationships and must thus include other actors who actually define the actor. Due attention needs to be paid to how parties mutually interpret their actions and reactions. In my study of interacting actors in 32 customer supplier relationships, I have chosen to frame the issues to investigate how actors attribute identity to others with whom they interact (La Rocca, 2011). This means that we have to abandon the traditional assumption that actors can be studied and their behaviours explained, by investigating their intent and assuming economic rationality. Since the identity of an actor cannot be derived from what the actor says about himself but derives from how others that interact with the actor perceive and imagine an actor’s identity, we have to take an outside-in view on the actor. Looking at an actor’s identity in business relationships from an interaction perspective calls for a bilateral approach. An actor’s identity must be explored in relation to the counterpart the actor is interacting with in a non-cumulative way. The issue here is that an actor has a distinct identity in a given relationship. That makes it impossible, or at least dubious and contradictory, to aggregate the various relational identities and deal with a sum of what the counterparts believe the actor is. Empirical study of mutual perceptions in interaction is anything but simple. It is a well-documented fact that, in a common customer supplier relationship in the business market, numerous individuals tend to be involved in the interaction in both organizations. That, of course, begs the question of who can


be considered a representative “informant”. In my case, I have opted to observe and survey sale and purchasing representatives. Such a choice is a convenience approach and in more complex relationships may distort the data collected. Again, it is common knowledge in marketing that perceptions and interpretations of the offering and of the supplier identities can vary considerably, depending on the roles in the purchasing team, and similar consideration cannot be excluded for the selling team. It remains an open issue how to aggregate these. The differences and incommensurability are a substantial fact and aspects of the problem, not just a nuisance. Furthermore, focusing on actors in the interaction process, there is the question of data collection method. Retrospective interviews with those involved are commonly used but have several limits, and the direct observation of how certain roles interact can be problematic. I have adopted a research design that had elements of ethnographic research based on observation as well as interviews and have chosen to combine bilateral interviews and the observations of the customer-supplier face-to-face encounters. That reflects the choice to consider the formal roles as best representing either party and having a key role in relationship development and to consider face-to-face interaction as the dominant interaction mode (disregarding more mediated interactions; e.g. phone calls, e-mails). Investigating the entity and role of actors in business relationships and developing the concept of “interactive actor” is likely to have various elements of “systematic combining” (Dubois & Gadde, 2002) and involves juxtaposition of conceptual frameworks and empirical evidence. Notwithstanding some claims to the contrary, the empirical observations are never “free of theory” that directs the attention of the research to certain circumstances rather than others. Conceptualizations and constructs related to the idea of the actor as an “inter-actor” available in current research are limited. On the question of the “relational identity” of actors in a business relationship, the extant literature offers little guidance on how to examine the role of actors representing the supplier and customer organizations and on which issues to focus. Review of the literature yielded only one study that explicitly addressed the issue of “relational identity” proposing to distinguish two components of relational identity: one that is role-based, drawing on the collective level and focusing on prototypical role occupants, and one that is person-based, reflecting the qualities of the role occupant (Sluss & Ashforth, 2007). I found this idea useful and compatible with the idea in symbolic interactionism, ANT, and IMP that an actor always is an “organized entity” acting on behalf of an organization. This idea can be translated into two aspects that can fit with customer/ supplier mutual perceptions: 1) an actor’s quality (related to role expectations) and 2) and actor’s personality (person’s fit with the role). As a shortcut, I then used some existing constructs used in marketing to profile supplier and customer quality and personality (more about that in the next section). This choice was linked to the very aim of my project, which was not only to develop the concept of inter-actor but also to examine how different actor profiles affect the interaction process and thus the development of business relationships. Departure from a more elaborate conceptualization of customer and supplier roles would have been helpful in directing both the observations and interviews with the informants. Whatever framework of departure, the problem remains how to identify the dimension of concern to the participants but also to investigate (and categorize) variation in these dimensions across actors. It has to be stressed that research approaching the actor from

an outside-in perspective is bound to be methodological complex and costly if compared to cross-sectional studies or experiments. Attempting the bilateral approach can also involve the problem of access to business situations and in particular the relationship interactions.

4.2 Investigating heterogeneity When the empirical research goes beyond conceptualizing the inter-actors and aims at investigating how actors affect relationship development, we need to address the issue of an actor’s specificity and uniqueness. Once we acknowledge the heterogeneity of actor roles as the root of relationship building, we are still left with the problem of how to address the uniqueness. Clearly cross-sectional studies that are so common in management and marketing are not likely to provide much insight into how inter-actors behave or how the economic organization is taking shape. The uniqueness and fluidity of actors from a relational perspective means thus that conceptualizing and theorizing must follow a somewhat different path. Dealing with intrinsic heterogeneity and uniqueness of actors in the business landscape requires focus on critical processes rather than generalizations about features and their impact. This, however, is easier said than done. Describing the processes implies the need to find dimensions to use in order to capture the differences, even if these are “inkind” regarding unique dimensions. Given that there is no “given actor”, we have to start from certain practices/processes (issues) and investigate the features of actors (and other variables) as they can be observed at work in these practices/processes. Investigating processes implies a different research focus and possibly has implications for research tools. One way out, as sometimes suggested, is the use of openended enquiry (e.g. open–ended interviews and unstructured observations) that tend to allow for richer findings and facilitate identification of genuine heterogeneity that tends to be about “differences in kind”. However, using a fully open data collection format tends to produce conspicuous material possibly comprising non-comparable dimensions. Also, as experience from ethnographic research suggests, while there is no such a thing as theory-free observation, ultimately the gist of the ethnographic research (for instance on innovation processes) becomes the “writing it up”, which amounts to more or less systematic selective interpretation of the material collected (Araujo & Hoholm, 2011). Addressing heterogeneity (in my particular case, addressing heterogeneous perceptions of counterparts’ identity) is problematic from a methodological perspective also because research in the more positivistic tradition has aimed at capturing similarities rather than accounting for differences and their consequences. It is clearly evident in most of the statistical methods that are nearly exclusively used to extract common regularities in observations. In my research, I have chosen to use structured constructs to collect perceptions of mutual identities among customers and suppliers and to capture how these changed after an interaction. More specifically, I have chosen to collect mutual perceptions related to the roles of the supplier and customer using the two constructs. One was the organization personality construct and relative scale proposed in the literature (Slaughter, Mohr, Zickar & Highhouse, 2004). Another was the construct of the perceived supplier and customer quality for which I used the Servqual scale (Parasuraman, Zeithaml & Berry, 2007) to assess


the supplier quality and an ad hoc scale to assess the quality of attractiveness of the customer (La Rocca, Caruana & Snehota, 2012). These served as the structured part of interviews that I carried out with the informants on both the customer and supplier sides. I kept part of the open ended interviews, but the structured part was aimed at and used as the main way to describe the differences in the mutually perceived entities of the actors. I used these standard scales as an alternative to pure qualitative research that would permit me to capture the true differences in kind but would likely produce indications on dimension that could be difficult to compare systematically. That certainly leads to loose validity in capturing the heterogeneity in non-comparable dimensions. The rationale, at least for me, was that such a way to proceed would permit me to identify a certain variety in identity perceptions. An open issue remains that doing so required the use of structured scales in unconventional ways for purposes other than that for which they have been developed. I was left with the problem of showing significant differences among various perceptions on the standard dimensions of constructs. My experience is that it is difficult to find statistical methods to achieve such an aim. Describing differences (and the heterogeneity) in an open-ended way is much easier than analyzing the differences (and the heterogeneity) systematically and attempting to categorize them. So, for instance, I found myself obliged to use cluster analysis, not to find homogeneous groups but to show that there was substantial heterogeneity in perceptions regarding the identity of even one and the same supplier. The heterogeneity assumption has been rather central in the IMP but also in the related streams of research on actors. Yet there is little systematic sharing of experience in researching the heterogeneity (in particular in reference to perceptions of actors’ identity and entity). If so, it either means that the problems are tall or that a systematic analysis of heterogeneity in perceptions is pointless. I am inclined to side with the former, but this points to the need to develop research methodology accordingly.

4.3 Capturing change Viewed from the interaction perspective, the actor’s entity appears to be continually re-enacted or re-interpreted. This property of the actor, when approached from an interaction perspective, requires research that takes into consideration evolution over time. A shift among interactions amounts to a shift from one definition of an actor to another. This emergent nature of the actor from an interaction perspective is emphasized both in the symbolic interactionism and ANT as well as in the IMP research regarding the actors in business relationships. Investigating how actors change over time implies longitudinal studies. Pure longitudinal studies of interaction in business relationships are probably difficult in practice. In abstract, this would mean to observe continuously multiple interaction processes and register how the mutual actor identities change over time. The minimum longitudinal element in the research design aimed at capturing the emergent nature of actors is then comparison of pictures of actors from at least two different points in time. In my study of how the identity of actors in business relationships changes, I employed research design that consisted in a three-step data collection. In the first step, some days before a scheduled meeting (interaction event), both the customer and the salesperson were interviewed using a structured questionnaire regarding the mutually expected performance and personality (using the constructs discussed in the previous section).

In the second step, I followed the sales force on their visits to customers observing their interactions. Finally, in the third step I re-interviewed salespersons and customers separately after the meeting using the same structured questionnaire employed in the first step of the data collection. Mutual perceptions before and after the meeting were compared, and significant changes were found in the perceptions before and after the face-to-face meeting. Incidentally, I used a t-test analysis that is commonly used to test the effects of a “treatment” on a sample (often involving a control group) assuming the context, and thus also the interaction effects, as neutral and with an experimental approach under laboratory conditions (In my case, I used the test considering the interaction as the “treatment” in order to show that a single interaction episode can have an effect on the perceived identity of the actor.) Such a way to proceed has the drawback of not allowing tracing differences that we earlier labeled “in kind”. Indeed, it does not permit tracking of possible changes in the dimensions of concern and thus no control for the possible consequences of the “genuine heterogeneity”. On the other hand, it permitted me to identify certain change patterns that might not have been observed through purely qualitative longitudinal research. The use of structured tools and data analysis in non-conventional ways has the advantage of revealing more easily unexpected results. For instance, looking at the change in perceptions through the use of quantitative scales allowed for the recognition that there was a trend in the direction of the changes. Indeed, most of the values decreased after interaction, hereby suggesting a “dropping” effect of interaction on perceptions. In other words, such a use of the scale permitted the capture of the change and also the direction of the change, even if it says nothing about the reasons for such a change. Addressing the need to assume a longitudinal perspective in empirical field studies, not unlike accommodating the need for an outside-in view, tends to lead to a methodologically complex research design which is likely to be costly and might entail problems related to its access. At the same time, assuming a longitudinal perspective is a necessary condition for research oriented toward a process view. The minimum step toward considering the time dimension is possibly to observe actors at two different points in time (for instance pre- and post-interaction events). Another issue that poses a specific problem to be addressed in research design with regard to the time dimension is that it requires bracketing of the flow of actions. Again, the claim common to the interaction perspective both of the symbolic interactionism, ANT, and IMP research is that there is no natural beginning point or a point of ending. That provides little guidance for “sensible” bracketing and deciding on observation points in time. Yet delimiting the phenomena (also in time) is likely to have significant consequences for the findings.

