Journal of the Academy of Marketing Science

, Volume 36, Issue 2, pp 191–201 | Cite as

A meta-analytic review of opportunism in exchange relationships

Original Empirical Research

Abstract

The potential to engage in opportunism is a central theme in institutional economics, yet prior research has not quantitatively reviewed the role of opportunism in marketing research. This study uses meta-analytic techniques to synthesize research on opportunism conducted over the last quarter century. The analysis of 183 effect sizes extracted from 54 publications from the period 1982 to 2005 offers some support to extant channel theory. The research also indicates that the informant’s frame of reference and the research design significantly influence the observed effects. Implications of the findings and future research directions are discussed.

Keywords

Institutional Economics Opportunism Meta-analysis 

Assumptions made about the nature of human behavior influence insights drawn from investigations of exchange relationships. Although a traditional economic perspective maintains that agents are self-interest seeking (Simon 1978), transaction cost economics assumes that agents are given to act opportunistically. Opportunistic agents offer incomplete or biased information designed to mislead, obfuscate, or otherwise confuse trading partners (Williamson 1985). Their action includes active or passive attempts to violate written or social contracts governing exchange (Wathne and Heide 2000).

The assertion that agents act opportunistically has prompted substantial research into channel interaction. Marketing research (e.g., John 1984) maintains that opportunism mitigates relationships between organizational properties and outcomes, yet these studies often produce inconsistent findings. Channel theory would benefit from identification of the nomological network of opportunism as well as the examination of the generalizability of relationships between opportunism and constructs to which it is most frequently related. Prior research has reviewed opportunism (e.g., Wathne and Heide 2000), yet no study has employed quantitative methods to integrate findings across this research. Meta-analysis provides the opportunity to combine results from multiple studies, explain inconsistencies in findings, and examine moderators of theoretical relationships (Rosenthal and DiMatteo 2001). Despite the potential gains, research to date has not employed meta-analytic techniques to examine opportunism.

The purpose of this study is to provide a meta-analytic review of opportunism. This study combines multiple theoretical approaches to build an integrative model of opportunism in marketing research. The review identifies and quantifies relationships between opportunism and other theory-based constructs. The results implicate under-researched variables and identify issues studied so extensively that more research is unnecessary. The review also identifies methodological factors that influence empirical results and presents avenues for empirical and theoretical contributions to research.

The manuscript is organized as follows: We initially outline the theoretical and methodological correlates of opportunism, and we subsequently present the research method and results. We conclude with a discussion of the implications of our findings.

Correlates of opportunism

Theoretical correlates

The research framework outlined by Heide (1994) serves as the theoretical basis for the study. This framework (see Fig. 1) includes institutional economics (Williamson 1985), resource dependence theory (Pfeffer and Salancik 1978), the behavioral research perspective (Stern 1969), and relational contracting theory (Macneil 1978).
Fig. 1

Conceptual framework for meta-analysis.

Institutional economics is a theoretical perspective that examines decision-making by business organizations (Cyert and March 1963). Two closely related theoretical orientations in institutional economics are transaction cost economics (Williamson 1985) and agency theory (Fama 1980). Transaction cost economics examines the deployment of governance structures to support economic transactions (Heide 1994). Agency theory similarly focuses on determining efficient contracts given characteristics of trading partners and environmental constraints (Bergen et al. 1992). Transaction cost economics implicates transaction specific investments and uncertainty as factors that lead to alternative governance structures. Transaction specific investments (TSIs) are specialized assets that have limited value outside their intended use (Williamson 1975). TSIs create an incentive to maintain the relationship as the investing party cannot leave the relationship without incurring economic loss. The ability of the non-investing party to use TSIs as hostages makes opportunistic behavior irrational for the investing party (Bradach and Eccles 1989). The non-investing party, however, can expropriate the value of the specific assets. Thus, TSIs should be related negatively to opportunism by the investing party and related positively to opportunism by the non-investing party.

Agency theory and transaction cost economics implicate two forms of uncertainty as relevant to governance decisions. Environmental uncertainty refers to unanticipated changes in circumstances associated with an exchange (Noordewier et al. 1990). Bounded rationality limits the ability to plan for contingencies a priori, resulting in adaptation problems in light of unforeseen circumstances. Behavioral uncertainty arises from difficulties associated with assessing performance and contractual compliance of exchange partners (Rindfleisch and Heide 1997). Agency theory recognizes that information asymmetries associated with behavioral uncertainty increase the likelihood that agents will shirk responsibilities (Bergen et al. 1992). Transaction cost economics similarly acknowledges that the inability to ascertain an exchange partner’s actions enhances the likelihood of opportunism (Heide 1994). Thus, behavioral and environmental uncertainty should be related positively to opportunism.

Transaction cost economics suggests opportunism will arise whenever it is feasible and profitable, and recognizes that individuals engage in this behavior to influence value creation and wealth distribution (Ghosh and John 1999). Wathne and Heide (2000) suggest that opportunism initially increases outcomes for the opportunistic party, but this behavior restricts value creation, increases costs, and decreases revenues for both parties in an exchange relationship. Consequently, we anticipate a negative association between opportunism and performance.

Resource dependence theory (Pfeffer and Salancik 1978) augments analysis of uncertainty with the investigation of dependency. An entity’s dependence is determined by its motivational investment in goals mediated by its exchange partner and the availability to achieve those goals outside the relationship (Emerson 1962). Under balanced dependence, both parties have much to lose if conflict erodes the relationship (Kumar et al. 1995a). To avoid economic losses and to maintain relational benefits, both parties should refrain from behaviors that promote conflict, including opportunistic behaviors (Lusch and Brown 1996). Asymmetric dependence, in contrast, is expected to engender opportunism. A dependence advantage enables one to expropriate resources from the more dependent party, resulting in a positive relationship between partner’s dependence and opportunism. The more dependent party, however, should refrain from opportunistic behavior to ensure relationship continuity.

