Suppliers’ network analysis under the perspective of structural, relational and cognitive embeddedness: an exploratory study

Paper aims: To analyze and characterize a supplier network under the analytical perspective of structural, relational and cognitive embeddedness. Originality: Analysis of a network of suppliers in a soft drink industry using Social Network Analysis (SNA) techniques. Research method: The case study was used as a research strategy. The analysis of structural perspectives, relational and cognitive embeddedness and their respective variables were considered, as well as measures of centrality and density of the network of suppliers. Main findings: The analysis allowed to establish the characteristics of the network of suppliers regarding structural, relational and cognitive embeddedness and their respective variables. However, it should be noted that the manuscript addresses a single case study, and it is not possible to generalize the results obtained. Implications for theory and practice: The findings of this paper can be used as a basis to understand how to use the main metrics of social network analysis to investigate the structural, relational and cognitive characteristics of a company and your network.

In this perspective, network theory has modified not only the way organizations operate within the market, but also the way they see themselves, since many have gone from an individualist view to a systemic view (Silva & Heber, 2013).
The question of how social relations affects economic action and organizations is one of the classic approaches in social theory (Granovetter, 1985). The notion that business-to-business transactions are embedded in a broader social context has become a recurrent concept for economic and organizational scholars (Borgatti & Foster, 2003). As is the case of Uzzi (1996) who through the empirical research of embeddedness focused has centered on a wide range of topics, such as business survival (Uzzi, 1996), price fixing (Baker & Faulkner, 1993), trust and opportunism in interorganizational relations (Batenburg et al., 2003), the social identity of firms (Rao et al., 2000), and the business behavior (Simsek et al., 2003).
The argument of social immersion thus, emphasizes the concrete role of personal relationships and the structures (or networks) of such relationships in generating trust in economic actions (Vasconcelos & Oliveira, 2012). The first type of relationship is called relational embeddedness and the second is structural embeddedness. There are other types of embeddedness cited in the literature, such as cognitive, which refers to the ability of two organizations to integrate effectively their respective knowledge (Gobbo, 2010).
The structural positioning of the actors in a network of relations is one of the primordial elements to understand the structure of exchange between the actors. In the centrality, for example, an actor centralizes the relationship with other actors of the network, being able to have access to resources, power and information (Sacomano Neto & Truzzi, 2009).
Another reason justifying the development of this research is the relationship that structural and relational embeddedness has with cognitive embeddedness (Figure 1). Although the interrelationships between the tree dimensions of embeddedness are of great interest among researchers, previous studies tend to focus on the effects of each dimension individually without investigating the relationship between them (Li et al., 2013). Therefore, the objective of this paper is to analyze and characterize the enterprise network of a soft drink company, under the analytical perspective of structural, relational and cognitive embeddedness. From the positional analysis of the actors present in the network, some problems were identified and some improvements were proposed. The network under analysis was structured with suppliers from the first and second tyer of the Focal Company, by means of a case study.
The choice of a single case project can be justified by the fact that the case selected for analysis is a representative or typical case (Yin, 2001), since the lessons learned from the soft drink company provided a wealth of information on the average experiences of this industry. The network of suppliers of a soft drink industry has a well-defined structure, according to studies such as that of Conceição & Quintão (2004). The purpose of the choice in this case was to have a consolidated basis to work with an incipient theme, as it was the investigation of the three embeddednesses.
It is valid to highlight that the choice of single case study is strongly related to the adoption of the roster-recall method (Morrison & Rabellotti, 2009) and by the research involving an ego-centered network. To help map the network and to explore its indicators, the software Gephi 0.9.2, Ucinet 6.357 and Net Draw 2.114 were used.

