A model for efficiency analysis of the customer satisfaction process

Transcription

1 A moel for efficiency analysis of the customer satisfaction process AUTHORS ARTICLE INFO Herbert F. Lewis Sanal K. Mazvancheryl Herbert F. Lewis an Sanal K. Mazvancheryl (2011). A moel for efficiency analysis of the customer satisfaction process. Innovative Mareting 7(1) JOURNAL "Innovative Mareting " FOUNER LLC Consulting Publishing Company Business Perspectives NUMBER OF REFERENCES 0 NUMBER OF FIGURES 0 NUMBER OF TABLES 0 The author(s) This publication is an open access article. businessperspectives.org

2 Herbert F. Lewis (USA) Sanal K. Mazvancheryl (USA) A moel for efficiency analysis of the customer satisfaction process Abstract It has been acnowlege that customer satisfaction is an important component of a firm s mareting strategies an tactics an has also been shown to be associate with improve outcomes for the firm lie greater maret share higher profitability an increase shareholer value. Current research in customer satisfaction while useful in etermining the effectiveness of customer satisfaction creation an its consequences oes not help managers measure the efficiency of their customer satisfaction efforts. This paper presents a moel for measuring the efficiency of the customer satisfaction process. It uses a Networ ata Envelopment Analysis (Networ EA) moel an extension of the basic EA approach as it is consistent with the literature on firm level customer satisfaction formation an its consequences. The authors apply the Networ EA methoology to the American Customer Satisfaction Inex framewor. They estimate this moel on 22 firms in the automobile inustry using the American Customer Satisfaction Inex (ACSI). Our results inicate that at least for certain firms there is room for improvement in efficiency of the customer satisfaction process. We further fin that our EA moel is not just correctly ientifying which firms are efficient in the customer satisfaction process but also which firms in their competitive set are using their resources more efficiently than others an whom they coul liely imitate. Keywors: Networ EA efficiency customer satisfaction. Introuction Customer satisfaction is an important component of a firm s mareting strategies an tactics (Fornell 2001). Recently there has been an increase acaemic interest in customer satisfaction (Anerson an Mittal 2000). Most acaemic research in customer satisfaction has focuse on the anteceents an consequences of customer satisfaction (Yi 1991). There is an establishe research stream looing at both iniviual-level as well as firm-level causes of customer satisfaction (Oliver 1980; Westbroo an Oliver 1991; Anerson Fornell an Lehmann 1994). There is also a growing boy of literature lining customer satisfaction with iniviual-level consequences (Anerson an Sullivan 1993) as well as accounting an financial measures of firm performance incluing operating margin (Rust Zahori an Keiningham 1994; Bolton 1998 Morgan an Rego 2006) return on investment (Anerson Fornell an Rust 1997; Zeithaml 2000) accounting returns (Ittner an Larcer; 1998 Ittner et al. 2009) an shareholer value (Anerson Fornell an Mazvancheryl 2004; Luo an Bhattacharya 2006). However there is very little wor on the efficiency of the customer satisfaction process as oppose to its effectiveness (Soteriou an Zenios 1999; Kamaura et al. 2002; Mittal et al. 2005). Managers want to now how efficient their current customer satisfaction efforts are before allocating any aitional resources to increase satisfaction levels (Reichhel an Sasser 1996; Klein an Einstein 2003). For example they woul lie to now if compare to other firms in their inustry they shoul be achieving a higher level of customer satisfaction an loyalty given their current level of spening. In the absence of such research managers Herbert F. Lewis Sanal K. Mazvancheryl The orer of authorship is alphabetical an oes not reflect the relative contribution of each author. are liely to remain ambivalent towars using satisfaction as a ey strategic metric (Srivastava Shervani an Fahey 1998; Ambler 2000). In this paper we present a moel for measuring the efficiency of the customer satisfaction process. We begin by applying the Networ ata Envelopment Analysis (Networ EA) methoology of Lewis an Sexton (2004a) to the American Customer Satisfaction Inex (ACSI) framewor of Fornell et al. (1996). A 3-stage Networ EA moel is evelope that allows us to measure the overall efficiency of the customer satisfaction process as well as efficiencies in each of the stages of the process. We use our moel to analyze the relative efficiency of the customer satisfaction process for firms in the automobile inustry an investigate if there are any systematic ifferences in efficiency in the ifferent stages of this process. Our finings inicate that a majority of the firms in the automobile inustry may be inefficient in their customer satisfaction efforts. Further there are significant ifferences in efficiency across the stages of the process. Our results have several important implications for managers in their usage an allocation of resources to customer satisfaction efforts. 1. The customer satisfaction process In the well-establishe customer satisfaction literature customer satisfaction is posite to be a function of customers expectation about a prouct or service (Churchill an Suprenant 1982; Westbroo 1981) the perceive quality of the prouct (Oliver 1980) as well as its perceive value (Johnson 1984). The immeiate consequences of increase customer satisfaction are lower levels of customer complaints (Hirshman 1970) an increase customer loyalty (Fornell an Wernerfelt 1988). This conceptual framewor is the basis of the wiely use ACSI methoology (Fornell et al. 33

