In the following section, we will review the literature on the determinant of export performance. We will, particularly, focus on the effects of regional clusters and innovation, and set up the main hypotheses of the study.

The participation in international markets represents important motivation means to the nations as the growth of economic activity, job opportunities, increased industrial productivity, profitability increase, and other relevant economic and social impacts (Guan & Ma, 2003; Moghaddam et al., 2012).

The export performance corresponds to a measurement element used by researchers to measure the company's internationalization process (Carneiro, Rocha, & Silva, 2007). Facing a great body of research in recent decades aiming at understanding the factors that led the companies to foreign markets, scholars have found heterogeneous results (Ayan & Percin, 2005; Chugan & Singh, 2014; Moghaddam et al., 2012; Souza, Martínez-Lopez, & Coelho, 2008).

The Figure 1 below synthesizes the main contributions of scholars on the relationships between regional clusters, innovation and export performance. Among these studies, the effects of regional clusters spillovers (Chugan & Singh, 2014; Wheeler, Ibeh, & Dimitratos, 2008) and innovation of firms (Chugan & Singh, 2014; Souza et al., 2008) have been pointed as determinants of export performance. In addition, studies have suggested other determinant factors, like organizational aspects such as company's size (Souza et al., 2008; Wheeler et al., 2008), international experience (Moghaddam et al., 2012; Papadopoulos & Martín, 2010) and the location of firms in greater industrial regional agglomerations (Mittelstaedt, Ward, & Nowlin, 2006).

Figure 1
Empirical Studies of Cluster, Innovation and Export Performance
Source: The authors.

The original discussion of industrial districts (or regional cluster) has its origins in the Alfred Marshall's (1890) book Principles of Economics (1890), who discussed the phenomenon of the agglomeration of specialized industries, as well the trade transactions occurring in such locations.

Marshall's work (1890) on agglomeration of industries offers four important contributions to the debate of industrial districts: (a) primarily, the location of spatial concentration of industries benefits from a pool of workers with specialized skills (access to skilled labor); (b) as a secondary contribution, this spatial agglomeration facilitates the development of specialized inputs and services; (c) provides for business opportunities of technological spillovers; and (d) access to higher market demand.

While the first three types of agglomeration externalities tend to benefit businesses by providing unique and efficient access to the provision of the necessary resources, the last advantage benefits particularly companies by increasing the demand for products or services (Marshall, 1890).

Subsequent to Marshall's work (1890), the effects of clustering were discussed by Alfred Weber (1929) in the paper entitled Theory of the Location of Industries (Weber, 1929). Weber (1929) distinguished three different categories of advantages in determining location of industrial production: (a) transportation costs; (b) labor; and (c) agglomeration.

More precisely, these advantages are divided into three categories. First, they can be generated by firms' economies of scale. Second, the agglomeration can occur because of the proximity to suppliers, a labor market in pool or localized knowledge diffusion. Finally, the concentration of production may give rise to external advantages, such as a highly developed infrastructure, lower energy costs, etc. - that is, which Weber calls urbanization economies (Weber, 1929).

In the economic and business literatures, several designations and concepts have been used to describe and address the phenomenon of economic agglomeration or the geographic concentration of industries: clusters, industrial districts, local cooperation networks, local productive systems, local clusters, Innovative milieu, technology parks, local innovation systems, among others. In some specific cases, the different concepts express different theoretical approaches, and in many times they are uses to describe the same phenomenon, pointing to the local perspective of addressing firms' competitiveness and/or limitations

Alfred Marshall (1890) was one the first economists to highlight the advantages of industrial agglomeration. He suggested that firms can benefit in several ways from agglomeration economies, like the specialization of labor, knowledge spillovers, access to specific resources, which can support firms' growth. Marshall called such advantages as externalities of economic agglomeration.

