7 The role of knowledge heterogeneity on the innovative capability of industrial districts
Abstract
This chapter seeks to contribute to the ongoing debate concerning the role of heterogeneity for the innovative capability of industrial districts. With this aim, using a knowledge-based approach, the chapter focuses on different sources of industrial district knowledge heterogeneity and studies how the different level of heterogeneity affects the innovative capability of industrial districts.
Four theoretical hypotheses concerning the effects of knowledge and knowledge heterogeneity on industrial district innovativeness are formulated. To test the hypotheses, an econometric analysis on 32 Italian district provinces is applied. Empirical results show that knowledge heterogeneity matter for increasing the innovative capability of industrial districts.
Introduction
Industrial districts (IDs) are geographically defined production systems, characterized by a large number of small and medium sized firms that are involved at various phases in the production of a homogeneous product family. These firms are highly specialized in a few phases of the production process, and integrated through a complex network of inter-organizational relationships. 1
The reasons underlying the ID competitiveness have been deeply studied in the related literature by adopting different theoretical perspectives coming from many research streams, namely regional economics, economic geography, political economy, and industrial organization. In particular, studies of regional scientists have explained the advantages of firms localized within IDs by using the notion of “localization economies,” which are benefits mainly due to the presence of specialized input suppliers, a local pool of specialized labor skills, and specialized knowledge concerning the particular industry (Becattini, 1990; Marshall, 1920). Studies on industrial economics have highlighted the reduction of the transactional costs on account of the geographical proximity of firms and informal and face-to-face contacts among them as one of the most important benefits for local firms (Mariotti, 1989). Studies on innovation management have pointed out that IDs found the competitive success on their innovative capability, which is on account of the presence of high specialized technical competencies, the existence of networks of formal and informal relationships, and the geographical proximity that creates an environment wherein information, codes, languages, routines, strategies, and knowledge are easy to be transferred and shared (Cainelli et al., 2007; Cooke and Morgan, 1998; Henry and Pinch, 2002; Lundvall and Johnson, 1994; Storper, 1997).
In recent years, where innovation and creative capacity are considered essential determinants of economic prosperity, and knowledge has become the most important production factor, a number of studies, both in the field of strategic management and in the field of regional economics (Grant, 1996; Krugman, 1991; Maskell and Malmberg, 1999; Saxenian, 1996), have widely recognized that the sustainable competitive advantage is strictly related to the ability of firms, and indeed IDs, to increase their innovative capability. In line with this, some scholars have rethought the ID production model shifting their attention from the cost-based benefits to the knowledge-based benefits. These works have proposed a knowledge-based theory of IDs (Maskell, 2001; Maskell and Malmberg, 2004), by investigating the nature of local knowledge (Tallman et al., 2004), the frequency and the effectiveness of the knowledge transfer processes among ID firms (Gordon and McCann, 2000; Mesquita, 2007), and the learning processes activated within IDs (Albino et al., 2005; Maskell, 2001).
By adopting a knowledge-based perspective, the innovative capability of a system depends on its ability to create new knowledge by both acquiring complementary knowledge owned by external sources of knowledge, and combining such external knowledge with its internal knowledge (Coombs and Hull, 1998; Leonard and Sensiper, 1998; Nonaka and Takeuchi, 1995). According to this perspective, the key source of IDs' competitive advantage is strictly related to their innovative capability, which in turn is due to their ability to have access to different sources of knowledge and to activate processes of knowledge transfer and creation.
Despite this wide literature, questions about how the IDs' innovative capability can be increased, how important is the ID technical and scientific knowledge, and how balanced should the local knowledge be with the external knowledge, are minimally addressed in the literature. Answering these questions seems to be particularly critical in the highly innovative and fast changing scenario in which firms operate and compete, where successful innovation depends more and more on the combination of different and heterogeneous pieces of knowledge.
Furthermore, questions around the role of the heterogeneity for the ID growth and innovativeness are still opened, both in the traditional and in the more recent literature (Beaudry and Schiffauerova, 2009; De Groot et al., 2009). In particular, the traditional literature (Jacobs, 1969; Glaeser et al., 1992), still pose the following questions: what kind of externalities – Marshall-Arrow-Romer (MAR) externalities or Jacobs externalities – is really the driving force of the local competitive advantage? Is the heterogeneity in the local industrial composition important for success and innovativeness? What types of IDs are more innovative, those specialized in a single industry or those spread across different industries?
Even in the latest works, in which IDs are interpreted as “cognitive systems” – namely social-productive systems in which knowledge, social experiences, mental models, and collective beliefs are accumulated in a specific space through time (Becattini and Rullani, 1996; Belussi and Pilotti, 2002) – the relevance of heterogeneity for the ID competitive advantage is an open question. In fact, heterogeneity in firms' competencies and capabilities evokes the notion of cognitive and technological proximity (Boschma, 2005; Schamp et al., 2004). The level of cognitive and technological proximity characterizing a system affects the effectiveness of inter-organizational knowledge exchanges and learning processes. In particular, it is widely recognized that too little cognitive and technological proximity increases the difference between the cognitive maps and technological capabilities of two firms and then decreases the capacity of one firm to identify, interpret, and exploit the knowledge possessed by the other firm (Cohen and Levinthal, 1990). However, too much cognitive and technological proximity may not be effective for learning and innovation, as it means a lack of novelty. As a consequence, a natural question arises: is there an optimal value for the level of heterogeneity in firms' competencies and capabilities within an ID?
Spurred into filling the gap of the literature and contributing to the ongoing debate, this chapter analyzes the role of knowledge and knowledge heterogeneity for the innovative capability of Italian IDs.
Drawing on various streams of the literature, the study tests whether the innovative capability of the Italian IDs is positively related both with the amount and the heterogeneity of the local knowledge, and the ID absorptive capacity (Asheim 2007; Asheim et al., 2011; Berliant and Fujita, 2011; Boschma and Iammarino, 2009). To do this, an econometric analysis on 32 Italian district provinces over the period 2000–08 is applied.
