How to Make Switching Costly: The Role of Marketing - iBrarian.netкод для вставки
1 HOW TO MAKE SWITCHING COSTLY: AN INVESTIGATION OF THE DRIVERS OF CONSUMER SWITCHING COSTS Yolanda Polo and F. Javier SesГ© 1 Abstract Customer switching costs have emerged as one of the fundamental drivers of customer retention. Although the consequences of these costs have been well documented in the literature, research on the determinants of switching costs remains limited. The present study seeks to address this issue by investigating the extent to which switching costs are influenced by marketing variables вЂ“price and advertisingвЂ“ and relationship characteristics. The authors develop a conceptual framework about the drivers of switching costs and test the framework empirically in the mobile phone industry using a hierarchical Bayes approach. The empirical results show that by using price and advertising вЂ“both service and brand advertisingвЂ“ firms are able to make switching costly for customers. Moreover, relationship characteristics significantly contribute to explaining consumersвЂ™ differences in the cost of switching. Finally, this study illustrates the key role played by competitorsвЂ™ marketing actions in affecting the cost of switching for customers of the focal firm. Implications for decision makers are discussed. Keywords: Switching costs, Customer retention, Price and advertising, Relationship characteristics, Hierarchical linear model. 1 Yolanda Polo is Professor of Marketing, Faculty of Economics, Marketing Department, University of Zaragoza, Spain, E-mail: firstname.lastname@example.org. F. Javier SesГ© is an Assistant Professor, Marketing Department, Faculty of Economics, University of Zaragoza, Gran VГa, 2, 50005, Zaragoza, Spain, Phone: +34 976761000, Fax: +34 976761767, E-mail: email@example.com. The authors gratefully acknowledge the financial support by the projects: SEC2005-05968 (MEyC, FEDER) and S09-PM062, from the Spanish Regional Government of AragГіn. 2 INTRODUCTION Switching costs have received increasing attention in the academic marketing literature. From a theoretical point of view, switching costs are recognized as an important driver of customer retention that leads to stable and long-lasting relationships (Dick and Basu, 1994; Ganesan, 1994; Bendapudi and Berry 1997). From an empirical point of view, researchers have shown the central role of switching costs in explaining customer retention in many service industries including banking, airlines, credit card and telecommunications (Burnham, Frels and Mahajan 2003; Harris and Uncles 2007; Jones, Mothersbaugh and Beatty 2000, 2002; Lam et al. 2004). In support of the theoretical arguments, most empirical studies conclude that switching costs significantly affect customer decisions to maintain a relationship either directly or indirectly through their moderating role on the associations between customer relationship perceptions (like satisfaction, service quality, trust or commitment) and customer loyalty (Bell, Auh and Smalley 2005; Patterson and Smith 2003; Sharma and Patterson 2000). Based on both theoretical and empirical work, there is a general consensus that switching costs are an important determinant of customer retention. While marketing literature contains extensive research devoted to the consequences of switching costs in terms of customer retention and customer switching behavior, little is known about the drivers of these costs. However, knowledge about the factors that generate and increase the level of switching costs can be of great interest to both academics and practitioners. From the point of view of academics, understanding customer retention and its antecedents is central to relationship marketing theories (Bolton, Lemon and Verhoef 2004; Sheth and Parvatiyar 1995; Verhoef 2003). So far, marketing scholars have primarily focused on satisfaction and/or commitment as the main motivations for customers to engage in longterm relationships (Garbarino and Johnson 1999; Mittal and Kamakura 2001; Morgan and Hunt 1994; Szymanski and Henard 2001). Customer switching costs are also a key driver of 3 customer retention, but have received much less attention in the literature. From a practical standpoint, executives are concerned about the negative consequences of customer switching in terms of market share and profitability (Rust and Zahoric 1993). To keep their customers in the relationship, firm strategies have mainly focused on satisfaction as the primary retention tool (Anderson and Sullivan 1993; Fornell 1992). Although customer satisfaction has been positively related to customer retention, satisfied or even very satisfied customers also defect (Jones and Sasser 1995). In addition, recent research has shown that the impact of switching costs on customer repurchase behavior is higher than that of satisfaction (Burnham, Frels and Mahajan 2003). Thus, identifying the strategies and factors that make switching costly is an important issue for marketing strategy. Intuition and anecdotal evidence suggest that relationship history and marketing variables influence the cost of switching providers. This is consistent with VerhoefвЂ™s (2003) hypothesis that customer retention is a function of both consumersвЂ™ past behavior and marketing practices. It is also consistent with recent empirical research on customer switching behavior (Wathne, Biong and Heide 2001) which shows that marketing programs and relationship variables are key elements in the choice of supplier. However, although these studies can serve as guidelines about the factors that may influence the cost of switching providers, an important question remains unanswered: How do marketing variables and relationship characteristics affect switching costs? Although some anecdotal evidence has been provided to support the thesis that firms can control, to a certain extent, the magnitude of customer switching costs (Shapiro and Varian 1999), there is no general consensus about whether and to what extent firm marketing practices affect these costs. Porter (1980) states that the impact of a firmвЂ™s strategy on switching costs should be evaluated. Chen and Forman (2006) emphasize the central role of firmsвЂ™ strategies in locking their customers into the relationship. However, most researchers 4 have primarily focused on customer relationship perceptions as the key drivers of customer retention, relegating marketing strategies to a secondary position (Gustafsson, Johnson and Roos 2005; Verhoef, Franses and Hoekstra 2002). CompetitorsвЂ™ marketing strategies are also a means of inducing customer switching (Keaveney 1995; Wathne, Biong and Heide 2001). For instance, offering better economic conditions will make customers more willing to change service providers. Although the important effect of competitorsвЂ™ marketing strategy on customer retention has been widely acknowledged, empirical research is limited mainly due to the lack of available information. Thus, more research is required to clarify the role of marketing вЂ“by both the focal firm and the competitorsвЂ“ in explaining customer switching costs. Customer past behavior is also a potential source of switching costs. Marketing scholars indicate that the interactions between buyers and sellers generate close interpersonal relationships and bonds that may act as a barrier to prevent customer switching (Lam et al. 2004). For instance, as the relationship develops, customers gain experience and familiarity with the company and its employees, which enable them to make quicker and more effective use of the services purchased. However, the literature does not offer much insight into whether past behavior is a significant predictor of switching costs. A notable exception is Burnham, Frels and MahajanвЂ™s (2003) article, which incorporates some customer relationship variables such as the breadth of the relationship. Given that both marketing instruments and customer relationship characteristics are two potential sources of switching costs, understanding their influence on these costs would be an interesting issue for marketing theory and practice. We intend to make two main contributions to the literature. First, we provide a conceptual framework about the drivers of switching costs that simultaneously considers marketing variables and relationship characteristics. In addition, we empirically investigate 5 the extent to which switching costs are influenced by these explanatory factors in order to assess their relative importance. Second, by virtue of having data about all the firms in the market, we can evaluate the impact of competitorsвЂ™ marketing strategy on customer switching costs in order to understand the role played by competition in explaining customer behavior. The remainder of this manuscript is organized as follows. In the next section, we provide the conceptual model and offer the research hypotheses. Subsequently, we discuss the research methodology and describe the data. Then we present the results from the empirical application and we conclude with a discussion of the main findings and contributions of the article and an assessment of managerial implications, limitations and directions for future research. CONCEPTUAL MODEL AND HYPOTHESES In this study, we investigate the cost that customers bear when switching service suppliers. We define the cost of switching as the disutility a customer experiences from changing product or service providers (Chen and Hitt, 2002). 2 Switching involves a process that can be difficult and costly. Customers wanting to switch service suppliers have to give up some relationship-specific investments (e.g. physical assets and organizational learning) (Klemperer, 1995). There are many barriers that consumers must overcome if they are to switch suppliers, including contractual, monetary, setup and psychological costs (Burnham, Frels and Mahajan 2003; Jones, et al. 2007). Although the importance of each of these costs varies with factors such as the industry or the customer-specific characteristics (Patterson and Smith 2003), together they represent a clear disincentive to switch suppliers. 2 Although we mainly use the term вЂњserviceвЂќ during the article, the discussion holds for both products and services. 