DETERMINANTS OF THE CHOICE OF MARKETING CHANNELS BY CORPORATE CLIENTS : AN ANALYSIS OF THE INFORMATION TECHNOLOGY SECTOR

Based on a survey of 505 corporate clients from the information technology industry, this study uses a logit statistical model to analyze whether such factors as loyalty, trust, client importance, intimacy between customer and supplier and the cost of changing suppliers may affect the choice of the marketing channel used by the client. The results show that the greater intimacy between a supplier and a corporate client is a determining factor for the client’s preference for being served by the manufacturer rather than an intermediary. In contrast, clients’ perception of their own importance to the supplier is not very relevant to their preference for being served directly by the manufacturer, and the other factors were not identified as (statistically important) determinants of the choice of the marketing channel. These results suggest that the real advantage of the direct sales channel over the indirect sales channel lies in its ability to strengthen the clientmanufacturer relationship, thereby contributing to the development of solutions that better serve the needs of corporate clients.


INTRODUCTION
The generalist nature of Information technology (IT), which economists refer to as a general-purpose technology (Helpman & Trajtenberg, 1998), has contributed to unprecedented changes for the business world and society as a whole (Brynjolfsson & Saunders, 2010).With the increasing transformation of digital media (Acker, Gröne & Schröder, 2012), these changes appear to be far from ending.As a way to understand part of the impact of IT on organizations, some scientific papers have already investigated the way in which companies manage information technology (e.g., Rodrigues, Maccari & Simões, 2009;Simões, Rodrigues, Maccari & Pereira, 2011) and analyzed the impact of IT on functional areas (e.g., marketing - Khauaja & Campomar, 2007) and organizational performance (e.g., Mahmood, 1994;Maçada, Becker & Lunardi, 2005).However, little is known regarding the relational dynamics in which organizations involve themselves to provide or gain access to information technology.Therefore, this study aims to fill this gap by analyzing how attributes of the supplier, client and their relationship affect the decisions of corporate clients in their choice between marketing channels to gain access to IT products.
Marketing or distribution channels have been considered for some time as economically relevant interorganizational arrangements, not only to directly involved organizations, but also to society (e.g., Reve & Stern, 1979).The competitive dynamic in many markets has led companies to pay more attention to channel management as a means to achieve competitive advantages (Cônsoli & Neves, 2008).As stated by Swift (2001), the greater intensity of competition and the emergence of new consumption patterns have induced many companies to redefine the value proposition that they offer to their clients.This behavior has been reflected in the evolution of research conducted during recent decades, and interest in the topic of 'marketing channels' has been increasing (Wang, Bradford, Weitz & Xu, 2008).The increasing attention devoted by the marketing literature to this subject is expected.Given that the marketing process is essentially characterized by an act of exchange and that the marketing channel is just the vehicle by which this exchange occurs (Gordon, 2003;Kotler, 2006), understanding how the agents of this relationship react to attributes of the channels or determining the antecedents that induce the choice of a particular channel is central to advancing our knowledge regarding this relationship (Sharma, Tzokas, Saren & Kyziridis, 1999;Sharma & Mehrotra, 2007).
The evolution of relationships between marketing channel agents is mostly due to the evolution of information technology and communication, which have allowed the development of new tools to bring suppliers and buyers closer and to make new exchange relationships viable (Liu, Wang & Chan, 2004).The result has been an increase in the number of companies (i.e., IT manufacturers) that adopt multi-channel strategies, i.e., that act in more than one marketing channel simultaneously (for example, having their own sales team and simultaneously, using independent intermediaries to market their products) (Wind & Mahajan, 2002).These structural forms are also called plural (or hybrid) forms, i.e., manufacturers that perform a particular activity on their own (in this case, the sale of goods and associated services) and, simultaneously, subcontract that activity to the market (Heide & John, 1992).
For many years, the analysis of this method of governance has been based on the theoretical constructs of transaction costs economics (TCE) developed by Williamson (1979).One of the assumptions of this theoretical model is that exchanges between the parties depend on a contract to oversee the relationship, and because the contract is unable to cover all contingencies, sometimes the transaction costs will be sufficiently high to justify the internalization of the activity (vertical integration).However, buyersupplier relationships are not always governed by contract; for example, it is not unusual that relationships of this kind are maintained and guided by relational exchange (Macneil, 1980;2000).Thus, given TCE's incapacity to explain relational governance, many scientific studies based on social exchange theory (SET) have emerged to explain transactions between companies (Dwyer, Schurr & Oh, 1987).As noted by Lambe, Wittmann and Speakman (2001), SET is particularly useful to explain the business-tobusiness (B2B) exchanges because it focuses on the quality of the relationship between the involved parties in addition to contractual issues.
However, most studies on the structure and management of channels have analyzed the relationship from the supplier's point of view, i.e., the factors deciding whether suppliers choose one or another method (or both) to distribute their products.Additionally, much about the literature regarding channels is related to the business-toconsumer (B2C) context, i.e., to the consumer market rather than the corporate market.The literature regarding marketing channels in the corporate (B2B) market is less significant than the literature regarding the consumer market.Furthermore, the marketing literature has studied the effects of the attributes of agents and relationships on the continuance of these relationships, but less emphasis has been placed on exploring much how these attributes effectively affect the client's decision to prefer one channel to another (Rosenbloom, 2007).The present study fills in this gap by identifying the determinants of the choice of the marketing channel used by corporate clients.
The study is based on the assumption, which has been empirically documented in the literature, that marketing channels can serve buyers' needs in different ways (Shipley, Egan & Edgett, 1991).Likewise, the present study is similar to studies on buyer's perception and how it influences the buyer's decision (e.g., Ulaga, Sharma & Krishnan, 2002).Particularly, the present study analyzes how corporate clients' perceptions about various aspects affect their decision of the choice of channel through which they will be served.These aspects are the following: their loyalty to the supplier, the trust they have in the supplier, the importance they are given by the supplier, the intimacy that they can develop with the supplier and the cost to change suppliers.Therefore, a survey of corporate clients from the information technology sector was conducted.The final sample included approximately 500 companies, and data were analyzed by statistical methods to ensure the validity of the constructs (i.e., principal component analysis) and to understand the impact of these constructs on the corporate client's decision of being served directly by the supplier or intermediaries (i.e., the logit model).The results revealed that the main determinant of these clients' preference for a channel is the intimacy level established in the relationship, and that, in this case, it was favorable to the manufacturer's own sales force channel.
The next section of this article reviews the literature that motivated the present research and the formulation of the hypothesis tested in this study.Then, the methodology is presented in detail.The subsequent section describes and discusses the results.Finally, conclusions are presented.