Literature review
Enterprise networks can be understood as interorganizational arrangements based on systematic, usually cooperative, links between formally independent firms, which give rise to a particular form of coordination of economic activities (Fusco, 2005). In many business environments, the network presents itself as an important solution to help companies respond quickly to market changes (Hallikas et al., 2004). On the other hand, there are the informal networks and according to Khodabandehloo (2014) the characteristics of these networks for regional development creates vagueness, and might lead to unclear results, in the end affecting the function of the network. In general, informal networks are more prone to personal and formal relationships to issues of control, hierarchy, and authority.
The concept of networks in the organizational sphere has been used in two perspectives: emphasis on its application as an analytical aspect and as a structure of governance (Lopes & Baldi, 2009). As an analytical perspective, the use of the network allows a contextualized approach of economic action, by which, researchers argue the social immersion in the economy. This inference is supported by the new economic sociology, which has in Granovetter (1985) its theoretical foundation (Swedberg, 2004).
The governance structure refers to the set of tools that coordinate participating organizations seeking to deliver results from the network (Grandori & Soda, 1995). Studies on this theme are related to the institutional mechanisms by which interorganizational relationships and the attributes of network actors and the form and content of their relationships in a particular institutional context (Oliver & Ebers, 1998).

Suppliers network
The supply chain is embedded in a complex network of relationships between suppliers, customers and various other stakeholders (Cruz & Liu, 2011). Thus, vendor selection is increasingly recognized as a critical decision in supply chain management (Millington et al., 2006).
In order to contribute to the management in question, social network theory emphasizes that the economic actions of a supplier are embedded in a network and that its results are substantially influenced by the current pattern of a relationship (Gulati, 1998;Granovetter, 2005). In this sense, the application of techniques on analysis of social networks in supplier's networks are becoming more frequent. Kim et al. (2011) demonstrated how to use social network analysis to investigate the structural characteristics of the supplier's networks, relating the main metrics of social network analysis, considering material flow and contractual relationships. Choi et al. (2001) observed the methodological potential of the SNA and addressed the study of supplier's networks from the perspective of the social network.
Later studies such as Borgatti & Li (2009) recognized the difficulty of collecting data at the supply network level, but argued that its imperative to manage supplies should be integrated with other management disciplines.
For these authors, a more systematic adoption of SNA is fundamental in the exploration of behavioral mechanisms of the entire supply network and allows us to better understand the operations of these networks, identifying how important are the companies individually considering their positions in the network and how the structural aspects affect the performance of individual companies.

Network properties
According to Castells (2003), network is a set of interconnected nodes, with several structural properties that can be exploited (Sacomano Neto & Truzzi, 2009). For Kilduff & Tsai (2007) these properties are centrality, structural equivalence, structural autonomy, density, cohesion, among others.
Network centrality defines the position of an actor within it, which means to what extent the focal actor occupies a strategic position in the network as a result of being involved with several significant links (Wasserman & Faust, 1994). For Mote (2005), the centrality can be determined by the number of connections established in each node, being calculated from Equation 1, described in Freeman (1979).
In structural autonomy, an actor mediates the relationship between other two, generating the same results of the centralized structure, such as: access to information, power, resources, status, among other aspects (Sacomano Neto & Truzzi, 2009). Gobbo (2010) brings this same approach, however highlights that the actor can occupy a structural hole between two other actors with whom he is connected. Burt (1992) understands that structural holes are areas within the network in which loops do not exist or the density is low.
The structural equivalence in turn, at the analysis level, concerns two actors who occupy similar positions in the network structure, a fact that generates similar actions among them (Sacomano Neto & Truzzi, 2009).
Density is the global level of connections between the members of the network (Provan et al., 2007), and cohesion is the intensity of the relationship (strong or weak), in which the effects are related to the gain of information, knowledge tacit, social control and reciprocity (Sacomano Neto & Truzzi, 2009). In quantitative terms, the density of a network can be estimated according to Equation 1 (Wasserman & Faust, 1994).
The network, object of study of this research, has input and output connectivity. In Equation 1, T indicates the number of ties in the network and n the number of nodes.
Cohesion is related to the intensity of the relationship, in which the effects are related to the gain of information, knowledge tacit, social control and reciprocity (Sacomano Neto & Truzzi, 2009). It concerns the volume of contacts established between the actors. The high volumes and high quality of information exchanged imply the high frequency of interaction and possibly greater cohesion relationship (Fusco, 2005). Polanyi (1944) is known as the creator of the term social immersion (embeddedness). Granovetter (1985), when retaking the concept of Polanyi (1944), directs its focus to the research of the institution of the market, its actors and contexts. Although the researchers have conceptualized embeddedness in different ways, Kim (2014) reiterates the context of the supply chain and defines embeddedness as the extent to which a company relies on a network of supplies, and perceive the influence of it. In this scenario, the expression embeddedness denotes the state of dependency of a company on its suppliers and customers within a specific supply network structure (Choi & Kim, 2008).

