Journal of Business Research 60 (2007) 285 – 295

Interfirm behavior and goal alignment in relational exchanges Andrew T. Stephen a,⁎, Leonard V. Coote b a

Columbia University, Graduate School of Business, Uris Hall, 3022 Broadway, Room 311, New York, NY 10027, USA b UQ Business School, University of Queensland, Brisbane, Qld, Australia Received 1 November 2005; received in revised form 1 August 2006; accepted 1 October 2006

Abstract This article considers the potential antecedents and performance-related consequences of a socialization-based approach to governing interfirm relationships. Relational behaviors, the manifestation of relational norms, are considered as a form of governance, in contrast to more formal and explicit contract-based approaches. The study addresses a gap in extant literature in relation to understanding how manageractionable behaviors (i.e., supportive leadership, behavior-based monitoring) influence relational behaviors, and whether this form of relationship governance has performance implications. Data from a mail survey of construction industry contractors and subcontractors are analyzed. The authors find that relational governance can be effective in achieving coordination in marketing relationships, which in turn improves financial performance. Contractors' supportive leadership and behavior-based monitoring efforts respectively play positive and negative roles in shaping relational behavior. However, the interaction of leadership and monitoring is crucial, with subcontractors tolerating monitoring when contractors employ a management approach that combines monitoring and supportive leadership. This research provides support for the notion that relational governance can be a solution to agency problems of hidden action. The findings have implications for the theory and practice of relationship management, particularly as firms seek out alternatives to formal approaches to relationship management. © 2006 Elsevier Inc. All rights reserved. Keywords: Relational exchanges; Goal alignment; Relational behaviors; Agency relationships

Relationships are an important way of organizing economic activities in marketing. Relationships between buyers and suppliers, clients and advertising agencies, and manufacturers and wholesalers, are examples of marketing relationships. Such relationships are agency relationships (Bergen et al., 1992). An agency theory perspective highlights a central problem for firms engaging in marketing relationships; that is, parties must work together and coordinate their activities in order to achieve desirable performance outcomes. The coordination problem in agency theory is a problem of hidden action or moral hazard (cf. Bergen et al., 1992; Eisenhardt, 1989). Solving this coordination problem is important, because agency theory suggests that the performance of individual parties can have a substantial impact on the overall performance of a relationship (Arrow,

⁎ Corresponding author. Tel.: +1 212 731 2012; fax: +1 734 758 2012. E-mail addresses: [email protected] (A.T. Stephen), [email protected] (L.V. Coote). 0148-2963/$ - see front matter © 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.jbusres.2006.10.022

1985; Eisenhardt, 1989; Heide, 1994). Much research examines formal approaches to coordinating marketing relationships, such as the use of explicit contracts and formal authority structures. However, cooperation in marketing relationships builds increasingly on socialization efforts (Heide, 1994; Morgan and Hunt, 1994). Relational governance approaches to managing relationships are the focus of relevant literature (e.g., Bello et al., 2003; Mishra et al., 1998; Wright et al., 2001). Relational governance attempts to socially embed exchange partners such that their goals are in close alignment and their joint utility is maximized. This effort requires that socialization efforts take place, which have the aim of promoting relational social norms. Relational behaviors, the manifestation of relational norms, are central to the process of relational governance (cf. Heide and John, 1992; Macneil, 1980). Specifically, relational behaviors create the conditions necessary for bringing goals into alignment. This logic is worth exploring further, because the logic links relational governance and behaviors to the effective functioning

286

A.T. Stephen, L.V. Coote / Journal of Business Research 60 (2007) 285–295

of relationships. However, the implementation of relational governance receives scant attention in extant literature (i.e., antecedents that can be implemented by managers are not well understood). Additionally, the literature does not document the performance implications of relational governance. To address these gaps, this article explores a socialization approach to relationship management. The article develops and tests a model of relational governance which extends the emerging stream of literature on socialization approaches to relationship management (e.g., Bello et al., 2003; Mishra et al., 1998; Wright et al., 2001). The aim is to understand the extent such an approach achieves coordination in relational exchanges and enhances performance following pro-social interactions between parties. The framework explicitly considers the impact of socialization efforts (i.e., supportive leadership, behaviorbased monitoring) on perceptions of relational behaviors, goal alignment, and financial performance. Thus, the remainder of this paper examines two central issues: (1) the influence of relational governance on perceptions of relational behaviors, and (2) the influence of relational behaviors on perceptions of goal alignment and performance. 1. Agency theory and relational exchange 1.1. Agency theory and relational governance Agency relationships are interactions where one party (the “buyer”) depends on another party (the “supplier”) to undertake some action on the buyer's behalf, and the outcomes achieved by the buyer are affected by the decisions made by and the performance of the supplier (Bergen et al., 1992; Wright et al., 2001). Key assumptions of agency theory are that buyers and suppliers are self-interest maximizers, and as a result will tend to have incongruent goals (Eisenhardt, 1989). Because agency theory posits that independent firms seek to maximize their self-interest, problems of coordination arise as the central issue. Much of the agency literature examines formal and contractual approaches to resolving coordination problems. By contrast, the emerging literature on relational governance explores informal mechanisms for achieving coordination (e.g., Mishra et al., 1998; Wright et al., 2001). Relational governance relies on socialization efforts for promoting relationalism, and encouraging close coordination and cooperation. Although the passage of time may play an important role, close and purposeful interactions are crucial to effective relational governance. Socialization efforts can become the basis for sustainable, long-term relationships. Parties seek complex, personal, and non-economic satisfactions from socialized exchanges (Dwyer et al., 1987). This search provides incentives to respect a partner, because of the valuable intrinsic benefits encapsulated in socialized exchanges. Through interactions that are prosocial in nature, the parties to a relationship can establish relational norms that govern exchanges without reference to explicit contracts. Exchange participants are therefore motivated to transmit positive sanctions to their partners through close interactions (Willer and Anderson, 1981). Exchanges can