The analysis here is of relevance to scholars who are interested in either ID- or firm-level outcomes. At the ID level, the study suggests ways to improve IDs' innovative capability. This knowledge is invaluable, considering the importance of regional innovativeness for a nation's competitive advantage (Porter, 1998; 2000). At the firm level, the investigation provides a better understanding of factors that enhance firms' knowledge creation. This is crucial since knowledge is considered to be a firm's most important competitive asset (Teece and Pisano, 1994; Winter, 1987).
The chapter is organized as follows. The next section provides the theoretical framework and develops the research hypotheses. The following section illustrates the methodology and the data employed to empirically test the hypotheses. The fourth section reports and discusses the econometric findings, while the final section concludes with summary remarks and suggestions for future research.
Knowledge and industrial districts
Knowledge-based industrial district competitive advantage
Recently, strategic management literature has pointed out that in today's economy the source of sustainable competitive advantage for firms cannot be limited to cost and differentiation advantages, and it has recognized the importance of knowledge as a fundamental factor in creating economic value and competitive advantage for firms (Barney, 1991; Grant, 1996; Leonard-Barton, 1995). What a firm knows, how it uses what it knows, and how fast it can develop new knowledge are key aspects for firm success (Hamel and Prahalad, 1994; Prusak, 1997). Therefore, knowledge is a key asset for competing firms and, consequently, learning is a key process. This, in fact, increases the firm's cognitive capital (knowledge stock).
These new strategic management theories have forced new studies in the field of economic geography and regional economics. In the last two decades, some scholars have rethought the ID production model, shifting their attention from competitive advantages that are based purely on lower costs to knowledge-based competitive advantages. These works have proposed a knowledge-based theory of IDs (Cooke, 2002; Etzkowitz and Klofsten, 2005; Maskell, 2001; Maskell and Malmberg, 2004) by investigating: the nature of knowledge circulating in IDs (Tallman et al., 2004), the role of physical proximity for the diffusion of local knowledge (Gerosky, 1995), the effect of knowledge spillovers and collective learning in local innovation processes (Audretsch and Vivarelli, 1994; Capello and Faggian, 2005), the effectiveness of the knowledge transfer processes among ID firms (Gordon and McCann, 2000; Mesquita, 2007), the learning processes activated by firms within IDs (Albino et al., 2005; Maskell, 2001), and the conditions to create “learning and innovating regions” (Etzkowitz and Klofsten, 2005).
According to these studies, the sustainable competitive advantage of IDs lies in their superior capacity to support processes of knowledge accumulation and creation, and to facilitate innovation.
The role of knowledge on the industrial district innovative capability: a theoretical framework
Using a knowledge-based perspective, IDs can be considered as cognitive systems (Maskell, 2001) with a knowledge-base that is the combination of information, know-how, skills, and capabilities owned by local firms and individuals. The ID's knowledge-base is not static but continually evolves and increases because of processes of knowledge creation, acquisition, combination, and diffusion.
In particular, new technological knowledge is created by R&D activities carried out by firms, university, and research centers. Yet, new knowledge results by processes of learning by doing and by using, which give rise to incremental improvements of the existing technological knowledge-base. Processes of knowledge acquisition and combination are activated by means of formal and informal interactions established both within the network of relationships among actors located inside the ID and through relationships with actors located outside the ID. In the first case, the consequent learning process (learning by interaction) is based on knowledge sources external to the firm but internal to the geographical area in which the ID is localized. In the second case, the knowledge sources are external both to the firm and to the geographical area. The spatial diffusion of knowledge is mainly based on knowledge spillovers and its effectiveness increases when the geographical proximity of firms increases (Anselin et al., 1997). All the above knowledge generating processes affect the competitiveness and foster the economic growth of IDs by increasing the IDs' knowledge base and then their innovative capability (Asheim, 1996; Tallman et al., 2004).
Drawing on the above, it is argued that the innovative capability of an ID rises with the amount of ID knowledge; this brings us to the following hypothesis:
Hp1: The higher the amount of the ID knowledge, the higher will be the ID innovative capability.
However, numerous studies show that IDs vary widely with respect to their competitive advantage.
In line with a knowledge-based approach, such a variance among IDs is explained by the fact that the accumulation of knowledge per se is not sufficient to create competitive advantage. In the current climate of rapid technological change and increased competition, no firm can rely entirely on its internal knowledge capacities and sources to develop successful innovations, and that innovative performance is strictly associated with the heterogeneity of knowledge present in the ID (Prahalad and Hamel, 1990; Teece and Pisano, 1994). Knowledge heterogeneity refers to the variety of knowledge, know-how, and expertise to which a system has access. Exposure to heterogeneous knowledge should improve both the creative potential of the ID as well as its ability to develop innovation (Rodan and Galunic, 2004).
According to regional scientists, the competitive differences among IDs are explained by the different types of agglomeration externalities (Glaeser et al., 1992; Saviotti, 1996): Marshall (1920)–Arrow (1962)–Romer (1986) (MAR) externalities, and Jacobs (1969) externalities). Where MAR externalities are on account of the concentration of a particular industry within a specific geographic region, Jacobs externalities regard inter-industry spillovers as the results of the agglomeration of different industries within an urban region. The former facilitates knowledge spillovers across firms and the latter fosters innovation because the heterogeneity of available local knowledge sources will spark creativity, new ideas, and innovations. In particular, as argued by Jacobs (1969) and confirmed by further empirical studies (Glaeser et al., 1992; Duranton and Puga, 2000), only in a context of industrial diversity, rather than industrial specialization, does the exchange of complementary knowledge lead to cross-fertilization of ideas and new knowledge combinations, which in turn favors innovation.