The behavioral research perspective examines control structures deployed to yield desirable outcomes, secure investments, and minimize opportunism (Stern 1969). Channel research into control structures has focused on centralization, formalization, coordination, and surveillance. Centralization is the concentration of decision making authority, and formalization refers to the use of explicit procedures to govern a relationship (John and Reve 1982). Centralization attempts to overcome opportunism by limiting behavioral discretion, whereas formalization removes transaction difficulties and constrains opportunism (Dwyer and Oh 1987). Coordination refers to the purposive organization of activities, resources, and information flows between exchange parties (Reve and Stern 1986). Coordination limits opportunism by curtailing adaptation problems and establishing congruent goals (Buvik and John 2000). Surveillance, or monitoring of a partner’s action, limits opportunism by reducing information asymmetry between partners (Wathne and Heide 2000). By limiting self-control and autonomy, these structures may exacerbate, rather than attenuate, opportunism (John 1984).

Relational contracting theory recognizes that the context of a relationship influences the manner in which transactions occur (Macneil 1980). Opportunism research in this vein includes analyses of norms, communication, and satisfaction. Norms are safeguards that specify limits on behavior and constrain opportunism. Norms shift the focus from an individual’s outcomes to joint benefits at the relationship level and thus should curtail opportunism (Rokkan et al. 2003). Defined as the sharing of meaningful, timely information (Anderson and Narus 1984), communication between exchange partners decreases information asymmetry. Due to its ability to decrease information asymmetry and promote goal congruence, communication should be associated negatively with opportunism. Satisfaction refers a positive affective state drawn from appraisal of a firm’s relationship with another firm (Geyskens et al. 1999). Opportunism impairs social interactions between parties and thereby should limit satisfaction.

Methodological moderators

Methodological factors that may influence the theoretical relationships include frame of reference, organizational context, research strategy, functional orientation, and number of industries. Frame of reference is a moderating variable designed to account for the perspective of the informant.1Partner-based opportunism refers to informant perceptions of the degree to which a trading partner engages in opportunism. For example, Rokkan et al. (2003, p. 215) ask informants to assess the accuracy of statements such as “On occasion, this supplier lies about certain things in order to protect their interests.” Self-reported opportunism refers to informant reports of their own opportunistic behavior. For example, John’s (1984, p. 288) informants indicate their level of agreement with statements such as “Sometimes, I have to alter the facts slightly in order to get what I need.” Self-reports are subject to social desirability bias (Jap and Anderson 2003). Given the nature of opportunism, these reports should yield lower levels of opportunism and weaker effect sizes in the nomological network.

Institutional economics and organizational sociology recognize that organizational context influences the theoretical relationships associated with opportunism. In contrast to market transactions, interaction among employees within the same organization operates within a shared institutional context that influences the meaning attributed to variables within the nomological network (Moran and Ghoshal 1996). Intrafirm relationships evince higher levels of specific assets, behavioral uncertainty, and environmental volatility (Anderson 1985). In addition, administration of intrafirm relationships benefits from heightened levels of bureaucratic controls and relational governance mechanisms that limit opportunism (Williamson 1991).

Research strategies vary in their ability to achieve generalizability, precision in measurement and control, and realism. Field studies offer heightened realism but sacrifice precision in measurement and control, whereas experiments exhibit less realism but have greater precision in measurement and control (McGrath 1982). The greater precision of experiments should yield larger effect sizes than those found in field studies (Sawyer and Ball 1981).

Functional orientation distinguishes between relationships initiated to procure or to provide product offerings to channel partners. Buyers and sellers are prone to opportunism, yet it is plausible for persons on one side of the dyad to evince higher levels of opportunism. Finally, the number of industries considers the breadth of the empirical domain. Research employing multiple industries is more generalizable than single industry studies, but should result in smaller observed correlations than research employing one industry (Geyskens et al. 1998).

Research method

Literature search

Electronic and manual searches identified studies measuring opportunism. These searches reviewed the Academy of Management Journal, Administrative Science Quarterly, European Journal of Marketing, Industrial Marketing Management, International Journal of Research in Marketing, Journal of International Marketing, Journal of Management, Journal of Marketing, Journal of Marketing Research, Journal of Retailing, Journal of the Academy of Marketing Science, Journal of the Market Research Society, Marketing Letters, Marketing Science, Psychology and Marketing, and Strategic Management Journal from 1975 to 2005. In addition, we searched ABI/Global Inform, Business Source Premier, and Dissertation Abstracts International. We subsequently contacted authors of opportunism research for working papers measuring opportunism. Finally, we used an ancestry approach in which references of studies identified in the preceding searches and key conceptual articles were searched. The search yielded 54 publications measuring opportunism (see Table 1). Two studies analyzed the same sample and six studies collected data from multiple samples, resulting in 59 independent samples.2
Table 1

Descriptive statistics

 

Overall

Self-reported

Partner-based

Frame of reference

   

 Self-reported

28a

 Partner-based

28

Organizational context

   

 Intraorganizational

10

6

4

 Interorganizational

49

21

25c

Research strategy

   

 Field Study

49

20

26c

 Experiment

10

7

3

Functional orientation

   

 Buyer

28

13

13c

 Supplier

18b

8

9c

Number of industries

   

 Single

25

14

8c

 Multiple

24b

6

18

Sample range

57–531

64–395

57–531

Total sample size

11,566

4,978

6,216

aRemaining studies used a mixture of self- and partner-based items.

bRemaining studies did not specify or moderating variable was not applicable.

cStudies employing a mixture of self-reported and partner-based items were omitted, resulting in a smaller number of studies when the data are separated by self-reported and partner-based opportunism.

Procedure

The effect size is Pearson’s product moment correlation, r. Similar to Geyskens et al. (1998), we categorize correlations based on the operationalization of constructs. When studies use multiple measures of a construct, the mean r is used in the analysis. In all, 183 effects sizes are analyzed via the procedures outlined by Rosenthal (1991). Multiple data coders would have facilitated inter-rater reliability analysis, but resource constraints prevented us from engaging in this effort.

