JAN B. HEIDE, KENNETH H. WATHNE, and AKSEL I. ROKKAN* This article examines the effects of monitoring on interfirm relationships. Whereas some research suggests that monitoring can serve as a control mechanism that reduces exchange partner opportunism, there is also evidence showing that monitoring can actually promote such behavior. The authors propose that the actual effect of monitoring depends on (1) the form of monitoring used (output versus behavior) and (2) the context in which monitoring takes place. With regard to the form of monitoring, the results from a longitudinal field study of buyer–supplier relationships show that output monitoring decreases partner opportunism, as transaction cost and agency theory predict, whereas behavior monitoring, which is a more obtrusive form of control, increases partner opportunism. With regard to the context, the authors find that informal relationship elements in the form of microlevel social contracts serve as buffers that both enhance the effects of output monitoring and permit behavior monitoring to suppress opportunism in the first place.
Interfirm Monitoring, Social Contracts, and Relationship Outcomes Monitoring programs are integral parts of many firms’ relationship strategies and are used to ensure that the value created through a firm’s marketing decisions can be claimed by the focal firm (Ghosh and John 1999). The recent availability of information technology that facilitates the acquisi-
tion of performance data has been predicted to motivate firms to increase their monitoring efforts in relation to their exchange partners (e.g., Jacobides and Croson 2001). Importantly, however, a general recommendation to increase monitoring activity may be premature because of conflicting theoretical perspectives and empirical evidence regarding its actual effects. On the one hand, transaction cost and related theories (Jensen and Meckling 1976; Williamson 1985) suggest that monitoring serves as a control mechanism that should suppress partner opportunism. On the other hand, other bodies of literature suggest an opposite scenario—namely, that monitoring may promote opportunism because of “reactance” types of effects (e.g., Barkema 1995; Deci, Koestner, and Ryan 1999; John 1984). Although previous research has suggested that monitoring can both suppress and promote opportunism, these discrepant views have not been reconciled. Indeed, prominent theorists (e.g., Jensen 1998, p. 98) have expressed concern about “a lack of a theory of monitoring.” Our current goal is to reconcile these divergent perspectives by considering (1) the effects of different forms of monitoring and (2) the context in which monitoring takes place. Regarding the first point, we examine whether monitoring a partner on the basis of output versus behavior has different consequences. Regarding the second point, we consider the possibility that monitoring, by its very nature, constitutes a formal governance device and, as such, may possess inherent limitations. Although formal devices
*Jan B. Heide is Irwin Maier Chair in Marketing, School of Business, University of Wisconsin–Madison, and Professorial Fellow, Department of Management and Marketing, University of Melbourne (e-mail: jheide@ bus.wisc.edu). (Part of this article was completed while Heide was Montezemolo Visiting Professor, Judge Business School, University of Cambridge.) Kenneth H. Wathne is Assistant Professor of Marketing, School of Business, University of Wisconsin–Madison (e-mail:
[email protected]. edu). Aksel I. Rokkan is a research fellow, Institute for Research in Economics and Business Administration, Norwegian School of Economics and Business Administration (e-mail:
[email protected]). (This article was completed while Rokkan was a research fellow, Bodø Graduate School of Business, Bodø University College.) The authors acknowledge the assistance of Sissel Adolfsen with data collection and thank the Marketing Science Institute and Bodø Graduate School of Business for financial assistance. They also thank Erin Anderson; Neeraj Arora; George John; Natalie Mizik; Sriram Venkataraman; seminar participants at Judge Business School at the University of Cambridge, Pennsylvania State University, University of Arizona, University of Melbourne, University of Michigan, City University of Hong Kong, Emory University, University of Texas, University of Western Ontario, Erasmus University, the Norwegian School of Management, and the Marketing Science Institute’s Young Scholars Program; and the anonymous JMR reviewers for their helpful comments. The authors are listed in random order. To read and contribute to reader and author dialogue on JMR, visit http://www.marketingpower.com/jmrblog.
© 2007, American Marketing Association ISSN: 0022-2437 (print), 1547-7193 (electronic)
425
Journal of Marketing Research Vol. XLIV (August 2007), 425–433
426 have received considerable attention, especially in economics, research in marketing has pointed to both (1) the potential dysfunctional consequences of formal relationship features (e.g., John 1984) and (2) the inherent importance of informal ones (e.g., Heide and John 1992; Kumar, Scheer, and Steenkamp 1995). Our key argument is that certain relationship outcomes (i.e., opportunism) depend not only on the deployment of individual formal governance mechanisms (e.g., monitoring), as is often assumed, but also on constellations of formal and informal governance processes. Specifically, we draw on the notion of a microlevel social contract (Dunfee, Smith, and Ross 1999; Locke 1690; Rousseau 1762) and posit that such informal agreements serve as buffers against potential reactance effects from monitoring and thus permit monitoring to reduce opportunism. We organized this article as follows: First, we present our conceptual framework and research hypotheses. Second, we describe the research method used to test the hypotheses and the empirical results. Finally, we discuss the implications of the findings and the study’s limitations and provide possible topics for further research. THE NATURE AND EFFECTS OF MONITORING The monitoring construct appears prominently in several different streams of literature, including marketing (Anderson and Oliver 1987), economics (Alchian and Demsetz 1972), and organization theory (Ouchi 1979). In general, monitoring is defined as an effort made by one party to measure or “meter” the performance of another. Most researchers recognize monitoring as a generic organizational process that can be built into relationships between independent firms, but the monitoring phenomenon remains poorly documented. Although a body of analytical (e.g., Lal 1990) and empirical (e.g., Stump and Heide 1996) literature exists on the antecedents of monitoring, research on its actual effects is still scarce. Moreover, the empirical evidence that exists shows an inconsistent pattern. The General Effects of Monitoring Conceptually, monitoring offers control by reducing information asymmetry between exchange parties (Eisenhardt 1985). For example, in a buyer–supplier relationship, monitoring enhances a buyer’s ability to detect supplier opportunism in the form of quality shirking. An underlying assumption in transaction cost and related theories is that the greater one party’s investment in monitoring, the lower is the other party’s opportunism (Alchian and Demsetz 1972; Jensen and Meckling 1976). Notably, however, some empirical evidence suggests an opposite causal sequence (e.g., Aiken and Hage 1966; Barkema 1995; Deci, Koestner, and Ryan 1999). For example, Murry and Heide (1998) find that manufacturer efforts to verify compliance with in-store promotional programs actually undermined retailer cooperation. A common explanation for such counterintuitive findings is that monitoring represents an obtrusive form of control that may offend another party’s sense of autonomy and cause reactance (Brehm 1966; Perrow 1986). For example, Deci and Ryan (1987) suggest that actors have inherent preferences for engaging in behaviors that are self-determined. Thus, mechanisms such as monitoring, whose goal is to regulate