5. Final reflections on approaching the actors in the interactive business landscape Looking at actors from the perspective of interaction in business relationships implies making distance from the assumption that the actor is a given entity and that, to understand his/her conduct, one has to get to know his/her intents and plans, which are central assumptions in most of the management research. Looking at actors “from action” implies defining the actor not from its features but from how it is perceived in interactions. This leads to a conceptual shift from conceiving actors as homogenous


and stable to looking at actors as heterogeneous and emerging largely defined from the meaning attributed to its action by those concerned with it. Approaching the actor from an interaction perspective implies using an “open” methodological approach, possibly multi-method, that allows for the deep involvement of the researcher in the research context in both space and time. The challenges of approaching an actor from an interaction perspective involves methodological choices predicaments and trade-offs, some of which were discussed in this paper. From a methodological point of view, how to cope with an actor when the actor is conceived as a variable entity is a very interesting problem that has been only marginally addressed so far. Both the conceptual framework and the tools for assessing and measuring identity perceptions in business markets need to be redefined if not re-invented. Set to investigate interaction in buyer-seller relationships, this process is likely to better focus on the specific dimensions of the roles involved in business relationships such as, for instance, identifying the expectations that, in different contexts, define a good and attractive customer and a good and attractive supplier. Assuming, as it is done in much of the research on business relationships, that various relationships and actor identities perceived in interaction are indeed heterogeneous and emergent requires allowing for heterogeneity and the transient (emergent) nature of the actors also in the methods to use for empirical research. Heterogeneity is about differences in kind and not only the differences in degree (size or value) and it is the other side of uniqueness of actors. Heterogeneous entities have thus, by definition, different, non-comparable dimensions. Heterogeneity is derived from different role expectations and from the attribution of different properties (kind) to the same actor over time and in different relationships. The capabilities sought in a given relationship by one actor (and the attributed capabilities) are not necessarily of interest for another actor, who is likely to seek and attribute other capabilities to the counterpart at a different point in time. Diversity is an important attractor in relationships. What makes an actor different from another is not primarily a different weight given to an a priori set of characteristics. Rather, what makes each actor different is the variation in the dimensions in which others are interested from their own specific angle. Capturing the heterogeneity empirically means devising methods to capture the differences in which lies the uniqueness of actors. As I have dealt with perceptions of identity among actors in my study, I reached the conclusion that it is particularly urgent and challenging to develop a specific approach, and not only a construct, to assess the heterogeneous perceived relational identities. Although I am convinced that a qualitative, open-ended research approach would offer richer findings, I have chosen to use closed questionnaires focusing on pre-defined dimensions and related quantitative scales, as I am equally convinced that it is useful to produce observations that can be aggregated and compared and thus to provide a more systematic account of diversity and heterogeneity. Aware of the shortcomings of using a set of dimensions of identity that fit all customers in evaluating their supplier and vice versa, I cannot find a better way out of the predicament. As shown in a few studies on network development (e.g. Coviello, 2005; Provan et al., 2007), combining the qualitative with quantitative approaches appears to be fruitful because it can overcome the challenge of making sense of a non-numeric data that can be eclectic, unstructured and messy (Langley, 1999). It is possible that developing better conceptualization of the roles in buyer-seller relationships could ease the problem and the

apparently contradictory approach. In the specific context of investigating actors in business relationships, it could be useful, in my opinion, to develop specific constructs (scales) for assessing personality (features and characteristics) and quality (role performance expectation) of customer and supplier. Since customer supplier relationships in business markets are substantially different from those in consumer markets, I would argue that there is a need to develop a more suitable instrument focusing on the dimensions that truly matter for business customers in relation to business suppliers than those I used in my study (e.g. Servqual and organizational identity). Another methodological consideration regards the role of interaction in the formation of identities. In order to explore the possible link between the two, it would be desirable to develop an approach to capture and analyse interaction encounters and then to relate these more systematically to change and identity perceptions. It is a bit surprising that so few attempts have been made to develop a framework to describe interaction strategies in business relationships and those available in the IMP research (Shurr, Hedaa & Geersbro, 2008; Johnsen & Ford, 2007) have not been much tested empirically. The shift from conceiving actors as homogenous and stable to looking at actors as heterogeneous and emerging appears to imply studies that allow for deep involvement of the researcher in the research context in both space and time. Looking for action, instead of declared intentions, in a longitudinal perspective involves the use of an “open” methodological approach, such as an ethnographic approach (Hoholm & Araujo, 2011) that implies that the researcher participates – overtly or covertly, watching what happens, listening to what is said, asking questions, and collecting any other relevant data (Hammersely & Atkinson, 1995). The ethnography usually entails the use of a varied repertoire of research techniques: analyzing spoken discourse, narratives, collecting and interpreting visual materials, collecting oral history and life material, and so on. Whatever technique is used, the central point of the ethnographic approach is that research remains firmly rooted in the first-hand exploration of the research setting, since the aim is to describe what happens in the setting and how the people involved see others’ actions, their own actions, and the context. The ethnographic approach is based on the idea, close to the ideas about interaction behaviors in ANT, symbolic interactionism, and IMP, that behaviors are based on meanings that people attribute to and bring to situations and that the meaning is continually constructed and reconstructed on the basis of people’s interpretations of the situations they are in (Blumer, 1969). Whatever approach we take in empirically investigating interaction in business relationships and the role of actors in it, we have to acknowledge that, in confronting situations that arise in interaction, the actors act upon the context. In acting, they learn and reformulate their ideas, and new questions arise which can lead to yet newer ideas. Interactionist ethnography is fruitful for studying phenomena in an interactive business landscape, because it allows for the exploration of open-ended processes. The methodological requirements for ethnography formulated by Baszanger and Dodier (1997) asserting “the need for an empirical approach, the need to remain open to features that cannot be listed prior to the study, and the need to ground phenomena observed in the field” (p. 8) appear well suited with the task of empirically studying relationships in the business landscape. Finally, the word “approach” (instead of “methodology”) suggests the importance of making use of multiple methodological tools as a means for getting in touch with phenomena in a re-


flexive way. Reflexivity, more than rigor, may also inspire “creativity through opening up for new perspectives and providing reference points for what one is doing and to avoid or minimize certain ‘harmful’ aspects of research that follow from lack of reflexivity” (Alvesson, Hardy & Harley, 2008, p. 497).

References Abbot, A. (1995). Things of boundaries. Social Research, 62(4), 857-882. Achrol, R. S., & Kotler, P. (2011). Frontiers of the marketing paradigm in the third millennium. Journal of the Academy of Marketing Science, 40(1), 35-52. Akrich, M., & Latour, B. (1992). A summary of a convenient vocabulary for the semiotics of human and nonhuman assemblies. In W. Bijker & J. Law (Eds.), Shaping technology, building society: Studies in sociotechnical change (pp. 259-264). Cambridge, Mass: MIT Press. Alvesson, M., Hardy, C., & Harley, B. (2008). Reflecting on reflexivity: Reflexive textual practices in organization and management theory. Journal of Management Studies, 45(3), 480-501. Baszanger, I., & Dodier, N. (1997). Ethnography. Relating part of the whole. In D. Silverman (Ed.), Qualitative Research. Theory, Method and Practice. London: Sage Publications. Biddle, B.J. (1979). Role theory: Expectations, identities, and behaviors. New York: Academic Press. Blumer, H. (1969). Symbolic interactionism: Perspective and method. Englewood Cliffs, NJ: Prentice-Hall. Callon, M. (ed.) (1998). The laws of the markets. London: Blackwell Publishers. Chatman, J.A., Bell, N.E., & Staw, B.M. (1986). The managed thought: The role of self-justification and impression management in organizational settings. In H.P. Jr. Sims & D.A. Gioia (Eds.), The thinking organization. San Francisco: Jossey-Bass. Coviello, N. E. (2005). Integrating qualitative and quantitative techniques in network analysis. Qualitative Market Research: An International Journal, 8(1), 39-60. Cyert, R. & March, J.G. (1963). A behavioral theory of the firm. Englewood Cliffs, NJ.: Prentice-Hall Inc. Dubois, A., & Gadde, L.-E. (2002). Systematic combining: An abductive approach to case research. Journal of Business Research, 55(7), 553−560. Emirbayer, M., & Mische, A. (1998). What is agency? The American Journal of Sociology, 103(4), 962-1023. Ford, D., Gadde, L.-E., Håkansson, H., Snehota, I., & Waluszewski, A. (2010). Analysing business interaction. IMP Journal, 4(1), 82-103. Goffman, E. (1967). Interaction ritual. Garden City, New York: Anchor. Goffman, E. (1974). Frame analysis: An essay on the organization of experience. New York: Harper and Row. Håkansson, H., & Snehota, I. (1995). Developing relationship in business networks. London: Routledge. Håkansson, H., Ford, D., Gadde, L.-E., Snehota, I., & Waluszewski, A. (2009). Business in networks. Chichester: John Wiley & Sons. Hammersley, M., & Atkinson, P. (1995). Ethnography: Principles in practice. London: Routledge. (First edition published by Tavistock in 1983) Hoholm, T., & Araujo, L. (2011). Studying innovation processes in real-time: The promises and challenges of ethnography. Industrial Marketing Management, 40(6), 933-939.

Johnsen, T., & Ford, D. (2007). Customer approaches to product development with suppliers. Industrial Marketing Management, 36(3), 300-308. Johanson, J., & Vahlne, J. (2011). Markets as networks: implications for strategy-making. Journal of the Academy of Marketing Science, 39(4), 484-491. Katz, D., & Kahn, R.L. (1978). The social psychology of organizations. New York, NY: John Wiley and Sons. Langley, A. (1999). Strategies for theorizing from process data. Academy of Management Review, 24(4), 691-710 La Rocca, A., Caruana, A., & Snehota, I. (2012). Measuring customer attractiveness. Industrial Marketing Management, 41,1241-1248. La Rocca, A. (2011). Interaction and actors identities in business relationships. Dissertation, Università della Svizzera Italiana, Lugano. Latour, B. (1987). Science in action: How to follow scientists and engineers through society. Milton Keynes: Open University Press. Law, J., & Hassard, J. (eds.) (1999). Actor network theory and after. Oxford Malden: Blackwell Marglin, S. (2008). Dismal science: How thinking like an economist undermines community. Cambridge, MA: Harvard University Press. Mintzberg, H. (1973). The nature of managerial work. New York: Harper & Row. Morgan, R.M, & Hunt, S.D. (1994). The commitment-trust theory of relationship marketing. Journal of Marketing, 58(3), 20-38. Parasuraman, A., Zeithaml, V.A., & Berry, L. (1994). Alternative scales for measuring service quality: A comparative assessment based on psychometric and diagnostic criteria. Journal of Retailing, 70(3), 201-231. Provan, K.G., Fish, A., & Sydow, J. (2007). Interorganizational networks at the network level: A review of the empirical literature on whole networks. Journal of Management, 33(3), 479-516. Schelling, T. C. (1978). Micromotives and macrobehavior. Cambridge, MA: Harvard University Press. Schurr, P.H., Hedaa, L., & Geersbro, J. (2008). Interaction episodes as engines of relationship change. Journal of Business Research, 61(8), 877–884. Simon, H. (1957). Administrative behavior: a study of decisionmaking processes in administrative organization. New York: MacMillan. Slaughter, J.E., Mohr, D.C., Zickar, M.J., & Highhouse, S. (2004). Personality trait inferences about organizations: Development of a measure and assessment of construct validity. Journal of Applied Psychology, 89(1), 85-103. Sluss, D.M., & Ashforth, B.E. (2007). Relational identity and identification: Defining ourselves through work relationships. Academy of Management Review, 32(1), 9-32. (Lead article). Stryker, S., & Statham, A. (1985). Symbolic interaction and role theory. In G. Lindzey & E. Aronson (Eds.), Handbook of social psychology (3rd ed.). New York: Random House. Thibaut, J.W., & Kelley, H.H. (1959). The social psychology of groups. New York: Wiley. Waluszewski, A., & Håkansson, H. (2006). The importance of angry actors. Paper presented at the 22nd IMP Conference, Milan. Waluszewski, A., Hadjikhani, A., Baraldi, E. (2009). An interactive perspective on business in practice and business in theory. Industrial Marketing Management, 38(6), 565-569.