Results

As a preface to the analysis of the theoretical correlates, we examined whether methodological factors influence the observed correlations. We employed generalized least squares regression to test the moderators because it minimizes undue influence from studies reporting multiple correlations (Raudenbush et al. 1988). The results (see Table 2) indicate that all the moderating variables influence the theoretical framework significantly. Intrafirm studies exhibit smaller effect sizes than interfirm studies (β = −0.121, z = −6.14), and field studies evince smaller effects than experiments (β = −0.140, z = −4.24). Buyer samples yield stronger effect sizes than supplier samples (β = 0.258, z = 13.67), and single industry studies reveal larger effects than multi-industry studies (β = 0.196, z = 11.88).
Table 2

GLS moderator results

 

Opportunism

Self-reported opportunism

Partner-based opportunism

Moderator

Beta

z value

Beta

z value

Beta

z value

Frame of reference

0.213

14.97*

 Partner-based—1

      

 Self-reported—0

      

Organizational context

−0.121

−6.14*

−0.081

−1.68*

−0.157

−5.82*

 Intraorganizational—1

      

 Interorganizational—0

      

Research strategy

−0.140

−4.24*

0.246

4.07*

−0.376

−4.20*

 Field Study—1

      

 Experiment—0

      

Functional orientation

0.258

13.67*

0.390

10.17*

0.115

3.63*

 Buyer—1

      

 Supplier—0

      

Number of industries

0.196

11.88*

0.337

6.05*

0.362

15.38*

 Single—1

      

 Multiple—0

      

*p < 0.05

In addition, the informant’s frame of reference significantly influences the theoretical correlates. Partner-based reports evince larger effect sizes than self-reports of opportunism (β = 0.213, z = 14.97). Given the perceptual and statistical contrasts associated with frame of reference, the remaining analyses distinguish between partner-based and self-reported opportunism.

Partner-based opportunism

Institutional economics

The results outlined in Tables 3 and 4 offer some support to institutional economics. Own TSIs are associated positively with partner-based opportunism (r = 0.13, p < 0.05), yet partner TSIs are associated negatively with the partner-based opportunism (r = −0.11, p < 0.05). In contrast to multi-industry studies (r = 0.03), single industry studies (r = 0.25) evince larger effect sizes for own TSIs (z = 2.91, p < 0.05). Similarly, environmental uncertainty is associated positively with partner-based opportunism (r = 0.06, p < 0.05), but the effect is greater (z = 3.05, p < 0.05) in single industry studies (r = 0.25) than multi-industry studies (r = 0.04). Although the overall of effect of behavioral uncertainty is not significant (r = −0.01, p > 0.05), organizational context moderates these results (z = 3.05, p < 0.05). Behavioral uncertainty is associated negatively with partner-based opportunism in intrafirm settings (r = −0.12) but positively with opportunism in interfirm settings (r = 0.10). Performance is associated negatively with partner-based opportunism (r = −0.32, p < 0.05), but this effect is more pronounced (z = −2.97, p < 0.05) in single industry studies (r = −0.46) than multi-industry studies (r = −0.30).
Table 3

Partner-based opportunism

Product moment correlations, confidence intervals, and chi-square tests of heterogeneity

 

ka

Total N

Mean r

Weighted mean r

Lower 95% CIb

Upper 95% CIb

Chi-squarec

Fail-safe statistic

Own TSIs

7

1,003

0.09

0.13*

0.07

0.19

24.30*

82.95

Partner TSIs

5

652

−0.14

−0.11*

−0.19

−0.03

18.81*

50.52

Environmental uncertainty

9

1,495

0.08

0.06*

0.01

0.11

58.43*

42.08

Behavioral uncertainty

5

882

0.05

−0.01

−0.08

0.05

17.47*

1.04

Performance

12

2,142

−0.36

−0.32*

−0.37

−0.29

147.53*

341.06

Own dependence

8

1,253

0.00

0.04

−0.01

0.10

16.83*

24.74

Partner dependence

4

868

0.03

0.06*

0.00

0.13

2.33

7.53

Formalization

11

1,690

−0.11

−0.11*

−0.15

−0.06

76.04*

104.97

Centralization

4

593

0.12

0.16*

0.08

0.24

6.09

59.44

Coordination

4

917

−0.43

−0.46*

−0.56

−0.43

15.90*

178.99

Surveillance

2

673

0.13

0.12*

0.04

0.19

13.17*

21.22

Norms

6

1,063

−0.44

−0.46*

−0.55

−0.43

18.53*

268.17

Communication

4

473

−0.26

−0.34*

−0.44

−0.26

40.93*

131.69

Satisfaction

2

520

−0.55

−0.46*

−0.59

−0.41

17.76*

90.40

*p < 0.05

aNumber of studies

bConfidence intervals for weighted mean r

cChi-square test of heterogeneity

Table 4

Partner-based opportunism

Univariate moderator resultsa

 

Organizational context

Research strategy

Functional orientation

Number of industries

 

Intra- vs. inter-organizational

Field study vs. experiment

Buyer vs. supplier sample

Single vs. multiple industry sample

Own TSIs

b

0.18 vs. −0.04

0.25 vs. 0.03*

Partner TSIs

−0.13 vs. −0.15

Environmental uncertainty

0.13 vs. 0.00

0.25 vs. 0.04*

Behavioral uncertainty

−0.12 vs. 0.10*

Performance

−0.41 vs. −0.31

−0.46 vs. −0.30*

Own dependence

0.02 vs. −0.15*

0.05 vs. 0.04

−0.15 vs. 0.08*

Formalization

−0.19 vs. −0.09*

−0.16 vs. −0.08

−0.35 vs. −0.03*

Coordination

−0.41 vs. −0.48

−0.41 vs. −0.46

Surveillance

Norms

−0.52 vs. −0.43*

−0.41 vs. −0.60*

−0.36 vs. −0.48

−0.43 vs. −0.41

Communication

−0.28 vs. −0.17

Satisfaction

−0.38 vs. −0.69*

−0.38 vs. −0.69*

−0.69 vs. −0.38*

*p < 0.05

aPartner dependence and centralization did not exhibit significant heterogeneity to warrant inclusion.

bInsufficient variation in studies for the moderating variable

Resource dependence theory

Contrary to this theory, the data indicate that partner dependence is associated positively with partner-based opportunism (r = 0.06, p < 0.05). Own dependence is not related to partner-based opportunism (r = 0.04, p > 0.05), but the research strategy (z = −1.73, p < 0.05) and number of industries sampled (z = −3.17, p < 0.05) moderate this relationship. Field studies evince a positive effect between own dependence (r = 0.02) and partner-based opportunism, but the effect is negative (r = −0.15) for experiments. By contrast, single industry studies yield a negative effect between own dependence and partner-based opportunism (r = −0.15), yet the effect is positive in multi-industry studies (r = 0.08).