As a consequence of these reflections, the heterogeneity of ID's knowledge-base can be considered as an important factor to explain the competitive success of IDs based on their superior innovative capability.
Drawing on the above, it is argued that the innovative capability of an ID rises with the heterogeneity of the ID knowledge. This heterogeneity derives from the complementarity of knowledge within the ID, and it is fed by the amount of knowledge coming from external knowledge sources (Asheim, 2007; Asheim et al., 2011; Berliant and Fujita, 2011; Boschma and Iammarino, 2009).
These lead the authors to formulate the following hypotheses:
Hp2: The higher the complementarity of knowledge within the ID, the higher will be the ID innovative capability.
Hp3: The higher the amount of the external knowledge brought into the ID, the higher will be the ID innovative capability.
The above reflections open a further issue concerning the ability of an ID to manage the knowledge heterogeneity. In this regard, the literature on knowledge management and organizational learning highlights that having access to a great amount of knowledge flowing into a system (whether it is an organization or a region) is not per se a sufficient condition for building new and complementary knowledge that increases the heterogeneity of the system knowledge base. Although new and complementary knowledge is important for an organization, its identification, acquisition, and, above all, its implementation are by no means simple processes and depend on the “absorptive capacity” of the organization (Veugelers, 1997). Absorptive capacity is the ability of any firm and region to acquire, assimilate, adapt, and apply new knowledge – that is, to learn (Tallman et al., 2004; Zahra and George, 2002).
Cohen and Levinthal (1990) argue that investment in R&D activities produces not only new knowledge but contributes also to the absorptive capacity of the firm by increasing the skills of the employees who have been involved in the R&D process. These stocks of skills or of prior knowledge determine the ability to assimilate and utilize external knowledge. Tripsas (1997) finds that a combination of internal investment in absorptive capacity of the firm, and an external communication infrastructure to facilitate the transmission of external knowledge, successfully enables the firm to integrate knowledge outside its boundaries. Rothwell and Dodgson (1991) emphasize the importance for small and medium sized enterprises in having highly qualified technical specialists, scientists, and engineers in order to access external knowledge.
A fourth hypothesis is formulated in order to consider the effect of the absorptive capacity as follows:
Hp4: The higher the absorptive capacity of the ID, the higher will be the ID innovative capability.
The empirical analysis
The data
The data set is represented by the 32 Italian district provinces (DPs). According to Becattini and Coltorti (2004), the DPs are those provinces that satisfy the following requirement: the percentage of manufacturing employees working in firms with fewer than 250 employees must be higher than the national average.2 In other words, DPs are geographical areas whose production system is predominately that of the ID, strongly characterized by the presence of small and medium sized firms operating in the manufacturing sectors. We choose the province as territorial unit of analysis instead of the local labor systems (LLSs), because provinces provide a more suitable level of analysis in terms of data availability.3
Having specified the territorial unit of analysis, it is worth explaining the industrial sector classification used in the empirical analysis. The 14 manufacturing industries have been grouped into four sectors of activity: food, household and personal goods (textile and clothing, leather and footwear, furniture), mechanics, and heavy industry (paper products, chemicals, rubber, transport equipment). In Table 7.1, each sector of activity is defined according to the two-digit NACE classification.4
To each DP (or ID) have been associated one ore more sectors of activity on the basis of the manufacturing specialization of the IDs located inside the ID. Table 7.2 summarizes the main characteristics of the IDs.
To build our dataset we used different statistical sources: the Italian Office of National Statistics (ISTAT) databases; the European Patent Office (EPO) database; the Ministry of University and Research (MIUR) database, and the UNIONCAMERE database. Data for import and export are provided by the Coeweb database from ISTAT; population and R&D investment come from ISTAT databases, data on employment in two-digit NACE codes are provided by the ASIA database from ISTAT, data on graduates and university are extracted by the MIUR database, data on firms in two-digit NACE codes come from the UNIONCAMERE database, finally data on patents are provided by the Intellectual Property Office (IPO) database.
Data have been collected for each ID and refer to the period 2000–08.
The econometric model
In order to estimate the relationship between regional innovation capability and its determinants, we apply a non-linear estimator, i.e. negative binomial. The reason for choosing such estimator is because of the special feature of our dependent
| Sector of activity | NACE two-digit codes |
| Household and personal goods | DB17–18, DC19, DD20, DN36, DI26 |
| Mechanics | DK29, DL30–31–32–33, DJ28 |
| Heavy industry | DJ27, DF23, DE21–22, DG24, DH25, DM34–35 |
| Food | DA15–16 |
Source: ISTAT data.
variable. The dependent variable patent application is a count data which is considerably over-dispersed because the sample variance is 43 times the sample mean, as reported in Table 7.4. In order to handle this situation, the literature suggests several models, such as negative binomial, zero-inflated negative binomial, and hurdle models (Cameron and Trivedi, 2008).5 The dependent variable has only one zero value (out of 287 observations). Therefore, the zero-inflated models are not necessary, intuitively. Even if there would be the excess of the zero in our data, it does not necessarily mean that zero-inflated models can be the best option (Cameron and Trivedi, 2008), since it must be possible to distinguish between “true zeros” and “excess zeros” in order to be reasonable to use zero-inflated models. The mechanism for distinguishing these two types of zero is not clear in the patent application data, hence the use of zero-inflated models seems to be implausible.6 Nevertheless, in order to be sure, we perform the Vuong test of zero-inflated negative binomial vs. (standard) negative binomial. The test is slightly in favor of negative binomial. The hurdle model is not a preferred model, too, for the same reason as for the zero-inflated models. The formulation for the innovation capability of ID r in year t is written as follows:
 |
(7.1) |
where
and yrt is the number of patent applications in ID r in year t, X1rt is the vector of the amount of the ID knowledge variables (both internal and external variables), X2rt is the variable of the ID knowledge heterogeneity variables, X3rt is the absorptive capacity variable, Zrt represents the control variables, and σrt is normally assumed to have a gamma distribution with mean 1 and variance alpha, which can be estimated from the data. Alpha is the over-dispersion parameter, which corrects for the over-dispersion by adjusting the variance independently from the mean (Cameron and Trivedi, 2008). We will use both pooled and panel application (random effect) of the negative binomialmodel in the subsequent analyses, while the panel application is the preferred one (more elaboration is provided in Table 7.5).7
The dependent variable
As the phenomenon under investigation is the innovativeness of Italian IDs, we use the number of patents developed in each ID during the period 2001–08 as an indicator of innovative performance (see Jaffe and Trajtenberg, 2005; Acs et al., 2002). Patents have been found to be a good proxy of innovative activity in general (Griliches, 1990) and for regional-level analysis in particular (Acs et al., 2002).8 This is because patents are granted for inventions which are novel, inventive, and have industrial application (Andersson and Lööf, 2011).
