A dynamic model of customer loyalty an empirical

A dynamic model of customer loyalty: an empirical evidence
M.Costabile *, M.A.Raimondo**, G.Miceli**

Affiliation:
* SDA BOCCONI, Milano;
Università degli Studi della Calabria – Campus di Arcavacata
** Università degli Studi della Calabria – Campus di Arcavacata

(Track indication: Relationship Marketing)

*

Università degli Studi della Calabria – Campus di Arcavacata, Rende (CS) – 87036 –
Italy Phone and Fax: +0984/492269; E-mail:


A dynamic model of customer loyalty: an empirical evidence
Abstract
Many studies have focused the attention on defining, operationalizing and measuring the
market relationship construct. According to these studies there are cognitive, affective and
behavioral constructs (satisfaction, trust, commitment, loyalty, etc.) that qualify several
types of relationship during different stages of the life cycle, starting from customer
satisfaction and all the way through to customer loyalty and partnership in the ideal-typical
path.
This paper aims to demonstrate that a dynamic model of customer loyalty can be
conceptually defined and empirically tested. In applying a structural equation model, the
authors show evidence of the multidimensionality of customer loyalty, obtaining some
findings about the effect of time on the loyalty construct dimensions.
Search Keywords: relationship, loyalty, trust, LISREL
1. Introduction
Marketing scholars have extensively analyzed the post-purchase consumer behavior (or

customer behavior) and the customer relationship development, both in the business to
business and in the business to consumer field. Nevertheless a conceptual model widely
accepted and empirically tested on how a relationship evolves from transaction up to loyalty
is not available, yet.
On the ground of the customer satisfaction and relationship marketing literature, this
paper describes a model on the customer loyalty development. Loyalty is considered a
multidimensional construct that emerges dynamically as the optimal stage evolution of the
relationship between a customer and a firm or a brand. The model is empirically tested, and
some interesting evidences are given, thanks to a structural equation models application,
even if a longitudinal analysis, appropriate to measure the dynamic of the construct, is not
still available (a panel research is running). The findings confirm that customer loyalty is a
multidimensional construct, and each of its dimensions have a different role over the time.
2. The literature on the customer buying behaviour and loyalty
Customer satisfaction theories are based on the social and experimental psychology
studies that discovered the foundation of the ‘confirmation/disconfirmation’ paradigm
useful to explain the perception of satisfaction and the link between satisfaction and trust
(Bitner, 1995; Costabile, 1998). This connection is allowed for the interpretation of the
dynamics of market relationships, and thus for the origin of trust and loyalty, both grounded
on the experience of satisfaction accumulated over time. Since then many scholars analyzed
the determinants of satisfaction, also attempting to understand the variables that intervene
between the perception of satisfaction and the choice of repurchasing a brand, or a set of
brands (Yi, 1990, Oliver, 1997). To sum up, it is possible to claim that there are clear casual
connections between trust and repurchase decisions, and under some conditions, loyalty
(Oliver, 1999).
Grounding on the psychology of interpersonal relationships and on the social network
theory, relationship marketing scholars have investigated exchange processes in dyadic and
network contexts, showing the central role of trust. The fundamental contribution of Dwyer,
Schurr and Oh (1987) identified trust as the critical factor for moving from discrete market
transactions towards exchange relationships.
Other specific researches have given evidence of the great importance of “relational

constructs”; among others, commitment – defined by Morgan and Hunt (1994) as the
“durable desire to maintain an important relationship” – and equity – defined as “the
perception of reciprocity in the relationship with the seller” (Costabile, 2000).


