Science Parks and Local Knowledge Creation: A Conceptual Approach and an Empirical Analysis in Two Italian Realities

Abstract

Thanks to their ability to perform knowledge-related tasks such as diffusing knowledge locally, promoting high-tech firms, establishing links between knowledge-creating bodies (e.g. universities, research centers) and knowledge-exploiting bodies (e.g. public and private firms, local institutions), science parks have long been considered efficient instruments of industrial and regional policy (Jones 1996; Martin 1997). They were expected to enhance the diffusion of new and advanced technologies/knowledge among firms, and consequently boost the competitiveness of firms and regions. At present, a significant number of studies show that most science parks have failed to perform their intended function (Appold 2004; Massey et al. 1992; Quintas et al. 1992; Vedovello 1997). Many reasons have been attributed to this failure. An important reason is the erroneous and misleading belief that simple geographical proximity between sources of knowledge and local firms is sufficient to foster the widespread spatial diffusion of information, technologies and new ideas (Macdonald 1987; Vedovello 1997). Another reason is the peculiar governance structure of science parks: in fact they may address as many objectives as there are main stakeholders, which in turn may lead to inconsistent policies (Monk et al. 1988; Löfsten and Lindelöf 2002).

Though the paper is the result of a joint effort of the two authors, R. Capello wrote Sects. 13.1, 13.2 and 13.5, while the remaining sections have been written by A. Morrison.

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Introduction

Thanks to their ability to perform knowledge-related tasks such as diffusing knowledge locally, promoting high-tech firms, establishing links between knowledge-creating bodies (e.g. universities, research centers) and knowledge-exploiting bodies (e.g. public and private firms, local institutions), science parks have long been considered efficient instruments of industrial and regional policy (Jones 1996; Martin 1997). They were expected to enhance the diffusion of new and advanced technologies/knowledge among firms, and consequently boost the competitiveness of firms and regions.

At present, a significant number of studies show that most science parks have failed to perform their intended function (Appold 2004; Massey et al. 1992; Quintas et al. 1992; Vedovello 1997). Many reasons have been attributed to this failure. An important reason is the erroneous and misleading belief that simple geographical proximity between sources of knowledge and local firms is sufficient to foster the widespread spatial diffusion of information, technologies and new ideas (Macdonald 1987; Vedovello 1997). Another reason is the peculiar governance structure of science parks: in fact they may address as many objectives as there are main stakeholders, which in turn may lead to inconsistent policies (Monk et al. 1988; Löfsten and Lindelöf 2002).

Beyond the reasons encountered in the literature, all with a sound scientific validation, we consider that two additional major elements can account for the questionable effectiveness of science parks which has been reported. First, the existing assessment literature pays little attention to the variety of science parks. They have different structures, missions, aims and functions: these factors have to be taken into account in the evaluation procedure. Second, we feel that relevant factors affecting the effectiveness of science parks have been inadequately addressed in the literature. In particular, the role that science parks can play in the diffusion of knowledge by supporting and stimulating spontaneous local knowledge creation and transfer channels, has been overlooked. Moreover, little is known about the gatekeeping function of science parks – that is their ability to search and scan the external environment in order to connect external knowledge sources to local actors ¹ – and their ability to connect local actors (i.e. to contribute to strengthen what regional economists have for a decade called “collective learning”).

This paper intends to fill in this gap. In particular, the paper aims at measuring the effectiveness of science parks; defined and measured as the ability of science parks to support spontaneous mechanisms of local knowledge transfer among local actors.

The paper is structured in two parts. In the first part we present a conceptual framework in order to classify science parks according to their “effectiveness” (Sect. 13.2). In the second part, we present the results of the empirical analysis. On the basis of a quantitative approach we test the effectiveness of science parks in helping to foster spontaneous mechanisms of local knowledge creation and transfer. For this purpose, we carry out our analysis on a sample of 160 firms located in two Italian regions where selected science parks operate (Sect.13.3). Findings suggest that our conceptual framework is well equipped to guide investigation into this issue (Sects. 13.4 and 13.5).

Science Parks and Local Knowledge

Science Parks: A Definition

The expression “Science Park” covers a large variety of research centers and innovation incubators. In general terms, a science park is defined as a geographical area in which firms, universities and research centers have a common location in order to exploit proximity advantages, knowledge spillovers and dynamic agglomeration economies. Examples of these kinds of Science Park are the American success stories of Silicon Valley and of Stanford Research Park, replicated in Europe, e.g. in the Sophia Antipolis Park in France and in the Cambridge Science Park in Great Britain.