JOURNAL OF MARKETING RESEARCH, AUGUST 2007 behavior externally, may influence a party’s attribution of causality and ultimately the motivation to perform. As a whole, the existing literature provides some different perspectives on the way monitoring affects relationship outcomes. We suggest that these differences can be explained on the basis of (1) the particular form of monitoring used and (2) the context in which monitoring takes place. Forms of Monitoring The extant literature often draws a distinction between output and behavior monitoring (Anderson and Oliver 1987). Output monitoring involves measuring the visible consequences of a partner’s actions, such as a supplier’s delivery time, order accuracy, and product quality. In contrast, behavior monitoring involves evaluating the processes that are expected to produce the focal outcomes. Notably, although this conceptual distinction is widely accepted, empirical evidence regarding each strategy’s effects is scarce. Indeed, Narayandas and Rangan (2004, p. 75) note that “a worthwhile area for further research is to identify whether and when performance evaluation based on outcomes or actions is more critical to the development of a relationship.” Consider first the effects of output monitoring. Under such a system, an exchange partner is left alone to choose the means by which to reach the relevant outcomes (Anderson and Oliver 1987). In an organizational purchasing context, a buyer may engage in extensive incoming inspection of a supplier’s components, but the monitoring process does not directly involve the focal supplier. As a consequence, there is little room for intervention that undermines the supplier’s self-control, and the likelihood of reactance effects is low. Thus, we expect the opportunism-reducing effect of monitoring described in transaction cost and agency theory to materialize. Formally, H1: A firm’s output monitoring decreases a partner’s opportunism.
In contrast with output monitoring, whose focus is on actual consequences, behavior monitoring is directed at a partner’s actions (Anderson and Oliver 1987). For example, a buyer may ascertain a supplier’s behavior through onsite inspections of manufacturing and quality-control procedures. Because behavior monitoring imposes strict guidelines on which activities are to be performed and how they should be performed, all else being equal, this form of monitoring risks being perceived as intrusive. Such a perception is particularly likely when the monitoring takes place in relationships between financially independent firms. Ultimately, behavior monitoring in such a situation may produce defensive attitudes and behaviors that are characteristically opportunistic in nature (e.g., Aiken and Hage 1966; Barkema 1995). Formally, H2: A firm’s behavior monitoring increases a partner’s opportunism.
A few additional comments on H1 and H2 are in order. First, the general premise of H1 is that because output monitoring does not directly regulate a partner’s actions, it is capable of reducing opportunism. At the same time, monitoring in general has been described as an inherently obtru-
Interfirm Monitoring, Social Contracts, and Relationship Outcomes sive management approach (e.g., Perrow 1986). As such, we expect that though output monitoring is capable of suppressing overt opportunistic tendencies, it leaves a residual reactance effect. We return to this issue in subsequent sections. Second, we view H1 and H2 as baseline hypotheses that express the general effects of each monitoring strategy, all else being equal. In the next section, we attempt to add precision to these predictions by considering conditions that may (1) make output monitoring more effective in suppressing opportunism and (2) change the actual nature of the effect of behavior monitoring (i.e., allow it to reduce partner opportunism). The Monitoring Context: Microlevel Social Contracts Early transaction cost literature (e.g., Williamson 1975) discussed monitoring in the context of a larger syndrome of hierarchical governance processes, whose expected aggregate effect (i.e., under complete vertical integration) was to suppress opportunism. Unfortunately, the early literature failed to articulate the specific nature of these processes clearly. However, subsequent theorizing within transaction cost theory’s “measurement branch” has begun to consider the possibility that hierarchies de facto rely on combinations of monitoring devices and informal agreements of various kinds. For example, Ouchi (1979) argues strongly that an underlying agreement between exchange parties constitutes a crucial cognitive foundation for ongoing monitoring efforts. The agreement construct has a long intellectual history, dating back to the early work on social contracts by Locke (1690) and Rousseau (1762). The focus in this work was on societal or macrolevel contracts, or “agreements among men” (Rousseau 1762), which ensured legitimate exercise of power and ultimately the protection of people from “anarchic invasion” (Locke 1690). Social contracts were viewed as being imposed on people at the societal level, and they served to regulate behavior even if the specific parties in question did not participate in their initial development. In Rousseau’s (1995, p. 14) terminology, these types of social contracts were “inherited at birth.” Subsequent work has extended the notion of a social contract to microlevel agreements that are purposely designed within individual exchange relationships (Dunfee, Smith, and Ross 1999; John 1984). For example, buyers and suppliers establish agreements about specific output and behavior standards as a foundation for their ongoing interactions. Such an agreement is not necessarily a legally binding document (though it may build on one) but rather one party’s belief about “salient relationship issues” in the interaction with another exchange partner (John 1984, p. 280).1 Despite the difference in level of analysis, microlevel social contracts retain the original notions of legitimacy and protection. In the context of monitoring, an existing contract not only grants a party (e.g., a buyer) monitoring rights within a specified range (Ouchi 1979) but also safeguards the exchange partner (e.g., a supplier) from random inter1A social contract differs from a relational contract (Macneil 1980), whose defining characteristic is not an agreement per se but rather particular types of exchange norms. It is also distinct from a psychological contract (Rousseau 1995), which typically refers to individual-level beliefs about organizational outcomes and obligations.