The Information technology Sector
Information technology (IT) can be defined as "the set of non-human resources dedicated to the storage, processing and communication of information, and the way in which these resources are organized into a system capable to perform a set of tasks" (Bakopoulos, 1985, p. 7).Its elements include hardware, software and adjacent services (Duarte & Castello Branco, 2001;Rodrigues Filho & Ludmer, 2005).
According to the IDC1 data, the decade from 2000 to 2010 was very prosperous for Brazil, and, according to the same source, the Brazilian IT market should continue to grow at twice the global average.For the Brazilian IT market in general, the IDC expects growth of 8.8% in 2012 compared to the previous year.The hardware segment is expected to grow by 7.2%, software 11.4%, services 11.1% and Telecom 7.2%.The IDC also estimates that there are almost 15,000 sales channels in the Brazilian information technology segment representing 199 IT manufacturers.
The IT department has a key role in both the design of the future operations of these IT companies and in implementing their strategies and achieving greater competitiveness (Venkatesan, Kumar & Ravishanker, 2007).Since the industrial revolution, technology has been considered a major factor of increased productivity and welfare (Helpman & Trajtenberg, 1998).According to Brynjolfsson & Saunders (2010), the IT sector is characterized by intense competition and consolidation because it has advantages related to learning, is strongly affected by globalization and experiences pressure for cost rationalization and a reduction in the life cycle of technology and products.For example, in the mid-1980s, the life cycle of a product was, on average, one year.A similar analysis carried out in 1999 observed a shrinking on this cycle to an average of three months (Curry & Kenney, 1999).Given these competitive and technological dynamics, companies in this sector have to frequently address price and margin reductions.To compensate for low margins, companies seek to increase their sales to achieve economies of scale (Gabrielsson, Kirpalani & Luostarinen, 2002).In the corporate segment, one of the actions adopted by IT equipment manufacturers in their attempts to scale up has been the use of multiple marketing channels (Dutta, Bergen, Heide & John, 1995;Merrilees & Fenech, 2007).

