Telecommunications Policy 25 (2001) 249}269

Customer retention, loyalty, and satisfaction in the German

mobile cellular telecommunications market
Torsten J. Gerpott *, Wolfgang Rams
, Andreas Schindler
Telecommunications Management, Department of Business Administration, Gerhard-Mercator-University Duisburg,
Lotharstr. 65, D-47057 Duisburg, Germany

Deutsche Telekom AG, Friedrich-Ebert-Allee 140, D-53133 Bonn, Germany
Deutsche Telekom MobilNet GmbH, Landgrabenweg 151, D-53227 Bonn, Germany
Received 1 September 2000; received in revised form 4 December 2000; accepted 8 December 2000

Abstract
Customer retention (CR), loyalty (CL), and satisfaction (CS) are important (intermediate) goals for
telecommunication network operators on their way to superior economic success in the liberalised German
market. Therefore, drawing on a sample of 684 residential customers of digital cellular network operators in
Germany this study tests hypotheses suggesting that CR, CL, and CS should be treated as di!erential
constructs which are causally inter-linked. LISREL analyses support a two-staged model in which overall CS
has a signi"cant impact on CL which in turn in#uences a customer's intention to terminate/extend the
contractual relationship with his mobile cellular network operator ("CR). Mobile service price and
personal service bene"t perceptions as well as (lack of) number portability between various cellular operators
were identi"ed as supply-related variables with the strongest e!ects on CR. Mobile network operators'
perceived customer care performance had no signi"cant impact on CR. The "ndings suggest that an important
lever for regulators to promote competition in cellular markets is the enforcement of e$cient number
portability procedures between mobile network operators.  2001 Elsevier Science Ltd. All rights reserved.
Keywords: Customer loyalty; Customer retention; Customer satisfaction; Mobile communications; Number portability;
Telecommunications marketing

1. Background
Since the 1990s, the telecommunications sector has become a dynamic key area for the economic
development of industrialised nations. This is the result of enormous technical progress as well as of
* Corresponding author. Tel.: #49-203-379-3109; fax: #49-203-379-2656.
E-mail addresses: [email protected] (T.J. Gerpott), [email protected] (W. Rams), [email protected] (A. Schindler).
0308-5961/01/$ - see front matter  2001 Elsevier Science Ltd. All rights reserved.
PII: S 0 3 0 8 - 5 9 6 1 ( 0 0 ) 0 0 0 9 7 - 5

250

T.J. Gerpott et al. / Telecommunications Policy 25 (2001) 249}269

the increased number of network operators and the intense competition that has developed. These
factors, in turn, are a consequence of the removal of monopoly rights, which were mainly enjoyed
by state-owned operators of public telecommunication networks. The increasing economic importance of telecommunications companies inspired many management scholars to devote more
teaching and research attention to this sector (see Szyperski & Loebbecke, 1999, pp. 485}486).
Speci"cally in the "eld of marketing strategies for telecommunications services it is frequently
pointed out that once customers have been acquired and connected to the telecommunications
network of a particular operator, their long-term links with the focal operator are of greater
importance to the success of the company in competitive markets than they are in other industry
sectors (see Wilfert, 1999, pp. 198}199; Gerpott, 1998, pp. 213}221; Knauer, 1998, pp. 510}511;
Harter, Ripsam, & Ruhl, 1997, p. 15; Booz. Allen & Hamilton, 1995, pp. 55}60). Nevertheless, there
is a dearth of empirical research into the extent of customer retention or the supply side retention
drivers in particular telecommunications markets. Therefore, this paper examines di!erences and
commonalties between the constructs `customer retentiona, `customer loyaltya, and `customer
satisfactiona and what supply side factors in#uence them both conceptually and empirically for the
German mobile communications market using data from a sample of 684 residential mobile
communications users.
This type of analysis is not only of interest to cellular network operators/service providers but
also to other industries for at least two reasons. First, a discussion of how customer retention,
customer loyalty, and customer satisfaction can be di!erentiated and how the three variables are
inter-linked is of general importance to corporate management. This is because without respect to
the industry concerned all three are (intermediate) objectives on the way to ensuring a company's
sales success which can be in#uenced by management action. Second, to date management
research on customer retention and satisfaction has been concerned almost exclusively with
over-the-counter goods sold in unconnected individual transactions in mature markets. In the
mobile communications market, however, usage-dependent and service-like contract goods are
sold. By focusing the analysis on this type of market a less `made-to-order "eld of researcha is
investigated.
The remainder of this paper is structured as follows: In Section 2, a brief description of
the German mobile communications market is provided. In Section 3, we explicate the concepts
of customer retention, customer loyalty, and customer satisfaction * "rst in general, and
then with reference to the mobile communications market in particular. Further, we develop
hypotheses on supply side determinants of these three constructs within the mobile
communications market. These hypotheses are then tested in an empirical study reported
in Section 4. Finally, in Section 5, we discuss implications of the empirical "ndings for
number portability regulation in mobile markets and for competitive strategies of mobile network
operators.

 See, for instance, the article collections edited by Bruhn and Homburg (1998), and Simon and Homburg (1997), in
which customer retention, customer loyalty, and customer satisfaction are discussed as success criteria for marketing
strategies in a wide range of di!erent industries.
 For reviews of relevant management research see Gerpott (2000), Homburg and Bruhn (1998), Homburg and
Rudolph (1997), and Dick and Basu (1994).

T.J. Gerpott et al. / Telecommunications Policy 25 (2001) 249}269

251

2. Pro5le of the German mobile cellular market

Mobile communications markets can be divided by the type of services provided and by the
telecommunications networks used for `productiona into the sub-markets for cellular
radiotelephony, paging, trunked mobile radio access, and satellite services and networks (see
Gerpott, 1998, p. 220; Knauer, 1998, p. 509; Stoetzer & Tewes, 1996, p. 304). In terms of sales
revenues and number of customers involved, the mobile cellular telephony market is by far the
most important sub-market for mobile communications. Thus, this study concentrates on this
part of the mobile market in Germany. In keeping with common practice, the terms `cellular
marketa and `mobile (communications) marketa will be used synonymously. The "rst non-military
public mobile communications network was introduced in 1958 into Germany. By the end of
1991, however, only 0.53 million radiotelephone access lines had been sold (see Gerpott, 1999,
pp. 60}61; Wilfert, 1999, pp. 188}189; Stoetzer & Tewes, 1996, p. 305). This was due to the high
prices of terminals and services, and to technical shortcomings of the analogue network generations used at the time (e.g., frequently no radio coverage within buildings.) Demand only started to
take-o! when digital mobile networks based on the GSM standard replaced analogue networks,
and Deutsche Telekom's monopoly in the "eld of mobile cellular networks and services was
dissolved.
Competition between four GSM operators promoted high network availability and the fast
introduction of user-friendly terminals. It also prompted considerable marketing e!orts. These
factors and others led to a dramatic increase in the number of mobile access lines in Germany
* from 0.92 million at the end of 1992 to 23.32 million at the end of 1999. In the 12 months from
the end of 1998 to the end of 1999 alone, the number of cellular access lines in Germany increased
by 9.46 million or by 68%. At the end of 1999, there were four network operators sharing the
market. These were Mannesmann/Vodafone with a customer market share of 42%, Deutsche
Telekom/T-Mobil with 40%, E-Plus with 15%, and VIAG Interkom with 3%. Because the four
operators use a network technology based on a common standard, the services they o!er are very
similar. As a result, the German mobile communications market has the structure of a narrow
oligopoly. In this oligopoly market, at least up to the end of 1999 none of the four network
operators adopted a very aggressive let alone a cutthroat competitive style (cf. Wilfert, 1999,
pp. 189}190; Gerpott, 1998, p. 220; Stoetzer & Tewes, 1996, p. 307).
Studies of customer retention, loyalty, and satisfaction have typically analysed markets for
over-the-counter goods sold in single discrete transactions (e.g., motor cars). In comparison, the
mobile communications market has two special features. First, access to a cellular network and the
handling of calls over that network represent a continuous contractual transaction carried out over
a long period. During this period, the buyer cannot be certain that the network operator will
always provide the promised service quality to the full extent. The contractual transaction is
normally structured in such a way that customers pay a basic monthly fee for the opportunity to
use the mobile communications network through access that has been provided for the purpose. In
addition, they pay ex post call charges in accordance with the amount of minutes they actually use

 For a general discussion of contractual/continuous vs. single/discrete transactions, see Oevermann (1996, p. 30) and
Jeschke (1995, p. 14).