become socialized to the extent that the parties behave as a single, socially embedded entity that works to maximize joint utility. This outcome corresponds to a clan type of governance, which is a highly socialized form of coordination (Jaworski, 1988; Ouchi, 1979). Hence, relational governance is a possible mechanism for achieving coordination and mitigating agency problems of hidden action. 1.2. Relational behaviors Relational social norms build from the expectation of mutuality of interest, prescribe stewardship behavior, and are designed to enhance the wellbeing of a relationship as a whole (Macneil, 1978, 1980). Although relational norms may appear in different forms, it is generally agreed that norms are shared behavioral expectations that develop between exchange partners (Heide and John, 1992). Relational norms fill gaps in explicit contracts and formal understandings, and are manifest in relational behaviors (Lusch and Brown, 1996). The willingness of parties to display relational behaviors is at the core of relational governance. Relational behaviors are central to creating a socially embedded relationship where the maximization of joint utility is promoted over self-interest. Recognizing common goals requires relational behaviors, including trustworthiness, as a prerequisite (Dwyer et al., 1987; Morgan and Hunt, 1994). Perceptions of relational behaviors are therefore a crucial mechanism in coordinating relationships and are a precursor to future exchange. Discussions of relational behaviors draw heavily on the relational contracting literature (Macneil, 1978, 1980). Types or dimensions of relational behaviors include flexibility, solidarity, information exchange, and trustworthiness (Dwyer et al., 1987; Heide and John, 1992; Stinchcombe, 1986). These behaviors are conceptualized as highly interrelated dimensions of a higherorder construct. Flexibility is a bilateral expectation of willingness to make adaptations as circumstances change (Noordewier et al., 1990). Solidarity is a bilateral expectation that joint rather than individual outcomes are highly valued (Heide and John, 1992). Information exchange refers to a bilateral expectation that parties will proactively provide information useful to their partners (Lusch and Brown, 1996). Trustworthiness is a bilateral expectation that parties will act with integrity and honesty throughout the course of a relationship (Dwyer et al., 1987). The dimensions clearly support one another and constitute a set of interlocking behaviors. 2. Conceptual model and hypotheses Fig. 1 presents the basic framework and hypotheses. A fundamental premise of the model is that perceptions of relational behaviors mediate the impact of socialization efforts on goal alignment and performance. The literature on relational governance implies that a process such as Fig. 1 shows is appropriate. The behavior of one party toward its partner is instrumental in shaping a relationship and the outcomes that the parties can ultimately achieve. However, successful relational exchange requires two parties working cooperatively

A.T. Stephen, L.V. Coote / Journal of Business Research 60 (2007) 285–295

287

Fig. 1. Hypothesized model. Key: SUPPT = Leadership Support; FEEDB = Leader Feedback; MONIT = Behavior-Based Monitoring; SUPT·MON = Product of support and monitoring; FEED·MON = Product of feedback and monitoring; FLEX = Flexibility; INFOEX = Information Exchange; SOLID = Solidarity; TRUST = Trustworthiness; GOAL = Goal Alignment; FINBUD = Financial (cost/budget) Performance.

to achieve shared performance outcomes (Anderson and Narus, 1990; Morgan and Hunt, 1994). The parties' perceptions of socialization efforts and relational behaviors are important and influence the extent to which coordination, and in turn, performance outcomes can be achieved. Interfirm relationships are dynamic and can require complex adjustments in response to changing internal and external environments and in response to shifting expectations of roles and responsibilities (Macneil, 1978, 1980). This necessitates an important role for relational behaviors because the coordination of activities and the alignment of goals require relational behaviors (e.g., solidarity, trustworthiness) as a fundamental prerequisite. 2.1. Leadership and relational behaviors Leadership is essential to socializing exchange partners and is expected to influence perceptions of relational behaviors. Buyers need to be proactive in signaling the importance of relationships to suppliers. This can be achieved through supportive leadership (i.e., support and encouragement, and constructive feedback). Support requires the provision of informal support and guidance, which can result in a positive motivational response (House, 1971). Feedback involves the provision of constructive feedback. Leadership helps to develop relational behaviors through sending positive signals, as well as by encouraging closeness among the parties. In order to be supportive and to provide feedback, parties need to be closely involved with each other throughout a relationship. By providing encouragement, guidance, and constructive feedback, buyers signal their interest in and willingness to participate actively with their suppliers. A positive motivational effect is expected, and will result in perceptions of greater flexibility and solidarity, and greater exchange of information and trustworthiness.