6 In this section, we provide a conceptual framework that describes and analyzes how and why marketing variables (from the focal firm and from competitors), as well as customer relationship characteristics, affect the cost of switching providers. This is a customer-specific analysis because it studies the cost of switching at the individual level and it is a dynamic model because longitudinal information on both marketing instruments and customer behavior is used. Thus, the main goal of this model is to ascertain the drivers of customer switching costs. The conceptual framework is displayed in Figure 1. FIGURE 1 Conceptual Model. Drivers of the Customer Switching Cost FOCAL FIRMвЂ™S MARKETING VARIABLES RELATIONSHIP CHARACTERISTICS COMPETITORSвЂ™ MARKETING VARIABLES Price Service Advertising Brand Advertising Length Depth Breadth Price Service Advertising Brand Advertising COST OF SWITCHING PROVIDERS CONTROL VARIABLES Type of Setting Gender Age In this model, we focus on two main groups of variables which potentially explain the disutility that customers experience from switching: (1) marketing instruments and (2) relationship characteristics. Marketing literature provides considerable insight into the key role played by marketing instruments in explaining customer behavior (Bolton, Kannan and 7 Bramlett 2000; De Wulf, Odekerken-SchrГ¶der and Iacobucci 2001; Liu 2007; Rust, Zeithaml and Lemon 2000; Verhoef 2003; Wirtz, Mattila and Lwin 2007). However, little is known about the process through which these instruments influence the cost of switching. We focus on two general-oriented marketing instruments: price and advertising. The latter is further divided into communication focused on the service вЂ“service advertisingвЂ“ and that focused on the brand вЂ“brand advertising. Price is one of the most important factors affecting customer choice behavior (Keaveney 1995; Roos 1999) and the economics literature has highlighted its important role in the management of customer switching costs (Farrell and Klemperer 2007). Advertising is another marketing tool that has a potential impact on customer switching costs and has been the subject of research in prior marketing studies dealing with switching costs (Shum 2004). In addition to examining the effect of the focal firmвЂ™s marketing practices, we assess the role of the competitorsвЂ™ marketing instruments in explaining customer switching costs. CompetitorsвЂ™ price and advertising (service and brand advertising) will therefore be considered. In the second group of variables, relationship characteristics describe the development of a relationship over time. Three factors are included: (i) the duration of the relationship вЂ“ lengthвЂ“, (ii) service usage вЂ“depthвЂ“ and (iii) cross-buying behavior вЂ“breadthвЂ“, which reflect the progression of the relationship and describe the customer purchase behavior (Bolton, Lemon and Verhoef 2004). This framework incorporates an additional group of variables in order to control for some observed factors such as the type of setting (contractual vs. noncontractual) and demographics (gender, age). Marketing Instruments Organizations invest in a wide range of marketing activities aimed at influencing and stimulating customer behavior. In the marketing literature, researchers have stressed the 8 important role of some marketing instruments in affecting customer behavior and loyalty (Bolton, Kannan and Bramlett 2000; Bolton, Lemon and Verhoef 2004). Prior research has also highlighted that, by using marketing strategies, firms are able to influence the magnitude of switching costs (Chen and Forman 2006). In this section we discuss the associations between two general-oriented marketing instruments, price and advertising, and the cost of switching providers. Price. To stimulate customer behavior, firms can act on this marketing instrument. The impact of price on customer behavior has been well documented in economics (Einhorn 1994; Smith and Brynjolfsson 2001). An important characteristic of these studies is that they usually highlight the central role of price in the acquisition of new consumers. From the point of view of marketing, researchers have recognized the importance of price in affecting the behavior of existing customers (Bolton and Lemon 1999; Rao and Monroe 1989; Rust, Zeithaml and Lemon 2000). Price has also been considered a key factor in the management of customer switching costs (Klemperer 1995) and there is ample evidence in the marketing literature that shows the key role played by price in customersвЂ™ decisions to switch service providers (Keaveney 1995; Roos 1999; Roos, Edvardsson and Gustafsson 2004). Below, we discuss the expected effect that the focal supplierвЂ™s and the competitorsвЂ™ price have on the cost of switching. The focal firm can use price to strengthen their relationships with current customers by offering them better tariffs. From an economic perspective, researchers have shown that the lower the price, the higher the customerвЂ™s probability of purchasing the product or service from the firm (Einhorn 1994; Smith and Brynjolfsson 2001). From a marketing point of view, scholars have long recognized that a low price leads to better customer price perceptions and expectations and to higher levels of satisfaction, which, in turn, leads to longer customer-firm relationships (Bolton and Lemon 1999; Kalyanaram and Winer 1995; Keaveney 1995). These 9 arguments suggest a negative effect of price on customer switching costs. In addition, we can argue that the magnitude of switching costs is also going to be affected by customer expectations about future economic gains. Krishna et al. (2002) indicate that comparing the price charged by the current firm with that charged by competitors influences perceived savings. Thus, we expect that the higher the focal firmвЂ™s price, the higher the potential monetary savings from switching service providers, which will lead to lower switching costs. Competitors will also use price to stimulate customer behavior. They are interested in attracting customers from the focal firm in order to increase their market share and profitability. One way to do this is by affecting the cost of switching providers for customers of competing companies. Competitors can use price to subsidize switching costs for these customers (Chen 1997). By reducing this marketing instrument, the disutility that switching generates will be partially compensated by the economic gains. Using the same argument as in the previous paragraph about customer expectations, the lower the competitorsвЂ™ price, the higher the potential monetary savings of changing providers (leading to a lower cost of switching). Based on the above discussion, we hypothesize the following: H1a: Focal providerвЂ™s price negatively influences the cost of switching. H1b: CompetitorsвЂ™ price positively influences the cost of switching. Advertising. Marketing researchers have recognized that advertising plays a key role in the acquisition of new customers (Bolton, Lemon and Verhoef 2004) and in affecting the behavior of existing ones (Manchanda et al. 2006; Prins and Verhoef 2007). Advertising can be used by firms with two main objectives (Prins and Verhoef 2007): (1) to inform existing and prospective customers about service attributes and characteristics (service advertising) and/or (2) to create favorable brand associations and brand preference (brand advertising). It 10 is important to consider the distinction between these two types of advertising as they are expected to have different effects on customer switching costs. Service advertising focuses mainly on informing existing and potential customers about the advantages and benefits of the service (Prins and Verhoef 2007). On the one hand, it is reasonable to think of focal supplierвЂ™s investments in service advertising as increasing customer switching inconvenience because positive aspects of the service are highlighted (Bolton, Lemon and Verhoef 2004). Customers will be aware of the benefits and advantages that the service provides and the relationship between the customer and the firm will be strengthened. On the other hand, focal supplierвЂ™s service advertising might have the opposite effect. This is because it provides customers with valuable information which can be used to better assess the different alternatives. The cost involved in searching for the information and in evaluating and comparing the alternatives will, therefore, be substantially reduced. This might make switching less costly. In support of this argument, Shum (2004) finds that advertising has a negative impact on the purchase probabilities of loyal customers. Therefore, there is no clear association between focal firmвЂ™s service advertising and the customer switching cost. CompetitorsвЂ™ service advertising is expected to reduce switching costs. Competitors will use service advertising to create awareness and knowledge among the customers of the focal firm of their services and to show their advantages and benefits. Exposure to this kind of advertising will make customers of the focal firm more informed about the services supplied by the competitors, reducing the cost involved in acquiring, evaluating and comparing the information and, at the same time, reducing the risk involved in switching suppliers. Moreover, highlighting the advantages and strengths of competitorsвЂ™ services might induce customers of the focal firm to switch, as they will be aware of the potential benefits of 11 switching providers. In this vein, Shum (2004) reports a significant reduction in the switching barriers as a consequence of competitorsвЂ™ advertising. Following the previous reasoning, we do not posit any directional hypothesis on the focal firmвЂ™s service advertising. We hypothesize that: H2a: The focal providerвЂ™s service advertising is related to the cost of switching. H2b: CompetitorsвЂ™ service advertising negatively influences the cost of switching. Brand advertising aims to change brand feelings positively so as to influence purchase behavior (Cramphorn 2006). The focal firmвЂ™s brand advertising will, therefore, focus on increasing brand awareness, improving brand attitudes and creating a positive brand image. This will reinforce the relationship between the customer and the firm, making the switching process more costly (Burnham, Frels and Mahajan 2003). Studies in the marketing literature have usually ignored the impact of brand advertising on retention or on the cost associated with switching. Notable exceptions are studies by Rust, Zeithaml and Lemon (2000), who acknowledge the effect of branding on customer equity, and by Bolton, Lemon and Verhoef (2004), who contend that brand advertising has a positive impact on relationship length. Taken together, these arguments suggest that brand advertising by the focal firm will make switching more costly. Competitors will also brand advertise in order to stimulate customer behavior. Exposure to this form of advertising can lead to a greater awareness and knowledge of the competing brands among existing customers of the focal firm. In addition, researchers