252

T.J. Gerpott et al. / Telecommunications Policy 25 (2001) 249}269

the network. As far as network operators are concerned, continuing the contractual relationship
with a current customer is more important with this sort of business structure than with transactions in purchase goods. The reason is that marginal income from the customer increases the longer
the business relationship lasts, because the incremental costs for each customer/cellular access line
and period are much lower than the basic access fee paid by the customer within the period.
The second special feature is that the German mobile communications market is still relatively
young. Apart from rapid growth in the number of customers, another general feature of young
markets is that the service category exchanged there is new, both for sellers and buyers (cf. Laker,
Pohl, & Dahlho!, 1999, pp. 88}93; Gerpott, 1998, pp. 215}216). For instance, none of the four
network operators had had any experience in competition with other providers of mobile networks, when they entered the German market between 1992 and 1998. Likewise, the large number
of new customers acquired by the four networks were predominantly (private) accounts who have
not used a mobile terminal before. Consequently, the level of uncertainty among prospective
customers in the mobile communications market may be higher * or their level of trust may be
lower * than in fully developed markets.
On the demand side, the mobile communications market is divided into residential and business
customers. Our own research excludes the market for business customers who mainly use mobile
communications services to earn income. Unlike residential customers, business users often do not
themselves make the decision to sign or extend a mobile subscription contract. Instead, there are
special purchasing departments within their companies that are responsible for doing this. In the
mobile communications market, therefore, there should be a clear distinction between the processes used to retain business accounts and those used to retain residential customers. To reduce
complexity, therefore, we restrict our analysis to the residential customer segment.
3. Customer retention, customer loyalty, and customer satisfaction: di4erentiation of the constructs,
their causal links, and their determinants
3.1. General research status
The phenomenon of customer retention encompasses a degree of `fuzzinessa since it represents
a theoretical construct which cannot be observed directly. There is considerable variance in the
ways in which customer retention is speci"ed conceptually and empirically by scholars and
practitioners. There are also great di!erences in the manner in which it is more or less separated

 Since 1997, operators of mobile communications networks in Germany have been o!ering their customers an
alternative contractual model, namely the ability to pay in advance for speci"c use of the network (such as 100 min of
outgoing calls) over a speci"ed period (for instance six months). (These are the so-called `pre-paid access cardsa). But
even this business model involves a contractual transaction in which the network operator has an (even greater) interest
in motivating the customer actively to renew the contractual relationship (`reloading of the pre-paid cardsa). This is
because the relationship will otherwise be brought to an end automatically when the mobile communications access is
deactivated once the contractual period has expired. For this reason, no further distinction will be made here between
pre-paid customers and conventional mobile communications customers, whose contracts are implicitly extended unless
explicitly terminated by at least one of the parties.

T.J. Gerpott et al. / Telecommunications Policy 25 (2001) 249}269

253

from related constructs such as `customer loyaltya, `customer satisfactiona, `customer enthusiasma, `customer trusta, and `customer obligationa (for similar views see Bliemel & Eggert, 1998,
p. 38; Diller, 1996, p. 81). Therefore, for the sake of clarity we must point out that we follow the
theoretical arguments of Homburg and Bruhn. They suggest distinguishing between the constructs
of customer retention, customer loyalty, and customer satisfaction which they see as linked by
a two-stage causal chain (cf. Homburg & Bruhn, 1998, pp. 8}10). Accordingly, customer satisfaction is a direct determining factor in customer loyalty, which, in turn, is a central determinant of
customer retention. Each of the three constructs mentioned is a!ected by other factors, which can
be divided into those on the side of the potential customer (e.g., need for variety) and those on the
side of the supplier of a product. On the provider side, companies have a direct in#uence
on construct determinants (e.g., the marketing mix). On the customer side, companies can, at
most, in#uence factors (e.g., demographic and psychographic features of the potential customer)
indirectly by selecting their target markets. Therefore, customer side factors are not analysed in this
paper.
If one, "rst of all, speci"es the customer retention construct at the end of the causal links
postulated by Homburg and Bruhn, it is possible to identify the following common core in the
wide-ranging de"nitions suggested in the literature: Customer retention (CR) is concerned with
maintaining the business relationship established between a supplier and a customer. This can be
achieved in two ways. The "rst is by subsequent purchases, or by extending the customer's contract
with the supplier over a speci"ed period of time (ex post perspective). The second is by the intention
of the customer to make future purchases from the provider, or to refrain from quitting the contract
(ex ante consideration) (cf. among others, Herrmann & Johnson, 1999, p. 583; Bliemel & Eggert,
1998, pp. 38}39; Me!ert, 1998, p. 119; KruK ger, 1997, pp. 19}22; Peter, 1997, p. 7; Diller, 1996,
pp. 83}84). It is not possible to derive clear threshold values * either for the frequency of
subsequent purchases, the continuation of contracts, or the intention to abandon a provider * that
could be used to classify a customer as `retaineda when they were exceeded or not reached. The
retention of a customer by a supplier therefore is represented by a continuous variable which can
take di!erent values over time (see in agreement Homburg & Bruhn, 1998, p. 10; Me!ert, 1998, pp.
119, 129; KruK ger, 1997, p. 27; Diller, 1996, p. 84).
On the one hand, a business relationship may be maintained involuntarily because a customer is
prevented by mobility barriers from changing suppliers or dispensing with a category of service
(cf. Herrmann & Johnson, 1999, pp. 585}586; Bliemel & Eggert, 1998, pp. 41}43; Me!ert, 1998,
pp. 127}128; Diller, 1996, p. 88). On the other hand, a customer may carry out subsequent
transactions because she has a favourable attitude towards the provider and the services he
supplies, and because he therefore wants to keep the business relationship going to their mutual
bene"t. Customer loyalty (CL) is the term used when business relationships are continued in the
latter way (cf. Bliemel & Eggert, 1998, pp. 39}41; Homburg & Bruhn, 1998, pp. 8}9; Weinberg,
1998, p. 49; KruK ger, 1997, pp. 20}21; Diller, 1996, pp. 88}89; Dick & Basu, 1994, p. 101). From this
distinction it follows that, although CL and CR may be strongly related in terms of cause and e!ect,
the existence of mobility barriers and mobility intensi"ers means that they are not completely
congruent.
The construct of customer satisfaction that comes at the beginning of the causal links assumed
by Homburg and Bruhn (1998) can be conceptually speci"ed as follows: Customer satisfaction (CS)
is an experience-based assessment made by the customer of how far his own expectations about the