The industrial concentrations may involve specialized suppliers, related industries and support entities (such as government institutions, governmental organizations, business associations, research centers, etc.) that, together, seek to promote and/or generate a more favorable environment for business enterprises. Thus, in an environment characterized by competition and co-operation, the companies develop various productive activities with the support of public policies and related sectors (Porter, 2009). It is important to emphasize that, under this perspective, cluster's externalities suppose a geographical concentration of specific industry and related industries, as well.

Authors have addressed the location advantages of regional clusters by pointing to different factors. Schmitz (1995) has highlighted the concept of collective efficiency, while Enright (2003) the importance of understanding the cluster development stage to determine the development of promotion policies (Enright, 2003). Altenburg and Meyer-Stamer (1999) have particularly emphasized that the positive economic externalities of regional clusters are driven by low transaction costs and the joint actions of firms. On the other hand, the economic agglomeration can generate a processes of collective learning (cooperation and innovative dynamics) through the join firm actions which may add specific dimensions for the competitiveness of firms to face the dynamics of market competition, as well by spreading information and knowledge to the whole society (Lastres & Cassiolato, 2003). This process can benefit the companies from the same industry, but also can strengthening other related industries operating in the same or closer locations.

The Marshallian industrial districts, characterized by a large number of companies involved in a given environment, aroused further studies of economic thought. In this sense, Becattini (2002) observed the sectorial specialization of locations in Italy highlighting the existing externalities in the formation of a given geographical area. The externalities represent an effect of environmental, social and economic change caused by formation of industrial districts.

In this perspective, Porter (1999) defines the industrial cluster as "a geographically concentrated group of interrelated companies and related institutions in a particular field, linked by common and complementary elements" (p. 211). Therefore, being located in a cluster allows contact with related technological industries, shared infrastructure, major innovations and entrepreneurial activities (Delgado et al., 2010), as well as access to specialized customers and suppliers, skilled labor, universities and institutions (Glaeser & Kerr, 2009).

Maskell, Bathelt and Malmberg (2006) discuss the interaction between companies, customers, suppliers at trade shows and events as means to absorb knowledge and explain the success of companies that are outside of clusters. However, it can be noted that the connection with other external performers is complementary, but does not replace the interactions in regional clusters. In collaboration with this perspective, Glaeser and Kerr (2009) indicate the integration of the cluster agents that promotes the development of new ideas and knowledge transfer. Overall, companies located in clusters hold access to shared resources that enable superior performance compared to companies outside clusters (Molina-Morales & Martínez-Fernández, 2004).

Yet, other studies highlight the advantages and benefits of cluster's resources as learning process, knowledge creation (Maskell & Malmberg, 1999), knowledge network (Giuliani, 2007, 2013), human capital, specialized research tools, suppliers, consumers, manufacturing facilities or services, knowledge spillovers and institutions (McCann & Folta, 2008). However, Hervás-Oliver and Albors-Garrigós (2007) address singularity of resources and capabilities for each cluster, such as skilled labor, social interactions, business sophistication, supplier linkages, network or external linkages, institutional linkages (training and R&D support).

Based on the concepts above discussed, with regards to the advantages and externalities of economic agglomeration that may benefit companies operating inside an industrial cluster, we conducted a classification that summarizes these externalities as specific resources that are available for all the companies in the clusters. We identified three sets of clusters resources: local resources, institutional resources and business network resources. While local resources include professionals, infrastructure and local market; institutional resources represent educational institutions, professionals as well as cluster governance institutions; and business network resources constitute the economic cluster stakeholders such as suppliers, customers, competitors and related companies.

The relationship between access to cluster resources and the export performance has been found positive and statistically significant in several studies (Belso-Martínez, 2006; Fernhaber et al., 2008; Mittelstaedt et al., 2006), but other empirical investigations have come to opposite correlations, and identified negative effects (Diez-Vial & Fernández-Olmos, 2014). In this study, we consider, however, that the access to the local, institutional and network cluster resources can represent a significant source of competitiveness, and, therefore, it may stimulate the firms to establish specific strategies of growth based on international markets.

