Abstract
Regional development policies to foster innovation and competitiveness have evolved towards a ‘soft’ focus on facilitating relationships of cooperation. This is demonstrated by the popularity of network and cluster policies. However, the development of these policies poses particular challenges since there is insufficient understanding of the factors in the social structure that underpin networking behaviour and network outcomes. The analysis of this social context provides an important base for policy learning and therefore for the development of networking policies. The paper makes both a theoretical contribution (in establishing the framework) and a methodological contribution (in exploring its implementation in an ongoing policy process). The case studied is that of the Basque aeronautics cluster, a medium-sized cluster with 35 members founded in 1997 within the Basque Country (Spain) cluster policy, one of the longest-running cluster policies in Europe. The participatory design carried out in the application of the theoretical framework to the case study enabled a deeper appreciation of the different realities and behaviour of targeted firms and supported strategies to improve policy effectiveness.
Introduction
Regional development policies designed to support innovation and competitiveness have evolved a focus on facilitating relationships of cooperation between firms and other agents in the system. This is in line with much contemporary academic analysis of territorial economic development. For example, it is now widely acknowledged that innovation takes place in a systemic context (Freeman, 1987; Lundvall, 1992; Nelson, 1993; Cooke et al., 1998). This provides a rationale for supporting relationships between firms and other agents alongside policies designed to change their internal behaviour (for example in R&D investment). Simultaneously there has been a resurgence of literature dating back to the work of Alfred Marshall (Marshall, 1907, 1919) on the benefits of conscious and unconscious relationships between geographically proximate firms. These relationships have been analysed from a range of different perspectives and contexts (Piore and Sabel, 1984; Becattini, 1991; Saxenian, 1994; Schmitz, 1995), but it is the cluster concept (Porter, 1990, 1998) that has been particularly influential among policy makers. Indeed, ‘cluster policies’ are widely implemented today in all corners of the world, despite varied criticism of their theoretical and empirical basis (Benneworth and Charles, 2001; Martin and Sunley, 2003; Lorenzen, 2005; Belussi, 2006; Pitelis et al., 2006).
Cluster policies, and more generally network policies, are examples of what can be termed ‘soft policies’ (Harrison and Rodríguez-Clare, 2010). Rather than dealing in ‘hard’ subsidies for specific production- or innovation-related activities, they focus support on fostering cooperative relationships between agents. The design, implementation and evaluation of these policies is particularly challenging. It is difficult to delimit their often intangible impacts, and overlapping policies are frequently found in different government departments and at different geographical scales. 1 Most evaluation attempts focus on observable networking behaviour stimulated by the policy (cooperative ties, joint projects, membership of associations), often seeking to establish a relationship with performance outcomes (productivity, employment, innovation). However, there is far less understanding around the relationship between networking behaviour and those factors in the social structure that underlie network cooperation. This is an important gap because networking policies are always implemented in the context of a specific social structure, which is likely to interact with the policy in terms of inducing behaviour, and in turn impact on performance outcomes. 2 In particular, the degree of social capital among agents underlies the effective development of networking, and is also impacted upon over time by that networking. Understanding these dynamics therefore offers important opportunities for learning among the agents involved, potentially leading to enhanced effectiveness of the networking policy and other related policies (Clarysse et al., 2009).
Fostering such policy learning requires a move away from traditionally static and ‘ex post’ evaluation approaches, towards the development of new processes and indicators. In this paper we propose a framework that sheds light on the social capital foundations for networking behaviour and whose application in a specific case has generated policy learning outcomes. As such, the paper makes both a theoretical contribution (in establishing the framework) and a methodological contribution (in exploring its implementation in an ongoing policy process). The framework was developed in the context of a participatory evaluation process involving stakeholders – government, cluster association, firms – in the Basque aeronautics cluster. As part of a wider project for evaluating the results and impacts of networking under the cluster policy, a new set of indicators were developed for gauging the social capital among agents that underlies the behavioural changes promoted by the policy. The participatory design of the process and the type of indicators collected have enabled a deeper appreciation of the different realities and behaviour of targeted firms, supporting the development of strategies to improve policy effectiveness. Reflections on the case highlight the advantages of integrating such processes in similar ‘soft’ competitiveness policies.