427
vention (Weber 1947). Halaby (1986) describes the underlying process in terms of so-called authority costs: If one party’s mode of governance is at odds with another party’s legitimacy beliefs, the latter party’s perceived authority costs will increase, and the overall value of the relationship will decrease. That is, if a buyer begins to evaluate a supplier on a set of quality standards that the parties did not agree on, it may impose a cost increase on the supplier, for example, in the form of investments in new quality-control systems. The focal supplier may reject the new standards entirely, or if the buyer’s monitoring continues, the supplier may attempt to restore the relationship’s equilibrium by engaging in actions that impose new costs on the focal buyer.2 However, if the party that is being monitored perceives the governance practices as conforming to existing beliefs, as established in the social contract, the party will assign lower authority costs to the relationship and, in general, be more willing to cooperate. This is consistent with the work of Simon (1951) and Williamson (1975), both of whom recognize that expression of authority in a relationship is an object of value (i.e., a cost factor) that is separate from the economic value of the “transaction” itself (e.g., work for pay). The notion of legitimacy from organization theory is related to the concept of procedural fairness from social psychology (e.g., Lind and Tyler 1988). In marketing, Kumar, Scheer, and Steenkamp (1995) apply this concept to manufacturer–reseller relationships and define it in terms of a reseller’s perception of a focal supplier’s procedures and processes. In our context, procedural fairness resides in the relationship between a supplier’s monitoring effort and the prevailing social contract. For example, monitoring may be perceived as fair to the extent that it is executed against the backdrop of an established agreement. The upshot of this discussion is that the actual effect of monitoring is influenced by the content of the relationship between the focal parties. Given the presence of a microlevel social contract, the general effect of monitoring on a party’s behavior should be a reduction in opportunism, consistent with the original transaction cost framework (Williamson 1975). In the absence of such a contract, however, monitoring will be considered an invasive form of control, which may produce retaliatory opportunistic actions. In technical terms, we propose that a microlevel social contract moderates the effect of monitoring on partner behavior. Importantly, however, we expect that the specific nature of the moderating effect differs for output and behavior monitoring because of the different baseline effects discussed previously.3 First, the main premise of H2 is that behavior monitoring causes a reactance effect and promotes opportunism (e.g., 2Importantly, although all the referenced research converges on the possibility of “reactance” behaviors, there are different perspectives on the process by which reactance comes about. For example, social psychological research (e.g., Brehm 1966) tends to focus on “pure” or immediate reactance. In contrast, extant organizational research (e.g., Halaby 1986) focuses on how monitoring affects a party’s perception of relationship value. Under the latter scenario, monitoring may induce reactance, but behaviors follow from an explicit calculation of relationship costs and benefits. 3Technically, H and H involve conditional effects (of monitoring on 1 2 opportunism).
428 Deci, Koestner, and Ryan 1999). However, if a sufficiently strong microlevel social contract exists, we expect that behavior monitoring will reduce opportunism, consistent with transaction cost theory (Williamson 1975). In the latter case, a microlevel social contract changes the direction of the monitoring effect. Formally, H3: A firm’s behavior monitoring decreases a partner’s opportunism, given the presence of a microlevel social contract (for behavior) between the focal firms.
Second, the logic behind H1 is that output monitoring decreases opportunism because of its relatively unobtrusive nature. We expect that the presence of a microlevel social contract will strengthen this effect. Recall from our previous discussion that though output monitoring is capable of controlling overt opportunistic actions, it may leave the partner with residual negative sentiments. However, we expect that the presence of a strong social contract will serve as a buffer against such sentiments and consequently enhance the ability of output monitoring to reduce opportunism, in accordance with transaction cost and agency theory. Formally, H4: The negative effect of a firm’s output monitoring on partner opportunism is strengthened in the presence of a microlevel social contract (for output) between the focal firms.