Note that in order to exclude endogeneity problems, the independent variables have been measured in the period 2000–07, lagged of one year in respect of the dependent variable.
The independent variables
The independent variables considered in the model refer to the conceptual framework illustrated in the second section. Therefore, they have been classified as proxies for the amount of the IDs' knowledge, the IDs' knowledge heterogeneity, and the IDs' absorptive capacity.
Amount of the IDs' knowledge
The amount of IDs' knowledge has been measured by the amount of R&D investments. The variable has been constructed by disaggregating at the province level the data available at the regional level. In particular, R&D by ID was assigned from regional data on the basis of the regional R&D investments per employee in each institutional sector (business sector, universities, and public administration) and multiplied by the employees in each ID (Boix and Galletto, 2009). Therefore, the value for each province has been computed by:
 |
(7.2) |
where 
is the amount of R&D investments in Region k where the District Province i is located, Ek is the total number of employees in Region k, and Ei is the total number of employees in ID i.
R&D expenditure has been frequently used as a proxy for the local capability to generate new knowledge. Cohen and Klepper (1991, 1992) point out that the greatest source generating new economic knowledge is generally considered to be R&D.
A critical problem associated to this variable is that R&D activities produce formal and codified knowledge that can be easily protected by firms (Brouwer and Kleinknecht, 1999) and cannot then be considered as a public good that increases the IDs' knowledge base. However, empirical studies prove the existence of R&D knowledge spillovers within geographically bounded areas. For example, Jaffe (1989), Acs et al. (1992), and Audretsch and Feldman (1996) found that the knowledge created through R&D activities spills over to contribute to the generation of new technical knowledge.
Having acknowledged this point, according to the conceptual framework, this variable is expected to impact positively on the IDs innovative capability.
ID knowledge heterogeneity
The ID knowledge heterogeneity has been captured by using two variables intended to explain two different sources of knowledge heterogeneity: one internal to the ID, related to the complementarity of knowledge within the ID, and the other external to the ID, because of the amount of knowledge coming from external knowledge sources.
The complementarity of knowledge within the ID has been associated to the degree of production diversity of the ID. Such production diversity has been measured, as in previous empirical studies (Greunz, 2004; van Oort, 2002; Paci and Usai, 1999; Feldman and Audretsch, 1999), by concentration measures. In particular, this study uses the Gini coefficient. The index aims at capturing the level of concentration of employees in a specific manufacturing sector.
The Gini coefficient is defined as follows:
 |
(7.3) |
where i indexes the manufacturing sector (i = 1, …, n), classified into the two-digit NACE codes,
 |
and |
Here 
is the cumulative sum of employees in each manufacturing sector, classified into the two-digit NACE codes, when the sector employment is ordered in increasing order and CE is the total number of employees.
The Gini coefficient ranges from a minimum value of zero to a theoretical maximum of one. Values of G_prod close to zero indicate that ID firms are specialized into different manufacturing sectors, in contrast values of G_prod close to one indicate that the ID manufacturing specialization is concentrated in very few sectors. Therefore, the G_prod aims at capturing Jacobs externalities; it decreases together with the production diversity of the firms. The lower is G_prod, the higher is the complementarity of knowledge within the ID, thus the higher is the ID knowledge heterogeneity.
The amount of external knowledge that is brought into the ID has been proxied by using the amount of the international trade linkages of each ID (Boschma and Iammarino, 2009). The variable has been calculated by the amount of imports and exports – expressed in monetary terms and weighted on the number of employees – recorded by each ID:
 |
(7.4) |
where Importi and Exporti are the amount of imports and exports of the ID i, respectively, and CE r is the number of employees in ID r.
Trade indicators have been traditionally very important in assessing knowledge flows in open economic systems, particularly when there is an emphasis on extra-regional linkages. Boshma and Iammarino (2009) used the degree of export and import diversification to measure the degree of knowledge variety into a region. Camagni (1991) pointed out that external connections bring new knowledge into a region. In this way, sectoral lock-in at the regional level may be counterbalanced by the inflow of a high degree of variety of knowledge through inter-regional relationships.
By using the international trade linkages of each ID as a proxy of the external knowledge brought into the ID, the higher the value of Ext_Knowledge, the greater the external knowledge that flows into the ID, and thus the higher the ID knowledge heterogeneity.
According to the conceptual framework, both the above variables are expected to impact positively on the ID's innovative capability.
Absorptive capacity
The ID absorptive capacity has been proxied by a variable expressing the availability of high-educated human resources within the ID. Namely, the variable has been measured by the incidence of graduates in technical–scientific fields on the total population of the ID:
 |
(7.5) |
where Graduatesr is the number of graduates in technical–scientific fields in the ID r and Populationr is the population of the ID r.
