Directory UMM :Data Elmu:jurnal:J-a:Journal Of Business Research:Vol49.Issue3.2000:

Effects of Supplier Reliability and
Benevolence in Business Marketing
Fred Selnes
NORWEGIAN SCHOOL OF MANAGEMENT–BI

Kjell Gønhaug
NORWEGIAN SCHOOL OF ECONOMICS AND BUSINESS ADMINISTRATION

The authors address the important issue of how supplier behavior in
terms of reliability and benevolence creates positive and negative affect,
influences customer satisfaction, and subsequently behavioral intentions
to be loyal to the supplier. Recent research suggests that affective responses
are important, but has mainly focused on products’ performance in consumer markets. The authors extend past research by examining affective
and cognitive responses to suppliers’ performance in an industrial context.
In a study of 150 established buyer–seller relationships in the industrial
telecommunication market, they found that customers’ affective responses
to supplier reliability were different from their responses to supplier
benevolence. Low supplier reliability was found to create negative affect,
while high supplier benevolence created positive affect. Supplier reliability
showed a strong positive effect on satisfaction with the supplier, which
subsequently increased loyalty. Supplier benevolence appeared to have

no direct effect on overall satisfaction, but was found to influence customer
loyalty indirectly through positive affect. Thus, both cognitive and affective
processes mediate the effects of supplier reliability and supplier benevolence
on loyalty. J BUSN RES 2000. 49.259–271.  2000 Elsevier Science
Inc. All rights reserved.

S

upplier reliability and benevolence have been identified
as two key factors in developing relationships based
on trust and commitment (e.g., Kumar, Scheer, and
Steenkamp, 1995). Reliability is the perceived ability to keep
an implicit or explicit promise. A supplier is perceived as
reliable when deliveries are made according to contract, when
relevant information is provided timely and accurately, when
members of the organization are knowledgeable about their
business and their products, and so on (e.g., Kumar et al.,
1995; Biong and Selnes, 1997). Supplier benevolence is the
perceived willingness of the supplier to behave in a way that
benefits the interest of both parties. A supplier is perceived

as benevolent if they are willing to make an extra effort when

unexpected problems arise (Kumar et al., 1995). Given that
the supplier has a goal to develop and maintain customer
loyalty, it is important to understand how reliability and benevolence can create such results. The purpose of this article
is to develop and test a model of how affective and cognitive
processes mediate the effect of supplier reliability and benevolence on behavioral intention to be loyal to the supplier.
Affective responses have received considerable attention in
recent research on consumer satisfaction (e.g., Westbrook,
1987; Westbrook and Oliver, 1991; Oliver, 1993; Mano and
Oliver, 1993). Despite growing acknowledgment of the importance of affect in consumer markets, modest attention has
been given to affect in industrial markets. Although it does
not make sense to talk about feelings as a characteristic of an
organization itself, members of the organizational buying team
may have feelings toward a supplier as an organization. Individuals involved in decision making are, as reflected in the
buying-center literature, influenced by among other things
their subjective experiences (Johnston and Bonoma, 1981;
Kohli, 1989). Although not explicitly discussed in the literature, it is reasonable to expect affective and cognitive responses
also to be present in buyer–seller relationships, and to influence decision making in industrial markets. Anecdotal evidence from business practice in industrial markets indicates
that marketers exert behaviors that are directed toward affective responses.

The remainder of the article is organized as follows: after
discussing underlying theory, we develop a model and derive
hypotheses related to affective and cognitive responses to supplier behavior. Next, we report the research method used and
the results of the empirical test. Finally we discuss future research possibilities and managerial implications of the findings.

Theory and Hypotheses
Supplier Behavior

Address correspondence to Fred Selnes, Professor of Marketing, Norwegian
School of Management-BI, Oslo, Norway. Tel.: 14722985101; E-mail:
fred.selnes@bi.no
Journal of Business Research 49, 259–271 (2000)
 2000 Elsevier Science Inc. All rights reserved.
655 Avenue of the Americas, New York, NY 10010

Ravald and Gro¨nroos (1996) propose that customers value
not only the focal product, but also the firm supplying the
ISSN 0148-2963/00/$–see front matter
PII S0148-2963(99)00017-X

260

J Busn Res
2000:49:259–271

product or service, and that the two entities represent different
processes in creating value. We will focus on the supplier’s
ability to create value through being reliable and showing
benevolence. Clearly, a supplier should be reliable and fulfill
what is promised to the customer. However, a supplier could
be benevolent in situations in which there are no explicit
or implicit promises. More precisely, customers expect the
supplier to solve unpleasant events that cause problems for
them when they believe such events to be within the supplier’s
domain (i.e., within the supplier’s responsibility, whether the
supplier’s promise is explicit or implicit). However, even if
unpleasant events are outside the supplier’s perceived responsibility, the supplier could help the customer. Showing benevolence is believed to be interpreted differently by the customers than is the supplier’s ability to be reliable, and hence
customers are likely to respond differently toward the seller
in these situations. When the focus is on a single transaction,
the distinction between reliability and benevolence will not

be important or relevant. However, in the perspective of an
ongoing relationship, the distinction may be crucial. When
committed to the continuity of a relationship, the supplier
may be willing to help or surprise positively the customer in
other ways, even when such efforts are outside what is implicitly or explicitly promised.

