Measuring the efficiency of customer satisfaction and loyalty for mobile phone brands with DEA

Expert Systems with Applications 39 (2012) 99–106

Contents lists available at ScienceDirect

Expert Systems with Applications journal homepage: www.elsevier.com/locate/eswa

Measuring the efficiency of customer satisfaction and loyalty for mobile phone brands with DEA Erkan Bayraktar a, Ekrem Tatoglu b, Ali Turkyilmaz c, Dursun Delen d,⇑, Selim Zaim e a

Bahcesehir University, Department of Industrial Engineering, Besiktas, Istanbul 34349, Turkey Bahcesehir University, International Trade and Business, Besiktas, Istanbul 34349, Turkey c Fatih University, Department of Industrial Engineering, Istanbul 34500, Turkey d Oklahoma State University, Department of MSIS, Tulsa, OK 74106, USA e Fatih University, Department of Management, Istanbul 34500, Turkey b

a r t i c l e

i n f o

Keywords: Customer satisfaction Customer loyalty DEA Mobile phone sector Turkey

a b s t r a c t The concept of customer satisfaction and loyalty (CS&L) has attracted much attention in recent years. A key motivation for the fast growing emphasis on CS&L can be attributed to the fact that higher customer satisfaction and loyalty can lead to stronger competitive position resulting in larger market share and profitability. Using a data envelopment analysis (DEA) approach, in this study we analyzed and compared CS&L efficiency for mobile phone brands in an emerging telecommunication market, Turkey. The constructs of European Customer Satisfaction Index (ECSI) model are treated and used as input and output indicators of our DEA model. Drawing on the perceptual responses of 251 mobile phone users, the DEA models reveal that from the top six mobile phone brands in Turkey, Nokia features as the most efficient brand followed by LG and Sonny Ericsson in terms of CS&L efficiency, while Motorola, Samsung and Panasonic rank as the least efficient brands. Ó 2011 Elsevier Ltd. All rights reserved.

1. Introduction Over the past two decades, organizations of all types have increasingly acknowledged the importance of customer satisfaction and loyalty. The marketing literature suggests that the long term success of a firm is clearly based on its ability to rapidly respond to changing customer needs and preferences (Narver & Slater, 1990; Webster, 1992). A key motivation for the increasing emphasis on customer satisfaction is that higher customer satisfaction can lead to have a stronger competitive position resulting in higher market share and profitability (Fornell, 1992), reduced price elasticity, lower business cost, reduced failure cost, and mitigated cost of attracting new customers (Chien, Chang, & Su, 2003). The principal focus of this study is on evaluating the efficiency of customer satisfaction and loyalty (CS&L) for existing mobile phone brands in Turkish mobile phone sector. Since the early 1990s, with the launch of the mobile phones, there has been a remarkable development both in their product sophistication and their rapid and widespread adoption. With more than three billion subscribers around the world, the extent of mobile phone diffusion in emerging markets has been increasingly larger than that in developed countries (Kalba, 2008). Turkey, being one of the fastest emerging market economies in the world, adopted mobile phone ⇑ Corresponding author. Tel./fax: +1 (918) 594 8283. E-mail address: [email protected] (D. Delen). 0957-4174/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.eswa.2011.06.041

technology in 1994. Since then, there has been a considerable increase in the level of mobile phone ownership, where the number of mobile phone users in the country is expected to reach around 70 million by the end of 2013, representing a penetration rate of over 90% (RNCOS, 2010). The significant rise in mobile phone usage can partially be attributed to the fact that Turkey has the youngest population in Western Europe. Turkey currently has the 6th largest young mobile phone user base in the world, with more than 11 million subscribers under the age of 25, providing a very lucrative market for mobile phone companies (Euromonitor International, 2010). It should however be noted that the penetration in this market at present is still below the EU average, indicating that the mobile phone sector is not saturated yet, and there is still space for new investors. Currently, there exist nearly more than 10 major mobile phone companies operating in the Turkish mobile phone sector, each having a relatively large product line. As of 2010, the top five mobile phone brands were Nokia, Samsung, LG, Motorola and Sony Ericsson and together they account for nearly 75% of overall market sales. As a new comer, iPhone is rapidly increasing its market share, but as of the start of this study, did not have a significantly large presence. In terms of market share, Nokia has been undisputedly the market leader (36.4% of sales) with Samsung featuring second (19.5%) and LG ranking third (10.1%) (Patron Turk, 2010). Commensurate to its widespread diffusion globally, there has been a growing worldwide academic interest in mobile phone

100

E. Bayraktar et al. / Expert Systems with Applications 39 (2012) 99–106

usage which focuses mainly on examining its contribution to social life, user preferences and its ergonomic features (Bagchi, Kirs, & Lopez, 2008). A number of empirical studies were also conducted within the context of Turkish mobile phone sector. The topics of these studies ranged from examining motivation of use (Dedeoglu, 2004; Ozcan & Kocak, 2003) to mobile phone selection (Isiklar & Buyukozkan, 2007), from customer satisfaction (Turkyilmaz & Ozkan, 2007) to brand loyalty (Simsek & Noyan, 2009). The methodology used in study to evaluate the relative CS&L efficiency of mobile phone brands is based on data envelopment analysis (DEA). The traditional DEA technique has long been utilized as an invaluable tool in the field of operations research and management science to solve problems in wide range of industries (Hu, Lai, & Huang, 2009; Lee, 2009; Lin, Lee, & Chiu, 2009) as well as in not-for-profit organizations (Mahajan, 1991; Wu, Liang, & Chen, 2009; Zhang, Huang, Lin, & Yu, 2009); but its diffusion into the field of marketing and related disciplines has been relatively slow. For instance, in the marketing field, DEA has recently been employed as a powerful tool for data analysis in measuring efficiency in retailing sector (Charnes, Cooper, Learner, & Phillips, 1985; Donthu & Yoo, 1998; Keh, 2000; Keh & Chu, 2003; Thomas, Barr, Cron, & Slocum, 1998), evaluating website marketing efficiency (Shuai & Wu, 2011), benchmarking marketing productivity (Donthu, Hershberger, & Osmonbekov, 2005; Kamakura, Ratchford, & Agrawal, 1988), and measuring relative market efficiency (Murthi, Srinivasan, & Kalyanaram, 1996) or service quality (Athanassopoulos, 1997; Soteriou & Staurinides, 1997). The assessment of CS&L has always been a major research item on the agenda of researchers in the marketing and related fields, because the issue of how efficiently a firm manages its marketing processes and their relationship with their customers is central to its ability to gain competitive edge vis-à-vis its rivals. The DEA approach adopted in this study illustrates how differences in CS&L efficiency between various mobile phone brands can be ascertained empirically, and thus helps management determine proper policies and courses of action. The rest of the paper is organized as follows. Section 2 reviews the recent literature on customer satisfaction and customer loyalty studies. Section 3 provides an in-depth description of our research methodology. Section 4 presents the results of our analysis. The last section (Section 5) summarizes our findings, describes managerial implications of the study and provides the concluding remarks.

