Network resources and social capital in airline alliance portfolios

1 NETWORK RESOURCES AND SOCIAL CAPITAL IN AIRLINE ALLIANCE PORTFOLIOS ABSTRACT The airline sector is a mature industry in which a wide range of competitive practices reflect highly intense levels of rivalry. This is evident from the alliance portfolios of airline companies that are composed of mutually advantageous agreements between firms in the aviation sector. In this context, we examine the influence that access to valuable network resources may have on company performance (in operational and financial terms). Moreover, as network resources come under the resource dimension of a firm’s social capital, we also study the way the resource dimension is affected by other dimensions of social capital (network position, network structure formed by the alliance portfolio and the quality of the relations). We study the alliances agreed between 214 mainstream airlines, up until 2007, during which time 351 airlines formed various types of partnerships. The results suggest, on the one hand, that access to partner resources is influenced by structural and relational factors. However, on the other hand, the structure of each airline’s alliance portfolio has more influence than its position in the global network of alliances. These results point to the need for satisfactory governance of the composition and management of an airline’s alliance portfolio. KEYWORDS: network resources, social capital, performance, alliance portfolio, airlines alliances, codeshare. 2 1. Introduction Over the past two decades, researchers have gathered evidence on interfirm alliances, interorganizational relations and their influence on firm performance (McEvily & Zaheer, 1999; Nohria & Garcia-Pont, 1991). In particular, studies have shown that alliances can improve innovative capacity (Ahuja, 2000a; Zaheer & Bell, 2005), market share (Shipilov, 2006), survival (Baum, Calabrese, & Silverman, 2000) and financial results (Rowley, Behrens, & Krackhardt, 2000). These investigations go further than the idea of the firm as an isolated actor. Moreover, not only have they sought to analyze how interorganizational relations affect performance, they have also attempted to explain the background to those relations (GarciaPont & Nohria, 2002; Gulati, 1995; Gulati & Gargiulo, 1999). From a theoretical viewpoint, the majority of these studies are founded on either the Resource Based-View (RBV) or Network Theory, or on the joint use of both approaches. The RBV began with the ideas of Penrose (1959) concerning the value of the resources to which the firm has access. Subsequently, these ideas were developed by a group of researchers (Barney,1991; Dierickx & Cool,1989; Grant, 1991; Rumelt, 1991; Wernerfelt, 1984), whose work suggested that each firm possesses a set of unique resources that distinguishes it from other firms and defines and maintains its competitive advantage (Barney, 1991). Many authors have used the RBV to analyze alliances and interorganizational relations (Ahuja, 2000a; Lavie, 2006; Zaheer & Bell, 2005). Studies on business alliances have definitively incorporated social network theory since Gulati (1995, 1998) proposed going beyond the study of the dyad to examine the entire network. From that point, the application of social networks to interorganizational relations has centred, above all, on comparing how the structure of a network of firms 3 (density, centrality, structural holes.) leads to different levels of performance (Ahuja, 2000a; Baum et al., 2000; Koka & Prescott, 2002; Rowley et al., 2000). The relational view (Dyer & Singh, 1998) complements this perspective by proposing the existence of rents generated by the alliance as a whole. In recent years, a new approach to business alliances - the alliance portfolio - has been gaining momentum. It is situated around the confluence of the RBV with network theory and promises to be very fruitful (Hoffmann, 2007; Kale & Singh, 2009; Wassmer, 2010). An alliance portfolio is “a focal firm’s egocentric alliance network” (Wassmer, 2010). This approach looks closely at the individual firm and at how it manages its relations with other firms, organizations and individuals in multiple alliances. Thus, the object of analysis moves from the alliance towards the firm and attention shifts from the complete network to the egocentric network of the focal firm. However, this spotlight on the individual firm and its egocentric network has attracted enormous attention in the previous literature on interorganizational relations seen in terms of social capital (Adler & Kwon, 2002; Koka & Prescott, 2002; Nahapiet & Ghoshal, 1998). The majority of studies on social capital have centred on the benefits that firms can obtain if they are immersed in a dense (Coleman, 1988) or a dispersed network (Burt, 1992), and on the type of relations that they maintain with other organizations (Koka & Prescott, 2002; Tsai & Ghoshal, 1998). Therefore, social capital as a construct is linked to network resources. Gulati, Nohria, and Zaheer (2000) proposed that both the structure of a firm’s ties and the characteristics of the links between them represent two network resources, although they also mentioned network members as a resource. Lavie (2006) expressed this idea in greater detail by pointing out how firms can obtain advantages from their alliances, depending on the resources with which their partners are endowed. 4 This study establishes a connection between the research agenda put forward by Lavie (2006, 2008) to study network resources, the alliance portfolio perspective and the concept of social capital. These are complementary approximations insofar as they are based on the egocentric networks of a firm that arise from various interorganizational relations. Thus, a new dimension of social capital is put forward for consideration as a multidimensional construct (Nahapiet & Ghoshal, 1998): the resource dimension, which may be added to the structural and the relational dimensions (Galán & Castro, 2004; Granovetter, 1992; Gulati, 1998). The resource dimension brings meaning to the idea that the value of business relations is conditional upon access to partners with valuable resources, over and above the quantity or the quality of those business relations and their structure (Casanueva & Gallego, 2010). There has been a constant proliferation of mergers, acquisitions and alliances in most sectors over the past two decades, but the airline business stands out because of the very real changes in the way in which competition takes place within it, due to