254

T.J. Gerpott et al. / Telecommunications Policy 25 (2001) 249}269

individual characteristics or the overall functionality of the services obtained from the provider
have been ful"lled. Satisfaction is higher or lower with respect to the extent to which what was
actually provided exceeds or falls short of what was expected. In the literature CS is discussed as
one important factor in determining CL which is not conceptually identical with CL. This is
because CL is also essentially in#uenced by the future conxdence of customers in the capabilities of
a supplier or his sales objects, and by their assessment of owers by competitors relative to those
which they experienced from their present vendor (cf. Herrmann & Johnson, 1999, p. 586;
Homburg, Fa{nacht, & Werner, 1999, p. 186; Bliemel & Eggert, 1998, p. 40; Weinberg, 1998,
pp. 48}49; Dick & Basu, 1994, pp. 101, 104).
The theoretical di!erentiation of CR, CL, and CS that can be derived from the literature, and the
two-staged causal links between these constructs will next be considered with regard to their
speci"c relevance for the German mobile communications market.
3.2. Hypotheses on interrelationships between CR, CL, and CS and their determinants
in the German mobile market
3.2.1. Customer retention and its determinants
The matrix in Fig. 1 displays four ideal combinations of CR and CL values as pertinent for the
German mobile communications market (cf. Diller, 1996, p. 88, Fig. 5). Accordingly, there are two
standard situations covering the following cases. First, that in which customers want to terminate
the contractual relationship with their mobile carrier and, from the point of view of their loyalty,
will distance themselves from their provider (see the customer type `wanderersa in Fig. 1). Second,
that in which customers want to maintain the contractual relationship and have a positive attitude
to the provider (see the customer type `loyal customersa in Fig. 1). In addition, it may happen,
however, that although customers do not feel any loyalty towards their network operator, they
nevertheless do not wish to terminate the contractual relationship (see the customer type `captive
customersa in Fig. 1). The main reason for this may be that customers in Germany are not able to
take their assigned cellular phone numbers with them to another network operator with whom
they would like to enter into a new contract once they have terminated their contract with their
present carrier (cf. Schwarz-Schilling & Stumpf, 1999, pp. 3}10, 37}65; Mellewigt, 1997, p. 576).
Customers, then, have to pay a price for `inconveniencea if a new mobile communications number
is assigned to them when they change providers. This is because (1) other persons in their
environment have to be informed of the new number so that contact is not lost, and (2) new
business cards, writing paper, address labels, etc. may have to be revised (see Berke, 1999, p. 94;
Knauer, 1998, pp. 516}517).

 For a discussion of the meaning of `customer satisfactiona, see Herrmann and Johnson (1999, pp. 582}583),
Homburg, Giering and Hentschel (1999, pp. 175}176), KruK ger (1997, pp. 43}48), and Woodru! (1997, pp. 142}143).
 In view of the distinction made here between CR, CL, and CS, the following point must be noted. The crosstabulations shown in Fig. 1 di!er from matrices in the previous literature that, at "rst glance, appear similar, but in which
CR and CL or CL and CS are not analysed separately (see e.g. Herrmann & Johnson, 1999, p. 585; Homburg, Fa{nacht
and Werner (1998, p. 406); Dick & Basu, 1994, p. 101).
 For a general discussion of the competitive relevance of mobile network operator number portability, see Gerpott
(1998, pp. 75, 244).

T.J. Gerpott et al. / Telecommunications Policy 25 (2001) 249}269

255

Fig. 1. Ideal combinations of customer retention and loyalty, and customer loyalty and satisfaction in the mobile
communications market.

Non-standard types of customers are the `advantage maximisersa and the `bad buyersa (see
Fig. 1). Although these credit their network operator with providing a thoroughly `respectable
technical servicea and have a positive attitude towards him, they nevertheless intend to terminate
their business relationship with the operator. A literature review and anecdotal insights which we
obtained over a 5-year period in consulting projects for cellular operators suggest that this low CR
in spite of high CL can be explained by three complementary lines of argument:
E In the German mobile communications market, it is common practice for customers to be sold
mobile handsets at a low price when they conclude a contract with a provider. This terminal
price is well below the one that has to be paid when no mobile subscription contract is involved.
The aim is to lower the barrier for new customers entering the young mobile communications

256

T.J. Gerpott et al. / Telecommunications Policy 25 (2001) 249}269

market. In return, the providers require that their customers sign a long-term access line contract
(typically for two years) to make up what they have lost on the terminal through the revenues
obtained from a customer (see Wilfert, 1999, pp. 198}199; Knauer, 1998, pp. 513}514; Booz.
Allen & Hamilton, 1995, p. 52). The result of this practice could be that after the minimum
contractual period has expired a substantial number of customers terminate their contract so
that they can once again obtain a subsidised terminal (see Wilfert, 1999, p. 199). Accordingly, the
strong desire of mobile communications customers to obtain a new terminal, irrespective of their
loyalty to their network operator, can have a negative e!ect on CR.
E Many residential customers only use mobile communications services for the xrst time after they
have signed a network access contract. Because of their lack of experience with such services, it is
possible that customers may overestimate the bene"ts they expect to obtain. Furthermore, the
requirements of potential customers can change to the extent that the reasons that led to
a contract in the "rst place no longer apply. In both cases, the reduced personal bene"ts that
a mobile communications connection brings can lead customers to change their intention of
continuing with the contract. This low CR level does not necessarily have to go hand in hand
with a negative attitude towards the network operator; for the customers may recognise and
acknowledge that the important reason for wanting to terminate the business relationship has
nothing to do with their contractual partner.
E The costs that customers have to pay for the use of mobile communications access are di$cult to
estimate when a contract is being signed. This holds particularly for customers who have no
experience of cellular services. For this reason, it could turn out that a customer has underestimated how much he/she will have to pay. Thus, there may be a large number of cellular
customers who, as a result of having to pay more than their threshold value, intend to terminate
the contractual relationship with their network operator to completely withdraw from the
market ("low CR). Nevertheless, because of the similar prices charged by all the competitors in
the German mobile communications market, these customers express complete loyalty to their
current business partner ("high CL).
To sum up, based on Fig. 1 and the preceding three paragraphs, we expect CR to be stronger in
the German mobile communications market when
E CL is also high (Hypothesis H ), although there will also be a considerable number of customers

who, with strong (weak) bondings to a network operator, show a low (high) degree of CL
(Hypothesis H );

E it is important for customers to retain their mobile access number over a long period (Hypothesis
H );

E fewer customers wish to obtain a new mobile terminal (Hypothesis H );

E customers rate highly the bene"t they obtain from the services supplied by their own mobile
communications provider (Hypothesis H );

E customers favourably rate the prices charged by their own mobile communications provider for
the services supplied (Hypothesis H ).

3.2.2. Customer loyalty and its determinants
In the literature (cf. Herrmann & Johnson, 1999, pp. 584}587; Homburg et al., 1999, p. 178; Dick
& Basu, 1994, p. 104) it is emphasised frequently that there is a signi"cantly positive correlation

T.J. Gerpott et al. / Telecommunications Policy 25 (2001) 249}269

257

between CS and the loyalty that customers feel towards a "rm, but that a high degree of CL does
not always have to go hand in hand with a high degree of CS. Here, it is usually assumed that CL
and CR can be equated. However, it has just been shown that CS has only an indirect e!ect on CR
via CL. It is therefore necessary to have separate discussions about typical ideal combinations of
CL and the overall satisfaction of customers (see the bottom half of Fig. 1). Accordingly, there is the
standard case of `disappointed customersa who are dissatis"ed with the services provided by their
network operator, and who consequently make a negative assessment of this provider with regard
to the future (see Fig. 1, standard case A). Similarly, we have a standard case of `impressed
customersa who are very satis"ed with the performance of their network operator to date, and who
therefore place great trust in him with regard to the future as well (see Fig. 1, standard case B). In
addition, however, there are plausible explanations for the occurrence of two rarer cases. First,
a network operator can have `optimistic customersa who are not really satis"ed with the services
supplied to date, but who nevertheless trust that their provider will improve his performance in the
future. Alternatively, they see him as a `lesser evila in comparison with his competitors (see Fig. 1,
atypical case A). Second, there may be `pessimistic customersa who are satis"ed with the services
they have received from their network operator, but who nevertheless expect, or assume, that it
might be possible to obtain better services from his competitors. Our arguments derived from
conceptual CR/CS studies can be condensed in the form of three hypotheses:
E CS has positive e!ects on CL (Hypothesis H ).