We, therefore, hypothesized that:

H1: The access to cluster's resources is positively associated with export performance.

Porter (1999) argues that companies located in regional or industrial clusters can benefit from several advantages, like advantage of information, knowledge, technological opportunities, market opportunities, in addition to the near relationship with companies and institutions. In this sense the dynamic process of interaction between firms located in the same and/or related cluster of companies, increases the degree of innovation.

On the other hand, Ferreira, Serra, Costa, Maccari and Couto (2012) show that the interaction between businesses, individuals and cluster agents are important mechanisms to facilitate innovations, although some clusters may record different results because of the heterogeneity of the available resources. This means that accessing shared resources may have a positive effect on the innovativeness of firms (Delgado et al., 2010).

In turn, Hervás-Oliver and Albors-Garrigós (2009) explore the role of combinations of internal and external resources to explain innovation performance of firms located in clusters. Based on a study using data from small and medium European companies, the results highlight the idea of internal resources as important factor to access external knowledge - called absorption capability, and to obtain innovation benefits through complementarity of internal and external resources.

Furthermore, industrial cluster environments can generate knowledge spillovers, which represent a key factor for the cluster's expansion and growth. The spillovers are understood as knowledge owners transiting inside companies operating in clusters, enabling the co-evolving of companies. Chyi, Lai and Liu (2012) investigated the cluster of high-tech Hsinchu Taiwan, and found out evidence that spillovers contribute to raising the performance of companies.

In line with the literature (Hervás-Oliver & Albors-Garrigós, 2009; Lai et al., 2014; F. B. Rodríguez & Valencia, 2008; Tristão et al., 2013), we consider that, through externalities and spillover effects, the agglomeration economies of a regional cluster can provide substantial resources and knowledge spillovers, that, in the long run may be converted in significant opportunity of innovation and growth.

We, therefore, hypothesize that:

H2: Access to the cluster's resources is positively associated with the degree of innovation of firms.

Innovation has been studied intensively by researchers in different fields of study for being a multidisciplinary concept (Damanpour & Aravind, 2012). Because of all the social changes of recent decades, the term innovation has been associated with the competitiveness of companies and nations (Porter & Rivkin, 2012).

Despite the existence of various definitions of innovation, for the purpose of this study we adopted the definition of the Organization for Economic Co-operation and Development (OECD, 2005). Therefore, innovation corresponds to the implementation of a new product or service, or significantly improved; a process; a new marketing method; a new organizational practice in organizations and external relations.

Product innovation is the differentiation or improvement of a product to meet the needs of external users, while proceedings innovation is to improve the production of a product or service process (Damanpour & Aravind, 2006, 2012).

Innovations can be promoted by organization's internal resources (Rosenbusch, Brickmann, & Bausch, 2011) as well as the knowledge acquired externally (Kostopoulos, Papalexandris, Papachroni, & Ioannou, 2011; Sáenz, Revilla, & Knoppen, 2014) or even by a random set of actors combined with internal resources of the organization - called open innovation (Chesbrough, 2007; Dahlander & Gann, 2010).

In the field of International Business, scholars have sought to establish the relationships between innovation and internationalization. Monreal-Pérez, Aragón-Sánchez and Sánchez-Marín (2012) conducted a longitudinal survey by Spanish companies and identified that exporter companies are more likely to develop more innovations and thus increase their activities in international markets.

Although studies didn't find statistically significant relationship between innovation and export performance (Mais, Carvalho, & Amal, 2014), we hypothesized, in line with the empirical study of Monreal-Pérez et al. (2012), as well as evidences from Cassiman, Golovko and Martínez-Ros (2010), Alegre, Pla-Barber, Chiva and Villar (2012), Becker and Egger (2013) and Tavassoli (2013), that, due to the gains in the competitiveness, innovation can be seen as a strategic advantage to overcome trade barriers and to establish a significant market position, as well in the domestic as in foreign markets:

H3: Innovation is positively associated with export performance of firms.