The paper is structured as follows. First, we provide a brief discussion of regional competitiveness policies, identifying the need for policy learning. Second, we argue for the centrality of social capital as a contextual foundation for the complex interactions and impacts of these policies, setting out key features of a social capital framework to network policy learning. Third, an application to the case of the Basque aeronautics cluster is presented and, finally, the results of this process are discussed in terms of policy learning outcomes and main conclusions.
Regional competitiveness policies: the need for policy learning
Two broad arguments are made in the literature to justify policy intervention related to innovation and competitiveness (Laranja et al., 2009). First is the neoclassical approach, in which policy responds to market failures. Second is the evolutionary approach, where government intervention is justified by wider system failures. It is in this second approach that we find justification for most contemporary network policies (Smith, 2000; Laranja et al., 2009).
The concept of ‘additionality’ is frequently invoked to justify policy interventions in terms of their complementary effects. Related primarily to market failures in a neoclassical framework are ‘input additionality’, where policy ensures additional resources to those of beneficiaries, and ‘output additionality’, where policy results in the generation of additional outputs from beneficiaries’ activities (Georghiou, 2002; Clarysse et al., 2009). ‘Behavioural additionality’, on the other hand, refers to situations where policy supports changes in routines and processes, either within firms or in terms of their interactions in the system.
It is this latter category that best characterizes ‘soft’ policies for competitiveness and innovation, among which there is a strong convergence given the centrality of innovation for competitiveness. Innovation processes have historically been analysed as linear: more inputs (for example, expenditure in R&D) produce more outputs, (for example, patents or new products), implying high relevance of input and output additionalities. However, a systemic approach has become recognized as increasingly important, whereby innovation is interpreted not only as a technical process but also as a social/interactive learning process developed across firms and between them and their environment (Lundvall, 1992). This has contributed to changes in innovation and competitiveness policies, which have evolved from focusing on improving elements of the system (for example, firms, universities, technology centres) to also upgrading the interactions among these elements (that is, network failures) and the interactions with other systems (that is, lock-in failures) (Parrilli et al., 2010).
Linked to the growth in network policies aiming to generate behavioural changes, the significance of context has become a critical aspect (Nauwelaers and Wintjes, 2008). As noted by Sanderson (2002), ‘there are social phenomena independent of cognition to be explained in terms of underlying mechanisms (which may not be directly observable)’ (Parrilli et al., 2010:8), and ‘the task of social science is to understand the way in which mechanisms work in conjunction with contextual factors to generate social outcomes’ (Sanderson, 2002:8).
A tailored portfolio of instruments is required in each context, implying the need for greater contextual understanding and a more interactive mode of policy formulation and implementation, with significant space for learning (Nauwelaers and Wintjes, 2002; Parrilli et al., 2010). Correspondingly, there is a strong rationale for more interactive and integrated approaches to policy evaluation that themselves reinforce desired behavioural changes targeted by the policies (Diez, 2001).
Above all, cluster and network policies can be seen as effective when they address the system failures that they are designed to tackle, stimulating behavioural changes that in turn have an impact on competitiveness outcomes. However, the systemic context in which the policies are implemented also interacts with the changes that they are designed to stimulate. In particular, the already existing social structure provides a foundation that both conditions networking behaviour and is changed by it during the course of the policy. Most evaluation attempts focus ex post on behavioural changes stimulated by the policy (cooperative ties, joint projects, membership of associations) and/or on associated performance outcomes (productivity, employment, innovation) (Huggins, 2001; Martin et al., 2008a, 2008b; Aranguren et al., 2010b; Bellandi and Caloffi, 2010; Tomlinson, 2010). There is much less analysis of the real-time interactions of the policies with the underlying social structure, despite these providing critical foundations for the immediate and ongoing success of these policies.
In response to this gap, we propose an analytical framework to explore the underlying social capital foundations of network policies, in order to facilitate deep-seated policy learning around their interactions with the policy. The analysis is based on the following assumptions. First, the social capital underlying a network plays a key role in network policy development (Maskell, 2001; Barrutia and Echebarria, 2010). Second, the social capital underlying a network is a collective asset developed by the behaviour of all the agents participating in the network (Woolcock and Narayan, 2000; Gedikli, 2009). As such, social capital can be seen as a foundation in which network policies are implanted, but not a static foundation. Rather, social capital will interact with the policy and will evolve during its course, meaning that it can also be interpreted as a ‘result’ of the policy.