However, very different types of institutional entity can be included under the generic label of science park, such as private/public institutions created with the aim of encouraging the formation and growth of innovative (generally science-based) businesses, and actively engaging in transfers of technology or business skills to a “customer” organisation (Colombo and Delmastro 2002). A science park of this kind is not necessarily located in a particular geographical area, but is represented by a formal institution running research activities and hosting research laboratories; Innovation and Technological Centers can be included within this category. Finally, a third typology of science park is represented by public/private institutions whose sole aim is to act as an intermediary body between knowledge creators (e.g. university, research centers) and knowledge users. In this case, most of the literature refers to Business Innovation Centers (BIC). ²

The deep differences in the nature of science parks explain the large variety of functions that they may develop with regard to:

• The diffusion and transfer of advanced technologies (e.g. best practice) among firms, sectors and regions, and support to the creative adoption of traditional innovation (a knowledge transfer function)

• The creation of radical innovation, aiming at contributing to a shift in the technological frontier (a knowledge creation function)

• The creation of a scientific environment, where firms accrue the benefits of being close (geographical proximity) to different sources of knowledge (a seedbed function)

• The creation of new technology-based firms (NTBFs), through spin-off processes where scientists move from research laboratories to privately-owned research activities (an incubator function for new firms)

A science park, regarded as a real-estate investment in a given geographical area, where R&D laboratories of public and private firms, research centers and universities are hosted, in principle performs all of the above functions, with the exception of the knowledge transfer function (Fig. 13.1.). A different and opposite case is represented by a science park which does not host research activities; but the most important function is that of knowledge transfer. Science and Research Parks, such as Innovation Centers, which host incubator functions and R&D laboratories, are generally able to fulfil most functions, although with more limited expected performance than: “geographical” science parks.

Fig. 13.1

Typology and functions of science parks

We maintain that measurement of science park effectiveness should take into account their peculiarities. For example, it would be misleading and wrong to measure the ability of a science park which does not host any R&D activity according to its ability to create new knowledge, or be a seedbed for innovation.

However, a word of caution has to be sounded. Although many science parks do not host or run R&D activities, they can still play a strategic role in supporting firms’ competitiveness by fostering firms’ capacity for innovation, and processes imitating and adopting best practice.

Science Parks and Learning Processes

Most empirical analyses on the effectiveness of Science Parks have mainly approached this issue by comparing samples of on-park and off-park firms and measuring their different innovative performance (Monk et al. 1988; Westhead and Batstone 1998; Löfsten and Lindelöf 2002). The main aim of these studies is to capture the role that science parks play in the creation of knowledge.

However, our main idea in this study is that science parks can play a strategic role in the innovative performance of firms by supporting, stimulating and increasing the number of channels through which knowledge develops and cumulates at a local level. In other words, science parks can play a very important role in enabling and reinforcing spontaneous mechanisms of local knowledge diffusion. In doing this, they would indirectly foster the innovative activity and performance of local firms.

The literature on local knowledge diffusion has recently been enriched by many conceptual and empirical studies. There is agreement that “physical proximity” among firms plays a crucial role in improving their innovative capacity (Audretsch and Feldman 1996; Jaffe 1989). Space matters due to the existence of “knowledge spillovers”. However, the way in which space is conceptualized through the concept of “knowledge spillovers” differs greatly, depending on the approach considered (Capello and Faggian 2005).

Geographers and industrial economists interpret space as simple physical proximity among innovative actors. Solid econometric results prove the importance of physical proximity. In practical terms, this approach sees proximity to university laboratories and other research centers as providing nearby firms with easier access to scientific expertise and research results, thus facilitating transfer of research into commercial application (Colombo and Delmastro 2002).

Many authors address this issue in their contributions. Audretsch and Vivarelli (1994), for instance, try to measure the effect of knowledge spillovers on innovation – measured in terms of new patents (using data on Italian firms) – and find a significant effect of these spillovers on small and medium sized firms. However, their definition of spillovers is quite limited, since it only covers physical proximity (physical distance) to universities or research centers. ³

Autant-Bernard (1999) extends the definition of spillovers to include sectoral aspects – firms are close if they belong to the same sector. Again, as in Audretsch and Vivarelli (1994), he finds a significant positive relationship between knowledge spillovers – measured in terms of R&D expenditure and researchers of firms in the local area – and the innovative performance of firms.