3 1996) that consists of a causal moel lining these six constructs (see Figure 1). Expectations Quality Value Satisfaction Complaints Loyalty Fig. 1. The ACSI moel of the customer satisfaction process We ientify the three istinct stages in the ACSI moel as well as the measures that serve as inputs an outputs in each stage. In the first stage referre to as the Value-Generation stage customer expectations an perceive quality lea to perceive value. In the secon stage which we call the Satisfaction-Creation stage this perceive value together with the level of customer expectations an quality leas to customer satisfaction. Finally in the thir stage calle the Loyalty-Builing stage customer complaints an loyalty are the consequences of satisfaction. This overall 3- stage framewor together with the input an output measures in each stage is shown in Figure 2 1. Expectation Firm Expectation Quality Value- Generation stage Value Satisfaction- Creation stage Satisfaction Loyalty- Builing stage Complaints Loyalty Quality 2. Efficiency measurement of the customer satisfaction process To measure the efficiency of the customer satisfaction process we turn to the technique of ata Envelopment Analysis (EA). EA has become a wiely use methoology for evaluating relative efficiency in a variety of contexts in such iverse fiels as eucation (Bessent et al. 1982) electricity prouction an istribution (Färe an Primont 1984) recreation (Rhoes 1982) an health care (Sexton et al. 1989; Hollingsworth et al. 1999). Tavares (2002) provies a comprehensive bibliography of the EA literature. In past years EA has been applie to stuy research issues in mareting incluing the prouctivity of retail outlets (Kamaura et al. 1996; onthu an Yoo 1998; hontu et al. 2005) benchmaring of avertising spening (Luo an onthu 2001) the evaluation of the efficiency of the service profit chain (Kamaura et al. 2002) an the revenue efficiencies of a hotel chain (Keh et al. 2006). In a recent stuy Mittal et al. (2005) investigate the impact of customer satisfaction efficiencies on firm financial performance. They fin that firms that achieve both revenue increases an cost performances simultaneously outperform those firms that o not pursue such a ual emphasis. 34 Fig. 2. The 3 stages of the customer satisfaction process EA s 1 mathematical evelopment can be trace to Charnes et al. (1978) who built on the wor of Farrell (1957). EA is esigne specifically to measure relative efficiency in situations in which there are multiple inputs an outputs an there is no obvious objective way of aggregating either inputs or outputs into a meaningful inex of prouctive efficiency. In its basic form EA consiers a collection of ecision-maing units (MUs) each of which uses certain levels of selecte inputs to prouce levels of selecte outputs. EA maes no assumptions regaring the manner in which a MU converts inputs into outputs; each MU is regare as a blac box with respect to its prouction process. EA allows for iffering assumptions regaring returns-to-scale. For example using the variable returns-to-scale assumption a MU will be compare to other MUs using similar levels of inputs an outputs. In aition EA moels may be input-oriente output-oriente or un-oriente. Input-oriente moels ientify input reuctions that woul enable a MU to become efficient while output-oriente moels ientify output increases that woul achieve the same effect. Un- 1 While a strict interpretation of the ACSI moel woul also require a linage between expectations an quality our simplifie framewor allows for a tractable 3-stage EA moel that preserves the relationships of the ACSI moel.