Although the proxy may appear rather rough, it takes into account an important aspect of IDs, namely the large presence of small and medium sized firms specialized in the low-tech sector that build up their absorptive capacity mainly on high-educated individuals working in the organization (Rothwell and Dodgson, 1991) rather than on R&D activities. In line with this, Mangematin and Nesta (1999) argue that high-educated employees, naturally by their daily tasks, will increase the stock of knowledge in the organization. Carter (1989) argues that higher educated employees are the main contributors of know-how trading on account of the high level of knowledge embodied in these people. This statement is further supported by Guellec (1996) who emphasizes skilled labor to be in a better position to generate new knowledge because they master the state of the art and thus are better able to manage new technology.
According to the conceptual framework, the variable is expected to impact positively on the ID's innovative capability.
Table 7.3 summarizes all the independent variables showing the used measure and the data source.
The control variables
The analysis includes several control variables. First, in order to take into account the general economic conditions of the district areas, the authors used a dummy for IDs localized in the North; that is a more advanced and developed area (Mariotti et al., 2008). Two other controls have been considered for factors that may affect the ID innovative capability through production of new knowledge. Specifically, the first variable is a density index that measures the geographic proximity among firms and it is considered as a factor that facilitates spillovers and the growth of knowledge (Audretsch and Feldman, 1996; Carlino et al., 2001, 2006; Knudsen et al., 2007). The index has been measured as the ratio between the population of manufacturing firms and the ID land area.
where 
is the number of manufacturing firms in ID r and Ar is the land area covered by ID r.
The second variable refers to the presence of universities within the district area, which are considered as sources of new knowledge (Benneworth and Hospers, 2007; Cooke and Piccaluga, 2004).
Results
Table 7.4 contains the descriptive statistics and the correlation matrix of the variables. The dependent variable patent application is considerably over-dispersed
because the sample variance is 43 times the sample mean for this variable. Hence, there is the lack of equidispersion for this variable, which violates the assumption of the Poisson model. The formal test of the significance of over-dispersion parameter (Alpha), reported in Table 7.5, provides further evidence for over-dispersion of our dependent variable. All explanatory variables are positively correlated with the dependent variable, except G_prod, which has a negative correlation as expected.
Given the high correlation value of some independent variables, we perform the variance inflation factor (VIF) test to check for multicollinearity between independent variables. The VIF score for each variable is well below five (and the mean VIF is 2.37), hence we can expect that multicollinearity does not substantially bias the regression results.9
For the dependent variable and most of the regressors, the vast majority of the variation in the data consists of the between-variation rather than the withinvariation. Therefore, the fixed-effects estimator may not be very efficient, since it relies on within-variation, and therefore we use the random-effects estimator in our panel models.10
The results of negative binomial estimation of determinants of patent application for Italian IDs over the period 2001 to 2008 are reported in Table 7.5. Both pooled and panel regressions (RE) are performed, while the panel models are the preferred ones, as discussed before. We add one explanatory variable in each model (column), while controlling for the set of control variables in all the models. This is done first for pooled models (columns 1 to 4) and then for panel models (columns 5 to 8). The likelihood-ratio (LR) test versus pooled is always in favor of panel models, hence the preferred models are columns 5 to 8. In the first model (column) we start with the amount of the ID knowledge, measured by the intensity of R&D activities. This variable is positive and significant and the results remained almost robust in the later models, especially for panel models.
In the second model G_prod is introduced. It is significant and negative and the results remained robust in all the later models. This is what we expected, as the lower G_prod means the higher the production diversity within an ID. In the third model, we introduce the variable used as proxy of the amount of external knowledge that is brought into the ID (Ext_knowledge). This variable is positive and highly significant, which remained robust in all the later models. The result confirms our third hypothesis: the greater is the external knowledge that flows into the ID, the higher will be the ID innovative capability. Finally in the fourth model, we include absorptive capacity of IDs and this variable also shows positive and significant effect on patent application of Italian IDs.
Among control variables, all of them are positive, while density and north dummy show more robust behavior in terms of significance. They are positively and significantly associated with the patent applications, which is in line with previous literature.
Describing the full model in the preferred estimator (model 8), all the determinants of patent application are significant and showing the expected sign.
Since we used the maximum likelihood estimation (MLE), in order to compare the models with each other, we used Bayesian information criterion (BIC) and Akaike information criterion (AIC). Both criteria get smaller when we move from model 1 to 4 (pooled models) and when we moved from model 5 to 8 (panel models). This means that by gradually adding the variables from model 1 to 4 (and 5 to 8), the models are getting better in terms of fitness, while there is no evidence of over-fitting. This is also obtained by LR test of restricted vs. unrestricted models, which is always in favor of unrestricted models. In other words, internal knowledge, external knowledge, and absorptive capacity of the region all together can produce the better fit for modeling the patent application in comparison with including only one or two of them. The values in panel models are lower than their counterparts in pooled models, indicating a better fit of panel models.
The alpha, over-dispersion parameter, is reported in Table 7.5 too, for the pooled models. When the over-dispersion parameter is zero, the negative binomial distribution is equivalent to a Poisson distribution. In our case, however, alpha is significantly different from zero in all (pooled) models and thus shows again that the negative binomial is a preferred estimation strategy over the Poisson or zero-inflated Poisson models.
Conclusions
This chapter seeks to contribute to the ongoing debate concerning the role of knowledge and knowledge heterogeneity for the innovative capability of IDs. With this aim, using a knowledge-based approach, the study tests whether the innovative capability of IDs is positively related both with the amount and the heterogeneity of the ID knowledge, and the ID absorptive capacity.
Four theoretical hypotheses concerning the effects of knowledge and knowledge heterogeneity on the ID innovativeness are formulated. To test the hypotheses, an econometric analysis on 32 Italian DPs is applied.
The chapter's findings confirm the hypothesized positive relation between knowledge and the ID innovative performance. In particular, results confirm our first hypothesis regarding the positive relationship between the availability of knowledge and the ID innovative performance. Concerning ID knowledge heterogeneity, we test the effects of both sources of knowledge heterogeneity, namely the internal source, related to the complementarity of knowledge within the ID, and the external one, due to the amount of knowledge coming from external knowledge sources. We find that ID knowledge heterogeneity positively affects the ID innovative capability, whether it is because of production diversity of the ID firms whether it is because of the external knowledge brought into the ID.