Another flourishing stream of studies has focused on agglomeration economies, in particular on the relative importance of localization versus urbanization economies.⁴ The two perspectives differ in the respect that the localization economies’ argument, following Marshall, points that specialization matters for the co-localization of firms and in turn for spurring knowledge externalities at local level (Glaeser et al. 1992). Conversely, the latter view, drawing on Jacobs, claims that diversity, as the one observed in cities, is crucial for cross-fertilization among different industries, and in turn for the generation of new ideas (Jacobs 1969).

Despite recognizing that proximity to universities, research centers and other firms – belonging to the same or to different sectors – is important, what emerges from a critical review of the literature is that the existence of knowledge spillovers is explained purely in terms of the probability of contacts between economic actors, which increases in a limited geographical space.

The second and older approach is linked to regional economic studies. In these studies the concept of proximity, in the form of agglomeration economies, has always been an important element in explaining location choices, and local economic performance.⁵ During the 1970s and 1980s agglomeration economies explained the performance of new industrial areas; in the literature using this approach, territory is analyzed as an active element in economic development, being a productive resource in itself and a source of advantage for firms. In more recent times, territory has been conceived as a support for innovation activity, being able to decrease uncertainty and risks accompanying innovative processes. In this view, space is a complex concept, which does not only refer to geographical proximity, measured in terms of both physical and time distance.⁶ Space is interpreted in terms of relational proximity, defined as the ability of local firms, institutions and people to put in place strong local relationships – market relationships, power relationships, and co-operation.⁷ These relationships, alternatively called “relation capital”, underlie any process of collective learning.

Collective learning is the territorial counterpart of learning in an industrial context; it is thought of as the vehicle for knowledge transmission, both in a temporal and in a spatial dimension. In the former dimension, the transfer of knowledge is guaranteed by continuity; in the latter by interaction among agents, which guarantees transmission among individuals and firms and which becomes, in the case of the milieu, an element for the spatial transfer of knowledge.⁸

The channels through which knowledge develops locally are considered to be:

• The local labor market. The local labor market plays an important role within the local production system, as the high internal turnover of specialized labor. Low external mobility assures cross-fertilization processes for firms and professional upskilling for individuals; local know-how grows through a collective and socialized process, but this is subject to the risks of isolation and locking-in unless external energy is also captured through selected external co-operation linkages.

• Stable linkages between suppliers and customers. Stable input–output relationships generate a codified and tacit transfer of knowledge between suppliers and customers – it accumulates over time and defines patterns of incremental innovation which feed a specific technological trajectory. In this case also, comparison with firms’ technological trajectory is straightforward. As Aydalot (1986) suggested, the innovation process in a territorial entity such as a milieu is a process of “rupture/filiation” (break and continuity): if an innovation is a break with a pre-existing situation, economic creativity and innovation potential have their seeds in the local accumulated knowledge and know-how acquired over time.

• Intense innovative interactions with suppliers and customers and by mechanisms of local spin-off. Theoretically, a spin-off is defined as a new independent firm fulfilling two criteria (Perhankangas and Kauranen 1996; Dahlstrand 2000): (a) the startup of a new business by an agent previously belonging to another local firm, and (b) the derivation of a new business idea due to the previous employment of the founder. Local milieux provide both the social and the market preconditions for this phenomenon to take place: from the social point of view, high trust and a common sense of belonging to the same cultural community make this process acceptable.⁹ Local market conditions, such as stable interactions with suppliers known from a previous job, receptive local demand for particular products developed in a previous job, and the presence of external economies, assure locational advantages, guarantee the achievement of profits and thus give rise to chances for survival on the local market.

Beyond this set of mainly informal, “un-traded” relationships – among customers and suppliers, among private and public actors – and a set of tacit transfers of knowledge taking place through individual professional mobility and inter-firm imitation processes, another knowledge acquisition channel has been highlighted in the literature. More formalized, mainly trans-territorial co-operation agreements – among firms, among collective agents and among public institutions – in the field of technological development, vocational and on-the-job training, infrastructure and services provision are important channels for achieving new knowledge. In trans-territorial networks, partners are single and selected economic units – enterprises, banks, research centers, training institutions or local authorities – where the locational element is only one of many elements defining the unit. At first glance, therefore, these networks only link together different economic actors, with no necessary relation to space. But when the location of a unit takes on significant meaning, inasmuch as it reveals a set of relations which generate territorial development and identity (e.g. Apple at Cupertino, Silicon Valley) and when these network relations start to multiply, they do become territorial. When carefully observed, the identity of the local milieu often prevails over the identity of the individual partner, stressing the importance of the territorial: the strategic importance of links with a company in Silicon Valley resides more in the opening of a “technological window” in Silicon Valley than in access to that specific company’s know-how.