E CS does not determine CL completely. There is, indeed, a considerable number of customers
who have a high (low) degree of CL while at the same time demonstrating a low (high) degree of
CS (Hypothesis H ).

E The more positive the image that customers have of their network operator's competitors, the
weaker will be the CL to their current contractual partner (Hypothesis H ).

3.2.3. Customer satisfaction and its determinants
Up to now, CS has been characterised here as an overall retrospective judgement about how far
expectations with regard to a service have been ful"lled in use situations. This qualitative
perception is in turn based on evaluative perceptions by the customer with regard to individual
purchase-relevant features or value drivers which facilitate or block the achievement of servicerelated personal goals of the customer. As far as mobile communications services are concerned,
previous research suggests the following four individual features as key drivers of the customer
value of cellular services (see Wilfert, 1999, pp. 191}194; Bolton, 1998, pp. 54}55; Gerpott, 1998,
pp. 282}283; Danaher & Rust, 1996, pp. 67}69; Booz. Allen & Hamilton, 1995, pp. 57}58):
E the network quality, which is re#ected in excellent indoor and outdoor coverage and in the clarity
of voice reproduction without any connection break-downs;
E the price paid for obtaining access to and using the network;
 On the importance of trust and the perception of competitors as factors determining CL, see also Peter (1997, pp.
121}124) in addition to Herrmann and Johnson (1999, p. 586), Homburg et al. (1999, p. 186), Bliemel and Eggert (1998, p.
40), Weinberg (1998, pp. 48}49), and Dick and Basu (1994, pp. 101, 104).
 For a general theoretical treatment of the customer value notion, see Woodru! (1997).

258

T.J. Gerpott et al. / Telecommunications Policy 25 (2001) 249}269

E the features summarised under the rubric of **customer care++, namely the quality of the exchange
of information between customer and supplier (1) in response to customer (telephone) enquiries
and (2) in the course of interactive activities initiated by the network operator (e.g., presentation
of an invoice).
Another point to be taken into account is that the overall satisfaction of customers with mobile
communications services is also in#uenced by how far they perceive these services to be of bene"t
to them personally. Even if a customer has a positive (negative) view of the services provided by his
network operator, his overall satisfaction may still be adversely (favourably) a!ected by the fact
that he rates the general value of the range of services to meet his needs * in other words, the
idiosyncratic bene"t he obtains from them * as low (high). Overall, then, it is to be expected that
CS will increase hand in hand with the positive assessment of network quality (Hypothesis H ),

mobile communications prices/costs (Hypothesis H ), customer care (Hypothesis H ), and the


personal bene"t obtained from mobile communications services (Hypothesis H ).

There is one "nal point that remains to be made from a practical point of view. What is of great
interest is not just the mere existence of e!ects of di!erent variables on CR, CL, and CS, but rather
the relative size e!ect of the various supplier side factors linked to the three focal criteria.

4. Empirical methods and results

4.1. Sample
Data for the present study were collected by an established market-research company. We
instructed this "rm to carry out a standardised telephone survey among customers of mobile
network operators in Germany. The following groups of individuals were excluded from the
survey:
E Customers of the network operator VIAG Interkom because this company only had a 0.2%
share of the customer market at the time the data were collected (January 1999). Furthermore,
the small number of VIAG customers had very little experience with the services supplied by
their operator, since the VIAG network had only been launched three months before.
E Customers of T-Mobil who are still using the analogue `C-Netza because `C-Netza only had
a 2.5% share of the customer market in January 1999, and was due to be shut-down by the end of
2000. Questions about CR were therefore of little signi"cance in this context.
In order to obtain a random sample of about 700 residential customers layered in proportion to
the market shares of the three established mobile operators in Germany phone numbers were
drawn at random from all the "xed-network telephone numbers in Germany. If the interviewers
found that one person in a household was a customer of a mobile network operator, the person
concerned was asked there and then, or when contacted later, to participate in the survey. Those
who agreed to do so were asked if they were able to o!set their mobile communications costs
against tax, and whether they used their mobile communications principally for business purposes.
Only those who answered `noa to both questions were classi"ed as residential customers, and were
then included in subsequent analyses.

T.J. Gerpott et al. / Telecommunications Policy 25 (2001) 249}269

259

Following this procedure, a total of 684 residential customers were interviewed by phone. Of
these, 41.1% were customers of Mannesmann, 39.9% of T-Mobil, and 19.0% of E-Plus. The
respondents' age ranged between 15 and 76, with a mean of 36.89 years (S"13.34; n"680). 69.9%
of the respondents were men. In 51.2% of the cases, the net monthly income in the household of the
interviewee exceeded DM 4000. The monthly mobile bill of the respondents was between DM 10
and DM 1000, with a mean value of DM 80.05 (S"66.81; n"639). For the German cellular
market data on the socio-demographic structure of all the residential customers are not publicly
available. However, we obtained such data from one mobile network operator. We were therefore
able to analyse with regard to the four socio-demographic variables reported above, to what extent
there were signi"cant deviations between the sub-sample of the customers of this operator included
in our survey and the total population of all customers of this provider. Appropriate tests revealed
no di!erences at the 10% level of statistical signi"cance.
4.2. Measurement of variables
4.2.1. Customer retention, customer loyalty, and customer satisfaction
Following earlier work (for instance, Grund, 1998; Joho, 1996; Keaveney, 1995; Rust & Zahorik,
1993), to measure CR, interviewees were asked, whether they would `terminate their contract [with
the relevant network operator] as soon as possiblea ("indicator y ). Answers had to be given on

a 5-point scale ranging from `completely righta (coded 1), to `partly righta (coded 3), to `not at all
righta (coded 5). Table 1 informs about the distribution of the CR indicator in our sample.
According to this, 59.8% of the respondents can be classi"ed as closely linked to a network
operator (scale level of 5), and 12.6% as only slightly linked or highly willing to quit the contract
(scale levels of 1 and 2).
To measure CL, previous research has mainly employed items that re#ect an individual's
intention to repurchase and his willingness to recommend a product to others (cf. Herrmann
& Johnson, 1999; KruK ger, 1997; Peter, 1997). Therefore, the residential customers were asked to
what extent, on the basis of their experience, they would
E reselect the same network operator or another provider (y ),
 
E recommend their own or another network operator to friends or acquaintances (y ).
 
Participants were provided with "ve answer categories ranging from `de"nitely opt for/recommend another network operatora (coded 1), to `undecideda (coded 3), to `de"nitely reselect/recommend the present network operatora (coded 5). As expected, both items were strongly
correlated (r"0.74). Consequently, they were averaged on a single CL scale with item weights
 In mid-January 1999 the distribution of all customers of the three German operators of mobile communications
networks mentioned was as follows: Mannesmann 44.2%; T-Mobil (without `C-Netza) 40.8%; and E-Plus 15.0%. As
a consequence, E-Plus customers are somewhat over-represented, and those of Mannesmann and T-Mobil somewhat
underrepresented. A
 test revealed that network-operator shares in our sample and in the market as a whole di!er
signi"cantly at the 5% level, but not at the 1% level (
"8.09; df"2).
 For measurement/indicator variables of dependent (independent) constructs to be explained (not to be explained),
we use the abbreviation y (x ). If a construct is measured by several indicators, the following notation holds:
G G
y (x )"measurement variable z to capture the endogenous (exogenous) latent variable y (x ).
G X G X
G G

260

T.J. Gerpott et al. / Telecommunications Policy 25 (2001) 249}269

Table 1
Distribution of answers for the variables customer retention, customer loyalty, and customer satisfaction
Percentage frequency of answers for scale level

Variables
1. Customer retention (y )

2. Customer loyalty (y )

3. Customer satisfaction (y )


(Low&&&&&&&&&&&&&&&&&&
High)
1
2
3
4
5

M

8.9
4.3
1.8

4.14
3.95
3.95

1.28
0.99
0.90

674
652
682

3.7
8.6
4.5

11.3
12.4
18.2

16.3
51.2
47.5

59.8
23.5
28.0

Customer loyalty was measured with two items using the average of the factor-weights of the items to calculate the
`customer loyaltya scale score. Possible intermediate values that are not whole numbers were rounded o! to simplify the
presentation of the distribution of `customer loyaltya answers. In all other statistical analyses, the exact, non-rounded
customer loyalty scale scores were used.