Weick, K.E. (1995). Sensemaking in organization. Thousand Oaks, CA: Sage.

Antonella La Rocca, Research Fellow, Institute of Marketing and Communication Management, University of Lugano - USI, Via G. Buffi 13, 6904 Lugano, Switzerland, Email: [email protected]


The Political Embeddedness of Business Networks in a Chinese Context: The Case of a Biopharmaceutical Business Network Åse Linné & Tommy Tsung-Ying Shih Uppsala University Lund University

Abstract

Several researchers point to the intense interaction between the Chinese government and the business sphere. This paper investigates the political embeddedness of biopharmaceutical business networks in China. The biopharmaceutical drug innovation process illustrates the political embeddedness of a business network. The analysis is based on the resource interaction framework (4R model) and identifies interactions between political, business and non-business actors in the network. The paper finds that the business network is deeply politically embedded and that the government actively engages research institutes and companies in the development and commercialisation of biopharmaceutical drugs in China.

Key words: Political embeddedness, innovation, China, resource interaction, biopharmaceutical drugs

1. Introduction Several scholars point to the Chinese government’s influence on business in China (e.g. Gibb & Li, 2003; Naughton, 1994). Redding and Witt (2009) note that the Chinese government controls resources and has the political power to direct industrial activities on a scale that no other nation in the world can emulate. Chinese companies, therefore, often seek and need to establish connections with governmental organisations in order to get access to specific resources, legitimacy and favourable treatment (Peng & Lou, 2000; Nee, 1992). In directing and steering the Chinese industrial landscape, the government relies to a large degree on proactive industrial policy-making. Industrial policies in recent years have focused on creating innovation capabilities and transforming China into a modern market economy (Liu & White, 2001). Innovation in high-tech areas, as acknowledged by the Chinese government, is particularly crucial for China to achieve economic, environmental and social sustainability (MOST, 2006). Studies on the possibilities of government, at both central and local levels, to direct and control innovation in China have predominantly focused on exploring the national innovation system (Liu & White, 2001; Motohashi & Yun, 2006). Moreover, a large number of studies have focused on regional innovation development and the establishment of science parks (Walcott, 2003; Wei & Leung, 2005; Leiftner et al., 2006; Zhou, 2008); these studies commonly investigate the establishment of an innovative milieu for Chinese high-tech industries, and the role of the government in creating a conducive infrastructure. Less attention, however, is given to how innovation becomes embedded in a networked environment and how government plays an active part in the innovation landscape; this paper aims to provide a deepened

empirical understanding of these matters and thereby contribute to the literature on the political embeddedness of business networks (e.g., Hadjikhani & Håkansson, 1996; Hadjikhani & Sharma, 1999; Welch & Wilkinson, 2004). To that end, this paper is based on a case study of the embedding of an innovation in the Chinese biopharmaceutical sector. The Chinese biopharmaceutical sector has, since the late 1990s, been promoted by the government as an industry of the future. Due to China’s enormous market and the pro-active industrial policies of the government, a biopharmaceutical industry has emerged, consisting of more than 400 research institutes and universities, 500 Chinese biopharmaceutical companies, and the launch of more than 30 biopharmaceutical drugs on the Chinese market (Chen et al., 2007). These impressive developments suggest that governments can play an important role in forming networks and embedding innovations in economic structures. Hence the question that we seek to answer is: how do government actors support and direct innovation processes in China? In the next section, we present the theoretical underpinnings, including the notion of the political embeddedness of business networks, and the characteristics of innovation processes. The subsequent section discusses the method used, followed by a description of the case study of the embedding of a biopharmaceutical drug. Lastly, we present the analysis and the conclusions.

2. Theoretical framework 2.1 Political embeddedness The Industrial Marketing and Purchasing approach has, since the 1970s, offered a perspective on the nature of business interac-


tion in networks (Håkansson & Waluszewski, 2007). Although IMP scholars acknowledge that the contextual setting, such as the political environment, is a part of influential network behaviour (e.g., Johanson & Mattsson, 1992), it has often not been considered explicitly in the analysis of industrial networks. The lack of studies concerning the ‘political environment’ and its effect on business interaction has been pointed out by researchers such as Michaelson (2007), Welch and Wilkinson (2004), and Hadjikhani & Sharma (1999). Political actors can both drive and impede business interaction and are an important component in understanding business networks (Hadjikhani & Håkansson, 1996). A main issue regarding the importance of governmentcompany relationships is that the governments often control resources, such as access to other organisations and business or non-business actors, which can be important for companies’ business development (Okhmatoviskiy, 2010; Hadjikhani & Håkansson, 1996). Moreover, a company’s relationship with government bodies can result in increased legitimacy (Welch & Wilkinson, 2004). Scholars argue that, in a business landscape where both market and planning systems work side-by-side, ties to government bodies are considered to be crucial (Krug, 2012; Li & Zhang, 2007). In an emerging economy like China, relationships with government bodies are seen as particularly important, due to a lack of established institutions and legal frameworks (e.g., Hoskisson et al., 2000; Peng & Heat, 1996; Xin & Pearce, 1996). Whitley (2000) also emphasises the central role of the Chinese government, by characterising the Chinese business landscape as a “state-organized” model. As such, Child (1994: 154) demonstrates that private companies in China, and their managers, have a “disproportionately greater contact” with government bodies than within other business systems. The intense interaction in China between the government and companies is arguably dependent on the political setting (Peng & Zhou, 2005). Compared to other emerging countries, China is steered by an authoritarian state whose legitimacy is closely linked to the issue of economic growth (Göbel, 2011); this has meant that the state has often had close ties to the business sphere (Peng & Heat, 1996). In order to understand the government-company relationship, we use the concept of political embeddedness, which refers to “a specific form of social embeddedness characterized by a mix of formal and informal ties with political agents” (Krug, 2012: 4), and “… bureaucratic, instrumental, or affective ties to the state and its actors” (Michelson, 2007: 352). Embeddedness is derived from economic sociology, where it is emphasised that organisations are not only embedded economically in other organisations, but also socially (Granovetter, 1985). From this perspective, organisations become embedded in each other through interaction, which creates opportunities but also constraints. According to Okhmatoviskiy (2010), the concept of political embeddedness can be used to see beyond the role of government as the regulator of the business landscape and view it rather as an economic actor within the business landscape. Within the IMP sphere, Welch and Wilkinson (2004) identify four main subjects for studies on political embeddedness in the IMP: 1) political institutions, 2) political actors, 3) political activities, and 4) political resources. This paper focuses mainly on the role of political actors and political resources in the embedding of innovation in networks. The above-mentioned scholars generally highlight political embeddedness as an important component of the business landscape. Moreover, several studies point to the importance of government in creating conditions for innovation (Edqvist,

1997). These innovation policies aim to create the foundation for future economic growth by creating a suitable milieu where industrial actors can commercialise scientific discoveries (Lundvall & Borrás, 2005). As a consequence, innovation is tightly interlinked with cooperation between government and business. However, there is a lack of studies where the role of the government in specific innovation processes at the network level is brought to the surface; most studies are focused at the macro level, and especially on the role of the government as a policymaker (c.f. Waluszewski, 2011). The following section proposes a framework, through which the role of government actors in innovation processes can be analysed.

2.2 The innovation process and interconnected resources Innovation is increasingly acknowledged as an interactive process, and as the result of resource interactions across time, space and organisational borders (Chesbrough, 2003; Håkansson & Waluszewski, 2002). As Håkansson & Waluszewski (2007) state, innovation does not only refer to the development of novel technological solutions, but also to how existing solutions are combined in new ways within specific contexts. Innovation thus refers to a broader phenomenon and process (Fagerberg, 2009). Generally, innovation entails the replacement of existing technologies, products and processes by solutions that are more efficient or otherwise considered better (Schumpeter, 1934). However, the literature does not agree on whether these solutions need to be ‘entirely new’ or ‘new to a particular context’ to be called an innovation. Several scholars support the notion that an already-existing solution introduced into a new context or market can be considered an innovation (Schumpeter, 1934; Håkansson & Waluszewski, 2002; Fagerberg et al., 2005; Rosenberg, 1982). If a narrow definition of innovation applied, innovation would only occur in the US, Japan, UK, France and Germany (Lundvall et al., 2009). A broader definition of innovation suggests that, for example, other countries such as Sweden and Taiwan can also be called innovative, as they have a highly developed capacity to absorb and use technology developed elsewhere (Lundvall et al., 2009). The nature of innovation suggests that, in emerging economies such as China, it is important to understand how new solutions and technologies get embedded in industrial networks of interlinked resources, i.e., the process of innovation. The argument behind this idea is that a solution’s introduction into a new context implies that it needs to adapt and assimilate to the unique context, with its existing resource structures including companies, products, technologies, users, institutions etc. Especially when a new solution is highly technical and based on scientific discoveries, it needs to connect and adapt to a wide range of activated resources in the new context (Håkansson & Waluszewski, 2002). Due to the interconnected web of resources, innovation processes have been described as the result of ‘trial and error’ (Basalla, 1988; Dosi et al., 1988). Others, such as Van de Ven et al. (1995: 5), describe innovation as a nonlinear dynamic system. Several scholars note that new solutions often fail to reach the market (e.g., Tidd et al., 2001; Pavitt, 1991; Cooper, 1975). According to von Hippel (1976), it is the creation of interfaces between the production and use of a new solution – something that is difficult to achieve in an innovation process (von Hippel, 1976). Moreover, several researchers emphasise that the determinant of an innovation is whether it gets widespread use (Akrich et al., 2002; Rosenberg, 1982).


The above understanding resonates with how IMP scholars treat the phenomenon of innovation. What IMP scholars highlight is that any new solution needs to find its place in three interconnected settings: the developing, producing and using settings (Linné, 2012; Ingemansson, 2010; Shih, 2009; Håkansson & Waluszewski, 2007); that is to say, in an innovation process, every new solution needs to be embedded in an economic structure where it can be used, produced and developed. This means that any new solution has to be adapted to resource structures and connected to other companies, products, technologies, individuals and so forth in each of these interlinked settings (Håkansson & Waluszewski, 2007). In short, the using, producing and developing settings shape and mould the new solution during its innovation journey. Consequently, embedding a new solution means to connect and adjust the new solution to already-activated resources in a specific context (Akrich et al., 2002). According to the IMP view, the innovation process combines resources spanning the developing, producing and using settings. Hence, it is through this resource-combining process that the new solution gets its life. To investigate the political embeddedness of a new drug innovation in China, this paper focuses on revealing the government resources that embed the new drug during its innovation process, and it uses the concept of resource interfaces, which refers to the ‘contact points’ between resources. It is through these interfaces that resources interact, influence and adapt to each other (Baraldi et al., 2012; Baraldi & Strömsten, 2006). This paper identifies the political actors as represented by governmental units at central, regional, and local levels. Hence, political embeddedness is discussed in terms of direct or indirect resource interfaces between these governmental units and the focal drug innovation. More specifically, the paper concentrates on revealing two types of resources discussed in the 4R model developed by Håkansson & Waluszewski (2002). The first type is physical/technical resources, which are represented by products and facilities; these are physical artefacts such as equipment, factories, prototypes etc. The second type is social/organisational resources, which are represented by organisational units and organisational relationships; these are resources with a social dimension such as knowledge, routines and experience. Both technical and organisational resources embed the new drug throughout its innovation journey, and this paper aims to reveal the direct or indirect interfaces with governmental

units and investigate their impact on the embedding resources in the developing, producing and using settings. The framework is illustrated below in Figure 1.