deregulation, globalization and, above all, the extensive use of alliances between businesses. Evans (2001) analyzed the determinants behind the proliferation of alliances in the sector. In general, airline companies attempt to appropriate or to use the valuable resources of their partners, in order to compete or to survive, making agreements that provide them with access to new routes, to customer loyalty programs and shared marketing, joint purchasing, and subcontracted operational processes, among other aspects. The process has gone even further and true competition is developing between coalition and allied groups in the sector (Gomes-Casseres, 1994). The objectives of this study are to observe the different dimensions of the social capital that each airline possesses, because of its membership of a network of alliances, and the way these relate to each other. It also examines how these dimensions affect the 5 performance of the airline companies. Nevertheless, we should point out that, by conducting this research, we are not seeking to confirm exact, specific relations such as identifying behavioural patterns that connect the dimensions of social capital and the alliance portfolio with the performance of each particular airline, on the basis of network resources. Instead, we wish to arrive at an initial approximation of the proposed model, in order to understand which relations are established and in what way. This will allow us to continue with the investigation of those relations in a more detailed manner, or to establish new plans in view of the results. In any event, the fundamental purpose of this investigation is twofold. In the first place, it aims to confirm the effect of the relations within the alliance portfolio of an actor on its access to the resources of its partners; in other words, how does the egonet structure influence access to the resources of the network. In second place, an attempt is made to confirm how access to network resources, understood as the resources of partners that belong to the alliance portfolio of an actor, affect both the operative and the financial results of that actor. Thus, this study of the alliances between mainstream airline businesses reveals the role of the resource dimension in the performance of these firms and how networkresource endowment is conditioned by structural and relational factors, considered in earlier studies on social capital. Evidence is therefore shown, which support the ideas of Lavie (2006), to the effect that the greatest advantage in business relations is to maintain links with businesses and organizations that have valuable resources for the focal firm. These findings help to consolidate the bridges between RBV and social network theory, linking the ideas of network resources to partner resource endowment (Lavie, 2006, 2008) and to proposals to study the alliance portfolio from a network perspective. They also suggest that airline governance should consider partner resources as a principal 6 element in the management of their alliance portfolio, as it affects their choices as well as the processes of assimilation and appropriation. This study continues with the theoretical presentation of the relations between social capital, the alliance portfolio and network resources and how these can affect the performance of the firm, by describing a model to understand their relations better. Subsequently, an explanation is given of the methods used to test the hypotheses that arise from the model, in the context of alliances between airline businesses. The results of the analysis are then described and, finally, the conclusions are presented, along with the study’s limitations and future lines of research. 2. Social Capital, Alliance Portfolio and Network Resources 2.1. Social Capital Social capital is a resource in the network of individual and organizational relations over time (Adler & Kwon, 2002). Like all forms of capital, it is able to generate value for whoever possesses it. Adler and Kwon (2002) analyzed the differentiating characteristics of social capital as against other forms of capital of the firm. Various works have studied the way in which social capital influences the performance of individuals (Burt, 1992; Podolny & Baron, 1997) and of firms (Koka & Prescott, 2002; Zaheer & Bell, 2005). Although some controversy exists over whether social capital is an individual or a collective asset (Adler & Kwon, 2002; Lin, 1999; Portes, 1998), it appears clear that the beneficiary of social capital is an individual actor, as it is primarily of benefit to the actors that possess it (Koka & Prescott, 2002; Kostova & Roth, 2003). 7 Social capital has been defined as a multidimensional concept (Batjargal, 2003; Inkpen & Tsang, 2005; Koka & Prescott, 2002; Nahapiet & Ghoshal, 1998). However, authors have identified and justified different dimensions that define and quantify social capital. Nahapiet and Ghoshal (1998) pointed to the existence of three dimensions of social capital, although they recognize that they are not independent, but that each one affects the others. Tsai and Ghoshal (1998) and Inkpen and Tsang (2005) used these same dimensions in their analyzes of product innovation and knowledge transfer. The structural dimensions refer to the ties in the network that each actor possesses and the actual layout of each network (in the sense of its structure and the patterns in the relations). The relational dimension is linked to the characteristics of the relations between the actors. Finally, the cognitive dimension refers to the codes, languages, narratives, visions and norms that are shared within the network. This last dimension covers the idea of social capital as a public asset (Tsai & Ghoshal, 1998). The cognitive dimension may therefore create the same advantageous opportunities for any of the members in the network and does not lend itself to the study of the different endowments of social capital in each firm in the network. Accordingly, there are constant references in the literature to various consequences that arise between the structural and relational aspects (Gulati, 1999; Inkpen & Tsang, 2005; Koka & Prescott, 2002), in line with the distinction between structural embeddedness and relational embeddedness, suggested by Granovetter (1992). Along with those last two dimensions, Batjargal (2003) added embeddedness derived from resources, as a new dimension of social capital. In line with earlier studies and the ideas of Lin (1999), Batjargal suggested that contacts in networks that offer useful resources for the actor are necessary to improve performance. His evidence for this was 8 taken from the experiences of a group of Russian entrepreneurs. However, he made no explicit proposal for the existence of a resource dimension of social capital. A characteristic feature of the resource dimension is that the rents or the advantages for the focal firm arise, not from its ties or its position in the network, but from the partner resource endowment with which it maintains relations. In other words, a firm can obtain a valuable result from its relations without occupying a central or an intermediary position in the network (it can be a peripheral actor) or without having links of particular intensity or quality, if the firm has access to a partner that holds the resources that it needs. This view of the resource dimension that flows from the ideas of Lin (1999) and of Batjargal (2003) directly connects with the concept of network resources defined by Lavie (2008). 