Investigating Consumer Responsiveness to Service Contract Attributes: a Choice Experimental Approach

Harmen Oppewal, University of Surrey
Davie J.I. Grant, University of Sydney
ABSTRACT - This paper investigates consumer preference for retail service contracts such as involved in the purchase of mobile phones, pay-TV, or rental apartments and demonstrates how choice experiments can be used for this purpose. A sample of 122 Australian teachers completed a choice experiment which tests effects of entry fee, monthly cost, contract specificity, and the transferability of a durable component. We find several effects on contract preference in general and some effects on preference for contract lengths. Results confirm that expected price increases lead to an increased preference for longer-term contracts; an opposite effect for price decreases is not found.
[ to cite ]:
Harmen Oppewal and Davie J.I. Grant (2002) ,"Investigating Consumer Responsiveness to Service Contract Attributes: a Choice Experimental Approach", in NA - Advances in Consumer Research Volume 29, eds. Susan M. Broniarczyk and Kent Nakamoto, Valdosta, GA : Association for Consumer Research, Pages: 133-138.

Advances in Consumer Research Volume 29, 2002     Pages 133-138

INVESTIGATING CONSUMER RESPONSIVENESS TO SERVICE CONTRACT ATTRIBUTES: A CHOICE EXPERIMENTAL APPROACH

Harmen Oppewal, University of Surrey

Davie J.I. Grant, University of Sydney

ABSTRACT -

This paper investigates consumer preference for retail service contracts such as involved in the purchase of mobile phones, pay-TV, or rental apartments and demonstrates how choice experiments can be used for this purpose. A sample of 122 Australian teachers completed a choice experiment which tests effects of entry fee, monthly cost, contract specificity, and the transferability of a durable component. We find several effects on contract preference in general and some effects on preference for contract lengths. Results confirm that expected price increases lead to an increased preference for longer-term contracts; an opposite effect for price decreases is not found.

INTRODUCTION

Many consumer services involve some ongoing relationship between service providers and consumers. Ways of establishing and maintaining these relationships have been an important topic of study in the service literature, as customer switching typically hasnegative effects on market share and profitability (Keaveney, 1995; Rust, Zahorik and Keiningham, 1995). Recent studies in this area have focussed on the distinction between transactional based and more relational based orientations of customers (Crosby, Evans and Cowles, 1990; Dwyer, Schurr and Oh, 1987; Garbarino and Johnson, 1999). Recent research has mostly focussed on relational orientations but there is also an extensive literature on transactio cost theory and its implications for business organizations (see Rindfleisch and Heide, 1997). Little research seems to have been done however on transaction-based aspects of consumer services, such as involved in service contracts.

Contracts represent a particular but widespread way of establishing and maintaining a transactional or functional relationship with customers. Many services can only be used if consumers agree and sign a contract, often after a trial period in which the service could be used at a reduced fee. Contracts specify the rights and duties of the provider and consumer and typically have a fixed time horizon. As such, the contract specifies the service that is bought by the consumer. Mobile phones, internet access, and cable or pay-TV subscriptions are examples of service contracts that typically have a fixed time period; other examples are real estate rental contracts, transport travel passes, and many financial services such as credit cards and insurance.

The literature in marketing has to our knowledge not addressed the problem of how to model consumer choice between contracts of a given length. The literature on stockpiling behavior and purchase quantity (eg, Chintagunta, 1993; Cripps and Meyer, 1994; Meyer and Assuntpo, 1990) has addressed the related but different problem of how to analyze continuous dependent measures that, by the way, were mostly observed for durables and consumer goods and not for services. The methodological concerns in the service literature have mostly been with service quality measurement and structural equation modeling, but not so much with choice.

In this paper we test ideas about contract preference and we explore and demonstrate how choice experiments (Louviere and Woodworth, 1983) can be used to model consumer choice of service contracts. We derive hypotheses for the effects of fixed versus variable service contract costs, uncertainty about the service performance, and the costs of switching providers in terms of asset specificity on general contract preference and on preferences for contracts of a specific length. In addition we test the effects of price expectations on consumer preference for contract length. To test these hypotheses, we set up a choice experiment in which we vary the various contract features within each of three price expectation conditions. The choice experiment involves pay-TV contracts and is conducted in Australia.

The paper is organized as follows. We first derive our hypotheses. We then present the details of our application to pay-TV contracts. After presenting the method and results, we discuss our results and their implications for further research into the area of service contracts.