M"arithmetic mean value; S"standard deviation; n"number of valid answers received.

taken from their loadings in a factor analysis. This scale had an acceptable factor reliability of 0.89
(see Fig. 2). As can be deduced from Table 1, the CL mean was 3.95. In other words, the average
residential respondents showed a relatively high degree of loyalty to their mobile operator.
To measure (overall) CS with their mobile communications network participants were asked to
comment on the statement `I am completely satis"ed with my current mobile communications
networka (y ), using a 5-point scale anchored with the reply options `completely righta (coded 5)

and `completely not righta (coded 1). It can be seen from Table 1 that only 6.3% of the respondents
were slightly satis"ed, or not satis"ed at all, with their mobile communications network (scale levels
of 1 and 2), and 28.0% were completely satis"ed (scale level of 5).
There were signi"cantly positive associations between the three endogenous variables. At 0.57,
the correlation between CL and CS reached approximately the order of magnitude found in earlier
studies of other markets with similar variable measurements (see, for example, Herrmann &
Johnson, 1999, p. 594; Peter, 1997, p. 221). An explorative factor analysis con"rmed the expected
three-factorial structure of the four items y , y , y , and y . Consequently, hypotheses H , H ,
    

 
H , and H formulated in Sections 3.2.1 and 3.2.2 cannot be classi"ed as statements already refuted


by an initial rough analysis. Therefore, subsequent con"rmatory detailed analyses can begin with
the assumption that the variables CR, CL, and CS represent overlapping but not identical constructs.
4.2.2. Other supply side determinants of the focal criteria
The three potential CR determinants phone number constancy, acquisition of a new terminal at
a favourable price, and personal benext from the mobile communications network of the current
contractual partner of a customer were each measured by single item questions. Their wording and
answer scales are shown in Table 2 (see variables x , x , and x in Table 2). As the fourth supply side
 

factor determining CR and CS, assessment of mobile communications prices was operationalised
with two items that are marked as indicators x and x in Table 2. An interviewee's perception
 
 
 For CR and CL: r"0.57 (n"645); for CL and CS: r"0.57 (n"651); for CR and CS: r"0.47 (n"674).

T.J. Gerpott et al. / Telecommunications Policy 25 (2001) 249}269

261

Table 2
Measurement of supply side variables expected to in#uence the focal three criteria
Variable name

Measurement

Phone number constancy

`It is important for me to be able to keep my present mobile phone number

once my contract with2has expired.a (x )

Five answer categories, for `completely righta ("5), to `partly righta ("3), to
`not at all righta ("1)
`It is important for me to obtain a new mobile phone at a very reasonable price
once my contract with 2 comes to an end.a (x )

Five answer categories, as shown above for item x

`What personal bene"t does the 2 network have for you?a (x )

Five answer categories, from `very substantial bene"ta ("5), to `medium
bene"ta ("3), to `no bene"ta ("1)
`How do you assess the prices, tari!s, and conditions of your 2 access?a (x )
 
`How do you assess the prices charged for connecting from your 2 access to
the "xed network?a (x )
 
Five answer categories, from `extremely gooda ("5), to `gooda ("3), to `bada
("1)
`I will present a number of statements to you that may describe providers of
mobile phone services. For each one, please indicate whether it applies particularly to 2[competitor of the interviewee's own provider].a
Customer orientation (x ): Total of the statements marked as particularly
 
applicable `is fair to his customersa, `is customer-friendlya, and `is nicea
Reliability (x ): Total of the statements marked as particularly applicable
 
`is seriousa, `is reliablea, and `provides securitya
Modernity (x ): Total of the statements marked as particularly applicable
 
`is technically up to datea, `is a modern "rma, and `is dynamica
Sum of the three image elements x , x , and x to form an overall image
   
 
score with higher values representing a better competitor image
`How do you assess the quality of the 2 network, based on your overall
experience in using the 2 network for phone calls?a (x )
 
`How do you assess the call quality of the 2 network ?a (x )
 
Five answer categories, as shown above for items x
and x
 
 
`How do you assess the quality of the customer care services of the 2 network?a (x )
 
`How do you assess the e!orts of your network operator to keep you up to
date? These e!orts include #yers sent with your bill, letters to customers, and
brochures.a (x )
 
Five answer categories, as shown above for items x
and x
 
 

New terminal

Personal bene"ts

Assessment of prices

Image of competitors

Assessment of network quality

Assessment of customer care

of the network operator's competitors (image of the competitors) was investigated by presenting nine
potentially positive characteristics of operators of mobile communications networks. Respondents
were then asked to indicate the extent to which these characteristics applied to a particular
competitor. Factor analyses revealed that the image of the competitors comprised the three
dimensions `customer orientationa, `reliabilitya, and `modernitya (see variables x , x , and
   
x in Table 2). These could, in turn, be aggregated to an overall scale image of competitors with an
 

262

T.J. Gerpott et al. / Telecommunications Policy 25 (2001) 249}269

Fig. 2. Completely standardised solution of the empirical causal model.

acceptable measurement reliability (see Section 4.3). Finally, the CS determinants of perceived
network quality and customer care performance were each measured with two items for which details
can also be found in Table 2 (see there variables x , x , x , and x ).
     
 

T.J. Gerpott et al. / Telecommunications Policy 25 (2001) 249}269

263

4.3. Analysis of causal links

4.3.1. Quantixcation and quality assessment of the postulated causal model
The hypotheses put forward in Section 3.2 can be interpreted as a model with multi-stage causal
links. In this model, CR is in#uenced by CL and by the supply side variables of `phone number
constancya, `new terminala, `personal (service) bene"ta, and `service pricesa. CL, in turn, is dependent
on CS and the image of the competitors. Finally, CS is a function of the assessment of price, network
quality, customer care, and the personal bene"t provided by the services of a mobile communications
operator. To wit, CS works indirectly through CL on CR, but does not a!ect CR directly.
Causal analysis methods can be used to test this multi-stage causal model. In causal analyses,
simultaneous tests are performed to see how far the following three models correspond to the
empirical data: (1) a measurement model designed to capture the (indicator-based latent) variables
addressed in the research's cause and e!ect hypotheses; (2) a structural equations model intended to
re#ect the cause and e!ect relationships between the latent variables; and (3) an overall model
consisting of the measurement model and the structural equations model. Therefore, the present
study employed a con"rmatory causal analysis test of the system of causal relationships de"ned by
hypotheses H }H using the LISREL 8 software package. To estimate the models, we selected
 