Affective Responses
Customers’ affective responses are important for several reasons. For example, recent advances in social cognition, cognitive psychology, and social psychology suggest that affective
processes constitute not only a powerful source of motivation,
but is also a major influence on information processing and
choice (see Lazarus, 1991, for an excellent overview). Affect is
generally understood to comprise a class of mental phenomena
characterized by a consciously experienced, subjective state
of feeling, commonly accompanying emotions and moods.
Several taxonomies have been proposed to classify the variety
of subjective feelings into a small set of fundamental, or primary effects (e.g., Izard, 1977). Russell (1980) suggests that
emotions can be described in terms of two primary dimensions
that define a circular configuration, commonly referred to as
a circumplex. The dimensions are pleasure/displeasure and
degree of arousal. Westbrook (1987) posited that affect has

a similar structure and can be classified according to valence,
that is either positive, neutral, or negative. Oliver (1993) found
that attribute satisfaction and attribute dissatisfaction had
crossover influences on both positive and negative affect.
These findings also confirm the “affect-balance theory” proposed by Bradburn (1969), which recognized that positive
experiences (or positive affect) are not necessarily inversely
correlated with negative experiences (or negative affect), predicting that positive and negative affect make independent
contributions to satisfaction (Oliver, 1993). This implies,
among others, that positive and negative affect about an object
(such as a supplier) can coexist at the same time and over time,

F. Selnes and K. Gønhaug

while negative and positive cognitions (such as satisfaction
and dissatisfaction with a supplier) rarely co-occur (Edell and
Burke, 1987).
Because suppliers should be reliable, it is reasonable to
expect negative emotions to occur when they are experienced
not to be so. Attribution theory may be useful in understanding
when negative and positive feelings carry over to positive and

negative affect toward the supplier (Deighton, 1992; Oliver,
1997, p. 281–282). The basic point of departure in attribution
theory is that the individual tries to make sense of the social
context in which he or she is embedded (see Folkes, 1988;
Harvey and Weary, 1984 for excellent reviews). Research has
demonstrated that attributions of motivation may be selfserving or hedonic, typically expressed as attributing favorable
outcomes to oneself and unfavorable outcomes to external
forces (cf. Miller and Ross, 1975). Deviance from promised
reliability (“the supplier should”) can be both positive and
negative. It follows from attribution theory that negative deviance from what is promised, particularly in routine or ordinary
activities and operations, would be attributed to the supplier.
Negative experiences may evoke affective responses of negative
valence (e.g., anger). Further, when poor performance is attributed to the supplier, the evoked negative feelings would
carry over to the supplier (Oliver, 1993). It also follows from
attribution theory that positive outcomes (i.e., high reliability)
would be attributed to “self,” that is, the buyer (e.g., “I did
well because I was careful to choose a good supplier”) or to
the specific situation (e.g., “I had a lucky day”). In such
situations, evoked positive affective feelings are not likely to
be carried over to the supplier.

H1: Perceived low supplier reliability leads to negative
affect toward the supplier.
Because benevolence is likely to be perceived as positive,
showing benevolence is likely to create positive affect. In cases
of supplier benevolence (i.e., situations in which “the supplier
could”), attribution processes are likely to be different from
experienced reliability. Recall that situations of demonstrated
benevolence are outside the contract or promise, but still
within the relationship. Experienced benevolence is likely to
be attributed to the supplier and perceived as a friendly act
(e.g., “He didn’t have to help me, but he did.”). Such a positive
experience is likely to create positive affective arousal and
even excitement (Russell, 1980). When the supplier does not
help the customer in an unpleasant situation, and thus is not
benevolent (only “neutral”), we predict that the outcome is
likely to be attributed to self and/or situation. We predict that
the affective arousal in this case is more in the direction of
depression and sadness attributed to oneself (see also Oliver,
1997, p. 280), and thus not a negative feeling toward the
supplier.

H2: Perceived supplier benevolence enhances positive affect toward the supplier.

Effects of Supplier Reliability and Benevolence

Cognitive Responses
Supplier behavior is predicted to influence cognitive responses
in terms of customer satisfaction and behavioral intention.
Customer satisfaction is believed to be a function of expectations and experienced performance of a product or service
offering (e.g., Oliver, 1980; Oliver and DeSarbo, 1988).
Churchill and Surprenant (1982) found that level of performance had a direct effect on satisfaction. Oliver (1997, p.
120) reported several other studies showing direct effects from
performance to satisfaction. Thus, we propose that performance of a supplier in terms of reliability and benevolence
will influence satisfaction with the supplier.
H3: Experienced supplier reliability positively influences
satisfaction with the supplier.
H4: Experienced supplier benevolence positively influences satisfaction with the supplier.
Westbrook (1987) found that positive and negative affect
carry over to satisfaction, which has been confirmed in subsequent studies (Evrard and Aurier, 1994; Mano and Oliver,
1993; Oliver, 1993). Thus, we hypothesize that:
H5: Positive affect toward the supplier increases satisfaction with the supplier.

H6: Negative affect toward the supplier reduces satisfaction with the supplier.
Reichheld (1996) pointed out that customer loyalty should
be the goal of a company and not satisfaction per se. Multiple
studies have found strong positive relationships between satisfaction and behavioral intentions such as expressed future
loyalty toward a supplier or product. Customer satisfaction
is important for several reasons. In highly competitive markets
customers will consider alternative sources of supply if not
satisfied, particularly when exit barriers from an existing relationship are modest (cf. Fornell and Wernerfelt, 1987). Enhanced customer satisfaction is argued to motivate the customer to continue to transact with the supplier (e.g., Fornell,
1992; Richins, 1983; Singh, 1988), and conversely to reduce
the likelihood of exit from the relationship with the supplier

Figure 1. Theoretical model of the hypothesized affective and cognitive responses to supplier behavior.

J Busn Res
2000:49:259–271

261

(e.g., Hirchman, 1970; Kelly and Thibaut, 1978). When buyers engage in product-related conversations, they seek and
give advice about products/services and suppliers. Of particular value for the marketer are customers who “witness” positively in favor of the supplier (e.g., Blodgett, Granbois, and

Walters, 1993; Richins, 1983). Finally, satisfaction with a
supplier may result in more business and a stronger commitment and thus motivation to expand the scope of the relationship (e.g., Johanson, Halle´n, and Seyed-Mohamed, 1991; Morgan and Hunt, 1994; Gundlach, Achrol, and Mentzer, 1995).
Thus, we hypothesize that:
H7: Customers’ satisfaction with a supplier enhances behavioral intention and motivation to be loyal to the
supplier.
Figure 1 summarizes the preceding discussion.