Positional network analysis -embeddedness
The works of Granovetter (1992) have still generated countless developments. Some authors have sought to identify the various possible types of embeddedness, such as Hess (2004), who identified three types (social, network and territorial); Halinen & Törnroos (1998), who identified the presence of six different types of embeddedness (social, technological, market, political, temporal and spatial). Zukin & DiMaggio (1994) sought to broaden the original concept by incorporating new dimensions such as cognitive, cultural and political dimensions. Recent researches characterize networks based on three types of embeddedness: structural, relational and cognitive (Lechner et al., 2010).

Structural embeddedness
Structural embeddedness describes the general architecture of the network, encompassing the properties of links between companies as a whole (Dacin et al., 1999), and emphasizing the informative role of the position that an organization occupies in the general structure of the network (Gulati,1998). Among the important facets of structural incorporation is the presence or absence of links between actors. The variables of this concept include closure, density, connectivity and hierarchy (Wasserman & Faust, 1994).

Relational embeddedness
The researchers conceptualized relational embeddedness from three perspectives: a direct bond, an asset, and the strength of an interaction (Kim, 2014). The first perspective considers relational incorporation as a direct cohesive link between two companies, also known as a cohesive social interaction and a personal relationship (Lin et al., 2009). In the second perspective, relational chaining is seen as an asset, being a form of trust, reciprocity and obligation, which is rooted in a relationship and is created through ongoing interactions (Tsai & Ghoshal, 1998). The third perspective highlights the strength aspect of incorporation: relational incorporation is seen as the strength or quality of a relationship (Dhanaraj et al., 2004).
The research on structural embeddedness focuses mainly on the impact of network density and network position on external and performance corporate activities (Gonzalez et al., 2014). Yang et al. (2011) argue that a strategic position in the network ensures non-redundant information and makes the company have bargaining power, formalizing alliances with its partners.
The dense network refers to an effective governance mechanism with a high level of interconnections and shared routines (Rowley et al., 2000). Dense network participants are strongly connected to each other and have many common contacts, developing intense communication and sharing redundant information, often obtained from multiple sources (Polidoro Junior et al., 2011).
In contrast, diffuse network participants have few common contacts and seek to increase the efficiency of information flow (Burt, 1992). Briefly, the degree to which firms benefit from structural embeddedness depends on the centrality and reach of the network (Van Eck et al., 2011). This research was based on Gobbo Junior & Vasconcellos (2008), which delimit some variables of structural embeddedness, as presented in Table 1. It is related to the proximity or distance of an actor in relation to other actors in a network (sum of geodesic distances). The calculation of a vertex proximity is done by the ratio of the total number of other vertices (n-1) by the sum of all the distances between that vertex and each of the other vertices, as can be seen in Equation 3:

Betweenness Centrality
The equation representing the centrality of a node is related to the quotient of the shortest number of paths between nodes in a network, which includes the node in question and also the number of all the short paths of this network. Equation 4: where: σi,j is the number of geodetic paths that connect the vertices i and j ; σi,j(v) represents the number of geodetic paths that connect the vertices i and j through v. Corresponds to the number of actors in the network. (Sluzki, 1997) Relational embeddedness also highlights the strength of ties (Tiwana, 2008). The strong tie refers to the degree of frequency of one company interacting with another and circulating knowledge and resources efficiently, while the weak link is characterized by a freely coupled relationship or the relative frequency of interaction (Choi & Kim, 2008).
A high degree of relational embeddedness promotes close observations and interactions, allowing companies to access specific information (Khoja et al., 2010). In addition, it is possible to point out other gains based on relational embeddedness, such as the improvement of understanding of partners' needs (Cheung et al., 2010).
Therefore, one of the ways to understand relational embeddedness is through the cohesion of the relations between the actors, and can be understood through the intensity (frequency) of the relationship.

Cognitive embeddedness
Cognitive embeddedness, in turn, refers to similarities in representations, interpretations and systems of meaning between enterprises (Nahapiet & Ghoshal, 1998), or to common contexts among different social actors in a network (Rost, 2011). This dimension involves concepts of shared norms, systems of meanings and values, and can directly affect the development of social capital and organizational relationships (Muniady et al., 2015).
Relationships developed with shared norms and values may be stronger (Moran, 2005). Weick (1995) stated that when there is convergence in goals and values and when interpretations are shared between organizational partners, cognitive capital becomes continuous, cumulatively favorable. The cognitive dimension reflects the concept that distinct networks develop unique terms, acronyms, and number interpretations.
To determine the potential of cognitive variety present in a network, it is necessary to analyze the density of loops. A large number of present loops, as a percentage of the total possible number (dense network), indicates the possibility of access to different types of knowledge possessed by other actors (Gilsing & Duysters, 2008).
Therefore, two variables related to embeddedness cognitive can be highlighted. The first is the absorptive capacity, which is the organizational ability to recognize the value of new and external information, to assimilate it, and to apply for commercial purposes (Cohen & Levinthal, 1990) and the congnitive distance, which refers to the difference between individuals with cognitively different frameworks in a network (Gilsing & Duysters, 2008). Nahapiet & Ghoshal (1998) studied the interactions between the three dimensions of embeddedness and suggested that even when these dimensions are analytically separated, they still have highly interrelated elements. Consequently, the relational and cognitive aspect should affect the structural, resulting in dynamic interrelationships.

Materials and methods
Exploratory studies are performed when the objective is to examine a subject or a research problem that is not well studied (Gil, 2008). In general, exploratory research involves a bibliographical survey, interviews with people who have had practical experiences with the problem researched and analysis of examples that stimulate understanding (Prodanov & Freitas, 2013).
As for technical procedures, a case study is an (research that investigates-changes) a contemporary phenomenon within its real-life context, particularly when the boundaries between phenomena and context are not clearly defined (Yin, 2001). The researcher does not intend to intervene on the object to be studied, but to reveal it as he perceives it (Fonseca, 2002).
The choice of single case study is directly related to two determining situations. The first is due to the adoption of the roster-recall method (Morrison & Rabellotti, 2009). The roster-recall method (Morrison & Rabellotti, 2009), which consists of providing a list of all the members to each actor in the network, was used so that he can mention with which he has relationship. The second justification is that the analysis involves an ego-centered network.
Another choice was the use of non-directed networks, since the applied questions referred to the contractual relationship and not following the flow of materials, as in the case of directed networks, where the geodesics from ni (origin) to nj (destination) may not be the same as those from nj to ni.
The authors followed the methodological assumptions of Yin (2001) and Miguel (2007) for the development of the case study. In the first stage, it was the theory development, in which it was sought to know the literature of networks that in turn helped in the work delimitation and in the identification of the variables to be analyzed. Shortly thereafter, the case planning was carried out, defining the following protocol presented in Table 2.
With the protocol elaborated the third stage was carried out, which consists of a pilot test applied with the soft drink company, denominated Focal Company. The test allowed checking the data quality and making the necessary adjustments. The fourth step, data collection, was performed at the Focal Company.
The Focal Company was founded in the 1950s and since then has been expanding its brand. It currently employs more than 80 employees and produces just over 6 million liters per year of soft drinks, bottled in PET or glass bottles of different sizes. The Focal Company, all its suppliers of raw material for the production of soft drinks in PET bottles (first tier suppliers) and the main second tier suppliers were considered as objects of analysis for this study.
The empirical study was based on three semi-structured interviews conducted in loco with the general manager of the soft drink factory.
All data collected were completed from direct observations and secondary sources (publications and reports related to the area of this research and databases). The use of secondary sources improves the validity of all data obtained (Yin, 2001). To fill the matrix, the scale (weight) used in the Focal Company's relations with its first tier suppliers had as criteria the years of partnership, being: grade 1 (up to 2 years of partnership); grade 2 (from 2 to 5 years of partnership); and grade 3 (more than 5 years of partnership). The scale used in the relations of the first tier companies with their respective suppliers (second layer companies) was defined as: grade 1 (common raw material supplier); grade 3 (strategic product provider).
Then, the techniques of Social Network Analysis (SNA) and Ucinet software 6.357 and NetDraw 2.114 (Borgatti et al., 2002) were used to reproduce the quantitative indicators. In this sense, the network was quantitatively analyzed according to three centrality indicators indicated in Table 1. Subsequently, in a qualitative way, the network was characterized under the analytical cut of structural, relational and cognitive embeddedness. Finally, through observations on the positioning of the actors in the network, the main problems were listed, and with some references, some improvements were suggested. Figure 2 depicts the procedures that were followed in the development of this work, in order to achieve the established objectives.