the `Generally Weighted Least Squares (WLS)a method because it typically provides valid
parameter estimates, even if * as in our study * a normal distribution of the indicators cannot be
assumed in the underlying population, and if the indicators have to be regarded as ordinal-scaled
rather than interval-scaled (cf. JoK reskog & SoK rbom, 1996, p. 240).
Fig. 2 gives details of the structural equation model calculated in accordance with our 13 e!ect
assumptions. It also gives details of the factorial measurement model on which it is based. Latent
variables are indicated by ellipses and measurement variables by rectangles. The arrows pointing
from the latent constructs to the measurement variables reveal the factor loading of the indicator
on the construct. The arrows pointing `from outsidea to the indicators quantify the residual part
for the relevant indicator that cannot be explained by the measurement model. The
arrows/standardised path coe$cients between the two latent variables represent the causal relationship between the two constructs.
Before the individual path coe$cients in Fig. 2 can be interpreted, the "t (1) of the measurement
model, (2) of the structural equation model, and (3) of the entire causal model with our data must be
checked (see Homburg & Baumgartner, 1995, pp. 162, 171}172). To be acceptable, an LISREL
measurement model has to ful"l the following criteria: (a) the reliability for each indicator must be
*0.4; (b) each factor reliability must be *0.6, and (c) the average factor variance explained must
be *0.5. Our measurement model meets requirement (a) for 15 of 16 indicators (the exception is
variable x ) and requirement (b) without restriction. Requirement (c) is met by four of the "ve
 
latent variables in Fig. 2 that are measured by at least two indicators. Only for the construct
`assessment of customer carea (x ) does the average explained variance of the variables for the

factor fail, with 0.46, to reach the aforementioned threshold value. Overall, then, the quality of our
measurement model can be said to be at least satisfactory, if not very good.
 Methodological introductions of statistical causal analysis techniques can be found in Homburg and Baumgartner
(1995, pp. 163}165), and Homburg and Hildebrandt (1998, pp. 16}43 and the sources quoted there).
 LISREL stands for Linear Structural Relationships.
 The threshold values for the various quality criteria discussed below are also established there.

264

T.J. Gerpott et al. / Telecommunications Policy 25 (2001) 249}269

The quality of structural equation models is classi"ed as acceptable if the squared multiple
correlation for each endogenous variable (to be explained) * in other words, for CR, CL, and CS in
our model * is *0.4. In our sample, the squared determination coe$cients of the model shown in
Fig. 2 are 0.78 for CR, 0.64 for CL, and 0.65 for CS. This means that our structural equation model
exhibits a good "t with the empirical data.
Common test criteria for an overall model are the following (see for details Homburg &
Baumgartner, 1995, pp. 165}172):
1. a root mean squared error of approximation (RMSEA) that should not be above 0.05;
2. a ratio of
 model-"t statistics by degrees of freedom that should not exceed 2.5;
3. goodness of "t indices, and particularly (a) the goodness of "t index (GFI), (b) the adjusted
goodness of "t index (AGFI), and (c) the comparative "t index (CFI), each of which has to exceed
a threshold value of 0.9 if a model is to be classi"ed as "tting the data well.
As can be seen from Fig. 2, the "rst two or the last three quality criteria for our overall model are
well below the highest recommended value or above the minimum suggested values. This means
that the overall model in Fig. 2 can be classi"ed as "tting the data very well.
A critical aspect in our theoretical reasoning is the separation of/ability to separate CL and CS.
Therefore, in spite of the good "t of the data with the causal model shown in Fig. 2, we tested how
much better would be an alternative model in which these two variables or the three indicators
y , y , and y were combined to form a new construct `customer tiesa, "tted to the data. This
   

alternative model was compared against the theoretically expected model in Fig. 2: Its RMSEA
value was 35% higher ("worse) and its
/degree-of-freedom ratio was 33% higher ("worse)
than the benchmark model values. Further, the revised model did not explain the CR construct as
well as our initial model. Therefore, it is both possible and necessary to retain the proposed
separation of the constructs CL and CS.
4.3.2. Discussion of factors inyuencing CR, CL, and CS
In line with H , the empirically derived causal model (see Fig. 2) suggests that the willingness of

mobile communications customers to continue the contractual relationship with their network
operator is strongly in#uenced by the extent to which they have a positive perception of him * in
other words by CL. At the same time, however, CL does not in any way fully explain the CR
variance in our sample. Thus, the data also support H . Further signi"cant direct determinants of

the degree of CR were the desire to leave one's phone number unchanged, the assessment of the
personal bene"t obtained from the services supplied by one's own mobile communications
provider, and his service prices (see Fig. 2). Our analyses therefore also con"rm H , H , and H .
 

Contrary to H , the strength of preference for the acquisition of a new terminal at a favourable

price did not have any signi"cant (negative) e!ect on CR.
With regard to possible determinants of CL, it can be seen from Fig. 2 that the causal links
postulated in H }H are in line with the empirical observations. Thus, for instance, an increase in
 
CS by one unit leads to a highly signi"cant increase in CL by 0.75 units (support of H ).

Nevertheless, although there is no doubt that mobile communications CS has an outstanding
in#uence on CL to their network operator, the CS variable cannot by any means completely
explain the CL variance. Consequently, it is reasonable to preserve the conceptual distinction
between CL and CS as suggested by H . In our sample, a second signi"cant factor determining CL


T.J. Gerpott et al. / Telecommunications Policy 25 (2001) 249}269

265

proved to be the image that the customer had of his network operator's competitors. The more
positive this image was, the lower the degree of CL (support of H ). This indicates, then, that the

degree of CL depends partly on processes in which customers compare their perceptions of their
own network operators with the perceptions they have of his competitors.
Our analyses supported the hypothesised two-stage causal link in which CS has only an indirect
e!ect on CR via CL: CS had a strong indirect e!ect of 0.375 (0.75;0.50) on CR, but its e!ect on CR
was completely mediated by the CL variable. Furthermore, three of the four remaining hypotheses
on determinants of CS could be maintained in our sample. As can be seen in Fig. 2, the perceived
network quality (H ), the assessment of mobile prices (H ), and the assessment of personal


bene"t obtained from mobile communications services (H ) were detected as signi"cant direct

determinants of the overall satisfaction of mobile customers. Contrary to H , we did not "nd

a signi"cant relationship between CS and perceived quality of customer care.
If the direct and indirect e!ects of the factors explaining the di!erences in CR in Fig. 2 are
summarised, it can be seen that CL is the strongest determinant of CR with a standardised e!ect of
0.50. But the overall CR e!ect of a customer's assessment of mobile communication prices amounts
to 0.44 (0.28#0.41;0.75;0.50) and is thus only slightly weaker. Following CS, perceived
personal bene"t obtained from mobile communications services from one's own provider had the
fourth-strongest impacts on CR with a total of 0.30 (0.20#0.25;0.75;0.50). The 47% di!erence
in the e!ect produced by the factors `pricea and `personal bene"ta can be explained by the
following consideration: Because of technological standardisation in digital mobile networks, it is
di$cult for mobile operators to di!erentiate their services once a comprehensive regional coverage
of their networks has been achieved. As a consequence, a customer's perception of the personal
bene"t obtained from the services of his own network provider could not be attributed to the
performance of the company concerned in particular, but to all providers in general. Therefore, in
turn, the personal bene"t obtained from services supplied by the customer's own contractual
partner has a weaker e!ect on CS and on CR with just that partner.
With an overall e!ect of 0.22, the mobility barrier caused by the lack of network operator
portability of mobile access numbers was the "fth-strongest determinant of CR. This barrier helps
network operators to `motivatea customers to continue their contractual relationship even if
they do not have a very positive attitude towards their mobile communications provider and/or
are not satis"ed with the services he supplies. With an e!ect size of 0.17 (0.46;0.75;0.50),
the sixth-strongest determinant of CR was the perceived quality of the network. Customers
appear to consider quality to be a feature that `can be taken for granteda. Therefore, this
variable does not have an outstanding e!ect on CR. Finally, the desire of a mobile communications
customer to obtain a new terminal and perceived quality of an operator's customer care processes
had practically no e!ect on the CR criterion. Customer care can therefore be taken to
represent a peripheral service feature that is of little importance in shaping CR compared to the
core service.