RESEARCH METHOD Empirical Context and Data Collection The empirical context for this study is business-tobusiness relationships between manufacturers (suppliers) of building materials (e.g., doors, windows, frames, stairs, roofing products) and their downstream buyers. Our focus is on a particular supplier’s reaction, in the form of opportunism, to various forms of buyer monitoring. We collected data through a longitudinal field study design that measured buyer monitoring and microlevel social contracts at Time 1 and supplier opportunism at Time 2. We used a three-year lag for the buyer’s monitoring efforts to manifest themselves fully at the supplier level in the form of identifiable (and potentially opportunistic) actions. The incorporation of a temporal lag into the research design provides at least two benefits. First, the temporal separation in measurement between the independent and the dependent variables can reduce a potential method bias that stems from measuring both sets of variables at the same time within the same “medium” (survey instrument). Podsakoff and colleagues (2003) refer to these as “measurement context effects.” Second, as Cohen and colleagues (2003) describe, our design places us in a unique position to examine systematically the relationship between posited ex ante governance variables (monitoring, microlevel social contract) and ex post relationship outcomes (opportunism). The sampling frame was a national mailing list containing names of managers of independent suppliers. We drew a random sample of 1300 names from the sampling frame and subsequently contacted each manager by telephone to locate an appropriate key informant within each firm. In total, we identified 550 managers who met Campbell’s (1955) informant criteria and who worked in companies deemed to be appropriate for the study. To maximize response rates and ensure a sufficiently large pool of
JOURNAL OF MARKETING RESEARCH, AUGUST 2007 informants for Time 2, we followed Yu and Cooper’s (1983) suggestion to combine a mail survey with a telephone option. Each informant who agreed to participate in the study was mailed a questionnaire packet, and if the informant desired, an appointment was made to conduct a telephone interview that paralleled the survey instrument. To avoid the risk of self-selection and still capture relationships that were important enough to be salient to the informants, we asked the managers to select and describe a particular relationship in which the firm was the third-largest buyer (in terms of annual dollar sales) for a particular item (Anderson and Narus 1990). Three hundred forty-two questionnaires were completed at Time 1, for a response rate of 62%. To assess whether systematic differences existed between the questionnaires administered by telephone (n = 129) and mail (n = 213), we tested the null hypothesis of no mean differences between the groups with respect to our study variables. We found no significant differences. Moreover, we compared our sample of manufacturers with the initial sampling frame with respect to demographics, such as company size and annual revenue, and found no significant differences. Three years later, we contacted the same companies for a second wave of data collection. At this stage, 24% of the firms contacted either could not be reached or refused to participate in the second round. To ensure that the informants were answering questions with respect to the same buyer in the second wave of data collection, we provided them with the name of the buyer and the product description. Ultimately, our final sample consisted of 105 matched questionnaires across Times 1 and 2 (31% of the 342 from Time 1), for an overall response rate of 19% (of the 550 companies deemed appropriate for the study; this represents 8% of the initial sampling frame). Overall, our response rates compare favorably with those of other studies of interfirm relationships (e.g., Rindfleisch and Moorman 2001). We tested whether there were systematic differences between the questionnaires administered by telephone (n = 38) and mail (n = 67) and compared our final sample of manufacturers with the initial sampling frame with respect to company size and annual revenue. We found no significant differences. Finally, we compared our final sample (who participated in both waves) with the manufacturers that participated only in the first wave with respect to our study variables. Again, we found no significant differences. Key Informant Checks We assessed the quality of our reports through a post hoc check of informant knowledge within the survey instrument, in accordance with Kumar, Stern, and Anderson’s (1993) approach. On the basis of this test, which employed a seven-point knowledge scale, we eliminated three companies at Time 1. The average knowledge score for the remaining cases was 6.5 (SD = .75). A concern with collecting data from the same company at two different times is attrition due to informant turnover. Recognizing this threat, we had each new informant complete a telephone assessment that verified the established criteria. In addition, each survey sent out to new informants included a post hoc check on the focal informants’ knowledge about the specific nature of the relationship over the past three years. We eliminated only one case as a result of
Interfirm Monitoring, Social Contracts, and Relationship Outcomes this check. The average knowledge score was 6.1 (SD = 1.14), indicating that the new informants were highly qualified to report on their firms’ relationships.
429
Construct Validity We subjected the reflective multi-item measures to a systematic assessment of internal consistency and unidimensionality. First, we evaluated each item set on the basis of item-to-total correlations and exploratory factor analysis. Second, we subjected the entire item set to confirmatory factor analysis using LISREL 8.5 (Jöreskog et al. 2001). Although the chi-square goodness-of-fit index was significant (χ2 = 370.57, p < .01), as is almost always the case in larger samples, the root mean square error of approximation (.07), the incremental fit index (.87), and the comparative fit index (.87) all meet the critical values for good model fit (Bollen 1989; Browne and Cudeck 1992). Moreover, all the factor loadings are large and significant (t-values > 2.00). As the Appendix shows, the composite reliability (CR) and average variance extracted (AVE) for all constructs met Fornell and Larker’s (1981) recommended thresholds. To assess discriminant validity, we calculated the shared variance (SV) between all possible pairs of constructs and demonstrated (see the Appendix) that they are lower than the AVE for the individual constructs. We also estimated several additional models in which each pair of factor correlations was constrained to unity. We then compared the fit of each new model with the original unconstrained model (Bagozzi and Phillips 1982). All the measures show evidence of discrimination.4 The correlation matrix for the variable set appears in Table 1. Note that though the mean level of opportunism is low, it is consistent with the levels reported in previous studies (e.g., Gundlach, Achrol, and Mentzer 1995).
Measures We operationalized the key study variables using multiitem reflective scales. The actual items, response formats, and key descriptive statistics appear in the Appendix. Supplier opportunism. The scale describes the extent to which the supplier engages in “self-interest seeking with guile” (see Williamson’s [1975, p. 6] conceptual definition). We derived the six items in part from those that John (1984) and Gundlach, Achrol, and Mentzer (1995) used, and we adapted them to our context. Monitoring. We used two different monitoring scales, which describe the buyer’s efforts to verify the supplier’s performance on output and behavior. The measures reflect the supplier’s perception. We derived them from those that Celly and Frazier (1996) and Stump and Heide (1996) used, and we adapted them to our context. We define “microlevel social contract” as one party’s perception of agreement on the terms that characterize the relationship with another party (Dunfee, Smith, and Ross 1999). We used two separate scales that describe the supplier’s perception of the extent to which there is an agreement in the relationship with the buyer regarding output and behavior standards. Recall from our previous discussion that (1) a microlevel social contract reflects a supplier’s perceptions about what the parties have agreed to and not necessarily an agreement per se, and (2) the supplier’s perceptions of the situation at hand drives the decision to pursue (or refrain from) opportunistic actions (see Scheer, Kumar, and Steenkamp 2003). Control variables. We included three control variables in the model. First, to account for the notion that larger buyers may be able to influence the partner’s behavior because of their superior bargaining position, we included a measure of relative firm size. Second, we included a measure of the supplier’s dependence on the buyer, which we captured through the perceived replaceability of the buyer for the product in question (Heide and John 1988). Third, although our theoretical focus is on whether monitoring (at Time 1) influences opportunism at Time 2, we included in the model a measure of opportunism at Time 1 to ensure that the estimated effects of the focal independent variables are independent of opportunism at Time 1.