Methods
The data used to test the hypotheses were collected in a
telephone survey of business customers of a telecommunication company. A professional research firm conducted the
interviews. Because companies differ in their use of telecommunication products and services depending on the nature
of the business and the information technology employed,
they represent a large variety of needs. We wanted the sample
of relationships to reflect complexity in interactions between
supplier and customer. In addition, we wanted the sample
to reflect established relationships in which customers had
sufficient experience with the supplier to evaluate relevant
attributes describing the supplier. To create such a sample of
relationships we asked the company’s account managers to
provide a list of customers with which they had an established
relationship. A random sample of 250 customer relationships
was drawn from the total list of reported relationships to avoid
systematic biases.
We also asked the account managers to provide the names
of the contact persons within the buyers’ organizations. Where
several contact persons were used, we asked for the name of

262

J Busn Res
2000:49:259–271

the person who had the most formal authority and knowledge
about the focal relationship. Qualitative interviews conducted
prior to the main study indicated that the identified contact
persons had a good overview and knowledge of the relationship with the supplier and that they were also motivated to
provide the type of information wanted. These observations
are in accordance with the experience of other researchers
studying industrial buyer–seller relationships (Anderson and
Weitz, 1989; Heide and John, 1990). Hence, the person contacted was expected to possess the qualifications necessary to
serve as key informant (Campbell, 1955).
The contact persons identified from the supplier’s list were
telephoned. Several callbacks were necessary to increase the
response rate. However, some prospective respondents were
not available at the time of data collection. Only a few people
refused to answer. All of the qualified persons contacted evaluated the relationship as complex. A total of 150 respondents
completed the interview (i.e., a response rate of 60%). On
average the interviews lasted about 15 minutes.
Respondents were asked to describe the relationship with
the supplier in terms of its relative cost for their firms. On a
six-point scale (1 5 “very little,” 6 5 “very much”), the mean
score was 3.61 with a standard deviation of 1.61. Hence,
the relationship with the supplier represented on average a
substantial cost to the firms investigated. On a similar scale,
we measured the strategic importance of the relationship. The
mean score was 4.46, and skewed toward high importance.
Hence, the sample consisted of buyers who viewed the relationships with their supplier as strategically important. Seventy-six percent reported that the relationship had lasted 10
years or longer.

Development of Measures
Supplier reliability (RELIAB) is defined as the supplier’s ability
to keep their promise. Building on prior work by Kumar et
al. (1995) and Biong and Selnes (1997), we used five items
to capture this construct: (1) ability to deliver according to
contract; (2) provision of enough and relevant information; (3)
trust in provided information; (4) trustworthiness (expertise);
and (5) overall reliability of the supplier. Responses were given
on six-point scales with endpoint “not at all” (1) and “very
much” (6).
Supplier benevolence (BENEVOL) is defined as a perceived
willingness of the supplier to behave in a way that benefits
the interest of both parties in the relationship. We adopted
five items used by Kumar et al. (1995) to measure supplier
benevolence. These were (1) willingness to support the customer if the environment causes changes; (2) consideration
of the customer’s welfare when making important decisions;
(3) responding with understanding when problems arise; (4)
consideration of how future decisions and actions will affect
the customer; and (5) dependable support on things that are
important to the customer. Responses were given on six-point
scales with endpoint “not at all” (1) and “very much” (6).
Affect was measured with two variables defined as the

F. Selnes and K. Gønhaug

positive (POSAFF) and negative (NEGAFF) feelings toward
the supplier (Mano and Oliver, 1993). Although positive and
negative affect are negatively related, several studies find that
these are actually two different constructs (e.g., Bradburn,
1969; Watson, Clark, and Tellegen, 1988; Russell, 1980),
and thus separate scales for positive and negative affect were
developed. Construction of the scales started with the list of
positive and negative items reported by Watson et al. (1988).
Qualitative interviews with three customers were conducted
to assess the applicability of those items in the study context.
We found that items directed toward a supplier became irrelevant when we used terms like “strong,” “proud,” “guilty,” and
“scared.” After the interviews we decided to keep four positive
and five negative items of the original scales. In addition, the
qualitative interviews suggested “useful” as a possible positive
item describing feelings toward the supplier. The positive
items were: (1) enthusiastic, (2) excited, (3) inspired, (4)
interested, and (5) useful. The negative items were: (1) distressed, (2) nervous, (3) jittery, (4) irritable, and (5) upset.
Subjects indicated on six-point scales (1 5 “not at all” to
6 5 “very much”) the degree to which the specific emotion
described their feelings toward the supplier.
Satisfaction (SATISFA) is defined as the (cognitive) overall
evaluation of the relationship with the supplier. We used
three items previously suggested by Fornell (1992), to assess
satisfaction with the supplier: (1) general satisfaction, (2) confirmation of expectations, and (3) the distance from the customer’s hypothetical ideal supplier. For the first two items,
subjects indicated on six-point scales (1 5 “not at all” to 6 5
“very much”) the degree to which they felt the specific question
described their relationship with the supplier. For the third
item, subjects indicated on a six-point scale the distance from
an ideal supplier (1 5 “far away” to 6 5 “very close”).
Behavioral intention (BEINTEN) is defined as the motivation to be loyal to the supplier. We used four items to assess
this variable. First item is intention to change the share of
business with a supplier (Kumar et al., 1995). We measured
this variable on a five-point scale (1 5 “much less than before,”
2 5 “less,” 3 5 “about the same,” 4 5 “more,” and 5 5
“much more”). In line with Hirschman (1970), we measured
continuity/exit and voice. Continuity/exit was measured with
one item, the likelihood of exit from the relationship within
two years, on a six-point scale ranging from 1 5 “not very
likely” to 6 5 “very likely” (Kumar et al., 1995). In Hirschman’s terminology, voice represents negative reactions toward
the supplier. In marketing, voice as word of mouth is directed
toward someone other than the supplier, and thus its effects
are external to the relationship but important to the supplier.
As in previous studies (Blodgett et al., 1993), word-of-mouth
was measured as the likelihood that the customer would give
positive recommendations if asked for an opinion about the
supplier. Two items were used, one reflecting telling professional colleagues outside the organization about experiences,