2. Background literature While customer satisfaction has been defined in various ways, the high-level conceptualization that appears to have gained the widest acceptance states that satisfaction is a customer’s postpurchase evaluation of a product or service (Cronin & Taylor, 1992; Westbrook & Oliver, 1991). Customer satisfaction is also generally assumed to be a significant determinant of repeat sales, positive word-of-mouth, and customer loyalty. It has also long been considered as one of the key antecedents of creating brand loyalty (Cronin, Brady, & Hult, 2000; Dick & Basu, 1994; Fornell, Michael, Eugene, Jaesung, & Barbara, 1996; Syzmanski & Henard, 2001). Satisfied customers return and buy more, and they tell other people about their experiences, both positive and negative (Fornell et al., 1996). Building on Hirschman’s (1970) exit-voice theory, weakly dissatisfied consumers would be of primary importance to a firm. While strongly dissatisfied consumers generally choose the exit option (i.e., they leave the firm), the weakly dissatisfied customers tend to stay loyal to the firm and rather employ the voice option, which implies overt complaints as an attempt to change the firm’s

practices or offerings (Fornell & Wernerfelt, 1988). Thereby, proper handling of customer complaints may ensure that weakly dissatisfied consumers remain loyal, and serve as an exit barrier (Fornell, 1992; Halstead & Page, 1992). The impact of loyal customers is considerable; for many industries the profitability of a firm increases proportionally with the number of loyal customers and up to 60% of sales to new customers can be attributed to the word of mouth referrals (Reichheld & Sasser, 1990). Within the existing literature on customer satisfaction research, various customer satisfaction models were developed based on a cumulative view of satisfaction. To this end, a number of customer satisfaction indices (CSIs) were designed with most prominent of those being Swedish Customer Satisfaction Barometer (SCSB), the American Customer Satisfaction Index (ACSI) and European Customer Satisfaction Index (ECSI). Of these CSIs, we employed the ECSI model as the backbone of our CS&L efficiency model in this study due to its recent popularity in the literature and its comprehensiveness in CS&L coverage. The ECSI is a structural model based on the assumptions that customer satisfaction is derived by a number of factors such as perceived quality, perceived value, expectations of customers, and image of a firm. These factors are the antecedents of overall customer satisfaction (Turkyilmaz & Ozkan, 2007). The model also estimates the results when a customer is satisfied or not. The four antecedents of customer satisfaction may also have direct effects on customer loyalty (Johnson, Gustafsson, Andreassen, Lervik, & Cha, 2001). Each construct in the ECSI model is a latent construct which is operationalized by multiple indicators (Chien et al., 2003; Fornell, 1992). The underlying constructs of the ECSI model are explained as follows: The image construct evaluates the underlying image of the company. Image refers to the brand name and the kind of associations customers obtain from the product/company (Andreassen & Lindestad, 1998). Martensen, Kristensen, and Grønholdt (2000) argue that image is an important dimension of the customer satisfaction model. Image is a consequence of being reliable, professional and innovative, having contributions to society, and adding prestige to its user. It is anticipated that image has a positive effect on customer satisfaction, customer expectations and customer loyalty. Customer expectations are the consequences of prior experience with the company’s products (Rotondaro, 2002). This construct evaluates customer expectations for overall quality, for product and service quality, and for fulfillment of personal needs. The customer expectations construct is expected to have a direct and positive relationship with customer satisfaction (Anderson, Fornell, & Lehmann, 1994). Perceived quality is evaluation of recent consumption experience by the market served. This construct evaluates customization and reliability of a given product or service. Customization is the degree to which a product or service meets a customer’s requirements, and reliability is the degree to which firm’s offering is reliable, standardized, and free from deficiencies. Perceived quality is expected to have a positive effect on customer satisfaction (Fornell et al., 1996). Perceived value is the perceived level of product quality relative to the price paid by customers. Perceived value is the rating of the price paid for the quality perceived and a rating of the quality perceived for the price paid (Fornell et al., 1996). Perceived value structure provides an opportunity for comparison of the firms according their price-value ratio (Anderson et al., 1994). In the model, perceived value is expected to have a positive impact on satisfaction. Customer satisfaction construct indicates how much customers are satisfied, and how well their expectations are fulfilled. This construct evaluates overall satisfaction level of customers,

101

E. Bayraktar et al. / Expert Systems with Applications 39 (2012) 99–106

Fig. 1. The DEA model of CS&L.

fulfillment of their expectations, and company’s performance versus the ideal provider. Customer loyalty is the ultimate factor in the ECSI model. Loyalty is measured by repurchase intention, price tolerance and intention to recommend products or services to others. It is expected that better image and higher customer satisfaction should increase customer loyalty. 3. Methodology This section presents the research methodology adopted in this study. The following subsections explain the survey instrument, the data collection procedure, and the DEA model. 3.1. Survey instrument The DEA model of CS&L, which is shown in Fig. 1, consists of the aforementioned constructs which are based on previous research and prominent theories in the field of consumer behavior. The constructs of the CS&L model are unobservable (latent) variables indirectly described by a set of observable variables which are called manifest variables or indicators. The constructs and their constituent items are shown in Table 1. The use of multiple measures for each construct increases the precision of the estimate as compared to an approach of relying on a single measure. In our CS&L efficiency model, all four antecedents of customer satisfaction and loyalty which include image, customer expectations, perceived quality and perceived value were treated as input variables, while the two constructs, namely customer satisfaction and customer loyalty were considered as output variables. The survey questionnaire was designed using a three-step process. First, the consumer behavior literature was extensively reviewed for the manifest variables. Secondly, the questionnaire items were prepared in Turkish and refined through a series of discussions with two senior marketing managers of a prominent mobile phone company and a number of experienced academics in the field of consumer behavior. Finally, the survey questionnaire was

subjected to extensive pre-testing and refinement based on a pilot study of 30 mobile phone users. Feedback from this pilot study indicated that some questions were ambiguous, difficult to understand, or irrelevant for mobile phone sector. This pilot study also served as a practical exercise for interviewers. The final questionnaire contained a total of 23 items pertaining to the CS&L. These 23 items appeared to have face validity as to what should be measured. All the items were measured on 10-point scales, with anchors ranging from 1 denoting a very negative view and 10 indicating a very positive view. Relying on 10-point scales enables customers to make better discriminations (Andrews, 1984). Table 1 The constructs and their constituent items in the CS&L model. Constructs