Hypotheses Tests and Results We tested our hypotheses following Cohen and colleagues’ (2003) suggested model specification for two-wave data. Specifically, we estimated an ordinary least squares regression model with supplier opportunism at Time 2 as a function of buyer output monitoring at Time 1 (H1), buyer behavior monitoring at Time 1 (H2), microlevel social con4Recall from our previous discussion that our conceptualization of microlevel social contracts is distinct from relational norms. To establish this formally, we estimated an additional confirmatory factor analysis model that, in addition to our focal variables, included measures of a commonly studied relational norms—namely, solidarity. We calculated the SV between all pairs of constructs and demonstrated that SV between solidarity and the two social contract measures (output and behavior) was only 4% and 7%, respectively.
Table 1 CORRELATION MATRIX Construct Supplier opportunism at Time 2 (SO2) Buyer’s behavior monitoring (BBM) Buyer’s output monitoring (BOM) Microsocial contract: behavior (MCB) Microsocial contract: output (MCO) Relative size_Dummy1 (S1) Relative size_Dummy2 (S2) Replaceability (R) Supplier opportunism at Time 1 (SO1) *p < .05 (two-tailed). **p < .01 (two-tailed).
SO2 1.0** –.05* –.26** –.21** –.16 .13** –.13** –.03** .27**
BBM
BOM
MCB
MCO
S1
S2
R
SO1
1.0** .20*** .06** .12** –.10** –.00** .06** .24**
1.0** .17** .27*** .09** –.07** .07** –.06**
1.0* .47** .12 –.11 –.01 –.21*
1.0** .03 .05 –.05** –.32**
1.0* * –.62** –.07** –.08**
1.0* .20* –.11
1.0 .05
1.0
430
JOURNAL OF MARKETING RESEARCH, AUGUST 2007
tract at Time 1 (output and behavior), the interaction between buyer behavior monitoring and microlevel social contract (behavior) (H3), the interaction between buyer output monitoring and microlevel social contract (output) (H4), opportunism at Time 1, and the two measures of power. In line with Cohen and colleagues’ procedure, we meancentered all the independent variables.5 Table 2 shows the estimated coefficients and associated tstatistics. The effect of buyer output monitoring on supplier opportunism is significant and negative (t = –2.99, p < .05), consistent with H1. Moreover, the effect of buyer behavior monitoring on supplier opportunism is significant and positive (t = 1.84, p < .05), consistent with H2. Although we did not hypothesize any direct effects of microlevel social contract, microlevel social contract (behavior) has a significant, negative effect on supplier opportunism. We also find that the interaction between behavior monitoring and microlevel social contract (behavior) is significant and negative (t = –2.68, p < .05), in support of H3. Similarly, the interaction between output monitoring and microlevel social contract (output) is significant and negative (t = –1.48, p < .10), in support of H4. As we expected, size has a significant effect on supplier opportunism (supplier > buyer; t = 1.47, p < .10). Recall that our hypotheses involve different contingency effects of buyer monitoring on supplier opportunism over the range of microlevel social contracts. We can show these effects more clearly by plotting the partial derivative of the
regression equation following Schoonhoven’s (1981) procedure. In Figure 1, Panel A, we show the effects of behavior monitoring. As the plot indicates, the effect is nonmonotonic over the range of the social contract, consistent with H2 and H3. For lower levels or weaker social contracts (on the mean-centered scale), an increase in buyer behavior monitoring has a positive effect on supplier opportunism, whereas the effect becomes negative for higher levels or stronger social contracts. In Figure 1, Panel B, we show the corresponding effect for output monitoring. Here, the relevant contingency effect is monotonic over the range of the social contract, consistent with H1 and H4. Although the effect does not change sign, the presence of a microlevel social contract enhances the effectiveness of output monitoring, as we expected. In combination, these effects suggest that monitoring and microlevel social contracts play distinct and complementary roles in a relationship. DISCUSSION The Monitoring Construct and Interfirm Relationships The theoretical construct of monitoring occupies a central position in several different literature streams. However, the theory of monitoring is still in its infancy. A particular problem is the existence of competing arguments about monitoring’s actual effects (Sewell and Baker 2006). On the one hand, authors such as Tenbrunsel and Messick (1999) have argued that systems with low detection ability (e.g., firms that do not engage in regular monitoring) are likely to produce undesirable behaviors on a regular basis. On the other hand, other researchers have told cautionary tales about monitoring, to the point of characterizing theories that incorporate the monitoring construct as “dangerous” (Perrow 1986) and “bad for practice” (Ghoshal and Moran 1996). Both arguments are most likely too categorical in nature. Our study shows that monitoring is capable of both increasing and decreasing partner opportunism, consistent with existing conceptual arguments. More important, however, we show that the emergence of a particular effect depends crucially on (1) the particular form of monitoring and (2) the context in which the monitoring initiative is imple-
5Because we control for opportunism at Time 1, our model is a conditional change model. We also considered the possibility of testing our hypotheses by means of an unconditional change-score model (i.e., method of first difference, which on the dependent variable side would use the difference between opportunism at Time 1 and Time 2). We decided against such an approach for several reasons. First, when measurement error is present (which is common with survey data), the reliability of the difference scores (i.e., Y2 – Y1) can be less than the reliability of either Y2 or Y1. Second, models with a differenced dependent variable make the restrictive assumption that the lagged dependent variable (i.e., Y1) has no influence on either Y2 or the change in Y. This assumption is often incorrect on both substantive and statistical grounds, and the effect is typically one of overcorrection of the postscore by the prescore (for a more detailed discussion, see Cohen et al. 2003). Finally, examining the effect of a change in monitoring on a change in opportunism is conceptually different from our current research hypotheses—namely, to examine the effect of monitoring at Time 1 on opportunism at Time 2.