While scholars have largely discussed the impacts of innovation on internationalization and export performance, it is noteworthy that there are still few studies that investigate the innovation as a mediator channel (Alegre et al., 2012). With the insertion of a mediating variable, we seek to investigate whether the innovations generated as a function of access to regional clusters is compatible with export performance. The effect of the mediation occurs by reducing the direct effect of one hypothesis (H1), rather than the joint ratio of the two hypotheses (H2) and three (H3).

We assume that the more innovative companies are more likely to access greater intensity of cluster resources and, therefore, to sustain higher export performance. Through this assertion it can be inferred that the clustering of industrial activities is relevant not only to strengthen the competitiveness of companies in international markets, but represents a generating mechanism of innovation, which can promote higher export performance.

In other terms, it is assumed that companies that record higher innovation performance in products and processes are more likely to possess specific capabilities to access on a higher intensity the shared resources of an industrial cluster, and consequently achieve higher export performance levels. For the purpose of testing the innovation as a mechanism of mediating relationships between regional cluster externalities and export performance, we will test the following research hypothesis:

H4: Innovation mediates positively the effects of cluster resources on companies export performance.

However, it seems that the spillovers and externalities generated by agglomeration economies are not uniform among industries. Empirical studies have find differences in the internationalization process according to the technological intensity of industries. Stoian, Rialp and Rialp (2011), in an empirical study of small and medium-sized (SMEs) Spanish companies, have found evidence that SMEs operating in high and medium-high technology industries recorded better export performance than firms in medium-low and low-tech industries.

These findings corroborate the J. Bell, Crick and Young (2004) and López-Rodríguez and García-Rodríguez (2005) studies. In addition, a literature review points to strong evidences of the effect of the technological intensity as a feature of relevant industry to explain the export performance of companies (Wheeler et al., 2008).

In accordance with previous studies, technological intensity is considered in this study as a relevant variable to moderate the model's relationships. Therefore, we will test the fifth hypothesis:

H5: The technological intensity of the sector moderates positively the relationship between cluster resources, innovation and export performance.

According to the literature review, the size of the company (Chugan & Singh, 2014; Wheeler et al., 2008), international experience (Chugan & Singh, 2014; Moghaddam et al., 2012) and the location in high industry agglomeration regions (Chugan & Singh, 2014) are important control variables to explain the export performance of organizations.

In addition, studies indicate organizational aspects such as company size (Souza et al., 2008; Wheeler et al., 2008), international experience (Moghaddam et al., 2012; Papadopoulos & Martín, 2010) and the fact of companies based in regions with higher industrial concentration (Mittelstaedt et al., 2006) as important factors that may influence the export performance of firms. Thus, we will control for size, international experience and industrial concentration for the relationships between cluster resources, innovation and export performance. In the following illustration, we will present the operational model of our study, with the corresponding hypotheses as discussed in the literature review.

Figure 2
General Framework of Export Performance Determinants
Source: The authors.

According to Figure 3, the result points to the evidence that there is an association between accessing the cluster resources and export performance. The more companies can benefit from institutional, local and business network resources, the more likely they can record a higher export performance. This points to a heterogeneity among companies regarding the level of export performance, with a factor loading of 0.306 (p-value = 0.002). We measured the export performance by two main variables: increase of market shares and market diversification.