2.2. Network Resources Lavie (2008) enters the debate by pointing to the existence of three definitions of network resources that vary with respect to the ‘scope’ and the ‘locus’ of the construct. The first is the network, as a resource that creates benefits for the firms that belong to it (Gulati et al., 2000; Kogut, 2000). The second refers to the valuable resources that may be obtained from ties with other firms, such that the nature of these ties becomes an essential and defining element in the definition of the firm. The third definition is seen as fundamental by Lavie, as it implies a change in perspective for the study of interorganizational relations. According to Lavie (2008: 548) “network resources are assets that are owned by the firm’s partners, but can potentially be accessed by the firm through its ties to those partners”. 9 These definitions of network resources correspond directly with the three categories of embeddedness and the three structural relational and resourcedimensions of social capital (Batjargal, 2003; Casanueva & Gallego, 2010). Thus, Lavie (2008: 549) commented that, “we can now distinguish between network structures, the nature of relationships, and the resources that are channelled through such network ties”. Gulati (2007) also makes explicit reference to the assets held by partners in his definition of network resources. This perspective not only explicitly describes a third dimension of social capital and of network resources, which had remained latent in earlier theoretical developments and empirical studies, but it also suggests that a focal firm can draw advantages from the partner network, which basically depend on partner resource endowment. From the theoretical point of view, considering the study of interorganizational relations from the Resource Based View and network theory, this new perspective suggests a series of interesting changes: a) The basic reference point for the study of alliances and interorganizational relations moves from the complete network to the egocentric network of the focal firm. b) If the importance of the analysis, in the habitual logic of network theory and its application to interorganizational relations, lay in the study of the ties and their patterns, ascribing relatively little value to the nodes, the importance is now moving from the ties to the nodes (and this is happening without abandoning the network approach). c) The structural and relational dimensions lose weight with respect to the resource dimension in the analysis of social capital and interorganizational relations. 16 will exercise a direct and powerful influence over access to network resources, which have been defined as the resources of their partners. Burt (1992) proposed the study of egocentric network structure as a measure of social capital. Preliminary research into alliance portfolios, understood as the ego-net of the focal actor, have also revealed how the structure of that egocentric network positively affects resources (Gulati, 2007; Wassmer, 2010). Some authors have made it clear that, faced with the limitation of finding internal resources, it is better to achieve them thanks to the relations with other partners that possess them and that this behaviour (Ahuja, 2000b; Hoffmann, 2007) should guide the management of the alliances. Lavie (2006, 2007) makes a direct connection between acceptable management of the alliance portfolio with access and mobilization of the resources of the partners, in other words with the resources of the network. Finally, the relational dimension was also analyzed from various perspectives (intensity of the relations, repetition of the links, relations of trust, etc.) as facilitating access to network resources (Kale, Singh, & Pelmutter, 2000; Koka & Prescott, 2002; Moran, 2005; Nahapiet & Ghoshal, 1998). Tsai and Ghoshal (1998) proposed a similar logic in their analysis of the influence of social capital in product innovations, when they centred on the capacity of interorganizational units to combine and exchange resources; this capacity being the real influence on innovation. Moreover, Rodan and Galunic (2005) pointed out that network structure has been used as a proxy to measure the content of the relations, bypassing direct measurement of their characteristics. These authors suggested a separation between the structure of the network and the contents of the relation and studied its links in the specific case of the heterogeneity of knowledge obtained from different colleagues that is considered valuable. Casanueva and Gallego (2010) explicitly analyzed the connection between the relational and resource dimension, establishing that the latter plays a catalytic role in the results of innovation. 17 Thus, the resource dimension may be considered as the central pivot around which the social capital of a firm revolves. In other words, the value of its ties will fundamentally be conditioned by access to the valuable resources that its partners hold and this access will be conditioned by its position in the global network, by the structure of its alliance portfolio and by the quality of its relations. Moreover, although still scarce, due to the recent conceptualization of network resources, empirical evidence exists that links the capacity to access partner resources with the performance of the firm, measured in different ways (Batjargal, 2003; Casanueva & Gallego, 2010; Lavie, 2007; Lee, 2007). Batjargal (2003) advanced evidence on how access and mobilization of certain resources such as financial ones affected the performance of firms. Analyzing the alliances of the tourism sector, Pansiri (2008) studied how the characteristics of the partners influenced performance to varying degrees. Casanueva and Gallego (2010) proposed showing the direct influence of the resource dimension on innovation capacity, as opposed to the indirect influence of the structural and relational dimensions. Lavie (2007) showed how financial and marketing resources held by partners have a positive effect on the market performance of the firm. Lazzarini (2007) proposed that operational performance in alliances between airlines is influenced by the use of resources shared with the partners. Finally, Wassmer discussed from a theoretical point of view, how network resources within an alliance portfolio can impact on the value of the focal firm and applied this to the airline sector (Wassmer, 2007; Wassmer & Dussauge, 2011). 