4 oriente moels ientify a mix of input reuctions an output increases that lea to efficiency. EA establishes an efficient frontier base on the observe best performing MUs among those in the analysis an evaluates the efficiency of each MU relative to this frontier. MUs that lie on the frontier are efficient. EA evaluates the efficiency of a MU that oes not lie on the frontier relative to a linear combination of the efficient MUs. The efficient MUs that receive positive weight in this linear combination constitute the efficient reference set for the inefficient MU. This linear combination represents an empirically feasible reference MU that ominates the inefficient MU uner evaluation in the sense that the reference MU uses no more of each input while proucing at least as much of each output as oes the MU uner evaluation. The EA moel fins the most prouctive reference MU an computes the efficiency of the MU uner evaluation relative to this reference MU. The basic EA moel for a specific MU can be formulate as a linear program. A complete EA requires that one such linear program be solve for each MU. The result is both the efficient reference set for each MU an its relative efficiency. 3. The Networ EA moel of customer satisfaction Recent research (Lewis an Sexton 2004a; Sexton an Lewis 2003; Castelli et al. 2001; Färe an Grossopf 2000) has extene the EA methoology to enable analysts to loo insie processes with complex internal structures. This research enables us to more accurately moel complex processes that inclue several stages such as the customer satisfaction process shown in Figure 2. We can thus measure the efficiency of both the overall process as well as the iniviual stages. We present a Networ EA moel of the customer satisfaction process base on the methoology in Lewis an Sexton (2004a). We first solve the stanar EA moel for each stage of the process. This gives us a relative efficiency measure for each stage. In orer to measure the organizational efficiency we then sequentially resolve the iniviual EA moels assuming that all previous stages are efficient. For example while solving the moel for the Satisfaction-Creation stage we use the level of perceive value that the Value-Generation stage woul have prouce ha it been efficient. Similarly while solving the moel for the Loyalty-Builing stage we use the level of customer satisfaction that the Satisfaction-Creation stage woul have prouce ha both the Value-Generation stage an the Satisfaction- Creation stage been efficient. We moel each stage of the process using an output orientation. Further each stage is assume to have a variable returns-to-scale. Let enote the inex of the MU uner analysis enote the inex of the MUs = 1 thus we efine that: 1 = Weight place on the Value-Generation stage in MU by the Value-Generation stage in MU ; 2 = Weight place on the Satisfaction- Creation stage in MU by the Satisfaction- Creation stage in MU ; 3 = Weight place on the Loyalty-Builing stage in MU by the Loyalty-Builing stage in MU ; 1 = Inverse efficiency of Value-Generation stage at MU ; 2 = Inverse efficiency of Satisfaction-Creation stage at MU ; 3 = Inverse efficiency of Loyalty-Builing stage at MU ; an E 3 = Efficiency of the Loyalty-Builing stage at MU. Next following Figure 2 we efine the inputs an outputs of the three stages: EXP = Expectation inex of MU ; QUA = Quality inex of MU ; VAL = Value inex of MU ; SAT = Satisfaction inex of MU ; COM = Complaints inex of MU ; an LOY = Loyalty inex of MU. We begin by calculating the inverse efficiency scores for each stage. This involves solving three EA moels for each MU. The Value-Generation stage uses two inputs (EXP an QUA) an prouces one intermeiate prouct (VAL). The EA moel for the Value-Generation stage at MU is: Maximize ( EXP) ( QUA) ( VAL) 1 1 subject to ( EXP) ( QUA) ( VAL) Moel 1 35

5 The Satisfaction-Creation stage uses two inputs (EXP an QUA) as well as one intermeiate prouct (VAL) to prouce one intermeiate prouct (SAT). Therefore the EA moel for the Satisfaction- Creation stage at MU is: Maximize ( EXP) ( QUA) ( VAL) ( SAT ) 1 2 subject to ( EXP) ( QUA) ( VAL) 2 ( SAT ) Moel 2 The Loyalty-Builing stage uses one intermeiate prouct (SAT) an prouces two outputs (LOY) an (COM) 2 The EA moel for the Loyalty-Builing stage at MU is: Maximize E E ( SAT) ( COM ) ( LOY) subject to ( SAT) E ( LOY) 3 3 ( COM ) Moel 3 Next we calculate the organizational inverse efficiency score for each MU. This involves solving two more EA moels for each MU. We resolve the moel for the Customer-Satisfaction stage using the level of VAL the Value-Generation 2 Note that COM is a reverse quantity i.e. larger values inicate lower output levels (Lewis an Sexton 2004b). stage coul have prouce ha it been efficient. We enote this by VAL *. Thus we moify the thir constraint in Moel 2 above as follows: 1 * ( VAL) ( VAL). ( 1) 2 Finally we re-solve the EA moel for the Loyalty- Builing stage using the level of SAT that the Satisfaction-Creation stage coul have prouce ha both the Value-Generation stage an the Satisfaction- Creation stage been efficient. We enote this by * SAT *. Thus the first constraint in Moel 3 above is moifie as follows: 1 * * ( SAT) ( SAT). ( 2) 3 The value of 3 for this EA moel is the organizational inverse efficiency. 4. An application to the U.S. automobile inustry 4.1. The ata. We use ata from the ACSI. The ACSI methoology provies a uniform inepenent customer-base cumulative firm level satisfaction measure for nearly 200 companies in 40 inustries over 7 sectors of the U.S. economy. It covers over 40% of the gross omestic prouct (GP) of the U.S. an inclues both the private sector an the public sector. The raw ata for the ACSI is collecte using ranom telephone surveys of customers (at least 200 customers per firm) who have recently consume the specific bran of prouct or service of the firm. Responents are ase questions on 15 measurement variables which are then use as inicators of the 6 latent variables or constructs shown in Figure 1. Each of these constructs can range from 0 to 100 an are comparable across firms. Fornell et al. (1996) escribes the latent variable estimation technique use. We apply the customer satisfaction efficiency moel escribe above to the U.S. automobile inustry. There are 22 such firms in the ACSI sample. We use the inexe ata from 1998 for each of the following six constructs: customer expectation (EXP) perceive quality (QUA) perceive value (VAL) customer satisfaction (SAT) complaints (COM) an loyalty (LOY). This ata is shown in Table Results Inustry results. We present the results of applying our output-oriente Networ EA moel to the ata shown in Table 1. Our analysis yiels an inverse efficiency for each of the 3 stages as well as an organizational inverse efficiency score for each of the 22 firms. These results as well as summary statistics are presente in Table 2.