have recognized that brand name and corporate image are important quality signals, mainly for intangible services (Davis, Buchanan-Oliver and Brodie 2000). All these factors will ultimately diminish the uncertainty associated with switching providers, leading to a global reduction of the cost 12 of switching. So far, empirical evidence on these effects is almost absent in the marketing literature. However, based on the above theoretical discussion, we hypothesize: H3a: The focal providerвЂ™s brand advertising positively influences the cost of switching. H3b: CompetitorsвЂ™ brand advertising negatively influences the cost of switching. Relationship Characteristics Relationship characteristics describe the development of a relationship over time. In this section we discuss the associations between the length, depth and breadth of the relationship and switching costs. Prior theoretical and empirical research has highlighted the central role of these variables in explaining various relational and behavioral constructs such as satisfaction (Bolton 1998), number of services purchased and customer referrals (Verhoef, Franses and Hoekstra 2002), profitable lifetime duration (Reinartz and Kumar 2003) and customer switching costs (Burnham, Frels and Mahajan 2003). However, to our knowledge this is the first study that integrates these three relationship characteristics into a single framework about the drivers switching costs. Length refers to the duration of a relationship. The early stages of a relationship are characterized by high uncertainty and low customer experience (Verhoef, Franses and Hoekstra 2002). In this phase of the relationship, specific investments are low and the interdependencies between the parties are not greatly developed (Dwyer, Schurr and Oh 1987; Geyskens, Steenkamp and Kumar 1999). As the relationship progresses over time: (1) trust develops between the parties (Gwinner, Gremler and Bitner 1998); (2) customers gain experience in using the products and services (Verhoef, Franses and Hoekstra 2001); (3) 13 familiarity and interdependencies between the parties grow (Dwyer, Schurr and Oh 1987); and (4) psychological attachment increases (Burnham, Frels and Mahajan 2003). Moreover, long-term customers are usually granted some benefits to enhance their loyalty and to lock them into the relationship. Some are economic benefits in the form of loyalty programs, which give customers financial rewards like discount coupons, points redeemable for prizes or other economic advantages (Sharp and Sharp 1997). Continuing customers also receive social benefits in the form of communication, cooperation, friendship and rapport, which seek to increase social bonds and interdependencies between the parties (Chiu et al. 2005). All these benefits are lost if the customer leaves the company and are not readily available elsewhere. Consistent with the previous reasoning, Verhoef, Franses and Hoekstra (2002) provide some arguments supporting the idea that switching costs may increase with relationship age. In addition, Shapiro and Varian (1999) offer some anecdotal evidence suggesting that customers become locked into the relationship as a consequence of the increasing switching costs. Hence, we hypothesize that: H4: The length of the relationship positively influences the cost of switching. The depth of a relationship is the frequency of service usage over time. It might be assumed that the depth of the relationship positively affects the cost of switching providers for the following reasons. First, using the firmвЂ™s services provides customers with a better knowledge of the company and the processes needed to use these services satisfactorily (Alba and Hutchinson 1987). Second, those who use the service more can develop non-transferable provider-specific skills from the time and effort invested in the relationship (Keaveney and Parthasarathy 2001). Third, as customers use the service more, they become familiar with the company and the interdependencies between the parties become stronger (Dwyer, Schurr and 14 Oh 1987). All these factors would make switching more costly. In support of these arguments, Bell, Auh and Smalley (2005) emphasize that, as relationships deepen, consumersвЂ™ switching costs increase; Keaveney and Parthasarathy (2001) report a negative effect of service usage on customer switching behavior; and Chen and Hitt (2002), in an online context, show a positive influence of service usage on the probability of customers staying with the company. Together, these studies suggest a positive association between service usage and the cost of switching. The previous argument implies that an extremely heavy user should bear the highest cost when switching suppliers. However, some researchers argue that intensive service usage allows customers to become more informed about the products and services and to evaluate options more rapidly and accurately (Alba and Hutchinson 1987). This might have the opposite effect on the customer switching cost. Heavy users are expected to be able to better evaluate and compare the available alternatives in the market, which will lead to a lower dependence on the supplier (Bendapudi and Berry 1997; Ganesan 1994). In addition, heavy users might also have greater incentives to look for the service provider that best fits their needs and provides more value (Chen and Hitt 2002). Taking future benefits into account would reduce switching costs and lead to a higher propensity to switch service providers. In support of this argument, Johnson et al. (2004), in an online context, report that more active shoppers tend to search across more alternatives. Based on this discussion, the only way to accommodate both arguments is by means of a non-linear relationship. Therefore, we propose an inverse U-shaped association between the depth of the relationship and the cost of switching, where intermediate service usage is associated with the highest switching costs. Thus: H5: The depth of the relationship is related to the cost of switching in an inverse U- shaped manner. Intermediate service usage is associated with the highest switching cost. 15 The breadth of a relationship is reflected in the number of additional products or services purchased from a company over time. Buying additional products or services from the firm provides customers with a better knowledge of the company (McCracken 1986), which leads to more accurate expectations and to less uncertainty (Keaveney and Parthasarathy 2001), and increases the number of points where customer and vendor connect (Kamakura et al. 2003). In addition, the switching process for customers who buy additional products or services is expected to be more complicated for the following reasons: first, the risk associated with the switching decision is higher because there are more services involved in the process (Burnham, Frels and Mahajan 2003); second, switching providers implies the need to compare alternative vendors in a greater number of attributes and characteristics (Shugan 1980); third, switchers need to invest additional resources in learning the new features of the switched-to providerвЂ™s products and services. Thus, we propose the following hypothesis: H6: The breadth of the relationship positively influences the cost of switching. Control Variables In the conceptual model, we include a set of control variables that may affect the cost associated with switching suppliers: type of setting, gender and age. Type of setting. Understanding the differences between contractual and non-contractual settings in terms of their impact on switching costs can be of interest to marketing literature. Although some studies have already acknowledged the differences between them and their different impacts on some relational constructs (Reinartz and Kumar 2000, 2003), empirical research testing for these differences is lacking. A contractual setting refers to a relationship which is governed by a contract or membership. Typical examples are Internet services or magazine subscriptions. In contrast, non-contractual relationships are those that are not 16 governed by a contract. Examples of non-contractual settings are grocery store purchases or hairdressing services. Due to the distinctive characteristics of these two settings, they are expected to have different impacts on the cost of switching. In a contractual setting, the customer signs a contract that specifies the details of the arrangement and that normally commits her/him to stay in the relationship for a specified period. Otherwise, the customer has to compensate the firm (normally by paying a monetary penalty). In addition, uncertainty about future exchanges is lower because some eventualities like price or service level are normally specified in the contract. In contrast, in a noncontractual setting customers are not committed to any firm and there is no penalty for switching providers. In this setting, switching behavior is more common and customers usually split their expenses among several firms (Dwyer 1997). Reinartz and Kumar (2000, 2003) have argued that switching costs are higher in contractual settings than in noncontractual ones. This discussion allows us to formulate the following hypothesis that associates the type of setting and the cost of switching: H7: Contractual settings positively influence the cost of switching. Our model also includes customer demographics which capture observed customer heterogeneity. The marketing literature has already stressed the important role played by socio-demographic variables in explaining the differences in profitable lifetime duration (Reinartz and Kumar 2003), perceived satisfaction levels (Mittal and Kamakura 2001), repurchase intentions and switching behavior (Keaveney and Parthasarathy 2001). The main motivation for introducing these variables is for statistical and segmentation purposes. We examine heterogeneity in terms of customersвЂ™ age and gender. However, we do not posit any directional association between these variables and the customer switching cost due to the absence of an appropriate theory. 