Table 2 ORDINARY LEAST SQUARES REGRESSION MODEL DEPENDENT VARIABLE: SUPPLIER OPPORTUNISM AT TIME 2 Independent Variables (Time 1)
Unstandardized Coefficients
Standardized Coefficients
Monitoring, output Social contract, output Monitoring, output × social contract, output Monitoring, behavior Social contract, behavior Monitoring, behavior × social contract, behavior
–.16 .03 –.09 .10 –.18 –.14
–.30 .04 –.15 .20 –.35 –.31
–2.99** .33 –1.48* 1.84** –2.82** –2.68**
Controls Relative size_Dummy1 (X times larger) Relative size_Dummy2 (X times smaller) Replaceability Supplier opportunism at Time 1
.30 –.02 .01 .16
.18 –.01 .04 .12 Adjusted R2 = .18
1.47* –.11 .37 1.10
*p < .10 (one-tailed). **p < .05 (one-tailed).
t-Value
Interfirm Monitoring, Social Contracts, and Relationship Outcomes Figure 1 ILLUSTRATIONS OF THE CONTINGENCY EFFECTS OF MONITORING A: Nonmonotonic Effect of Buyer Behavior Monitoring on Supplier Opportunism
B: Monotonic Effect of Output Monitoring on Supplier Opportunism
mented—more specifically, on the strength of the social contract that characterizes a given relationship. From a theoretical standpoint, the study responds to the call for research on the “complex interactions” between formal and informal relationship elements (e.g., Gibbons 1999). Specifically, we provide evidence that informal organizational arrangements, such as microlevel social contracts, are important prerequisites for the deployment of formal ones (e.g., monitoring), consistent with Barnard’s (1938) early (but largely untested) theoretical arguments. Moreover, our findings imply that the general tendency in empirical research to focus on individual relationship dimensions in isolation (e.g., monitoring) may obscure crucial interrelationships (e.g., between monitoring and social contracts) that were implied in the original literature (Williamson 1975). Recently, researchers have specifically advocated that governance systems be designed with “inter-
431
nally consistent attributes” (Williamson 2002, p. 175) and with careful attention to “complementarities across design variables” (Roberts 2004, p. 283). To date, however, empirical evidence has been limited. Our results tell a cautionary tale about theoretical disaggregation in general and about conceptual and empirical shifts away from the notion of hierarchical exchange in particular. From a theoretical standpoint, our study shows the importance of beginning to “reassemble” governance systems and of explicitly considering constellations of governance processes. Implications for Practice This study also serves as a cautionary tale about nondiscriminating investments in monitoring. In particular, this is the case for behavior monitoring, which we showed has the distinct potential to promote opportunism. However, we also showed that even potentially obtrusive forms of monitoring may suppress opportunism, provided that they are supported by microlevel social contracts. From a practical standpoint, social contracts are important because they are likely to be “design variables,” which can be purposely and easily crafted in a relationship. In contrast, other informal relationship elements, such as solidarity norms, may require extensive prior socialization and time commitments. In general, given the costs associated with a monitoring program, both in terms of (1) direct surveillance efforts and (2) the opportunity costs associated with the partner’s reactions, firms are well advised to create contexts proactively that support their ongoing governance initiatives. Limitations and Further Research Some limitations of the current research should be pointed out. Importantly, some of them represent worthwhile avenues for further research. First, we limited our focus to the supplier’s reactions to the buyer’s monitoring efforts. Further research could compare the supplier’s perspective on these processes with that of the buyer. For example, a buyer may have fundamentally different hypotheses about the drivers of opportunism than those that underlie the supplier’s actual decision calculus. As such, it is possible that a buyer could pursue costly governance initiatives that are counterproductive or may involve resource misallocations. Second, although we argued that a microlevel social contract can be an actual design variable, our current data do not permit us to document the specific manner in which they come about. However, this raises questions, for example, about the extent to which such contracts are supported by relationship-specific processes, formal contracts, and industry macrocultures. Finally, further research could usefully be directed toward exploring even more complex constellations of governance processes. In the current article, we limited our attention to the interaction between monitoring and microlevel social contracts. However, additional insights may be gained by taking this line of thought further, for example, by examining the effects of particular incentive structures. Recall from our prior discussion that monitoring has potential cost implications for an exchange partner (Halaby 1986). As such, if the role of incentives is accounted for, some of the focal effects may change in notable ways. For
432
JOURNAL OF MARKETING RESEARCH, AUGUST 2007
example, the opportunism-enhancing effect of behavior monitoring in the absence of a strong microlevel social contract may further increase in the presence of incentives. In addition, in the presence of incentives that punish partner actions, output monitoring, which constitutes a relatively unobtrusive strategy, may actually take on an obtrusive character and ultimately promote opportunism. APPENDIX: MEASURES Supplier Opportunism at Time 1 (Reliability = .73) and Time 2 (Reliability = .87) Please evaluate the degree to which the following statements accurately describe your company by circling the most appropriate number on the scale (seven-point scale: 1 = “completely inaccurate description,” and 7 = “completely accurate description”; Time 1: M = 1.38, SD = .48; CR = .80; AVE = 41%; and SV = 12%. Time 2: M = 1.56, SD = .62; CR = .88; AVE = 55%; and SV = 14%): 1. On occasion, we lie about certain things in order to protect our interests. 2. We sometimes promise to do things without actually doing them later. 3. We do not always act in accordance with our contract(s). 4. We sometimes try to breach informal agreements between our companies to maximize our own benefit. 5. We will try to take advantage of “holes” in our contract to further our own interests. 6. We sometimes use unexpected events to extract concessions from the buyer.