6 Table 1. Customer satisfaction ata for the automobile inustry Firm name Expectation Quality Value Satisfaction Complaints Loyalty Volvo Mercees Benz BMW Subaru Hyunai Maza Volswagen Nissan Hona Toyota Chrysler-Jeep/Eagle Chrysler-oge-Plymouth Chrysler-Chrysler For-Lincoln-Mercury For-For GM-Saturn GM-Pontiac GM-Chevrolet-GEO GM-Buic GM-Olsmobile GM-Caillac GM-GMC Table 2. Inverse efficiency scores for each automobile firm an inustry-wie summary statistics Firm name Value- Generation Satisfaction- Creation Inverse efficiency scores Loyalty- Builing Firm-level Volvo Mercees Benz BMW Subaru Hyunai Maza Volswagen Nissan Hona Toyota Chrysler-Jeep/Eagle Chrysler-oge-Plymouth Chrysler-Chrysler For-Lincoln-Mercury For-For GM-Saturn GM-Pontiac GM-Chevrolet-GEO GM-Buic GM-Olsmobile GM-Caillac GM-GMC MEAN S MINIMUM MEIAN MAXIMUM

7 We expect that a mature an well-evelope inustry lie the U.S. automobile inustry woul be fairly efficient. Our results confirm this as the mean inverse efficiencies for the 3 stages are an respectively. The organizational mean inverse efficiency is While these inverse efficiency scores seem relatively small we must remember that the implications of improving efficiency by even a small amount can have a significant impact on the financial performance of a firm. For example Kamaura et al. (2002) fin a strong lin between increasing customer loyalty an increasing firm profitability. In aition Anerson Fornell an Mazvancheryl (2004) fin that increasing customer satisfaction by just one point (for firms in the ACSI sample) increases shareholer value on average by $ 275 million. Also only a few of the firms are efficient in any given stage inicating that most firms in our sample have at least some room for improvement. Specifically only 2 out of 22 firms (9.1%) are efficient in the first or the Value-Generation stage. Only 7 out of 22 firms (31.8%) are efficient in the Satisfaction-Creation stage an only 8 out of 22 firms (36.3%) are efficient in the final or Loyalty- Builing stage. Organizationally just 4 firms out of 22 (18.1%) are efficient. Even among these 4 firms only 2 (GM-Saturn an Hyunai) are efficient in all 3 stages. The other 2 organizationally efficient firms (GMC an GM-Caillac) can improve their efficiency in the Value-Generation stage. In particular GMC has an inverse efficiency score of in the Value- Generation stage an therefore may want to focus efforts on this stage. The sprea of the inverse efficiency scores as measure by the stanar eviation shows that the Loyalty- Builing stage has the highest stanar eviation (0.034) relative to the Value-Generation stage (0.0135) an the Satisfaction-Creation stage (0.0135). The stanar eviation of organizational inverse efficiency scores is Thus in orer to improve the overall efficiency of the customer satisfaction process managers might nee to focus their efforts on Loyalty- Builing as well as on Value-Generation. This fining is also supporte by the literature which suggests that builing loyalty might be harer than generating value or creating customer satisfaction (Bloemer an Kasper 1995) Firm-level results. For each inefficient MU in each stage as well as for the organization the Networ EA moel provies us with an efficient reference MU which is a linear combination of efficient MUs in that stage. This reference MU uses no more of each input an prouces at least as much of each output as oes the inefficient MU. The efficient MUs in each stage reference themselves with a weight of 1. The reference sets for the Value-Generation stage are shown in Table 3 for the Satisfaction-Creation stage are shown in Table 4 an for the Loyalty-Builing stage are shown in Table 5. Finally the reference sets for the organization are shown in Table 6. Table 3. Value-Generation stage reference set weights an inverse efficiency scores Firm name Reference set weights Hyunai GM-Saturn Inverse efficiency Volvo Mercees Benz BMW Subaru Hyunai Maza Volswagen Nissan Hona Toyota Chrysler-Jeep/Eagle Chrysler-oge-Plymouth Chrysler-Chrysler For-Lincoln-Mercury For-For GM-Saturn GM-Pontiac GM-Chevrolet-GEO GM-Buic GM-Olsmobile GM-Caillac GM-GMC

8 Table 4. Satisfaction-Creation stage reference set weights an inverse efficiency scores Firm name BMW Subaru Hyunai Chrysler- Jeep/Eagle Reference set weights GM-Saturn GM-Caillac GM-GMC Inverse efficiency Volvo Mercees Benz BMW Subaru Hyunai Maza Volswagen Nissan Hona Toyota Chrysler-Jeep/Eagle Chrysler-oge-Plymouth Chrysler-Chrysler For-Lincoln-Mercury For-For GM-Saturn GM-Pontiac GM-Chevrolet-GEO GM-Buic GM-Olsmobile GM-Caillac GM-GMC Firm name Table 5. Loyalty-Builing stage reference set weights an inverse efficiency scores Mercees Benz Reference set weights Hyunai Hona For-For GM-Saturn GM-Buic GM-Caillac GM-GMC Inverse efficiency Volvo Mercees Benz BMW Subaru Hyunai Maza Volswagen Nissan Hona Toyota Chrysler-Jeep/Eagle Chrysler-oge-Plymouth Chrysler-Chrysler For-Lincoln-Mercury For-For GM-Saturn GM-Pontiac GM-Chevrolet-GEO GM-Buic GM-Olsmobile GM-Caillac GM-GMC