18 3. Methods 3.1. Network data and selection This research analyzes the alliances that occur in the airline sector at an international level, in order to study the relations detailed in the theoretical model. The airline sector is a mature industry in which a wide range of competitive practices reflect highly intense levels of rivalry. These are firms of a large-size, because of the industry’s own logic, that are present throughout the world. It is likewise a sector with a dynamic movement towards inter-firm relations that range from commercial agreements to integration processes (acquisitions and mergers). These dynamic relations between airline companies have a history that dates back to early airline activity. However, the construction of large-scale strategic alliances between companies that spur competition between groups at a global level has speeded everything up over the past ten years. Thus, these alliances have been the subject of various studies (Evans, 2001; Gimeno, 2005; Gomes-Casseres, 1994; Lazzarini, 2007; Shah & Swaminathan, 2008). The airline sector is especially appropriate for analysis with the proposed model. Airlines are integrated in multiple horizontal cooperative agreements that allow them to set up alliance portfolios to gain access to the resources of their partners and to achieve their goals. The configuration of an airline alliance portfolio gives an airline access to important network resources such as the destinations that complete or complement its network, technological capacity to perform operations away from its bases, reputation in new markets, etc. In this context, it is interesting to examine whether the structure of the alliance portfolios, the quality of partner relationships or the structure of the alliances 19 among all the competitors affect access to network resources and whether this access affects the individual performance of each airline. In line with earlier research (Chen & Chen, 2003; Evans, 2001; Gimeno, 2005), data on the main airlines published by the journal Airline Business was used to study alliances in the airline business sector. A list of 200 leading firms from the sector, published in 2010 and based on income in 2009, was used to choose the sample. Several business groups appeared in that ranking, a detailed analysis of which led to the consideration of a total of 214 airline businesses. All the companies had an annual turnover at the time of over 50 million dollars. The majority of these airlines are from Europe (73), Asia (55) and the USA (51). Moreover, data had to be gathered and identified on all those airlines with which the 214 maintained direct contact, to apply the definition of network resources. In total, data on 351 airline companies and almost 700 alliances from all over the world were analyzed. The data on those companies and on those of the airlines with which they have entered into alliances were extracted from the ATI (Air Transportation Intelligence) database and from the ICAO (International Civil Aviation Authority), provided by the company Flightglobal. These included the broadest datasets from the sector obtained from multiple secondary data sources and bespoke surveys. Data collection and processing took place in April and May 2011. Various types of relations arise among the airline companies at an international level: global alliances, plane transfers or loans, codeshares, cargo alliances, joint Frequent Flyer Programmes (FFP), etc. In this work, the ties considered for the study of alliances between airline companies were the existence of agreements on codeshares and cargo alliances (Gimeno, 2005; Ito & Lee, 2007; Rajasekar & Foust, 2009; Wassmer & Dussauge, 2012). These are agreements on cooperation that start up joint operations between the airlines (commercial, operative, financial, etc.), 20 which require a high degree of coordination. In practice, the start of a relation between two airlines in a codeshare or cargo alliance between both has a limited scope of routes and destinations from one or the other airline and is not necessarily equitative. However, the future of these agreements is the slow expansion of the destinations and the routes that both airlines jointly operate. A matrix of relations was constructed in which the codeshares were considered jointly with cargo alliances still in force, which were entered into between 1960 and 2007. Agreements on regional cooperation and on charter flights do not appear in these data. Neither are other agreements of a commercial nature included, nor marketing agreements such as joint participation in FFP, nor global alliances, which will nevertheless be considered for the construction of some indicators. Consideration of data on airline alliances up until 2007 is because of the effects of cooperation between firms, which will only be felt a certain time after its implementation. Hence, the analysis of the results is for the time span running from 2008 to 2010. 3.2. Structural equation modelling Partial Least Squares (PLS) structural equation models are used, in order to test the hypotheses of the models which relate, on the one hand, to the different antecedents of the structure and the network relations needed for partner access to resources, and, on the other, to how access to these network resources affects performance. This technique is acceptable for this investigation, since theoretical knowledge is still not fixed, the purposes of the investigation are of causal-predictive nature and a complex model needs to be estimated (Barclay, Higgins, & Thompson, 1995; Chin, Marcolin, & Newsted, 1996). For all these reasons, we consider that it is an appropriate tool in an investigation 21 that seeks to describe relational patterns between different concepts, rather than delve into the nature of the relations that are identified. 