The Relationship Structure of a Transnational, Ego-centered, and Innovation-Oriented Network

This study identifies the relationship structure of an ego-centered network by examining a transnational organization in the telecommunications sector that is oriented toward innovation. The study considers the dimensions of density, centrality, structural equivalence, and relationship content. Following the literature, we correlate the underlying concepts of network innovation substantiated through the technique of network analysis as a research method. The results suggest that the position of organizations in relation to the distance and pathways to other organizations in the network can influence the network’s capacity for innovation, conditioned by the strategic positioning of the focal company. This does not seem to be the case with the company considered by this study, as the strategic positioning of the matrix considers only innovative products with a worldwide scope, and not those with a more limited, local, or regional interest. The main contribution of this study is the identification of the relationship structure of a transnational, ego-centered, and innovation-oriented network with the aid of a statistical tool.


Introduction
Networks can be analyzed from the perspective of their structural or morphological characteristics and can be treated either comprehensively-according to the totality of the network-or by separating the connections between actors, as in the analysis of pairs (dyads) in this study.These characteristics play a significant role in the construction of scientific knowledge, as they provide a general overview of the network based on density, centrality, equivalences, and structural holes; thus, allowing for an understanding of the content flowing through the relationships in an organization's network (Gilsing et al., 2016).
Social Network Analysis (SNA) allows treatment from different perspectives and can be considered a metaphor, theory, or method (Powell;Smith-Doerr, 1994;Reinert;Maciel, 2012;Mäntymäki & Riemer, 2016).The network analysis used in this study is not fundamentally a research strategy but rather a set of techniques typically used in the treatment of relational data (variables of ties between social actors) at different levels.These techniques can be used with several designs (Reinert & Maciel, 2012).key factor in the survival of organizations operating in this market, especially for transnational organizations, as this market's field of business has rarely been limited to domestic markets.
The company in the ego-centered network examined in this study was chosen not only because it is one of the largest global players in telecommunications, with operations in 150 countries and annual revenue of 14 billion euros in 2011, but also because it is positioned with a focus on innovation and with perspectives related to global performance.The various classifications of ego-centered networks and networks in general can be used in an interrelated way, as the analysis of the data on a complete network has an ego-network with each of the relationships measured.Thus, the network formation in this research is egocentric, but the network data are analyzed in terms of dyadic (peer) relations.In addition to the focal organization, the network includes 18 companies considered strategic and fundamental in the development of its activities, selected based on the technique of snowballing (Biernacki & Waldorf, 1981).These include eight subsidiaries located in six different countries: Argentina, Singapore, Colombia, Mexico, China, and Brazil.The network also consists of two categories of organization: five suppliers and four logistics partners, with headquarters in different countries, such as France, the US, Switzerland, Colombia, Slovenia, and Brazil.Germany is also part of the network because, although it is not the headquarters of the joint venture, it is considered part of the network as a parent company as it owns half the shares of the entity created by a merger between the companies.
Network analysis was used to identify the configuration of the ego-centered network and its relational content, with an emphasis on the information base, products innovation, and entry into new markets.All data were statistically treated using UCINET, a software program for network analysis developed by Borgatti, Everett, and Freeman (2002).The boundaries of the ego-centered or self-centered networks are defined during data collection using the naming technique (Marsden, 2005).The network of relationships was defined based on input from a focal organization of the telecommunications sector formed by a joint venture between two companies based in Finland and Germany.This paper is divided into five sections.After a brief introduction, including the purpose of the study, its theoretical basis is described with regard to the relationship structure and innovation in interorganizational networks.The methodological and analytical procedures underpinning the data analysis are then presented.Afterwards, the research results are discussed and final considerations are presented, along with a discussion of the limitations and recommendations of the research.

Literature Review
Studies on interorganizational networks are occupying an increasingly significant space in the academic field (Balestrin, Verschoore, & Reyes Junior, 2010), and include various theoretical perspectives and currents of thought.The emphasis on an empirical-theoretical framework is supported by Wasserman and Faust's perspective (1994).They argue that methods related to network analysis can be categorized into structural properties, roles, and positions, as well as statistical analyses of relationships and innovation from the network perspective.