3. Method and research design This paper departs from a larger empirical study of drug innovation in China’s biopharmaceutical industry (Linné, 2012). In total, five embedded cases (Yin, 1994) were illustrated by following a biotechnology instrument, ÄKTApilot, supplied by GE Healthcare. The instrument is used to separate molecules at a scale between laboratory and large-scale production, and thus is used for the identification of drug projects in late development. Drug innovation processes have three interlinked dimensions: the developing, producing and using settings. The analysis aims to identify government involvement and its direct or indirect interfaces with the drug innovation. We investigate how interfaces to governmental units affect the embedding of technical and organisational resources throughout the whole innovation process, i.e., the developing, producing and using settings. We used a single case study for our investigation. As Eisenhardt (1989) points out, case studies are appropriate when seeking to investigate change and development processes. Dubois and Araujo (2007) also stress that a case study approach is suitable for understanding interactions in processes over time and space. The case study used in this paper illuminates the development, production and use of an antibody drug. This case was chosen due to the participation of political actors (represented by governmental units) in all three settings at various levels, including national and local levels; government related interfaces stretched across the whole innovation process, therefore the case can provide a better understanding of the role of the Chinese state in new drug innovation processes. At the time the drug was under development, there were a total of 13 drugs in late clinical trials in China (Loüet, 2004). The main empirical data were collected between 2004 and 2007, and a total of 17 interviews were conducted with respondents directly and indirectly involved in the drug innovation process. The interviewees included a production manager of the company producing the drug and representatives of GE Healthcare involved in the sales of laboratory and production equipment . Moreover, a number of interviews were conducted with respon-

Figure 1: Framework


dents who had extensive knowledge of the biopharmaceutical industry in China. The interviews with the first group of respondents aimed at understanding the interaction between the actors involved directly in the drug’s development, production and use. The interviews with the second group focused on understanding the innovation process and the institutional factors affecting the development, production and use of drugs in China generally. Secondary sources were also used, such as official data from the Ministry of Science and Technology (MOST), the Chinese Academy of Science (CAS), and homepages related to the main producing company and its owners.

4. Case study: Development, production and use of an antibody drug The promotion of the Chinese biotechnology industry dates back to the mid-1980s, when the Chinese government promulgated policies intended to strengthen biotechnology science. These measures increased the number of research institutes and universities focusing on biotechnology. Since the mid-1990s, the Chinese government has also emphasised the need for China to use its established science base to improve China’s innovation capacity, which has resulted in an increase in the number of science parks and industry-business interactions. In 2007, there were more than 400 universities and research institutes that had research projects related to biotechnology and more than 500 Chinese firms were engaged in developing or producing biopharmaceutical drugs (Chen et al., 2007). The following case focuses on the embedding of a drug based on antibodies used to target diseases such as cancer. The economic potential of drugs based on antibodies has led to a steady increase in antibody research and the number of companies commercialising such drugs in the United States and Europe. Since the late 1990s, several Chinese companies have also become involved in developing drugs based on antibodies. The commercialisation of such drugs requires large-scale production and high-end production equipment. As China is a relative newcomer to the biopharmaceutical field, the government has actively participated in creating an infrastructure, as well as providing resources to companies involved in the development and commercialisation of drugs.

4.1 The development of an antibody drug in China The Chinese military has long played a central role in China’s science and technology system. After the Cultural Revolution (1966-1976), the Chinese military was the only surviving research institution. With the introduction of the ‘open-door policy’ in 1978, the Chinese government encouraged military research organisations to support the foundation of the “socialistic market economy” and contribute to the modernisation of China (Karmel, 1997). This new directive led to an decrease in funding for research supported under the banner of the Chinese military, and encouraged military organisations to instead increasingly interact with society and business. Chinese military research organisations became more involved in the commercialisation of research and the establishment of new companies (Bitzinger, 2007). The Shanghai Military Medical University (SMMU) – a leading research institute related to antibody research in China – has formally established ties with business actors. In the mid-1990s, it initiated the development of a new biopharmaceutical drug

based on antibodies targeting arthritis. The drug candidate was new to the SMMU, but it was based on an American drug with worldwide annual sales of more than US$ 3 billion. Hence, the drug candidate was a so-called ‘bio-similar’ drug, sharing some structural similarities with an already-existing drug (Scheinberg & Kay, 2012). By developing a bio-similar drug, the SMMU saw the opportunity to develop a cheaper Chinese version. With the drug, the SMMU also hoped to generate money for further research. The researchers at the SMMU had experience of conducting research on antibodies, but had no prior experience in drug development. They used the original drug as a reference point from which to initiate the development. While the SMMU could use published papers on the original drug as a point of departure, the exact details concerning the development were only accessible to the original developer. Moreover, as with any new biopharmaceutical product, the drug needed to go through development and production according to regulatory statutes, and also become established within distribution channels. The first stage of clinical trials of the new drug was initiated in the late 1990s. At this stage, the SMMU also needed to scale production of the active ingredient of the drug, as well as distributing it to public hospitals and patients all over China. To achieve these goals, the military institute collaborated with a pharmaceutical company, Shanghai MAB Pharmaceutical (MAB).

4.2 Finding a company to continue commercialisation of the new drug Shanghai MAB Pharmaceutical was formed in 1998 as joint venture between a Chinese import/export (IE) company and a Hong Kong-based real-estate company. Both companies had no previous experience in biotechnology or pharmaceuticals. The Hong Kong-based company is a subsidiary of one of China’s largest state-owned conglomerates. This conglomerate, established in 1979, was appointed by the State Council to attract capital and investment in new ventures. Since the introduction of the ‘open-door policy’, the group has played a major role in China’s business landscape. MAB was established without having any drug projects to develop. Thus, the company needed to find suitable drug projects to integrate into its business operations. While searching for suitable drug projects, MAB came across the SMMU, which suggested that MAB license the anti-arthritis drug candidate that was in clinical trials. In 2000, the two parties signed an agreement concerning further cooperation and development of the drug. The agreement stated that MAB would focus on commercialisation and setting up a production facility, while the SMMU would continue focusing on research and clinical trials. To set up the production process and handle the clinical trials, MAB employed several researchers from the SMMU on a part-time basis. After the agreement was signed between the two parties, MAB focused on setting up both R&D and production facilities at the company’s location in the Shanghai Zhangjiang High-Tech Park – a science park established by the Shanghai municipality in 1996, accommodating more than 130 biotechnology and pharmaceutical companies. Being located in the science park, MAB was granted tax-reductions and other benefits from the Shanghai municipality. The R&D and production departments were located within the same facility to facilitate close interaction. The R&D department focused on producing samples for clinical trials, and the production process of the new drug was set up using mainly high-end imported equipment. In order to spur the


increase of Chinese companies involved in high-tech industries, the Chinese government introduced a reduction on the import tax of such high-tech equipment, which MAB could benefit from. The production process of the new drug was divided into three separated production steps: the phases of fermentation, purification and formulation. The fermentation phase, where the mass culture of proteins take place, was set up with bioreactors from Germany. The purification phase, the actual cleaning of the proteins, was performed using bioprocess systems and ÄKTApilot supplied by GE Healthcare. In the formulation phase, where the proteins are mixed with other chemicals and transformed into an injectable drug, domestic Chinese suppliers supplied the main equipment. By the time the drug project was purchased by MAB, the SMMU had already conducted the first clinical trials. With MAB taking over the drug project, the company’s internal R&D department took over responsibility for producing samples for clinical trials, and for dealing with the public regulatory agency (the SFDA), and SFDA-approved public hospitals. In 2003, MAB was approved by the SFDA to continue the late clinical trials that were being performed at various public hospitals. The actual approval of the new drug was somewhat delayed, due to a corruption scandal in 2005 within the SFDA. However, by late 2005, the Chinese SFDA approved MAB’s New Drug Application (NDA) for an antibody drug for the Chinese market.

4.3 The new drug as the reason for the establishment of a new research facility in cooperation with the Shanghai municipality With the NDA’s approval, only the final approval of the current Good Manufacturing Practice (cGMP) of the production facility remained. However, in order to receive a cGMP-approved production facility, MAB needed to separate the R&D and production departments. The physical separation of the departments was also necessary, as more bio-similar drug projects had been transferred from the SMMU to MAB. Since 2000, several drug projects have been transferred from the SMMU to MAB for further commercialisation. In 2008, MAB was the largest Chinese company involved in developing drugs based on antibodies. In order to facilitate the development of the increased number of drug projects, MAB needed larger facilities to handle the clinical production of drugs further ‘upstream’. The separation of the departments was realised in 2006, and MAB launched the antibody drug later that year. The Science and Technology Commission of Shanghai Municipality (STCSM) played an integral role in separating the departments within MAB. The STCSM became a new partner to both MAB and the SMMU as it funded the larger part of a new R&D department within MAB – the MAB Biotechnology Institute. The MAB Biotechnology Institute was equipped with highend imported equipment that could screen several antibody drugs at the same time. This capability improved the possibility of handling more projects from the SMMU. The MAB Biotechnology Institute became responsible for handling all the research, along with the clinical trials of all the drugs belonging to MAB and the SMMU. The STCSM’s funding of the new biotechnology institute was contingent on other Chinese companies and research groups being invited to use the R&D facility. The larger aim of the STCSM was to strengthen the development of antibody drugs within the Shanghai region. The head of the MAB Biotechnology Institute was a professor at the SMMU, and the main users of the R&D facility were researchers also from the SMMU that were involved in drug projects related to MAB.

4.4 Users of the antibody drug and achieving Chinese ‘blockbuster status’ Since the launch of the antibody drug in China, sales have gradually increased each year. Due to the improved sales performance, MAB scaled up the production process and invested in larger production equipment in 2008. That same year, sales of the drug reached RMB 140 million (US$ 17.5 million), and exceeded the ‘blockbuster’ cut-off point of annual sales above RMB 100 million (US$ 12.5 million). The drug was used to treat patients in public hospitals in over 30 provinces all over China. In comparison to the original American drug, MAB’s antibody drug was sold at a discount of 60 percent, making it a more affordable drug on the Chinese market. By 2012, over 50,000 patients had received treatment with the antibody drug. Priced at approximately US$ 6000, however, the drug remained relatively expensive, as patients in China are not eligible for any reimbursements from the national health insurance system for biopharmaceutical drugs (Shanghai MAB Pharmaceutical Homepage, 2012). Given its relative success in financial terms and its broad use in clinical settings, the drug and MAB have received recognition and support from the Chinese government. The national Ministry of Science and Technology (MOST) – a department directly under the State Council – provided funding of RMB 10 million (US$ 1.25 million), for the commercialisation of the drug via the 863-programme in 2005. Additional funding was also granted through other schemes within the MOST. In 2006, acknowledgment in the form of the National Award of Technical Innovation and National New Crucial Product was awarded to the new drug and the manufacturer, MAB. There has been a massive infusion of financial capital into the development of the drug; MAB estimates that, up to 2007, more than RMB 1.2 billion had been invested in the company. This includes the investments from the two main public sector owners – a trading company and a real estate company – and the investments made by MOST, the Shanghai local government, and the SMMU.