Items

Image (IM)

IM1: IM2: IM3: IM4: IM5: IM6:

Expectations (EXP)

EXP1: need EXP2: EXP3: EXP4:

Being reliable Being professional Social contributions to society Customer relations Innovative and forward looking Adding value to user (prestige) Expectations for fulfillment of personal Expectations for overall quality Expectations for product quality Expectations for service quality

Perceived quality (PQ)

PQ1: PQ2: PQ3: PQ4: PQ5:

Overall quality Product quality (technical) Service quality Customer services Appropriateness to intent of use

Perceived value (PV)

PV1: Price/performance PV2: Performance/price

Customer loyalty (CL)

CL1: Repurchase intention CL2: Recommendation to others CL3: Price tolerance

Customer satisfaction (CS)

CSI1: Overall satisfaction CSI2: Fulfillment of expectations CSI3: Compare with ideal

102

E. Bayraktar et al. / Expert Systems with Applications 39 (2012) 99–106

Besides the model items, some demographic variables such as age, gender, education level, etc. were also included in the questionnaire. 3.2. Sample and data collection The CS&L model was implemented in Turkish mobile phone sector, where the fierce competition within the industry resulted in a dynamic product innovation to satisfy rapidly increasing demand for new products and product features. As noted earlier, the mobile phone sector in Turkey is characterized by a high degree of concentration. There are currently not more than 10 major mobile phone companies, where the top five brands hold over 75% of the total sales. Data were gathered based on street interception type personal interviews. Although the sampling frame for this study included all mobile phone users in Turkey, the sample was drawn from the greater metropolitan region of Istanbul. As the survey setting with around 15 million inhabitants, Istanbul metro-region has become a mega-city, ranking eighth out of 78 OECD metro-regions in terms of population size. Istanbul is also the leading economic epicenter in Turkey and has established itself as the industrial, financial and logistics center of the country, producing almost one-third of the national output and absorbing the bulk of foreign direct investment (OECD, 2008). A total of six main districts within Istanbul metro-region were identified as a basis for selecting subjects. These six districts were chosen to minimize the potential bias that might be present in data drawn from a single territory or a district. The survey was conducted in major traffic nodes and busy shopping areas. Over a one-week fieldwork, a total of 282 usable questionnaires were collected and used for this study. Data concerning the study’s main constructs were compared and contrasted with respect to the demographic characteristics of the respondents (e.g. age, gender and education level) in order to ensure that there is no systematic variation between the study’s constructs and the demographic profile of the respondents, and showed no clear differences (p > 0.1). Our sample respondents were categorized in terms of their choice of a particular mobile phone brand. While a total of 10 mobile phone brands were identified, four brands were eliminated due to small sample size. Our sample finally included a total of 251 questionnaires pertaining to top six mobile phone brands, which constituted the subject of our data analysis in terms of CS&L efficiency comparison. These brands were namely Nokia, Motorola, Panasonic, Samsung, LG and Sony Ericsson holding together over eighty per cent of the Turkish mobile phone market. 3.3. The DEA model DEA is a linear programming based model which was first proposed by Charnes, Cooper, and Rhodes (1978) 20 years after Farrell’s seminal work for evaluating activities of not-for-profit entities participating in public programs (Farrell, 1957). Since then, a wide variety of DEA applications have been successfully executed for evaluating the performances of various kinds of entities engaged in many different activities in many different contexts in many different countries (Cooper, Huang, & Li, 2004). DEA can be used to assess the comparative efficiency of homogeneous multiinput multi-output organizational units, such as bank branches, schools, airports, tax offices, and hospitals (Cook & Seiford, 2009; Thanassoulis, 1999). The efficiency score is usually denoted by either a number between 0 and 1 or 0 and 100%. The efficiency score of 1 or 100% of a decision making unit (DMU) shows that DMU is efficient relative to other units in the research sample. In addition to providing meaningful scalar efficiency values, DEA is designed to determine the sources of inefficiencies and estimate

their amounts that might be present in the various output and input factors (Charnes et al., 1978). Since DEA identifies inefficiencies in DMUs by comparing them with similar DMUs regarded as efficient, it can be used as a valuable benchmarking tool (Avkıran, 2006). Unlike other benchmarking tools that rely mostly on the managers’ observations, DEA is capable of identifying best practices that are too complex to be accurately identified via observations (Sherman & Ladino, 1995). The most important advantage of DEA over other traditional econometric methods is that it does not require prior assumption about the analytical form of the production function (Banker, Charnes, & Cooper, 1984; Cooper et al., 2004; Lee, Lee, & Kang, 2005). On the other hand, the main problem with DEA is that, it is a non-parametric method making it sensitive to the measurement problems (Al-Sharkas, Hassan, & Lawrence, 2008). While the standard DEA model is assumed to have multiple incompatible input and output variables that are of a quantitative nature, it can also be applied to a number of wide variety of problem settings where qualitative data have been heavily used, including for not-for-profit oriented organizations and service firms (Cook, 2004; Lin, 2009; Zhu, 2003). In line with previous literature (Bayraktar, Koh, Tatoglu, Demirbag, & Zaim, 2010; Demirbag, Tatoglu, Glaister, & Zaim, 2010; Lin, Madu, Kuei, & Lu, 2004; Narasimhan, Talluri, & Das, 2004; Swink, Talluri, & Pandejpong, 2006), this study employs a DEA based methodology that rely on qualitative survey data obtained from mobile phone users. An output oriented envelopment model of DEA, known as CCR (Charnes et al., 1978) in the literature, is defined for each of the mobile phone users as follows:

Max

/bo þ e

n X

ebio þ

i¼1

Subject to

Ul B X X

m X

! b

djo

ð1Þ

j¼1

klu xliu þ ebio ¼ xbio ;

i ¼ 1; . . . ; n

ð2Þ

l¼1 u¼1 Ul B X X

b

klu ylju  djo ¼ /bo ybjo ;

j ¼ 1; . . . ; m

ð3Þ

l¼1 u¼1 b

ebio ; djo ; klu P 0;

for all i; j; u; l

ð4Þ

where /bo : Efficiency score of mobile phone brand b according to user o. xliu : Value of input i for the mobile phone brand l according to user u. ylju : Value of output j for the mobile phone brand l according to user u. b ebio ; djo : The amounts of excess input i and deficit output j for mobile phone brand b according to user o, respectively. e > 0: Predefined non-Archimedean element. klu : Dual variable utilized to construct a composite ideal mobile phone brand to dominate mobile phone brand under evaluation. B: The number of mobile phone brands. Ul: The number of mobile phone users of brand l. m: The number of outputs. n: The number of inputs. The objective function (1) assesses the efficiency score (/bo ) of the mobile phone brand b according to user o. The term related to the slacks in the objective function assures full-efficiency for the mobile phone brand found to be efficient by the user o (/bo ¼ 1), while enforcing all slack values (output deficits and input excesses) to be zero. Constraint (2) ensures that the input i for mobile phone brand b according to user o is a linear combination of the inputs for each mobile phone brand (l) users (u) and its surplus value. Constraint (3) states that the optimal output of j for a mobile

103

E. Bayraktar et al. / Expert Systems with Applications 39 (2012) 99–106

phone brand b according to user o is a linear combination of the outputs for each mobile phone brand (l) users (u) minus its shortage. In the optimal solution of model (1)–(4), mobile phone brand b b according to user o is efficient if /bo ¼ 1 and ebio ¼ djo ¼ 0 for all input excesses and output deficits (Cooper et al., 2004). If /bo ¼ 1 but b either ebio or djo is non-zero, the mobile phone brand b according to user o is called weakly efficient. The mobile phone brand found efficient in the solution of the model (1)–(4) forms the efficiency frontier which is called as reference set for mobile phone brand b P PU i i according to user o. If the constraint Bi¼1 u¼1 ku ¼ 1 is adjoined to the model (1)–(4), it may be called as output oriented BCC (Banker et al., 1984).

Table 2 Unidimensionality of the constructs.

4. Analysis and findings

DEA was run separately for each brand. The average input and output values for each mobile phone brand are shown in Table 3. The output-oriented BCC (BCC-O) efficiency scores on the last column of Table 3 also illustrate the mean value of brand efficiencies of how mobile phone users assess and perceive their own choice of a particular mobile phone brand. Higher mean values of brand efficiency refer to the homogeneity of particular mobile phone brand users, and indicate the consistency and reliability of the survey instrument. Among the mobile phone brands, Panasonic was found to have the highest efficiency score (0.936), and Panasonic users were highly homogeneous in the evaluation of the CS&L efficiency achieved with respect to the efforts on image, customer expectations, quality and value perceptions. In contrast, Motorola users were noted as the least homogeneous ones (0.857). High standard deviation (0.195) on the efficiency scores of Motorola also tends to support this finding. In the second step, mobile phone users who evaluated their brand as inefficient, were projected onto the efficiency frontier of their brand by adjusting the value of their input and output variables. Therefore, the assessment and perception differences among the particular group of users were removed. In the third step, a pooled DEA with all mobile phone users at their adjusted efficiency levels was conducted. Based on the pooled DEA, average efficiency score of each brand was calculated and shown in the second column of Table 4. In order to test whether there exist statistically significant differences among mobile phone brands in terms of the efficiency levels of CS&L, a Kruskal–Wallis rank test was applied in step 4 (Brockett & Golany, 1996; Sueyoshi & Aoki, 2001). A Kruskal–Wallis rank test statistic (KW) incorporating the effect of the ties was computed as follows (Conover, 1999; Sueyoshi & Aoki, 2001):

In DEA, the problem of discrimination may occur when the number of variables is relatively large compared to the number of observations. The exploratory factor analysis (EFA) was suggested as a solution to reduce the number of variables in DEA with minimum loss of information (Adler & Golany, 2001, 2002). Therefore, each construct in the ECSI model was subjected to EFA separately, which indicated that the items in each of the six constructs formed a single factor. In order to verify the unidimensionality and reliability of each construct in the model, one may choose to use of there potential techniques: principal component analysis of the construct, Cronbach’s alpha and Dillon–Goldstein’s q. A construct is essentially unidimensional if the first eigenvalue of the correlation matrix of the construct manifest variables is larger than 1 and the second one is smaller than 1, or at least very far from the first one. From principal component analysis, first eigenvalue was noted to be greater than 1 and second eigenvalue is less than 1 for each construct. A construct is also considered as unidimensional when Cronbach’s alpha and Dillon–Goldstein’s q values are close to or greater than 0.7 (Tenenhaus, Vinzi, Chatelin, & Lauro, 2005). For the data set, Cronbach’s alpha and Dillon–Goldstein’s q values of each construct were found to be greater than 0.80. These results lead to an acceptance of the unidimensionality of all constructs used in this study as shown in Table 2. Since DEA does not allow the use of negative numbers, the EFA values of the model constructs were transformed to scales with values ranging from 0 to 10 using a well-known normalization formula.1 These positive values constituted the new values of input and output indicators of our CS&L efficiency model. Through EFA, a total of four inputs and two outputs latent variables were identified for each mobile phone user. Among them, image (IM), expectation (EXP), perceived quality (PQ), and perceived value (PV) were considered as inputs, while customer satisfaction (CS) and customer loyalty (CL) were determined as outputs for the DEA in the evaluation of the CS&L efficiency of mobile phone brands. An output-oriented BCC model of DEA was selected to maximize the value of the outputs (CS&L) while keeping the level of inputs currently achieved. A practical technique to determining the efficiency differences among the groups of DMUs is proposed by Brockett and Golany (1996). The technique involves a four-step process called ‘‘the program evaluation procedure’’ to determine programmatic efficiency differences among the groups of DMUs sharing certain joint characteristics. A similar approach was also applied here to determine the efficiency differences of mobile phone brands relying on user perceptions. In the first step, the mobile phone users were grouped according to their choice of a particular mobile phone brand, and zzmin 1 Z ¼ ðzmax zmin Þ  10, where Z is new value and z refers the actual value in the latent variable. zmax and zmin represent the maximum and minimum values of the dataset.