Finally, results highlight the important role played by the absorptive capacity of the ID, confirming that the higher the absorptive capacity of the ID, on account of the higher density of graduates in technical–scientific fields, the higher will be the ID innovative capability.
In addition to this consideration, it is interesting to note that the G_prod variable, which measures the ID knowledge heterogeneity due to the complementarity of knowledge within the ID, has the highest magnitude in comparison with other variables. This is a remarkable outcome of our analysis for two main reasons. First, it is a counterintuitive result with respect to the traditional studies that recognize that R&D investment has the major driver of patent applications. Second, since the G_prod variable captures Jacobs externalities, our result emphasizes the important role played by externalities that may come from a diversity of related industries in IDs (Frenken et al., 2007). Thus, the more variety across related sectors in an ID, the higher the number of technologically related sectors, and the more learning opportunities there are for local industries. This will result in more inter-sectoral knowledge spillovers, which enhance the ID innovative performance.
Notes
1 The industrial district is a specific type of geographical cluster. The latter is defined by Porter (1998) as a geographically proximate group of interconnected companies and associated institutions (for example universities, standards agencies, and trade associations) in particular fields, linked by commonalities and complementarities. Geographical clusters also promote knowledge sharing, learning processes, and innovation development. Therefore, the topic we investigate in the chapter is of interest not only in Italy but also in other countries.
2 Becattini and Coltorti (2004) have classified the 103 Italian provinces into four groups: provinces of large firms, district provinces, residual provinces, and mixed provinces.
3 In our study, DPs (district provinces) is the same as IDs (industrial districts).
4 NACE is the statistical classification of economic activities in the European Community.
5 Since the mean and variance are not equal, the estimations based on Poisson and Zero-inflated Poisson models are not the preferred options.
6 An example of a situation where it is possible to distinguish between true zeros and excess zeros is when we want to explain the amount of cigarettes smoked per day, while we have a survey containing both smokers (causing true zeros) and non-smokers (causing excess zeros) (Cameron and Trivedi, 2008: 584).
7 The Hausman test reveals that the difference in coefficients between random effect and fixed effect are not systematic.
8 Acs et al. (2002) compared innovation and patents across US regions and concluded, “the empirical evidence suggests that patents provide a fairly reliable measure of innovative activity” (p. 1080).
9 The VIF test is performed after the conventional OLS regressions. There is no formal threshold for VIF score, but as a rule of thumb the VIF score below 10 is said to be the evidence of quite mild multicollinearity.
10 Nevertheless, the formal Hausman test reveals that the difference in coefficients between random effect and fixed effect are not systematic.
References
Acs, Z. J., Audretsch, D. B., and Feldman, M. P. (1992), “Real Effects of Academic Research,” American Economic Review, 82: 363–7.
Acs, Z. J., Anselin, L., and Varga, A. (2002), “Patents and Innovation Counts as Measures of Regional Production of Knowledge,” Research Policy, 31: 1069–83.
Albino, V., Carbonara, N., and Giannoccaro, I. (2005), “Industrial Districts As Complex Adaptive Systems: Agent-Based Models Of Emergent Phenomena,” in C. Karlsson, B. Johansson and R. Stough (Eds), Industrial Clusters and Inter-Firm Networks, Northampton: Edward Elgar Publishing: 73–82.
Andersson, M. and Lööf, H. (2011), “Small Business Innovation: Firm Level Evidence from Sweden,” Journal of Technology Transfer, DOI: 10.1007/s10961-011-9216-9.
Anselin, L., Acs, Z. J., and Varga, A. (1997), “Local Geographic Spillovers Between University Research and High Technology Innovations,” Journal of Urban Economics, 42: 422–48.
Arrow, K. (1962), “The Economic Implication Of Learning By Doing,” Review of Economic Studies: 29.
Asheim, B. T. (1996), “Industrial Districts as ‘Learning Regions’: A Condition for Prosperity,” European Planning Studies, 4 (4): 379–400.
Asheim, B. T. (2007), “Differentiated Knowledge Bases and Varieties of Regional Innovation Systems,” The European Journal of Social Sciences, 20 (3): 223–41.
Asheim, B. T., Boschma, R., and Cooke, P. (2011), “Constructing Regional Advantage: Platform Policies Based on Related Variety and Differentiated Knowledge Bases,” Regional Studies, 45 (7): 893–904.
Audretsch, D. and Vivarelli, M. (1994), “Small Firms and Spillovers: Evidence from Italy,” Small Business Economics, 8: 249–58.
Audretsch, D. B. and Feldman, M. P. (1996), “R&D Spillovers and the Geography of Innovation and Production,” The American Economic Review, 86 (3): 630–40.
Barney, J. B. (1991), “Firm Resources and Sustained Competitive Advantage,” Journal of Management, 17: 99–120.
Beaudry, C. and Schiffauerova, A. (2009), “Who's Right, Marshall or Jacobs? The Localization Versus Urbanization Debate,” Research Policy, 38: 318–37.
Becattini, G. (1990), “The Marshallian Industrial District as a Socio-Economic Notion,” in F. Pyke, G. Becattini and W. Sengenberger (Eds), Industrial Districts and Inter-firm Co-operation in Italy, Geneva: International Institute for Labour Studies: 37–51.
Becattini, G. and Rullani E. (1996), “Local Systems and Global Connections: The Role of Knowledge,” in F. Cossentino, F. Pyke and W. Sengenberger (Eds), Local Regional Response to Global Pressure: The Case of Italy, Geneva: Ilo.