5. Discussion In the following discussion, we reveal the degree of political embeddedness by identifying governmental units and their direct and indirect interfaces with organisational and technical resources throughout the developing, producing and using settings.

5.1 Interfaces between governmental units and organisational resources The case illustrates that the initiation of the development of the antibody drug was indirectly related to the changed interface between the Chinese military and the Chinese state government, the State Council. The development of the drug was a response to the decline in funding for military research from the central government. With less government funding for research, and an increased pressure to contribute directly to development of the economy, the main research unit, SSMU, sought ways to commercialise its research. As an established research organisation with experiences from several other drug development projects, the SSMU was in a favourable position to develop a biopharmaceutical drug. However, the interface between the research unit and the main producing unit, MAB, was also important in further downstream development. MAB – a company with excess


capital that was linked to the State Council – sought to invest in a new business area. The interface between the SSMU and MAB thus combined financial resources and biopharmaceutical expertise not only to proceed with the development of the drug but also to establish it into production. The SSMU had experience in scale-up production and assisted MAB in establishing a production facility. In the use of the drug, an important interface concerns the drug and public hospitals. Public hospitals not only provide a platform for conducting clinical trials, but they are also the main users and prescribers of the drug. Thus, these clinical units can be seen as spanning the whole innovation process, including the developing, producing and using settings. Moreover, the government is important in embedding the new drug in the using setting by deciding the price of the drug. The price needs to be set at a lower price level than the ‘original’ drug to stimulate sales. The antibody drug displays strong interfaces to the regulating unit, the SFDA, which regulates the development, production and use.

5.2 Interfaces between governmental units and technical resources The development stage of the drug displays the close ties between the organisational resources of the business and government spheres. In the production of the drug, the importance of combining technical resources becomes more apparent. There are strong interfaces with physical resources related to the main producing unit, MAB. However, in being able to establish MAB as a producing unit, there are also strong interfaces with the government. In the producing setting, the government is the main financer, but the financing originates from different government sources. First, both national and local governmental units support the new drug innovation, for instance through preferable taxation rates for the import of high-tech equipment, along with financing the establishment of a new biotechnology institute through the STCSM and through the 863-plan by MOST. Secondly, the main public owners provide the financing for both the development and the production of the new drug. Through these two types of government funding sources, it was possible to produce the new drug. The purchase of equipment, along with setting up a production process, was possible by employing suitable staff with the right experience. Thus, the interface with these governmental financing units resulted in the possibility of investing in technical resources, such as equipment and setting up a factory, and establishing the organisational units, such as establishing relationships with suppliers, attracting new personnel, and so forth. The strong interface between the production of the drug and government financing units is important in creating legitimacy, not only for the drug and the producing company, but also for both national and local governments.

5.3 Government influence over the innovation process The drug innovation process illustrated in this paper displays a strong political embeddedness. The units belonging to the governmental apparatus are active participants in the innovation process, and play a central role in the formation of new interfaces both within and between the developing, producing and using settings. It is evident that the relationships to Chinese governmental bodies are an important component in supporting drug innovation processes, and government is actively

involved in creating both technical and organisational resource interfaces. The drug innovation process is closely related to established research institutions, with research units belonging to organisations dating back more than 50 years. The government also encourages companies to engage in commercialisation of biopharmaceutical drugs; it does so by providing essential resources, such as financing parts of the development or production. However, even though the companies and organisations involved in biopharmaceutical drugs are encouraged to innovate, these units are allowed the freedom to experiment with new resource combinations.

6. Conclusions Drug innovation processes are risky and uncertain ventures, and encompass various resource interactions across the stages of development, production and use. This study has presented a case study of a drug innovation process and its political embeddedness. The heavy political embeddedness of business networks in China can create stability of interaction, due to the government’s provision of financial resources and its addition of legitimacy to drug projects. The role of the government is multifaceted. The government is a significant actor in creating vision and allocating resources to achieve what it ideates. For, in the case of innovation, the government can influence established organisations to pursue new goals related to biopharmaceutical development. Its authoritarian rule allows for a certain amount of control over the economy and, in this case, it creates the possibility for government-related actors to influence the innovation process. This means that the government takes a proactive position in networks and, at local and national levels, it is an active mediator in the innovation process. The findings of this paper suggest how the government plays an influential part in the whole innovation process. In order to reduce the risk and uncertainty of innovation, companies establish relationships with government actors. The government functions as a mediator in the network and creates interfaces between key actors, supports innovation processes, provides legitimacy to projects and supplies resources that can fill gaps in the value chain. The political context provides the government with the tools to form networks and dictate general business behaviour, along with creating legitimacy for its own work in sustaining long-term economic growth within China.

6.1 Further research The management of networks has been a heavily debated issue in the industrial marketing literature. The case of China suggests that the government is capable of forming networks, along with managing and steering innovation processes. However, one key factor of the ‘steering capacity’ of the Chinese government is due to its authoritarian character. The role of the state in development processes and the formation of business networks has not been studied extensively in the IMP literature, so more study should be devoted to this topic.

References Akrich, M., Callon, M. & Latour, B. (2002). The Key to Success in Innovation Part II: The Art of Choosing Good Spokespersons. International Journal of Innovation Management, 6:


207-225. Bitzinger, R.A. (2007). Reforming China’s Defense Industry: Progress in Spite of Itself? The Korean Journal of Defense Analysis, 19: 99-118. Basalla, G. (1988). The Evolution of Technology. New York: Cambridge University Press. Baraldi, E., Gressevold, E., Harrison, D. (2012). Resource Interaction in Inter-organizational Networks: Foundations, Comparison, and a Research Agenda. Journal of Business Research, 65: 266-276. Baraldi, E., Strömsten, T. (2006). Embedding and Utilising Low Weight: Value Creation and Resource Configuration in the Networks around IKEA’s Lack Table and Holmen’s Newsprint. Advances in Business Marketing and Purchasing, 14: 251-316. Chen, Z., Wang, H-G., Wen, Z-J. & Wang, Y-H. (2007). Life sciences and biotechnology in China. Philosophical Transactions of the Royal Society B, 362: 947-957. Chesbrough, H. (2003) Open Innovation: The New Imperative for Creating and Profiting from Technology. Boston: Harvard Business School Press. Child, J. (1994). Management in China During the Age of Reform. Cambridge: Cambridge University Press. Cooper, R.G. (1975). Why New Industrial Products Fail. Industrial Marketing Management, 4: 315-326. Dosi, G., Freeman, C., Nelson, R., Silverberg, G. & Soete, L. (eds.) (1988). Technical Change and Economic Theory. London and New York: Pinter Publishers. Dubois, A. & Araujo, L. (2007). Case Research in Purchasing and Supply Management: Opportunity and Challenges. Journal of Purchasing & Supply Management, 13: 170-181. Dubois, A. & Gadde, L.-E. (2002). Systematic Combining: An Abductive Approach to Case Research. Journal of Business Research, 55: 553-560. Easton, G. (1995). Methodology and Industrial Network. In: Möller, K & Wilson, D. T. (eds.) Business Marketing: An Interaction and Network Perspective. Boston, Dordercht, and London: Kluwer Academic Publishing. Edquist, C. (1997). Systems of Innovation. Technologies, Institutions and Organizations. London and Washington: Pinter. Eisenhardt, K. (1989). Building Theories From Case Study Research. The Academy of Management Review, 14: 532-550. Fagerberg, J. (2009) Innovation Systems, Technology and Development: Unpacking the Relationships, in Lundvall, B.Å., Joseph, K.J., Chaminade, C. and Vang, J. (eds) Handbook of Innovation Systems and Developing Countries, Cheltenham: Edward Elgar. Gibb, A. & Li, J. (2003). Organizing for enterprise in China: what can be learned from the Chinese micro, small and medium enterprise development experience? Futures, 35: 403-421. Granovetter, M. (1985). Economic Action and Social Structure: The Problem of Embeddedness. American Journal of Sociology, 91: 481-510. Göbel, C. (2011). Authoritarian consolidation. European Political Science, 10: 176-90. Hadjikhani, A. & Håkansson, H. (1996) Political actions in business networks: a Swedish case. International Journal of Research in Marketing, 13: 431-447. Hadjikhani, A. & Sharma, D. (1999) A view on political and business actors. Advances in International Marketing, 9: 24357. Halinen, A. & Törnroos, J.-Å. (2005). Using case methods in the study of contemporary business networks. Journal of Business

Research, 58: 1285-1297. Hendriks, J., Liang, Y. & Zeng, B. (2010). China’s emerging vaccine industry. Human Vaccines, 6: 602-607. Hoskisson, R.E., Eden, L., Lau, C.M. & Wright, M. (2000). Strategy in emerging economies, Academy of Management Journal, 43: 249–267. Håkansson, H. & Waluszewski, A. (2002). Managing Technological Development: IKEA, the Environment and Technology. London: Routledge. Håkansson, H. & Waluszewski, A. (eds.) (2007). Knowledge and Innovation in Business and Industry - The Importance of Using Others. London: Routledge. Ingemansson, M. (2010). Success as Science but Burden for Business? On the Difficult Relationship Between Scientific Advancement and Innovation. Doctoral Thesis, Department of Business Studies, Uppsala University. Johanson, J. & Mattsson, L-G. (1992). Network positions and strategic action—An analytical framework. In Axelsson, B. & Easton, G. (Eds.), Industrial networks: A new view of reality (pp. 205–214). London: Routledge. Karmel, S.M. (1997). The Chinese Military Hunt for Profit. Foreign Policy, 107: 102-113. Krug, B. (2012). Political Embeddedness in China. In: Capitalisms and Capitalism in the Twenty-First Century. Morgan, G. & Freeman, R. (eds) Oxford Scholarship online: sep-12. Li, H., & Zhang, Y. (2007). The role of managers’ political networking and functional experience in new ventures performance: Evidence from China’s transition economy. Strategic Management Journal. 28: 791-804. Linné, Å. (2012). China’s Creation of Biopharmaceutical Drugs: Combining Political Steering, Military Research and Transnational Networking. Doctoral Thesis, Department of Business Studies, Uppsala University. Liu, X. & White, S. (2001). Comparing innovation systems: a framework and application to China’s transitional context. Research Policy, 30: 1091-1114. Lundvall, B.-Å, K. J. Joseph, C. Chaminade and J. Vang (2009). Handbook of Innovation Systems and Developing Countries: Building Domestic Capabilities in a Global Setting, Cheltenham/Northampton: Edward Elgar. Lundvall, B-Å., Borrás, S. (2005) Science, Technology and Innovation Policy. In: Fagerberg, J. Mowery, D., Nelson, R. (eds). The Oxford Handbook of Innovation. Oxford: Oxford University press. Michaelson, E. (2007). Lawyers, Political Embeddedness, and Institutional Continuity in China’s Transition from Socialism. American Journal of Sociology, 113: 352-414. MOST (2006). http://www.most.gov.cn, viewed January 19, 2012. Naughton, B. (1996). Growing Out of the Plan: Chinese Economic Reform 1978-1993. Cambridge: Cambridge University Press. Okhmatovskiy, I. (2010). Implications of Ties to the Government and SOEs: A Political Embeddedness Perspective. Journal of Management Studies, 47: 1020-1047. Pavitt, K. (1991). Key Characteristics of the Large Innovating Firm. British Journal of Management, 2: 41-50. Peng, M. & Heat, P.S. (1996). The growth of the firm in planned economies in transition: Institutions, organizations, and strategic choice. Academy of Management Review, 21: 492–529. Peng, M. & Zhou, J. (2005). How network strategies and institutional transitions evolve in Asia. Asia Pacific Journal of Management, 22: 321-336.