Results and discussion
This section aims to describe and analyze the set of data obtained from the network mapping of suppliers of the soft drink factory.

Characterization of the supplier network
This research presents an example of application in a network of companies of a soft-drink factory. In order to better characterize the network, Figure 3 is divided into business segments, where the companies supplying both raw material and packaging for the company's soft drink production are presented.
The suppliers of raw material are consolidated in the market, most of them being large national or multinational companies. Few companies serve only a certain region. In supremacy, they are leading and pioneering organizations in the segment in which they operate, recognized for the quality of their products and for cooperation with their customers. Figure 4 shows the enterprise network.  The Focal Company was designated as actor H (Figure 4). The first and second tier suppliers were represented as actors F and C, respectively. The size of the sphere, varied according to the degree of centrality of each actor. With respect to the edges (lines that connect the knots), the thickness varied according to the weight assigned to each relation (the larger the weight, the greater the line thickness). Figure 4 shows the clear representativeness of F1, F2, F7, F9, F10, F11, F12, suppliers in which the focal company has partnered for more than 5 years. F3, F6 and F8, F13, F14, F15, companies with a partnership varying between 3 and 5 years with company H. Firstly, F1 is a supplier of sugar, an indispensable raw material not only in the production of soft drinks, but also in several other products, which explains the various F1 connections in the network.
The companies F9, F10, F11 and F12 are responsible for supplying the products that will give the soft-drinks its characteristic flavor, such as: aromas, conservatives, juices, emulsion, grape preparation, pineapple preparation and guarana extract. These companies share information with each other and maintain contact with several other second tier suppliers.
The companies F13, F14 and F15 provide chemical inputs (citric acid, ascorbic acid, sodium benzoate and caramel dye) and their high importance in the network is associated with the strategic product that is commercialized, and it can be produced with formulation exclusively for each client. In turn, the second layer companies C1 and C44 are also highlighted in the network (Figure 4) by the highest level of centrality (Table 3), due to the good service provided and the efficient logistics, seeking to streamline and optimize processes of movement of between the companies' partners. Although the graphical analysis is important, then the quantitative results will be presented for a further deepening of this analysis.