Social capital foundations for ‘soft’ policies
Different strands in the literature support the argument that network relationships among co-located firms play a key role in supporting innovation and competitiveness (Porter, 1998, 2000; Schmitz and Nadvi 1999; Becattini et al., 2003; Eisingerich et al., 2010). Raines (2003), for example, states that cluster dynamics emerge from the links between individual business behaviour, collective action and wider economic impacts. Changes in company performance are attributable to cluster promotion only when actually caused by the network membership; thus, looking at corporate figures or linkage patterns does not clearly reveal impact. Indeed, the impact of cluster initiatives on the competitiveness and performance of firms emerges theoretically from collective dynamics and relationships that help to source complementary assets, innovation or learning. Thus ‘soft’ qualities such as the atmosphere of the cluster are important intangible assets that facilitate socially embedded learning and trust among cluster members, encouraging joint activities (Fromhold-Eisebith and Eisebith, 2008). In this sense, network policies seek to develop a social infrastructure (Flora et al., 1997) defined in terms of the presence or creation of social capital, and there have been several recent empirical studies analysing the effects of social capital development on cluster and network success from different perspectives (Casper, 2007; Gilsing et al., 2008; Krätke, 2010; Eisingerich et al., 2010).
Social capital increases the efficiency of action (Nahapied and Ghoshal, 1998) and of information diffusion (Burt, 1992), reduces the costs of monitoring processes and transactions, and encourages the cooperative behaviour necessary for innovation and value creation (Fukuyama, 1995). Yet, there is little consensus around what should be included in social capital (Inkeles, 2000). The concept brings together under one label very different elements that have been given different emphasis depending on the perspective and focus of each author (Adler and Kwon, 2002; Lorenzen, 2007). Thus social capital is not a single entity but a variety of different entities with two characteristics in common: some aspects of social structure; and the capacity to facilitate certain actions of agents within that structure (Coleman, 1988).
Given our interest in social capital as an underlying foundation for networking policies, we locate ourselves in the bonding view of social capital (Putnam, 2000; Adler and Kwon, 2002; Cooke et al., 2005), taking in linkages among individuals in a system and the features that give cohesiveness and facilitate the achievement of collective goals. We employ a correspondingly broad concept of social capital that accounts for all the elements that constitute a resource for achieving cooperative outcomes. Thus the social capital underlying a network is defined as a set of actual and potential resources available in the structure of relationships among the involved agents (firms, government and other agents in the system). However, given the multidimensional nature of the concept (Putnam, 1995; Nahapied and Ghoshal, 1998), the development of a framework should also account for different dimensions.
Several authors distinguish two main dimensions (Coleman, 1988; Putnam, 2000; Stone, 2001; Adler and Kwon, 2002): social structures or networks; and the norms or values governing those social structures. Others, however, differentiate three dimensions (Nahapied and Ghoshal, 1998). A structural dimension refers to the overall pattern of connections between actors. A relational dimension corresponds to behavioural assets such as values, norms and expectations, created through a history of interactions. Finally, a cognitive dimension refers to those resources providing shared interpretations, vision, language and codes among parties. This third dimension is very relevant given the focus of many network policies on developing common projects and vision among agents, and it is also used to analyse social capital in intra-firm networks (Tsai and Ghoshal, 1998). We therefore propose a social capital framework in which these three dimensions are reinforced dynamically among themselves through the interactions of the network agents and will also inevitably interact with the implementation and behavioural outcomes of any network policy. The three dimensions of social capital are now considered in greater detail before presenting the overall framework.
Relational dimension
For a social link to be established, actors must have something in common – some norms, values or preferences, or some minimum degree of mutual trust (Maskell and Lorenzen, 2004). There is broad agreement that trust in particular is critical for the development of social structures, providing a remarkably efficient lubricant to economic exchange and cooperation (Sako and Helper, 1998; Woolcock and Narayan, 2000; Huggins, 2001; Maskell and Lorenzen, 2004; Cooke et al., 2005). Social relations rather than institutional arrangements or generalized morality are mainly responsible for the production of trust in economic life (Granovetter, 1985). Indeed, trust is defined as an expectation held by an agent that a partner will behave in a mutually acceptable manner (Sako and Helper, 1998).