Rosenberg, N. (1982). Inside the Black Box: Technology and Economics. Cambridge: Cambridge University Press. Schumpeter, J. (1934) The Theory of Economic Development, Cambridge, MA: Harvard University Press. Shih, T. (2009). Scrutinizing a Policy Ambition to Make Business Out of Science. Doctoral Thesis, Department of Business Studies, Uppsala University. Tidd, J., Bessant, J. & Pavitt, K. (2001). Managing Innovation. Chichester: Wiley. Walcott, S. & Xiao, W-B. (2000). High-tech parks and development zones in metropolitan Shanghai: From the industrial to the information age. Asian Geographer, 19: 157-179. Waluszewski, A. (2011). The Policy Practitioners Dilemma. The National Policy and the Transnational Networks. Vinnova Report VR 2011: 07. Wei, Y. D. & Leung, C-K. (2005). Development Zones, Foreign Investment, and Global City Formation in Shanghai. Growth and Change, 36: 16–40. Welch, C. & Wilkinson, I. (2004). The political embeddedness of international business networks. International Marketing Review, 21: 216 – 231.

Whitley, R. (2000). The Institutional Structering of Innovation Strategies: Business Systems, Firm Types and Patterns of Technical Change in Different Market Economies. Organization Studies, 21: 855-886. Van de Ven, A., Polley, H., Garud, D. & Venkataraman, S. (1999). The Innovation Journey. New York: Oxford University Press. Von Hippel, E. (1976). The Dominant Role of Users in the Scientific Instrument Innovation Process. Research Policy, 5: 212–239. Xin, C.R. & Pearce, J.L. (1996). Guanxi: connections as substitutes for formal institutional support. Academy of Management Journal, 39: 1641–1658. Yin, R. (1994). Case Study Research: Design and Methods. Thousand Oaks, CA, Sage.

Åse Linné, Researcher, Uppsala STS , Uppsala University, P.O. Box 513, SE-751 20, Uppsala, Sweden, email: [email protected] Tommy Tsung-Ying Shih, Senior Lecturer , Department of Business Studies, Lund University, P.O. Box 7080, SE-220 07, Lund, Sweden. Email: [email protected]


The Role of Tie Strength, Relational Capability and Trust in the International Performance of High Tech SMEs. Breda Kenny & John Fahy Cork Institute of Technology University of Limerick

ABSTRACT

This study identifies and examines the relationship between network characteristics and international performance of High Tech SMEs (HTSME) in the telecommunications industry in Ireland. The network characteristics construct for this paper comprises three dimensions: strength of ties, relational capability and trust. Empirical research was carried out using a mail survey in which 154 firms completed and returned the questionnaire. Five hypotheses were analysed using structural equations modelling using LISREL. The hypothesis stating that stronger ties are more influential on international performance than weak ties was supported. Strong ties and trust were positively associated with international performance, but non-significant. Weak ties and relational capability were negatively associated with international performance. Therefore, these findings indicate that the level of interdependence between the firms in this research is not prominent. The findings also imply that the effects of networks are contingent: they can present both strengths and constraints to firms.

KEY WORDS: Networks, Strong and Weak ties, Relational Capability, Trust, High Tech SMEs, International Performance

1. Introduction Reliance on networks and inter-organisational relationships has grown considerably in recent years, while partnerships with external actors have become a central strategy for many organizations in a wide range of industrial contexts (Gulati, 1995). Within the IMP perspective, progress has been made at identifying, describing, discussing and conceptualizing connected relationships. This body of work confirms that relationships and firms do not exist in isolation, but rather that they are connected to form a network of direct and indirect relationships (Holmen and Pedersen, 2003). These relationships form part of a firm’s network horizon and network concept. The network concept comprises all firms and relationships (direct and indirect) that a focal firm considers relevant (Anderson et al, 1994). The network horizon, according to Holmen and Pedersen (2003) comprises those other firms and relationships of which a focal firm is aware – whether or not it considers them relevant. The part of the overall network, which a focal firm is not aware of, is its environment. The part, which the focal firm is aware of, is its network horizon. The part of the network horizon that is considered relevant therefore, is the firm’s network context. The network horizon and context is relevant in this study as relationships are seen as important in the international environment that requires participation of both partners to perform activities jointly (Achrol et al. 1988). Furthermore, in the context of High Tech SMEs, relationships allow firms to cope with increasing technological dependence on others and the need to develop and tailor offerings. Each firms gains benefits and incurs costs from the network and from the investments and actions of all the parties involved (Håkansson

and Ford, 2002). A relationship between actors or partners in this context can be characterized in terms of the strength of their social ties, their level of trust, and the extent to which they share common processes and values (Kale et al., 2000; Cohen and Prusak, 2001). The IMP literature refers to nodes and threads in relation to connections between actors (Håkansson and Ford, 2002). The content of the threads is the result of investments by both actors. The development of the threads gives opportunities to both nodes, however the existence of the threads also impose restrictions on the actors. The strength of tie literature provides ample evidence of benefits that can be derived from both strong and weak ties. Kale et al (2000) argues that relational capital has important performance implications for partners. However, empirical research to support the bias that trust in particular, in international relationships enhances performance is limited and equivocal (Katsikeas et al. 2009). The primary objective of this study is to shed light on the question of whether tie strength, relational capability and trust actually matter when it comes to the international performance of High Tech SMEs? Networks allows firms to access foreign markets, therefore, the categorisation of strong and weak ties outlined in this paper will be operationalized by using mode of entry. Three dimensions of network characteristics are proposed in the model, namely tie strength, relational capability and trust. The strengthof-ties construct deals with the nature of the relational bond and the interdependencies between firms in the network. Strong and weak ties differ in terms of frequency of contact, resources committed and the social dimension of the relation. While a firm is likely to have a mix of strong and weak ties, Kale et al (2000)


argue it will benefit from a portfolio of ties favouring one type more than the other depending on the conditions surrounding the firm. The relationships examined in this paper range from low to high levels of interdependencies depending of the mode of foreign market entry used by the firm. The model proposed in this paper, argues that as strong ties are more beneficial in terms of execution and integration, they are more likely to lead to a higher level of international performance than weak ties. Relational capability is the ability to interact with other companies and refers to the degree of reciprocity and closeness among firms. Holmen and Pedersen (2003) further define this ability as being able to analyse and influence partners’ joining, relating and insulating functions. Issues such as mutual respect, social skills, communication skills (language and culture) and level of cooperativeness are covered under relational capability. Trust affects the depth and richness of exchange relationships and is an essential prerequisite for most forms of interdependent relationships (Moran 2005). This model proposes examining relational/ interpersonal trust as independent of other structural characteristics of the network. This was based on strong evidence in the literature to the importance of trust in achieving behavioural and market performance objectives in inter-organisational partnerships, especially in cross-border relationships where hierarchical control may not be a viable alternative. The quantitative analysis and the qualitative comments portray managers in High Tech SMEs that appear not to buy the network concept as outlined in this study. Specifically, stronger ties emerge as more influential on international performance than weak ties. Strong ties and trust were positively associated with international performance, but non-significant. Weak ties and relational capability were negatively associated with international performance. Therefore, it becomes clear in the conclusions that the level of interdependence between the firms in this research is not prominent. This finding is somewhat at odds with IMP literature, since IMP stress interdependencies rather than independence. That is, IMP does not stress that interdependence, relationships and networks are inherently good, but that they impact on firm activities. The rest of the paper is organised as follows: Firstly, a discussion on internationalisation and the HTSME is presented. Secondly, the derivation of the hypotheses is presented. Thirdly, the research method and the empirical results are outlined. The final section provides a discussion of these findings along with some limitations and future research directions.

2. Internationalisation and the high tech SME Although there is no single agreed definition of HTSMEs, these are generally characterised by small and medium-sized firms with advanced knowledge and capabilities in technology, an educated workforce, and the ability to adapt quickly to fast changing environments (Crick and Spence, 2005). These characteristics facilitate the internationalisation of HTSMEs which have been known to act quickly when windows of opportunity in foreign markets present themselves (Lindell and Karagozolu, 1997; Baldwin and Gellatly, 1998). SMEs within the high tech sector frequently operate within a narrowly defined market niche. Firms operating in, e.g. wireless data security cannot afford to target only a single (home) market. If the company is to take full advantage of the market potential this means simultaneous penetration to all markets (Saarenketo et al., 2004). The internationalisation process of small and specialized high technology firms is often different from that of more mature in-

dustries (Saarenketo et al., 2004). In dynamic high tech markets, one of the factors influencing high performance appears to be speed of internationalisation. Recent reports (Fan and Phan, 2007) show that these firms are growing and expanding their operations to other countries at a relatively faster pace than others. Consequently, HTSMEs may not necessarily have the time to integrate prior knowledge and fully develop their international strategies before implementing them as suggested by Johanson and Vahlne (1977). Instead, these companies need to react rapidly, develop mechanisms to assess opportunities quickly and allocate resources to take advantage of them. The results of these actions, some being previously labelled ‘reactive strategies’ have become the basis for survival in dynamic environments (Eisenhardt and Martin, 2000). Interpersonal and inter-organizational relationships are viewed as the media through which these firms can gain access to a variety of resources (Hoang and Antoncic, 2003). Furthermore, firms that pursue opportunities in foreign markets enabled by network resources may experience greater international growth that those that do not. Spence and Crick (2005) substantiate this argument by claiming that the internationalisation of HTSMEs is, in fact an inherently entrepreneurial act in itself, whereby firms seek out new potential resources and resource combinations in networks to exploit in foreign markets.

3. Hypotheses 3.1 Strength of Ties It is widely acknowledged that networks of different structural and relational characteristics have specific strengths, and hence a composition of network ties is required to support business development (Granovetter 1973; Johannisson 1988; Dubini and Aldrich 1991; Gargiulo and Benassi 1999). Håkansson and Ford’s (2002:134) work conceptualising the network as nodes and threads argue that the content of the threads or ties between actors in a network is the result of the investments by both counterparts in the relationship. The greater the investment the more substantial will be the content of the thread. The stronger the threads are- the more content there is and hence, the more important they are in giving life to the node, but also the more restrictions that are imposed on changes to the nodes or firms in the network. The strength of ties literature is primarily concerned with the nature of the relational bond between two or more social actors, as well as the effect this bond has on information sharing activities (Granovetter 1973; Frenzen and Nakamoto 1993; Uzzi 1997; Hansen 1999). Tie strength researchers typically classify the relationship between social actors as being linked by either a strong tie or a weak one (Rindfleisch and Moorman 2001). By treating strong and weak ties as separate constructs rather than degrees of one another, Rowley et al (2000) state that this captures richness in the data, which past researchers deem important in understanding network effects and firm behaviour. Rowley et al (2000) conceptualize strong and weak inter-organisational ties as separate constructs, different in kind rather than degree based on Contractor and Lorange’s (1988) original ordinal scale. They categorize equity alliances, joint ventures, and non-equity cooperative (R and D) ventures as strong ties, while defining marketing agreements, and licensing and patent agreements as weak ties, thereby capturing the strength of interfirm relationships on the basis of the partners’ typical levels of interaction in, and resource commitment to, each alliance type.