FACTORS AFFECTING CONTRACT PREFERENCE

Various factors can be expected to affect consumers’ preference for service contracts. From the perspective of the consumer, the decision whether to sign a service contract of certain length involves a trade-off between various costs, risks, and benefits. Transaction costs are one of the costs involved in arranging or rearranging a service contract and clearly one would expect that a contract with lower transaction costs is more attractive than one with higher transaction costs. One way in which a consumer can reduce transaction costs is by purchasing a longer term contract, as renewals and contract extensions of shorter term contract all incur some kind of transaction cost in addition to the initial transaction cost. Our first hypothesis is therefore that increased contract renewal costs make contracts less attractive, especially if these contracts are relatively short-term (H1).

Initial or contract entry costs will also affect contract preference. Larger entry costs will result in a lower overall preference for the contract. In addition however, we expect that higher entry costs will make shorter-term contracts particularly unattractive (H2). This is because with a longer contract the fixed costs of the transaction can be offset against a longer usage period.

One source of risk involved in signing a contract concerns uncertainty of service performance during the contract period. The consumer purchases the service in advance but, by the nature of the service, the consumer will remain uncertain about the actual service quality until the service has been delivered. The longer the contract, the larger is this risk. Consumers would therefore tend to prefer shorter contracts. We also expect however that if a contract is more specific with respect to the quality of the service to deliver then consumers are more willing to sign up for a longer term (H3). This is because the risk of a deterioration of service quality over time is more limited if the terms of delivery are specified in the contract.

A specific feature of many services is that in order to be able to use the service, the consumer has to purchase a durable component. One example of such "bundled systems" (Wilson, Weiss and George, 1990) of durable and service components is a pay-TV package where the cable access is bundled with a period of programming (cf. Chae, 1992). The durable (cable) is a one-off purchase whereas the service (programming or communication service) is paid for on a period by period basis as designated by the contract, which can be renewed at the completion of each period. If the consumer does not want to renew but would prefer to switch to another provider, then switching costs will be involved. The switching costs are lower if the durable is compatible with the competing provider’s system than if the durable is not transferable. In the latter case the durable component is a "relationship specific asset" (e.g., Crawford, 1990; Rindfleisch and Heide, 1997) that no longer has any value to the consumer if the contract is terminated. Hence, the durable price is a sunk cost.

Under conditions where the durable component is market compatible, this problem does not exist for there is the option to switch to a competitor’s product. Crawford (1990) argues that in contracts between businesses the existence and size of a "relationship specific asset" is positively related to the length of a contract. For example, the supplier of headlamp fittings to a car manufacturer requires the purchase of industry specific machinery. The supplier favors a long-term relationship for it is dependent upon the manufacturer as the sole purchaser. Had a long-term contract not been negotiated, then the risk of investment might have been too great and the contract not signed. We hypothesize that in retail service contexts this condition will also hold. Consumers investing more in the "system specific durable" will prefer to lock the supplier into a longer term agreement to reduce risk of changes to conditions at a re-negotiation point in the future. Therefore, if the durable is nontransferable, there is an increased reluctance to buy a contract, especially if the price of this durable is high. This effect will in particular occur when the contract is of short duration (H4).

A final factor that we consider is market price expectation. If prices are expected to increase, one can expect that consumers are more willing to sign up for a contract than if prices are expected to fall. This is because delay of purchase in conditions of increasing prices results in higher purchase costs when the product is bought eventually. Conversely, when prices are expected to go down it will pay off to wait before taking out the service on offer. When prices are expected to rise it will also be more attractive to purchase a contract of longer duration than if prices are expected to fall (cf. Hutchinson and Meyer, 1994; Jacobson and Obermiller, 1990; Krishna, 1994; Loewenstein, 1988). This inclination to buy now if prices rise not only affects contract preference in general but also the preference for contract duration, longer contracts being more attractive when prices are expected to rise (H5).

In this paper we set out to test these ideas within the framework of random utility theory using the multinomial logit model (MNL; e.g., Ben-Akiva and Lerman, 1985). We design a choice experiment in which respondents choose between contracts of different length, or they choose to not purchase a contract (Louviere and Woodworth, 1983). In the random utility framework the utility derived from any alternative is a function of this alternative’s attributes. The utility for each alternative has a stochastic component that is assumed independent and following an identical extreme value Type I distribution. This assumption leads to the well-known formulation for the choice probability for alternative i given choice set A:

EQUATION

In line with most experimental choice studies, we will assume that the deterministic component of the utilities Vj is a linear function of the alternatives’ attributes xjk, which represent the factors outlined above:

EQUATION

where the ¯’s are the parameters to estimate. Our hypotheses above will be tested by testing for the significance of the corresponding attribute parameters in our models.