The dynamic perspective in the analysis of the customer-supplier relationship was
proposed by Ford (1980). Other relevant contributions came from Dwyer, Schurr and Oh
(1987), Wilson (1995) and Iacobucci and Zerrillo (1997).
To sum up, customer behavior literature and relationship marketing literature supply a
wide and solid background to analyze the relational continuum and the variables that play a
role on the customer perceptions, attitudes, beliefs and behaviors. The purpose of this paper
is to propose and test a dynamic model of customer loyalty. The model was developed
thanks to the above mentioned literature background and was aimed to explain the customer
growing involvement in the relationship.
3. The dynamic model of customer loyalty
The dynamic model of customer loyalty (Costabile, 2000) proposed in this paper is
shown by Figure 1.
LOYALTY

tn

DYADIC VALUE ANALYSIS

REPURCHASE

TRUST

SATISFACTION
t1

t0
t2, 3, … EXPECTED VALUE

MENTAL LOYALTY

PURCHASE

PERCEIVED VALUE

tm
BEHAVIOURAL LOYALTY

MONADIC VALUE ANALYSIS

Figure 1. The dynamic model of customer loyalty.

The model concerns the following stages:
the “satisfaction to trust” stage: in exchange processes, when perceived value, in the
buyer’s perspective, meets or exceeds the expected value satisfaction is generated; the flows
of satisfaction enhance the perception of seller’s reliability, increasing the stock of positive
attitude defined as trust and allowing the development of relationship towards loyalty
(Morgan e Hunt, 1994);
• the “trust to behavioral loyalty” stage: the growth in the stock of trust has a positive
influence on the customer intention to repurchase (Boulding, Kalra, Staelin and Zeithmal,
1993), in particular thanks to the opportunities of economies on cognitive, emotional,
operational and structural costs (Costabile, 2000); considering this stage, in a dynamic
perspective, these repurchase savings generate a behavioral loyalty, dependant on
competitive and technological pressures that operate on customer involvement;
• the “mental loyalty” stage: the behavioral loyalty is not a never ending type of loyalty.
During the life cycle some kind of “conflict” could emerge (Iacobucci and Zerrillo, 1997),

in the shape of a new comparison between the value offered and experienced by the firm
and the value proposition advanced by its competitors (Woodruff and Gardial, 1996). The
positive resolution of the conflict based on monadic value analysis – regarding the ratio
“customer benefits/customer sacrifices” compared with the same ratio offered by
competitors
– pushes customers towards mental loyalty, defined as a strong belief in the seller’s capability
in offering always “the best value”, in order to satisfying the customer requirements over
time. The mentally loyal customer is a very strong committed one. He will generate positive
word-of-mouth and he will be unlikely to switch to another seller;
•


•

the “customer loyalty” stage: a long time customer has enough information to compare the
perceived value he has received from the relationship and the value that the longevity of the
latter has created for the seller. Both the common sense belief that loyalty over time creates
extra-value for the firm, and the learning opportunities about the products, the firm and its
economics that are given to the customer by a long relationship push the customer to check
the perception of equity of the relationship. At this advanced stage of the relationship, when
the customer knows very well the firm’s behavior, a dyadic value analysis becomes the
central focus of his evaluation: a second type of conflict arises with a focus on the
evaluation of the supplier “fairness” (versus “opportunism”). The positive resolution of this
second conflict – the belief that the supplier adopted a fair behavior over the time – is
determinant to reach the optimal stage of the relational continuum: the customer loyalty. A
loyal customer is a collaborative one, and his behavior will be aimed to co-operate in
building a long lasting relationship, also when contextual or competitive variables plays
against the loyalty option (proactive loyalty - Oliver, 1997 and 1999).
4. The research design
To empirically test the model a structural equation model has been applied. The