Capaldo (2007) builds on this previous strength-of-ties research at the inter-organisational level of analysis, wherein three major aspects of partnering behaviour have been advanced to express tie strength: the amount of time that characterizes the tie (Kraatz 1998), the partners’ level of resource commitment (Rowley et al. 2000), and the social contents which develop at both interpersonal and inter-organisational levels (Rindfleisch and Moorman 2001). This duration, frequency and intensity dimension, therefore, synthesizes the resource and social dimension of the tie strength. According to Håkansson and Waluszewski (2002), interaction within networks can refer to resources that are not only technical or physical, but also include resources that are social in origin, such as the skills and knowledge of individuals or groups. In fact, the relationships that form a network is a resource in its own right (Håkansson and Snehota, 1995: 27). These relationships comprise routines, information, memories, and expectations. Relationships also include intangible resources such as various kinds of knowledge (Baraldi et al, 2012: 124) As referred to earlier, networks allows firms to access foreign markets, therefore, the categorisation of strong and weak ties outlined above, will be extended to include entry modes. Internationalisation ‘mode’ refers to the organisational structure used to enter and penetrate a foreign market. Often, modes are organised according to the resource commitments they require and the level of control over international operations that the firm can afford (Johanson and Vahlne 1977). Internationalisation modes include: indirect exporting (i.e. via domestic intermediary); direct exporting; exporting via foreign intermediary; sales and/or manufacturing joint venture; sales and/or manufacturing subsidiary; and licensing and franchising (Calof and Beamish 1995; Petersen and Welch 2002). In terms of the firm’s commitment of resources, exporting modes are lower commitment modes, while foreign joint ventures and subsidiaries are higher commitment modes. Agndal and Chetty (2007) looked at changes in mode strategy where relationships were an important influence. Most of the mode changes in their research were gradual in terms of commitment of resources rather than leaps in forms of multiple steps at once, thus supporting Johanson and Vahlne (1977) that internationalisation occurs incrementally. As the firms gained more knowledge and experience in their international markets they often switched to a higher commitment mode, which was often a change from a distributor to a sales subsidiary (Agndal and Chetty 2007). This research addresses the concerns of Agndal and Chetty (2007) who feel that although some researchers have focused on the firm’s network positions and connections and how these affect internationalisation (Axelsson and Johanson 1992), mode selection has been neglected. One perspective on internationalisation focuses on organisational learning, which is based on Penrose’s (1959) ideas. For example, scholars such as Johanson and Vahlne (1977) focus on the issues of knowledge as a resource and mode selection. They argue that as firms become more experienced with conducting international activities, they become more willing to commit additional resources to these activities. The entry mode as a formal part of the internationalisation process, and indicative of the competitive stance of SMEs in international markets, would seem fundamental to a fuller understanding of international entrepreneurship according to Jones and Young (2009), who reviewed over 140 international entrepreneurship and found that over 80 failed to accommodate any discussion on the role of entry mode or mode of operation as a component of international venturing. Specifically in net-

work studies, they feel that entry modes tends to be neglected and underplayed as concern is focused on the development of relationships rather than the governance of business activities. This study explicitly addresses this gap in previous studies as it uses mode to entry to operationalize the tie strength construct when measuring the elements of network characteristics. In the literature, strong ties are shown to provide organisations with two primary advantages. First, strong ties are associated with the exchange of high-quality information and tacit knowledge. Uzzi (1996) observed in his study of the New York apparel industry that firms participating in strong ties were able to exchange fine-grained knowledge. In the development of strong ties, inter-firm partners learn about each other’s organisation, become more dependent on one another and develop relational trust (Larson 1992). Based on a deeper understanding of a partner’s operations, tacit knowledge is more readily transferred across organisational boundaries, which are blurred by close contact (Hagg and Johanson 1983). Second, strong ties serve as part of the social control mechanism, which governs partnership behaviours. Firms enter strategic alliances with competitors to gain access to external resources, share risks and cost, or pool complementary skills (Hagg and Johanson 1983; Kogut 1988; Hagedoorn 1993). Larson (1992) shows that strong ties incrementally promote and, in turn enhance, trust, mutual gain, reciprocity, and a long-term perspective. Consequently, partners are more likely to forego individual short-term interests, exercise voice (rather than exit), and develop joint problem-solving arrangements (Powell 1990; Uzzi 1996). Strong ties produce and are governed by relational trust and norms of mutual gain and reciprocity, which grow through a history of interactions (Powell 1990; Larson 1992). Similar to Powell’s (1990) assertion that networks represent a separate and distinct organisational form, Uzzi (1996) refers to this alternative governance system based on trust as the logic of embeddedness, and argues that it is the product of cohesive/intense ties. These strong tie benefits are different from the advantages gained through weak ties. Granovetter (1973) argues that weak ties are conduits across which an actor can access novel information. Weak ties are more likely than strong ties to be ‘local bridges’ to distant others possessing unique information. The strength of weak ties argument is as much about structural embeddedness as it is about relational embeddedness. A weak tie can be beneficial, because it is more likely to embed an actor in (or provide access to) divergent regions of the network rather than to a densely connected set of actors. For example, according to Granovetter’s (1973) argument, an actor’s collection of weak ties is more likely to be a sparse structure reaching divergent regions of the surrounding network. The substantial support for the benefits derived from both strong and weak ties suggests that neither type is unconditionally preferred. Indeed, strong and weak ties have different qualities, which are advantageous for different purposes. Tiwana (2007) found that weak ties provide innovation (exploration) potential for firms, but lack integration (exploitation) capacity, and strong ties provide integration capacity but lack innovation capacity. In the context of international trade, it can be argued that strong ties are more beneficial than weak ties since they allow for greater volume of resources to move between actors (Podolny 2001), have greater motivation to be of assistance and are typically more easily available (Granovetter 1983), more willing to take the time to carefully explain, detail, or listen to novel or complex ideas (Granovetter 1985; Uzzi 1996; Hansen 1999; Moran 2005), and ultimately, as strong ties are more


beneficial in terms of execution and integration, they are more likely to lead to performance related outcomes, such as contracts signed, sales and market share attainment. Furthermore, it has been widely accepted that resources being tacit in nature cannot easily be transferred by arm’s length transactions (Kogut and Zander 1992). Especially internationally dispersed intangible resources are difficult to access by arm’s-length transactions (Zander 1999), thereby calling on the use of closer, stronger ties between firms. Therefore, it can be argued that a portfolio of strong and weak ties have benefits for international trade, also it is possible to argue that as a firm develops stronger ties with other partners, they are more likely to commit more resources and have a higher level of commitment, which in turn can lead to exploiting more opportunities for international trade. Thus: Hypothesis 1: There is a positive relation between strong ties and international performance Hypothesis 2: There is a positive relation between weak ties and international performance Hypothesis 3: The relation with international performance is stronger in strong ties than in weak ties.

3.2 Relational Capability Relationships are not only a necessity for firms to transfer knowledge and capabilities and to co-ordinate the transfer process (Johanson and Mattsson 1987; Forsgren 1990), for instance by reducing uncertainty (Sölvell and Birkinshaw 1999); they are also means to create new knowledge and capabilities (Hallen et al. 1991; Håkansson and Snehota 1997). It has been stated, though, that any focal organisational unit will maintain close, intense, and frequent relationships only to a limited number of network partners within its business network (Holm et al. 1996; Håkansson and Snehota 1995; Forsgren 2002). Units that actually maintain close, intense, and frequent relationships are thereby considered as being embedded in their business network (Andersson et al. 2001). It is assumed that the closer the relation and the higher the number of close relationships, the higher is the unit’s degree of embeddedness within its business network. The possibility to assimilate new knowledge and to modify or generate capabilities is stated to be positively related to the degree of embeddedness of the focal unit within its business network (Andersson et al. 2001). Relational embeddedness, which is an essential dimension of a strong tie, refers to the degree of reciprocity and closeness among firms. Networks that are characterized by high relational embeddedness are networks of organisations that have strong socializing relationships and share similar attitudes and behavioural norms. Firms within such highly cohesive networks tend to be active in communication processes and thus share more common information and same understandings. Past research indicates that a high level of relational embeddedness in network relationships can enhance the level of access and transfer of finegrained information and, more importantly, tacit knowledge and know-how among firms within the network (Gulati 1998; Hansen 1999; Lorenzoni and Lipparini 1999). When firms are close to one another, they tend to develop interaction routines with more frequency and intensity, resulting in more willingness in information sharing and greater ability of firms to absorb and act on the new information and knowledge in a timely manner (Dyer and Singh 1998; Hansen 1999). Furthermore, Kapasuwan (2006) found a positive relation between relational embeddedness, organisational learning and international performance. Past

research has operationalised relational embeddedness (Bonner et al. 2005; Grundlach et al. 1995; Heide and John 1992; Rindfleisch and Moorman 2001), relational skills (Walter et al. 2006), relational competence (Loxton and Weerewardena 2006), relational capital (Badaracco 1991; Inkpen 1994; Mohr and Spekman 1994; Madhok 1995; Gulati 1995; Dyer, 1996; Dyer and Singh 1998; Kale et al. 2000) using similar constructs. Relational skills, also referred to as social competence (Baron and Markman 2003) includes such aspects as communication ability, extraversion, conflict management skills, empathy, emotional stability, self-reflection, sense of justice, and cooperativeness (Browne 1996; Tushman and Nader 1996; Foray 1997; Marshall et al. 2003; Ritter and Gemünden 2003). Social qualifications in a cross cultural setting are of special interest, skills such as cultural awareness and foreign language competency are important for interpersonal interaction in the international trade arena (Kenny and Sheikh 2000). Kale et al (2000) refers to mutual trust, respect and friendship that reside at the individual level between alliance partners as relational capital. Trust will be dealt with separately in the next section. Furthermore, Kale et al (2000) argues that relational capital has important performance implications for alliance partners. Lorerenzo and Lipparini (1999) regard ‘relational capability’ as the capability to interact with other companies, a capability that is based on absorption, combination and coordination. According to Holmen and Pedersen (2003: 411), the effectiveness of the exchange between two companies relates to how well it dynamically acts and reacts in a network context depicted by change and stability. Håkansson and Snehota (1989: 530) go further by stating that the efficacy of companies performance in a network is contingent not only on how well the focal firm performs in interaction with its direct counterparts, but also on how these counterparts in turn manage their relations with third parties. This study will focus on relational capability, which essentially is a measure of the quality of the relationship and is an amalgam of each of the terms mentioned above. The background literature consequently leads to the development of the following hypothesis: Hypothesis 4: The higher the level of relational capability of a firm within the network the greater the impact on International performance

3.3 Trust The assimilation of knowledge and the generation of critical capabilities require intense, close, and frequent relationships because knowledge and capabilities are intangible resources characterised by a high degree of tacitness. The transfer of tacit resources is only feasible in an atmosphere of trust between the entities involved in the transfer process (Grabher 1993). Thus, the transfer of tacit resources has to be interpreted as a social phenomenon rather than a market transaction (Tyre and Von Hippel 1997). Trust between partners is often cited as a critical element of network exchange that in turn enhances the quality of the resource flows (Larson 1992; Lorenzoni and Lipparini 1999). Other scholars have also defined network governance by the reliance on ‘implicit and open-ended contracts’ that are supported by social mechanisms, such as power and influence (Thorelli 1986) and the threat of ostracism and loss of reputation (Portes and Sensenbrenner 1993; Jones and George 1998) rather than legal enforcement.