This second kind of network can define a process of “learning through networking”. Through strategic alliances, non-equity agreements and technological cooperation, firms are able to capture some of the necessary assets from outside, overcoming the costs of internal development. This model is in a sense intermediate between internal and collective learning, it opens the firm to the general context, but maintains it within a set of selected and targeted relationships. ¹⁰ The continuity element is generated by the relative complexity of processes involved in setting up the terms of the cooperation contract, the clauses and sanctions for excluding opportunistic behavior; this promotes a long-term horizon and relative stability in the agreements. On the other hand, the knowledge transfer element is generally seen as rapid and powerful due to the complementarity of the different partners in the cooperation network. This learning channel helps local firms to avoid being locked into old knowledge, since it allows new and advanced knowledge to be introduced into the area.

When science parks are created to perform innovation and knowledge diffusion functions, their primary role should be to support and foster direct, but more importantly indirect channels, i.e. all socialized processes of knowledge creation and diffusion. In other words, science parks should be able to participate in processes of:

• Creating both vertical and horizontal stable linkages among firms, both at the local and at the international level

• Helping the transparency and information of the local labor market

• Giving support to spin-off activities

Figure 13.2 reports the role science parks may play in supporting local relational activities, on the one hand, and external trans-territorial networks, on the other. Different combinations of the two give rise to different learning trajectories, namely:

• A “network learning trajectory”, when science parks are able to strengthen relationships between local actors and agents outside the area, through which new and advanced knowledge can be acquired by local firms.

• A “localized collective learning trajectory”, when, conversely, science parks support solid and long lasting relationships among local actors, both vertically and horizontally. In this way, knowledge will cumulate in a socialized (collective) way around a well defined technological trajectory, giving rise to what industrial economists call “localized learning”. Firms will search for new technological solutions around well known technological and geographical boundaries if no new and radical knowledge is inserted locally.

• A “collective learning trajectory”, when science parks are able to strengthen both local and external relationships. In such a case, socialized learning processes also encompass new technological solutions which drive towards paradigmatic change in technological trajectories brought into the area and shared among local actors.

• A “no-learning trajectory through science parks”, when science parks do not play any role either supporting local relationships or external networking.

Fig. 13.2

The role of science parks in learning processes: possible alternatives

In our understanding, the effectiveness of science parks can be defined as their ability to enter into and support socialized processes of knowledge creation. The greater their capacity to have an active role in “collective learning” processes – by supporting channels of local knowledge transfer, such as relationships among local actors, and by enhancing long distance relationships – the greater their effectiveness. When science parks play a role in localized collective learning processes, they risk helping firms’ competitiveness only in the short term, but do not help local actors to avoid lock-in mechanisms or to jump to more advanced technological trajectories, which would assure long-term competitiveness. On the contrary, when science parks only act on networking processes of learning, they neglect the innovation transfer function, and are thus not actively involved in socialized mechanisms of knowledge diffusion.

Effectiveness of Science Parks: An Overview of Propositions

In the previous sections a definition of the effectiveness of science parks was proposed. From this definition we can easily deduce our first testable proposition.

Proposition 1

We expect the innovative capacity of local firms to be influenced by the active involvement of science parks in socialized processes of local knowledge diffusion and in networking activities.

This general statement is true under certain conditions. The first one we refer to is firm size. Socialized processes of knowledge diffusion are important mainly for small firms. In large enterprises, large-scale R&D functions and engineering departments act as information storage agents and select decision-making routines, because they are long-term units. More importantly, R&D functions serve as a firm’s memory, where knowledge is cumulated, embedded in routines, and transferred as tacit knowledge in the process of searching for new technological innovations, giving rise to specific technological trajectories. In these organizational structures, therefore, there are no reasons for processes of knowledge socialization to occur.

Instead, in small firms the innovation searching function does not exist due to diseconomies of scale and the unpredictable and relatively short life of small firms. It is in this type of productive system that knowledge cumulates in a socialized way (e.g. in the local labor market; in the network of local customers and suppliers) (Camagni 1995). From these considerations, a second research proposition emerges:

Proposition 2

The active involvement of science parks in socialised processes of local knowledge diffusion and in networking plays a greater role in the innovative activity of small firms than large firms.