Relationship Structure in Interorganizational Networks
Network analysis at the interorganizational level considers its own relationships as components of a structure of relationships within a social order (Matinheikki et al., 2016).Nelson (1984) argues that the structure of relationships "allows us to study the interactions between formal and informal relationships, as well as other types of relationships, which is a long recognized need, but for which no methodology existed" (p.151).Nelson (1984) suggests that the most elaborate definition of "structure" is the one that deals with interconnected behaviors.The key to this definition is the repetition of interdependent, but not necessarily reciprocal, relationships of a cyclical nature that persist over time, such as frequency, degree of symmetry, configuration, content, and the nature of relationships.
One of the difficulties concerning network analysis is the definition of the network boundaries and the resultant configuration of the structure, especially when the network is considered as a whole (i.e., in full network design).For ego-centered networks, the boundaries are defined during data collection using name generation (Marsden, 2005) or, as in this study, by analyzing a focal organization.According to Wasserman and Faust (1994), an ego-centered network consists of a focal actor (the "ego"), a set of other actors with ties to the ego, and the measurements of the bonds between these actors.
The various classifications of ego-centered networks and networks as a whole can be used in an interrelated way.According to Kirke (1996), when an ego-centered network is considered dense, the entire network can be built through egocentric data.Marsden (2005) argues that the data analysis of a complete network has ego-networks with each of the relationships measured.Thus, the formation of a given network may be egocentric, but the analysis can be developed from dyadic (peer) relationships.Scott (2000) points out some important measures of centrality provided by the network analysis technique: centrality of degree, centrality of proximity, and centrality of intermediation.The centrality of degree is measured as the number of an actor's connections; it is not restricted to indirect relations, since degrees can be "indegrees" and "outdegrees."The original concept of centrality, based on degrees, treats connections as symmetrical-that is, the relationships are reciprocal (Knoke, 1994;Gilsing et al., 2016).
The measure of centrality of proximity allows one to go beyond direct relationships and consider the minimum number of steps an actor needs in order to establish contact with other actors in the network.From this perspective, the more central the actor is, the more likely the actor is to reach other actors in the network.Centrality of intermediation emphasizes the actors that serve as links between actors in the network.This is an important measure, since the identification of central intermediate actors allows one to determine how much they can influence the structure of the network, since they can control the routes through which resources and information pass among other actors.The degree of intermediation is represented by the Betweenness measure and reflects a node's capacity to intermediate the communications between pairs of nodes (Scott, 2000;Araujo et al., 2016).
Another important measure of centrality is the Eingenvector, which is usually used to find the centrality of each actor.The fewer the connections to nearby actors, the lower the actor's prestige in the network.In addition to the centrality measures mentioned above, Effsize is a measure of "structural holes," a term coined by Burt (1995).A structural hole is the space between actors, which, in this study, can indicate that an organization in the ego-centered network lacks a relationship with sources of resources or information.Wasserman and Faust (1994) emphasize the importance of structural equivalence in the analysis of positions in the network, such that two social actors are structurally equivalent when they have identical ties with other actors in a network.