3.2.1. Dependent variables In our study, we have considered two dependent variables: one formed by the data on airline activity and the other by the data on performance. The data for the calculation of the indicators included in both variables were taken from the ATI database and refer to the different types of business results and data on operations and flights of the 214 airline companies in the sample over the period 2008e2010. Data from earlier years was not used, as it was considered that the effects of social capital are not immediate but will become manifest when they are effectively developed. Thus, data on social capital was considered for the period prior to 2008 and data on performance from 2008. The indicators were constructed on the basis of those three years. Financial Performance. The variable Financial Performance attempts to collect information on the performance of each company in terms of performance and margins on their activities (Cappel, Pearson, & Romero, 2003; Park & Cho, 1997). The use of measures relating to this variable such as Return on Assets is common. However, complete data on many airlines included in the sample could not be obtained for the period under consideration. Finally, the financial indicators that the variable includes are as follows: a) Mresoper: Operating income. A profitability indicator, calculated as the difference between operating revenue and operating costs. If a firm has no non-operating income, the operating income coincides with Earnings Before Interest and Taxes. 22 b) MRusNet: Net results. Net benefit or net result is the profitability ratio of a firm after taking all costs into account. It is calculated by subtracting all direct costs, indirect costs, depreciation, repayments, interests, taxes, and rates, from the sales income. c) MMargenNet: Profit margin or net profit margin. This is a profitability indicator that is calculated by dividing net profit by revenue,and expressing the result as a percentage. It is an indicator of how a firm controls its costs. Investors use the net profit margin to compare firms from the same sector and between sectors, to determine which are the most profitable. Operational Performance. One alternativeway of measuring the performance of an airline company is through its level of activity, taking physical as well as monetary criteria into account (Assaf & Josiassen, 2011). Four indicators are used to measure it: a) MPasaj: Passengers. Total number of passengers transported by an airline in a set period of time. b) MIngPasKm: Passengers-Km. Transported (PKTs). Total number of passengers transported in a plane by kilometre flown. This is a standard unit of production for airline transport, which measures demand. c) MTonneKm: Tons-Km. Transported (TKTs). Tons transported per kilometre flown. It includes all paid cargo, i.e. passengers, goods and post. It is a standard production unit for airline transport that measures the payload, which is equivalent to PKTs. d) MRevTot: Total revenue. This is a financial indicator that covers total revenue obtained by each airline. It includes revenue from the transport of passengers and freight, as well as other revenue. 23 3.2.2. Independent Variables Four independent variables were used to construct our model, which correspond to the different dimensions of social capital that it includes: one that gives information on the structure of the network of airlines, in other words, on the network position of each firm; a second that refers to the structure of the alliance portfolio of each airline; a third, which reveals the quality of the ties that the airlines have with the firms integrated in the network; and, finally, the fourth, which tells us the network resources to which an airline has access through its partners in the network. Structure. The first of the independent variables is intended to capture the combined effects of the position of a firm, within the global network of alliances in the sector, that arise from the different ways of understanding centrality. The indicators in use correspond in great measure to those proposed by Borgatti et al. (1998) for the calculation of social capital as a function of the measures of centrality, all of which are standardized versions. In all, five indicators were calculated: a) nFreemanDegree: Degree centrality. A measure of the direct connections of an actor with the others. b) nCloseness: Closeness centrality. A measure of centrality that takes the indirect relations into account and that is calculated from the shortest distances to access the different actors. c) nBetweenness: Betweenness centrality. An indicator of the betweenness capacity of an actor in the network, calculated by taking into account the dependence of other actors on the focal actor, in order to reach their contacts along the shortest route. 24 d) nEigenvector: Eigenvector centrality. This indicator calculates centrality by taking the existence of different factors into account and refers it to the principal component. e) nFlowBet: Flow centrality. Measures the position of betweenness by considering all the routes and giving greater weight to the shortest. The five variables were calculated from the matrix made up of all the alliances (codeshares and cargo) between the 351 firms under analysis that remained in business up until 2007. We used the Ucinet programme for their calculation (Borgatti, Everett, & Freeman, 2002). Portfolio. We considered a series of indicators referring to the egocentric networks of the airline companies under study, when creating the variable that gave us information on the alliance portfolios of the airlines. The egocentric network is made up of the direct contacts of each actor. In other words, the egocentric network of a focal airline is made up of the relations between all the companies that are allied with that focal airline. The following indicators were used: a) Size: Size of the egocentric network of each actor. In other words, the number of direct partners of the actor, also known as its neighbourhood. b) Density: Density of the egocentric network. Number of actual ties between the partners of the focal actor divided by the number of possible ties. c) 2StepReach: Number of actors at a distance of two positions from the focal actor. d) nEgoBetween: Measure of the centralization of betweenness in the egonet. e) nWeakComp: Number of components (unconnected parts of a network) of the egocentric network. 