9 40 Firm name Table 6. Firm-level reference set weights an inverse efficiency scores Mercees Benz Hyunai Hona Chrysler- Chrysle Reference set weights For-For GM-Saturn GM-Buic GM- Caillac GM-GMC Inverse efficiency Volvo Mercees Benz BMW Subaru Hyunai Maza Volswagen Nissan Hona Toyota Chrysler-Jeep/Eagle Chrysler-oge-Plymouth Chrysler-Chrysler For-Lincoln-Mercury For-For GM-Saturn GM-Pontiac GM-Chevrolet-GEO GM-Buic GM-Olsmobile GM-Caillac GM-GMC Consier the example of Maza. We efine Maza has an inverse efficiency score of in the Value- Generation stage. In this stage Maza references two efficient firms Hyunai an Saturn with weights of an respectively. A reference MU forme by combining Hyunai an Saturn in the respective proportions uses hypothetical input levels of an of quality an expectation respectively an prouces a hypothetical output of value equal to If we loo at the actual inputs an Table 7. Stage-level EA results for Maza Inverse efficiency of the Value-Generation stage output of Maza in this stage (quality = 84.01; expectation = an value = 81.09) we see that the reference MU uses no more of each input while proucing a higher level of output. The ratio of the hypothetical output to the actual output is the inverse efficiency score for Maza. Thus Maza has room for improvement in creating value. Table 7 presents this analysis along with similar analyses for the Satisfaction-Creation an the Loyalty- Builing stages. Firm name Maza Hypothetical Ratio Firm name Hyunai GM-Saturn Expectation Expectation Quality Aj Quality Aj Value Value Inverse efficiency Weights Inverse efficiency of the Satisfaction-Creation stage Firm name Maza Hypothetical Ratio Firm name Subaru GM-Saturn GM-GMC Expectation Expectation Quality Quality Value Value Satisfaction Satisfaction Inverse efficiency Weights Inverse efficiency of the Loyalty-Builing stage Firm name Maza Hypothetical Ratio Firm name Hyunai For-For GM-GMC Satisfaction Satisfaction Complaints Complaints Loyalty Loyalty Inverse efficiency Weights

10 In many applications of EA there are factors calle site characteristics that can influence the ability of a MU to operate efficiently but that are beyon the control of the MU s management. In our application we consier whether the automobile firm is omestic or foreign as a site characteristic. Sexton et al. (2002) an Frie et al. (1999) present similar but istinct approaches for ealing with site characteristics in stanar EA moels. The metho of Sexton et al. (2002) requires us to ajust quality in the Value- Generation stage. Thus in Table 7 above an Table 8 below we use the ajuste level of quality as an input to the Value-Generation stage. None of the other ACSI ata requires ajustment. In etermining the organizational inverse efficiency scores the moel uses the output of the reference MU in a stage as the input to the next stage assuming all previous stages of the process are efficient. Thus in Maza s case the level of value (= 83.16) that woul have been prouce by the Value-Generation stage ha it been efficient is use as an input to the Satisfaction-Creation stage. In turn the level of satisfaction (= 79.45) that woul have been prouce by the Satisfaction-Creation stage ha both the Value-Generation Table 8. Firm-level EA results for Maza Organizational inverse efficiency Innovative Mareting Volume 7 Issue an Satisfaction-Creation stages been efficient is use as an input to the Loyalty-Builing stage. This prouces the levels of complaints an loyalty of a reference MU equal to an respectively. The organizational inverse efficiency score of is the minimum of the so-calle factor inverse efficiencies of each variable. In this particular case the factor inverse efficiencies for both loyalty an complaints are ientical. However this is not necessarily the case in general. Table 8 shows the results of this analysis. Since we are using a variable returns-to-scale moel the reference sets consist of firms that operate at a comparable scale i.e. they use similar levels of inputs an outputs. For example if we loo at Mercees in the Satisfaction-Creation stage we see that it references two of the efficient firms in that stage BMW an Caillac (see Table 4) with weights of an respectively. However Mercees oes not reference Hyunai which is another efficient firm. As shown in Table 9 the input an output levels of both BMW an Caillac are very similar to those of Mercees. On the other han the input an output levels of Hyunai are very ifferent from those of Mercees. Firm name Maza Hypothetical Ratio Firm name Hyunai GM-Saturn Expectation Expectation Quality Aj Quality Aj Value Value Inverse efficiency Weights Firm name Maza Hypothetical Ratio Firm name Subaru GM-Saturn GM-GMC Expectation Expectation Quality Quality Value Value Satisfaction Satisfaction Inverse efficiency Weights Firm name Maza Hypothetical Ratio Firm name Hona For-For GM-GMC Satisfaction Satisfaction Complaints Complaints Loyalty Loyalty Inverse efficiency Weights Table 9. Satisfaction-Creation stage input/output ata for Mercees relative to BMW Caillac an Hyunai Firm name Expectation Quality Value Satisfaction Mercees Benz BMW GM-Caillac Hyunai iscussion an conclusions The above research presents an appropriate methoology i.e. Networ EA to measure the overall efficiency as well as stage-by-stage efficiency of an im- portant mareting process i.e. the customer satisfaction process. This efficiency measurement complements the existing research on the consequences of customer satisfaction that focuses almost exclusively on the effectiveness of the customer satisfaction process. Further we emonstrate an application of our methoology using ata on the automobile inustry from a wiely accepte customer satisfaction atabase i.e. ACSI. We fin that while efficiency scores are relatively goo across the inustry there are some ifferences in efficiency scores across the ifferent stages of the process as well as across iniviual firms. 41