Constructs

Number of Cronbach Dillon– First Second indicators alpha Goldstein q eigenvalue eigenvalue

Image Customer expectations Perceived quality Perceived value Customer satisfaction Customer loyalty

6 4

0.836 0.831

0.885 0.884

3.337 2.673

0.793 0.554

5

0.875

0.917

3.373

0.625

2 3

0.871 0.834

0.938 0.902

1.785 2.258

0.221 0.522

3

0.831

0.897

2.246

0.548

h #

KW ¼

12 UðUþ1Þ

PB

R2i i¼1 U i

i

 3ðU þ 1Þ  PB 3  ðsi  si Þ 1  i¼1 U 3 U

In Eq. (5), Ri represents the rank sum of the ith mobile phone brand, where there are B brands under consideration, and U total number of mobile phone users participating in the survey. Ui of U participants uses ith mobile phone brand, and si among Ui users has same efficiency scores (tied scores). Conover (1999) states that exact distribution of KW is too cumbersome to work with, and the chi-squared distribution with B  1 degrees of freedom approximates well enough to the null distribution (H0) of KW as long as Ui > 5. According to the pooled DEA results of the mobile phone users calculated in the earlier step, overall efficiency scores were ranked in an increasing order. The smallest efficiency score was ranked by 1. Then, chi-squared test statistic and associated p-value of the test along with mean ranks of each brand were calculated, and shown in the last three columns of Table 4. The null hypothesis, H0, that there is no difference among all the mobile phone brand efficiencies, is rejected (p-value = 0.000).

104

E. Bayraktar et al. / Expert Systems with Applications 39 (2012) 99–106

Table 3 Average input and output values along with BCC-O individual brand efficiencies.a Brand name

LG Motorola Nokia Panasonic Samsung Sony Ericsson a b

Inputs

BCC-O efficiency scoresb

Outputs

Image

Customer expectations

Perceived quality

Perceived value

Customer satisfaction

Customer loyalty

6.435 (1.541) 5.036 (1.051)

7.0697 (1.765) 5.577 (1.749)

7.203 (1.733) 5.627 (1.629)

6.718 (2.226) 4.858 (2.332)

6.863 (1.812) 4.539 (2.064)

6.809 (2.151) 4.011 (1.782)

0.928 (0.088) 0.857 (0.195)

6.961 (1.272) 5.209 (1.666)

7.50 (1.431) 6.197 (1.608)

7.219 (1.393) 6.124 (1.449)

6.027 (2.132) 5.867 (2.211)

6.534 (1.499) 5.766 (1.555)

7.297 (2.220) 4.939 (2.149)

0.887 (0.114) 0.936 (0.096)

5.318 (1.845) 6.516 (1.616)

6.702 (2.225) 7.127 (1.585)

6.482 (1.741) 7.020 (1.520)

5.328 (2.362) 6.375 (1.834)

5.615 (1.764) 6.477 (1.686)

4.858 (2.266) 6.253 (2.109)

0.917 (0.094) 0.871 (0.135)

The numbers within parentheses indicate the standard deviation of each item. Geometric mean of the group-wise efficiencies of the mobile phone brands evaluated by their users.

Table 4 Overall efficiency scores and Kruskal–Wallis rank test results. Brand name

Overall efficiency scoresa

Mean ranks (Ri/ui)

KW

p-Value

LG Motorola Nokia Panasonic Samsung Sony Ericsson

0.943 0.853 0.980 0.882 0.892 0.938

117.78 73.85 172.72 85.12 80.96 114.07

68.295

0.000

a Geometric mean of the overall efficiency scores of all the projected mobile phone brands onto the efficiency frontier.

Table 5 Rank sum test results for multiple comparisons between mobile phone brands. Mobile phone brands (i)

Mean rank (Ri)

Mobile phone brands (j)

Mean rank (Rj)

KW#

Nokia Nokia Nokia Nokia Nokia LG LG LG LG Sony Ericsson Sony Ericsson Sony Ericsson Panasonic Panasonic Samsung

172.72 172.72 172.72 172.72 172.72 117.78 117.78 117.78 117.78 114.07 114.07 114.07 85.12 85.12 80.96

LG Sony Ericsson Motorola Samsung Panasonic Sony Ericsson Motorola Samsung Panasonic Motorola Samsung Panasonic Samsung Motorola Motorola

117.78 114.07 73.85 80.96 85.12 114.07 73.85 80.96 85.12 73.85 80.96 85.12 80.96 73.85 73.85

4.768*** 5.202*** 6.390*** 6.583*** 6.190*** 0.283 2.606*** 2.381** 2.086** 2.409** 2.167** 1.871* 0.237 0.601 0.382

The null hypothesis (H0), that there is no efficiency difference between mobile phone brand i and j is rejected, if KW# > t1a/ 2,UB where t1a/2 represents t-distribution test statistic at both sided a significance level. In Table 5, multiple comparison rank sum test results between mobile phone brands were listed. Nokia, which has the highest mean rank, was also found statistically different (p < 0.01) from the other mobile phone brands. Based on these findings, it would not be incorrect to infer that Nokia is the best mobile phone brand in terms of the CS&L efficiency. Next to Nokia, the performance levels of LG and Sony Ericsson on CS&L were significantly much better (p < 0.01) than the other brands. This would put both brands to the second and third ranks in the list, respectively. Finally, pair-wise comparisons among Panasonic, Samsung, and Motorola brands in rank sum test results produced no significant differences (p > 0.1). Their efficiencies on CS&L were not significantly different from each other and the worst among the six brands compared. In addition to ranking the mobile phone brands in terms of their CS&L efficiency, DEA provides further insights about the sources of inefficiencies for each mobile phone brand. The amount of inefficiency is determined by the magnitude of excess resources (inputs) and/or deficient outputs produced. Excess inputs or deficient outputs are calculated by subtracting the actual input/output values of a given mobile phone user from the ideal values of the composite (best practice) user. Input excesses and output deficits for mobile phone brand according to the brand’s users are calculated from the brand specific DEA model solved at the first step of our calculations. These excesses and deficits indicate the average inefficiencies resulting from the differences on brand users’ perceptions and evaluation errors, and are shown in Table 6. The Panasonic and LG

Table 6 Average brand inefficiencies with respect to brand’s users.