25 f) EffectiveSize: Measure of Burt’s structural holes (1992) that are calculated by taking the number of partners into account minus the average degree of the other actors within the egonet. g) Efficiency: Another indicator of Burt’s structural holes that is calculated by dividing the EffectiveSize by the number of others in the egocentric network. As happened with the indicators for the variable ‘Structure’, the above were calculated from the matrix formed by all the alliances (codeshares and cargo) using the Ucinet programme. QRelation. We made use of two indicators for the intensity and duration of the ties that the airlines maintain with their partners, in order to measure their quality. We employed data on different types of ties in the calculations provided by the ATI database, taking, for each airline, the data on the years that had elapsed since the alliance was set up until 2007. The choice of 2007 as the final year was due to the fact that our study considered social capital as an antecedent of performance, the data on which we had extracted from 2008 up until 2010. a) Intensity. We centred on seven types of alliances, when analyzing the intensity of the ties that an airline maintains with its partners, all of which were considered in our study: global alliances, marketing, FFP, codeshare, franchise, feeder and cargo alliances. These types of alliances were divided into three groups, given that we thought that some of them implied different degrees of intensity in the ties than others. Thus, we considered that one type of tie of a commercial type (marketing and FFP) was of lower intensity and another of greater intensity, which are operational type (codeshare, cargo, franchise and feeder), and finally, the global alliances were given separate 32 STRUCTURE PORTFOLIO QRELATION RESOURCES FINANCIAL PERFORMANCE OPERATIONAL PERFORMANCE NFLOWBET 0.9499 0.7794 0.6214 0.3269 -0.3037 0.5535 NBETWEENNESS 0.9375 0.7252 0.5689 0.2812 -0.3241 0.5566 NEIGENVEC 0.9037 0.6835 0.6538 0.4413 -0.205 0.4848 NWEAKCOMP 0.7673 0.9138 0.6464 0.5014 -0.2394 0.4727 2STEPREACH 0.8642 0.8888 0.8002 0.6552 -0.2513 0.5457 NEGOBETWEEN 0.4268 0.8601 0.5749 0.5615 -0.0968 0.2863 INTENSITY 0.644 0.6879 0.8832 0.584 -0.1853 0.4036 DURATION 0.5361 0.6675 0.8799 0.5765 -0.1295 0.4407 MSPUPPLIER 0.4312 0.6748 0.6718 0.9515 -0.1264 0.2981 MFLEET 0.3284 0.5765 0.6171 0.8753 -0.0744 0.256 MLEADER 0.3384 0.5746 0.5535 0.8569 -0.1237 0.1801 MRRHH 0.3284 0.4601 0.4427 0.7023 -0.061 0.2218 MDESTIN 0.0892 0.2952 0.3151 0.670 -0.0684 0.0558 MRUSNET -0.1785 -0.1642 -0.1416 -0.1204 0.9263 -0.2595 MMARGENNET -0.388 -0.2699 -0.153 -0.0786 0.7604 -0.2118 MRESOPER -0.1347 -0.0858 -0.1452 -0.0504 0.6411 -0.219 MINGTOT 0.6057 0.5043 0.4891 0.2609 -0.3003 0.975 MINGPASKM 0.5174 0.4355 0.4393 0.2195 -0.3662 0.9245 MPASAJ 0.4004 0.3213 0.3595 0.1879 -0.3102 0.8547 MTONNEKM 0.4148 0.4313 0.3567 0.2326 -0.0408 0.7068 Table 3: Cross-loadings of the measurements. Having demonstrated the reliability and the validity of the measurement model, in the second stage, it is necessary to evaluate the structural model itself. It is now a question of showing the relations between the different variables involved in the model and the degree of significance of those relations. Graph 2 shows the explained variance of the dependent variables of the model (R2) within the circle that represents it and the standardized regression path coefficients (b weights) for each of the relations in the model. Following Chin (1998), the non-parametric Bootstrap technique was used with 1000 sub-samples and the t student was calculated to determine the significance level of the (b) path coefficients. 33 *** p>0.001; ** p>0.01; * p>0.05 Graph 2: Results of the structural model Graph 2 shows us that the explanatory power of the variables in use (R2) is high, 0.543, in the analysis of the different aspects of the interorganizational relations and their influence on access to network resources, understood as resources held by the partners. Studying the coefficients of each variable, it may be seen that the relations are significative (with greater clarity in the case of the structure of the complete network, for the structure of the alliance portfolio, and for the quality of relations). The positive relation, with a high coefficient (0.613), is prominent between the structure of the alliance portfolio of a company and its access to resources. However, the relations are positive between the structure of the alliance portfolio and the quality of the relations in the alliances and the endowment of network resources, and they are negative between the positions of the firm in the network and the resource endowment of the partners. 34 Differences with regard to the inverse sense of the relations between the structure of the complete network and the structure of the alliance portfolio of an actor, lead us to suggest that attention should be focused on the set of relations that arise between the firms in the portfolio of the focal firm (between its partners), more so than on the general position of that firm in the network of alliances in its sector, in order to understand its more favourable access to resources and its greater social capital. Moreover, the influence of network resources on business performance also varied in the two groups of results: the relation was positive and significative when the influence of the endowment of network resources on operational performance was analyzed (using income, operational indicators and passenger volume). With respect to their influence on financial performance, the results of the model do not suggest any immediate conclusions. In general, the financial results are influenced by a very wide set of variables, many of which of a temporary character, such that access to partner resources is perhaps not a variable that influences it more directly and more in the short term. In all cases, the explained variance for both variables is very small, the R2 of the influence on the activity of the airline being only 0.069. Hence, there are other factors that can explain the behaviour of this variable better than the endowment of business network resources. 