H1. An increase in perceptions of supportive leadership leads to an increase in perceptions of relational behaviors. 2.2. Monitoring and relational behaviors Addressing coordination problems in exchange relationships can involve the traditional control system approaches of providing incentives (outcome-based strategies) or closely monitoring behaviors (behavior-based strategies) (cf. Anderson and Oliver, 1987). While more formal mechanisms may help to define goals, such approaches may limit the willingness of parties to engage in relational behaviors (Ghoshal and Moran, 1996). Because they lack a social or relational dimension, formal and explicit approaches can result in unsustainable, shorter-term relationships. In the case of monitoring, effective governance is based on close supervision of a supplier as it completes its work. However, suppliers may resent buyers' attempts to closely monitor their activities. Close monitoring may foster antisocial behaviors, such as the concealment of underlying motives and intentions in subtle and hard to detect ways (Ghoshal and Moran, 1996). A party's perception of being closely monitored might foster a perceived loss of self-determination or a psychological reactance-type response (cf. Brehm, 1966). This might undermine a party's motivation to act in the best interests of the relationship. Hence, relational behaviors may not be fostered through close monitoring and supervision. H2. An increase in perceptions of behavior-based monitoring leads to a decrease in perceptions of relational behaviors. 2.3. Interaction of leadership and monitoring The expectation is that a combination of leadership and monitoring has a positive motivational effect and promotes

288

A.T. Stephen, L.V. Coote / Journal of Business Research 60 (2007) 285–295

relational behaviors (Anderson and Oliver, 1987; Challagalla and Shervani, 1996). Supportive leadership is pro-social and has the potential to offset the negative aspects of monitoring. Monitoring in isolation can be a deterrent to relational behaviors. However, such monitoring does provide socialization opportunities by potentially bringing parties close together (Anderson and Oliver, 1987; Eisenhardt, 1989). When supportive leadership and close monitoring are combined, monitoring will less likely result in feelings of unwarranted scrutiny (cf. Challagalla and Shervani, 1996). A purpose of socialization efforts is to encourage closer interactions and to send positive signals (Heide, 1994). An approach that simultaneously combines leadership and monitoring satisfies this purpose, because of the nurturing and close environment brought about through supportive leadership. Hence, a combination of leadership and monitoring can have a positive influence on perceptions of relational behaviors. Through closer and more supportive exchanges, the parties establish norms for relational exchange and work cooperatively (Heide and John, 1992; Lusch and Brown, 1996; Rodríguez and Wilson, 2002). This discussion suggests the following hypothesis. H3. Perceptions of supportive leadership moderate the negative relationship between perceptions of behavior-based monitoring and perceptions of relational behaviors, with monitoring having a less negative effect on relational behaviors under higher levels of leadership. 2.4. Relational behaviors and goal alignment Relational behaviors help to achieve a state of mutuality of interests between parties where they are at common purposes, values, and expectations (Jap, 1999). Parties that display relational behaviors should more easily reach acceptance and agreement, and have a common understanding of aims and objectives (Anderson and Weitz, 1989). Relational behaviors have positive social benefits that draw parties closer together, embedding them in a social framework that promotes cooperation (Stinchcombe, 1986; Thibaut, 1968). When parties embed socially, coordination is more likely, because of the willingness of the parties to be flexible and recognize joint interests, and to exchange information and act with high integrity (Mishra et al., 1998; Wright et al., 2001). Through the display of relational behavior, parties can develop greater respect and understanding for each other and begin to reach a common purpose. Hence, goal alignment requires relational behaviors as a prerequisite. The expectation is that leadership and monitoring do not have a direct influence on goal alignment. Instead, these attempts at exchange partner socialization will have an indirect effect on goal alignment through their impact on relational behaviors. Relational behaviors establish a social framework in which parties can reach a mutual understanding and are closely coordinated (Macneil, 1978, 1980). Under more formal and explicit forms of governance, explicit contracts can perform this coordinating role (Heide, 1994; Lusch and Brown, 1996). Under relational governance, however, goal alignment and

close coordination requires socialization efforts and the relational behaviors they promote. H4. An increase in perceptions of relational behaviors leads to an increase in perceptions of goal alignment. 2.5. Goal alignment and performance It is generally thought that parties to a relationship will achieve superior performance because of a willingness to display relational behaviors. Although the results from past studies are mixed, many scholars have argued in support of, and have attempted to establish, this link (e.g., Hewett and Bearden, 2001; Nevin, 1995; Siguaw et al., 2003). Establishing this link is important if relational governance is to be considered as an effective form of governing exchange relationships. The conflicting results from past research, however, allow for the possibility that the effect of relational behaviors on performance outcomes is indirect. As Fig. 1 shows, goal alignment mediates the positive relationship between relational behaviors and performance. To achieve desired outcomes, the conditions necessary for coordination must occur first (cf. Lusch and Brown, 1996). This implies alignment of goals. The role of relational governance is to socialize parties and to encourage the display of relational behaviors, which in turn, makes possible joint action and close coordination (Bello et al., 2003; Mishra et al., 1998; Wright et al., 2001). Hence, aligning the goals of the parties through relational behaviors achieves superior performance. The expectation is that a direct positive relationship exits between goal alignment and financial performance. H5. An increase in perceptions of goal alignment leads to an increase in perceptions of financial performance. 3. Method 3.1. Research setting The context for this empirical study is the construction industry. Major construction companies in Standard Industrial Classification Groups 15 and 16 were considered for study. Group 15 comprises contractors who construct buildings and Group 16 comprises contractors who perform heavy construction work other than buildings, such as constructing bridges, tunnels, roads, and pipelines. Many major construction companies come under both groups. The focal relationships are those between construction contractors and the subcontractors hired to complete specific and specialized parcels of work on contractors' projects. Relationships between contractors and subcontractors are much more prevalent in the construction industry than relationships between clients and primary contractors. The practice of engaging small- and medium-sized companies as subcontractors is commonplace among large construction firms. Subcontractors have a substantial influence over projects and their completion, because much of the technical expertise on major construction projects is provided by subcontractors. For these reasons it is