11 Our inustry-level results emonstrate that even in a mature an competitive inustry lie the U.S. auto inustry customer satisfaction efforts of the firms are overall not efficient with the inefficiencies being greatest at the Loyalty-Builing stage. This suggests that recent efforts by automobile manufacturers to buy customer loyalty through large amounts of cash iscounts an other financial incentives might actually be a very inefficient tool if the objective is to buil customer loyalty rather than increase short-term sales. The firm-level results are also illuminating. One of the interesting results is that firms that are efficient at the Value-Generation stage (Hyunai an Saturn) are relatively recent entrants that are positione in the value segment of the U.S. automobile maret. Similarly firms that have long been positione in the luxury segment of the maret lie Mercees an GM- Caillac also seem to be the most efficient in the Loyalty-Builing stage. While further research is neee to confirm this our finings seem to suggest that a firm s successful position is relate to its efficient use of resources in creating that positioning. The reference weights that are shown in Table 6 are also interesting. The inefficient firms typically place most of their reference weights on efficient firms that woul be consiere as being in the same com- petitive set. For example Maza places most of its reference weight on Hona an GMC. Mercees on the other ha places most of its weight on Caillac. This suggests that our EA moel is not just correctly ientifying which firms are efficient in the customer satisfaction process but also which firms in their competitive set are using their resources more efficiently an whom they coul liely imitate. Finally we use Figures 3 an 4 to present perceptual maps comparing measures of efficiency an effectiveness in the automobile inustry. In both figures we use the firm-level inverse efficiency as the efficiency measure. In Figure 3 we use satisfaction as the measure of effectiveness while in Figure 4 we use loyalty as the measure of effectiveness. The vertical an horizontal lines in each map represent meian values. Firms in the lower-right region of the maps (Buic Caillac Hona Lincoln/Mercury Mercees Benz Saturn Subaru Toyota an Volvo) are both efficient an effective. Firms in the upper-right region of the maps (BMW an Olsmobile) are inefficient but effective. Firms in the upper-left region of the maps (Chevrolet/GEO Chrysler oge/plymouth For Jeep/Eagle Maza Nissan Pontiac an Volswagon) are inefficient an ineffective. Finally firms in the lower-left region of the maps (GMC an Hyunai) are efficient but ineffective Pontiac 1.12 Jeep/Eagle Nissan Maza Firm inverse efficiency Volswagen Chrysler Chevrolet/GEO For oge/plymouth Olsmobile BMW Volvo Subaru Lincoln/Mercury 1.03 Hona Toyota Buic Mercees 1 Hyunai GMC Saturn Caillac Satisfaction Fig. 3. Perceptual map of firm-level inverse efficiency versus customer satisfaction 42