Behavioral research perspective

The results offer mixed support for the behavioral research perspective. Formalization (r = −0.11, p < 0.05) and coordination (r = −0.46, p < 0.05) are related negatively to partner-based opportunism. Organizational context (z = 2.07, p < 0.05) and the number of industries (z = −4.78, p < 0.05) mitigate the influence of formalization given that intrafirm effects (r = −0.19) exceed interfirm effects (r = −0.09) and single industry studies (r = −0.35) exceed multi-industry studies (r = −0.03). Centralization (r = 0.16, p < 0.05) and surveillance (r = 0.12, p < 0.05) are associated positively with partner-based opportunism.

Relational contracting theory

This theory is supported by the observation that norms (r = −0.46, p < 0.05), communication (r = −0.34, p < 0.05), and satisfaction (r = −0.46, p < 0.05) are related negatively to partner-based opportunism. Organizational context (z = 1.78, p < 0.05) and research strategy (z = −2.33, p < 0.05) influence the role of norms given that intrafirm studies (r = −0.52) yield stronger effects than interfirm studies (r = −0.43) and experiments (r = −0.60) yield stronger effects than field studies (r = −0.41). Organizational context (z = −4.21, p < 0.05), research strategy (z = −4.21, p < 0.05), and number of industries sampled (z = −4.21, p < 0.05) moderate the relationship between satisfaction and partner-based opportunism, with interfirm studies, experimental studies and single industry studies evincing larger effect sizes.

Self-reported opportunism

Institutional economics

The results outlined in Tables 5 and 6 offer limited support to institutional economics. Own specific investments are unrelated to self-reported opportunism (r = 0.04, p > 0.05), but the number of industries sampled (z = −3.45, p < 0.05) moderates this relationship. Multi-industry samples evince positive effect sizes between self-reported opportunism and own TSIs (r = 0.27), yet the effect is negative for single industry samples (r = −0.01). Partner TSIs are also unrelated to self-reported opportunism (r = 0.01, p > 0.05), yet the observed correlation is mitigated by the organizational context (z = 9.16, p < 0.05). Intrafirm studies yield negative associations between partner TSIs and self-reported opportunism (r = −0.36), whereas interfirm studies yield positive associations between the constructs (r = 0.33). Number of industries sampled also influences the observed effect (z = 2.84, p < 0.05), given that single industry studies evince stronger associations (r = 0.26) with opportunism than multi-industry studies (r = 0.02). Environmental uncertainty is also unrelated to self-reported opportunism (r = 0.08, p > 0.05), but this effect is moderated by the organizational context (z = 2.92, p < 0.05). Self-reported opportunism is related positively to environmental uncertainty in interorganizational studies (r = 0.32), but it is not related to environmental uncertainty in intraorganizational studies (r = 0.00). Performance is related negatively to self-reported opportunism (r = −0.09, p < 0.05), but this effect is moderated by the organizational context (z = 2.49, p < 0.05). Performance is related negatively to self-reported opportunism in intrafirm settings (r = −0.15) and positively to opportunism in interfirm settings (r = 0.12).
Table 5

Self-reported opportunism

Product moment correlations, confidence intervals, and chi-square tests of heterogeneity

 

ka

Total N

Mean r

Weighted Mean r

Lower 95% CIb

Upper 95% CIb

Chi-squarec

Fail-safe statistic

Own TSIs

4

998

0.06

0.04

−0.03

0.10

24.23*

10.39

Partner TSIs

3

710

0.10

0.01

−0.07

0.08

92.38*

−0.49

Environmental Uncertainty

2

434

0.16

0.08

−0.02

0.17

8.55*

13.77

Behavioral Uncertainty

d

       

Performance

4

541

−0.09

−0.09*

−0.18

−0.01

12.45*

32.93

Own Dependence

7

1,287

−0.02

−0.05*

−0.11

0.00

35.09*

29.06

Partner Dependence

d

       

Formalization

10

1,683

−0.06

−0.03

−0.07

0.02

86.14*

15.02

Centralization

6

1,182

0.24

0.22*

0.16

0.28

33.21*

123.38

Coordination

7

1351

−0.23

−0.23*

−0.28

−0.18

13.98*

150.73

Surveillance

6

1202

0.08

0.08*

0.02

0.13

30.41*

39.07

Norms

8

1557

−0.17

−0.17*

−0.23

−0.12

46.42*

130.61

Communication

d

       

Satisfaction

5

885

−0.37

−0.34*

−0.42

−0.28

99.12*

162.88

*p < 0.05

aNumber of studies

bConfidence intervals for weighted mean r

cChi-square test of heterogeneity

dBehavioral uncertainty, partner dependence and communication do not have a sufficient number of studies to warrant inclusion.