Within the relational dimension, social capital is also promoted by norms of generalized reciprocity (Adler and Kwon, 2002). Reciprocity can be defined as ‘a series of bi-lateral transfers independent from each other and, at the same time, interconnected’ (Kolm, 1994: 68). 3 It therefore requires the existence of trust but implies a broader range of trustful behaviour, expanded over a longer period of time (Sako and Helper, 1998). This serves to solve problems of collective action and transforms individuals from self-seeking with little sense of obligation into members of a community with a shared interest.
Finally, the notion of ‘unity’ has been analysed alongside those of trust and reciprocity (Western et al., 2005). This is defined as the feeling of belonging to a network. In the framework presented in Figure 1, we prefer the term ‘commitment’. Organizational commitment in general has been widely studied and it refers to the feeling of attachment to the goals and values of the organization (Cook and Wall, 1980).
Social capital framework for network policy learning.
Structural dimension
Stronger actual ties and relationships have an impact on the values of the network, in particular by increasing expectations from and belief in cooperation. This in turn augments the possibility of cooperative outcomes. The organization’s network is defined as its set of relations, both horizontal and vertical, with other actors that are of strategic significance (Gulati et al., 2000). Thus the network’s structural characteristics arguably constitute the building blocks of competitive advantage, and the most important facets of this dimension are the presence or absence of network ties between actors (Scott, 1991; Wasserman and Faust, 1994). The network configuration then describes the pattern of linkages in terms of density, connectivity, hierarchy, and so on (Granovetter, 1973; Uzzi, 1996, 1997). Along with these characteristics of the actual relationship structure – actual ties and network configuration – we also consider desired ties as a relevant element. This is included because the complementarity of actors is an essential precondition for the development of actual relationships. Thus, declared willingness to cooperate with other agents is an indicator of the future evolution of the existing relationship structure.
Cognitive dimension
In the cognitive dimension, the shared beliefs of a community are considered to be a relevant element of social capital (Nahapied and Ghoshal, 1998). Indeed, given that network policies typically pursue the enhancement of competitiveness through cooperation, the existence of shared beliefs related to cooperation heavily conditions the evolution of the network. Along with a common vision of the benefits of the network policy, a shared vision of the network itself is important. We highlight in particular the proximity of the goals of different agents and the willingness and ability of actors to identify and share collective goals. Moreover, linked to this shared vision of the policy and network, expectations are essential for relationships to prosper. Actors are driven by instrumental motivations that incentivize them to cultivate social capital in order to obtain certain advantages (Adler and Kwon, 2002). In this sense, expectations of the efficiency of cooperation and of the network as a means to enhance competitiveness are highly relevant elements in motivating the actors to interact.
In summary, since network policies are conditioned by, and interact with, the existing underlying social capital, an understanding of this social capital in all its dimensions (relational, structural and cognitive) is a valuable departure point for reflection among the different stakeholders involved in a network policy. It can provide the basis for policy learning and continual improvement in network dynamics and performance. However, such a reflection requires new indicators that can capture and then track over time both the underlying social capital (in each of its dimensions) associated with the network and the specific outcomes that the network policy aims to achieve. In this paper we develop these measurements in the context of the case of the Basque aeronautics cluster and its corresponding policy, distinguishing between social capital indicators (SC) and network policy outcomes (NPO).
Social capital indicators measure the existing social capital among cluster association members in the three aforementioned dimensions of social capital, with data collected from a firm questionnaire. In the relational dimension, validated existing scales for measuring trust (Sako and Helper, 1998), reciprocity (Krishna and Shrader, 1999) and commitment (Cook and Wall, 1980) have been adapted to the case. The cognitive dimension is measured using specific questionnaire items that capture the agents’ perceptions of the commonalities of goals in the network and reasons for belonging to the network. Four items refer to a shared vision of policy goals and two to a shared vision of the cluster association. Finally, in the structural dimension we measure the network itself, in particular the actual and desired relationships declared by each cluster association member.