Marketing channels
The focus of the present study is on computer products (hardware), which are categorized as tangible assets associated with services.Examples of computer products include printing and imaging products, such as printers and scanners; access products such as desktops (desktop computers) and laptops (mobile computers) and enterprise products such as servers.These products are categorized as tangible assets associated with services because the products consist of a tangible asset associated with one or more services.Consequently, the greater the complexity of these products, the more their sales depends upon the quality and availability of services for the client (Kotler, 2006).
Manufacturers use different marketing channels so that these products can reach corporate clients.Marketing channels are characterized by a set of interdependent organizations involved in the process of making a product or service of a company available for consumption or use (Coughlan, Anderson, Stern & El-Ansary, 2001), i.e., the marketing channel is the contractual organization that the company operates to achieve its distribution goal (Rosenbloom, 2002).Because decisions regarding distribution are more costly to change than decisions regarding prices and even advertising, the formatting of marketing channels reflects the long-term commitment of a company to the market, and the efficient and effective use of these channels has proven to be an important component of companies' competitive strategy (Rangan, 1987;Sharma & Mehrotra, 2007).
Marketing channels may present different configurations, but, in general, they can be grouped in two ways: i) direct distribution to clients without the use of intermediaries (i.e., wholesalers/ retailers); and ii) indirect distribution, which assumes the sale to some intermediary/representative who will subsequently sell the product (Coughlan et al., 2001).The direct sales team is formed by employees of the supplier that perform pre-sales services, the sale itself and after-sales services.The main difference between the manufacturer's own team and intermediaries is the lack of an employment relationship with the manufacturer in the former case.Another difference between direct distribution and intermediaries is exclusivity regarding the manufacturer, which limits the products portfolio of the direct sales force.In addition, supervision and monitoring by the manufacturer of its own sales team are much more intense than the supervision and monitoring of sales representatives (i.e., intermediaries).The greater ease of monitoring by manufacturers leads them to prefer the use of their own sales team.Nevertheless, the market discipline imposed on intermediaries (i.e., their survival depends on their performance) partly mitigates the problems of control and monitoring (Dutta et al., 1995).
Sales representatives (i.e., intermediaries) are responsible both for representing the supplier to the client and the client to the supplier, i.e., they are sales specialists.In general, representatives are not exclusive.In other words, representatives represent multiple manufacturers offering related products, i.e., they represent multiple manufacturers to multiple buyers.Representatives have historically behaved as distributors not only of assets, bus also of services (Anderson & Weitz, 1989;Dutta et al., 1995;Stem & El-Ansary, 1982).The profile of the sales representatives varies according to the portfolio of products offered.In general, in the industrial market, representatives are engineers with some experience and a technical profile.In the marketing of products to the final consumer, the commercial component is predominant (Corey, Cespedes & Rangan, 1989).
Additionally, the hybrid (or mixed) sales structure, also referred to as a "multichannel" structure, consists of the simultaneous use of two groups (Coughlan et al., 2001).The growth of this hybrid structure occurs mainly because companies need to adapt their products and services to the needs of their target audience (Kotler, 2006), and this is no different in the corporate market (Rosenbloom, 2007).As noted by Sharma and Mehrotra (2007), using more than one channel increases the exposure of consumers to manufacturer offers and facilitates consumer access to those deals.However, operating with two marketing channels simultaneously is not a trivial task for the manufacturer because conflicts may exist between the channels due to the interest in both of these channels by the same buyers.To mitigate these conflicts, channels were developed to meet the needs of specific customers and are typically segmented by their sizes (Wiertz, de Ruyter, Keen & Streukens, 2004).However, marketing channels originally have distinct purposes.The use of intermediaries, for example, generates gains in efficiency through their specialization in activities that the direct sales force is not directly concerned with (e.g., inventory and transportation).In contrast, an emphasis on efficiency may compromise the effectiveness of manufacturers' offers that do not fully meet the needs of customers (Gordon, 2003).
Indeed, evidence exists that marketing channels have different attributes according to consumer perception.Shipley, Egan and Edgett (1991), for example, concluded that the direct sales force of a manufacturer has a potential advantage in terms of product (i.e., specifications setting and technical knowledge) and price (i.e., negotiating flexibility) when compared to independent channels (i.e., intermediaries).In contrast, according to the authors, these intermediate channels have an advantage in developing relationships.They also discovered that almost half (48.2%) of the buyers exhibited no preference for repurchasing using either channel.However, the vast majority of buyers that exhibited a preference for a specific channel showed interest in buying directly from the manufacturer rather than the intermediary (44.7% vs. 7.2%).According to the authors, the reason for the difference is that the buyers' purchase choice is mainly driven by economic motivations, in which the direct sales channel performs better than intermediaries.These results raise questions about the importance of relationships in the choice of the channel by the corporate client.
Despite the efforts of Shipley et al. (2001) to understand how the channels differ in several aspects in the perception of corporate clients, they did not analyze whether these differences, individually, make buyers prefer one channel to another.In addition, considering that their results suggest that the purchasing decision is mainly affected by economic motivation, a gap exists in our understanding of how other attributes may affect the buyer's decision by either channel.As Jain, Benyoucef and Deshmukh (2009) observed, the selection of the supplier is a major purchasing decision; therefore, according to the authors, initiatives that promote greater understanding regarding how this selection may be affected are welcome.However, as observed by Katsikeas, Paparoidamis and Katsikea (2004), research efforts rarely identify how the channels are perceived in the buyer ´s selection criteria.Similarly, although many studies have examined the collaborative relationships between supplier and buyer, little is known about how attributes of the buyer, the supplier and the buyer-supplier relationship affect the choice of the service channel by the buyer; the present study aims to fill this knowledge gap.

Buyer-supplier Relationship
In the last two decades, the corporate or B2B relationship has been studied in the industrial marketing literature.This literature generally divides the corporate relationship into two main groups: transactional (or discrete) and collaborative relationships.The transactional relationship is characterized by low interdependence, short-term commitment, trading conditions prescribed in contracts (terms and conditions), few communication channels, low level of confidence and low asset specificity.In contrast, the collaborative relationship is characterized by strong interdependence, high levels of trust and commitment, continuity for long periods, high transaction costs with terms and conditions strictly specified and high asset specificity (Macneil, 2000).
According to Bunduchi (2008), the transactional relationship can be categorized by economic exchanges of goods and services between the parties whereas the collaborative nature of inter-organizational relationships depends on the logic of the transaction cost economics associated with the need for trust and interdependence between the parties, as contemplated by the SET.That is, the transactional relationship has been studied mainly from the perspective of TCE, whereas relational exchanges are studied, even if implicitly, based on the theory of social exchanges (e.g., Lambe, Wittmann & Speakman, 2001;Morgan & Hunt, 1994).Recently, a new theoretical model has been discussed in the marketing literature that particularly emphasizes the need to address the area's theoretical propositions based on a logic dominated by services (service-dominant logic) (Vargo & Lusch, 2004).This initiative is somewhat developed in relation to trade between companies (e.g., Paulin & Ferguson, 2010;Sheth & Sharma, 2008), but is still incipient.Thus, the present study will be based on more developed theories of transaction cost and social exchange to substantiate the arguments proposed.
TCE has as its main premise that companies will internalize the activities for which the costs of transacting in the market are high, i.e., when transaction costs are high, the governance structure of the relationships will be more efficient if the activity has been vertically integrated.According to this theory, higher transaction costs are derived from the inability of certain contracts to contemplate all future contingencies and, as a result, specific investments in the buyer-supplier relationship are subjected to opportunistic behavior by one of the parties (Williamson, 1986).In the specific case of marketing channels, the decision would be between a company having its own sales force or outsourcing this function to the market, i.e., delegating the distribution and marketing of products to intermediaries.Although transaction cost theory treats this decision as binary (i.e., make or buy), it recognizes the existence of hybrid structures of governance, such as plural forms in which companies use both alternatives, i.e., provide their own sales force and also utilize intermediaries (Williamson, 2002).This hybrid form benefits from the use of information regarding the costs and profitability of internal channels to assist in negotiations with independent channels and promotes greater motivation for the improvement of the performance of internal channels when facing threats of increased use of the market by the company.Similarly, such a structure allows the development of internal supply channels as hold-up safeguards of independent suppliers (Williamson, 1991).
While this theoretical model is instructional in explaining why the supplier establishes one or another channel (or both), it is not very enlightening regarding the buyer's decision.Similarly, this model assumes that buyer-supplier relationships only occur due to the existence of a contract.However, buyer-supplier relationships are not always governed by a contract or guided by transactional relationships.Thus, given the inability of TCE to explain the relational governance, many scientific studies have emerged based on the social exchange theory to explain the collaborative relationship between buyer and supplier (Claro & Claro, 2010).The basic premise of SET is that agents enter into and maintain relationships because they have an expectation that they will benefit from them (Blau, 1964).According to Lambe et al. (2001), SET postulates that exchange interactions involve both economic and social results.Although economic rewards are important, social rewards (e.g., satisfaction) are also important, and both are simultaneously analyzed and compared to an alternative (Homans, 1958).This judgment about the rewards depends on the perception of the involved parties.In cases where the buyer is involved, the judgment should be guided both by their own attributes and by the attributes of the supplier and the attributes of their relationship (Palmatier, 2008).This study does not intend to be exhaustive and, therefore, merely investigated factors commonly cited in the literature as influences on the relationship: loyalty (Liu, Su, Li & Liu, 2010), trust (Palmatier, 2008), importance (Bunduchi, 2008), intimacy (Román & Martín, 2008) and switching costs (Dutta et al., 1995).