5. Conclusions, Limitations and Future Lines of Research The purpose of this investigation has been to understand the role that network resources play as part of the firm’s social capital and how these resources affect firm performance. The first interesting result is the differentiation of four components or dimensions in the social capital of the firms arising from the composition, over time, of their alliance portfolios. The widely used structural embeddedness dimension was 35 divided, on the one hand, into a dimension linked to the position of each firm in the global structure of alliances within the sector and, on the other, into the structural patterns of the alliance portfolio of each company, understood as the characteristics of its egocentric network. But above all, the resource dimension, a reflection of network resources (Gulati, 2007; Lavie, 2006), has shown itself to be a basic reference in the study of social capital, to the extent that it centres on the appropriation of resources present in a network, more than on its creation or on the endowment of the resources of each firm. The analysis of the airline sector at an international level confirms that the endowment of company network resources, understood as the resources held by the partners with which a firm has direct relations, is linked to other dimensions of social capital arising from the set of alliances of a particular firm. In particular, the structure of the alliance portfolio positively affects the endowment of different types of resources (physical, human, technological, marketing, and reputation) that may be obtained from partners. Likewise, the characteristics of the dyad ties in the alliances, measured in terms of the duration of the alliances and the intensity of the relations in the different forms of alliances, also affect the endowment of the network resources of the airlines. On the contrary, the position of an airline in the complete network of the sector at an international level is negatively related to access to network resources. This negative relation appears to cast doubt on the idea of network structure as an antecedent of access to resources and their mobilization (Casanueva & Gallego, 2010; Lin, 1999). However, it does have a positive effect on the structure of the egocentric network. As pointed out earlier, the firm’s position in the global structure of the network does not necessarily have to be linked to the structural patterns of its egocentric network. Those firms that occupy the most central positions in the global network, both in the sense of 36 access to other network positions and in the sense of serving as a path for different connections between its members, have a smaller endowment of network resources in relative terms; a fact that may be explained in various ways. The first is that these more central firms are usually linked to a broader number of partners; in other words, they have larger neighbourhoods, in which there are firms with a greater endowment of resources and others with smaller endowments and, as relative measurement indicators were used, the average value is reduced on account of the latter firms. The second explanation could be that the resource endowment of the more central firms is already sufficient and the choice of partners in their alliance portfolios is not conditioned by access to those resources, but by other parameters such as the quality and the complementarity of the relations. Finally, indicators were eliminated from the structural equation models (to guarantee validity, and the reliability of the scales of measurement in use) that might vary the composition of a variable and its relations with others. If airlines try to mobilize the strategic resources of their partners (destinations that complete their network, technological support and maintenance, reputation, etc.) they should manage the composition and structure of their egonet in an acceptable way (Alliance portfolio), and concern themselves less with their position in the global sector. Moreover, the data obtained from the alliances in the airline sector also allow us to confirm that the resource endowment of the network partially affects the performance of a firm. This finding is consistent with earlier studies which have shown that performance is linked to access to the firm’s resources. For example, Batjargal (2003) using a generator of positions, arrived at the same conclusions. However, the results only 37 showed this link in a significative form in the case of the activity indicators of the firm (its volume of operations), but not for its financial results. The majority of studies on alliance management have centred on the relations that arise between two isolated firms (Chen & Ren, 2007; Evans, 2001; Iatrou & Skourias, 2005; Lazzarini, 2007; Rajasekar & Foust, 2009; Shah & Swaminathan, 2008;Weber, 2005). These are two-to-two relations. Given that what is of real interest in the creation of an alliance is what the agreement brings to its members, various studies have centred on establishing whether alliances between airlines have an impact on some performance variable (Chen & Ren, 2007; Evans, 2001; Iatrou & Skourias, 2005; Lazzarini, 2007; Rajasekar & Foust, 2009; Weber, 2005). Our study shows that what really affects the results are not the dual relations, but the possibility of mobilizing the resources in an alliance network. These links in the network of alliances allow access to the valuable resources of the partners. As Wassmer and Dussauge (2012) have also pointed out, that kind of mobilization of resources is not produced in the global network of alliances, but in the egonet of the firm. This work shows clear evidence of that idea. Therefore, instead of centring, as airlines do, on their bidirectional relations with their allies or on the management of the global network, it would be more advantageous for airlines to centre on the management of their alliance portfolio (De Man, Roijakkers, & De Graauw, 2010), given that this would allow them to mobilize resources that will improve their results. In the airline sector, in which the alliances between firms is in constant growth, each airline will possess an alliance portfolio that comprises other airlines with which it has maintained business relations over time. The governance of this alliance portfolio can condition the results of the airline, at least with regard to operational aspects, which will 38 have consequences for its medium and long-term performance. This work has shown that, rather than global positioning in the network of alliances in the sector, it is advisable for a firm to have an acceptable structure in its alliance portfolio, in which it appears that both direct and indirect contacts are very important, in order to access partners with valuable resources (human, material, marketing, technological and reputation). The performance of airlines with partners that have more resources appears to be better, hence the choice of partners should be based on their resource endowment. The complexity involved in the analysis of the resource endowment of potential partners and the governance of the networks and sub-networks that the airlines attempt