METHOD: THE CASE OF PAY-TV CONTRACTS

Pay-TV contracts are a relatively new product in Australia. This category was selected because the package designs sold in the Australian market are consistent with a bundled durable/services system and many urban respondents have had to make decisions about pay-TV purchases within the last couple of years.

Choice task design.

In order to build a predictive model that allows testing of the various hypotheses, respondents were asked to make a choice in scenarios that each present one six-month and one one-month pay-TV package (or contract) (see Figure 1). Respondents were asked their preference for six-month and one-month contracts and could choose to buy either the six-month, one-month or not choose to buy a package.

The attributes included and the design developed was an outcome of the hypotheses to be tested, a reflection of real pay-TV packages available on the market and a result of ensuring the task was not beyond the consumers time commitment, or involvement. There were seven attributes included in the task for each of the two contract lengths. Each attribute was varied at two levels. A description of each of these attributes and the levels was provided for the respondents and is displayed in Figure 2.

The attributes included a "connection fee," a "service charge," conditions with regard to "transferability of service to competitors," and "guarantees of programming over the period of the contract". Four programming bundles were offered, which closely resembled bundles offered by a major pay-TV operator in the study area. The packages offered were a "standard," a "movie," a "sports" and a "deluxe" package. The latter three packages extended the standard package to include movies, or sports programming or both, respectively. There was a fixed price premium for the movie, sports and deluxe packages that reflected current market prices for such programming. Respondents were provided with a pamphlet that outlined the programming for the respective bundles. Price was varied for the one and six-month packages separately over two levels. Respondents were told that they would pay for the service at the beginning of each month. Respondents were also made aware of the penalty for discontinuing the six-month package before the contract had ended. This penalty was the equivalent to the value of the remaining months in the contract.

A fraction of the 214 factorial design with 32 treatments (scenarios) was set up to independently vary the attributes within and across the one- and six-month contracts. The design allows the estimation of main effects and selected two-way interactions independently of one another but not of unobserved higher order interactions. To reduce the task of the respondents from evaluating thirty-two scenarios to sixteen an additional independent design factor was used to divide the thirty-two scenarios into blocks of sixteen. The order of the scenarios within these blocks was randomized.

The design in thirty-two scenarios was nested under a between-subjects master design consisting of three different price expectation conditions. The price expectation conditions controlled for consumer price expectations of pay-TV service costs through the use of a mock newspaper article. The three conditions were an expected price rise of 15 percent over the next few months, no change to the price and an expected price decrease of 15 percent over the next few months. Respondents were randomly allocated to one of these conditions.

FIGURE 1

EXAMPLE CHOICE SET

FIGURE 2

ATTRIBUTES AND ATTRIBUTE LEVELS

Procedure

The experiment took the form of a self-completion take home survey and consisted first of general questions regarding TV watching, experience of and attitude towards pay-TV. The mock newspaper article outlining "expert" opinions of future price behavior for pay-TV followed this. Respondents were then asked about their expectations for price of pay-TV in the next few months. After seeing a mock pamphlet detailing the four pay-TV packages, respondents saw sixteen different choice sets. For each they were asked to indicate whether they preferred the presented six month or one month pay-TV contract and whether they would choose to buy the six month, one month, or not make a purchase. The questionnaire concluded with a series of questions concerning the basic demographics of the respondent.

Sample

A convenience sample of staff at an Australian secondary school was asked to participate in the experiment. This sample represented a cross section of people who would have been or in the nar future would be making a decision about pay-TV purchasing. As an incentive, a $100 raffle was offered. Of 135 staff, 122 completed the questionnaire. The age range in the sample was 17 to 65 years, with the mode lying in the 36-45 age group. Gender was distributed 60:40 men to women. A high proportion of the sample had experienced pay-TV with over 68 percent indicating that they had watched pay-TV in the past. The sample generally however had not formed strong attitudes towards pay-TV relative to free-to-air TV. This was evidenced by a tendency of respondents to circle the neutral response on a battery of Likert scale statements regarding pay-TV’s performance relative to free-to-air TV. When asked, 45 percent of respondents indicated that they would consider buying pay-TV in the future, 55 percent indicated that they would not. When asked about how long they would expect to sign up for pay-TV if they did undertake a contract, answers ranged from a month to two years. The most common answer to how long they expected to buy a contract for was a year; 16 percent expected to sign a contract for six months.