Table 6

Self-reported opportunism

Univariate moderator resultsa

 

Organizational context

Research strategy

Functional orientation

Number of industries

 

Intra- vs. inter-organizational

Field study vs. experiment

Buyer vs. supplier sample

Single vs. multiple industry sample

Own TSIs

b

0.05 vs. 0.16

−0.01 vs. 0.27*

Partner TSIs

−0.36 vs. 0.33*

0.40 vs. 0.26

0.26 vs. 0.02*

Environmental uncertainty

0.00 vs. 0.32*

Performance

−0.15 vs. 0.12*

−0.18 vs. 0.12*

Own dependence

−0.03 vs. 0.00

−0.01 vs. −0.05

−0.20 vs. 0.04*

Formalization

−0.05 vs. −0.06

−0.25 vs. 0.13*

−0.05 vs. 0.11*

Centralization

0.03 vs. 0.28*

0.39 vs. 0.21*

0.29 vs. 0.14*

Coordination

−0.25 vs. −0.22

−0.28 vs. −0.17*

−0.26 vs. −0.03*

Surveillance

−0.02 vs. 0.18*

0.15 vs. 0.01*

0.10 vs. 0.15

Norms

−0.17 vs. −0.18

−0.14 vs. −0.16

−0.30 vs. −0.03*

Satisfaction

−0.04 vs. −0.44*

−0.52 vs. −0.10*

−0.40 vs. −0.16*

*p < 0.05

aBehavioral uncertainty, partner dependence and communication do not have a sufficient number of studies to warrant inclusion.

bInsufficient variation in studies for the moderating variable

Resource dependence theory

Consistent with theoretical predictions, own dependence is associated negatively with self-reported opportunism (r = −0.05, p < 0.05). These results are tempered by the number of industries sampled (z = −3.71, p < 0.05). Single industry studies exhibit a negative effect size (r = −0.20), whereas multi-industry studies exhibit a positive effect (r = 0.04) with opportunism.

Behavioral research perspective

The results offer some support to the behavioral research perspective, but they are influenced by methodological factors. Although formalization is not related to self-reported opportunism (r = −0.03, p > 0.05), functional orientation (z = 5.69, p < 0.05) and number of industries sampled (z = −2.19, p < 0.05) moderate these results. Formalization is related negatively to self-reported opportunism among buyers (r = −0.25) and related positively to opportunism among suppliers (r = 0.13). Formal procedures are associated negatively with self-reported opportunism in single industry studies (r = −0.05) but associated positively with opportunism in multi-industry studies (r = 0.11). Coordination is associated negatively with self-reported opportunism (r = −0.23, p < 0.05), but these results are also moderated by functional orientation (z = 1.81, p < 0.05) and number of industries sampled (z = −2.78, p < 0.05). Buyer samples (r = −0.28) reveal stronger effect sizes than supplier samples (r = −0.17), and single industry studies (r = −0.26) yield larger effect sizes than multi-industry studies (r = −0.03).

Centralization is associated positively with self-reported opportunism (r = 0.22, p < 0.05), but these results are influenced by organizational context (z = 3.88, p < 0.05), functional orientation (z = −2.65, p < 0.05), and the number of industries sampled (z = 2.37, p < 0.05). Interfirm studies (r = 0.28) of centralization produce stronger effect sizes than intrafirm studies (r = 0.03), whereas buyer samples (r = 0.39) yield stronger effects than supplier samples (r = 0.21). In addition, single industry studies (r = 0.29) evince greater effect sizes than multi-industry studies (r = 0.14). Surveillance is associated positively with self-reported opportunism (r = 0.08, p < 0.05), but organizational context (z = 3.42, p < 0.05) and research design (z = −2.45, p < 0.05) influence these correlations. Surveillance is associated positively with opportunism in interfirm settings (r = 0.18) but negatively in intrafirm contexts (r = −0.02). Field studies (r = 0.15) of surveillance yield stronger effects sizes than experimental studies (r = 0.01).

Relational contracting theory

The results offer support for relational contracting theory as norms (r = −0.17, p < 0.05) and satisfaction (r = −0.34, p < 0.05) are associated negatively with self-reported opportunism. Number of industries mitigates the influence of norms (z = −4.46, p < 0.05) given that single industry studies evince greater effect sizes (r = −0.30) than multi-industry studies (r = −0.03). The relationship between satisfaction and self-reported opportunism is moderated by organizational context (z = −4.76, p < 0.05), functional orientation (z = 6.67, p < 0.05), and number of industries (z = −3.30, p < 0.05). Interfirm studies (r = −0.44) yield stronger effect sizes for satisfaction than intrafirm studies (r = −0.04). Buyer samples (r = −0.52) reveal stronger effect sizes than supplier samples (r = −0.10), and single industry studies evince stronger effect sizes (r = −0.40) than multi-industry studies (r = −0.16).

Discussion and conclusion

The goal of this study has been to provide a quantitative review of opportunism research. The meta-analysis uncovers several consistencies across empirical efforts. Analyses of variables drawn from institutional economics indicate consistent positive associations of own specific investments and environmental uncertainty with partner-based opportunism, yet negative associations of partner-based opportunism with partner specific investments and performance. Analyses of variables drawn from the behavioral research perspective indicate consistent positive associations of partner-based opportunism with centralization and surveillance, but negative associations of partner-based opportunism with formalization and coordination. Similarly, self-reported opportunism is associated positively with centralization and associated negatively with coordination. Research employing a relational contracting perspective indicates a consistent negative association of partner-based opportunism with norms, communication, and satisfaction. Self-reported opportunism yields similar results as it is associated negatively with norms and satisfaction.

The meta-analysis also uncovers some inconsistencies that warrant additional theoretical considerations and empirical efforts. The results indicate that organizational context influences observed correlations with variables drawn from institutional economics. Organizational context influences the relationship between partner-based opportunism and behavioral uncertainty, and it influences associations of self-reported opportunism with partner specific investments, environmental uncertainty, and performance. The research strategy influences the resource dependence results given that the relationship between partner-based opportunism and own dependence varies with the research design. The functional orientation influences the behavioral research analyses as the association of formalization with self-reported opportunism varies with the role responsibilities (i.e., buyer versus seller) of the respondent. The number of industries sampled influences results associated with institutional economics, resource dependence, and the behavioral research perspective. The association of self-reported opportunism with own specific investments and formalization varies based on the number of industries in the analysis. In addition, the number of industries sampled influences associations of own dependence with partner and self-reported opportunism.

The review also reveals some relationships that have not been empirically assessed. Surprisingly, empirical research has not examined the relationship between behavioral uncertainty and self-reported opportunism. This appears to be a glaring omission as the association between behavioral uncertainty and opportunism is a central tenet to institutional economics (Rindfleisch and Heide 1997). Two additional oversights in extant empirical research are the relationship between partner dependence and self-reported opportunism as well as the relationship between communication and self-reported opportunism. Given their potential to contribute to channel theory and practice, these relationships warrant inclusion in future opportunism studies.