Structural analysis
The network is made up of sixty-six actors (including the Focal Company). The first level of the supplier layer has fifteen actors. According to Sluzki (1997), small networks are less effective in situations of overload or long-term stress, since members begin to avoid contact to avoid overload or, on the contrary, tend to overload. For the author, numerous networks run the risk of being ineffective, since there will always be a chance that other individuals will already be taking care of the same problem.
After completing the relations matrix in Ucinet 6.357, the data were imported into NetDraw 2.114 to generate the results of the centrality indicators. To facilitate the interpretation of the data, the results were divided into categories, grouping of percentages into ranges of values from 0.000 to 100.000, as shown in Table 3.
In Table 3, it is possible to notice the existence of three indicators of centrality: centrality degree; closeness centrality and betweenness centrality. By first analyzing the centrality degree, it is verified that the F9 supplier is the one that receives the most and transmits information throughout the network, a fact that can be proven due to the importance and specificity of the product it provides. Its position within the network is strategic, being considered central actor.
As for the degree of closeness centrality, the Focal Company has a higher degree. In fact, by analyzing the network ( Figure 4) the actor H is in a more central position as to the smallest path he has to go to reach the other nodes of the network. This measures their independence from the control of other actors (Pineyrua et al., 2016). According to Michelon (2015), generally, the node with the highest degree of proximity will have better knowledge of the network and greater power of dissemination of information in comparison with the other nodes.
The last indicator of centrality to be observed is the degree of intermediation. It is possible to note that the first-tier providers (and the C44 supplier, second layer) present the highest value. An important factor that a higher degree of intermediation is the comparison of the price of the products between the companies that it intermediates. Using Equation 1 (section 2.2) and taking into account that the network had 66 actors and 195 ties, the density obtained was 9.09%. In this way, the network is considered diffuse. This presupposes the existence of difficulties for the companies to make decisions and to solve problems together (Coleman et al., 1966). According to Sacomano Neto (2004), in the diffuse networks the number of nodes is greater, which harms and lessens the need for supplier development.

Relational analysis
The duration of Focal Company contacts with its first tier suppliers is, for the most part, of long years, as shown in Figure 5. The long years of partnership with supplier companies is also a lot due to the time the general manager has been in Focal Company, for more than ten years. In all this time the manager, who is responsible for the purchasing department, maintains an intense contact with the suppliers, in order to intermediate the internal needs of the organization and the external availability of its partners.
In Figure 5, it can also be observed that the supplier's network is cohesive, which according to Sacomano Neto & Truzzi (2009) can be understood through the intensity of the relationship (strong or weak, long term relationships, among others). As highlighted in section 2.2.2, the strong relationship between firms refers to the degree to which they interact with one another. Therefore, Focal Company maintains strong relationships with several first-tier suppliers, since knowledge and new information circulate efficiently among these companies, in addition to the long period in which they have a commercial agreement, as shown in Figure 5.

Cognitive analysis
The cognitive distance between suppliers F9, F10, F11 and F12 is smaller, because the organizational differences in the know-how, perceptions and skills shared between these companies are smaller (Gobbo Junior & Vasconcellos, 2008). Consequently, the absorptive capacity between these companies tends to be greater, since the greater the frequency of communication, the new skills are developed and the faster the feedbacks.
In Table 4, the problems regarding the degree of positioning of the actors in the network are presented. It is noted that many companies still do not have close relationships with their suppliers due to the distrust they have to share managerial and technological knowledge. In contrast, the cognitive variety of the network is low, considering that the network is diffuse (density of 9.09%). Cognitive embeddedness is also positively associated with relational embeddedness. As companies develop a closer agreement, incorporating their beliefs, assumptions, and expectations into the network, reciprocity rules are likely to grow stronger, making actors develop common standards of behavior. These behavioral similarities lead to the development of similar beliefs, that is, cognitive similarities (Simsek et al., 2003).
Most of the problems identified in Table 4 are summarized in the lack of collaboration and cultural differences among companies part of the network. Therefore, it is first necessary to get participants in these alliances to change their minds, and then to adapt to new ways of sharing knowledge. New organizational structures based on knowledge are emerging, among these structures, Galbraith (1997) presents the "Innovative Organization", in which two organizational structures must coexist: operational and innovative; whose purpose is to promote the creation and application of knowledge. It is up to the innovative structure to design ideas and the operational structure to implement them. The feasibility of the suggested proposals should be analyzed by the managers and directors of the factory.