A further important element, which helps to assure the effectiveness of science parks, is the degree of relational capital existing in the area. This stems from a strong sense of belonging and a highly developed capacity for cooperation which is typical of institutions and agents with similar culture (Capello and Faggian 2005). In areas where this attitude is absent, the chances of a science park developing local cooperation is expected to be very limited. In areas where relational capital is very intense and functions efficiently, science parks risk having a superfluous role. From these considerations another testable proposition can be envisaged.

Proposition 3

We expect that greater the role science parks play in the socialized processes of local knowledge diffusion and in networking, greater is the degree of relational capital existing in the local area.

Last, but not least, a third important condition under which science parks play an active role in socialized processes of knowledge creation/diffusion is the reactive capacity of firms. As has been widely suggested by the literature, a firm’s capacity to exploit new knowledge depends to a significant extent on the level of prior related knowledge stored within the firm, which enables the value of new information to be recognized, assimilated and applied for commercial purposes. These abilities collectively constitute what has been labelled as “absorptive capacity” (Cohen and Levinthal 1990), and more recently reinterpreted in terms of knowledge-relatedness in order to take account of technological diversification processes occurring within firms (Breschi et al. 2003).

These considerations bring us to a fourth testable proposition, i.e.:

Proposition 4

Greater the involvement of Science Parks in socialized processes of local knowledge diffusion and in networking, greater is the absorptive capacity of a firm.

In the following sections we test these research propositions through an empirical analysis.

Database and Methodology

The Sample

The empirical analysis was carried out in two areas where science parks operate, namely Pisa and Genova. In the Pisa area we selected firms from a variety of science parks.¹¹ Overall these technological centers to a large extent internalize the research function. According to our taxonomy they perform most of the functions attributed to the ideal Science Parks (e.g. transfer function, incubator function, – see Sect. 13.2.1). Only one Science Park operates in the Genova area, the Science Park of Liguria.¹² As against the Science Park from Pisa, it does not own research facilities, acting mostly as coordinator and promoter of best practices and innovation-related initiatives.¹³

A sample of tenant firms was investigated along a vast array of dimensions.¹⁴ For this purpose we elaborated a questionnaire covering the following areas:¹⁵

• Measures of a firm’s characteristics: year of establishment, number of employees, a variety of indicators of economic performance (e.g. sales, exports), competitive position, knowledge base, etc.

• Measures of a firm’s innovative behavior: input (R&D expenditure; licences) and output (patents, percentage of firm’s turnover related to product and process innovation) indicators of innovation

• Measures of milieu and networking learning processes: information about the importance of local and external knowledge sources (e.g. competitors, providers, clients, universities, knowledge facilitators), which contributed to recent product and process innovations

• Measures of Science Park effectiveness: the role of science parks in connecting clients with relevant actors for developing the innovation (e.g. local and external competitors, providers, clients, universities, knowledge facilitators)

• Measures of relational capabilities: the percentage of relationships with local actors, in particular with customers and providers and their contribution in terms of relevant knowledge; the characteristics of the local labor market

• Measures of a firm’s linkages with science parks: frequency, typology of information/knowledge accessed through science parks, obstacles to knowledge acquisition, etc.

The survey was conducted over a period of 3 months covering 160 firms equally distributed between the two areas Genova and Pisa (80 firms each).¹⁶ A large number of firms belong to high-tech sectors (Table 13.1), although each geographic area presents some peculiarities. For example, we observed that sampled firms from Genova are skewed towards “old economy” sectors (e.g. oil, metal products, machinery), whereas in Pisa the majority of firms fall into the high-skill service sector. As far as size is concerned, few differences emerge: the vast majority are small or medium-sized firms and most established in the last two decades.

Table 13.1 Sample distribution by firm size, sector and location

Description of Variables

Table 13.2 reports the description of the variables included in the empirical analysis. We built a set of conventional explanatory variables, a set of key variables – measuring the role of science parks in socializing knowledge – and an independent variable, which measures a firm’s performance.

Table 13.2 Description of variables

The following “explanatory variables” are taken into consideration:

• A variable of absorptive capacity (AC), constructed as a dummy variable coded 1 if a firm’s R&D expenditure is above the sample mean, 0 otherwise.

• A variable of firm size (DIM), which is coded 1 if a firm’s size is above 20 employees, 0 otherwise.