Effects of Supplier Reliability and Benevolence

J Busn Res
2000:49:259–271

263

Table 1. Initial Measurement Model—Standardized Coefficients
BEINTEN
SHARE
EXIT
POSWO1
POSWO2
SATISFA1
SATISFA2
SATISFA3
POSAFF1
POSAFF2
POSAFF3
POSAFF4
POSAFF5
NEGAFF1
NEGAFF2
NEGAFF3
NEGAFF4
NEGAFF5
RELIAB1
RELIAB2
RELIAB3
RELIAB4
RELIAB5
BENEVOL1
BENEVOL2
BENEVOL3
BENEVOL4
BENEVOL5

SATISFA

POSAFF

NEGAFF

RELIAB

BENEVOL

0.43
0.69
–
0.93
0.82
0.90
0.70

and one related to telling others inside the buyer’s organization
(measured on six-point scales from 1 5 “not very likely” to
6 5 “very likely”). Thus, the four items reflecting behavioral
intention are: (1) share of supply, (2) exit from the relationship, (3) positive word-of-mouth to external colleagues, (4)
positive word-of-mouth to internal colleagues.
A total of 27 items were used to capture the six constructs.
The means, standard deviations, skewness, and kurtosis of
the items are reported in Appendix 1. The correlation matrix
is reported in Appendix 2. To avoid multicollinearity we removed items POSWO1, POSAFF1 and BENEVOL5 as these
correlated above 0.80 with other items.
Validity of the measurement model was addressed in several steps. First, by using the maximum-likelihood procedure
in LISREL VIII, we estimated the fit of the measurement model.
Second, we removed items that loaded low on their respective
latent variable or items that loaded high on other latent variables as indicated by modification indexes (m.i.). Finally we
assessed the overall fit of the purified model.
The initial measurement model consisted of 24 items and
six latent variables. The overall fit of the initial measurement
model is low, with a chi-square of 920.56 (d.f. 5 237), rootmean square error of approximation (RMSEA) of 0.14, an
adjusted goodness-of-fit index (AGFI) of 0.66, and a normed
fit index (NFI) of 0.71. All items load significantly on their
respective factors, but four items had low (standardized) load-

–
0.83
0.85
0.72
0.73
0.77
0.72
0.88
0.84
0.89
0.56
0.61
0.68
0.75
0.77
0.74
0.74
0.88
0.83
–

ings below 0.7. The loading-matrix is reported in Table 1.
We removed SHARE, RELIAB1, RELIAB2, and RELIAB3. The
measurement model was re-estimated with 20 items. The
overall fit of the model is still poor and we inspected the
modification indexes in order to identify items that appeared
to correlate high with other constructs. As modification indexes are interrelated we removed items sequentially and reestimated the model after each removal. This procedure was
repeated until modification indices were below five for all
remaining items. In this process we removed POSAFF3, POSAFF5, NEGAFF1, NEGAFF2, NEGAFF4, and BENEVOL2.
The measurement model was re-estimated with 14 items and
six latent variables. The overall fit of the purified model is
now satisfactory with a chi-square of 102.88 (d.f. 5 62 and
p 5 0.00086), root-mean square error of approximation
(RMSEA) of 0.067, an adjusted goodness-of-fit index (AGFI)
of 0.86, and a normed fit index (NFI) of 0.93. The loadingmatrix of the estimated measurement model is reported in
Table 2.
We computed two reliability indices for each construct
based on the estimated measurement model. The first reliability measure is a composite score, whereas the second is an
indication of average variance extracted for the construct (Dillon and Goldstein, 1984, p. 480). The reliabilities for each
construct are reported in the bottom rows of Table 2. As can
be seen, all reliability 1 coefficients except POSAFF are 0.7

264

J Busn Res
2000:49:259–271

F. Selnes and K. Gønhaug

Table 2. Estimated Measurement Model: Standardized Coefficients
BEINTEN
SHARE
EXIT
POSWO1
POSWO2
SATISFA1
SATISFA2
SATISFA3
POSAFF1
POSAFF2
POSAFF3
POSAFF4
POSAFF5
NEGAFF1
NEGAFF2
NEGAFF3
NEGAFF4
NEGAFF5
RELIAB1
RELIAB2
RELIAB3
RELIAB4
RELIAB5
BENEVOL1
BENEVOL2
BENEVOL3
BENEVOL4
BENEVOL5

–
0.70
–
0.92

Reliability 1
Reliability 2

0.80
0.67

SATISFA

POSAFF

NEGAFF

RELIAB

BENEVOL

0.82
0.91
0.69
–
0.77
–
0.65
–
–
–
0.89
–
0.88
–
–
–
0.72
0.77
0.76
–
0.87
0.79
–
0.85
0.66

or higher which indicate internal consistency among the measures. POSAFF has reliability 1 coefficient of 0.67, which is
close to the recommended level. All reliability 2 indices were
above the recommended 0.5. In sum, we concluded that the
measurement model was satisfactory.

Results
We estimated the hypothesized model by using the maximumlikelihood procedure in LISREL VIII. Supplier behavior performance, RELIAB and BENEVOL were included as exogenous
variables and the others as endogenous variables. The hypothesized model has an acceptable fit with a chi-square of 134.23
(d.f. 5 69), an RMSEA of 0.08, an AGFI of 0.84 and a NFI
of 0.91. Five of the seven structural paths in the hypothesized
model are statistically significant and in the expected direction
(see Table 3). Neither the direct hypothesized path from benevolence on satisfaction nor the indirect effect from positive
affect on satisfaction is confirmed. Thus the hypothesized
effect of supplier benevolence on behavioral intention mediated through affect and satisfaction is not confirmed. As expected, supplier benevolence has a significant effect on positive affect. Supplier reliability has an effect on satisfaction, both
directly and indirectly through negative affect. As expected,
satisfaction has a significant effect on behavioral intention.