*

p < 0.10. p < 0.05. *** p < 0.01 (two-tailed). **

Brands LG

Since the mobile phone brands were found to have statistically significant differences in terms of efficiency levels of CS&L achieved, the following test statistic (KW#) was calculated to identify the different pair(s) (Conover, 1999):

   Ri Rj  Ui  Uj  ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi KW # ¼ r  ffi UðUþ1ÞðU1KWÞ 1 1  U U 12ðUBÞ i

j

ð6Þ

Input excesses (%) Image Customer expectations Perceived quality Perceived value Output deficits (%) Customer satisfaction Customer loyalty

Motorola Nokia Panasonic Samsung Sony Ericsson

2.53 1.22

5.24 6.18

8.77 4.26

4.00 2.41

2.73 3.51

5.06 0.31

1.05 0.63

2.45 5.53

5.06 8.48

3.79 1.42

0.55 3.67

2.56 7.19

9.83 28.93

15.08

9.44

9.94

16.53

38.73 14.28

26.64

19.52

14.83 32.92

E. Bayraktar et al. / Expert Systems with Applications 39 (2012) 99–106 Table 7 Average relative inefficiencies for mobile phone brands. Brands LG Input excesses (%) Image Customer expectations Perceived quality Perceived value Output deficits (%) Customer satisfaction Customer loyalty

Motorola Nokia Panasonic Samsung Sony Ericsson

3.82 1.02 0.90 1.26

4.59 0.41

4.85 2.11

0.77 0.24

1.98 0.91

0.00 0.19 1.55 1.65

0.47 0.59

1.84 0.84

1.68 1.26

0.12 1.06

19.68 6.27

2.20

6.90

12.67

14.15

36.00 8.15

3.18

9.86

21.55

19.55

brand has lower input excesses and output deficits relative to the other brands. Therefore, it appears reasonable to state that the Panasonic and LG users in our sample have a stronger consensus in their evaluations. A similar argument is also valid for the users of Samsung brand. A strong support for such an argument also stems from the high efficiency scores of these brands as shown in the last column of Table 3. On the other hand, with the exception of perceived value Nokia was noted to have the largest input excesses in three of the four input categories, and also had the largest output deficit regarding customer loyalty and the second largest deficit concerning customer satisfaction after Motorola. These findings are also confirmed by the efficiency scores where Motorola and Nokia were ranked as the least efficient brands, respectively, as shown in Table 3. These findings might also be construed as such that the users of both brands tend to have some conflicting perceptions regarding their own mobile phones relative to the other brands. After removing the perceptual and evaluation errors among the brand users, a pooled DEA including all mobile phone users at their adjusted efficiency levels in step 3 was used in the calculation of input excess and output deficits to measure and to benchmark the relative inefficiencies among the brands. The average relative input excesses and output deficits were summarized in Table 7. The finding that both Panasonic and Nokia brands had relatively high levels of input excesses on image is worthy of note. This might be construed as such that while Panasonic and Nokia exert a good deal of efforts to raise their image, they do not fully take the advantage of their endeavors in enhancing customer satisfaction and loyalty. With the exception of image, Nokia was noted to have relatively lower levels of input excesses, while it had the lowest output deficits with respect to customer satisfaction and loyalty. So, on the whole this performance would make Nokia be the most efficient brand. LG might appear as the runner up brand in terms of CS&L efficiency as such that while LG had relatively higher levels of image and perceived value excesses; it was found to have the highest level of deficits in terms of customer satisfaction and loyalty. With the current mix of input factors, Table 7 reveals that both Sony Ericsson and Samsung brands have potential to improve their customer satisfaction by 14.15% and 12.15% and customer loyalty by 19.55% and 21.55%, respectively.

5. Conclusion and implications Commensurate with growing interest in the applications of efficiency analyses in management science, this study has attempted to develop a DEA based methodology for examining CS&L efficiency using the example of major cellular phone brands operating in Turkish mobile phone sector. Drawing on ECSI model, EFA was first applied to reduce the number of variables, and to determine

105

the exogenous and endogenous latent variables that were used as input and output variables in the CS&L efficiency model. Second, output-oriented BCC DEA was implemented to identify the CS&L efficiency of mobile phone brands. The evidence obtained from the analysis of relative CS&L efficiency revealed that out of the top six mobile phone brands Nokia featured as the most efficient brand followed by LG and Sony Ericsson, while Motorola, Samsung and Panasonic ranked as the least efficient brands. This study has important implications for practice. Without a doubt, the competition for greater market share is intensifying within the mobile phone industry in Turkey. A more focused approach to building up a novel competitive edge is vital for success (or mere survival) in this volatile market. One of the best and obvious ways of achieving this is through a scientifically sound marketing and customer retention strategy. It is more costly to attract new customers than to retain the existing ones. Therefore, the key focus in managing customer satisfaction is to identify the core satisfaction determinants from the user’s perspective and then to assess the company’s performance in addressing each of these determinants. One of the challenging tasks that existing brands face is how to improve their image and satisfy the expectations of their customers. They must strive to improve features and enhance product and service quality so that they can improve customers’ experiences with mobile phones and by doing so improves overall customer loyalty. Brand managers should also focus on enhancing their company’s image and develop advertising and promotional messages that could encourage customers to think about their experiences with the actual product. This needs to be accomplished simultaneously as reducing costs, if the existing manufacturers desire to maximize customer satisfaction and gain high market penetration through brand loyalty. While this study provides a useful methodology built on DEA modeling to assist managers in making accurate and timely decisions based on measuring CS&L efficiency, its limitations should also be acknowledged. First, the input and output measures of the study were determined using subjective measures largely drawn from the ECSI model. Alternatively, in a potential future study, some additional subjective and objective inputs and outputs measures may be added to enrich the comprehensiveness of the model in investigating the CS&L efficiency of the companies. Another potential future study may employ a two stages approach, which combines DEA and multiple regression analysis, to obtain further insights into some control variables such as age, region, and price level. Finally, given the relative paucity of marketing research in emerging countries, there is an obvious need for comparison of similar studies in other developed countries. Also, it would be interesting to conduct similar studies for comparison purposes at other emerging markets and countries that have commonalities with Turkey, such as India, Brazil, China and Russia.