different constructs defined by the model have been operationalized and measured realizing
a survey with two distinct groups of Italian customers of mobile telephone services. These
two groups differ by the longevity of their relationship with the supplier. The “elders” are
customers that repurchase from their principal provider at least since three years; the
“youngs” are customers that repurchase from their principal provider since less than three
years. The cut off in the duration of the relationship has been defined analyzing the data,
having found a median value of three years in the time distribution concerning the
relationship length. The aggregate sample has been created adopting a criterion based on
convenience. Collection of data has been carried out in two towns, Milan – in Northern of
Italy – and Cosenza – in Southern of Italy –.
The purpose of this research design was to test the main model hypotheses defined as
follows:
HP1 – confirmation or positive disconfirmation of expectations has a positive influence on
satisfaction;
HP2 - emotions have a positive influence on
satisfaction; HP3 - satisfaction has a positive influence
on trust;
HP4 - trust has a positive influence on customer loyalty with a greater intensity in
the early stages of relationship life cycle (therefore in the “youngs”);
HP5 - trust has a positive influence on mental loyalty;
HP6 - mental loyalty has a positive influence on customer loyalty with a greater
intensity in the late stages (therefore in the “elders”);
HP7 - the buyer’s perception of equity has a positive influence on customer loyalty with a
greater intensity in the late stages (therefore in the “elders”);
HP8 - customer’s sacrifices and seller’s sacrifices (both in the customer’s perceptions)
have a negative influence on buyer’s perception of equity;
HP9 - customer’s benefits and seller’s benefits (both in the customer’s perceptions)
have a positive influence on buyer’s perception of equity;
For some of the hypotheses (HP4, HP6 e HP7) a longitudinal study would be
appropriate. A cross sectional research does not give the opportunity to verify if the

differential intensity of some connections was already high when the “elder” customers
were “youngs”, and viceversa (“young” customers will show the same intensity of the
hypothesized connections when they will become “elders”?). Nevertheless a longitudinal
study require an early test of the model aiming to verify if the loyalty construct is a
multidimensional one and if there are


significant differences over the time. Then it will be possible to design a longitudinal study
to isolate the effect of time.
By the way, an exploratory factor analysis – explained variance = 55.46%; sig. = .000 –
has been run to purify measurement scales, built both on the ground of a literature review
and on the accomplishment of some qualitative interviews to generate contextual items (i.e.
variables affecting satisfaction towards mobile phone services). Promax rotation criterion
shows a factorial structure with no cross-loadings and with significant loadings on the
hypothesized latent factors; in a comprehensive view, very good levels of discriminant and
convergent validity are obtained. Satisfaction has been measured with an overall single item
because of the bad factorial performance of the scale proposed by Oliver (1993). The
satisfaction item hasn’t any significant loading on the other constructs. Cronbach alphas
grater than .71 have given evidence of reliability in all measurement scales. Grounding on
Bagozzi and Heatherton (1994) and considering the complexity of the model it has been
applied a partial aggregation approach on observed variables: eight of ten multiple-items
measurement scales have been synthesized into two indicators - our Y and X in our
structural equation model tested - by a process based on simple means on two, three or four
observed variables. Constructs have been operationalized on the following basis:
• Customer Loyalty (CL): two indicators (Y8 and Y9) are the output of the aggregation
process of the items regarding repurchase intention, word-of-mouth, spontaneous
collaboration, positive answer to requested collaboration, and two brand identification
items; this construction is based on a multidimensional definition of loyalty (Morgan and
Hunt, 1994; Dwyer, Schurr and Oh, 1987; Wilson, 1995; Costabile, 2000).
• Customer Satisfaction (SA): grounding on Oliver (1993) we have operationalized

Customer Satisfaction – as a per-se construct – distinguishing it from its antecedents,
from disconfirmation of expectations and from emotions; as above mentioned given the
bad factorial performance of items used by Oliver, Customer Satisfaction has been
measured with a single overall item (Y1).
• Trust(TR): five items expressing reliability and trustworthiness of the supplier (Morgan
and Hunt, 1994; Smith and Barclay, 1997) have been aggregated on two indicators (Y2
and Y3).
• Mental Loyalty (ML): three items concerning the provider capability in offering good
prices, promotions, and new value added services over time have been aggregated on
two indicators (Y4 and Y5).
• Perception of Equity (EP): eight items concerning reciprocity in benefits, revenues and
sacrifices, in the customer perspective (Oliver e Swan, 1989), have been aggregated on
two indicators (Y6 and Y7).
• Disconfirmation (DI): three items regarding the provider capability in confirming
expectations about good prices, promotions and new value added services (Oliver, 1980
and 1997) have been aggregated on two indicators (X1 and X2).
• Emotions (EM): seven semantic-differential items concerning emotions that Oliver (1993
and 1997) found as good antecedents of satisfaction and that have been confirmed in
qualitative interviews have been aggregated on two two indicators (X3 and X4).
• Customer Sacrifices (CS): two items (X5 and X6) have been used to operationalize the
information costs and the competitor evaluation costs.
• Customer Benefits (CB): six items regarding the level of perceived customer care, the
completeness of information, the goodness of communication, the problem-solving
capability, the readiness of innovations and the attention paid to the customer
requirements have been aggregated on two indicators (X 7 and X8).
• Supplier Sacrifices (PS): three items concerning the provider’s investments in advertising,
promotions and marketing research have been aggregated on two indicators (X9 and X10).