Network policy outcome indicators measure the level of accomplishment of the specific behavioural changes that the Basque government’s cluster policy aims to promote. The objective of the policy implies the development of strategic actions in cooperation among cluster agents. Two indicators are employed. First, we measure ‘associative maturity’ as a reflection of the degree of advancement in the development of strategic projects in cooperation. Associative maturity is directly linked with the social capital mechanisms that underlie network cooperation (Isham et al., 2002) and favour the sharing of information, the reduction in transaction costs, the reduction in barriers to collective action, the improvement of collective decision-making processes, and so on. We measure it using a 12-item question designed to reflect where each participant is situated in the key progress stages towards the policy goal of strategic cooperation. Thus the indicator acts as a thermometer of the cooperation stage in which the cluster is currently located. From Stage 0, which implies that members simply share information, to Stage 4, where they create stable cooperative groups, the indicator supports a diagnosis of the resulting level of associative maturity within the network. Our second measure of NPOs captures the observed projects in cooperation among the network, including the number of projects, their nature and the perceived value generated by them.
We bring these SC and NPO indicators together in a framework that is designed to consider how underlying social capital interacts with the implantation of a network policy and with the specific outcomes that such a policy aims to achieve. This is illustrated in Figure 1. The figure highlights the existence of two types of impact: direct impacts of the policy both in terms of its desired outcomes and on the underlying social capital context; and indirect impacts from the underlying social capital to the network policy outcomes and vice versa. It is important to note that the framework is designed not to distinguish in practice between the strength of these different impacts but to highlight the importance for policy learning of understanding the co-evolution of underlying social capital and network policy outcomes. In the remaining sections of the paper we illustrate how this approach has been applied to the specific network and policy in question and discuss the key outcomes of this process.
The Hegan case study
A case study explores ‘a contemporary phenomenon in its real context, where the limits between the phenomenon and the context are not well defined, and in which multiple sources of evidence are used’ (Yin, 1989: 23). It is hence an ideal method for exploring the effects of network policies, which we have argued cannot be separated from their context and which require exploration in depth (Eisenhardt, 1989). Thus we have employed a case study informed by the general framework presented in the previous section, developing a process targeted at a specific networking policy.
The case study relates to one of Europe’s longest-running cluster policies, that of the autonomous region of the Basque Country in northern Spain. At the beginning of the 1990s the Basque government was a pioneer in establishing a Porterian cluster policy that remains in operation today. The specified aim of the policy is the improvement of the competitiveness of firms and the region through cooperation in strategic projects related to three main areas: technology, quality management and internationalization. This is operationalized in support for cluster associations (CAs), institutions whose main objective is to improve each cluster’s competitiveness by facilitating and fostering cooperation among its members, which include firms, R&D centres and universities. There are currently 21 ‘clusters’ or ‘pre-clusters’ supported by the policy, the most recent of which were established in 2010.
Selection of the specific cluster for this study was guided by two main criteria: first, the motivation of the management team of the CA to be involved in an innovative pilot process, following a presentation of the project to all associations; second, the timing of the project in terms of the cycle of activities of the associations that demonstrated their interest, in particular to benefit from synergies with their own strategic reflection processes. This led to the selection of Hegan, the Basque aeronautics cluster, a medium-sized cluster established in 1997.
Hegan is a private, non-profit association that groups together 35 Basque entities, and it was set up to foster, promote and stimulate the aeronautics and space sector of the Basque Country. Its members are mainly aero-structure producers, engine producers and firms focused on systems and equipment. In terms of their size profile, 58 percent of the members have fewer than 50 employees, 17 percent have 50–500 employees and only 25 percent have more than 500 employees. The evolution of the cluster has been very positive during the past two decades, with big increases in turnover, exports and R&D expenditure. The analysis reported in this paper covers a three-year period from 2008 to 2010, although the participative evaluation on which the analysis draws is an ongoing process. Some basic statistics for the Hegan cluster in 2010 are presented in Table 1.
Hegan basic statistics, 2010.
Source: Hegan, 2010 Annual Report, URL: http://www.hegan.com.