Buyer Behavior Monitoring (Reliability = .74) Please indicate the extent of ongoing monitoring undertaken by this customer for the supply of this product in the areas listed below (seven-point scale: 1 = “minimal monitoring by buyer,” and 7 = “extensive monitoring by buyer”; M = 2.24, SD = 1.28; CR = .77; AVE = 53%; and SV = 5%): 1. Production schedules. 2. Storage and handling practices. 3. Quality-control procedures.
Buyer Output Monitoring (Reliability = .68) Please indicate the extent of ongoing monitoring undertaken by this customer for the supply of this product in the areas listed below (seven-point scale: 1 = “minimal monitoring by buyer,” and 7 = “extensive monitoring by buyer”; M = 5.55, SD = 1.16; CR = .73; AVE = 49%; and SV = 14%): 1. Product quality. 2. Delivery timeliness. 3. Order accuracy.
Microlevel Social Contract: Behavior (Reliability = .90) Please indicate the extent to which a mutual agreement exists between your company and this customer on standards in the areas listed below (seven-point scale: 1 = “low degree of mutual agreement,” and 7 = “complete mutual agreement”; M = 5.87, SD = 1.21; CR = .93; AVE = 82%; and SV = 27%):
1. Production schedules. 2. Storage and handling practices. 3. Quality-control procedures.
Microlevel Social Contract: Output (Reliability = .84) Please indicate the extent to which a mutual agreement exists between your company and this customer on standards in the areas listed below (seven-point scale: 1 = “low degree of mutual agreement,” and 7 = “complete mutual agreement”; M = 6.29, SD = .76; CR = .86; AVE = 67%; and SV = 27%): 1. Product quality. 2. Delivery timeliness. 3. Order accuracy.
REFERENCES Aiken, Michael and Jerald Hage (1966), “Organizational Alienation: A Comparative Analysis,” American Sociological Review, 31 (4), 497–507. Alchian, Armen and Harold Demsetz (1972), “Production, Information Costs, and Economic Organization,” American Economic Review, 62 (5), 777–95. Anderson, Erin and Richard L. Oliver (1987), “Perspectives on Behavior-Based Versus Outcome-Based Salesforce Control Systems,” Journal of Marketing, 51 (October), 76–88. Anderson, James C. and James A. Narus (1990), “A Model of Distributor Firm and Manufacturer Firm Working Partnerships,” Journal of Marketing, 54 (January), 42–58. Bagozzi, Richard P. and Lynn W. Phillips (1982), “Representing and Testing Organizational Theories: A Holistic Construal,” Administrative Science Quarterly, 27 (3), 459–89. Barkema, Harry G. (1995), “Do Top Managers Work Harder When They Are Monitored?” Kyklos, 48 (1), 19–42. Barnard, Chester I. (1938), The Functions of the Executive. Cambridge, MA: Harvard University Press. Bollen, Kenneth A. (1989), Structural Equations with Latent Variables. New York: John Wiley & Sons. Brehm, Jack W. (1966), A Theory of Psychological Reactance. New York: Academic Press. Browne, Michael W. and Robert Cudeck (1992), “Alternative Ways of Assessing Model Fit,” Sociological Methods and Research, 21 (4), 230–58. Campbell, Donald T. (1955), “The Informant in Quantitative Research,” American Journal of Sociology, 60 (4), 339–42. Celly, Kirti S. and Gary L. Frazier (1996), “Outcome-Based and Behavior-Based Coordination Efforts in Channel Relationships,” Journal of Marketing Research, 33 (May), 200–210. Cohen, Jacob, Patricia Cohen, Stephen G. West, and Leona S. Aiken (2003), Applied Multiple Regression/Correlation Analysis for the Behavioral Sciences. Hillsdale, NJ: Lawrence Erlbaum Associates. Deci, Edward L., Richard Koestner, and Richard M. Ryan (1999), “A Meta-Analytic Review of Experiments Examining the Effects of Extrinsic Rewards on Intrinsic Motivation,” Psychological Bulletin, 125 (6), 627–68. ——— and Richard M. Ryan (1987), “The Support of Autonomy and the Control of Behavior,” Journal of Personality and Social Psychology, 53 (5), 1024–1037. Dunfee, Thomas W., Craig N. Smith, and William T. Ross Jr. (1999), “Social Contracts and Marketing Ethics,” Journal of Marketing, 63 (July), 14–33. Eisenhardt, Kathleen M. (1985), “Control: Organizational and Economic Approaches,” Management Science, 31 (2), 134–49. Fornell, Claes and David F. Larcker (1981), “Evaluating Structural Equation Models with Unobservable Variables and Measure-