17 RESEARCH METHODOLOGY Studies investigating customer retention have normally used regression analyses or structural equation modeling to assess the impact of the independent variables on customer repurchase intentions/behavior (Burnham, Frels and Mahajan 2003; Lam et al. 2004; Verhoef, Franses and Hoekstra 2002), whereas studies aimed at analyzing switching behavior mostly use techniques such as discriminant analysis or the critical incidence technique (CIT, or some extensions) (Ganesh, Arnold and Reynolds 2000; Keaveney 1995; Keaveney and Parthasarathy 2001; Roos 1999). In this study, we investigate the drivers of the cost of switching by implementing a hierarchical linear model (Rossi and Allenby 2003; Snijders and Bosker 1999). Hierarchical linear modeling has been increasingly applied in recent years to a wide variety of marketing problems (see Rossi and Allenby 2003) ranging from demand analyses (Allenby, Arora and Ginter 1998) to heterogeneity in customer responses to marketing interventions (Rust and Verhoef 2005). We employ this methodology to study customer switching costs for the following reasons. First, the data set provides information at two different levels: occasions (level 1) nested in individuals (level 2). Hierarchical data structures require multilevel techniques to be applied instead of traditional regression methods (Snijders and Bosker 1999). Second, this is a customer-specific analysis of switching costs. Bayesian methods allow us to estimate choice model specifications with coefficients of explanatory variables varying over decision units (Solgaard and Hansen 2003). Third, Bayesian methods are able to estimate complex behavioral models taking into account the heterogeneity across individuals and other sources of uncertainty (Allenby, Bakken and Rossi 2004). Finally, recent developments in computational and modeling methods have made it possible to calculate the posterior distribution of a wide variety of models, including the one employed in this study. 18 We specify the hierarchical linear model as follows. At the occasion level (level 1), we use a technique to measure switching costs similar to Lee et al. (2006). According to the random utility framework (McFadden 1974), an individual derives utility by choosing the alternative that gives her/him the highest value. Therefore, in each period t (out of T), customer i (out of I) faces a choice of J alternatives in her/his choice set. The selected option will be the one with the highest utility. The utility that customer i obtains from selecting firm j in period t (Uijt) is formulated as a function of the attributes associated with the alternative j (systematic component) and of a random disturbance (random component): U ijt = X ijt ОІ i + Оµ ijt (1) where Xijt is a vector of service attributes associated with alternative j, ОІi is the parameter vector that measures the impact of service attributes on customer choice decisions, and Оµijt is a random term that is iid extreme value. Similar to Lee et al. (2006), the cost of switching is introduced into the model as follows. The X vector contains a switching dummy variable, labeled Sijt, which takes the value one in period t if customer i chooses an alternative service provider to her/his present supplier, and zero otherwise. Therefore, the parameter accompanying the variable Switching (Sijt), ОІ i , is S interpreted as the marginal utility that customer i derives from switching service providers. It is important to note that this is a customer-specific parameter as a value per customer is obtained from this model. To obtain the customer-specific parameters we adopt a Bayesian mixed logit specification, as in Train (2003). 3 3 A Bayesian mixed logit model is a highly flexible model that overcomes the three main limitations associated with the standard logit model as it allows for: (1) random taste variation, (2) unrestricted substitution patterns and (3) correlation in unobserved factors over time (Train 2003). 19 At the second layer of the hierarchy, customer-related variables are introduced to explain the heterogeneity in the first-level parameters. In this study, our focus is on the switching costs parameters. This leads to the following expression, ОІ i S = Z iО± + П„ i (2) where Zi is a vector containing customer-related variables (relationship characteristics, control variables) and marketing instruments, О± is the parameter vector measuring the impact of these variables on the customer switching cost, and П„i is a normally distributed disturbance term. Conventional marketing literature usually includes consumer characteristics at the second level of the hierarchy in order to explain heterogeneity in the first level parameters. However, in this study we also consider marketing variables at the second level as we aim to fully capture the conditions under which switching took place for each customer. That is, price and advertising are treated as customer-specific variables because a different value pertains to each customer depending on (1) the period in which they switched providers and (2) the firm they switched from. Thus, in order to better explain the heterogeneity in the cost of switching, Equation 2 analyzes not only the role of customer characteristics in the switching period, but also the role of marketing actions implemented by both the focal firm and the competitors in that period. Estimation of this hierarchical linear model is accomplished by using Markov Chain Monte Carlo (MCMC) methods, which approximate the posterior distribution by sampling from the full conditional distribution. 20 Data description Our empirical study is carried out in the Spanish mobile phone industry using data from three different sources. Information on relationship characteristics and control variables was provided by TNS Global. The database covers a four-year window (from 2001 to 2004) on a monthly basis (48 periods) and consists of a total number of observations of 287 mobile users who switched service providers during the observation period. We combine this information with monthly data on marketing instruments and firm attributes for all the firms competing in the market during the period under analysis: Movistar (the former public monopoly), Vodafone and Amena (currently Orange). This information was provided by two distinct databases: the Global Wireless Matrix database and Infoadex. Two important notes about the data are worth mentioning. First, although the available sample covers only a portion of the market, namely, the customers who switch, it is important to note that considering only these customers will probably provide a better picture of the actual cost of switching, as they are completely aware of the cost and disutility that switching providers causes. They have gone through the process of switching and have experienced the inconvenience associated with it. For customers who have never switched, the exact magnitude of the switching cost is not known: in the Spanish mobile phone industry, most customers ignore how much information they need to get in order to take an informed decision and/or they do not know much about how long the process of switching is. The real cost is not known unless switching occurs. In addition, customers who have never switched tend to overestimate the magnitude of these costs, leading to a biased analysis of switching costs. Instead of focusing on switching cost perceptions, this study investigates the real cost of switching for customers in the mobile phone industry. Second, most studies dealing with switching costs use self-reported and/or cross-sectional data. Although this kind of data might be useful for obtaining customer preferences for hypothetical alternatives (Lee et al 2006), it 21 may lead to an overestimation of the considered relationships and to serious methodological problems (Verhoef 2003). In addition, switching is a dynamic process (Roos 1999). Longitudinal information should therefore be used to better capture the underlying mechanisms that generate the cost of switching for customers. We have selected the mobile phone industry for the empirical analysis for the following reasons. First, the mobile phone industry is considered one of the most important sectors, not only in the communications field but also in the economy as a whole. Second, in most developed countries the penetration rate is very high, 4 which indicates that the mobile phone industry is reaching the mature stage of its lifecycle. Retention strategies, therefore, gain importance over acquisition strategies, leading to an intense interest in the drivers of customer retention. Third, the high fixed-cost structure of customer acquisition in information-intensive markets makes customer retention a strategic mandate for firms in order to compensate the initial sunk investments, which increases the relevance of studying retention mechanisms, such as switching costs, in these contexts (Chen and Hitt 2007). Fourth, the Spanish mobile phone market has recently been liberalized. Studying switching behavior in this type of market is a subject of great interest to marketing theory and practice (Wieringa and Verhoef 2007). Fifth, from a public policy perspective, switching costs are considered a threat to achieving competition in the mobile phone market (Grzybowski 2005). Finally, the mobile phone industry has been the focus of attention in prior studies dealing with switching costs and consumer retention (e.g. Watson et al. 2002; Lee et al. 2006), which shows the interest among scholars in understanding customer behavior in this specific industry. 4 In the Spanish market, the penetration rate at the end of 2006 was 103% (Source: Global Wireless Matrix database). 