Becattini, G. and Coltorti, F. (2004), “Aree di Grande Impresa ed Aree Distrettuali Nello Sviluppo Post-Bellico dell'Italia: un'esplorazione Preliminare,” Rivista Italiana degli Economisti, 1: 61–102.
Belussi, F. and Pilotti, L. (2002), “Learning and Innovation by Networking Within The Italian Industrial Districts: The Development of an Explorative Analytical Model,” Sinergie, 20 (58): 3–43.
Benneworth, P. and Hospers, G. J. (2007), “The New Economic Geography of Old Industrial Regions: Universities as Global-Local Pipelines,” Environment and Planning C: Government and Policy, 25: 779–802.
Berliant, M. and Fujita, M. (2011), “The Dynamics of Knowledge Diversity and Economic Growth,” Southern Economic Journal, 77 (4): 856–84.
Boix, R. and Galletto, V. (2009), “Innovation and Industrial Districts: A First Approach to the Measurement and Determinants of the I-District Effect,” Regional Studies, 43 (9): 1117–33.
Boschma, R. and Iammarino, S. (2009), “Related Variety, Trade Linkages, and Regional Growth in Italy,” Economic Geography, 85 (3): 289–311.
Boschma, R. A. (2005), “Proximity and Innovation: A Critical Assessment,” Regional Studies, 39: 61–74.
Brouwer, E. and Kleinknecht, A. (1999), “Innovative Output, and a Firm's Propensity to Patent. An Exploration of CIS Micro Data,” Research Policy, 28: 615–24.
Cainelli, G., Mancinelli, S., and Mazzanti, M. (2007), “Social Capital and Innovation Dynamics in District-Based Local Systems,” Journal of Socio-Economics, 36 (6): 932–48.
Camagni, R. (Ed.) (1991), Innovation Networks. Spatial Perspectives, London, New York: Bellhaven Press.
Cameron, A. C. and Trivedi, P. K (2008), Microeconometrics: Methods and Applications, New York: Cambridge University Press.
Capello, R. and Faggian, F. (2005), “Collective Learning and Relational Capital in Local Innovation Processes,” Regional Studies, 39 (1): 75–87.
Carlino, G., Chaterjee, S., and Hunt, R. (2001), “Knowledge Spillovers and the New Economy of Cities,” Working Paper No. 01–14, Philadelphia, PA: Federal Reserve Bank of Philadelphia.
Carter, A. P. (1989), “Know How Trading as Economic Exchange,” Research Policy, 18: 1–9.
Cohen, W. M. and Levinthal, D. A. (1990), “Absorptive Capacity: A New Perspective on Learning and Innovation,” Administrative Science Quarterly, 35: 128–52.
Cohen, W. M. and Klepper, S. (1991), “Firm Size Versus Diversity in the Achievement of Technological Advance,” in Z. J. Acs and D. B. Audretsch (Eds), Innovation and Technological Change: An International Comparison, Ann Arbor: University of Michigan Press: 183–203.
Cohen, W. M. and Klepper, S. (1992), “The Trade-Off Between Firm Size and Diversity in the Pursuit of Technological Progress,” Small Business Economics, 4 (1): 1–14.
Cooke, P. (2002), Knowledge Economies, London: Routledge.
Cooke, P. and Morgan, K. (1998), The Associational Economy. Firms, Regions, and Innovation, Oxford: Oxford University Press.
Cooke, P. and Piccaluga, A. (Eds) (2004), Regional Economies as Knowledge Laboratories, Cheltenham: Edward Elgar Publishing.
Coombs, R. and Hull, R. (1998), “Knowledge Management Practices and Path-Dependency in Innovation,” Research Policy, 27 (3): 237–53.
De Groot, H. L. F., Poot, J., and Smit, M. J. (2009), “Agglomeration Externalities, Innovation and Regional Growth: Theoretical Perspectives and Meta-Analysis,” in Capello, R. and Nijkamp P. (Eds), Handbook of Regional Growth and Development Theories, Northampton, MA: Edward Elgar Publishing.
Duranton, G. and Puga, D. (2000), “Diversity and Specialisation in Cities: Why, Where and When does it Matter?,” Urban Studies, 37 (3): 533–55.
Etzkowitz, H. and Klofsten, M. (2005), “The Innovating Region: Toward a Theory of Knowledge-Based Regional Development,” R&D Management, 35 (3): 243–55.
Feldman, M. and Audretsch, D. (1999), “Innovation in Cities: Science-Base Diversity, Specialization and Localized Competition,” European Economic Review, 43: 409–29.
Frenken, K., van Oort, F. G., and Verburg, T. (2007), “Related Variety, Unrelated Variety and Regional Economic Growth,” Regional Studies, 41 (5): 685–97.
Geroski, P. (1995), “Innovation and Competitive Advantage,” OECD Economics Department Working Papers, Paris: OECD Publishing: 159.
Glaeser, E., Kallal, H., Scheinkam, J., and Shleifer, A. (1992), “Growth in Cities,” The Journal of Political Economy, 100: 1126–52.
Gordon, I. R. and McCann, P. (2000), “Industrial Clusters, Complexes, Agglomeration and/or Social Networks?,” Urban Studies, 37: 513–32.
Grant, R. M. (1996), “Prospering in Dynamically-competitive Environments: Organizational Capability as Knowledge Integration,” Organization Science, 7 (4): 375–87.
Greunz, L. (2004), “Industrial Structure and Innovation – Evidence from European Regions?,” Journal of Evolutionary Economics, 14: 563–92.
Griliches, Z. (1990), “Patent Statistics as Economic Indicators: A Survey,” Journal of Economic Literature, 28 (4), 1661–707.
Guellec, D. (1996), “Knowledge, Skills and Growth: Some Economic Issues,” STI Review, 18: 17–38.
Hamel, G. and Prahalad, C. K. (1994), Competing for the Future, Boston, MA: Harvard Business School Press.