Theoretical implications

Institutional economics

This research underscores the importance of examining partner-based opportunism as it is related to specific investments, uncertainty, and performance. Specific investments are deployed to generate value (Ghosh and John 1999), yet our results indicate that their deployment increases the likelihood of opportunism. Environmental uncertainty further exposes the firm to opportunism that limits performance. The implication is to pursue research into mechanisms that facilitate value generation while simultaneously reducing exposure to opportunism. These mechanisms may include bilateral specific investments (Anderson and Weitz 1992), pre-contractual screening of agents (Bergen et al. 1992), and incentive alignment (Tosi et al. 1997).

In contrast to partner-based opportunism, self-reported opportunism is not related to specific investments or uncertainty. Self-reports may suffer from social desirability bias. Alternatively, institutional economics may be limited in its focus on controlling an exchange partner’s behavior. Williamson (1985) contends that firms can safeguard themselves from opportunistic exploitation by implementing greater control over their partner’s behavior through hierarchical governance. Similarly, agency theory focuses on bringing about the best possible returns for the principal (Bergen et al.1992). Future research should examine the extent to which governance structures increase the potential for opportunism across the dyad.

Research examining self-reported opportunistic behavior may also benefit from the logic of organizational justice research (Adams 1965; Greenberg 1990). This research suggests the investing party will require a greater portion of the outcomes to achieve an equitable state. An equitable state will elicit satisfaction and positive behavioral responses (Konovsky and Cropanzano 1991; Lind 2001) and inhibit opportunism (Adams 1965). In contrast, an inequitable state will elicit high levels of distress, and firms will be motivated to engage in opportunism to restore the relationship to its formerly equitable position. Supporting this contention, Greenberg (1990) found that perceived inequity promotes employee theft. It should be noted, however, that outcomes may have a direct effect on opportunism aside from relationship inputs or investments (Kumar et al. 1995b). A party is likely to refrain from opportunism when the level of outcomes is high in comparison to alternative options (even in light of perceived inequity) to sustain the future stream of income.

More recent research (e.g., Skarlicki and Folger 1997) underscores the importance of procedural and interactional justice (i.e., fair procedures and fair interpersonal treatment, respectively) in constraining opportunism. Skarlicki and Folger (1997) report a negative relationship between procedural and interactional justice and retaliatory behaviors (e.g., intentionally working slower). More importantly, they found that the relationship between distributive justice and retaliatory behaviors was significant only under low levels of procedural and interactional justice. Hence, implementing fair procedures and treating partners fairly may be an effective governance mechanism in inequitable relationships. Our results and these complementary studies demonstrate the need to examine perceived justice as it may influence opportunism in exchange relationships.

Resource dependence theory

Our findings offer some support to resource dependence theory as own dependence is related negatively to self-reported opportunism. By contrast, partner dependence is related positively to partner-based opportunism. This finding does not correspond to the theory, as a dependent partner should be less likely to engage in opportunistic behaviors (Lusch and Brown 1996). This research does not, however, examine interdependence asymmetry or total interdependence in the exchange relationship. Examining one side of the dyad paints an incomplete picture; it is the level of interdependence that influences each party’s motivations, behaviors, and perceptions (Gundlach and Cadotte 1994; Kumar et al. 1995a). Moreover, it is important to disentangle the effects of interdependence asymmetry and total interdependence as these variables have differential effects on trust, commitment, and conflict (Kumar et al. 1995a), and thus are likely to have differential effects on opportunism. Consequently, future studies should augment dependency analyses via investigation of dyadic interdependence.

Behavioral research perspective

The results are mixed for bureaucratic structures. Formalization is related negatively to partner-based opportunism, and coordination is related negatively to both partner-based and self-reported opportunism. Contrary to this theory, centralization and surveillance are related positively to both forms of opportunism. Firms employ centralized decision-making structures to secure economic returns from exchange relationships, and they use surveillance to ensure appropriate implementation of operating procedures (Reve and Stern 1986). Research should examine whether these objectives can be achieved while simultaneously controlling opportunism.

Relational contracting theory

This research supports relational contracting as a governance mechanism. Norms attenuate opportunism across a variety of settings, measures, and samples; and opportunism is related negatively to satisfaction. Relationships, however, develop over time and the efficacy of norms may vary over the relationship life cycle. Norms emerge in the early stages of the relationship life cycle (Dwyer et al. 1987), yet close relationships provide an opportunity for covert activities systematically designed to cheat a partner (Anderson and Jap 2005). Consequently, norms may become a less effective mechanism for governing behavior later in the relationship life cycle. Research is needed that longitudinally tracks the ability of relational contracting properties to quell opportunism in exchange relationships.

Methodological implications

Our findings underscore the need to consider the informant’s frame of reference when designing empirical studies. Frame of reference is a central issue that separates opportunism research into two different constructs. Partner-based studies ask informants to reflect on their perceptions of the degree to which a trading partner engages in opportunism. Self-reports, by contrast, reflect the informants’ willingness to reveal the extent to which they engage in miscreant behavior. Researchers designing future studies should recognize that self-reports are susceptible to social desirability bias and ordinarily evince weaker effect sizes. Hence, social desirability bias should be controlled through routines that detect and prevent its occurrence (Nederhof 1985) and via deployment of divergent data sources within a study (Jo 2000).

The review of other methodological factors highlights the influence of research design on empirical results. Interorganizational settings generally evince higher levels of opportunism (partner-based and self-reported) than intraorganizational settings. The research strategy also moderates the results given that experiments reveal larger effect sizes for partner-based opportunism and smaller effect sizes for self-reported opportunism. Interestingly, buyer and supplier results do not vary for partner-based opportunism. For self-reported opportunism, however, buyers yield stronger effect sizes than suppliers for the constructs associated with the behavioral research perspective. Finally, single industry studies generally produce stronger effect sizes than multi-industry studies in each theoretical domain.