0.67
0.51

0.88
0.79

0.71
0.55

0.85
0.65

Although overall fit of the hypothesized model is acceptable, we decided to examine modification indexes to explore
if the model can be improved through better specifications of
the structural paths. The examination revealed three potential
paths with relatively large modification indexes. These were:
(1) an effect from positive affect on behavioral intention (m.i 5
9.87); (2) an effect from behavioral intention on satisfaction (m.i. 5 16.75); and (3) an effect from benevolence on
behavioral intention (m.i. 5 10.64). As these were quite large
indications that the model was not correctly specified, we
decided to explore how the overall fit could be improved by

Table 3. Estimated Structural Model: Standardized Coefficients
Path
SATISFA → BEINTEN
POSAFF → SATISFA
NEGAFF → SATISFA
RELIAB → SATISFA
RELIAB → NEGAFF
BENEVOL → POSAFF
BENEVOL → SATISFA
RELIAB ↔ BENEVOL
** p , 0.01; * p , 0.05.

BE(1,2)
BE(2,3)
BE(2,4)
GA(2,1)
GA(4,1)
GA(3,2)
GA(2,2)
PHI(1,2)

Standardized

t-value

0.86
0.19
20.20
0.79
20.56
0.95
20.06
0.50

8.78**
0.58
23.88**
2.16*
24.62**
8.78**
20.12
5.96**

Effects of Supplier Reliability and Benevolence

J Busn Res
2000:49:259–271

265

Figure 2. Estimated adjusted model for affective and cognitive responses to supplier
reliability and benevolence.

following the suggestions given by the modification indexes.
As modification indexes will change depending on new structural paths we followed a procedure of sequential re-specification. The purpose of our research was to find the mechanisms
by which supplier behavior in terms of reliability and benevolence affects behavioral intentions. Our hypotheses were that
theses effects were mediated through satisfaction as a cognitive
route from affect to behavioral intention. A competing hypothesis, as discussed in Edell and Burke (1987), is that affect has
an independent route on motivation to be loyal. Thus, we
decided to test out if the effect of positive affect has a direct
effect on behavioral intention.
The modified model produced a significantly better fit with
an improved chi-square of 13.01 (d.f. 5 1). The structural
path from positive affect on behavioral intention was significant (t 5 3.8). A new inspection of the modification indexes
revealed that only the path from negative affect on behavioral
intention had a large (greater than five) modification index
(m.i. 5 11.38). Given the competing hypothesis of an independent affective route and the suggestion provided by the
modification index, a revised model was estimated where the
path from negative affect on behavioral intention was opened.
The second modified model also produced a significant
improvement in chi-square of 12.15 (d.f. 5 1). Seven of the
nine structural paths in the modified model were significant.
As in the hypothesized model, the effect from benevolence
on satisfaction and the effect from positive affect on satisfaction, were not significant. We finally estimated a revised model
where we closed the two non-significant paths. The fit of the
final modified model produced a satisfactory fit with a chisquare of 109.53 (d.f. 5 69 and p 5 0.0014), an RMSEA of
0.063, an AGFI of 0.87, and a NFI of 0.92. The estimated
path coefficients were all significant with t-values greater than
2.6. The estimated paths are reported in Figure 2.
Thus, both the effect of reliability and the effect of benevolence on behavioral intention are mediated by respectively
positive and negative affects. Only the effect of reliability on
behavioral intention is mediated by satisfaction. Admittedly
post-hoc, we may speculate that the mediating effects of satisfaction and affect from supplier behavior on behavioral inten-

tion operate in parallel through both a cognitive and an affective process. However, the finding is supported by Zajonc
(1980) who argues that thinking and feeling are two independent evaluation systems. For supplier reliability, the effect
on behavioral intention works both through satisfaction and
through negative affect. This implies, among others, that when
the customer dislikes the supplier (negative affect), behavioral
intention to be loyal is lower even if the customer is satisfied
with the supplier. For supplier benevolence, the effect works
only through positive affect. Thus, benevolence appears to
create liking for the supplier that next may create a kind of
bonding to the supplier. This latter bonding may thus operate
even when the customer is not satisfied with the supplier.
As can be seen in Figure 2, behavioral intention to be
loyal appears to be driven highly by affective states, whereas
satisfaction (as a cognitive evaluation) appears to have a fairly
modest effect on intentions of loyal behavior when we control
for affects. The effect from positive affect on behavioral intention is 0.50 and the effect from negative affect on behavioral
intention is 20.17, whereas the effect from satisfaction is
“only” 0.28. Thus, motivation to be loyal in a buyer–seller
relationship appears to be more influenced by affect (i.e.,
liking/disliking the supplier) than by satisfaction with the
supplier. The effect of reliability on behavioral intention is
0.446 [(0.92*0.28) 1 (20.57*20.16) 1 (20.57*20.17)]
and the effect of benevolence on behavioral intention is 0.475
(0.95*0.50). Thus, supplier benevolence and reliability appears to have about equal effect in developing buyer–seller
relationships.

Discussion
The reported findings support our argument that: (1) affective
responses are present in buyer–seller relationships; (2) affective responses differ according to whether they are related
to reliability or benevolence; and (3) affect and satisfaction
responses to supplier reliability and supplier benevolence have
strong effects on behavioral intention to be loyal to a supplier.
Supplier reliability has a strong effect on satisfaction and subsequently the buyer’s desire to continue the relationship and