A.T. Stephen, L.V. Coote / Journal of Business Research 60 (2007) 285–295

worthwhile to consider this type of relationship for further study. Moreover, variation in the measured variables is expected because of the varying operational structures of the projects in which these relationships exist. 3.2. Measurement Referring to the conceptual model, data were collected on nine latent constructs (i.e., leadership support, leader feedback, behavior-based monitoring, flexibility, information exchange, solidarity, trustworthiness, goal alignment, and financial performance). Existing measures from past research in marketing were used wherever possible. All of the constructs are measured using multiple items. An exception is the financial performance construct, which is measured by a single indicator. All of the constructs have a reflective measurement structure. The measures reflect their posited constructs, but are not necessarily defining characteristics of the constructs (cf. Jarvis et al., 2003). Details of the actual measures used and their origins are provided in the Appendix. 3.3. Sampling frame A mail survey of project managers working for major construction firms across three major metropolitan areas was used to capture data on the variables of interest. Project managers run construction projects and are directly responsible for hiring and managing subcontractors. However, access to these individuals is not straightforward. All of the construction firms within a defined geographic area were screened for eligibility. These firms had to be major construction firms involved in larger-scale projects on a regular basis. A key industry association, company websites, trade press, and firsthand knowledge of these firms were used to compile an initial list of eligible firms. Many of these firms operate multiple, semi-autonomous business units that were eligible for inclusion. Thirty-eight firms or business units were considered suitable for the study. To recruit firms, letters were sent to vice presidents outlining the study's objectives and data collection requirements. Of the 38 firms contacted, 19 participated (50%). Vice-presidents in each participating firm were asked to supply the names of eligible project managers. In total, 143 project managers were included in the sampling frame. 3.4. Data collection Data were collected using mail surveys. Each of the 143 project managers received a mail questionnaire. They were asked to complete it with respect to a completed construction project for which they were managerially responsible. They were encouraged to consider their most recently completed project, so that performance data could be provided and to reduce the potential bias associated with memory effects. In addition to the questionnaire itself, each survey pack included a cover letter explaining the purpose of the study and instructions on how to complete the survey. A prepaid envelope was included with each questionnaire. The nature of the study was

289

described in general terms and no attempt was made to deceive or mislead respondents about its purpose. To further improve data quality, each questionnaire included a post hoc check of respondent knowledge on the subject matter covered in the questionnaire. The mean knowledge score for respondents was 6.25 (standard deviation [SD] = 1.14) on a seven-point scale, indicating that the respondents were qualified to participate. 4. Results 4.1. Sample profile Of the 143 mailed questionnaires, 76 completed questionnaires were returned for a response rate of 53%. All 19 firms that initially agreed to participate were represented in the sample. The sample is representative of the types of firms operating in this industry with respect to firm size and scope of operations. While the actual sample size is modest, the response rate is relatively high for this type of research and data collection strategy. A check of early versus late respondents revealed no significant differences, suggesting a minimal likelihood of non-response bias (cf. Armstrong and Overton, 1977). The respondents have an average of 15 years experience (SD = 9.3) in managing subcontractors in the construction industry. No questionnaires were discarded for incompleteness. There were only six cases for which data were missing. The missing data were completely random (Little's MCAR test: χ2 = 2.03, df = 325, p = 1.0). An EM procedure was used to replace missing values, thus maximizing the effective sample size (Allison, 2001). To check the equivalence of the sample with imputed values (n = 76) with the listwise sample (n = 70), ttests for equivalence of means and F-tests for homogeneity of variance were calculated. The results of both tests were nonsignificant for each measured variable, indicating that the sample with imputed values was acceptable for use in further analysis. 4.2. Measurement validation A procedure similar to Anderson and Gerbing's (1988) twostep model testing approach was followed to test the hypothesized model. The first step involved testing the measurement model. Exploratory factor analyses were estimated to check the dimensionality of each of the multi-item scales. All of the measures operated as expected, with the exception of a small number of measures for flexibility and information exchange. Several items were trimmed from these scales at this stage. The next step in measure validation involved confirmatory factor analysis (CFA) using LISREL 8.54 (Jöreskog and Sörbom, 1996). Given the sample size of 76 cases and the constraints on the ratio of sample size to estimated parameters, however, separate CFA models were estimated: one for the exogenous constructs (i.e., support, feedback, and monitoring), and one for the endogenous constructs minus the single-item financial performance measure (i.e., flexibility, information exchange, solidarity, trustworthiness, and goal alignment). Note