• A variable of relational capital (REL), which assumes value 1 if the share of local linkages of the firm is above the sample mean, 0 otherwise.

• A variable of local collective learning (RELLOC), measured in terms of collaborations with local firms aimed at developing a specific innovation. It is a binary variable assuming value 1 if firms assign a high score to local firms, 0 otherwise.

• A variable of external collective learning (RELEXT), measured in terms of collaborations with external research centers aimed at developing a specific innovation. It is a binary variable assuming value 1 if firms assign a high score to external relationships with these centers, 0 otherwise.

As far as the key determinants are concerned, a number of variables interpreting the ability of science parks to connect local firms to either local or external sources of knowledge (the bridging and networking functions respectively) are introduced. Dummy variables are constructed as follows:

• The role of Science Parks in supporting local or external linkages, PSTLOC and PSTEXT respectively, are coded 1 if firms value science parks as a relevant factor for searching and identifying local or external sources of knowledge respectively (e.g. competitors, clients, universities), 0 otherwise.

• The role of Science Parks in supporting local or external linkages for small and medium firms, PSTLOCSME and PSTEXTSME respectively, are coded 1 if SMEs value Science Parks as a relevant factor for searching and identifying local or external sources of knowledge respectively (e.g. competitors, clients, universities), 0 otherwise.

• The role of Science Parks in supporting local or external linkages for firms having a high absorptive capacity, PSTLOCAC and PSTEXTAC respectively, are coded 1 if firms with high absorptive capacity value Science Parks as a relevant factor for searching and identifying local or external sources of knowledge respectively (e.g. competitors, clients, universities), 0 otherwise.

The dependent variable (INNOPD) measures the innovative performance of a firm. It is constructed as a binary variable coded 1 if firms introduced at least one radical product innovation over the last 5 years, 0 otherwise.

The Bridging and Networking Functions of Science Parks: Some Descriptive Results

In this section we report some data describing the role of science parks in carrying out bridging and networking functions. Results show that the effectiveness of science parks is strongly affected by firm’s characteristics; in particular to which we refer to relational capital, absorptive capacity and firm size.

Figure 13.3 shows the percentage of respondents divided by firm size. Results undoubtedly show that small firms benefit more from the bridging function of science parks; nevertheless science parks seem to support larger firms in building arm’s-length relationships (networking function).

Fig. 13.3

The importance of SP in establishing linkages by firm size

Moving to the second dimension (i.e. relational capability), Fig. 13.4 shows that the greater the firm’s relational capability, the greater is the ability of science parks to connect these firms to either local or external sources of knowledge. Results suggest that the higher the relational capability, the higher the share of firms benefiting from the gatekeeping ability of science parks.

Fig. 13.4

The importance of SP in establishing linkages by degree of relational capital

Similar results can be seen in Fig. 13.5. This shows that the higher the absorptive capacity, the greater the importance assigned to science parks in carrying out their functions (i.e. bridging and networking).

Fig. 13.5

The importance of SP in establishing linkages by degree of absorptive capacity

Science Parks and Knowledge Transfers: Interpretative Results

The Determinants of Firms’ Innovativeness: The Role of Science Parks

The theoretical propositions set out above have been analyzed here through econometric techniques; the aim is to test the impact of science parks on firms’ innovative performance. A logit model is run in order to estimate the probability of a firm introducing an innovation in terms of some explicative variables (e.g. firm size, relational and absorptive capacity, interaction with science parks). The model can be expressed with the help of the following equation:

$$ INNOPD = \alpha + \beta_1 DIM + \beta_2 AC + \beta_3 RELOC + \beta_4 RELEXT + \beta_5 PSTLOC + \beta_6 PSTEXT + \varepsilon_1, $$

(13.1)

where INNOPD is the radical product innovation, RELOC is the local relational capital; RELEXT is the external relational capital, DIM is the firm size, AC is the absorptive capacity, PSTLOC is Science Park’s bridging function between clients and other local firms and PSTEXT is Science Park’s networking function between clients and external research institutions.

Results of the estimates are presented in Table 13.3. Initially we test the explicative power of the locational explanatory variables (13.1). We then introduce different key and firm’s explanatory variables one at a time to capture their contribution to an explanation of a firm’s propensity to innovate.¹⁷

Table 13.3 Results from logit estimations

Local and external relational capital variables (RELLOC, RELEXT) always have a positive and significant effect on a firm’s propensity to introduce a radical product innovation. Our results confirm the established wisdom on this issue: strong relational capital enhances a firm’s innovativeness.