266

J Busn Res
2000:49:259–271

inclination to talk favorably about the supplier. Lack of reliability creates negative emotions and negative affect from the
customer toward the supplier (i.e., dislike) and subsequently
reduces the motivation to be loyal. Supplier benevolence appears to invoke a different type of response than supplier
reliability, and we found that it influences behavioral intention
to be loyal only indirectly through positive affect. Supplier
benevolence thus appears to create a “liking” for the supplier
that promotes loyal behavior independent of the customers’
satisfaction with the supplier.
With reference to Herzberg’s two-factor motivation hygiene
theory (Herzberg, 1966) and Oliver’s (1997, pp. 151–153)
discussion of monovalent satisfiers and monovalent dissatisfiers, we propose that perceived benevolence is a “motivator”
with a potential to create positive affect, and that perceived
reliability is a “hygiene” with a potential to create negative
satisfaction and negative affect. For example, a buyer who
thinks a supplier should deliver a product on time will be
dissatisfied if the delivery is late. Delivery on time (as it should
be) is expected to have little or no effect on positive affect,
that is liking of the supplier. Showing benevolence is a psychological “extra” (Oliver 1997, p. 152) causing positive affect
only when fulfilled. For example, if the demand for the buyer’s
product suddenly increases and the additional parts needed
exceed the number set in their contract, the supplier could
charge a higher price for the extra parts. However, the supplier
may demonstrate flexibility and willingness to help, forgoing
a short-term profit. Showing benevolence is likely to cause
positive mental activity similar to gratitude and a sense of
friendship.
Our findings have several managerial implications. Both
supplier reliability and supplier benevolence is important,
influencing the customer’s motivation to behave loyally toward
the supplier. We suspect that most companies have emphasized development of reliability as indicated by the interest
in ISO certification processes. However, even in professional
business-to-business relationships the customers value benevolent behavior. The important finding is that the effect of
benevolence appears to be about equally strong as the effect
of reliability. Suppliers should therefore take care to manage
both reliability and benevolence in buyer–seller relationships.
Avoiding negative affective responses and dissatisfaction can
be achieved by carefully monitoring and developing the reliability attributes of the supplier. Reliability is for example improved through carefully managing procedures and systems
involved in order-delivery processes and customer service
operations. The most important is to fulfill promises, and to
avoid promising what cannot be delivered. An important task
of account managers is therefore to ensure delivery of what
is promised and orchestrate the daily interaction between the
two firms (Biong and Selnes, 1997). In addition, managers
should be aware that customers often make implicit assumptions about promises that may be equally important as explicit

F. Selnes and K. Gønhaug

promises. It follows that suppliers should try to understand
fully their customer’s expectations to achieve reliability.
Benevolence, on the other hand, requires a different managerial handling than reliability. As opportunities to show benevolence are tied to unpredictable situations in the ongoing
relationship, this kind of behavior must be secured through
other mechanisms than standard operating procedures. One
way to increase benevolence is to give frontline personnel
authority to provide the customer with flexible solutions. Another avenue may be to create a relationship-oriented culture
where benevolent behavior is desired and valued by the organization. The supplier need not incur expenses, but should
try to be flexible and creative in finding good solutions that
will benefit both parties. Most importantly is that the customers should feel that the suppliers care about their problems
and that they are motivated to help in a manner equivalent
to friendship.
Future research should address two issues. First, the external validity of the findings should be examined through replications in other industrial settings. The advantage by using
data from only one industry and customers of the same supplier, as in the present study, is that the potential confounding
effect of the industrial context is reduced. However, a stronger
test of the model would be to test the hypothesized model
across several industries and types of relationships. Second,
future research should address how the theoretical constructs
can be better measured. In particular, we need better measures
of affective states. The measures of affect employed in the
present study were borrowed from studies conducted in other
contexts, and in the future we need measures that better
capture the situation in buyer–seller relationships. An interesting approach to develop better affect measures would be to
conduct in-depth interviews with buyers and sellers and perform content analysis of their verbal protocols to capture how
they describe and understand their business partners.
The authors are grateful for the financial support from Telenor and valuable
assistance by Kjetil Aasdal and Tore Haraldstad Sandmoen. The authors thank
two anonymous reviewers for their thoughtful comments on earlier versions
of this article.

References
Anderson, Erin, and Weitz, Barton: Determinants of Continuity in
Conventional Industrial Channel Dyads. Marketing Science 8 (Fall
1989): 310–323.
Biong, Harald, and Selnes, Fred: The Strategic Role of the Salesperson
in Established Buyer-Seller Relationships. Journal of Business-toBusiness Marketing 3(3) 1997: 39–78.
Blodgett, Jeffery G., Granbois, Donald H., and Walters, Rockney G.:
The Effects of Perceived Justice on Complainants’ Negative Wordof-Mouth Behavior and Repatronage Intentions. Journal of Retailing
69(4) (1993): 399–429.
Bradburn, Norman M.: The Structure of Psychological Well-Being. Aldine, Chicago, IL. 1969.

Effects of Supplier Reliability and Benevolence

Campbell, Donald T.: The Informant in Quantitative Research. American Journal of Sociology 60 (1955): 339–342.
Churchill, Gilbert A., and Surprenant, Carol: An Investigation into
the Determinants of Customer Satisfaction. Journal of Marketing
Research 19 (November 1982): 491–504.
Deighton, John: The Consumption of Performance. Journal of Consumer Research 19 (December 1992): 362–373.
Dillon, William R., and Goldstein, Matthew: Multivariate Analysis—
Methods and Applications. John Wiley & Sons, Inc., New York.
1984.
Edell, Julie A., and Burke, Marian Chapman: The Power of Feelings
in Understanding Advertising Effects. Journal of Consumer Research
14 (December 1987): 421–433.
Evrard, Yves, and Aurier, Philippe: The Influence of Emotions on
Satisfaction with Movie Consumption. Journal of Consumer Satisfaction, Dissatisfaction and Complaining Behavior 7 (1994): 119–
125.
Folkes, Valerie S.: Recent Attribution Research in Consumer Behavior: A Review and New Directions. Journal of Consumer Research
14 (March 1988): 548–565.
Fornell, Claes: A National Customer Satisfaction Barometer: The
Swedish Experience. Journal of Marketing 56 (January 1992):
6–21.
Fornell, Claes, and Wernerfelt, Birger: Defensive Marketing Strategy
by Customer Complaint Management. Journal of Marketing Research 24 (November 1987): 337–346.
Gundlach, Gregory T., Achrol, Ravi S., and Mentzer, John T.: The
Structure of Commitment in Exchange. Journal of Marketing 59
(January 1995): 78–92.
Harvey, J.H., and Weary, G.: Current Issues in Attribution Theory
and Research. Annual Review of Psychology 35 (1984): 427–459.
Heide, Jan, and John, George: Alliances in Industrial Purchasing:
The Determinants of Joint Action in Buyer-Supplier Relationships.
Journal of Marketing Research 54 (February 1990): 24–36.
Herzberg, Fredrick: Work and Nature of Man, World Publishing,
Cleveland, OH. 1966.
Hirschman, Albert O.: Exit, Voice and Loyalty Responses to Declines
in Firms, Organizations and States. Harvard University Press, Cambridge, MA. 1970.
Izard, Carrol E.: Human Emotions, Plenum Press, New York. 1977.
Johanson, Jan, Halle´n, Lars, and Seyed-Mohamed, Nazeem: Interfirm
Adaptation in Business Relationships. Journal of Marketing 55
(April 1991): 29–37.
Johnston, Wesley J., and Bonoma, Thomas V.: The Buying Center:
Structure and Interaction Patterns. Journal of Marketing 45(3)
(1981): 143–156.
Kohli, Ajay: Determinants of Influence in Organizational Buying: A
Contingency Approach. Journal of Marketing 53(3) (1989): 50–65.
Kelly, Harold H., and Thibaut, John W.: Interpersonal Relations, Wiley, New York. 1978.