Constructive validity of the case analysis was ensured by the use and triangulation of multiple sources of evidence and the contrasting of results with key agents in the case (Yin, 1998). The chain of evidence was constructed from press items in the period 2008–10, the CA’s annual report, detailed interviews with policy makers in the Basque government and the CA management team, information collected in workshops developed with the members of the CA, information collected by a questionnaire to the CA members, and complementary information from the CA’s internal strategic reflection process. Internal validity was ensured by the design of a dedicated framework, based on the relevant literature and on preliminary assumptions. Nevertheless, external validity was not confirmed in this study, given that we are treating an isolated case. Thus we should be cautious about generalizing our findings. However, our analytical framework was designed specifically to foster policy learning outcomes in the context of its application. In this sense, the research design represents an important contribution from which insights may be drawn for other policy contexts. The process was developed in three main phases: design and planning; application; and reflection on the principal results and learning process.
During the planning stage (autumn 2008), two pilot face-to-face interviews were conducted with policy makers and the CA management based on open-ended, moderately directive questions. We then explored the extent to which the respondents agreed with our theoretical logic, searching for common themes among the transcripts rather than developing a formal coding strategy. We found initial support for the two core assumptions underpinning the framework.
Supporting the assumption that social capital is a key element in network policy development, for example, a policy maker argued that: ‘It is an intelligent environment that, as opposed to an environment of isolated agents, can build on what already exists to create the conditions for generating a virtuous circle.’ Moreover it was argued that ‘the intensity of the policy depends on the degree of acceptance (the voluntary adherence to the cluster policy). If a cluster really exists it is because the members want it to, and if they stop believing (in the cluster policy) it will disappear . . . The weakness of the model is the weakness in the relations of the network.’ A member of the CA management team also stressed the importance of the ‘team mentality and cohesion’, pointing out that this is something that should exist and often needs policy stimulus.
There was also support for the assumption that the social capital of the network is a collective asset developed by the behaviour of all the agents participating in the network. From the policy-maker side, it was suggested that ‘we look to define what is good for the cluster between all of us’, recognizing that this involves the integration of a range of different preferences and aims. These sentiments were echoed by the CA management, signalling the importance that ‘there is willingness to agree, alignment within the diversity, respect for each agent, but all with a common objective’.
These initial findings added to the theoretical justification for the development and application of an evaluation framework centred on learning from underlying social capital dynamics, alongside a more conventional focus on networking outcomes and performance impacts. The evaluation was operationalized using a participative process structured around a series of workshops among stakeholders.
Three workshops were organized during 2009 to which all stakeholders in the CA were invited: policy makers, CA management and CA members. These workshops served three purposes: first, to explain to all stakeholders the novelty and benefits of an evaluation approach centred on learning rather than control; second, to design, agree and approve the framework and the indicators that would be employed to analyse the cluster; third, to support social capital improvement in terms of facilitating greater policy understanding, commitment and common purpose among stakeholders themselves. As a result of this workshop process, the framework created included two kinds of measurement that are relevant for the analysis in this paper: social capital indicators and network policy outcomes. 4
Data to construct these indicators were collected through a simple online software application that firms were asked to respond to at the beginning of 2010. Of Hegan’s 35 members, 21 answered the survey (60 percent) and 22 provided information on the projects in which they participated (63 percent). In line with the objective to stimulate a learning process of evaluation, the initial results were presented to stakeholders to continue the participative reflection process of earlier workshops. This will continue periodically as it becomes embedded in the annual processes of the CA and other complementary indicators reflecting impact and performance are established. In the next section we reflect on the policy learning results.
Policy learning outcomes
The learning outcomes generated during the process have been classified into two levels: general and specific. General learning refers to the CA globally. In particular, it includes improved policy understanding about the overall level of associative maturity, the strategic alignment of collaboration projects and the social capital base within the network. Specific learning refers to distinct groups within the CA. It highlights the heterogeneity of the network, within which certain groups of organizations have been identified as facilitating the development of more tailored networking strategies.