22 Variable operationalization For each month in the study period, we observe the customersвЂ™ choice of mobile phone operator. In Equation 1, utility is an unobserved latent variable that is manifested through customer choices (the alternative selected in each period is the one that maximizes customer utility). The dependent variable (Uijt) is then operationalized as a dummy variable that takes two possible values: one if customer i chooses firm j in period t, and zero otherwise. Customer choices are based on an evaluation of firm attributes over time. We introduce three explanatory variables into the utility model: price, firm size and switching costs. These variables have been widely used in prior studies that investigate customer choice behavior, and the inclusion of them in the utility function is similar to Lee et al. (2006) and Birke and Swann (2006). These variables are time-varying and are measured and updated each month. Price (Pjt) is operationalized as the Average Revenue per User (ARpU). Mobile phone prices are very complex and normally involve non-linear tariffs associated to different call types, which makes it difficult to obtain a single price per firm. This has led market research companies to develop some measures in order to approximate the price charged by each mobile phone operator. One measure that has received particular attention among academics is ARpU (McCloughan and Lyons 2006), and it has been previously used in research on switching costs (Shy 2002). We define firm size (Tjt) as the logarithm of the number of subscribers of firm j in each time period. The switching dummy variable (Sijt) has already been defined. In Equation 2, the criterion variable is the customer-specific estimate of the cost of switching providers and, thus, only a single event associated with each customer exists. Marketing instruments are introduced as follows. We define the focal firmвЂ™s price as the ARpU of the focal firm in the period in which switching takes place for customer i. CompetitorsвЂ™ price is measured as the average ARpU of the two competing firms in that 23 period. The variable introduced into the analysis (PRi) is the difference between them (Focal FirmвЂ™s Price вЂ“ Average CompetitorsвЂ™ Price). The advertising measurement follows the same rule. We use data on service and brand advertising expenditures (in millions of euros) by the focal firm in the switching period to test for the effect of the focal firmвЂ™s advertising on switching costs. To analyze the effect of competitorsвЂ™ advertising, we use the average service and brand advertising expenditures by competitors in that period. We introduce the difference between the focal firmвЂ™s and the competitorsвЂ™ advertising expenses in both categories (service advertising SAi, and brand advertising BAi) into the analysis. Relationship characteristics are introduced in the following way. Relationship length (LEi) is measured as the duration, in months, from the beginning of the relationship until the period in which customer i switches to a competitor. To measure relationship depth (DEi), we compute the average mobile phone consumption of customer i from the beginning of the relationship until the switching period. We define the breadth of the relationship (BRi) as a dummy variable taking the value one if customer i is subscribed to an additional service offered by the company in the period when switching takes place, and zero otherwise. 5 In addition, our research includes a set of covariates for statistical control purposes. Type of setting (Tsi) is operationalized as a dummy variable. It takes the value one if the customerfirm relationship takes place in a contractual setting (customer i has a post-paid subscription in the switching period), and zero otherwise (the relationship takes place in a non-contractual setting, that is, the customer has a pre-paid subscription in the switching period). With respect to customer demographics, we define the variable age as the consumerвЂ™s age, in years, in the switching period (Agi), and we set the gender dummy variable to zero for male customers (Gei). Table 1 describes the variables included in the analysis. 5 Mobile phone companies offer additional services which allow customers to call in special conditions or to benefit from special offers. For instance, Vodafone offers the A2 package, under which customers can communicate with one selected contact in special conditions (lower rates). 24 TABLE 1 Variable Measurement LEVEL 1: CHOICE MODEL Variable Utility Hypothesized Directional Impact Label Dummy variable: 1 = customer i chooses firm j in period t 0 = otherwise. Uijt Price Pjt No directional hypothesis Tjt No directional hypothesis Sijt вЂ“ Average Revenue Per User (ARpU) of firm j in period t Firm Size Logarithm of the number of subscribers of firm j in period t Switching Dummy variable: 1 = customer i chooses an alternative service provider in period t. 0 = otherwise LEVEL 2: DRIVERS OF THE COST OF SWITCHING Variable Label Hypothesized Directional Impact Focal Firm ARpU вЂ“ Average Competitors ARpU (both measured in the switching period) PRi вЂ“ SAi No directional hypothesis BAi + Duration of the relationship (in months) from the beginning until the switching period. LEi + Depth DEi Marketing Variables Price Service Advertising Focal Firm Service Advertising Expenses вЂ“ Average Competitors Service Advertising Expenses (both measured in the switching period and in millions of euros) Brand Advertising Focal Firm Brand Advertising Expenses вЂ“ Average Competitors Brand Advertising Expenses (both measured in the switching period and in millions of euros) Relationship Characteristics Length Average mobile consumption (in euros) from the beginning of the relationship until the switching period Breadth (DEi) 2 Inverse U-shaped relationship BRi + Tsi + Dummy variable: 1 = Female 0 = Male Gei No directional hypothesis Age Agi No directional hypothesis Dummy variable: 1 = Acquisition of an additional service from the company 0 = Otherwise Control Variables Type of Setting Dummy variable: 1 = Contractual setting (post-paid subscription) 0 = Non-contractual setting (pre-paid subscription) Gender Age (in years) in the switching period 25 EMPIRICAL FINDINGS All computations and simulations for this study were carried out using the WinBUGS package (Spiegelhalter et al. 2003). We ran 60,000 iterations, of which 50,000 were used for burn-in and the remainder (10,000) for estimation purposes. Autocorrelation in the sample was reduced by thinning the iterationsвЂ“taking only one in five. The 2,000 retained draws were used to conduct inference. The parameter estimates for the hierarchical linear model are reported in Table 2. Based on both graphical and numerical techniques, we can conclude that all the parameters converged to their posterior distributions. Statistical significance is not a concern of Bayesian analyses, but we can construct an analog in order to assess parameter importance. The table reports (1) the mean, (2) the standard deviation, (3) the Monte Carlo standard error and (4) the mean divided by this error. Parameter significance depends on this quotient being higher than two вЂ“in absolute valueвЂ“ (Rust and Verhoef 2005), which is satisfied by all the parameters included in the model. The parameters measuring the cost of switching providers are our main concern in this analysis. We find that the marginal utility from switching service providers is negative (ОІ3[Avg] = -4.24). That is, switching service providers is costly: it implies an important reduction in the utility level. We also find that there are significant differences in switching costs across customers, as shown by the standard deviation (ОІ3[Std] = 0.69). Figure 2 illustrates the distribution of switching costs in the sample. As we can see from the figure, the marginal utility from switching suppliers is quite heterogeneous, ranging from 2.48 to 6.06 (in absolute values). Although it is not the main concern of our research, price and firm size also contribute to explaining customersвЂ™ choices. On average, price negatively influences consumer utility (ОІ1[Avg] = -0.06), while firm size has a positive impact on it (ОІ2[Avg] = 0.54). 26 TABLE 2 Hierarchical Model Parameter Estimates Parameter Mean Standard Deviation Monte Carlo S.E. Mean/MCSE ОІ1[Avg] ОІ2[Avg] ОІ3[Avg] -0.0561 0.5378 -4.2425 0.0201 0.0712 0.6893 0.0005 0.0042 0.0053 -98.77 128.63 -792.23 О±1 -0.0122 0.0022 0.0010 -12.21 О±2 0.0265 0.0038 0.0002 160.71 О±3 0.0321 0.0104 0.0005 69.63 О±4 О±5 О±6 0.0737 0.0053 -0.0006 0.0068 0.0036 0.0001 0.0002 0.0001 0.0000 312.51 33.36 -274.27 О±7 0.5578 0.1176 0.0047 117.43 О±8 0.0823 0.0908 0.0082 10.08 О±9 О±10 0.1868 0.0318 0.1375 0.0034 0.0016 0.0001 110.08 197.52 CHOICE MODEL Price Firm Size Switching Cost DRIVERS OF THE COST OF SWITCHING Marketing Variables Price (Focal firm ARpU вЂ“ Competitors ARpU) Service Advertising (Focal firm investments вЂ“ CompetitorsвЂ™ investments) Brand Advertising (Focal firm investments вЂ“ CompetitorsвЂ™ investments) Relationship Characteristics Length (Duration) Depth (Consumption) (Depth)2 Breadth (Additional Service: 1 = acquisition; 0 = not acquisition) Control Variables Type of setting (1 = Contractual; 0 = Noncontractual) Gender (1 = female; 0 = male) Age In order to ascertain the drivers of customer switching costs, two main groups of variables were introduced into the model: marketing instruments and relationship characteristics. To facilitate the analysis and discussion of the results, absolute values of switching costs were taken. The results show that marketing instruments play a significant role in explaining switching cost differences across customers. Consistent with hypotheses 1a and 1b, the difference between the focal firmвЂ™s and the competitorsвЂ™ price negatively impacts customer switching costs (О±1 = -0.01). That is, while the focal firmвЂ™s price negatively influences the cost of switching providers, the competitorsвЂ™ price has a positive impact on it. 27 The parameter accompanying the variable service advertising has a positive sign (О±2 = 0.03), which means that the difference between the focal firmвЂ™s and the competitorsвЂ™ expenditures on service advertising positively influences customer switching costs. This is consistent with hypothesis 2b in that competitorsвЂ™ service advertising reduces the disutility from switching providers. We find support for hypotheses 3a and 3b in that the difference between the focal firmвЂ™s and the competitorsвЂ™ expenditures on brand advertising has a positive effect on the customer switching cost (О±3 = 0.03). While the focal firmвЂ™s brand advertising increases the cost of switching, the competitorsвЂ™ brand advertising reduces it. FIGURE 2 Switching Cost Distribution in the Sample 30% Frequency % 25% 20% 15% 10% 5% 0% < 3.0 3.1вЂ“3.5 3.6вЂ“4.0 4.1вЂ“4.5 4.6вЂ“5.0 5.1вЂ“5.5 > 5.6 Switching Costs (in absolute value) Relationship characteristics play a significant role in explaining switching cost differences as well. Consistent with hypothesis 4, we find that relationship duration is positively associated with the customer switching cost (О±4 = 0.07). We find support for hypothesis 5, in that there is a non-linear, inverse U-shaped relationship between mobile phone consumption and the cost of switching (О±5 = 0.01; О±6 = -0.00). Hypothesis 6 is also 28 supported because cross-buying behavior positively influences the disutility derived from switching service providers (О±7 = 0.56). Finally, the estimation results show that the parameter accompanying the variable Type of setting has the expected sign (О±8 = 0.08) and customers in a contractual setting (post-paid subscription) experience higher switching costs than customers in a non-contractual one (prepaid subscription). This is consistent with hypothesis 7. The results also show that female customers bear higher switching costs than males (О±9 = 0.19) and that customer age is positively associated with the cost of switching providers (О±10 = 0.03). DISCUSSION AND IMPLICATIONS In this research we study the drivers of customer switching costs. In particular, we focus on investigating the extent to which customer switching costs are influenced by marketing instruments and relationship characteristics. Although