Henry, N. and Pinch, S. (2002), “Spatializing Knowledge: Placing the Knowledge Community of Motor Sport Valley,” in Huff A. S. and Jenkins, M. (Eds), Mapping Strategic Knowledge, London: SAGE Publications Ltd: 137–69.
Jacobs, J. (1969), The Economy of Cities, New York: Vintage.
Jaffe, A. B. (1989), “Real Effects of Academic Research,” American Economic Review, 79: 957–70.
Jaffe, A. B. and Trajtenberg, M. (2005), Patents, Citations and Innovations. A Window on the Knowledge Economy, Cambridge, MA: MIT Press.
Knudsen, B., Florida, R., Gates, G., and Stolarick, K. (2007), “Urban Density, Creativity, and Innovation,” Working Paper, Toronto: The Martin Prosperity Institute, University of Toronto.
Krugman, P. (1991), “Increasing Returns and Economic Geography,” Journal of Political Economy, 99: 483–99.
Leonard, D. and Sensiper, S. (1998), “The Role of Tacit Knowledge in Group Innovation,” California Management Review, 40 (3): 112–32.
Leonard-Barton, D. (1995), Wellsprings of Knowledge: Building and Sustaining the Sources of Knowledge, Boston: Harvard Business School Press.
Lundvall, B. A. and Johnson, B. (1994), “The Learning Economy,” Journal of Industry Studies, 1: 23–42.
Mangematin, V. and Nesta, L. (1999), “What Kind of knowledge Can a Firm Absorb?,” International Journal of Technology Management, 18 (3/4): 149–72.
Mariotti, S. (1989), “Efficienza Dinamica e Sistemi di Imprese,” Economia e Politica Industriale, 64: 91–123.
Mariotti, S., Mutinelli, M., and Piscitello, L. (2008), “The Internationalization of Production by Italian Industrial Districts' Firms: Structural and Behavioural Determinants,” Regional Studies, 42 (5): 719–35.
Marshall, A. (1919), Industry and Trade, London: Macmillan.
Marshall, A. (1920), Principles of Economics, London: Macmillan.
Maskell, P. (2001), “Towards a Knowledge-Based Theory of the Geographical Cluster,” Industrial and Corporate Change, 10: 921–43.
Maskell, P. and Malmberg, A. (1999), “Localized Learning and Industrial Competitiveness,” Cambridge Journal of Economics, 23: 167–85.
Maskell, P. and Malmberg, A. (2004), “The Elusive Concept of Localization Economies: Towards a Knowledge-Based Theory of Spatial Clustering,” in G. Grabher and W. W. Powell (Eds), Networks, Cheltenham: Edward Elgar Publishing.
Mesquita, L. F. (2007), “Starting OverWhen The Bickering Never Ends: Rebuilding Aggregate Trust Among Clustered Firms Through Trust Facilitators,” Academy of Management Review, 32: 72–91.
Nonaka, I. and Takeuchi, H. (1995), The Knowledge-Creating Company, New York: Oxford University Press.
Paci, R. and Usai, S. (1999), “The Role of Specialisation and Diversity Externalities in the Agglomeration of Innovative Activities,” CRENoS Working Paper, 99/15. Available at: http://crenos.unica.it/crenos/en/node/142 (accessed October 1999).
Porter, M. (1998), “Clusters and the New Economics of Competition,” Harvard Business Review, 76 (6): 77–90.
Porter, M. (2000), “Location, Competition and Economic Development: Local Clusters in a Global Economy,” Economic Development Quarterly, 14 (1): 15–34.
Prahalad, C. K. and Hamel, G. (1990), “The Core Competence of The Corporation,” Harvard Business Review, 68 (3): 79–91.
Prusak, L. (1997), Knowledge in Organizations, Boston: Butterworth-Heinemann.
Rodan, S. and Galunic, C. (2004), “More Than Network Structure: How Knowledge Heterogeneity Influences Managerial Performance and Innovativeness,” Strategic Management Journal, 25: 541–62.
Romer, P. M. (1986), “Increasing Returns and Long-Run Growth,” Journal of Political Economy, 94: 1002–37.
Rothwell, R. and Dodgson, M. (1991), “External Linkages and Innovation in Small and Medium-sized Enterprises,” R&D Management, 21: 125–37.
Saviotti, P. P. (1996), Technological Evolution, Variety and The Economy, Cheltenham: Edward Elgar Publishing.
Saxenian, A. (1996), Regional Advantage. Culture and Competition in Silicon Valley and Route 128, Boston: Harvard University Press
Schamp, E. W., Rentmeister, B., and Lo, V. (2004), “Dimensions of Proximity in Knowledge-Based Networks: The Cases of Investment Banking and Automobile Design,” European Planning Studies, 12: 607–24.
Storper, M. (1997), The Regional World, London: Guilford Press.
Tallman, S., Jenkins, M., Henry, N., and Pinch, S. (2004), “Knowledge, Clusters and Competitive Advantage,” Academy of Management Review, 29: 258–71.
Teece, D. and Pisano, G. (1994), “The Dynamic Capabilities of Firms: An Introduction,” Industrial and Corporate Change, 3 (3): 537–56.
Tripsas, M. (1997), “Surviving Radical Technological Change Through Dynamic Capability: Evidence from the Typesetter Industry,” Industrial and Corporate Change, 6 (2): 341–77.
van Oort, F. (2002), “Innovation and Agglomeration Economies in the Netherlands,” Journal of Economic and Social Geography, 93: 344–60.
Veugelers, R. (1997), “Internal R&D Expenditures and External Technology Sourcing,” Research Policy, 26: 303–15.
Winter, S. (1987), “Knowledge and Competence as Strategic Assets,” in: D. Teece (Ed.), The Competitive Challenge, Cambridge, MA: Ballinger.
Zahra, S. A. and George, G. (2002), “Absorptive Capacity: A Review, Reconceptualization, and Extension,” Academy of Management Review, 27: 185–203.