ANALYSIS AND RESULTS

The effects of the various attributes manipulated in the experiment correspond to the hypotheses we formulated. To test our hypotheses, we therefore estimate and compare various specifications of the MNL model. We first however tested the effects of the manipulated price expectations.

Effects of Price Expectations

As a check on the manipulation of price expectations an ANOVA was run upon people’s stated expectations of price movement. This analysis indicated that the assigned price trend conditions had created the intended differences in price expectations. The mean expected price trends were significantly different (F=55.66; d.f.=2,117; p<.001). The mean price expectation in the 'price fall’ condition was a decrease by 11.1 percent, the mean for the control condition was 2.8 percent price increase, and the mean expectation in "price rise" condition was that the contract prices would show a 12.6 percent increase in price.

TABLE 1

PARAMETER ESTIMATES FOR BINARY CHOICE MODELS

Model Estimations

First a model was estimated from the binary choices regarding the most preferred contract. The model results are displayed in Table 1. The model fits the data well (Rho2=.40) and predicts better than chance (Chi2 is 1700.28, with 12 d.f., p<.001). We also estimated a model with all parameters made specific for the two different contract lengths. Though this model did not perform significantly better than the constrained model (Chi2 s 1710.08, with 21 d.f.; the model improvement is 9.08, with 9 d.f., n.s.), it does show where differences for specific attributes occur between the respective contract lengths.

Table 1 shows that several factors are having a significant impact on consumer contract preference. These are price rise expectations, sports and movie package dummies, connection fees, transferability, programming guarantees, monthly fee, and the interaction between durable price (connection fee) and transferability (programming). In the model with specific parameters for the one-month and six-month contracts the preference for a one-month contract depends on these same factors, except that transferability is not significant and the connection fee by transferability interaction term is only marginally significant in this model. The six-month contract is only significantly dependent on the sports and movie package dummies and the connection fee.

Hypothesis Tests

To interpret the results the appropriate dummy values need to be substituted into the equation. This firstly reveals that the effects of the service charge on contract preference are only marginally significant and are not significantly different between the one-month and six-month contracts. Hypothesis H1 is therefore not confirmed.

A high connection fee of $50 (instead of $20) decreases the utility of a contract relative to the other contract but there is no significant difference between these parameters for the one-month and six-month contract. We thus cannot confirm hypothesis H2 that the connection fee has a larger impact on the shorter-term contract.

Offering a programming guarantee has the effect of increasing the utility of the contract. In contrast to our expectations this effect specifically occurs for the one-month and not for the six-month contracts. So, programming guarantees make a contract more attractive but there is no evidence to support hypothesis H3 that programming guarantees lead to an increased preference for a longer-term contract.

The interaction between the durable price and the transferability is found to influence the purchase of a contract as hypothesized in H4. The effect can be seen in Figure 3, which indicates that the likelihood of being preferred decreases at a greater rate as connection fees rise for the compatible (i.e., transferable) product relative to the product with a non-transferable durable component. This effect seems to occur in particular for the one-month contract, for which we observe a marginally significant effect. However, the effect is not significantly different between the one-month and six-month contracts.

A four-dollar increase in monthly fee decreases the utility of a contract. When tested separately for the one-month and six-month contracts there is a significant effect for the one-month contract but no such effect is observed for the six-month contract, though the six-month contract concerns a commitment to pay this amount six times. This indicates that price-sensitivity is larger for a shorter-term contract.

Finally, as consumers expect price to rise by 15 percent, their utility for the one-month contract decreases, thereby reducing the odds of preferring the one-month contract relative to the six-month contract. However, as they expect price to decrease by 15 percent, their utility does not change relative to the no-change condition. Hence, our hypothesis H5 about the effect of price expectations is confirmed for price rises but not for price falls.