marketing literature has long emphasized the significant consequences of switching costs in terms of customer retention and customer switching behavior, research on the drivers of switching costs remains limited. To the best of our knowledge, this study represents the first attempt to integrate marketing variables and relationship characteristics into a single framework about the drivers of switching costs and to document the effect of competition on the magnitude of these costs. This approach offers an interesting contribution to marketing literature as a better understanding of customer retention and its drivers is central to relationship marketing theories and to marketing strategy. The present study reveals that marketing variables вЂ“price and advertisingвЂ“ are important drivers of switching costs. This is an interesting result, because prior research has mainly focused on relationship variables as the key elements leading to high switching barriers. Our 29 study results show that marketing instruments also contribute to making switching service providers costly. Specifically, price has a negative influence and both service and brand advertising have a positive effect on switching costs. While the negative effect of price and the positive impact of brand advertising on switching costs are consistent with our hypotheses, the positive effect of service advertising is more remarkable. There were reasons for both a positive and a negative association between service advertising and the cost of switching. On the one hand, service advertising highlights the advantages and benefits of choosing the focal firmвЂ™s service, which makes switching less attractive. On the other hand, this kind of advertising increases customersвЂ™ information about the characteristics and attributes of the service, making switching easier. Our study results show that the first effect dominates over the second and thus, investments in service advertising lead to high switching costs. These findings are also important for the customer relationship management literature. Some studies indicate that, after the customer-firm relationship has been established, the role of price and advertising become less prominent (Kordupleski, Rust and Zahorik 1993; Rust, Zahorik and Keiningham 1995). Our investigation shows that price and advertising are also important to stimulate customer behavior once the relationship has been established. In addition to investigating the effect of marketing strategies by the focal firm, we explored the role of the competitors in customer switching costs. Marketing scholars have long emphasized that understanding customer responses to competitive actions is essential to marketing theory and practice. However, research in this area is very limited mainly due to the lack of available information. We have access to a database that provides information on price and advertising for all the companies operating in the market, which enables us to extend current research on this subject by investigating the extent to which switching costs are affected by competitorsвЂ™ marketing strategies. Our study results reveal that, by acting on price and advertising (both service and brand advertising), competitors are able to influence the cost 30 of switching providers for customers of the focal firm. Specifically, our results show that a price cut is an effective strategy for competitors in order to attract customers from the focal firm as it subsidizes the cost of switching they bear. This result is in line with previous research showing that firms вЂњpay customers to switchвЂќ (Chen 1997). We find a negative association between both competitorsвЂ™ service and brand advertising and the cost of switching providers. An explanation for the first effect is that customers of the focal firm exposed to competitorsвЂ™ service communication efforts are more informed about the characteristics of the service, which leads to a reduction of the switching costs. More notable is the finding of a negative association between competitive brand advertising and the cost of switching as it shows that even marketing communication efforts not directed at the service itself affect the cost of switching providers for customers of the focal firm. A possible explanation for this result is that customers of the focal firm who are exposed to this type of advertising can develop positive feelings toward competing brands, leading to a decrease of the uncertainty associated with changing providers. In a recent article, Prins and Verhoef (2007) show that competitive communication efforts affect the adoption timing of a new product among existing customers of the focal firm. We extend current knowledge on the effect of competitive communication actions by showing that they also influence customer switching costs. We also examined the extent to which switching costs are influenced by relationship characteristics. This is an interesting contribution to the literature as, to our knowledge, this is the first study that simultaneously investigates the impact of the length, depth and breadth of the relationship on customer switching costs. The results reveal that both relationship length and breadth are positively related to the cost of switching. This is a rather interesting result as it shows that increasing the duration of the relationship and/or cross-buying additional products or services from the focal firm make switching more costly for customers. This is in 31 line with prior research showing that switching providers is easier in the early stages of the relationship (Ongena and Smith 2001; Verhoef, Franses and Hoekstra 2002). We reasoned that relationship depth might have a non-linear association with the cost of switching. The results confirm this expectation, showing that relationship depth and switching costs are related in an inverse U-shaped manner. This is a notable finding as it shows that intermediate service consumption is associated with the highest switching cost, while both lowconsumption and high-consumption customers find switching providers less costly. Overall, these results provide further evidence that the history of the relationship plays an important role in explaining customer heterogeneity in relational constructs (Reinartz and Kumar 2003). Although the control variables were not our primary focus, our study results show that the type of setting is an important factor explaining the cost of switching as well. In a contractual setting, where the relationship is governed by a contract or membership, customers find switching providers more costly than in a non-contractual setting. While this finding is consistent with our prior expectations, to the best of our knowledge no study to date has simultaneously considered both types of setting in a single study and empirically evaluated their different effects on relational variables. Finally, from a database and modeling perspective, our study uses longitudinal information on customer behavior and marketing variables and applies a hierarchical linear model to study switching costs. The use of this kind of data avoids some methodological and statistical problems associated with other sources of information (self-reported, crosssectional) and allows us to better capture the dynamic and heterogeneous nature of switching costs. In addition, the availability of marketing data for all the firms in the market enables us to examine the impact of competitorsвЂ™ marketing practices on customer switching costs. In terms of the hierarchical linear modeling, although there has been increasing interest in recent years in using these techniques to study marketing problems, this has been less prominent in 32 the investigation of customer retention. Based on this modeling approach, we were able to (1) handle the multilevel structure of the data available to us (occasions nested in individuals), (2) obtain individual-specific parameters and (3) fully capture customer heterogeneity (Rossi and Allenby 2003). Classical procedures often fail to capture customer heterogeneity and, with complex behavioral models (like the one specified in this study), the maximization of the simulated likelihood function can be difficult (Train 2003). Thus, our study extends current knowledge on the use of hierarchical modeling by applying it to the investigation of customer switching costs. Managerial Implications Achieving customer retention is essential for many firms in order to increase revenue and profitability. This is especially true in information-intensive businesses where the high fixedcost structure of customer acquisition increases the importance of retaining customers in order to overcome the initial sunk investments. However, successful customer retention requires a proper understanding of its drivers. In the information economy, customer switching costs have become one of the most powerful and effective mechanisms to achieve stable and longlasting relationships (Chen and Hitt 2007). The question is: how can firms make switching costly? By understanding the determinants of switching costs, a firm can identify the key factors that must be addressed to prevent future defections. Our study results show that firms can use two general-oriented marketing instruments, price and advertising, to increase the cost of switching suppliers. However, the effect of advertising вЂ“both service and brand advertisingвЂ“ on switching costs is greater than that of price. This suggests that making the process of switching costly for customers should be done mainly by communicating the service and the brand. This strategy, at the same time, can help the company to increase the size of its customer base as it reduces the cost of switching for 33 customers of competing vendors. Another strategy might be to focus on price. Our study results reveal that, by reducing the level of this marketing instrument, switching becomes more difficult for existing customers. Similar to advertising, this has an additional positive effect on the firmвЂ™s customer base as it compensates the cost of switching for customers of competing suppliers. An important result of our research is that competitorsвЂ™ marketing actions also influence the cost of switching. In particular, price cuts and investments in service and brand advertising by competitors may induce customers of the focal firm to switch providers because these strategies subsidize, at least in part, their switching costs. This suggests that, to prevent future defections, firms should react in a timely manner to competitorsвЂ™ marketing movements. A proper understanding of the history of the relationship is also needed to optimally manage customer switching