290

A.T. Stephen, L.V. Coote / Journal of Business Research 60 (2007) 285–295

that Monte Carlo studies of maximum likelihood estimation show no bias in parameter estimates in samples as small as 50 observations (Gerbing and Anderson, 1985). All of the λ coefficients for each item in the final CFA models achieved significance (p < .05). The significance of the λ coefficients indicates convergent validity. A small number of items were trimmed from the solidarity, trustworthiness, and leadership support scales, because their λ values were non-significant or because of low squared multiple correlations (< .50), or both. A final pool of 38 items plus the single indicator of financial performance were used to measure the theoretical constructs of interest. Scale reliability was further assessed using the final pool of measures as determined by the CFA models. Coefficient alphas and variance extracted estimates are reported in the Appendix. All scales achieve high reliability, as indicated by coefficient alphas and variance extracted estimates (Fornell and Larcker, 1981; Nunnally, 1978). Discriminant validity was assessed by comparing the squared correlations and average variance extracted estimates for each pair of latent constructs (Fornell and Larcker, 1981). Discriminant validity was achieved for all possible pairs. The fit of the final CFA models was then evaluated. The fit statistics for the CFA model of exogenous constructs are: χ2 (df, p) = 199.19 (101, .00), normed χ2 (χ2 / df) = 1.97, standardized root mean square residual (SRMR) = .09, goodnessof-fit index (GFI) = .73, comparative fit index (CFI) = .92, and incremental fit index (IFI) = .93. For the CFA of endogenous constructs, the fit statistics are: χ2 (df, p) = 276.68 (199, .00), normed χ 2 = 1.39, SRMR = .08, GFI = .72, CFI = .95, and IFI = .95. All of these indices suggest that both measurement models fit the data well (cf. Bentler, 1990; Bollen, 1989).

parameters. Modeling all 39 observed variables from the final CFAs would not have been particularly parsimonious. Moreover, model fit has been shown to be substantially improved when using composites as compared to including all items (Landis et al., 2000). Weights for the individual items were estimated using factor score regressions from the CFAs. This approach has the advantage of retaining measurement information from the original variables (Werts et al., 1978). Correlations, means, and standard deviations for all composites and the singleitem performance measure are reported in Table 1. As a final stage of measurement validation, a higher-order CFA for the supportive leadership construct (support and feedback) and the relational behaviors construct (flexibility, information exchange, solidarity, and trustworthiness) was estimated. Note that these higher-order constructs are represented in the final model with multiple measures as indicated by their respective dimensions. The higher-order factor model shows good overall fit to the sample data, and supports the hypothesized factor structure of the higher-order constructs (χ2 [df, p] = 11.16 [8, .19], χ2 / df = 1.40, SRMR = .07, GFI = .95, CFI = .98, IFI = .98). 4.4. Hypothesis tests A path analysis model with latent and observed variables, as Fig. 1 shows, tests the research hypotheses. A latent construct representing the interaction of leadership and monitoring appears in the model that Fig. 1 shows. The products of the composite indicators of leadership and monitoring were used as measures of this latent construct (Kenny and Judd, 1984; Ping, 1995). The hypothesized model has acceptable fit to the data in terms of overall model fit, and absolute and comparative fit measures are either within, or close to, accepted thresholds (χ 2 [df, p] = 105.66 [40, .00], χ 2 / df = 2.64, SRMR = .09, GFI = .76, CFI = .62, IFI = .63). Completely standardized path estimates and t-values are reported in Table 2.

4.3. Composite variables Composite variables were created from each set of multi-item scales, to maintain an acceptable ratio of cases to estimated

Table 1 Correlations, means, and standard deviations Observed variable a

1

2

3

4

5

6

7

8

9

10

11

1. Cost performance 2. Goal alignment 3. Flexibility 4. Information exchange 5. Solidarity 6. Trustworthiness 7. Leadership support 8. Leader feedback 9. Monitoring 10. Support · monitoring 11. Feedback · monitoring Mean b Standard deviation b

1.0 .21 .24 .05 .06 .19 .18 .07 .05 .16 .08 3.91 1.21

1.0 .40 .24 .28 .41 .27 .08 .24 .29 .15 4.92 1.02

1.0 .44 .62 .59 .37 .26 .10 .28 .19 4.79 1.21

1.0 .43 .45 .49 .47 .47 .54 .51 5.00 .94

1.0 .65 .40 .24 .36 .43 .32 4.73 1.15

1.0 .49 .30 .29 .45 .31 4.74 1.14

1.0 .58 .57 .91 .63 5.66 .89

1.0 .67 .70 .92 5.44 .91

1.0 .85 .90 5.86 .79

1.0 .85 33.54 8.17

1.0 32.33 8.35

Notes: r > .33 are significant at p < .01 (two-tailed), r > .26 are significant at p < .05 (two-tailed), and r > .23 are significant at p < .10 (two-tailed). a Variables are composites of measured variables retained after confirmatory factor analysis of measurement models. b Measured on seven-point scales, except for Support · monitoring (10) and feedback · monitoring (11), which are on 49-point scales (72).