12 1.15 Pontiac 1.12 Jeep/Eagle Nissan Maza Firm inverse efficiency Chrysler Chevrolet/GEO For BMW Volswagen Olsmobile oge/plymouth Subaru Lincoln/Mercury Volvo 1.03 Hona Toyota Buic Mercees 1 Hyunai GMC Saturn Caillac Loyalty Managerial implications Fig. 4. Perceptual map of firm-level inverse efficiency versus loyalty Our Networ EA moel may allow managers to allocate resources across the stages of the customer satisfaction process so as to better utilize the resources at their isposal. Managers can also complement such an analysis with internal profitability measures to ecie whether or not to allocate aitional resources to satisfaction an if so where. Current thining has focuse almost solely on increasing customer satisfaction or loyalty levels without much thought for the resources being utilize in achieving such levels (Hauser et al. 1994). This coul be one of the reasons for the conflicting finings on whether or not satisfaction or loyalty leas to higher levels of profitability (Klein an Einstein 2003). Efficiency analyses such as ours have the potential to increase profitability by pointing to areas in which current resources can be more efficiently utilize. But there are practical issues to be consiere as well. To be able to implement this approach managers nee etaile ata on all stages of the customer satisfaction meaning that they will nee to put into place a customer satisfaction process similar to the one escribe in Fornell (2001). This may prove to be both expensive an time consuming for managers in smaller firms. Limitations an future research It is important to remember that our analysis oes not inicate why a certain firm is efficient or inefficient nor oes it provie us with specific actionable guielines to improve efficiency. For example our research may inicate that for a given level of customer satisfaction the firm coul have achieve a higher level of loyalty an a lower level of complaints. However this alone cannot tell managers how to choose from the several possible alternative choices they may have to achieve these levels of outputs. An unerstaning of the context of a process is necessary to help managers mae such ecisions. An important extension of our research woul be to lin the efficiency analysis of the customer satisfaction process with a firm s effectiveness in creating profits an long-term value. It woul be useful to see if the efficient firms are the ones that are effective as well. Also are there any stages of the customer satisfaction process where efficiency is more crucial to a firm being more effective than others? Recognizing that time effects such as lags an persistence play an important role in the customer satisfaction process it may be useful to exten the Networ EA methoology to incorporate such effects using multi-perio ata. Finally another important area for future research is to apply the Networ EA moel to other inustries an contexts either using ACSI ata or to stuy relate processes lie the creation of customer lifetime value an profits (Rust et al. 2002). 43

13 Acnowlegments The authors gratefully acnowlege the ata provie by Professor Claes Fornell an the National Quality Research Center at the Ross School of Business at the University of Michigan. This wor has References 44 benefite substantially from the comments of Professor Thomas R. Sexton as well as participants at research seminars at the College of Business at Stony Broo University an the 2003 INFORMS- Mareting Science Conference at the University of Marylan. 1. Ambler Tim (2000). Mareting an the Bottom Line Financial Times Prentice Hall Publishing. 2. Anerson Eugene W. Claes Fornell an onal R. Lehmann (1994). Customer Satisfaction Maret Share an Profitability Journal of Mareting 56 (July) pp Anerson Eugene W. Claes Fornell an Sanal K. Mazvancheryl (2004). Customer Satisfaction an Shareholer Value Journal of Mareting 68 (4) pp Anerson Eugene W. Claes Fornell an Rolant T. Rust (1997). Customer Satisfaction Prouctivity an Profitability: ifferences between Goos an Services Mareting Science 16 (2) pp Anerson Eugene W. an Vias Mittal (2000). Strengthening the Satisfaction-Profit Chain Journal of Service Research 3 (20) pp Anerson Eugene W. an Mary Sullivan (1993). The Anteceents an Consequences of Customer Satisfaction For Firms Mareting Science 12 (2) pp Bessent A. W. Bessent J. Kennington an B. Reagan (1982). An application of mathematical programming to assess prouctivity in the Houston Inepenent School istrict Management Science 28 (12) pp Bloemer J.M.M. an J..P. Kasper (1995). The complex relationship between consumer satisfaction an bran loyalty Journal of Economic Psychology 16 pp Bolton Ruth N. (1998). A ynamic moel of the uration of the customer s relationship with a continuous service provier: the role of satisfaction Mareting Science 17 (1) pp Castelli C. R. Pesenti an W. Uovich (2001). EA-lie moels for efficiency evaluations of specialize an interepenent units European Journal of Operational Research 132 pp Charnes A. W.W. Cooper an E. Rhoes (1978). Measuring the efficiency of ecision maing units European Journal of Operational Research 2 (6) pp Churchill Gilbert A. an Carol Surprenant (1982). An Investigation into the eterminants of Consumer Satisfaction Journal of Mareting Research 19 (November) pp onthu N. an B. Yoo (1998). Retail prouctivity assessment using ata envelopment analysis Journal of Retailing 74 (1) pp Emun Hershberger an Talai Osomonbeov (2005). Benchmaring Mareting Prouctivity Using ata Envelopment Analysis Journal of Business Research 58 (November) pp Färe R. an S. Grossopf (2000). Networ EA Socio-Economic Planning Sciences 34 pp Färe R. an. Primont (1984). Efficiency measures for multiplant firms Operations Research Letters 3 (5) pp Farrell M.J. (1957). The measurement of prouctive efficiency Journal of the Royal Statistical Society Series A 3 pp Fornell Claes (2001). The Science of Satisfaction Harvar Business Review March-April pp Fornell Claes Michael. Johnson Eugene W. Anerson Jaesung Cha an Barbara Bryant (1996). The American Customer Satisfaction Inex: escription Finings an Implications Journal of Mareting 60 (4) pp Fornell Claes an Birger Wernerfelt (1988). A Moel For Customer Complaint Management Mareting Science 7 (Summer) pp Frie H.O. S.S. Schmit an S. Yaisawarng (1999). Incorporating the Operating Environment Into a Nonparametric Measure of Technical Efficiency Journal of Prouctivity Analysis 12 pp Hauser John R uncan L. Simester an Birger Wernerfelt (1994). Customer satisfaction incentives Mareting Science 13 (4) pp Hirschman Albert O. (1970). Exit Voice an Loyalty Responses to ecline in Firms Organizations an States Cambrige MA: Harvar University Press. 24. Hollingsworth B. P. awson an N. Maniaais (1999). Measurement of health care: a review of nonparametric methos an applications Health Care Management Science 2 (3) pp Ittner Christopher. an avi F. Larcer (1998). Are non-financial measures leaing inicators of financial performance? An Analysis of Customer Satisfaction Journal of Accounting Research 36 pp Ittner C.. Larcer. Taylor (2009). The stoc maret s pricing of customer satisfaction Mareting Science 28 (5) pp Johnson Michael. (1984). Consumer Choice Strategies for comparing non-comparable alternatives Journal of Consumer Research 11 pp Kamaura W. T. Lenartowicz an B. Ratchfor (1996). Prouctivity assessment of multiple retail outlets Journal of Retailing 72 (4) pp