This means that a higher access to the cluster resources increases the international market shares and the satisfaction with the diversification of foreign markets. These results indicate the access to cluster resources provides the advantages of localization. The results show that all the resources have been found statistically significant and positively correlated, but business network and location resources have shown relatively more loading factor than institutional resources according to the managerial perception. The differences in the factor loading can be explained by several reasons. It seems that firms are more likely to benefit in a larger manner from business network and location resources than institutional resources. Firms can explore business networks using their own capacities. The more firms can be involved with their business partners, the more likely to assess specific market opportunities. Firms that are committed in the network can, therefore, better benefit from knowledge and the growth strategies of their suppliers and customers in the international market. This means that the knowledge, technology and growth are not restricted to the domestic market, but rather to the experiences and knowledge gathered in a larger scale. The second reason is that the factor location resources (professionals, infrastructure and local market) has a high impact on costs and efficiency. Thus, firms can take advantages from the different externalities of the location, particularly those that provide conditions to benefit from the economies of scale, either from the upstream, or the downstream side of the value chain. Finally, institutional resources are not directly related to the competitiveness of firms. These resources use to be important triggers to promote external growth and mechanisms of cooperation governance, however their impacts are relatively low particularly in the short run perspective. Institutional resources can drive and promote the commitment of firms with export and internationalization activities, however, their impacts on performance are more likely to be limited.

Thus, the hypothesis H1 of the positive association between cluster resources and export performance is supported. This result is in line with other studies, like Belso-Martínez (2006), Mittelstaedt, Ward and Nowlin (2006), and Fernhaber, Gilbert and McDougall (2008). In general, the cluster's resources as institutional, local resources and business network are important sources to contribute for development of internal resources and capabilities of firms (McCann & Folta, 2008; Molina-Morales & Martínez-Fernández, 2004).

The relationship between resources in cluster and innovation in structural model demonstrates a factor loading of 0.398 (p-value = 0.000). These evidences indicate that companies with higher performance of innovations in products and processes are more likely to have better access to greater intensity of clusters resources (Delgado et al., 2010).

These evidences are in line with other empirical studies (F. B. Rodríguez & Valencia, 2008; Hervás-Oliver & Albors-Garrigós, 2009; Lai et al., 2014; Tristão et al., 2013), and support the hypothesis Access to cluster resources positively influences the degree of innovation (H2). Thus, Companies benefit from different regional clusters externalities, such as institutional, local and business network resources. These represent important mechanisms for the firms' integration with external agents to promote the development of own internal resources. The role of such externalities mechanisms has been found highly significant in influencing as well the innovation performance, as the export performance of firms in international markets.

This result is in line with the findings of Hervás-Oliver and Albors-Garrigós (2009), which point to the complementarity of external with internal resources to obtain innovation benefits. We assess the impacts of using external mechanisms as a factor of innovation performance of firms, however, we consider that the role of internal resources on the process of innovation development is still a determinant driver.

The results of modeling represented in Figure 3 indicate positive factor loading with 0.116 in the relationship between innovation and export performance (p-value = 0.277). These results show, however, no evidence about the impacts of innovation in product and services on the export performance of firms when measured by market diversification and the satisfaction of market diversification. This finding is not in line with the results of other studies (Alegre et al., 2012; Becker & Egger, 2013; Cassiman, Golovko, & Martínez-Ros, 2010; Tavassoli, 2013).

Therefore, the hypothesis Innovation positively influences the export performance (H3) is not supported, in line with the results of Mais, Carvalho and Amal (2014). This result may reflect, first, the use of a perception variable (satisfaction), instead of other measurement variables of export performance as export intensity (Alegre et al., 2012; Tavassoli, 2013) which was excluded from the model because of its low internal consistency, and even the propensity to export (Becker & Egger, 2013; Cassiman et al., 2010; Tavassoli, 2013). This result can be understood as characteristic of emerging countries, because their export performance is less associated with innovation, but likely more to cost specific leadership. On the other hand, this result can also reflect the limitations of using a subjective variable of the product and process innovation. Future studies can consider using the number of patents as a variable to capture innovation performance of firms (Archibugi & Pianta, 1996).