Once inserted into (13.1), variables measuring the bridging function of science parks play a positive and significant role in explaining a firm’s propensity to innovate. In particular this is true for a subset of small firms [see PSTLOCSME, (13.2) and (13.3)]. This latter result strongly supports our first proposition. Science parks play a greater role in transferring knowledge to small firms than to large ones.

Variables capturing the networking function of science parks (PSTEXT and PSTEXTSME), however, exhibit significant, though negative coefficients [see (13.4) and (13.5)], suggesting that by connecting tenant firms to external sources of knowledge, science parks reduce the probability that they will develop innovative products. Although counterintuitive, such a finding appears quite reasonable if compared with other survey outcomes, showing that firms distrust such collaborations for fear of losing strategic information.

Equations (13.6)–(13.8) introduce a last but important explanatory variable, the absorptive capacity (AC), which exhibits significant and positive coefficients (13.6). The same result is obtained with the variable (PSTLOCAC) interpreting the role science parks play in promoting local learning through the bridging function (for firms having a high absorptive capacity); (13.7) in fact shows a positive and significant sign of this variable. When the role of science parks in generating external linkages – for firms having a high absorptive capacity – is examined, the networking function of science parks hinders rather than boosts local learning (PSTEXTAC). On this latter point, we may presume that the networking function can function effectively for a subset of firms. We will test this proposition later in the paper.

Overall, our findings can be summed up as follows. The propensity of science park tenants to introduce radical product innovations is positively influenced by a firm’s characteristics (e.g. absorptive capacity). We confirm that the ability of science parks to perform a bridging function is relevant, especially for small firms. Nevertheless we still have to clarify some unresolved points. First, it is still unclear whether firm-specific characteristics, such as relational capital, affect science park mechanisms of transferring knowledge. Second, we have to understand what factors prevent science parks from effectively performing their networking function.

In order to shed light on these latter points we carried out a cluster analysis. This aimed to single out homogeneous groups of firms – in terms of innovative performance, degree of relational capital, size and degree of absorptive capacity – which share similar patterns of learning.

Firm Size, Relational Capital and Absorptive Capacity in Science Park’s Processes of Knowledge Socialization

In this section we aim to test the influence of firm size, relational capital and absorptive capacity on science park processes of knowledge socialization. We also intend to identify what factors explain how the networking activity of science parks functions properly. Cluster analysis allows homogeneous groups of firms presenting heterogeneous behavior to be singled out. Hence, it allows identification of those groups of firms assuming coherent or distant behavior with respect to our research proposition and their characteristics. Four main clusters are identified in our analysis (Table 13.4):¹⁸

• The first group includes firms specialized in traditional sectors with networking behavior: this cluster is characterized by low innovative performance and large firms operating in scale intensive or supplier dominated sectors. External linkages, either developed directly or mediated by science parks, appear to be the main channel for accessing knowledge. Although they benefit from the gatekeeping function of science parks, their innovative capacity is poor.

• The second group is populated by innovative firms with networking and milieu behavior: this cluster matches almost exactly with sample mean values. It includes small innovative firms operating in both high tech and traditional sectors. Local and external networks of linkages with knowledge sources fuel learning mechanisms. Science parks play a relevant role in strengthening these relationships, in particular those at local level.

• The third group contains firms characterized by weak innovative performance coupled with weak local and external linkages: this cluster identifies a group of firms which are isolated with respect to both local and external sources of innovation. Their knowledge base is also poor, preventing them from seeking and absorbing complementary technological inputs. In addition, their poor ability and propensity to relate with the external environment may prevent them from benefiting from Science Parks services. Firms within this cluster are equally distributed over the two areas (Genova and Pisa) and represent almost half our sample.

• In the last group we have small dynamic firms specialized in high tech sectors and with milieu and networking behaviour: firms within this cluster define a small group of highly innovative firms specialized in ICT. They have entered a virtuous learning trajectory, where internal competencies enable them to establish fruitful linkages with local and external sources of knowledge. In addition they do not ignore the importance of knowledge facilitators; they have in fact established strong connections with science parks, exploiting more then any other cluster the bridging and networking functions provided by science parks.

Table 13.4 Cluster analysis results

These results confirm our research propositions for a subset of sample firms. As Table 13.4 clearly shows, a group of small firms (cluster 4) has higher relational and absorptive capacity, accompanied by a strong innovative performance and stable linkages with science parks. These firms are inserted into a fast learning path.¹⁹ We may conclude that our research propositions related to science park involvement in knowledge processes are satisfied for at least a small, but significant, subset of firms.