References Adler, N., & Golany, B. (2001). Evaluation of deregulated airline networks using data envelopment analysis combined with principal component analysis with an application to Western Europe. European Journal of Operational Research, 132(2), 18–31. Adler, N., & Golany, B. (2002). Including principal component weights to improve discrimination in data envelopment analysis. Journal of the Operational Research Society, 53, 985–991. Al-Sharkas, A. A., Hassan, M. K., & Lawrence, S. (2008). The impact of mergers and acquisitions on the efficiency of the US Banking Industry: Further evidence. Journal of Business Financial Accounting, 35, 50–70. Anderson, E. W., Fornell, C., & Lehmann, D. R. (1994). Customer satisfaction, market share, and profitability: Findings from Sweden. Journal of Marketing, 58, 53–66. Andreassen, T. W., & Lindestad, B. (1998). The effects of corporate image in the formation of customer loyalty. Journal of Service Marketing, 1, 82–92. Andrews, F. M. (1984). Construct Validity and error components of survey measures: A structural modeling approach. The Public Opinion Quarterly, 48(2), 409–442.

106

E. Bayraktar et al. / Expert Systems with Applications 39 (2012) 99–106

Athanassopoulos, A. D. (1997). Service quality and operating efficiency synergies for management control in the provision of financial services: Evidence from Greek Bank branches. European Journal of Operational Research, 98, 300–313. Avkıran, N. K. (2006). Developing foreign bank efficiency models for DEA grounded in finance theory. Socio-Economic Planning Sciences, 40(4), 275–296. Bagchi, K., Kirs, P., & Lopez, F. (2008). The impact of price decreases on telephone and cell phone diffusion. Information and Management, 45, 183–193. Banker, R. D., Charnes, A., & Cooper, W. W. (1984). Some model for estimating technical and scale inefficiencies in data envelopment analysis. Management Science, 30(9), 1078–1092. Bayraktar, E., Koh, S. C. L., Tatoglu, E., Demirbag, M., & Zaim, S. (2010). Measuring the efficiency of supply chain management and information systems practices: A comparison of Turkish and Bulgarian SMEs. International Journal of Production Research, 48(2), 425–451. Brockett, P. L., & Golany, B. (1996). Using rank statistics for determining programmatic efficiency differences in data envelopment analysis. Management Science, 42(3), 466–472. Charnes, A., Cooper, W. W., & Rhodes, E. (1978). Measuring the efficiency of decision making units. European Journal of Operational Research, 2, 429–444. Charnes, A., Cooper, W. W., Learner, D. B., & Phillips, F. Y. (1985). Management science and marketing management. Journal of Marketing, 49(2), 93–105. Chien, T. K., Chang, T. H., & Su, C. T. (2003). Did your efforts really win customers’ satisfaction? Industrial Management and Data Systems, 103(4), 253–262. Cook, W. D. (2004). Qualitative data in DEA. In W. W. Cooper, L. M. Seiford, & J. Zhu (Eds.), Handbook on Data Envelopment Analysis. Norwell, MA: Kluwer Academic Publishers. Cook, W. D., & Seiford, L. M. (2009). Data envelopment analysis (DEA) – Thirty years on. European Journal of Operational Research, 192(1), 1–17. Cooper, W. W., Huang, Z., & Li, S. (2004). Chance constraint DEA. In W. W. Cooper, L. M. Seiford, & J. Zhu (Eds.), Handbook on Data Envelopment Analysis. Norwell, MA: Kluwer Academic Publishers. Conover, W. J. (1999). Practical Nonparametric Statistics (3rd ed.). New York: John Wiley & Sons. Cronin, J. J., & Taylor, S. A. (1992). Measuring service quality: A re-examination and extension. Journal of Marketing, 56(3), 55–68. Cronin, J. J., Brady, M. K., & Hult, G. T. M. (2000). Assessing the effects of quality, value and customer satisfaction on consumer behavioral intentions in service environment. Journal of Retailing, 76(2), 193–218. Dedeoglu, A. O. (2004). The symbolic use of mobile telephone among Turkish consumers. Journal of Euromarketing, 13(2), 143–162. Dick, A. S., & Basu, K. (1994). Customer loyalty: Toward an integrated conceptual framework. Journal of the Academy of Marketing Science, 22, 99–113. Donthu, N., & Yoo, B. (1998). Retail productivity assessment using data envelopment analysis. Journal of Retailing, 74(1), 89–105. Donthu, N., Hershberger, E. K., & Osmonbekov, T. (2005). Benchmarking marketing productivity using data envelopment analysis. Journal of Business Research, 58, 1474–1482. Euromonitor International (2010). Young population drives Turkey’s mobile market. Accessed 29.03.10. Farrell, M. J. (1957). The measurement of productive efficiency. Journal of the Royal Statistical Society, Series A: General, 120(3), 253–281. Fornell, C. (1992). A national satisfaction barometer: The Swedish experience. Journal of Marketing, 56, 6–21. Fornell, C., & Wernerfelt, B. (1988). A model for customer complaint management. Marketing Science, 7(3), 287–298. Fornell, C., Michael, D. J., Eugene, W. A., Jaesung, C., & Barbara, E. B. (1996). The American customer satisfaction index: Nature, purpose and findings. Journal of Marketing, 60, 7–18. Demirbag, M., Tatoglu, E., Glaister, K. W., & Zaim, S. (2010). Measuring strategic decision making efficiency in different country contexts: A comparison of British and Turkish firms. Omega, 38(1/2), 95–104. Halstead, D., & Page, T. J. (1992). The effects of satisfaction and complaining behavior on consumer repurchase intentions. Journal of Consumer Satisfaction, Dissatisfaction and Complaining Behavior, 5, 1–11. Hirschman, A. O. (1970). Exit, voice, and loyalty – Responses to declines in firms, organizations and states. Cambridge: Harvard University Press. Hu, W. C., Lai, M. C., & Huang, H. C. (2009). Rating the relative efficiency of financial holding companies in an emerging economy: A multiple DEA approach. Expert Systems with Applications, 36, 5592–5599. Isiklar, G., & Buyukozkan, G. (2007). Using a multi-criteria decision making approach to evaluate mobile phone alternatives. Computer Standards and Interfaces, 29, 265–274. Johnson, M. D., Gustafsson, A., Andreassen, T. W., Lervik, L., & Cha, J. (2001). The evolution and future of national customer satisfaction index models. Journal of Economic Psychology, 22(2), 217–245. Kalba, K. (2008). The adoption of mobile phones in emerging markets: Global diffusion and the rural challenge. International Journal of Communication, 2, 631–661. Kamakura, W. A., Ratchford, B. T., & Agrawal, J. (1988). Measuring marketing efficiency and welfare loss. Journal of Consumer Research, 15, 289–302. Keh, H. T. (2000). Measuring retailer performance: Towards an understanding of productivity. Journal of Targeting, Measurement and Analysis for Marketing, 9(2), 160–173.