5. Implications for competitive strategies of mobile network operators

The purpose of this study was to analyse levels, di!erences, and causal links of the three
constructs CR, CL, and CS, using a sample of 684 residential mobile communications customers in

266

T.J. Gerpott et al. / Telecommunications Policy 25 (2001) 249}269

Germany. Overall, respondents scored relatively high on the three focal constructs. The results
further indicate that it is conceptually justi"able, empirically tenable, and practically helpful for the
contract business under consideration if CR is not equated with CL and/or CS. Instead, a twostage causal mechanism should be assumed in which CS drives CL which in turn has impacts on
CR. To illustrate the need to distinguish between the three focal variables, respondents were
divided along the median values of the three target constructs in the sample in cases with a low or
a high value on a particular construct. Cross-tabulations of the CR}CL or the CL}CS answer
combinations of the respondents revealed that 34 or 31% of the customers belong to the atypical
cases 1 and 2 or A and B shown in Fig. 1. In other words, these individuals had opposite CR and
CL or CL and CS variable values.
A prerequisite for the di!erentiation between the three target constructs also having practical
relevance is that it should help to identify supply side management variables that only selectively
a!ect the three target constructs. In our study, such parameters were the variables `phone number
constancya, `image of competitorsa, and `network qualitya (see Fig. 2 and Table 2). In particular,
a lack of number portability between operators of mobile communications networks appears to act
as a barrier that prevents customers from terminating the contract with their network operator,
even if their loyalty or satisfaction is low. This is because they would lose their current number if
they were to change to another provider. For operators of mobile communications networks with
a large customer base, this "nding implies that they should organise professional political lobbying
activities in every national market that help to reduce competitive pressures by perpetuating
a regulation that does not enforce portability of phone numbers in current GSM networks and in
the next generation of technology for mobile communications networks (UMTS/IMT-2000). In
Germany such lobbying activities of mobile operators were recently unsuccessful since the German
regulatory authority RegTP decided to enforce complete number portability between operators of
mobile communications networks in February 2002. According to our "ndings this change is likely
to increase considerably the intensity of competition in Germany at a stage of market development
where the number of new "rst-time customers is declining and where acquisition of old customers
from their competitors will become an increasingly important ingredient of mobile operators'
competitive strategies.
In order to develop sound competitive and marketing strategies, it would seem advisable for
mobile network operators not just to rely on general indicators of CL and CS when analysing the
threat of migration by their existing customers. They should rather seek to improve their
measurements of customers' perceptions of characteristics of the core services they themselves o!er.
There are two elements that provide the most important early-warning signals for the degree of CR
and which also act as a `levera to motivate customers to continue their contractual relationship
with a provider. These are, "rst, the customer assessment that the prices charged by their supplier
are `good and faira (compared against competitors), and, second, the customer perception of the
functional bene"t of mobile communications services. The use of pricing policies to achieve positive
CR e!ects is especially di$cult for operators of mobile communications networks because, up to
now, competitors were very quick in neutralising temporary advantages in price level or price
structure by introducing modi"ed pricing schemes of their own (see for details Wilfert, 1999, p. 197;
Gerpott & KnuK fermann, 1998, pp. 144}147; Stoetzer & Tewes, 1996, p. 307). However, customerspeci"c discounts * somewhat similar to `bonus milesa in aviation * can be made dependent on
the duration of a contract and on the amount that services are used. Operators of mobile networks

T.J. Gerpott et al. / Telecommunications Policy 25 (2001) 249}269

267

who "rst succeed in creating the complex billing procedures for this type of pricing strategies and
who manage to communicate these plans in a way that a!ects the mass market, will be able to
strengthen the ties of their residential customers to their "rm because such strategies have hardly
been used in the past in Germany (cf. Gerpott, 1998, pp. 213}214, 294; Knauer, 1998, pp. 515}516).
Perceived personal bene"t obtained from mobile communications services has considerable
direct and indirect e!ects on CR. This suggests that mobile operators within the residential
customer market in Germany ought to strive to become pioneers in service innovations that
improve the core functionalities of mobile communications services (fast, simple, and trouble-free
transfer of all types of information from any place at any time). More speci"cally, if, between 2001
and 2004, an operator of mobile communications networks in Germany is the "rst provider to
introduce functioning fast(er) mobile data services and the next mobile network generation with
broadband multimedia services onto the market he will have the opportunity to forge stronger ties
with residential customers as well.
Finally, it should be noted that two `non-"ndingsa have practical implications for competitive
strategies of mobile operators in the residential customer market. First, there appear to exist only
a few residential customers who terminate the contract with their network operator because they
want to obtain a new terminal at a low price by switching to another provider once their contract
comes to an end. Consequently, it may not be necessary for mobile operators to o!er their existing
residential customers up-to-date models of terminals at a price that is below the initial cost of the
device. Second, the quality of customer care * whether provided over the phone by call centre
service agents, or by a continuous supply of written product or company information * has no
signi"cant direct or indirect e!ect on CR. To wit, popular claims of consultants (for others, see
Harter et al., 1997, pp. 160}162; Heidemann & Zynga, 1997, pp. 19}21) that, even in the case of
standard telecommunications services for the mass market, customer care is a very important
variable for network operators in improving customer acquisition and CR were not substantiated.
This "nding may be taken to indicate that the quality of care for residential customers in the case of
standard telecommunication services represents a `hygiene factora: If customer care is poor, it will
contribute to dissatisfaction; but if it is good, it will not improve CS. Accordingly, mobile operators
would be well advised to maintain the telco industry standard in customer care processes, but
should not attempt to obtain a competitive strategic advantage by providing an extraordinary level
of customer care.

6. Implications for future research

Directions for future research on retention, loyalty, and satisfaction of cellular service buyers in
particular and of buyers of service-like innovative contract goods in general follow from four major
limitations of the present study. First, our analysis relies on cross-sectional data. Thus, to provide
an even more convincing case for causal interpretations of variable correlations, additional
longitudinal research is needed in which exogenous factors are captured before data on endogenous
criteria are collected. Second, in the present study the focal constructs CR and CL were measured
by questionnaire items asking for behavioural intentions and attitudes related to service o!erings
of cellular operators. New research can help by expanding this measurement approach to include
indicators of actual customer usage behaviours such as number of months passed since a customer

268

T.J. Gerpott et al. / Telecommunications Policy 25 (2001) 249}269

"rst signed the contract with her cellular operator, customer air time minutes/revenues per year or
number of times a customer recommended her operator to another person within a speci"ed
period.
Third, our work did not take into account observations made by Bolton (1998, pp. 52}62) in the
US cellular residential market, which suggest that the magnitude of correlations between cellular
service attributes and CS on the one hand and CL and CR on the other hand may be contingent on
the past duration of the contractual relationship between a customer and his cellular service
operator. As a consequence, additional studies are needed exploring CR and CL determinants in
various residential customer sub-samples which di!er in terms of the duration of their previous
relationship with a mobile operator. Fourth, the present investigation placed a deliberate focus on
supply side mobile service attributes a!ecting CR, CL, and CS. Therefore, further research should
include customer side variables re#ecting speci"c individual customer goals and motivations in
using cellular services as well as general psychological constructs such as need for variety which
past work discussed in the context of repeat purchase behaviours. Similarly, empirical research
exploring determinants of CR, CL, and CS in other telecommunications service market segments (e.g.,
dial-up internet access) would be bene"cial to learn to what extent our "ndings can be generalised
beyond the cellular market.
To sum up, given the paucity of previous research on CR, CL, and CS in competitive (mobile)
telecommunications service markets there exists ample opportunity for management scholars and
practitioners alike to contribute towards improved carrier pro"tability by expanding our understanding of antecedents and consequences of the duration of an operator's contractual relationship
with his mass market customers.