A.T. Stephen, L.V. Coote / Journal of Business Research 60 (2007) 285–295

Supporting H1, the relationship between leadership and relational behaviors is positive and significant (γ31 = .25, t = 3.03, p < .01). Consistent with H2, the relationship between monitoring and relational behaviors is negative and marginally significant (γ32 = − .43, t = − 1.81, p < .10). The hypothesized relationship between the interaction of leadership and monitoring, and relational behaviors (H3) is strongly positive and significant (γ33 = .65, t = 3.90, p < .01). This pattern of results supports the prediction of a moderating effect of leadership on the negative relationship between monitoring and relational behaviors. As discussed subsequently, it appears that monitoring is more easily tolerated when supportive leadership is present. The relationship between relational behaviors and goal alignment is positive and significant (β23 = .45, t = 3.10, p < .01), as predicted by H4. Lastly, there is support for a positive relationship between goal alignment and financial performance (β12 = .21, t = 2.12, p < .05), as predicted by H5. The model explains 33.1%, 20.4%, and 4.3% of the variation in the endogenous constructs, relational behaviors, goal alignment, and financial performance, respectively. As a final check of the hypothesized process, the analysis includes estimating two rival models. First, a model without the relational behaviors construct was estimated, to demonstrate the mediating role of this construct. This model specifies direct effects from leadership and monitoring to goal alignment. This rival model has acceptable fit to the sample data (χ2 [df, p] = 79.33 [11, .00], χ2/df = 7.21, SRMR = .07, GFI = .74, CFI = .36, IFI = .37), although it does not fit the sample data better than the hypothesized model (Δχ2 [Δdf, p] = 26.33 [29, .61]). Moreover, the path estimates from leadership and monitoring to goal alignment are not supported (p > .05). Building from these formal tests, relational behaviors appear to have a mediating role, as the model of Fig. 1 represents. Second, the study includes estimating a rival model with a reverse model sequence. This rival model reverses the sequencing of relational behaviors and goal alignment (cf. Jap, 1999). This rival model has poor fit to the sample data (χ2 [df, p] = 147.58 [40, .00], χ2/ df = 3.69, SRMR = .17, GFI = .70, CFI = .58, IFI = .43), and it has significantly poorer fit to the data than the hypothesized model (Δχ2 [Δdf, p] = 41.92 [0, .00]). Finally, none of the path estimates in this rival model achieve significance (p > .05). These tests provide further evidence that the hypothesized process is a useful representation. Table 2 Standardized path estimates Path

Standardized estimate

Leadership → relational behavior Monitoring → relational behavior Leadership · monitoring → relational behavior Relational behavior → goal alignment Goal alignment → financial performance R2 − relational behavior R2 − goal alignment R2 − financial performance

.25 (3.03) a − .43 (−1.81) c .65 (3.90) a .45 (3.10) a .21 (2.12) b .33 .20 .04

a b c

Significant at p < .01 (one-tailed). Significant at p < .05 (one-tailed). Significant at p < .10 (one-tailed).

291

5. Discussion The relationship management literature is beginning to embrace a socialization approach to the management of marketing relationships. However, the nature of socialization efforts, their impact on relational behaviors, and their performance implications, if any, are not well documented. A basic premise of this paper is that relational governance can improve coordination and performance outcomes. However, the impact of relational governance on goal alignment is indirect. A core finding is that relational behaviors mediate the relationship between relational governance and goal alignment. Because the goals of relationship partners can be brought into close alignment through socialization efforts, relational governance can therefore be an effective solution to agency problems of hidden action. This is one of few studies to have shown an effective socialization approach for mitigating agency problems. Hence, the study achieves its main aim and extends the marketing literature, which has traditionally focused on formal approaches to relationship management. Empirical support for the hypothesized process indicates a positive financial impact of relational governance. 5.1. Implications for theory A key contribution of this study is to link socialization efforts with relational behaviors. This study shows that supportive leadership and behavior-based monitoring impact on exchange partners' perceptions of relational behaviors. However, the socialization efforts studied here are complex and there are systematic differences in their impacts on relational behaviors. Leadership that is characterized by efforts to support a partner and to provide valuable feedback promotes relational behaviors. Close monitoring of an exchange partner erodes perceptions of relational behaviors. However, the interaction of leadership and monitoring is crucial to establishing relational behaviors. Leadership and monitoring provide socialization opportunities by bringing parties closer together. Because supportive leadership is prosocial, however, monitoring is more readily tolerated when leadership is present. Supportive leadership may suppress the perceived loss of self-determination that relates sometimes with close monitoring (cf. Challagalla and Shervani, 1996; Ghoshal and Moran, 1996), while promoting a closeness of exchange that promotes relational behaviors. Hence, the negative aspects of monitoring can be offset by simultaneously engaging in supportive leadership. Recognizing the complex interplay between different socialization efforts is crucial to understanding the impacts of relational governance. This study shows that relational governance is a prerequisite for achieving coordination in relationships and desired performance outcomes. Parties can implement relational governance to establish relational behaviors and to improve cooperation. However, the positive impact of buyers' socialization efforts on goal alignment is mediated by relational behaviors. Relational behaviors are central to the process of achieving coordination through relational governance. Relational behaviors establish