FIGURE 3

CONNECTION FEE AND COMPATIBILITY EFFECT ON CONTRACT UTILITY

We also attempted to estimate a model that relates to consumers’ choices between the six-month, one-month and the no contract option. The choice model explained a reasonable amount of variation in the choices of the respondent (Rho2=.41) and predicted better than chance with an observed Chi2 of 1747.60 at 22 degrees of freedom (the corresponding 95 percent critical Chi2 value is 33.92). There were however only very few parameters significant in this model (the constants, connection fees, and movie and sports package dummies), none of which related specifically to our hypotheses.

The lack of results for this model is most likely due to the large number of choice sets resulting in a "no contract" selection. Across all choice sets, 79 percent of choices were for the "no contract option" where as 12 percent of selections was for the six-month option and 9 percent for the one-month option. So, the respondents expressed an overall reluctance to sign up for contracts under the specified conditions. Noteworthy in this respect is that the number of purchases was larger for the six-month contracts than for the one-month contracts. This suggests that a one-month trial is only attractive to a minority in our sample and only under specific circumstances.

CONCLUSIONS AND DISCUSSION

This paper aimed to test several ideas regarding consumer preference for service contracts. To test the hypotheses a choice experiment was designed for the case of pay-TV. Pay-TV is just one example of a service contract and the approach that we demonstrated can be readily applied to other services, such as mobile phones, rental apartments, internet service providers, rental cars and other equipmentCthough of course each new application will require that appropriate stimuli are carefully developed. The ideas to test concerned effects of transaction costs, transferability of a durable component, the specificity of the contract and consumers price expectations on consumer preference of contracts of certain duration.

We estimated a choice model based on respondents’ choice of their most preferred contracts in each choice set. The results only partly confirmed the hypotheses. Our first hypothesis was that a higher service charge would make contracts less attractive, in particular short-term contracts. No effects were found however for the service charge attribute in this experiment. A possible explanation for this is that respondents overlooked that even though the service charge seems small, the five dollars is charged every time a contract is renewed.

The second hypothesis was that higher entry costs would make a contract less attractive, again, especially if the contract is for a short term only. The results from the experiment partly support this idea as there was a significant and negative effect of entry costs (connection fee) on contract attractiveness, however this effect was not significantly different for short and longer term contracts. Programming guarantees were expected to make the contracts more attractive, especially a longer term contract, because it reduces risk and uncertainty about the service delivery. We indeed found that contracts that were giving specific guarantees about the service delivery are more attractive, however again we found no difference between short and longer-term contracts.

We also expected that a service contract that requires a consumer to invest in the service through the purchase of a durable component would be more attractive if the durable is transferable to other products and companies. In addition we expcted that the required purchase of such a relationship-specific asset would make consumers prefer a longer-term contract. Our results did show the hypothesized interaction of a larger price effect occurring for a non-compatible durable, but the difference in effect size between the short and longer-term contract was only marginally significant.

Our final hypothesis concerned the effects of consumer expectations about the market price trend for the service. Expected price falls would result in consumers giving more preference to shorter-term contracts and/or in consumers delaying their purchase. Expected price increases were supposed to have an opposite effect. Our results confirm these ideas for expected price risesCin these conditions there was a higher preference for longer-term contracts. Expected price falls had no effect however and the hypothesized effects on overall purchase did not show up. The purchasing of a longer contract when consumers expect prices to rise is consistent with the argument that expected price rises relative to the contract price are framed as a potential gain to the consumer (Thaler, 1985). This indicates that consumers do adopt "service stockpiling strategies" in a similar way to that which was suggested by Meyer and Assuntpo (1990) for consumer goods.

While the observation that consumers frame and act upon gains was consistent with our price expectation hypothesis, the results relating to price expected falls were not. It may be that the optimal strategy a consumer wishes to make when expecting a price fall is the delay of a purchase, however we did not observe effects to support this idea in the choice models.

To summarize, most of the contract attributes were found to affect contract preference but only few affected the choice between short and longer-term contracts. This may be due however to our choice to compare one and six month contracts, one month being too short for most people to consider even as a trial, and possibly to some of the manipulations being too weak. In addition, the overall reluctance of our sample to sign up for pay-TV seems to have hampered this study. Despite all this, the impact of the price-related attributes varied with the manipulated price range. This suggests that providers of bundled durable/services systems should give careful consideration to the pricing policies of their systems. There may be scope to induce customers to purchase packages through mixed leader strategies, where discounts upon the durable are offset against premiums on the service fee (Guiltinan, 1987).