A number of scholars have asserted that these distinctive elements of network governance can create cost advantages in comparison to coordination through market or bureaucratic mechanisms (Thorelli 1986; Jarillo 1988; Jones and George 1998; Lipparini and Lorenzoni 1999). In particular, mutual trust as a governance mechanism is based on the belief in the other partner’s reliability in terms of fulfilment of obligation in an exchange (Pruitt 1981). Trust allows both parties to assume that each will take actions that are predictable and mutually acceptable (Powell 1990; Uzzi 1997; Das and Teng 2000). These expectations reduce transaction costs—for example, monitoring and renegotiating the exchange in reaction to environmental changes—particularly in highly complex tasks facing strong time constraints (Jones and George 1998). The presence of inter-firm trust is an extraordinary lubricant for alliances that involve considerable interdependence and task coordination between partners, firms with prior network connections are likely to have greater awareness of the rules, routines, and procedures that each needs to follow (Gulati et al. 2000). Trust also affects the depth and richness of exchange relationships, particularly with respect to the exchange of information (Saxenian 1990; Lorenzoni and Lipparini 1999; Hite 2003). For example, a qualitative study of vertical relationships involving the purchase and supply of goods or services between networked firms revealed that the nature of the information exchange extends far beyond a discussion of price and quantity. Uzzi (1997) found that information exchange between clothing manufacturers and their ‘embedded’ small suppliers tended to be more holistic in nature. Because of its positive impact on information flows, trusting behaviour is cited as a critical factor in enhancing innovation through inter-firm collaboration (Hausler et al. 1994) and an integral reason for inter-firm networks’ longevity (Saxenian 1990; Lipparini and Lorenzoni 1999). Essentially, trust can be viewed as the basic active ingredient of social capital, the condition that allows an actor to reliably expect to obtain and use the resources made available through one’s contacts (Ring and Van de Ven 1994; McAllister 1995; Nahapiet and Ghoshal 1998). The focus here is on relational or interpersonal trust (McAllister 1995; Rousseau et al. 1998). Such trust is constructed through personal interactions and experiences with the other party. Conditions for this form of trust include the assessed integrity of the contact, their competence in ongoing exchanges, and their predictability through the alignment of goals and values (Butler 1991; Hosmer 1995; Rowley et al. 2000). Turning towards the contributions this research makes to the trust literature, Zaheer et al (1998, p. 141) note: “considerable ambiguity is evident in the literature about the precise role of trust as it operates at different levels of analysis and its influence on performance.” As mentioned earlier, this study measured trust independent of structural characteristics of the network. In the international context, there is evidence that the choice of foreign partner is often mandated by the host government, or that firms do not choose optimal partner firms due to the information asymmetries about long-term partner objectives during the initiation stage. Aulakh et al (1996) points out – little systematic research attention has been given to identifying the determinants of inter-organisational trust. Katsikeas et al (2009) notes that of particular interest are the findings that external uncertainty is not directly related to trust, but enhances a party’s opportunistic inclinations. One line of speculation for the lack of a direct link between external uncertainty and trust pertains to the adaptation problem created by turbulent environmental condi-

tions (Rindfleisch and Heide 1997). Environmental uncertainty can limit decision-makers’ predictive abilities, make elaborate contracts difficult and costly, and render even the most detailed agreements inadequate. In an attempt to adjust more readily to changing conditions surrounding international exchanges, importing firms may opt for developing relational norms that promote actions toward relation preservation (Heide and John 1992) and thus facilitate trusting behaviours. Although external uncertainty could be argued to undermine trust, one might also suggest that, from a normative perspective, the inherent need for trust is greater under high levels of external uncertainty. However, Becerra et al (2008) suggest that risk or uncertainty does not necessarily need to be present for trust to exist or to be meaningful. Strategically, firms may find it prudent to work jointly with their foreign suppliers on contingency plans covering those cases where environmental changes seem potentially imminent. When circumstances change radically (i.e., outside the boundaries of existing agreements), there may be ambiguity about how to actually define an opportunistic action. It may be possible that fluctuations in the environment affect the parties’ understanding of what constitutes opportunism in the first place? As Zaheer and Zaheer (2006) note, there is still only the barest appreciation of the role of trust in cross-border relationships. Those studies where trust has been explicitly considered in social capital research concern redundant, cohesive networks, where the visibility of actions places enormous sanctions on opportunistic behaviour and thus engenders a form of calculated trust (Coleman 1988). What has been considered, in other words, is network structure (i.e., closure) as a substitute for trust and not the trust associated with interpersonal relationships. Trust, then, is often left unmeasured or else its presence is assumed to be associated with a certain structural form (Moran 2005), such as strong ties or relational embeddedness. One exception to this was Wincent (2005) who measured trust in the context of networking width and depth inside the SME network and found trust to be related to corporate entrepreneurship. To the extent that trust is an important element and is engendered through interpersonal experiences (Granovetter 1985; Uzzi 1996; Rowley et al. 2000), Moran (2005) contends that it is important to measure it and determine its value, independent of structural characteristics of the network. Besides the role of trust as a behavioural deterrent of opportunistic behaviour and as an alternative to ownership control (Aulakh et al. 1996), there is also evidence that building trust in inter-organisational partnerships has important market performance and efficiency implications (Parkhe 1993). Therefore, it can be argued that in a network, firms that trust their partners are more likely to engage, combine resources, and trade together to enhance performance in international markets. Hence: Hypothesis 5: The higher the level of trust between partners in a network the greater the impact on international performance.

4. Measures The scales used in this study were sourced from the literature, and in some cases were modified for the current research context.

4.1 Strength of ties

Networks allow firms to access foreign markets, therefore, strong and weak ties are measured through foreign market entry modes. Internationalisation ‘mode’ refers to the organisational structure used to enter and penetrate a foreign market. Often, modes are organised according to the resource commitments


they require and the level of control over international operations that the firm can afford (Johanson and Vahlne 1977). Internationalisation modes include: indirect exporting (such as, via domestic intermediary); direct exporting; exporting via foreign intermediary; sales and/or manufacturing joint venture; sales and/or manufacturing subsidiary; and licensing and franchising (Calof and Beamish 1995). In terms of the firm’s commitment of resources, exporting modes are lower commitment modes and treated as weak ties, while foreign joint ventures and subsidiaries are higher commitment modes and are considered strong ties.

4.2 Relational Capability Relational capability consists of a twelve item scale measuring the quality of the relationships within the network. The relational embeddedness dimension is measured by one item from Bonner et al’s (2005) scale and two items from Rindfleisch and Moorman (2001). Social competence in a network setting is measured by two items from Walter et al’s (2006) scale on relational skills, one item on level of interaction between partners from Kale et al’s (2000) relational capital scale, two items from Loxton and Weerewardena, (2006) relational competence scale and three items from Ritter and Gemünden’s (2003) social scale. Social competence in this study follows Baron and Markman (2003) and included such aspects as communication ability, extraversion, conflict management skills, empathy, emotional stability, self reflection, sense of justice, and cooperativeness (Tushman and Nader 1996; Browne 1996; Foray 1997; Marshall et al. 2003). Social competence in a cross cultural setting are of special interest, skills such as cultural awareness and foreign language competency are important for interpersonal interaction in the international trade arena (Kenny and Sheikh 2000) and an additional item is included to capture this.

4.3 Trust Three items developed from Sividas and Dwyer (2000) captured a firm’s trust in its cooperative partners inside the SME network. Three additional items from Moran (2005) captured the dimension of relational trust and included the perception of honesty

and truthfulness in exchange, perceptions of competence in ongoing interactions and alignment of goals and values.

4.4 International market performance Three dimensions have been identified to capture the firm’s level of international market performance. These dimensions are based on the company’s marketplace performance (Jaworski and Kohli, 1993), financial performance (Narver and Slater, 1990), and levels of customer satisfaction (Walter et al., 2006). The first two dimensions relate to a more objective analysis of performance and are based on marketplace indicators (i.e. sales growth over the past three years and the market share of the firm’s number one product) and financial indicators (i.e. average return of investment, revenue and pre-tax profitability). For customer satisfaction, respondents were asked to consider the extent to which they felt their firm had varying levels of customer satisfaction and retention.

5. Method This study adopted Dillman’s (2007) Tailored Design Method (TDM), which asserts that survey response can be explained in terms of the theory of social exchange. According to Fahy (2001) the appeal of TDM is that it provides the researcher with a comprehensive set of theoretically based and empirically tested guidelines for survey design, questionnaire construction and questionnaire implementation. SMEs are the focus of this study as they are a key economic sector in Ireland, where they constitute 97% of enterprises and contribute to the flexibility and resilience of the economy as well being active in international markets (SBA, 2008). This study draws on research from HTSMEs, in the telecommunications and internet services sectors in Ireland. This industry was selected as it is considered a global industry with a complex value chain. It is also an industry with high levels of inter-firm network and export activity. For the current study, the population comprised of all compa-

Table 1 Means, Standard deviations and correlations


Table 2: CFA and Constructs Reliability

nies in the telecommunications, internet and related industries. In order to compile a relevant sampling frame, data from the Irish Central Statistics Office (CSO) and The Commission for Communications Regulation of Ireland (ComReg), Business and Finance and Dunn and Bradstreet were used. The focus was on a single industry in one country to control for industry- and country-specific factors affecting international performance.

In relation to time, this study is cross sectional and the survey was carried out over a six week period in April/May 2008. This study was based on a mail survey of 458 SMEs (with more than 3 and less than 250 employees) drawn from this population. The questionnaire underwent multiple pre-tests. Whenever possible, multiple-item measures were used to minimize measurement error and to enhance the content coverage for constructs. State-


Table 3: Items removed during Purification Process

ment-style items were measured on seven-point Likert-scales. The overall response rate for this study was 40.39% with a useable response rate of 33.64%. The specific activities carried out by respondent firms include the following: computer consultancy, computer services, computing and bureau services, data communications, Internet services and Web design, telecommunications and telephone cost management. A missing data process in accordance with Hair et al. (2006) is not deemed an issue in this study as only one questionnaire was returned incomplete. However, further analysis of missing data was performed using Prelis 2.80 to impute individual missing values using the estimated means algorithm following Du Toit and Du Toit (2001). Using a t-test, early and late respondents were compared on several key characteristics such as importance of relationships, percentage of revenue derived from international markets, importance of international markets to overall performance and number of years exporting. No significant difference was found at the 0.05 level. Thus, based on these results and considering that the response rate was relatively high, it was concluded that non-response bias is not a significant problem. Furthermore, size and location differences between respondents and non-respondents revealed no significant differences between the sample and the population under investigation. Harman’s single factor test was performed to test for the presence of common method variance bias (Harman, 1967; Podsakoff et al., 2003; Chang et al., 2010). All variables were entered into an unrotated principal components analysis. The results of the analysis indicated nineteen items with eigenvalues greater than 1 and no single factor accounted for more than 33.7% of the covariation. Only one variable accounted for 18% of the variance. The results indicate that common method variance, though probably present to some degree, does not affect the results in this study. Although this study collected mainly quantitative data through the questions and scales used, respondents also had the opportunity to provide additional qualitative comments at the end of the questionnaire. Some of the quotations from the analysis of this data are used later in the discussion section of this paper.

5.1 Scale Validation Table 1 provides the descriptive statistics and correlations for the composite variables used in this study. The measures of strong

and weak ties, relational capability and trust were positively correlated with the measures of international performance with correlations ranging from 0.091 to 0.832. Table 2 displays the results obtained from the estimation of the CFA model. An inspection of these results shows that all items loaded on their specified constructs. Convergent validity is evidenced by the large and significant (t