Conclusion
The present article had as main objective to analyze and to characterize the companies' network of a soft drink factory, under the analytical perspective of the structural, relational and cognitive embeddedness. This paper may contribute to future research providing knowledge, especially about cognitive embeddedness, which is still little explored in the literature. Previous studies have suggested that the structural dimension is the origin of the emergence of relational and cognitive capital (Liao & Welsch, 2005). Khodabandehloo (2014) points out that informal networking should be avoided as it is inaccurate and makes it difficult to obtain tangible results. This research differs from the author's opinion and agrees with Jericó et al. (2008) that formal and informal communication are complementary to an organizational diagnosis and more flexible actions.
As for the relationship between embeddedness, it can first be emphasized that cognitive embeddedness is positively associated with relational embeddedness, that is, as companies develop an agreement embodying their convictions, premises and expectations within the network, reciprocity will likely become stronger, causing actors to develop common behavioural patterns. These behavioural similarities lead to the development of similar beliefs, or cognitive similarities As for the second relationship, that is, structural embeddedness with cognitive, a factor that deserves attention is the question of cognitive distance. It can be affirmed from the results obtained that the cognitive distance is small among companies that maintain strong ties in the network (degree of cohesion), that is, they are constantly learning and innovating their processes through the information and expertise they share. Consequently, the absorptive capacity between these companies is greater, since the greater the frequency of communication, the new skills are developed and the faster the feedbacks. In contrast, the cognitive variety of the network is Table 4. Main Problems Identified and Proposed Solutions.

Main Problems Identified Proposed Solutions
The location of the suppliers is distant. With this, the contacts usually occur by telephone, not allowing more detailed information exchange The use of information technology (IT) helps to bring the company closer to its suppliers. The goal of IT is to provide companies with the main input generated by information technology and then transformed into a subsidy for market actions (Mckenna, 1992). In the case of marketing, products and services are invented through collaboration between companies in digital corporate networks (Brum, 2017).
Little sharing of information and knowledge among packaging suppliers The dissemination of cognitive culture may be the progress toward a system of collaboration between these companies. Some actions should be taken, such as defining a radar of external competences, aiming at accessing specialized knowledge with competitive interest, establishing a structured process based on decision gates (Marques, 2015).
The departmentalization in some supplier companies creates internal barriers. This prevents them from getting a clear view of their own processes, not concentrating efforts on the needs and requirements of customers To structure the organization by processes, allowing more autonomy and polyvalence of the workers so that they can solve problems without having to resort to another department (flexible organization) (Gonçalves, 2000).
Lack of indicators for strategic selection of supplies It could be used, for example, the model of Handfield et al. (2000). The ultimate objective of the matrix is to optimize the relationship between costs and risks. The matrix crosses two dimensions: impact on the financial result and uncertainty of supply, generating four quadrants for the categorization of commodities small, considering that the density of loops, another indicator of structural embeddedness, presented a low percentage. In this way, the access to diverse types of knowledge in the network, in a global way, is restricted. The first limitation found in this study was related to the number of actors present in the second layer, and only the main ones were studied. Another limitation is the type of embeddedness that has not been addressed, such as: social, technological, market, political, temporal and spatial (Halinen & Törnroos, 1998).
Thus, it is suggested that future research addresses other types of embeddedness, as well as involving a quantitative study to deepen different analysis. Another suggestion is to cover a larger number of actors and to replicate the proposed methodological approach in networks of other economic sectors in order to increase the reliability of the results of this research.
In addition, new case studies using the same methodological approach would help to identify discrepancies between different economic sectors, for example by identifying the downstream character of a network in the service sector, where relational aspects such as the frequency of communication (embeddedness) with their customers would improve absorptive capacity in identifying behavioral changes and consumer market trends.