One issue to address in further research is that choice of contract length might be determined by a difference between planning horizons of a repeat purchase and that of a trial purchase. Evidence emerging in the open section of the survey indicated inherent differences in contract preference even within trial purchase occasions. Some respondents reported that they would purchase for only one month on the basis that they simply wished to trial the product. Others said that nothing short of six months was an appropriate time period for trial of the product. Other examples that consumers identified as influencing the choice of contract duration included the seasonal nature of sport, and patterns of holiday television watching. Evidently there may be heterogeneous factors at work in the marketplace that play a significant role in driving a purchase decision amongst contract lengths. These factors were not accounted for in the present models, which reduced the statistical power in our study. A study focusing upon actual users of pay-TV may also provide more conclusive results. Future studies could also focus on applications such as mobile phones or rental apartments as an area in which to demonstrate and test the general ideas we posed about consumer preference and service contract length. By conceptualizing service contract length as a product pre-purchase and by demonstrating how choice experiments can be applied to this area, we hope this paper will stimulate further research in the area of consumer behavior regarding retail service contracts.

REFERENCES

Ben-Akiva, Moshe and Steven R Lerman (1985), Discrete Choice Analysis. Cambridge, MA: MIT Press.

Chae, Suchan (1992), "Bundling Subscription TV Channels, A Case of Natural Bundling," International Journal of Industrial Organization, 10, 213-230.

Chintagunta, Pradeep K. (1993), "Investigating Purchase Incidence, Brand Choice and Purchase Quantity Decision of Households," Marketing Science, 12 (2), 185-208.

Crawford, Vincent P. (1990), "Relationship-Specific Investment," The Quarterly Journal of Economics, 105 (May), 561-574.

Cripps, John D. and Robert J. Meyer (1994), "Heuristics and Biases in Timing the Replacement of Durable Products", Journal of Consumer Research, 21, 304-318.

Crosby, Lawrence A., Kenneth R. Evans, and Deborah Cowles (1990), "Relationship Quality in Services Selling: An interpersonal Perspective," Journal of Marketing, 54 (July), 68-81.

Dwyer, F. Robert, Paul H. Schurr, and Sejo Oh (1987), "Developing Buying-Seller Relationships," Journal of Marketing, 51 (April), 11-27.

Garbarino, Ellen and Mark S. Johnson (1999), "The Different Roles of Satisfaction, Trust, and Commitment in Customer Relationships," Journal of Marketing, 63 (April), 70-87.

Guiltinan, Joseph P. (1987), "The Price Bundling of Services: A Normative Framework", Journal of Marketing, 51 (April), 74-85.

Hutchinson, Wesley and Robert J. Meyer (1994), "Dynamic Decision Making: Optimal Policies and Actual Behavior in Sequential Choice Problems," Marketing Letters, 5 (4), 369-382.

Jacobson, Robert and Carl Obermiller (1990), "The Formation of Expected future Price: A Reference Price for Forward-Looking Consumers," Journal of Consumer Research, 16 (March), 420-432.

Keaveney, Susan M. (1995) "Customer Switching Behaviour in Service Industries: An Exploratory Study," Journal of Marketing, 59 (April),17-82.

Krishna, Aradhna (1994), "The Effect of Deal Knowledge on Consumer Purchase Behaviour," Journal of Marketing Research 31 (February), 76-91

Louviere, Jordan J. and George Woodworth (1983), "Design and Analysis of Simulated Consumer Choice or Allocation Experiments: An Approach based on Aggregate Data," Journal of Marketing Research, 20 (November), 350-367.

Loewenstein, George F. (1988), "Frames of Mind in Intertemporal Choice," Management Science, 34 (2 February), 200-214.

Meyer, Robert J and Joao Assuntpo (1990) "The Optimality of Consumer Stockpiling Strategies," Marketing Science, 9 (Winter), 18-41.

Rindfleisch, Aric and Jan B. Heide (1997), "Transaction Cost Analysis: Past, Present, and Future Applications," Journal of Marketing 61 (October), 30-54.

Rust, R. T., Zahorik, A. J. and Keiningham, T. L. (1995), "Return on Quality (ROQ): Making Service Quality Accountable," Journal of Marketing, Vol. 59 (April), 58-70.

Thaler, Richard (1985) "Mental Accounting and Consumer Choice," Marketing Science, 4 (Summer), 199-214.

Wilson, Lynn O, Allen M. Weiss, and John George (1990), "Unbundling of Industrial Systems," Journal of Marketing Research, 27 (May), 123-13.

----------------------------------------