to set up, means it is advisable to set up and/or develop a specific function within each airline business. Satisfactory exploitation by an airline of its alliance portfolio should take into account its composition as well as its management. Thus, it would be interesting if airlines set up a unit or created posts that specialize in the management of alliance portfolios. This is in fact already happening and is a reality in some airlines, such as KLM, where Henk de Graauw directs a department, which is principally concerned with the strategic management of alliances. Finally, the limitations of this investigation should be pointed out. Among others, it is clear that the set of firms from the airline transport sector is very diverse in size, scale, strategic position, marketing strategies, services offered, cost structure, pricing policy, etc. for which reason the aggregate treatment of all the firms can give rise to incorrect conclusions, which would make it necessary to consider strategies that would allow these effects to be monitored. Another important limitation is that time has not been taken into account in the investigation. The effects at different moments in time of the variables and the evolution in the relations would improve with an analysis of the results in which time was taken into account. Besides, not all of the factors have been considered that might 39 affect performance, nor have larger models that include different effects from those that arise within the different dimensions of social capital. Neither have other possible links been established between the different dimensions of social capital under consideration. We have not studied the influence of each concrete indicator on the dependent variables, but instead their influence on a factor. Finally, there are pertinent indicators, from the theoretical point of view, that have been eliminated to assure the goodness of the measurement scales. We were aware of these limitations even before the results were obtained, because as we indicated at the start of this work, our interest did not reside in conducting an exhaustive study of the relations that arise between the variables, but of observing the relational patterns that arise between them. Future lines of research are therefore linked to the abovementioned limitations. One possible way of extending this work would be to conduct a detailed analysis of the effect that the model has on different segments or strategic groups in the airline transport sector, such as flag carriers, low cost sector, air freight transport, etc. In second place, given that data is available over various years, the temporal effects could be studied on the endowment of network resources as well as on the relations between social capital and the results of the firms. In third place, the scope of the investigation may be increased by incorporating new variables that would help to explain the observations, through the use of other statistical tests that might analyze the indicators in greater depth or through the study of other possible relations between the dimensions of social capital. And, finally, we wish to look more deeply at the direct impact of the most relevant indicators, in order to construct the independent variables from the indicators of the dependent variables, by applying more appropriate techniques such as linear regression. 40 References Adler, P., & Kwon, S. (2002). Social capital: prospects for a new concept. Academy of Management Review, 27(1), 17-40. Ahuja, G. (2000a). Collaboration networks, structural holes, and innovation: a longitudinal study. Administrative Science Quarterly, 45(3), 425-456. Ahuja, G. (2000b). The duality of collaboration: inducements and opportunities in the formation of interfirm linkages. Strategic Management Journal, 21(3), 317-343. Assaf, A. G., & Josiassen, A. (2011). European vs. U.S. airlines: performance comparison in a dynamic market. Tourism Management, 33(2), 317-326. http://dx.doi.org/10.1016/j.tourman.2011.03.012. Barclay, D., Higgins, C., & Thompson, R. (1995). The partial least squares (PLS) approach to causal modelling: personal computer adoption and use as an illustration. Technology Studies. Special Issue on Research Methodology, 2(2), 285-309. Barney, J. B. (1991). Firm resources and sustained competitive advantage. Journal of Management, 17(1), 99-120. Batjargal, B. (2003). Social capital and entrepreneurial performance in Russia: a longitudinal study. Organization Studies, 24(4), 534-556. Baum, J. A. C., Calabrese, T., & Silverman, B. S. (2000). Don’t go it alone: alliance network composition and startups’ performance in Canadian biotechnology. Strategic Management Journal, 21(3), 267. Borgatti, S. P., Everett, M. G., & Freeman, L. C. (2002). Ucinet 6 for Windows: Software for social network analysis. Natick: Analityc Technologies. Borgatti, S. P., Jones, C., & Everett, M. G. (1998). Network measures of social capital. Connections, 21. 41 Borgatti, S. P., & Li, X. (2009). On network analysis in a supply chain context. Supply Chain Management, 45(2), 5-22. Burt, R. S. (1992). Structural holes: The social structure of competition. Cambridge MA: Harvard University Press. Cappel, S. D., Pearson, T. R., & Romero, E. J. (2003). Strategic group performance in the commercial airline industry: strategic response to structural disequilibrium. Journal of Management Research, 3(2), 53-60. Carmines, E. G., & Zeller, R. A. (1979). Reliability and validity assessment. Sage University paper series on quantitative applications in the social sciences. Beverly Hills: Sage. N. Casanueva, C., & Gallego, A. (2010). Social capital and individual innovativeness in university research networks. Innovation: Management, Policy & Practice, 12(1), 106-118. Chen, Ch., & Ren, Y. (2007). Predictive methods for using capacity data to estimate market shares and extent of risk pooling by airline alliance partners under parallel codesharing. Transportation Journal, 46(4), 21. Chen, F. C.-Y., & Chen, C. (2003). The effects of strategic alliances and risk pooling on the load factors of international airline operations. Transportation Research, 39, 19-34. Chin, W. W. (1998). The partial least squares approach to structural equation modelling. In G. A. Marcoulides (Ed.), Modern methods for business research. New Jersey: Mahwah: Lawrence Erlbaum Associates Publisher. Chin, W. W., Marcolin, B. L., & Newsted, P. R. (1996). A partial least squares latent variable modeling approach for measuring interaction effects: results from a Monte Carlo simulation study and voice mail emotion/adoption study. In J. I.