Buyer's attributes
One of the most discussed elements in marketing literature is loyalty.Loyalty is a deep commitment to the consistent repurchasing of a product/service in the future, thus causing repeated purchasing of the same brand or set of brands without concern for situational influences and marketing efforts that have the potential to cause changes in behavior.Thus, loyalty is characterized mainly by a repurchasing commitment in a manner that is consistent and not concerned about situational factors (Olivier, 1999;Vidal, 2012).Reichheld (1996) believes that loyalty should be treated as a process of value creation and notes that loyalty is intrinsically linked with value creation.Loyalty measures whether a company has offered superior value or not: either customers return to buy more from the same supplier or they buy from another supplier.As a result, loyalty promotes a series of "economic effects" such as an increase in revenues and market share, improved quality, improved productivity and greater efficiency in customer service (Flint, Blocker & Boutin Jr., 2011;Liu et al., 2010).
When sales channels (direct or indirect) are considered, the range of offerings presented by the channel with intermediaries (i.e., indirect) can give additional value to the client through higher subsidies for their decision on what they will acquire.In other words, the corporate client may compare different value propositions being offered by different manufacturers.In contrast, when being served directly by the manufacturer, the client may have more conditions to receive concessions during negotiation (Shipley et al., 1991).In addition, by becoming loyal to a manufacturer, it is expected that the client demands a closer relationship given that the company-client link is already consolidated, thereby resulting in a perception of a lower variation in value compared to alternatives offered by competitors.Thus, in a multi-channel context, it is expected that the corporate client prefers to buy directly from the manufacturer rather than third-party representatives; therefore, the following hypothesis is formulated: Hypothesis 1: The greater the loyalty shown by a client, the higher the probability of this client being served by direct sales channels.