General learning outcomes
The level of associative maturity is one of the most interesting general outcomes extracted from the application of the methodology. Although in the short term this indicator is difficult to change, it represents an important long-term gauge of how underlying social capital and the network policy are interacting in developing the conditions for strategic cooperation, which is the aim of the policy. In our case study, the associative maturity of the cluster is found to be generally low and far from meeting the stated aim of the policy, the aforementioned Stage 4 in which cooperative groups are created and functioning (Figure 2). Although the policy is observed to be far from meeting its aim of generating cooperation in strategic projects, its potential is demonstrated in the 83 percent of firms that engage to some extent in identifying synergies among firms in the network (Stage 2) and the 82 percent of firms that prioritize collective over individual interests to some extent (Stage 3).
Associative maturity.*
Complementing this indicator of associative maturity, there is also much to be learned from an evaluation of the strategic alignment of existing collaboration projects. Although projects in collaboration is a basic outcome of the network policy, concern to demonstrate numbers of projects can lead to overlooking the alignment of these projects to the goals defined in the network’s strategic plan. This is a particularly important indicator of whether the network is developing towards cooperation in more strategic (and therefore often sensitive for cooperation) areas. Analysis of this question suggests that the projects were in general well aligned with the broad areas of action defined in the strategic plan of the CA. However, it was also revealed that the projects themselves tend to focus on more operational rather than strategic activities.
Finally, in terms of general learning outcomes, we can analyse the relationship between these outcomes (associative maturity and projects incorporation) with indicators reflecting the three dimensions of social capital. Although there are insufficient observations to establish the significance of correlations, the results paint an intuitively interesting picture. For example, we can observe a broadly positive correlation between associative maturity and each of the three social capital dimensions (see Figures 3–5), with this relationship most marked in terms of relational social capital. Similarly, we can observe a broadly positive relationship between the cognitive and structural dimensions of social capital and the number of projects in which members collaborate (Figures 7 and 8) but a less clear relationship between relational capital and project collaboration (Figure 6). A tentative interpretation of these patterns might be that, whereas relational social capital is important for the development of general attitudes that reflect greater associative maturity, the existence of a shared vision and established or desired ties between members is important for the emergence of concrete projects in collaboration.
Relational social capital and associative maturity.
Cognitive social capital and associative maturity.
Structural social capital and associative maturity.
Relational social capital and project collaboration.
Cognitive social capital and project collaboration.
Structural social capital and project collaboration.
Specific learning outcomes from revealed heterogeneity
Alongside general insights into the levels of social capital and behavioural outcomes such as associative maturity and project cooperation, the evaluation process has also demonstrated to the CA and policy agents the degree of heterogeneity among network members in these indicators. This is particularly interesting for policy learning because it offers opportunities for designing more sophisticated strategies, tailored to different groups, for improving the functioning and ultimately the impacts of the network. Moreover, with respect to the structural dimension, the process has led to the identification of firms by their ‘desire for cooperation’ and by their ‘attractiveness for cooperation’ as perceived by others. Table 2 illustrates this, along with an aggregated indicator reflecting overall ‘cooperation potential’.
Structural dimension of social capital: Indicators of network ties.
Here we can identify a group of six members (highlighted) with both high desired cooperation and high attractiveness, and it is interesting to note that four of these firms also demonstrated particularly high values in the relational and cognitive social capital indicators. There is one firm (A20) with a strong desire to cooperate but low attractiveness, and a further three (A9, A1, A15) that are reasonably attractive but demonstrate low desired cooperation. In the reflection process with stakeholders, these indicators were acknowledged as extremely useful for the CA in terms of identifying structural hotspots and cold spots in the network, thus defining strategies to build on strengths and address weaknesses. Furthermore, they can be supported by graphical analyses of the network configuration, an example of which is drawn in Figure 9 to illustrate the network structure in terms of reciprocal cooperation relationships (actual and desired). The centrality of associate number 17 is highlighted in this configuration, and there are four isolated members with no actual or desired reciprocal relationships. Most strikingly, the network does not exhibit a great deal of density in reciprocal relationships between different agents, but rather takes the form of a hub-and-spoke network with one central firm and two small clusters where there exists limited relationship density among spoke firms.
The Hegan network configuration.