292

A.T. Stephen, L.V. Coote / Journal of Business Research 60 (2007) 285–295

goal alignment, which in turn, makes possible the achievement of performance outcomes. Few past studies have linked governance approaches to the financial outcomes of parties in interfirm exchanges (cf. Heide, 1994). The initial findings presented here provide evidence that relational governance contributes in part to financial performance outcomes. 5.2. Managerial implications The implications for relationship managers are twofold. First, the study identifies managerial actions useful for establishing relational behaviors. Second, the study offers evidence that relational governance can be strategically valuable. Leadership characterized by support and feedback is important for developing relational behaviors. Monitoring is also important, however, the impact of monitoring appears to be more complex than previously thought. Monitoring has the potential to undermine a system of relational governance. Parties must ensure that monitoring does not erode relational behaviors, by combining monitoring with supportive leadership. Relational behaviors, coordination, and performance outcomes, can be achieved through the interplay of leadership and monitoring. A second implication is that skills in managing relational exchanges are a strategically valuable resource. Recall that under relational governance, the parties engage in a social process and derive complex non-economic satisfactions from their exchanges. The strong relational bonds underpinning relational exchanges may be difficult to break. Moreover, managing relational exchanges involves tacit knowledge. These factors make strategic partnerships difficult for competitors to imitate (cf. Barney, 1996; Hunt and Morgan, 1994). Hence, it is possible that sustainable competitive advantages could accrue to firms that possess superior relational governance capabilities. A firm's skill and expertise in managing relational resources, as reflected in the implementation of a socialization-based approach to governance, is the key to leveraging relational resources. 5.3. Limitations Consider several limitations when reviewing the findings. One potential limitation relates to the size of the sample (n = 76). Although there is no evidence of non-response bias, the sample may not be representative of the target population. Much effort was directed at encouraging participation in the study. Yet, a sample much larger than 76 respondents could not reasonably be expected. Recall that the sampling frame comprised of only 143 individuals. A second limitation is that a single industry was selected for study (i.e., major construction). The results may not generalize to other settings, and attempts to generalize the results to other contexts would be difficult even with a larger sample. However, restricting the sample to a single context has the advantages of controlling for extraneous sources of variation and simplifying the sampling procedures. A third possible limitation is that only the perspectives of contractors were examined. The method of relying on single

informant reports from one side of a dyadic exchange gives rise to two possible limitations in particular (cf. Kumar et al., 1993). The first issue is the problem of key informant competency. A check of informant quality revealed that the respondents were knowledgeable and qualified to participate. Thus, key informant quality is not likely to be a problem. The second issue is the potential problem of differences in viewpoints among parties to a relationship. As part of the initial data collection process, contractors were asked to identify subcontractors and provide information for possible contact. Sampling subcontractors yielded 26 responses, and a measure of agreement was calculated for each of the 26 paired responses (Cohen, 1988). This analysis reveals a positive and significant relationship between contractors and subcontractors' perspectives (r = .10, p < .05). Hence, the potential bias of reporting only on contractors' perspectives is likely to be minimal. A final limitation relates to measurement of the financial performance construct. The single-item measure for this construct was industry-specific and captured cost performance. A possible consequence of using a single-item measure is attenuation in the path from goal alignment to financial performance. Thus, the single measure makes for a stronger test of the relationship between coordination and financial performance. However, there are other dimensions of financial performance that could have been captured to improve the validity and reliability of this measure (e.g., revenue and profitability). Additionally, the performance measure captures project performance, and not relationship performance. This might explain the small proportion of explained variance in the performance measure. 5.4. Suggestions for further research First, further research might explore other consequences of relational governance. For example, other forms of coordination need to be considered, such as the alignment of tasks as well as goals. This is particularly important given that parties often have joint and overlapping responsibilities in marketing relationships (Heide, 1994). Moreover, further work is required to confirm the positive impact of relational governance on performance outcomes. This study provides an initial finding to support this indirect relationship, but much further work is needed to understand how socialization efforts impact relationship and firm performance. A second avenue for further research is to consider other conceptualizations of relational behaviors. The relational behaviors construct in the model of Fig. 1 has foundations in the literature on relational social norms (Macneil, 1978, 1980). However, other conceptualizations might be developed in future research. For example, a hierarchy of relational behaviors might exist in which some dimensions, such as trustworthiness, are more fundamental than others. Modeling the interrelationships among the relational behaviors is worthy of further consideration. Additionally, leadership and monitoring are by no means exhaustive of potential socialization efforts. One useful avenue would be to pursue a finer-grained analysis of the role played by leadership in fostering relational behaviors. A starting point is to

A.T. Stephen, L.V. Coote / Journal of Business Research 60 (2007) 285–295

consider the components of leadership such as supportiveness, encouragement, guidance, and different types of feedback (i.e., positive and negative). Finally, much work remains in extending the unit of analysis to networks of multiple buyer–supplier relationships. Considering single dyads in isolation is a simplification, yet very few studies have considered networks of marketing relationships (cf. Wathne and Heide, 2004). Research in this direction might consider feedback among buyers and suppliers, as both parties can contribute to socialization efforts. Additionally, a network approach may generate new insights into the behavior of exchange participants, and the performance implications of relational governance. The insights from this initial study might provide a basis for theorizing about how systems of relational governance operate in networks of marketing relationships. Acknowledgement This research was completed while the first author was an undergraduate at the University of Queensland working under the supervision of the second author. The authors thank the editor, associate editor, and the anonymous JBR reviewers for their helpful comments and suggestions. The authors also acknowledge the assistance of the Queensland Major Contractors Association in helping to collect data, and input from Dr Robert Day (civil engineering) at the University of Queensland. Appendix A. Measurement items

Leadership support (α = .87, vee = .64)

House (1971), Netemeyer et al. (1997)

1. We tried to make things easier and less complicated for the subcontractor. 2. We treated the subcontractor as an equal. 3. We were supportive of the subcontractor's work activities. 4. We treated the subcontractor with respect. Leader feedback (α = .88, vee = .54) Challagalla and Shervani (1996), Jaworski and MacInnes (1989) 1. We provided the subcontractor with feedback about its work as it was completing its tasks. 2. We made sure that we gave performancerelated feedback to the subcontractor throughout its involvement. 3. The feedback given to the subcontractor was useful in helping it to be more productive. 4. We informed the subcontractor about whether or not its progress lived up to our expectations. 5. We gave the subcontractor suggestions and ideas relating to its work that it might have found helpful. 6. The feedback we gave to the subcontractor about its work was constructive.