Supplier attributes
Another recurrent factor in the literature on channel relationship is trust.Trust is present in interpersonal relations and is defined as the desire of an agent to increase their vulnerability to the actions of another agent.The same can be applied to relationships between organizations (Moorman, Deshpande & Zaltman, 1993).Anderson and Narus (1990) defined trust as the belief that a company will carry out actions that will bring added value to the company and will tend not to take emergent actions that result in value destruction.Trust is seen as a key ingredient to a successful relationship (Dwyer, Schurr & Oh, 1987;Morgan & Hunt, 1994).Some studies (e.g., Palmatier, 2008) note that trust is strongly linked to the perception of values such as honesty and integrity in the other party and contributes to the stability of the relationship, i.e., trust is essential to relationships between clients and suppliers (Bunduchi, 2008;Claro & Claro, 2004;Doney, Cannon & Mullen, 1998;Galford & Drapeau, 2002).The companies that most value trust are the most interested in maintaining long-term and more stable relationships.If trust is built on partnership, opportunistic behavior is less likely to occur given that partners will envision the longterm implications of such behavior rather than seeing only the short-term return (Gassenheimer, Hunter, & Siguaw, 2007).That is, the behavioral repertoire is more directed at cooperation than opportunism (Hill, 1990).As a second role, trust may also be considered as a substitute for hierarchical governance.The idea is to meet organizational goals between companies constituting the partnership when the propertybased control (authority) is not strategically or economically viable (Paulin & Ferguson, 2010).The existence of trust or normative governance provides an environment in which related parties can work together, even in the presence of uncertainty (Roy, Sivakumar & Wilkinson, 2004).
Thus, the trust that a client places in their supplier will make the client more receptive to accepting the value proposition of the supplier, eventually without questioning its accuracy.Therefore, when an offer is presented to the client through multiple channels (direct or indirect), the client understands that the service mediated by an independent representative may be limited when compared to the service provided by the manufacturer, even if the client has some trust in the intermediary.In the direct relationship with the supplier, the client must perceive a more favorable scenario by the supplier's interest in more effectively resolving any obstacle that hinders the supply of the product to the client.This is expected because, in the indirect service mode, the intermediary company may have some restriction on service because situations exist whose nature precludes direct interference of the intermediate.Another factor that favors greater trust in the manufacturer's portfolio consists in the knowledge the manufacturer has about the offered products.Therefore, in a context where the buyer can opt for one or another channel: Hypothesis 2: The greater the trust placed by the client in the supplier, the higher the probability of being served by direct sales channels.
The strengthening of the relationship may represent the value that suppliers give to clients, i.e., the importance given to the relationship.This is a key point in establishing a good relationship between the parties.Exclusivity provides credibility, making the client-supplier relationship much closer and providing not only immediate but also future returns (Churchill, 1996).The importance that the provider gives to the client is associated with the client's long-term investment in the relationship (Kotler, 2006).This greater importance given to the client provides the client with the needed security to make more important purchases from this supplier (Patterson & Dawes, 1999).Clients' perception of their own importance to the supplier also provides them more safety regarding not being subjected to hostile behavior by the supplier (Gassenheimer, Hunter, & Siguaw, 2007).
Both manufacturers and intermediaries can demonstrate the importance of the clients.The manufacturer, however, is responsible for serving a more extensive number of clients than the representative intermediary because, in theory, the manufacturer's customer portfolio includes all customers of all the sales channels and, hence, the proportional attention paid to each client will be smaller.Moreover, the manufacturer has a limited ability to fully serve all the clients.As a result, the manufacturer should focus on the most important or strategic clients and answer them directly, leaving the remaining clients to be served by intermediaries.In contrast, independent representatives (i.e., intermediaries) can serve customers more homogeneously.However, even noting these differences, what the client supposedly wants is to be given priority of service by the manufacturer (Shipley et al., 1991;Katsikeas et al., 2004).Therefore, given an intention to purchase a product, the customer will prefer to be served by the manufacturer rather than third-party sales representatives.Thus, in evaluating the importance of the client when faced with a multi-channel structure, the following hypothesis is formulated: Hypothesis 3: The greater the importance to the supplier perceived by the client, the higher the probability of being served by direct sales channels.

Attributes of the dyad
Another critical factor in the relationship between channels is the intimacy established between supplier and clients.Companies that are close to their clients succeed when they become experts in their clients' business and in the creation of solutions.These companies differentiate themselves through deep customer knowledge.As a result, these clients establish, on a continuous basis, a certain dependence.Murillo and Annabi (2002) claim that intimacy focuses on capitalizing on clients' needs to improve their satisfaction and thus increases purchasing behavior.Treacy and Wiersema (1993) show that intimacy with the client is achieved by the accurate segmentation of the market and the offering of appropriate products.Therefore, combining customer information and process flexibility becomes important for these suppliers.Wiersema (1996) further develops the question of intimacy by demonstrating that the successful implementation of the relationship strategy depends on a culture that supports not only joint effort but also emphasis on learning.According to the author, this intimacy strategy requires characteristics such as sensitivity to new opportunities, company willingness to serve the many needs of clients, joint effort aiming at cooperation among employees and among functional subsystems, learning the needs of the customers and the structuring of transactions to improve results.Therefore, developing intimacy is an option for creating competitive advantage by the supplier (Román & Martín, 2008).
The intermediary channel tends to have a smaller client portfolio than the manufacturer and, therefore, the ability of the former to serve and thus create conditions to establish a closer contact with clients is greater than that of the latter.However, clients tend to notice that intimacy with the supplier increases the closer they are to the main source of the object of exchange, i.e., the manufacturer (Merrilees & Fenech, 2007).Thus, despite the advantage that third-party representatives have in establishing this relationship characteristic, in the perception of corporate clients, intimacy will exist when they are served by the manufacturer.In summary, the greater the intimacy that clients have with the supplier, the closer they will be to the manufacturer.Therefore, we present the following hypothesis: Hypothesis 4: The greater the intimacy between client and supplier, the higher the probability of being served by direct sales channels.
Last but not least, the relationship management of marketing channels may be affected by the bargaining power of the supplier.In other words, if the relationship between client and supplier is ruled by the inability of the client to switch suppliers, then negative effects on the relationship can be experienced (Dutta et al., 1995).The concept of switching cost represents the difficulty imposed by a possible replacement of a supplier when the party responsible for the customer relationship cannot be easily replaced (Hamel & Prahalad, 1994); i.e., the client gets stuck with a sales representative, and the main problems are sunk investments (sunk costs).
The creation of switching costs can be a strategic option for the supplier.Switching costs create barriers for the entry of new competitors, and therefore, clients are unable to pit suppliers against supplier (Porter, 1986).Sengupta, Krapfel and Pusateri (1997) showed that switching costs causes certain inertia that induces customers to remain in relationships with their current suppliers.Following the same line of argument, Dutta et al. (1995) supports the notion that customers who invest significantly in the relationship with their supplier's representatives have a potential onus related to the lack of efficiency by the entry of a substitute/additional vendor.The authors show that these client companies attempt to request more guarantees from the solution providers.
Therefore, in the context of the switching costs of the supplier, clients will prefer to buy from the same vendor after they have invested in the relationship with this supplier to purchase products.Changing manufacturers has a high cost to clients because, in this case, they would have to adapt the portfolio of another manufacturer to their needs.When switching to another sales intermediary but retaining the same manufacturer, the associated costs are only connected to the sales process itself.Thus, it is expected that clients prefer a direct relationship with the manufacturer rather than using a third-party intermediary.So, it is expected that high switching costs are positively related to the preference of selecting a direct channel with the manufacturer rather than a third-party channel, as indicated in the following hypothesis: Hypothesis 5: The higher the switching costs incurred by a customer, the higher the probability of being served by direct sales channels.