Keh, H. T., & Chu, S. (2003). Retail productivity and scale economies at the firm level: A DEA approach. Omega, 31, 75–82. Lee, C. C. (2009). Analysis of overall technical efficiency, pure technical efficiency and scale efficiency in the medium-sized audit firms. Expert Systems with Applications, 36, 11156–11171. Lee, S., Lee, K., & Kang, I. W. (2005). Efficiency analysis of controls in EDI applications. Information and Management, 42, 425–439. Lin, H. T. (2009). Efficiency measurement and ranking of the tutorial system using IDEA. Expert Systems with Applications, 36, 11233–11239. Lin, C., Madu, C. N., Kuei, C. H., & Lu, M. H. (2004). The relative efficiency of quality management practices. International Journal of Quality and Reliability Management, 21(5), 564–577. Lin, T. T., Lee, C. C., & Chiu, T. F. (2009). Application of DEA in analyzing a bank’s operating performance. Expert Systems with Applications, 36, 8883–8891. Mahajan, J. (1991). Data envelopment analytic model for assessing the relative efficiency of the selling function. European Journal of Operational Research, 53, 189–205. Martensen, A., Kristensen, K., & Grønholdt, L. (2000). Customer satisfaction measurement at post Denmark: Results of application of the European customer satisfaction index methodology. Total Quality Management, 11(7), 1007–1015. Murthi, B. P. S., Srinivasan, K., & Kalyanaram, G. (1996). Controlling for observed and unobserved managerial skills in determining first-mover market share advantages. Journal of Marketing Research, 33(3), 329–336. Narasimhan, R., Talluri, S., & Das, A. (2004). Exploring flexibility and execution competencies of manufacturing firms. Journal of Operations Management, 22(1), 91–106. Narver, J. C., & Slater, S. F. (1990). The effect of a market orientation on business profitability. Journal of Marketing, 54(4), 20–35. OECD (2008). OECD territorial reviews: Istanbul, Turkey. OECD Policy Review. . Ozcan, Y. Z., & Kocak, A. (2003). Research note: A need or a status symbol? Use of cellular telephones in Turkey. European Journal of Communication, 18(2), 241–254. Patron Turk (2010). Market shares of mobile phone companies in Turkey Accessed 22.03.10. Reichheld, F., & Sasser, W. E. (1990). Zero defections: Quality comes to services. Harvard Business Review, 68(September/October), 105–111. RNCOS (2010). Turkey telecom sector forecast to 2012. Accessed 29.03.10. Rotondaro, R. G. (2002). Defining the customers expectation’s in E-business. Industrial Management and Data Systems, 102(9), 476–482. Sherman, H. D., & Ladino, G. (1995). Managing bank productivity using data envelopment analysis. Interfaces, 25(2), 60–73. Shuai, J. J., & Wu, W. W. (2011). Evaluating the influence of E-marketing on hotel performance by DEA and grey entropy. Expert Systems with Applications, 38(7), 8763–8769. Simsek, G. G., & Noyan, F. (2009). A model of brand loyalty in Turkish cellular phone market. METU Studies in Development, 36(1), 121–159. Soteriou, A. C., & Staurinides, Y. (1997). An internal customer service quality data envelopment analysis model for bank branches. International Journal of Operations and Production Management, 17(8), 780–789. Swink, M., Talluri, S., & Pandejpong, T. (2006). Faster, better, cheaper: A study of NPD project efficiency and performance tradeoffs. Journal of Operations Management, 24(5), 542–562. Sueyoshi, T., & Aoki, S. (2001). A use of a nonparametric statistic for DEA frontier shift: The Kruskal and Wallis rank test. Omega, 29, 1–18. Syzmanski, D. M., & Henard, D. H. (2001). Customer satisfaction: A meta analysis of the empirical evidence. Journal of the Academy of Marketing Science, 29(1), 16–35. Tenenhaus, M., Vinzi, V. E., Chatelin, Y. M., & Lauro, C. (2005). PLS path modeling. Computational Statistics and Data Analysis, 48, 159–205. Thanassoulis, E. (1999). DEA and its use in the regulation of water companies. European Journal of Operational Research, 127(1), 1–13. Thomas, R. R., Barr, R. S., Cron, W. L., & Slocum, S. W. (1998). A process for evaluating retail store efficiency: A restricted DEA approach. International Journal of Research in Marketing, 15, 487–503. Turkyilmaz, A., & Ozkan, C. (2007). Development of a customer satisfaction index model: An application to the Turkish mobile phone sector. Industrial Management and Data Systems, 107(5), 672–687. Webster, F. E. (1992). The changing role of marketing in the corporation. Journal of Marketing, 56(4), 1–17. Westbrook, R. A., & Oliver, R. L. (1991). The dimensionality of consumption emotion patterns and customer satisfaction. Journal of Consumer Research, 18, 84–91. Wu, L., Liang, L., & Chen, Y. (2009). DEA game cross-efficiency approach to olympic rankings. Omega, 37(4), 909–918. Zhang, X., Huang, G. H., Lin, Q., & Yu, H. (2009). Petroleum-contaminated groundwater remediation systems design: A data envelopment analysis based approach. Expert Systems with Applications, 36, 5666–5672. Zhu, J. (2003). Imprecise data envelopment analysis (IDEA): A review and improvement with an application. European Journal of Operational Research, 144(3), 513–529.