J Busn Res
2000:49:259–271

267

Kumar, Nirmalya, Scheer, Lisa K., and Steenkamp, Jan-Benedict
E.M.: The Effects of Supplier Fairness on Vulnerable Resellers.
Journal of Marketing Research 32 (February 1995): 54–65.
Lazarus, Richard S.: Cognition and Motivation in Emotion. American
Psychologist 46 (April 1991): 352–367.
Mano, Haim, and Oliver, Richard D.: Assessing the Dimensionality
and Structure of Consumption Experience: Evaluation, Feeling,
and Satisfaction. Journal of Consumer Research 20 (December
1993): 451–466.
Miller, D.T., and Ross, M.: Self-Serving Biases in Attribution of Causality: Fact or Fiction. Psychological Bulletin 82 (1975): 213–225.
Morgan, Robert M., and Hunt, Shelby D.: The Commitment-Trust
Theory of Relationship Marketing. Journal of Marketing 58 (July
1994): 20–38.
Oliver, Richard L.: Theoretical Bases of Consumer Satisfaction Research: Review, Critique, and Future Directions. Theoretical Developments in Marketing, in Charles W. Lamb, Jr., and Patrick M.
Dunne eds., American Marketing Association, Chicago, IL. 1980,
pp. 206–210.
Oliver, Richard L.: Cognitive, Affective, and Attribute Bases of the
Satisfaction Response. Journal of Consumer Research 20 (December
1993): 418–430.
Oliver, Richard L.: Satisfaction, A Behavioral Perspective on the Consumer, McGraw-Hill, New York. 1997.
Oliver, Richard L., and DeSarbo, Wayne S.: Response Determinants
in Satisfaction Judgments. Journal of Consumer Research 14 (March
1988): 495–507.
Ravald, A., and Gro¨nroos, Christian: The Value Concept and Relationship Marketing. European Journal of Marketing 30(2) (1996):
19–30.
Reichheld, Fredrick F.: Learning from Customer Defections. Harvard
Business Review (March–April 1996): 56–69.
Richins, Marsha L.: Negative Word-of-Mouth by Dissatisfied Consumers: A Pilot Study. Journal of Marketing 47 (Winter (1983):
68–78.
Russell, James A.: A Circumplex Model of Affect. Journal of Personality
and Social Psychology 39(6) (1980): 1161–1178.
Singh, Jagdep: Consumer Complaint Intentions and Behavior: Definitional and Taxonomical Issues. Journal of Marketing 52 January
(1988): 93–107.
Westbrook, Robert A.: Product/Consumption-based Affective Responses and Postpurchase Processes. Journal of Marketing Research
24 (August 1987): 258–270.
Westbrook, Robert A., and Oliver, Richard L.: The Dimensionality
of Consumption Emotion Patterns and Consumer Satisfaction.
Journal of Consumer Research 18 (June 1991): 84–91.
Watson, David, Clark, Lee Ann, and Tellegen, Auke: Development
and Validation of Brief Measures of Positive and Negative Affect:
The PANAS Scales. Journal of Personality and Social Psychology 54
(June 1988): 1063–1070.
Zajonc, Robert B.: Feeling and Thinking: Preferences Need No Inferences. American Psychologist 35 (February 1980): 151–175.

268

J Busn Res
2000:49:259–271

F. Selnes and K. Gønhaug

Appendix 1. Statistical Metrics of Items
Item

Mean

S.D.

Skewness

Kurtosis

SHARE
EXIT
POSWO1
POSWO2
SATISFA1
SATISFA2
SATISFA3
POSAFF1
POSAFF2
POSAFF3
POSAFF4
POSAFF5
NEGAFF1
NEGAFF2
NEGAFF3
NEGAFF4
NEGAFF5
RELIAB1
RELIAB2
RELIAB3
RELIAB4
RELIAB5
BENEVOL1
BENEVOL2
BENEVOL3
BENEVOL4
BENEVOL5

3.29
2.34
3.87
3.93
4.28
3.83
2.92
3.24
3.14
3.31
4.23
4.53
1.96
0.83
0.87
2.40
0.93
3.93
3.89
4.36
3.86
4.67
4.34
3.24
4.36
3.70
3.96

1.32
1.49
1.31
1.24
1.14
1.11
0.82
1.18
1.25
1.20
1.09
1.11
1.87
1.66
1.71
1.88
1.80
1.31
1.31
1.27
1.07
1.27
1.26
1.43
1.57
1.24
1.31

0.22
1.09
20.38
20.25
20.07
20.15
0.23
0.05
0.51
20.01
0.05
0.15
0.40
1.03
0.92
0.02
0.85
0.20
0.15
20.12
20.28
20.36
20.06
0.59
0.16
0.21
0.18

0.94
0.39
20.13
0.03
0.15
0.25
1.85
20.02
0.18
20.09
0.04
20.12
20.71
0.01
20.44
21.09
20.63
20.18
20.32
20.28
0.13
0.22
20.02
20.14
20.80
0.10
20.21