Data Collection and Sample
The IT sector was chosen as the context for the research because corporate clients of this sector have access to both direct and indirect channels (Roccato, 2008).
To test the hypotheses formulated in the previous section, a survey of the perceptions of corporate clients of information technology (IT) equipment, i.e., a database originated from information on industry associations, generating a list of approximately 3,000 companies that purchased IT products or services in 2008.The respondent of each company was selected based on whether they had permission to recommend or approve purchasing decisions.
Before administering the questionnaire, preliminary interviews were conducted with key people related to IT sales channels to validate whether the questions used in the questionnaire were adequate to capture the phenomenon of interest.The questionnaire was then sent electronically to each respondent, and follow-ups were conducted to stimulate a response.The return rate of the requests to answer the questionnaire was 32%, resulting in an initial sample of 1,088 companies.Given the presence of missing values (i.e., no response) and invalid answers, the sample was reduced to 678 observations.However, the regression was based on 505 companies that had complete information for all variables of interest.
Source: Elaborated by the authors.

Figure 1 -Distribution of the number of employees in the sampled companies by type of channel
Figure 1 shows the distribution of companies in the sample by size, as represented by the number of employees.Approximately half of the sample belongs to the micro and small companies group, 2/3 to the medium-sized companies group and 1/3 to the large companies group.Nevertheless, large companies have a greater propensity to obtain direct service rather than service via an intermediary, whereas micro, small and medium-sized companies are more frequently served via an intermediary rather than directly serviced.
Table 1 shows the distribution of sectors for the sampled companies.The main representatives are companies engaged in manufacturing, digital industry, service companies and the financial industry (79% of participants).Important points were also observed regarding the type of service.First, the financial market is served directly by the manufacturer twice as frequently than by independent representatives.This can be explained by the fact that these companies are generally large or strategic, and it is important for manufacturers to directly meet this demand rather than including an external mediator in the relationship.In the sector of service providers, the reverse occurs.The main reason may be the heterogeneity of service companies, which are medium-sized companies or companies located outside large capitals, and therefore they justify being served by external channels.Source: Elaborated by the authors.
Figure 2 illustrates the distribution of companies by volume of expenses for IT.The category of lower volume of investments is most commonly served by representatives.In contrast, the major investor categories are more often served by the manufacturer, which can be justified by the preference of manufacturers to invest directly in major sales accounts.
Note: Legend expressed in thousands of US dollars.Source: Elaborated by the authors.

Variables
The dependent variable was generated using an approach similar to that adopted in the work of Dutta et al. (1995), and therefore, its construction was derived from the following question in the questionnaire: "What is the primary marketing channel used by your company?"To this question, two possible answers were offered: a) Manufacturer's direct sales force or b) Third-party sales channel.Therefore, the dependent variable, 'Type of Channel' is discrete and dichotomous (dummy), indicating whether the client's primary service model is the direct sales force of the manufacturer or an independent sales representative (i.e., intermediary).It is worth noting that the customer has the option of being served by both channels, direct and indirect, and that the response identifies the primary sales channel.
The relationship attributes (i.e., independent variables) identified in the literature review were analyzed based on an existing questionnaire whose items are considered valid and reliable by others (Sengupta et al., 1997, Roy et al., 2004;Rosenbloom, 2007).These items were measured on a Likert scale.To confirm the quality of the scale used, reliability tests and analysis of the principal components were carried out and will be explained below.
Loyalty -This attribute was represented as a behavior that can be measured by the amount of repurchasing, i.e., the intention that the client has to purchase products or services of a particular category from the same supplier.In the present study, loyalty was treated as a construct resulting from four items, and all four were measured using a five-point Likert scale characterized by the following options: strongly disagree, partially disagree, neither agree nor disagree, partially agree, totally agree.The Cronbach's alpha coefficient was 0.77, indicating high internal consistency of the items related to the loyalty construct.
Trust -This attribute was analyzed by a proxy that identified the perception that the customer had in relation to their trust in the supplier.The questionnaire used the following question to generate the proxy trust: "Is your supplier reliable?"The measurement was structured using a five-point Likert scale (the scale was similar to that used for the loyalty attribute).
Importance -This attribute was characterized by how clients interpret the importance given to them by the suppliers and was measured as a construct of three items (all measured using a five-point Likert scale).The Cronbach's alpha coefficient was 0.69, confirming the high internal consistency of these items.
Intimacy -This attribute was measured using a construct of six items, which aimed to evaluate the closeness of the relationship between client and supplier.All items were measured using a ten-point Likert scale, where "1", referred to the inexistence of a relationship between the parties and the maximum value, and "10" represented an excellent or solid relationship.The Cronbach's alpha coefficient was 0.92, indicating high internal consistency of the items measuring the intimacy construct.
Switching cost -This attribute was measured by three items that sought to capture the significance of additional costs incurred in switching the customer's current supplier to any other.The same criteria were used for all items using a five-point Likert scale, which contained similar options to those shown for loyalty, trust and importance.The Cronbach's alpha coefficient was 0.65.To ensure the reliability of the constructs, the results of a principal component analysis were used (Table 2).The principal component analysis was used to combine the set of items that measure each of the constructs into a single variable, i.e., an index that is a linear combination of the items, the weights of which are shown in Table 2. Thus, standard indices are used with the regression model to represent each of the constructs of interest (loyalty, importance, intimacy and switching cost) rather than using each questionnaire item separately, which would make the testing of the hypotheses proposed in the present study unfeasible.These indices were obtained according to the data dependency structure and in such a way that they would retain as much information as possible, i.e., in the way that would explain most of the total variability (Reis, 2001).
It is important to note that the weights of the items that measure the same construct are similar in the index composition, which causes the items to have approximately the same importance in the measurement of constructs.In addition, the percentage of variability explained by each of the indices (Table 2) is considered suitable because it represents at least 60% of the total variability of the items that measure the construct (Anderson, Tatham & Black, 2005).