costs. Our results suggest that deploying strategies based on the customer relationship characteristics will also make switching costly. Relationship length and breadth were found to have a positive influence on the disutility derived from switching. Useful strategies might involve encouraging customers to stay in the relationship (e.g. giving them economic or relational benefits), and/or cross-selling additional products or services (e.g. extending the current offer of the firm and persuading customers to acquire new services вЂ“Internet services, TV). Concerning the depth of the relationship, firms should be aware that both customers who spend less and those who spend more find switching less costly. Our study results emphasize the relevance of the type of setting in explaining the cost of switching providers. In a contractual setting customers find switching providers more costly than in a non-contractual setting. In our empirical application, this distinction was based on the type of subscription the customer had: pre-paid or post-paid. Our results suggest that an interesting strategy for mobile phone firms might be to encourage existing pre-paid customers 34 to change to post-paid subscriptions, as this will increase the cost of switching. In addition, firms can persuade competing firmsвЂ™ customers or prospective ones to choose post-paid contracts. This study also has implications for public policy. Customer switching costs are an important threat to market competition (Grzybowski 2005). In most developed countries, the regulator has implemented policies in recent years aimed at reducing the magnitude of these costs. Number portability is a well-known example. However, this study clearly demonstrates the existence of high switching costs in the mobile phone market in spite of these efforts. Our results suggest that more measures should be taken by public policy officials in order to further reduce switching costs and to increase competition in the market. Finally, we should add a few words of caution regarding the management of customer switching costs. Although increasing the magnitude of these costs is an effective strategy for retaining customers in the relationship, it might lead to negative reactions from customers if they feel locked into a firm in which they would prefer not to be (Jones et al. 2007). Research Limitations and Further Research Some limitations of our study should be noted and they suggest possible directions for future research. First of all, in our empirical application, we only consider customers who have switched service providers. Although this is an interesting contribution to the literature because the actual cost of switching can be computed, the data set covers only a portion of the market. Future research might also consider consumers who have never switched providers and study the differences between switching and non-switching customers in order to obtain a better understanding of customer retention and customer switching behavior. 35 Second, we tested our hypotheses in one particular context, namely, the mobile phone industry. While this industry has some distinctive characteristics that make it relevant to the study of switching costs (e.g. recently liberalized market, high cost of customer acquisition), it would, of course, be important to investigate switching costs in other industries as well. In addition, the mobile phone industry is in the mature stage of its lifecycle. Thus, it might be interesting to study the cost of switching in less established industries where new consumers progressively enter the market. A promising avenue for further research would consist of analyzing firmsвЂ™ trade-offs between acquiring new customers вЂ“creating long customer bases to obtain higher rents in the futureвЂ“ and retaining existing ones вЂ“harvesting the current customer base to obtain higher rents in the present (Farrell and Klemperer 2007). Clearly, more research is required to address these issues. Third, our study shows that switching costs are affected by marketing strategy. Specifically, we focused on price and advertising. Clearly, firms have a wider range of marketing activities at their disposal to influence and stimulate customer behavior. Customer loyalty programs and/or product design strategies are good examples. Prior research has already emphasized their importance in affecting customer behavior and value (Bolton, Lemon and Verhoef 2004; Verhoef 2003). Thus, a promising avenue for future research would be to study the role of relationship marketing instruments in customer switching costs. Finally, public policy officials are very concerned about the negative consequences of switching costs on market competition. We have acknowledged that regulatory policies might also play an important role in explaining switching costs in this particular industry. For instance, prior research has shown that mobile number portability has significantly reduced the magnitude of these costs (Lee et al. 2006). Our study, however, does not include any regulatory policy. Thus, an interesting extension of this research would be to analyze the role of the regulator in driving customer switching costs. 36 REFERENCES Alba, Joseph W. and J. Wesley Hutchinson (1987), вЂњDimensions of Consumer Expertise,вЂќ Journal of Consumer Research, 13 (March), 411вЂ“54. Allenby, Greg M., Neeraj Arora, and James L. Ginter (1998), вЂњOn the Heterogeneity of Demand,вЂќ Journal of Marketing Research, 35 (August), 384вЂ“89. Allenby, Greg M., David G. Bakken, and Peter E. Rossi (2004), вЂњThe HB Revolution: How Bayesian Methods Have Changed the Face of Marketing Research,вЂќ Marketing Research, 16 (2), 20вЂ“25. Anderson, Eugene W. and Mary W. Sullivan (1993), вЂњThe Antecedents and Consequences of Customer Satisfaction for Firms,вЂќ Marketing Science, 12 (Spring), 125вЂ“43. Bell, Simon J., Seigyoung Auh, and Karen Smalley (2005), вЂњCustomer Relationship Dynamics: Service Quality and Customer Loyalty in the Context of Varying Levels of Customer Expertise and Switching Costs,вЂќ Journal of the Academy of Marketing Science, 33 (2), 169вЂ“183. Bendapudi, Neeli and Leonard L. Berry (1997), вЂњCustomersвЂ™ Motivations for Maintaining Relationships With Service Providers,вЂќ Journal of Retailing, 73 (1), 15вЂ“37. Birke, Daniel and G. M. Peter Swann (2006), вЂњNetwork Effects and the Choice of Mobile Phone Operator,вЂќ Journal of Evolutionary Economics, 16 (1-2), 65вЂ“84. Bolton, Ruth N. (1998), вЂњA Dynamic Model of the Duration of the Customer's Relationship with a Continuous Service Provider: The Role of Satisfaction,вЂќ Marketing Science, 17 (1), 45вЂ“65. Bolton, Ruth N., P.K. Kannan, and Matthew D. Bramlett (2000), вЂњImplications of Loyalty Program Membership and Service Experiences for Customer Retention and Value,вЂќ Journal of the Academy of Marketing Science, 28 (Winter), 95вЂ“108. Bolton, Ruth N. and Katherine N. Lemon (1999), вЂњA Dynamic Model of CustomersвЂ™ Usage of Services: Usage as an Antecedent and Consequence of Satisfaction,вЂќ Journal of Marketing Research, 36 (May), 171вЂ“86. Bolton, Ruth N., Katherine N. Lemon, and Peter C. Verhoef (2004), вЂњThe Theoretical Underpinnings of Customer Asset Management: A Framework and Propositions for Future Research,вЂќ Journal of the Academy of Marketing Science, 32 (Summer), 271вЂ“92. Burnham, Thomas A., Judy K. Frels and Vijay Mahajan (2003), вЂњConsumer Switching Costs: A Typology, Antecedents and Consequences,вЂќ Journal of the Academy of Marketing Science, 31 (Spring), 109вЂ“26. Chen, Yongmin (1997), вЂњPaying Customers to Switch,вЂќ Journal of Economics & Management Strategy, 6 (4), 877вЂ“97. Chen, Pei-Yu and Chris Forman (2006), вЂњCan Vendors Influence Switching Costs and Compatibility in an Environment with Open Standards,вЂќ MIS Quarterly, 30 (Special Issue), 541вЂ“62. 37 Chen, Pei-Yu and Lorin M. Hitt (2002), вЂњMeasuring Switching Costs and Their Determinants in Internet Enabled Businesses: A Study of the Online Brokerage Industry,вЂќ Information Systems Research, 13 (3), 255вЂ“74. Chen, Pei-Yu and Lorin M. Hitt (2007), вЂњInformation Technology and Switching Costs,вЂќ in T. Hendershott, ed., Handbook of Information Systems Economics. Chiu, Hung-Chang, Yi-Ching Hsieh, Yu-Chuan Li, and Monle Lee (2005), вЂњRelationship Marketing and Consumer Switching Behavior,вЂќ Journal of Business Research, 58, 1681вЂ“89. Cramphorn, Spike (2006), вЂњHow to Use Advertising to Build Brands: in Search of the Philosopher's Stone,вЂќ International Journal of Market Research, 48 (3), 255вЂ“75. Davis, Robert, Margo Buchanan-Oliver, and Roderick J. Brodie (2000), вЂњRetail Service Branding in Electronic Commerce Environments,вЂќ Journal of Service Research, 3 (2), 178вЂ“86. De Wulf, Kristof, Gaby Odekerken-SchrГ¶der, and Dawn Iacobucci (2001), вЂњInvestments in Consumer Relationships: A Cross-Country and Cross-Industry Exploration,вЂќ Journal of Marketing, 65 (October), 33вЂ“50. Dick, Alan S. and Kunal Basu (1994), вЂњCustomer Loyalty: Toward an Integrated Conceptual Framework,вЂќ Journal of the Academy of Marketing Science, 22 (2), 99вЂ“113. Dwyer, F. Robert (1997), вЂњCustomer Lifetime Valuation to Support Marketing Decision Making,вЂќ Journal of Direct Marketing, 11 (4), 6вЂ“13. Dwyer, F. Robert, Paul H. Schurr, and Sejo Oh (1987), вЂњDeveloping Buyer-Seller Relationships,вЂќ Journal of Marketing, 51 (April), 11вЂ“27. Einhorn, Michael A. (1994), вЂњOptimal Nonuniform Pricing with Generalized Consumer Choice,вЂќ The Journal of Industrial Economics, 42 (March), 105вЂ“13. Farrell, Joseph and Paul Klemperer (2007). вЂњCoordination and Lock-In: Competition with Switching Costs and Network Effects,вЂќ in M. Armstrong, and R. Porter ed., Handbook of Industrial Organization, Volume 3, Elsevier. Fornell Claes (1992), вЂњA National Customer Satisfaction Barometer: The Swedish Experience,вЂќ Journal of Marketing, 56 (January), 6вЂ“21. Ganesan, Shankar (1994), вЂњDeterminants of Long-Term Orientation in Buyer-Seller Relationships,вЂќ Journal of Marketing, 58 (April), 1вЂ“19. Ganesh, Jaishankar, Mark J. Arnold, and Kristy E. Reynolds (2000), вЂњUnderstanding the Customer Base of Service Providers: An Examination of the Differences Between Switchers and Stayers,вЂќ Journal of Marketing, 64 (July), 65вЂ“87. Garbarino, Ellen and Mark S. Johnson (1999), вЂњThe Different Roles of Satisfaction, Trust, and Commitment in Customer Relationships,вЂќ Journal of Marketing, 63 (April), 70вЂ“87. Geyskens, Inge, Jan-Benedict E. M. Steenkamp, and Nirmalya Kumar (1999), вЂњA Meta-Analysis of Satisfaction in Marketing Channel Relationships,вЂќ Journal of Marketing Research, 36 (May), 223вЂ“38. 