Innovation in Interorganizational Networks
Innovation seen from the perspective of interorganizational networks presupposes a complex process of interaction and cooperation that goes beyond the limits of the intraorganizational dimension and the limits created by an exclusive focus on technology or research and development.Smolka and Côrtes' (2006) research on cooperation networks for innovation in products in technology-based companies (TBCs) indicates that a network analysis provides important inputs for an extensive understanding when considering other organizations immersed in the network of relationships.Smolka and Côrtes (2006) revealed important relationships between TBCs and universities or research centers but identified few cooperative initiatives among the TBCs themselves.In addition, other actors with regulatory roles that configure the network, such as regulators, governmental agencies, and congressional bodies, influence the innovation of the TBCs by having levels of centrality that are more elevated and differentiated than are those of other agents in the network, providing them with a special understanding of the innovative environment.
Innovation is a multidimensional process that requires novelty, highlighting the need to combine three important research chains: internationalization, innovation, and networks (Chetty & Stangl, 2010).Freire, Baldi, and Lopes' (2010) work on innovation networks describes the innovative process as being collectively created based on interorganizational relationships at the local, national, or global levels, based on the convergence of social ties (Granovetter, 1973) and innovation networks (Hage & Hollingsworth, 2000;Veiga et al., 2016).This approach to innovation networks is justified by the growing need to understand and promote processes of innovation in competitive global networks, which are configured as forms of governance of new patterns of interorganizational relationships.This approach is visualized in the work of Weymer (2013), who noted that innovative ideas and products from the subsidiaries of an interorganizational telecommunications network are seen as legitimate only if they are aligned with the parent company's overall strategy.
Innovation goes beyond the structural aspects and the combination of objectives, resources, and people; it also brings to the fore an analysis that goes beyond organizational and geographical limits and emphasizes the content flowing in social relationships.This does not mean one can neglect the importance of innovation in favor of intraorganizational aspects.On the contrary, an analytical understanding of the internal context is the basis for understanding innovation from the network perspective and within the global context.Although the term "globalization" is sometimes used indiscriminately due to the somewhat reductionist perception of a single economy regulated by a new and unprecedented economic system, the world is clearly generating a "new global economy" based on a dynamic of its own, in which national borders are becoming much less restrictive (Ohmae, 2006).It is within this highly complex context that extending cross-cultural research to deepen understanding of the value of collaboration across domestic boundaries is urgently needed, particularly considering partnerships between organizations (Hillman, 2009;Matinheikki et al., 2016).

Methodological Procedures
The analysis of networks in this study consists of a set of techniques most often used in the treatment of relational data (variables of ties between social actors) at different levels.According to Reinert and Maciel (2012), such techniques can be employed with various designs and are not solely characterized as a research strategy.

Characterization of the Research
Although the starting point is a subsidiary company, this study employs surveys as a research tool as various agents from several organizations were considered.These provided 342 possible relationships, including partnerships and subsidies, for subsequent descriptive analyses.The degrees of importance of the 342 connections generated a 19 × 19 square matrix (minus the focal organization, as it cannot relate to itself).With this information, the UCINET software generated the representative sociogram of the ego-centered interorganizational network.

Population and Sample
The population consisted of all organizations that maintain relationships with the focal organization, and the sample consisted of the organizations named by the subsidiary.The level of analysis was thus interorganizational due to the network of relationships of the subsidiary, which is considered a focal organization.The unit of analysis was each organization in the network, represented by key actors (individuals).
The criteria for choosing the focal company were the following: it had to be a transnational company with a subsidiary in Brazil; one of its main objectives had to be innovation for internationalization; it had to be a global player; and it needed to be willing to provide information about its network of relationships.After identifying the company, three professionals at the strategic level of the subsidiary located in Curitiba, Paraná, Brazil, were identified.The confidentiality of the focal organization identity was a condition of the data collection and formally communicated to all organizations interviewed.
The three professionals, considered gatekeepers, were interviewed so that they could describe the scenario in which the company was immersed and its strategic positioning at a global level.During the interview, the gatekeepers listed the primary organizations they considered fundamental for innovation and described these organizations' activities.This led to the generation of additional contacts that were contacted and who named other organizations.Finally, this produced 19 network actors, including the Curitiba subsidiary.Among these companies, eight are subsidiaries located in the six countries, namely Argentina, Singapore, Colombia, Mexico, China, and Brazil), two São Paulo subsidiaries (one with an emphasis on development), and one subsidiary from Rio de Janeiro, in addition to the subsidiary in Curitiba, the focal company.Germany is also part of the network as a parent company, despite not being home to the joint venture's headquarters, because it owns half the shares of the entity created by a merger between the companies.The network was composed as shown in Table 1.
As Table 1 shows, all the network organizations pursue international activities involving 10 countries, including Brazil.Some of these organizations are large global players, with revenues exceeding US $5 billion for logistical partners and US $720 million for the company with the highest revenues in 2011.Considering that communication with most of the companies was conducted in English, it was decided to present terminologies and acronyms in that language, which will be used later in the network configuration analysis.The network also consists of two categories of organization: five suppliers and four logistics partners, with headquarters in countries such as France, the US, Switzerland, Colombia, Slovenia, and Brazil.