Statistical model
To effectively observe the effects of the determinants on the choice of the supplier channel by the client, a statistical analysis based on the Logistic regression model, or Logit, was performed.This modeling technique aims to discriminate between two groups of observations within a sample to obtain the probability that an observation belongs to a certain set, depending on the behavior of the independent variables (Hair Jr., Anderson, Tatham & Black, 1998).In the present study, the estimators will indicate by how much the probability of a corporate client being served by an independent sales representative channel (i.e., intermediary) varies according to the variation of the explanatory variables.

RESULTS
Table 3 shows the descriptive analysis of the items used to measure the constructs of interest for the total sample by service channel.Based on Table 3, the averages of all items used in the questionnaire are higher for direct service by the manufacturer than for service by an independent channel.Thus the results are generally consistent with what would be expected given the preference for direct service, especially because all of the hypotheses imply a higher likelihood for the company to choose to be served by direct sales channels.
To test these hypotheses, we must consider the statistical significance of the results, assuming a certain level of significance.Thus, this study applies regression analysis to capture the effects of the factors more robustly.It is worth noting that to obtain a better fit of the model, the standardized indices constructed earlier as determinants of the choice of the channel will be used, each index representing a construct of interest: loyalty, trust, importance, intimacy and switching costs.This article estimated a binomial logit regression to identify factors that affect the client's service preference between direct service by the manufacturer and service by an independent sales representative (dichotomous decision).Table 4 shows the results of the binomial logit regression.The model consists of five explanatory variables that represent the five factors that determine the types of service discussed in this study (loyalty, trust, importance of the customer, intimacy and supplier switching costs) as well as eight control variables (after-sales, product type, number of employees, client IT expense, respondent´s job, product modularity, use of the product and client's industry sector).
Table 4 shows that the importance of the client has, even if marginal, an impact on the service model used.The results indicate that the greater the importance given to the client by the supplier, the higher the probability of the client being served directly by the manufacturer, given that the estimated coefficient is negative.On average, a onepoint increase in the importance factor reduces the odds of the client being served by an intermediary representative by 19%, i.e., the results are supportive of the corporate customer's preference for direct service.This reinforces the idea that the client believes him or herself to be the most important among all clients.Thus, clients' real interest is to be served according to their expectations, regardless of how manufacturers control their list of priorities.In this case, all other variables remaining constant, the client will prefer to be on the manufacturer's list of direct clients.Therefore, there is no influence of third parties on the relationship.
The determinant factor intimacy presents a more (statistically) significant result.This means that the intimacy level between buyer and supplier affects the choice of the marketing channel, and the sign of the coefficient is consistent with the expectations, i.e., the higher the intimacy between buyer and supplier, the more the corporate client will prefer to negotiate directly with the manufacturer.Thus, the results are consistent with the hypothesis 4, indicating a preference for direct sales by the corporate client.In summary, it is confirmed that a one-point increase in the intimacy factor reduces the chances of the client being served by a representative intermediary by, on average, 25%.Therefore, these results indicate that the ability of intermediaries i) to develop strong relationships with clients, ii) to help clients select the best solution according to their needs, iii) to keep clients informed of new products and technologies, iv) to seek alternatives that add value to clients' businesses and v) to know the customers' industries are much more limited than the abilities of manufacturers in these respects.
It is possible that the manufacturers have more knowledge to suit the needs of clients than do intermediaries.In addition, the results suggest that corporate clients expect the ability to closely relate with the manufacturer.Although clients occasionally recognize the limitations that manufacturers have in providing services throughout the Brazilian territory, they disregard this limitation when asked about their preferences for the service model.Furthermore, it is possible that manufacturers focus most of their relationship efforts on clients with whom they have direct contact, leaving intermediaries to serve the remaining clients.This should reduce the client portfolio with which the manufacturers have direct contact and therefore allow them to develop more intimacy with clients than do the intermediaries.
Regarding loyalty, the negative coefficient for this variable is aligned with preference expectations by the client.However, its statistical insignificance means that the effect of loyalty on the choice of the channel adopted is marginal.Although loyal customers may prefer, under equal terms, the same supplier when compared to the competition (Oliver, 1999), this appears irrelevant to the choice of the channel.When providing direct services, manufacturers may be better positioned to give concessions for their products.However, intermediary representatives, although not capable of offering the same concessions, may offer the same products that their loyal clients prefer.Therefore, even if it is not possible to reject hypothesis 1, the results suggest that loyal clients remain loyal regardless of the channel used to serve them.
Although the coefficients of the factors trust and switching cost show positive signs, which would be contrary to the formulated hypotheses two and five, respectively, they are statistically insignificant, i.e., they do not impact the choice of the type of channel.Regarding the customer's trust in the supplier, the positive coefficient has proven not to be aligned with the expectations for direct sales.Therefore, although the client sees the manufacturer as the greatest source of knowledge about the products and understands that direct services provided by the manufacturer may be better in solving potential problems, when clients develop trust in their representative intermediary, the corporate client will have no interest in being served directly by the manufacturer.
Results for the determinant switching cost do not reject hypothesis 5, which indicates a higher probability of the client being served by the manufacturer.The expectation was that an increase in switching costs would reduce the odds of the client wanting to be served by a representative intermediary.However, the results obtained were the opposite of those expected.It is possible that in situations where the switching costs of the supplier are high, clients begin to question their relationship with the supplier.Even if clients have unanticipated additional costs in the short term, they are more inclined to seek less costly alternatives in the long term to extricate themselves from the situation of entrapment with the current supplier.However, given the statistical insignificance of the coefficient, this study concludes that the effects of switching cost do not interfere in the choice of the channel.
In general, the theory suggests that the creation of competitive advantage by companies depends in part on the ability of these companies to adequately manage the channels through which customers are served (Taylor, Kim, Ko, Kim & Park, 2008).Therefore, companies should not treat the transaction as an isolated phenomenon, but rather as a relational phenomenon in which supplier and buyer are engaged in a social exchange that involves personal satisfaction, not only economic incentives (Dwyer et al., 1987).Thus, it is expected that several attributes of the relationship are responsible for the success of a marketing channel (Bendapudi & Berry, 1997;Bolton & Lemon, 1999;Bunduchi, 2008;Rosenbloom, 2002).However, heretofore the development/appearance of many attributes has been considered independent of the channel; i.e., the development/appearance of a given attribute would not be potentiated/mitigated by the characteristics of the channel.
The results of this study show that the particularities of a given attribute may be more favorable for the service provided by a given channel.In particular, it was observed that a greater sense of importance by the corporate client leads to a higher probability of being served directly by the product manufacturer.In the same way, and with stronger statistical significance, corporate clients who perceive a greater intimacy with the supplier are more likely to be served directly.These results may be reflections of the manufacturers' own channel strategy of directly serving clients with the highest purchasing potential, leaving the other clients to be served by intermediaries (Palmatier, 2008;Shipley et al., 1991).Indeed, the control variables reinforce this aspect because the most important clients (both in terms of the number of employees and in purchase volume) are more likely to be served directly.Similarly, clients who experience greater customization of products are those most likely to be served directly by the Determinants of the choice of marketing channels by corporate clients: an analysis of the information technology sector 535 JISTEM, Brazil Vol. 9, No. 3, Sept/Dec.2012, pp.515-542 www.jistem.fea.usp.brmanufacturer, which may reinforce the perception of intimacy that the customer develops.