• Supplier Benefits (PB): two items (X11 and X12) have been used to operationalize the greater

quantity of purchased services, and the positive worth-of-mouth stemming from long- term
relationships with the firm.
5. Research findings
Results of running LISREL 8.50 syntax are shown in Table 1. In both analyses,
model fit was acceptable, with better values in the “elders” group. This means that the
model proposed is a good approximation to the data, even if some improvements are
further possible. Hypotheses are not completely confirmed. Some structural parameters
are not statistically significant, this is probably, the expression of a lack in the
measurement process, more than in the specification of the model, given the necessary
hard purification of measurement scales. Nevertheless, the central hypotheses of the
model are confirmed in direction, sign and strength; these are the hypotheses
concerning the role of trust, monadic value and dyadic value in determining customer
loyalty. The findings support the stronger influence of trust on customer loyalty during
the early stages of the relationship than in the late ones (Table 1: TR€CL); and the
greater impact of monadic and dyadic value during the advanced stages of the relationship life
cycle than in the initial one (Table 1: ML€CL and EP€CL). The causal path from
mental loyalty and from equity perception to customer loyalty are not significant in
“youngs” group - the null hypothesis that the relative parameters are equal to zero
can not be rejected -: this could be explained considering that in the early stage of the
relationship the customer is not self-confident in giving evaluations on monadic value
and on dyadic value. In spite of a good factorial performance, emotions have not been
confirmed as a significant antecedent of customer satisfaction.
Table 1. Results of LISREL application

Paths

DI €
EM €
SA €
TR €

TR €
ML €
EQ €
CS €
PS €
CB €
PB €

SA
SA
TR
CL
ML
CL
CL
EQ
EQ
EQ
EQ

“Elders” (N=246)

“Youngs” (N=219)

χ2 (158 df) = 249.85

χ2 (158 df) = 309.87

NFI = .92 NNFI = .96 CFI = .97

NFI = .88 NNFI = .91 CFI = .94

GFI=.91 AGFI= .87 RMR = .06

GFI=.88 AGFI= .82 RMR = .08

Structural
parameters
.93*
.07 (ns)
.94*
.55*
.59*
.20*
.28*
-.12***
-.12 (ns)
.62*
.14 (ns)

Hypotheses
HP1 supported
HP2 not supported
HP3 supported
HP4 supported
HP5 supported
HP6 supported
HP7 supported
HP8 partially
supported

Structural
parameters
.98*
- .08 (ns)
.94*
.67*
.62*
.11 (ns)
.12 (ns)
-.05 (ns)
.17 (ns)
.30*
.19**

Hypotheses
HP1 supported
HP2 not supported
HP3 supported
HP4 supported
HP5 supported
HP6 supported
HP7 supported
HP8 not supported

HP9 partially
HP9 supported
supported
ns = not significant; * significant at p <.01 level; ** significant at p < .05 level; *** significant at p < .08 level

Consistently with the satisfaction literature (Yi, 1990; Oliver, 1997; Costabile, 1998),
confirmation or positive disconfirmation of expectations has been verified as a strong
determinant of customer satisfaction. Measurement of trust has been the best one in the
whole analysis, despite the multiform contributions in the relationship marketing literature.
The reliability/trustworthiness definition, consistent with the model (Costabile, 2000), has
been supported by the analysis and offers a clear signal for the unidimensional definition of
trust.