Interfirm Monitoring, Social Contracts, and Relationship Outcomes ment Error,” Journal of Marketing Research, 18 (February), 39–50. Ghosh, Mrinal and George John (1999), “Governance Value Analysis and Marketing Strategy,” Journal of Marketing, 63 (Special Issue), 131–45. Ghoshal, Sumantra and Peter Moran (1996), “Bad for Practice: A Critique of the Transaction Cost Framework,” Academy of Management Journal, 21 (1), 13–47. Gibbons, Robert (1999), “Taking Coase Seriously,” Administrative Science Quarterly, 44 (1), 145–57. Gundlach, Gregory T., Ravi S. Achrol, and John T. Mentzer (1995), “The Structure of Commitment in Exchange,” Journal of Marketing, 59 (January), 78–92. Halaby, Charles N. (1986), “Worker Attachment and Workplace Authority,” American Sociological Review, 51 (5), 634–49. Heide, Jan B. and George John (1988), “The Role of Dependence Balancing in Safeguarding Transaction-Specific Assets in Conventional Channels,” Journal of Marketing, 52 (January), 20–35. ——— and ——— (1992), “Do Norms Matter in Marketing Relationships?” Journal of Marketing, 56 (April), 32–44. Jacobides, Michael G. and David C. Croson (2001), “Information Policy: Shaping the Value of Agency Relationships,” Academy of Management Review, 26 (2), 202–223. Jensen, Michael C. (1998), Foundations of Organizational Strategy. Cambridge, MA: Harvard University Press. Jensen, William H. and Michael C. Meckling (1976), “Theory of the Firm: Managerial Behavior, Agency Costs, and Ownership Structure,” Journal of Financial Economics, 3 (4), 305–360. John, George (1984), “An Empirical Investigation of Some Antecedents of Opportunism in a Marketing Channel,” Journal of Marketing Research, 21 (August), 278–89. Jöreskog, Karl G., Dag Sörbom, Stephen du Toit, and Mathilda du Toit (2001), LISREL 8: New Statistical Features. Hillsdale, NJ: Lawrence Erlbaum Associates. Kumar, Nirmalya, Lisa K. Scheer, and Jan-Benedict E.M. Steenkamp (1995), “The Effects of Supplier Fairness on Vulnerable Resellers,” Journal of Marketing Research, 22 (February), 54–65. ———, Louis W. Stern, and James C. Anderson (1993), “Conducting Interorganizational Research Using Key Informants,” Academy of Management Journal, 36 (6), 1633–51. Lal, Rajiv (1990), “Improving Channel Coordination Through Franchising,” Marketing Science, 9 (4), 299–318. Lind, Allan E. and Tom R. Tyler (1988), The Social Psychology of Procedural Justice. New York: Plenum. Locke, John (1690), Two Treatises of Government. London: Printed for Awnsham Churchill. Macneil, Ian R. (1980), The New Social Contract. New Haven, CT: Yale University Press. Murry, John P. and Jan B. Heide (1998), “Managing Promotion Program Participation Within Manufacturer–Retailer Relationships,” Journal of Marketing, 62 (January), 58–69.
433
Narayandas, Das and V. Kasturi Rangan (2004), “Building and Sustaining Buyer–Seller Relationships in Mature Industrial Markets,” Journal of Marketing, 68 (July), 63–77. Ouchi, William G. (1979), “A Conceptual Framework for the Design of Organizational Control Mechanisms,” Management Science, 25 (9), 833–48. Perrow, Charles (1986), Complex Organizations: A Critical Essay. New York: Random House. Podsakoff, Philip M., Scott B. MacKenzie, Jeong-Yeon Lee, and Nathan P. Podsakoff (2003), “Common Method Biases in Behavioral Research: A Critical Review of the Literature and Recommended Remedies,” Journal of Applied Psychology, 88 (5), 879–903. Rindfleisch, Aric and Christine Moorman (2001), “The Acquisition and Utilization of Information in New Product Alliances: A Strength-of-Ties Perspective,” Journal of Marketing, 65 (April), 1–18. Roberts, John (2004), The Modern Firm: Organizational Design for Performance and Growth. Oxford: Oxford University Press. Rousseau, Denise M. (1995), Psychological Contracts in Organizations: Understanding Written and Unwritten Agreements. Thousand Oaks, CA: Sage Publications. Rousseau, Jean-Jacques (1762), Du contrat social, ou, Principes du droit politique (The Social Contract or Principles of Political Right). Geneva: Archives de la Société Jean-Jacques Rousseau. Scheer, Lisa K., Nirmalya Kumar, and Jan-Benedict E.M Steenkamp (2003), “Reactions to Perceived Inequity in U.S. and Dutch Interorganizational Relationships,” Academy of Management Journal, 46 (3), 303–317. Schoonhoven, Claudia B. (1981), “Problems with Contingency Theory: Testing Assumptions Hidden Within the Language of Contingency ‘Theory,’” Administrative Science Quarterly, 26 (3), 349–77. Sewell, Graham and James R. Baker (2006), “Coercion Versus Care: Using Irony to Make Sense of Workplace Surveillance,” Academy of Management Review, 31 (4), 934–61. Simon, Herbert A. (1951), “A Formal Theory of the Employment Relationship,” Econometrica, 19 (3), 293–305. Stump, Rodney L. and Jan B. Heide (1996), “Controlling Supplier Opportunism in Industrial Relationships,” Journal of Marketing Research, 33 (November), 431–41. Tenbrunsel, Ann E. and David M. Messick (1999), “Sanctioning Systems, Decision Frames, and Cooperation,” Administrative Science Quarterly, 44 (4), 684–707. Weber, Max (1947), The Theory of Social Economic Organization. New York: The Free Press. Williamson, Oliver E. (1975), Markets and Hierarchies: Analysis and Antitrust Implications. New York: The Free Press. ——— (1985), The Economic Institutions of Capitalism. New York: The Free Press. ——— (2002), “The Theory of the Firm as Governance Structure: From Choice to Contract,” Journal of Economic Perspectives, 16 (3), 171–95. Yu, Julie and Harris Cooper (1983), “A Quantitative Review of Research Design Effects on Response Rates to Questionnaires,” Journal of Marketing Research, 20 (February), 36–44.