CONCLUSIONS
The use of intermediaries to market and distribute products generates benefits concerning the division of tasks and economies of scale.Conversely, adoption of intermediaries generates costs associated with the intermediary (Payne & Frow, 2004).As marketing channels are increasingly characterized by the simultaneous presence of intermediaries and the direct sales force of the manufacturer, and meeting the clients' expectations becomes crucial to the survival and sustainable differentiation of companies, it is important to understand which factors induce corporate clients to choose a particular channel.However, we wanted to know which factors are the most relevant to the choice of marketing channel by the corporate client.To answer this question, the present study conducted a survey of 505 corporate clients of the Brazilian IT market.Based on the logit statistical model, this research investigated whether the factors loyalty, trust, importance of the customer, intimacy between client and supplier, and switching costs of suppliers may affect the choice of the marketing channel by the client.
The results revealed that the customer-supplier intimacy level is the main determinant of the choice of corporate clients to be served by a direct sales channel.In turn, clients' perception of their own importance to the supplier is hardly relevant to their preference for being served directly by the manufacturer; the remaining factors were not identified (statistically) as determinants of the choice of marketing channels.
The present study contributes to our theoretical understanding by showing that the development of relationship attributes may influence the choice of the channel, but a channel may not always be the first option for the client's choice.Particularly, the survey results indicate that the real relational advantage of the direct sales channel over the sales representative channel lies in its ability to make the buyer-manufacturer relationship closer (i.e., more intimate), thereby contributing to the development of solutions that better serve the needs of corporate clients.In contrast, the contribution of this research to practice is to draw the suppliers' attention to the limits of their efforts to develop certain aspects of their relationship with clients (corporate).Although the value proposal offered to clients depends on the relationship that is established in the channels, the characteristics of these channels must be understood before defining what efforts will be used to improve the relationship.
As usual the present study is not free from limitations.Just as the constructs (although valid) may not perfectly capture the phenomena they represent, the analysis was based on the assumption that the corporate client chooses only one marketing channel, even if further alternatives exist.Thus, future studies should seek to explore not only other determinants (and their antecedents) for the choice of corporate clients but also the creation of more options to allow further investigation of the impact of those determinants on the simultaneous use of marketing channels by the corporate client.

Figure 2 -
Distribution of the IT expenses of the sampled companies by type of channel e reference category 'inexistent'; f reference category 'minimum'.Source: Elaborated by the authors.

Table 1 -
Distribution of the sector of the sampled companies by type of channel

Table 2 -
Principal component analysis Source: Elaborated by the authors.

Table 3 -
Descriptive analysis of the items used to measure the constructs

Table 4 -
Determinants of the choice of the marketing channel (sales representative channel)