Limitations and conclusion

Two limitations influence the results of this study. First, the Rosenthal (1991) approach does not adjust for statistical artifacts. Alternative methods (e.g., Hunter and Schmidt 1990) accommodate these artifacts, but they exacerbate Type I error rates when the number of correlations analyzed is small (Spector and Levine 1987). Hence, we chose to use Rosenthal’s (1991) approach. Second, similar to other meta-analytic reviews (e.g., Geyskens et al. 1999), this review grouped similar constructs into one overarching construct. The relationship between opportunism and its correlates may be moderated by the latter’s operationalization. Future research should examine the potential moderating effects associated with the operationalization of these constructs.

The purpose of our study has been to summarize opportunism research and identify avenues for future research. Our findings offer modest support to institutional economics, resource dependence theory, the behavioral research perspective, and relational contracting theory. Furthermore, the results indicate that the frame of reference and research design moderate the theoretical relationships. We hope these findings offer renewed insight into the analysis of opportunism in marketing research.

Footnotes

  1. 1.

    We thank a reviewer for this suggestion.

  2. 2.

    The technical appendices and a bibliography of studies included in the analysis are available from the first author.

Notes

Acknowledgement

The authors thank the editor, David W. Stewart, and the three anonymous JAMS reviewers for their insightful critiques of this manuscript. In addition, the authors thank Leslie Vincent for her valuable comments on an earlier version of this manuscript. Both authors contributed equally to this study. The Von Allmen Center for Electronic Commerce at the University of Kentucky provided partial funding for this research.