Data Collection
The questionnaire produced structured data in a matrix, constructed based on the theoretical foundation and focus area of this research.Prior to the referral, the questionnaire underwent a validation process with eight professors with doctorates in administration studies.
As all the companies in the network are transnational and eight of the respondents are located in other countries, the questionnaire was translated into English and sent by e-mail, with frequent follow-up by telephone.A six-month response period was employed for two reasons: (1) transnational corporations must follow legal protocols when releasing data, and (2) only three of the companies were located in Curitiba/Paraná, with the other companies located in foreign countries (eight companies) or in other states in Brazil (eight companies).
From these data, it was possible to identify the measures of centrality and structural holes, as well as the relational content, with an emphasis on the exchange of information, product innovation, and entry into new markets.

Presentation and Data Analysis
The starting point for the network configuration was choosing as the focal company a transnational firm in a position of global leadership in telecommunications services and a focus on innovation.The focal company has a complete portfolio of mobile, fixed, and convergent network technology, as well as professional services, including consulting and systems integration, set-up, maintenance, and managerial services.It operates in 150 countries and is one of the most important companies in the world in this sector, with annual revenue of 14 billion euros in 2011.The company was formed through a joint venture announced in 2006 between two large companies with headquarters in Germany and Finland; their total stock was divided in half between the companies.
The term "transnational" is based on the work of Bartlett and Ghoshal (1998), who suggest that many global industries have transformed since the 1980s, when there were traditional categories of global, multinational, or international corporations.The more current, transnational form requires companies operating globally to simultaneously deal with the configuration of assets and capacities, the role of operations abroad, and the development and diffusion of knowledge to remain competitive.
The analysis treated the network symmetrically-that is, without considering the directional differences in the relationships, known as the "indegree" and "outdegree."According to Scott (2000), "the indegree of a point is the total number of other points that have lines directed to it; the outdegree is the total number of other points to which it directs lines" (p.69).As this study seeks to identify flows from relationships regardless of reciprocity, the degree satisfies this need, since it indicates the number of direct relationships among actors (in this case, network organizations), as well as one of the elements for the calculation of centrality (Borgatti, Everett, and Freeman, 2002).The network of relationships among the 19 organizations of the ego-centered network, originating from the S_CWB organization (a subsidiary in Curitiba), can be illustrated as in Figure 1.

Density
The Source: Survey data According to Borgatti, Everett, and Freeman (2002), the degree indicates the number of direct relationships between organizations.It is also one of the elements for calculating centrality.The third column, represented by NrmDegree, indicates the degree of centrality divided by its possible maximum value, expressed as a percentage.
In this case, of the first 10 companies in the ranking, eight are subsidiaries, one is a supplier, and another is a logistics partner.
The presence of subsidiaries in the top 10 seems to be a natural trend for organizations that are linked to a parent company and that have common long-term global goals.It is worth mentioning here that the company most frequently named concerning information exchange was logistics partner LP_4.
This centrality can be explained by the fact that the company is a global player, is of French origin, operates in the logistics field, and serves almost all the organizations in the network (94.44%) in the international arena.As it is a logistics services provider, the content of information exchange is related more to negotiation than to the integration of knowledge, which has implications for the adaptations of partner organizations concerning innovative activities.
Another important piece of complementary data is related to the contribution of relationships to product innovation for each company.Considering the total number of effective relationships in the network, the perception of each company when dealing with its peers has little direct influence on the innovation of new products (61% of survey responses were between "low" and "very low").Only the subsidiaries from Singapore and Germany acknowledge receiving high levels of contributions from certain Subsidiaries.Additionally, worth considering is the organizations' influence on entry into new markets, which can be visualized in Figure 5. organizations in the same social context.This would allow for not only an analysis of dyads but also a structural analysis enabling the inclusion of other relevant information available from the network analysis technique.The structural properties of the network could be considered independent variables and be related to other variables, according to the interest of the researcher.

Table 1 .
Organizations in the network Table indicates that the organizations named most frequently are those considered important sources of informational exchange.These organizations occupy more central places in the network and can be better visualized in Table3.

Table 4 .
Contribution to product innovation