38 Grzybowski, Lukasz (2005), вЂњRegulation of Mobile Telephony across the European Union: An Empirical Analysis,вЂќ Journal of Regulatory Economics, 28 (1), 47вЂ“67. Gustafsson, Anders, Michael D. Johnson, and Inger Roos (2005), вЂњThe Effects of Customer Satisfaction, Relationship Commitment Dimensions, and Triggers on Customer Retention,вЂќ Journal of Marketing, 69 (October), 210вЂ“18. Gwinner, Kevin P., Dwayne D. Gremler, and Mary Jo Bitner (1998), вЂњRelational Benefits in Services Industries: The CustomerвЂ™s Perspective,вЂќ Journal of the Academy of Marketing Science, 26 (Spring), 101вЂ“14. Harris, Jennifer and Mark Uncles (2007), вЂњModeling the Repatronage Behavior of Business Airline Travelers,вЂќ Journal of Service Research, 9 (4), 297вЂ“311. Johnson, Eric J., Wendy W. Moe, Peter S. Fader, Steven Bellman, and Gerald L. Lohse (2004), вЂњOn the Depth and Dynamics of Online Search Behavior,вЂќ Management Science, 50 (3), 299вЂ“308. Jones, Michael A., David L. Mothersbaugh, and Sharon E. Beatty (2000), вЂњSwitching Barriers and Repurchase Intentions in Services,вЂќ Journal of Retailing, 76 (2), 259вЂ“274. Jones, Michael A., David L. Mothersbaugh, and Sharon E. Beatty (2002), вЂњWhy Customers Stay: Measuring the Underlying Dimensions of Services Switching Costs and Managing their Differential Strategic Outcomes,вЂќ Journal of Business Research, 55, 441вЂ“450. Jones, Michael A., Kristy E. Reynolds, David L. Mothersbaugh, and Sharon E. Beatty (2007), вЂњThe Positive and Negative Effects of Switching Costs on Relational Outcomes,вЂќ Journal of Service Research, 9 (4), 335вЂ“355. Jones, Thomas O., y W Earl Sasser Jr. (1995), вЂњWhy Satisfied Customers Defect,вЂќ Harvard Business Review, 73 (6), 88вЂ“100. Kalyanaram, Gurumurthy and Russell S. Winer (1995), вЂњEmpirical Generalizations from Reference Price Research,вЂќ Marketing Science, 14 (3), G161вЂ“69. Kamakura, Wagner A., Michel Wedel, Fernando de Rosa, and Jose Afonso Mazzon (2003), вЂњCross-Selling through Database Marketing: A Mixed Data Factor Analyzer for Data Augmentation and Prediction,вЂќ International Journal of Research in Marketing, 20, 45вЂ“65. Keaveney, Susan M. (1995), вЂњCustomer Switching Behavior in Service Industries: An Exploratory Study,вЂќ Journal of Marketing, 59 (April), 71вЂ“82. Keaveney, Susan M. and Madhavan Parthasarathy (2001), вЂњCustomer Switching Behavior in Online Services: An Exploratory Study of the Role of Selected Attitudinal, Behavioral, and Demographic Factors,вЂќ Journal of the Academy of Marketing Science, 29 (Fall), 374вЂ“90. Klemperer, Paul (1995), вЂњCompetition when Consumers Have Switching Costs: An Overview with Applications to Industrial Organization, Macroeconomics, and International Trade,вЂќ Review of Economic Studies, 62, 515вЂ“39. 39 Kordupleski, Ray, Roland T. Rust, and Anthony J. Zahorik (1993), вЂњWhy Improving Quality DoesnвЂ™t Improve Quality (Or Whatever Happened to Marketing?),вЂќ California Management Review, 35 (2), 82вЂ“95. Lam, Shun Yin, Venkatesh Shankar, M. Krishna Erramilli, and Bvsan Murthy (2004), вЂњCustomer Value, Satisfaction, Loyalty, and Switching Costs: An Illustration from a Business-to-Business Service Context.вЂќ Journal of the Academy of Marketing Science, 32 (Summer), 293вЂ“311. Lee, Jongsu, Yeonbae Kim, Jeong-Dong Lee, and Yuri Park (2006), вЂњEstimating the Extent of Potential Competition in the Korean Mobile Telecommunications Market: Switching Costs and Number Portability,вЂќ International Journal of Industrial Organization, 24, 107вЂ“124. Liu, Yuping (2007), вЂњThe Long-Term Impact of Loyalty Programs on Consumer Purchase Behavior and Loyalty,вЂќ Journal of Marketing, 71 (October), 19вЂ“35. Manchanda, Puneet, Jean-Pierre DubГ©, Khim Yong Goh, and Pradeep K. Chintagunta (2006), вЂњThe Effect of Banner Advertising on Internet Purchasing,вЂќ Journal of Marketing Research, 43 (February), 98вЂ“ 108. McCloughan, Patrick and Sean Lyons (2006), вЂњAccounting for ARPU: New Evidence from International Panel Data,вЂќ Telecommunications Policy, Vol. 30, 521вЂ“532. McCracken, Grant (1986), вЂњCulture and Consumption: A Theoretical Account of the Structure and Movement of the Cultural Meaning of Consumer Goods,вЂќ Journal of Consumer Research, 13 (June), 71вЂ“84. McFadden, Daniel (1974), вЂњConditional Logit Analysis of Qualitative Choice Behavior,вЂќ in Frontiers in Econometrics, P.Zarembka, eds. Academic Press, NY. Mittal, Vikas and Wagner A. Kamakura (2001), вЂњSatisfaction, Repurchase Intent, and Repurchase Behavior: Investigating the Moderating Effect of Customer Characteristics,вЂќ Journal of Marketing Research, 38 (February), 131вЂ“42. Morgan, Robert M. and Shelby D. Hunt (1994), вЂњThe Commitment-Trust Theory of Relationship Marketing,вЂќ Journal of Marketing, 58 (July), 20вЂ“38. Ongena, Steven and David C. Smith (2001), вЂњThe Duration of Bank Relationships,вЂќ Journal of Financial Economics, 61 (3), 449вЂ“75. Patterson, Paul G. and Tasman Smith (2003), вЂњA Cross-Cultural Study of Switching Barriers and Propensity to Stay With Service Providers,вЂќ Journal of Retailing, 79 (Summer), 107вЂ“120. Porter, Michael E. (1980), Competitive Strategy: Techniques for Analyzing Industries and Competitors. New York: Free Press. Prins, Remco and Peter C. Verhoef (2007), вЂњMarketing Communication Drivers of Adoption Timing of a New E-service among Existing Customers,вЂќ Journal of Marketing, 71 (April), 169вЂ“83. Rao, Akshay R. and Kent B. Monroe (1989), вЂњThe Effect of Price, Brand Name, and Store Name on Buyers' Perceptions of Product Quality: An Integrative Review,вЂќ Journal of Marketing Research, 26 (August), 351вЂ“57. 40 Reinartz, Werner J. and V. Kumar (2000), вЂњOn the Profitability of Long-Life Customers in a Noncontractual Setting: An Empirical Investigation and Implications for Marketing,вЂќ Journal of Marketing, 64 (October), 17вЂ“35. Reinartz, Werner J. and V. Kumar (2003), вЂњThe Impact of Customer Relationship Characteristics on Profitable Lifetime Duration,вЂќ Journal of Marketing, 67 (January), 77вЂ“99. Roos, Inger (1999), вЂњSwitching Processes in Customer Relationships,вЂќ Journal of Service Research, 2 (1), 68вЂ“85. Roos, Inger, Bo Edvardsson, and Anders Gustafsson (2004), вЂњCustomer Switching Patterns in Competitive and Noncompetitive Service Industries,вЂќ Journal of Service Research, 6 (3), 256вЂ“71. Rossi, Peter E. and Greg M. Allenby (2003), вЂњBayesian Statistics and Marketing,вЂќ Marketing Science, 22 (Summer), 304вЂ“28. Rust, Roland T. and Peter C. Verhoef (2005), вЂњOptimizing the Marketing Interventions Mix in Intermediate-Term CRM,вЂќ Marketing Science, 24 (3), 477вЂ“89. Rust, Roland T. and Anthony J. Zahorik (1993), вЂњCustomer Satisfaction, Customer Retention, and Market Share,вЂќ Journal of Retailing, 69 (2), 193вЂ“215. Rust, Roland T., Anthony J. Zahorik, and Timothy L. Keiningham (1995), вЂњReturn on Quality (ROQ): Making Service Quality Financially Accountable,вЂќ Journal of Marketing, 59 (April), 58вЂ“70. Rust, Roland T., Valarie A. Zeithaml, and Katherine N. Lemon (2000), Driving Customer Equity: How Customer Lifetime Value Is Reshaping Corporate Strategy. New York: The Free Press. Shapiro, Carl and Hal R. Varian (1999), Information Rules: A Strategic Guide to the Network Economy. Harvard Business School Press: Boston, MA. Sharma, Neeru and Paul G. Patterson (2000), вЂњSwitching Costs, Alternative Attractiveness and Experience as Moderators of Relationship Commitment in Professional, Consumer Services,вЂќ International Journal of Service Industry Management, 11 (5), 470вЂ“490. Sharp, Byron and Anne Sharp (1997), вЂњLoyalty Programs and their Impact on Repeat-Purchase Loyalty Patterns,вЂќ International Journal of Research in Marketing, 14 (5), 473вЂ“86. Sheth, Jagdish N. and Atul Parvatlyar (1995), вЂњRelationship Marketing in Consumer Markets: Antecedents and Consequences,вЂќ Journal of the Academy of Marketing Science, 23 (4), 255вЂ“271 Shugan, Steven M. (1980), вЂњThe Costs of Thinking,вЂќ Journal of Consumer Research, 7 (September), 99вЂ“ 111. Shum, Matthew (2004), вЂњDoes Advertising Overcome Brand Loyalty? Evidence from the Breakfast-Cereals Market,вЂќ Journal of Economics and Management Strategy, 13 (2), 241вЂ“72. Shy, Oz (2002), вЂњA Quick-and-Easy Method for Estimating Switching Costs,вЂќ International Journal of Industrial Organization, 20 (1), 71вЂ“87. Smith, Michael D. and Erik Brynjolfsson (2001), вЂњConsumer Decision-making at an Internet Shopbot: Brand Still Matters,вЂќ The Journal of Industrial Economics, 49 (December), 541вЂ“58. 41 Snijders, Tom A. B. and Roel J. Bosker (1999), Multilevel Analysis: An Introduction to Basic and Advanced Multilevel Modeling. Sage, London. Solgaard, Hans S. and Torben Hansen (2003), вЂњA Hierarchical Bayes Model of Choice between Supermarket Formats,вЂќ Journal of Retailing and Consumer Services, 10 (3), 169вЂ“80. Spiegelhalter, David J., Andrew Thomas, Nicky G. Best, and Dave Lunn (2003), вЂњWinBUGS Version 1.4 User Manual,вЂќ [available at: http://www.mrc-bsu.cam.ac.uk/bugs/winbugs/manual14.pdf]. Szymanski, David M. and David H. Henard (2001), вЂњCustomer Satisfaction: A Meta-Analysis of the Empirical Evidence,вЂќ Journal of the Academy of Marketing Science, 29 (1), 16вЂ“35. Train, Kenneth E. (2003), Discrete Choice Methods with Simulation. Cambridge University Press. Verhoef, Peter C. (2003), вЂњUnderstanding the Effect of Customer Relationship Management Efforts on Customer Retention and Customer Share Development,вЂќ Journal of Marketing, 67 (October), 30вЂ“45. Verhoef, Peter C., Philip Hans Franses and Janny C. Hoekstra (2001), вЂњThe Impact of Satisfaction and Payment Equity on Cross-Buying: A Dynamic Model for a Multi-Service Provider,вЂќ Journal of Retailing, 77, 359вЂ“378. Verhoef, Peter C., Philip Hans Franses and Janny C. Hoekstra (2002), вЂњThe Effect of Relational Constructs on Customer Referrals and Number of Services Purchased: Does Age of the Relationship Matter?вЂќ Journal of the Academy of Marketing Science, 30 (Summer), 202вЂ“16. Watson, Richard T., Leyland F. Pitt, Pierre Berthon, and George M. Zinkhan (2002), вЂњU-Commerce: Expanding the Universe of Marketing,вЂќ Journal of the Academy of Marketing Science, 30 (Fall), 333вЂ“ 347. Wathne, Kenneth H., Harald Biong and Jan B. Heide (2001), вЂњChoice of Supplier in Embedded Markets: Relationship and Marketing Program Effects,вЂќ Journal of Marketing, 65 (April), 54вЂ“66. Wieringa, Jaap E. and Peter C. Verhoef (2007), вЂњUnderstanding Customer Switching Behavior in a Liberalizing Service Market: An Exploratory Study,вЂќ Journal of Service Research, 10 (2), 174вЂ“186. Wirtz, Jochen, Anna S. Mattila, and May Oo Lwin (2007), вЂњHow Effective Are Loyalty Reward Programs in Driving Share of Wallet?вЂќ Journal of Service Research, 9 (4), 327вЂ“334.