References

  1. Adams, J. S. (1965). Inequity in social exchange. In L. Berkowitz (Ed.) Advances in experimental and social psychology. New York, NY: Academic.Google Scholar
  2. Anderson, E. (1985). The salesperson as outside agent or employee: A transaction cost analysis. Marketing Science, 4(3), 234–254.CrossRefGoogle Scholar
  3. Anderson, E., & Jap, S. D. (2005). The dark side of close relationships. MIT Sloan Management Review, 46(3), 75–82.Google Scholar
  4. Anderson, E., & Weitz, B. (1992). The use of pledges to build and sustain commitment in distribution channels. Journal of Marketing Research, 29, 18–34, (February).CrossRefGoogle Scholar
  5. Anderson, J. C., & Narus, J. A. (1984). A model of the distributor’s perspective of distributor–manufacturer working relationships. Journal of Marketing, 48(4), 62–74.CrossRefGoogle Scholar
  6. Bergen, M., Dutta, S., & Walker Jr., O. C. (1992). Agency relationships in marketing: A review of the implications and applications of agency and related theories. Journal of Marketing, 56(3), 1–24.CrossRefGoogle Scholar
  7. Bradach, J. L., & Eccles, R. G. (1989). Price, authority and trust: From ideal types to plural forms. Annual Review of Sociology, 15, 97–118, (August).CrossRefGoogle Scholar
  8. Buvik, A., & John, G. (2000). When does vertical coordination improve industrial purchasing relationships? Journal of Marketing, 64(4), 52–64.CrossRefGoogle Scholar
  9. Cyert, R. M., & March, J. G. (1963). A behavioral theory of the firm. Englewood Cliffs, NJ: Prentice-Hall.Google Scholar
  10. Dwyer, F. R., & Oh, S. (1987). Output sector munificence effects on the internal political economy of marketing channels. Journal of Marketing Research, 29(4), 347–58.CrossRefGoogle Scholar
  11. Dwyer, F. R., Schurr, P. H., & Oh, S. (1987). Developing buyer–seller relationships. Journal of Marketing, 51(2), 11–27.CrossRefGoogle Scholar
  12. Emerson, R. M. (1962). Power-dependence relations. American Sociological Review, 27, 31–41.CrossRefGoogle Scholar
  13. Fama, E. F. (1980). Agency problems and the theory of the firm. Journal of Political Economy, 88(2), 288–307.CrossRefGoogle Scholar
  14. Geyskens, I., Steenkamp, J.-B. E. M., & Kumar, N. (1998). Generalizations about trust in marketing channel relationships using meta-analysis. International Journal of Research in Marketing, 15, 223–48.CrossRefGoogle Scholar
  15. Geyskens, I., Steenkamp, J.-B. E. M., & Kumar, N. (1999). A meta-analysis of satisfaction in marketing channel relationships. Journal of Marketing Research, 36(2), 223–38.CrossRefGoogle Scholar
  16. Ghosh, M., & John, G. (1999). Governance value analysis and marketing strategy. Journal of Marketing, 63(Special Issue), 131–145.CrossRefGoogle Scholar
  17. Greenberg, J. (1990). Organizational justice: Yesterday, today, and tomorrow. Journal of Applied Business Research, 16(2), 399–432.Google Scholar
  18. Gundlach, G. T., & Cadotte, E. R. (1994). Exchange interdependence and interfirm interaction: Research in a simulated channel setting. Journal of Marketing Research, 31(4), 516–532.CrossRefGoogle Scholar
  19. Heide, J. B. (1994). Interorganizational governance in marketing channels. Journal of Marketing, 58(1), 71–85.CrossRefGoogle Scholar
  20. Hunter, J. E., & Schmidt, F. L. (1990). Methods of meta-analysis: Correcting error and bias in research findings. Newbury Park, CA: Sage.Google Scholar
  21. Jap, S. D., & Anderson, E. (2003). Safeguarding interorganizational performance and continuity under ex post opportunism. Management Science, 49(12), 1684–701.CrossRefGoogle Scholar
  22. Jo, M.-S. (2000). Controlling social-desirability bias via method factors of direct and indirect questioning in structural equation models. Psychology and Marketing, 17(2), 137–148.CrossRefGoogle Scholar
  23. John, G. (1984). An empirical investigation of some antecedents of opportunism in a marketing channel. Journal of Marketing, 21(3), 278–89.CrossRefGoogle Scholar
  24. John, G., & Reve, T. (1982). The reliability and validity of key informant data from dyadic relationships in marketing channels. Journal of Marketing Research, 19(4), 517–24.CrossRefGoogle Scholar
  25. Konovsky, M. A., & Cropanzano, R. (1991). Perceived fairness of employee drug testing as a predictor of employee attitudes and job performance. Journal of Applied Psychology, 76(5), 698–707.CrossRefGoogle Scholar
  26. Kumar, N., Scheer, L. K., & Steenkamp, J.-B. E. M. (1995a). The effects of perceived interdependence on dealer attitudes. Journal of Marketing Research, 32, 348–356, (February).CrossRefGoogle Scholar
  27. Kumar, N., Scheer, L. K., & Steenkamp, J.-B. E. M. (1995b). The effects of supplier fairness on vulnerable resellers. Journal of Marketing Research, 32, 54–65.CrossRefGoogle Scholar
  28. Lind, E. A. (2001). Fairness heuristic theory: Justice judgments as pivotal cognitions in organizational relations. In J. Greenberg, & R. Cropanzano (Eds.) Advances in organizational justice. Stanford, CA: Stanford University Press.Google Scholar
  29. Lusch, R. F., & Brown, J. R. (1996). Interdependency, contracting, and relational behavior in marketing channels. Journal of Marketing, 60(4), 19–38.CrossRefGoogle Scholar
  30. Macneil, I. R. (1978). Contracts: Adjustments of long-term economic relations under classical, neoclassical and relational contract law. Northwestern University Law Review, 72(6), 854–905.Google Scholar
  31. Macneil, I. R. (1980). The new social contract: An inquiry into modern contractual relations. New Haven, CT: Yale University Press.Google Scholar
  32. McGrath, J. E. (1982). Dilemmatics: The study of research choices and dilemmas. In J. Martin, J. E. McGrath, & R. A. Kulka (Eds.) Judgment calls in research. Newbury Park, CA: Sage.Google Scholar
  33. Moran, P., & Ghoshal, S. (1996). Theories of economic organization: The case for realism and balance. Academy of Management Review, 21(1), 58–72.CrossRefGoogle Scholar
  34. Nederhof, A. (1985). Methods of coping with social desirability bias: A review. European Journal of Social Psychology, 15, 263–280.CrossRefGoogle Scholar
  35. Noordewier, T. G., John, G., & Nevin, J. R. (1990). Performance outcomes of purchasing arrangements in industrial buyer–vendor relationships. Journal of Marketing, 54(4), 80–93.CrossRefGoogle Scholar
  36. Pfeffer, J., & Salancik, G. (1978). The external control of organizations: A resource dependence perspective. New York: Harper and Row.Google Scholar
  37. Raudenbush, S. W., Becker, B. J., & Kalaian, H. (1988). Modeling multivariate effect sizes. Psychological Bulletin, 103(1), 111–120.CrossRefGoogle Scholar
  38. Reve, T., & Stern, L. W. (1986). The relationship between interorganizational form, transaction climate, and economic performance in vertical interfirm dyads. In L. Pellegrini, & S. K. Reddy (Eds.) Marketing Channels. Lexington, MA: Heath.Google Scholar
  39. Rindfleisch, A., & Heide, J. B. (1997). Transaction cost analysis: Past, present, and future applications. Journal of Marketing, 61(4), 30–54.CrossRefGoogle Scholar
  40. Rokkan, A. I., Heide, J. B., & Wathne, K. H. (2003). Specific investments in marketing relationships: Expropriation and bonding effects. Journal of Marketing Research, 40(2), 210–24.CrossRefGoogle Scholar
  41. Rosenthal, R. (1991). Meta-analytic procedures for social research. Newbury Park, CA: Sage.Google Scholar
  42. Rosenthal, R., & DiMatteo, M. R. (2001). Meta-analysis: Recent developments in quantitative methods for literature reviews. Annual Review of Psychology, 52, 59–82.CrossRefGoogle Scholar
  43. Sawyer, A. G., & Ball, A. D. (1981). Statistical power and effect size in marketing research. Journal of Marketing Research, 18(3), 275–90.CrossRefGoogle Scholar
  44. Simon, H. A. (1978). Rationality as process and as product of a thought. American Economic Review, 68(2), 1–16.Google Scholar
  45. Skarlicki, D. P., & Folger, R. (1997). Retaliation in the workplace: The roles of distributive, procedural and interactional justice. Journal of Applied Psychology, 82(3), 434–443.CrossRefGoogle Scholar
  46. Spector, P. E., & Levine, E. L. (1987). Meta-analysis for integrating study outcomes: A Monte Carlo study of its susceptibility to type I and type II errors. Journal of Applied Psychology, 72(1), 3–9.CrossRefGoogle Scholar
  47. Stern, L. W. (1969). In L. Stern (Ed.) Distribution channels: Behavioral dimensions. New York: Houghton-Mifflin.Google Scholar
  48. Tosi, H. L., Katz, J. P., & Gomez-Mejia, L. R. (1997). Disaggregating the agency contract: The effects of monitoring, incentive alignment, and term in office on agent decision making. Academy of Management Journal, 40(3), 584–602.CrossRefGoogle Scholar
  49. Wathne, K. H., & Heide, J. B. (2000). Opportunism in interfirm relationships: Forms, outcomes, and solutions. Journal of Marketing, 64(4), 36–51.CrossRefGoogle Scholar
  50. Williamson, O. E. (1975). Markets and hierarchies, analysis and antitrust implications: A study in the economies of internal organization. New York: Free Press.Google Scholar
  51. Williamson, O. E. (1985). The economic institutions of capitalism: Firms, markets, relational contracting. New York: Free Press.Google Scholar
  52. Williamson, O. E. (1991). Comparative economic organization: The analysis of discrete structural alternatives. Administrative Science Quarterly, 36(2), 269–96.CrossRefGoogle Scholar

Copyright information

© Academy of Marketing Science 2007

Authors and Affiliations

  1. 1.College of Business and EconomicsWest Virginia UniversityMorgantownUSA
  2. 2.Carol Martin Gatton College of Business and EconomicsUniversity of KentuckyLexingtonUSA
  3. 3.Norwegian School of ManagementCentre for Advanced Research in Retailing0442 OsloNorway

Personalised recommendations