According to Figure 3, the indirect effect of cluster resources on the export performance through innovation occurs by the accumulation of factor loadings discussed in the hypothesis 2 and 3 together. Thus, while the relationships between cluster resources and innovation has recorded a factor loading of 0.398, and statistically significant at 1% (p-value <0.001), the association between innovation and export performance with a factor loading of 0.116, was not statistically significant (p-value = 0.277). Finally, we could find no indirect effect between cluster resources and the export performance. This mediating relationship recorded a factor loading of 0.046, but was not statistically significant (p-value = 0.183).

To ensure the validity of mediation, when the construct innovation was included in the relationship between cluster and export performance resources, the direct effect of cluster resources should be reduced to zero (Hair et al., 2009; Iacobucci, Saldanha, & Deng, 2007), which has been not shown in the model.

On the other hand, it seems that the spillovers of regional clusters resources on the innovation performance generate no direct effects on the international competitiveness of firms (export performance). So these results suggest that the effects of cluster direct effect on export performance, and are uniform among firms no matter their level of innovation performance.

We can, therefore, consider that innovation does not operate as mediator in the relationship between cluster resources and export performance. This result provides no support for the hypothesis Innovation mediates the influence of cluster resources on the export performance of firms (H4).

Furthermore, these results also suggest that innovation, in the Brazilian case, has no direct, either indirect effects on export performance. That is to say, according to the survey respondents' perception, the fact that access to resources in clusters results in higher importance to determine the export performance compared to innovations in products and processes. We understand that innovation, as suggested by the theories, may have a positive effect on the export and internationalization of firms. However, in the case of emerging economies (the Brazilian case), the patterns of internationalization is mainly driven by local and regional factor endowments, based in large extent by natural resources and labor intensive industries. These local advantages may have higher impacts on the export performance of firms. However, we also suggest that in the case of Brazil the relationship between innovation and export performance and internationalization is rather pointing to reverse relationship, by which it is more likely that the internationalization of firms that drive innovation performance. This can be explained by the different opportunities firms can explore to transfer knowledge through their foreign subsidiaries, or through a mechanism of learning process when interacting with customers, suppliers and competitors in the global markets.

The model estimates show that the access to resources in clusters to obtain a heterogeneous export performance and innovations is uniform among industries, and that the level of technological intensity of the industry does not matter. The same can be observed in the relationship between cluster resources and innovation. Such effect seems to be homogeneous among different industries, as can be seen in Table 2 below.

Table 2
Effect of Moderation of Technology Intensity

These results suggest that industries with high or low knowledge intensity can benefit from shared resources of clusters to increase the export performance and innovations. In addition, neither industry benefit from innovations in products and processes to increase the export performance.

These evidences do not corroborate with the findings of the studies of J. Bell et al. (2004), J. L. Rodríguez and Rodríguez (2005) and Stoian et al. (2011), which shows the effect of technological intensity on export performance. We performed a Student T test, which certified that the average observed did not point to a statistically significant difference among the industries. Therefore, the hypothesis The technological intensity of the sector moderates the relationship between cluster resources, innovation and export performance (H5) cannot be supported. In contrast with other studies of developed economies, this evidence indicates similar benefits in cluster's resources for firms independently of industry and technological intensity.

Besides the results presented by the general model of export performance determinants, we attempted to investigate whether there are control variables that may present constraints to the model. According to Figure 4, among all the control variables, only international experience was found positive and statistically significant at 10% with a loading factor of 0.411 (p-value = 0.076), as can be seen in the Figure 4. This finding indicates that experience in international markets has a direct effect on the export performance of firms in accordance with studies of Papadopoulos and Martín (2010) and Stoian et al. (2011). This result may suggest that firms with higher international experience, may benefit better from the regional clusters resources and the corresponding institutional, geographical, and network externalities.

Figure 4
Effect of Control Variable
Source: Research data.

This evidence confirms that unlike the company's size and geographical agglomeration, the most experienced companies internationally access in greater intensity the cluster resources to get higher export performance. Therefore, the larger the international experience of firms, the higher its capacity to explore the positive impacts of economies of agglomeration on export performance.