14 29. Kamaura W. V. Mittal F. e Rosa an J. Mazzon (2002). Assessing the service-profit chain Mareting Science 21 (3) pp Keh H.T. S. Chu an J. Xu (2006). Efficiency effectiveness an prouctivity of mareting in services European Journal of Operational Research 170 pp Klein Mar an Arthur Einstein (2003). The Myth of Customer Satisfaction Briefs Spring Lewis Herbert F. an Thomas R. Sexton (2004a). Networ EA: efficiency analysis of organizations with complex internal structure Computers an Operations Research 31 pp Lewis Herbert F. an Thomas R. Sexton (2004b). ata envelopment analysis with reverse inputs an outputs Journal of Prouctivity Analysis 21 pp Luo X. an N. onthu (2001). Benchmaring avertising efficiency The Journal of Avertising Research 41 pp an C.B. Bhattacharya (2006). Corporate Social Responsibility Customer Satisfaction an Maret Value Journal of Mareting 70 (July) pp Mittal Vias Eugene W. Anerson Ain Sayra an Panu Taiamalla (2005). ual Emphasis an the Long- Term Financial Impact of Customer Satisfaction Mareting Science 24 (4) pp Morgan N.A. L.L. Rego (2006). The value of ifferent customer satisfaction an loyalty metrics in preicting business performance. Mareting Science 25 (5) pp Oliver Richar L. (1980). A Cognitive Moel of the Anteceents an Consequences of Satisfaction ecisions Journal of Mareting Research 17 (November) pp Reichhel Fran an Leonar Sasser (1996). The Loyalty Effect Harvar Business School Press Boston MA. 40. Rhoes E.L. (1982). A Stuy of U.S. National Par Service Performance Variations: an Application of ata Envelopment Analysis. Woring Paper Series No Buffalo: School of Management State University of New Yor. 41. Rust Rolan T. Christine Moorman an Peter R. icson (2002). Getting Return on Quality: Cost Reuction Revenue Expansion or Both? Journal of Mareting 66 (October) pp Rust Rolan T. Anthony J. Zahori an Timothy L. Keiningham (1994). Return on Quality: Measuring the Financial Impact of Your Company s Quest for Quality Chicago: Probus. 43. Sexton T.R. A.M. Leien A.H. Nolan S. Liss A.J. Hogan an R.H. Silman (1989). Evaluating managerial efficiency of Veterans Aministration Meical Centers using ata envelopment analysis Meical Care 27 (12) pp Sexton Thomas R. an Herbert F. Lewis (2003). Two-stage ata envelopment analysis Journal of Prouctivity Analysis 19 pp Sexton T.R. R.N. Norton an R.H. Silman (2002). Firm-Specific Prouctive Efficiency Offsets in the evelopment of a Price Cap Formula The Electricity Journal 15 (10) pp Soteriou Anreas an Stavros A. Zenios (1999). Operations quality an profitability in the provision of baning services Management Science 45 (9) pp Srivastava Rajenra K. Tasauq A. Shervani an Liam Fahey (1998). Maret-Base Assets an Shareholer Value: a Framewor for Analysis Journal of Mareting 62 (January) pp Tavares Gabriel. A Bibliography of ata Envelopment Analysis (2002) RUTCOR Research Report RRR Rutgers University Piscataway NJ. 49. Westbroo Robert A. (1981). Sources of Consumer Satisfaction with Retail Outlets Journal of Retailing 57 (3) pp Westbroo R.A. an R.L. Oliver (1991). The imensionality of consumption emotion patterns an consumer satisfaction Journal of Consumer Research 18 pp Yi Youjae (1991). A Critical Review of Customer Satisfaction Review of Mareting 1989 Valarie A. Zeithmal (Eitor) American Mareting Association: Chicago IL. 52. Zeithaml Valerie A. (2000). Service Quality Profitability an the Economic Worth of Customers: What We Know an What We Nee to Learn Journal of the Acaemy of Mareting Science Winter. 45