A significant improvement in the measurement of equity has been given in respect to
findings of Oliver and Swan (1989); their results enhanced a “preference” form of equity.
The partial support offered by our results to the equity condition – in which the ratios
“benefits/sacrifices” of both parts of a dyad might be equal – gives new stimuli for further
researches.
The measurement model has pointed out a good overall performance. All lambda
parameters have resulted significant and consistent with the hypothesized constructs.
Customer loyalty has been confirmed as a multidimensional construct by the good
measurement performance, in accordance with the theoretical basis of the model. In regards
to “elders” group, only in four out of twenty-one cases the indicators error variance is
greater than the squared multiple correlation; in “youngs” group, this happened in five cases
out of twenty-one. Squared multiple correlations for structural equations in both groups are
acceptable, except for the one concerning the mental loyalty equation – in both groups –
and for the one concerning the perception of equity equation – in the “youngs” group –.
This condition signals a lack in the definition of constructs, due to a hard purification of the
measurement scales in the preliminary analysis.
Further to these considerations, more analyses are needed in order to improve
operationalization process of relational constructs and to clarify how the strength of
relations among loyalty and its dimensions can vary over time. A longitudinal study is
needed.
6. Conclusive remarks

The dynamic model of customer loyalty offers new insights in understanding the life
cycle of the customer: from transaction to loyal relationship. Its dynamic nature would
require a longitudinal study, but, so far, the multidimensionality of the customer loyalty
construct, and the different role of the construct dimensions over the time have been
empirically tested.
Considering the future directions the adoption of a dynamic perspective both in the
academic research, that should focus on longitudinal analysis of such a kind of construct,
and in the loyalty management practices can be considered very relevant. Many loyalty
programs and customer relationship management projects does not take into account the
time effect on the customer perceptions and behavior, so firms do not differentiate
effectively their loyalty policies.
7. Key References
Bagozzi, R.P., Heatherton, T.F., 1994. A general approach to representing multifaceted personality constructs:
application to state self-esteem. Structural Equation Modeling 1, 35-67.
Costabile, M., 1998. Customer satisfaction and trust into the resource-based perspective. Reseach propositions
endorsing the confirmation/disconfirmation paradigm. Competitive paper presented at the 27 th EMAC
Conference, 20th-23thof May, Stockholm.
Costabile, M., 2000. A Dynamic Model of Customer Loyalty. Competitive paper presented at the 16 th Annual
IMP Conference, Bath (UK), 7 th-9 th of September.
Dwyer, F.R., Schurr, P.H., Oh, S., 1987. Developing buyer-seller relationships. Journal of Marketing 51, 11-27.
Ford, D., 1980. The development of buyer-seller relationship. European Journal of Marketing 14, 339-354.
Morgan, R.M., Hunt, S.D., 1994. The commitment-trust theory of relationship marketing. Journal of Marketing
58, 20-38.
Oliver R.L., 1993. Cognitive,Affective and Attribute Bases of the Satisfaction Response. Journal of Consumer
Research 20, December, pagg. 418-430.
Oliver, R.L., 1997. Satisfaction. A behavioral perspective on the consumer. McGraw-Hill, New York.
Oliver, R. L., 1999. Whence Consumer Loyalty?. Journal of Marketing, special issue, 33-44.
Oliver, R.L., Swan, J.E., 1989. Consumer Perceptions of Interpersonal Equity and Satisfaction in
Transactions: A Field Approach., Journal of Marketing, April, 21-35.
Yi, Y., 1990, A Critical Review of Customer Satisfaction, in V.A. Zeithaml (edited by), Review of Marketing.

68-123, AMA Chicago.
Wilson, D.T., 1995. An Integrated Model of Buyer-Seller Relationship. Journal of the Academy of marketing
Science, 4, 335-345.