Effects of Supplier Reliability and Benevolence

J Busn Res
2000:49:259–271

269

Appendix 2. Correlation Matrix of All 30 Items

SHARE
EXIT
POSWO1
POSWO2
SATISFA1
SATISFA2
SATISFA3
POSAFF1
POSAFF2
POSAFF3
POSAFF4
POSAFF5
NEGAFF1
NEGAFF2
NEGAFF3
NEGAFF4
NEGAFF5
RELIAB1
RELIAB2
RELIAB3
RELIAB4
RELIAB5
BENEVOL1
BENEVOL2
BENEVOL3
BENEVOL4
BENEVOL5

SHARE

EXIT

POSWO1

POSWO2

SATISFA1

SATISFA2

SATISFA3

1.00
0.18
0.34
0.42
0.22
0.35
0.33
0.32
0.32
0.29
0.18
0.27
20.33
0.01
20.20
20.42
20.23
0.24
0.25
0.28
0.35
0.26
0.25
0.28
0.29
0.34
0.28

1.00
0.63
0.65
0.48
0.63
0.45
0.33
0.38
0.34
0.34
0.40
20.44
20.25
20.40
20.53
20.38
0.29
0.35
0.42
0.46
0.51
0.57
0.25
0.47
0.44
0.48

1.00
0.89
0.67
0.73
0.55
0.60
0.65
0.57
0.54
0.61
20.58
20.31
20.49
20.58
20.53
0.50
0.47
0.55
0.50
0.53
0.57
0.39
0.65
0.56
0.64

1.00
0.63
0.73
0.55
0.58
0.66
0.57
0.48
0.60
20.60
20.30
20.49
20.59
20.51
0.50
0.47
0.51
0.56
0.62
0.66
0.44
0.70
0.59
0.66

1.00
0.75
0.55
0.44
0.52
0.41
0.38
0.54
20.49
20.10
20.44
20.55
20.41
0.36
0.43
0.60
0.53
0.64
0.53
0.44
0.54
0.51
0.58

1.00
0.63
0.51
0.56
0.49
0.38
0.57
20.57
20.27
20.42
20.62
20.42
0.44
0.38
0.56
0.61
0.64
0.60
0.42
0.61
0.52
0.60

1.00
0.42
0.46
0.32
0.29
0.47
20.52
20.21
20.43
20.52
20.35
0.38
0.43
0.50
0.48
0.47
0.46
0.26
0.47
0.38
0.43
(continued)

270

SHARE
EXIT
POSWO1
POSWO2
SATISFA1
SATISFA2
SATISFA3
POSAFF1
POSAFF2
POSAFF3
POSAFF4
POSAFF5
NEGAFF1
NEGAFF2
NEGAFF3
NEGAFF4
NEGAFF5
RELIAB1
RELIAB2
RELIAB3
RELIAB4
RELIAB5
BENEVOL1
BENEVOL2
BENEVOL3
BENEVOL4
BENEVOL5

POSAFF1

POSAFF2

POSAFF3

POSAFF4

POSAFF5

NEGAFF1

NEGAFF2

NEGAFF3

NEGAFF4

NEGAFF5

1.00
0.87
0.85
0.52
0.47
20.40
20.08
20.21
20.41
20.29
0.41
0.41
0.35
0.47
0.42
0.49
0.47
0.57
0.54
0.56

1.00
0.80
0.50
0.49
20.37
20.05
20.23
20.45
20.30
0.43
0.52
0.43
0.50
0.51
0.49
0.50
0.60
0.59
0.63

1.00
0.61
0.53
20.38
20.06
20.22
20.38
20.22
0.37
0.38
0.32
0.41
0.46
0.55
0.58
0.61
0.60
0.65

1.00
0.74
20.36
20.11
20.22
20.33
20.31
0.35
0.36
0.27
0.43
0.42
0.56
0.39
0.52
0.48
0.56

1.00
20.35
20.19
20.35
20.36
20.27
0.35
0.42
0.41
0.55
0.51
0.59
0.49
0.59
0.57
0.64

1.00
0.50
0.70
0.73
0.61
20.31
20.41
20.47
20.45
20.44
20.44
20.35
20.46
20.32
20.44

1.00
0.74
0.51
0.69
0.00
20.17
20.10
20.24
20.12
20.19
0.05
20.20
20.08
20.14

1.00
0.69
0.79
20.13
20.38
20.31
20.31
20.32
20.30
20.05
20.20
20.23
20.25

1.00
0.78
20.38
20.38
20.43
20.43
20.55
20.45
20.30
20.43
20.50
20.46

1.00
20.31
20.39
20.37
20.38
20.41
20.33
20.12
20.26
20.30
20.23

F. Selnes and K. Gønhaug

(continued)

J Busn Res
2000:49:259–271

Appendix 2. continued

RELIAB1
SHARE
EXIT
POSWO1
POSWO2
SATISFA1
SATISFA2
SATISFA3
POSAFF1
POSAFF2
POSAFF3
POSAFF4
POSAFF5
NEGAFF1
NEGAFF2
NEGAFF3
NEGAFF4
NEGAFF5
RELIAB1
RELIAB2
RELIAB3
RELIAB4
RELIAB5
BENEVOL1
BENEVOL2
BENEVOL3
BENEVOL4
BENEVOL5

1.00
0.44
0.40
0.37
0.39
0.40
0.36
0.48
0.43
0.47

RELIAB2

1.00
0.53
0.56
0.41
0.39
0.24
0.41
0.29
0.41

RELIAB3

1.00
0.51
0.48
0.38
0.20
0.48
0.41
0.44

RELIAB4

1.00
0.55
0.55
0.39
0.60
0.46
0.57

RELIAB5

1.00
0.57
0.48
0.59
0.61
0.69

BENEVOL1

1.00
0.54
0.62
0.56
0.69

BENEVOL2

1.00
0.64
0.72
0.73

BENEVOL3

1.00
0.73
0.82

BENEVOL4

1.00
0.83

BENEVOL5

Effects of Supplier Reliability and Benevolence

Appendix 2. continued

1.00

J Busn Res
2000:49:259–271

271