Figure 13.6 sums up the main findings: the vertical and horizontal axes represent the science park networking and bridging functions and our four clusters are depicted according to the role science parks play in each of them. Clusters 2 and 4 clearly identify those firms following a collective learning trajectory. Interestingly, in this case we can observe a positive relation between the role of science parks in supporting local and external linkages and firm’s characteristics – small firms with strong relational capital, a high absorptive capacity and a high degree of innovativeness.

Fig. 13.6

Different learning trajectories sustained by SPs

Cluster 1 identifies a group of large firms that benefit from the networking function of science parks, though they are not inserted in collective learning processes. In conclusion, cluster 3 identifies a group of firms with a passive learning behavior. It includes firms having extremely low innovative performance coupled with a poor knowledge base. In this case, science parks fail to enhance any learning process. This passive behavior is somehow a by-product of firm-specific factors, such as low relational capital, coupled with low internal competencies, which also reduce their ability to search for external knowledge inputs. In these cases science parks seem unable to provide feasible alternatives and fail to implement effective measures or design appropriate services for solving firms’ needs.²⁰

Conclusions

The purpose of this work was to evaluate the effectiveness of science parks, measured as the ability to support and enhance spontaneous mechanisms of local knowledge creation, (i.e. to support networking among local actors and with external agents). The effectiveness of science parks in supporting innovation activities has been widely studied in the literature. Our contribution to the large scientific debate on this subject has been developed in two main research directions.

The first line of research suggests it is important to clearly define the typology of a science park when evaluating it. Indeed, science parks may differ to a great extent in terms of their statutory mission, so some have the broad aim of contributing to the breeding of a local economy, while others may have a narrower focus, e.g. that of incubating high-tech start ups. In addition, the motivation behind the establishment of science parks has sensibly changed over the last 30 years. In the early stages of their development, they were primarily intended as policy tools to re-industrialize or renew urban or regional areas. More recently, science parks have narrowed their scope of intervention, and specialized in few and more focused functions. For example, many recent experiences were related to the creations of ICT firms. Such a variety of aims entail that any attempt to assess science parks’ effectiveness has to clearly state the functions subject to evaluation. Thus, for instance, if a science park does not develop any research activity, it would be misleading and wrong to investigate its role in generating new knowledge or in creating a seedbed for innovation. The transfer function will be its main objective and it should be evaluated accordingly.

The second issue we highlight is that science parks run the transfer function in different strategic ways. On the one hand, science parks are able to support and stimulate spontaneous local knowledge creation and transfer channels; on the other hand, they maintain an arm’s-length relationship with external sources of knowledge. These issues, however are still largely uncovered. Our empirical analysis, by providing prima facie evidence on role of two science parks in creating relationships among local actors, shows that firms’ characteristics are key determinants in explaining science parks effectiveness. We found that only a bunch of clients composed of small firms effectively makes use of science parks activities. Moreover, results also show that firms with high absorptive and relational capacity are those that benefit more from science parks’ bridging and networking functions.

The purpose of this study was not to directly assess the policy measures designed to support or implement a science park though the findings from the empirical analysis and the wider investigation over the customers of the science parks provide some useful suggestions in this direction. Overall, it clearly emerges that science parks are far from being easy policy instruments for promoting innovation activities. Science parks in principle should have the aim of making firms aware of their technological needs and scan the environment in order to find solutions for those needs. However, it could be the case that such intervention is superfluous, that is local firms would have, independently from public intervention, satisfied their technological needs. This entails that the public money has been misallocated. It suggests that policy makers, and more broadly the stakeholders and promoters of science park initiatives, should carefully interact with the potential beneficiaries and monitor their needs, and consequently identify the specific kind of services which could match the local demand. However, we should also consider that science parks, as stated above, often pursue several competing goals, and accordingly their members adhere to an initiative for staying motivated in diverse ways. This may explain why our sample includes both clusters of firms nearly disconnected from science parks functions and clusters with strong links. In this case, it is reasonable to say that the former, which mainly include large firms, have become clients for some other institutional reasons other than knowledge transfer (e.g. prestige).

All in all, any intervention to promote science parks should take into account the characteristics of the area where they are supposed to be settled and operate, in addition to the needs of potential customers. Only by tailoring their functions and mission to local needs can they be effective tools of innovation policy and local economic development.