The increased awareness among the CA and policy makers around the need for differentiated approaches towards, members can be illustrated in this case by the identification and characterization of two broad groups of firms. First, a group of firms with lower levels of social capital than the average tended to demonstrate lower associational maturity. In general, members of this group also participate less in projects and they have a worse perception of the contribution of the projects to their competitiveness. The reflection process has established that there is considerable heterogeneity among the firms included in this group. For example, whereas some have been observed to have inherently fewer synergies with other members owing to the kind of activities in which they are specialized, others are considered to have high synergies and attractiveness as cooperation partners.
Secondly, there is a group of firms with higher levels of social capital indicators, alongside a tendency to participate more intensely in projects and to have a better perception of their value. Among this group of active firms, the outstanding behaviour of a group of small firms has been detected, despite the well-known barriers to collaboration faced by small firms. It is particularly interesting to explore how these ‘active’ firms could play a role in facilitating the participation of less active firms where clear synergies exist.
The analysis of these two groups of firms has led to reflection among researchers and policy stakeholders that has yielded (and we suggest will continue to yield) various other learning outcomes. For example, the identification of groups of members with different levels of relational social capital, measuring respectively trust, reciprocity and commitment, has raised consciousness around potential barriers in developing cooperative projects. It was also concluded that an important factor in increasing social capital is likely to be an increase in the extent to which different people from each organization participate in CA projects and activities. This is a further area, therefore, where policy makers and the CA are seeking to adjust their strategies for facilitating member cooperation.
Conclusions
The role of underlying social capital in conditioning and interacting with ‘soft’ policies that aim to facilitate relationships of cooperation has not typically entered into the analysis and evaluation of these policies. Yet it is critical, because the cooperation outcomes desired by network policies must always be achieved in a specific social context. As such, this paper provides both theoretical and methodological contributions. It sets out a theoretically informed framework for analysing the co-evolution of underlying social capital foundations and network policy outcomes when a network policy is implemented. Through specific application to a case, the paper demonstrates how policy learning can be supported by participative processes that integrate a deeper appreciation of the different realities and behaviour of targeted firms in terms of their social capital.
The social capital framework presented here forms part of a wider evaluation of the results and impacts of networking under the Basque cluster policy. As such, we have shown how an integrated and participative process of design, data collection and reflection with stakeholders can play an important role in supporting policy learning. Specifically as a result of this process, policy makers and the cluster association have been able to identify groups of agents among which different strategies can be employed to enhance the cooperative behaviour, and potentially the impact, of the network. It has also led to improved understanding of the critical role that small firms play in the network under study, despite the tendency of policy to look first towards larger firms in initiating projects. Finally, reflections on the results have led to greater appreciation of the centrality of ‘people’ alongside ‘organizations’ in such policies, encouraging the development of strategies to broaden and deepen the pool of participants. These findings clearly have specific practical implications for the agents that form part of this case, and their discovery in this context opens the way to further research with the potential for more general reach.
Indeed, the paper’s main limitation is its case specificity, an issue that should be addressed in the future by employing this framework in other cases. Nevertheless, at this stage important general lessons have emerged regarding the benefits of delving deeper into the relationships that characterize networks so as to foster learning among stakeholders that can improve network policies. The learning stimulated by this participative process is evidenced by the decision of the CA to continue to use the generated tool to analyse the evolution of different indicators, initiating a long-term reflection around their strategy to meet the aims of the policy. This intention was signalled by a presentation of some of the preliminary results reported here at their annual general meeting in 2010 in order to generate consciousness of current weaknesses and strengths as a base for future learning. From the research side, repetition of the process will provide interesting data for a temporal analysis of the relationships tackled by this preliminary analysis. Moreover, the learning tool will now be opened up to the other clusters that come under the Basque government’s cluster policy, with the aim of generating a culture of evaluating for learning. Alongside supporting further policy learning outcomes, the resulting data will facilitate further scientific analysis of the under-explored relationship between social structure and ‘soft’ policies.
Footnotes
Acknowledgements
We would like to acknowledge the contributions of Marian Diez and Miren Larrea to the design of the project underlying this paper, and the research assistance of Edgardo Cruzado with regard to the data analysis. We are also grateful to Hegan (the Basque aeronautics cluster association) and the Department of Industry, Trade and Tourism of the Basque government for its collaboration in the project.