References
Berke, J. (1999). Oligopol sprengen: Mehr Wettbewerb im Mobilfunk. Wirtschaftswoche, 53(41), 94.
Bliemel, F. W., & Eggert, A. (1998). Kundenbindung * die neue Sollstrategie? Marketing, 20, 37}46.
Bolton, R. N. (1998). A dynamic model of the duration of the customer's relationship with a continuous service provider:
The role of satisfaction. Marketing Science, 17, 45}65.
Booz Allen, & Hamilton (1995). Mobilfunk. Frankfurt/M.: IMK.
Bruhn, M., & Homburg, C. (Eds.). (1998). Handbuch Kundenbindungsmanagement. Wiesbaden: Gabler.
Danaher, P. J., & Rust, R. T. (1996). Indirect "nancial bene"ts from service quality. Quality Management Journal, 3(2),
63}75.
Dick, A. S., & Basu, K. (1994). Customer loyalty: Toward an integrated conceptual framework. Journal of the Academy of
Marketing Science, 22, 99}113.
Diller, H. (1996). Kundenbindung als Marketingziel. Marketing, 18, 81}94.
Gerpott, T. J. (1998). Wettbewerbsstrategien im Telekommunikationsmarkt (3rd ed.). Stuttgart: SchaK !er-Poeschel.
Gerpott, T. J. (1999). Strukturwandel des deutschen Telekommunikationsmarktes. In D. Fink, & A. Wilfert (Eds.),
Handbuch Telekommunikation und Wirtschaft (pp. 49}75). MuK nchen: Vahlen.
Gerpott, T. J. (2000). Kundenbindung * Konzepteinordnung und Bestandsaufnahme der neueren empirischen
Forschung. Die Unternehmung, 54, 23}42.
Gerpott, T. J., & KnuK fermann, M. (1998). Mobilfunk * Entwicklung des liberalisierten deutschen Markts. Betriebswirtschaftliche Bla( tter, 47, 140}150.
Grund, M. A. (1998). Interaktionsbeziehungen im Dienstleistungsmarketing. Wiesbaden: Gabler.
Harter, G., Ripsam, T., & Ruhl, M. (1997). Customer care * Kundenbindung zum Nulltarif. In Booz. Allen, & Hamilton
(Eds.), Telekommunikation in der Welt von morgen (pp. 151}166). Frankfurt/M.: IMK.

T.J. Gerpott et al. / Telecommunications Policy 25 (2001) 249}269

269

Heidemann, D., & Zynga, A. M. (1997). Wettbewerbsvorteile im Telekommunikationsmarkt durch `World-Class

Customer Carea. Telekom Praxis, 74(2), 18}22.
Herrmann, A., & Johnson, M. D. (1999). Die Kundenzufriedenheit als BestimmungsgroK {e der Kundenbindung.
Schmalenbachs Zeitschrift fu( r betriebswirtschaftliche Forschung, 51, 579}598.
Homburg, C., & Baumgartner, H. (1995). Beurteilung von Kausalmodellen. Marketing, 17, 162}176.
Homburg, C., & Bruhn, M. (1998). Kundenbindungsmanagement * Eine EinfuK hrung in die theoretischen and praktischen Problemstellungen. In M. Bruhn, & C. Homburg (Eds.), Handbuch Kundenbindungsmanagement (pp. 3}35).
Wiesbaden: Gabler.
Homburg, C., Fa{nacht, M., & Werner, H. (1998). Operationalisierung von Kundenbindung and Kundenzufriedenheit.
In M. Bruhn, & C. Homburg (Eds.), Handbuch Kundenbindungsmanagement (pp. 389}410). Wiesbaden: Gabler.
Homburg, C., Giering, A., & Hentschel, F. (1999). Der Zusammenhang zwischen Kundenzufriedenheit and Kundenbindung. Die Betriebswirtschaft, 59, 174}195.
Homburg, C., & Hildebrandt, L. (1998). Die Kausalanalyse. In L. Hildebrandt, & C. Homburg (Eds.), Die Kausalanalyse
(pp. 16}43). Stuttgart: SchaK !er-Poeschel.
Homburg, C., & Rudolph, B. (1997). Theoretische Perspektiven zur Kundenzufriedenheit. In H. Simon, & C. Homburg
(Eds.), Kundenzufriedenheit (2nd ed.) (pp. 31}54). Wiesbaden: Gabler.
Jeschke, K. (1995). Nachkaufmarketing. Frankfurt/M: Lang.
Joho, C. (1996). Ein Ansatz zum Kundenbindungsmanagement fu( r Versicherer. Bern: Haupt.
JoK reskog, K. G., & SoK rbom, D. (1996). Lisrel 8: User's reference guide. Chicago: SSI.
Keaveney, S. M. (1995). Customer switching behavior in service industries. Journal of Marketing, 59(2), 71}82.
Knauer, M. (1998). Kundenbindung in der Telekommunikation: Das Beispiel T-Mobil. In M. Bruhn, & C. Homburg
(Eds.), Handbuch Kundenbindungsmanagement (pp. 507}522). Wiesbaden: Gabler.
KruK ger, S. M. (1997). Proxtabilita( tsorientierte Kundenbindung durch Zufriedenheitsmanagement. MuK nchen: FGM.
Laker, M., Pohl, A., & Dahlho!, D. (1999). Neue MaK rkte: Kunden gewonnen * Was kommt dann? In H. H.
Hinterhuber, & K. Matzler (Eds.), Kundenorientierte Unternehmensfu( hrung (pp. 85}97). Wiesbaden: Gabler.
Me!ert, H. (1998). Kundenbindung als Element moderner Wettbewerbsstrategien. In M. Bruhn, & C. Homburg (Eds.),
Handbuch Kundenbindungsmanagement (pp. 115}133). Wiesbaden: Gabler.
Mellewigt, T. (1997). Kommentar zu A 43 TKG. In W. BuK chner, J. Ehmer, M. Geppert, B. Kerkho!, H.-J. Piepenbrock,
R. SchuK tz, & F. Schuster, (Eds.), Beck'scher TKG-Kommentar (pp. 556}578). MuK nchen: Beck.
Oevermann, D. (1996). Kundenbindungsmanagement von Kreditinstituten. MuK nchen: FGM.
Peter, S. I. (1997). Kundenbindung als Marketingziel. Wiesbaden: Gabler.
Rust, R. T., & Zahorik, A. J. (1993). Customer satisfaction, customer retention, and market share. Journal of Retailing, 69,
193}215.
Schwarz-Schilling, C., & Stumpf, U. (1999). Netzbetreiberportabilita( t im Mobilfunkmarkt (Report of the WIK for RegTP,
hectograph manuscript, VI#68pp.).
Simon, H., & Homburg, C. (Eds.). (1997). Kundenzufriedenheit (2nd ed.). Wiesbaden: Gabler.
Stoetzer, M.-W., & Tewes, D. (1996). Competition in the German cellular market? Telecommunications Policy, 20,
303}310.
Szyperski, N., & Loebbecke, C. (1999). Telekommunikationsmanagement (TKM) als betriebswirtschaftliche Spezialdisziplin. Die Betriebswirtschaft, 59, 481}495.
Weinberg, P. (1998). Verhaltenswissenschaftliche Aspekte der Kundenbindung. In M. Bruhn, & C. Homburg (Eds.),
Handbuch Kundenbindungsmanagement (pp. 39}53). Wiesbaden: Gabler.
Wilfert, A. (1999). Der Wettbewerb auf dem Mobilfunkmarkt in Deutschland. In D. Fink, & A. Wilfert (Eds.), Handbuch
Telekommunikation und Wirtschaft (pp. 187}202). MuK nchen: Vahlen.
Woodru!, R. B. (1997). Customer value: The next source for competitive advantage. Journal of the Academy of Marketing
Science, 25, 139}153.