Behavior-based monitoring (α=.88, vee=.56) 1. We made sure the subcontractor knew what it had to achieve. 2. We maintained close contact with the subcontractor as it worked. 3. We were up-to-date and well informed of activities undertaken by the subcontractor. 4. We spent a lot of time observing the subcontractor as it worked. 5. We were interested in how the subcontractor was actually completing its work. 6. We closely monitored the job-related behaviors of the subcontractor throughout the project. Flexibility (α = .87, vee = .65) 1. We were able to make adjustments in our relationship without significant disputes, conflicts, or uncooperative behaviors. 2. When unexpected situations arose, we preferred to work out a new arrangement rather than holding each other to the original arrangement. 3. When unexpected events occurred, both parties were open to modifying previous agreements. 4. The subcontractor and our firm were flexible in response to requests made by each other throughout the course of the project. Information exchange (α = .85, vee = .57) 1. Both parties were willing to provide proprietary information if it helped each other. 2. The parties kept each other informed about any events or changes that may have affected them. 3. Information was exchanged willingly. 4. Each party provided proprietary information that was helpful to the other. Solidarity (α = .88, vee = .66) 1. Problems that arose in the course of the project were treated as joint rather than individual responsibilities. 2. Both parties were committed to improvements that may have benefited the project as a whole and not only the individual parties. 3. Problems were solved jointly, rather than pushing problems solely onto one party. 4. The responsibility for making sure that the relationship worked for all parties was shared jointly. Trustworthiness (α = .92, vee = .71)

293 Oliver and Anderson (1994)

Heide and John (1992)

Heide and John (1992)

Heide and John (1992)

Doney and Cannon (1997), Morgan and Hunt (1994)

1. Both parties trusted each other throughout the course of the project. 2. We had a trust-based relationship with our subcontractor. 3. We were confident in relying upon our subcontractor. 4. We felt that our subcontractor and our firm acted with integrity and honesty towards each other throughout the course of the project. (continued on next page)

294

A.T. Stephen, L.V. Coote / Journal of Business Research 60 (2007) 285–295

Goal alignment (α = .92, vee = .66) Jap (1999) 1. We had very similar project-related goals. 2. We had compatible project-related goals. 3. We supported each other's objectives. 4. We generally agreed upon project-related goals. 5. Our attitudes towards what needed to be achieved were very similar. 6. Our goals were in close alignment. Project financial performance Alarcón and Ashley (1996) 1. Final project budget/cost.

References Alarcón Luis F, Ashley David B. Modeling project performance for decision making. J Constr Eng Manag 1996;122(3):265–73. Allison Paul D. Missing data: series on quantitative applications in the social sciences. Newbury Park: Sage Publications; 2001. Anderson James C, Gerbing David W. Structural equation modeling in practice: a review and recommended two-step approach. Psychol Bull 1988;103 (3):411–23. Anderson James C, Narus James A. A model of distributor firm and manufacturer firm working partnerships. J Mark 1990;54(1):42–58. Anderson Erin, Oliver Richard L. Perspectives on behavior-based versus outcome-based salesforce control systems. J Mark 1987;51(4):76–88. Anderson Erin, Weitz Barton. Determinants of continuity in conventional industrial channel dyads. Mar. Sci 1989;8(4):310–23. Armstrong Scott J, Overton Terry S. Estimating nonresponse bias in mail surveys. J Mark Res 1977;14(3):396–402. Arrow Kenneth J. The Economics of Agency. In: Pratt J, Zeckhauser R, editors. Principals and agents: the structure of business. Boston: Harvard Business School Press; 1985. p. 37–51. Barney Jay B. The resource-based theory of the firm. Org Sci 1996;7 (5):469–501. Bello Daniel C, Chelariu Cristian, Zhang Li. The antecedents and performance consequences of relationalism in export distribution channels. J Bus Res 2003;56(1):1–16. Bentler Peter M. Comparative fit indices in structural models. Psychol Bull 1990;107(2):238–46. Bergen Mark, Dutta Shantanu, Walker Jr Orville C. Agency relationships in marketing: a review of the implications and applications of agency and related theories. J Mark 1992;56(3):1–24. Bollen Kenneth A. Structural equations with latent variables. New York: John Wiley and Sons; 1989. Brehm Jack W. A theory of psychological reactance. NewYork: Academic Press; 1966. Challagalla Goutam N, Shervani Tasadduq A. Dimensions and types of supervisory control: effects on salesperson performance and satisfaction. J Mark 1996;60(1):89–105. Cohen Jacob. Statistical power analysis for the behavior sciences. 2nd ed. Hillsdale, NJ: L. Erlbaum Associates; 1988. Doney Patricia M, Cannon Joseph P. An examination of the nature of trust in buyer–seller relationships. J Mark 1997;61(2):35–51. Dwyer F Robert, Schurr Paul H, Oh Sejo. Developing buyer–seller relationships. J Mark 1987;51(2):11–27. Eisenhardt Kathleen M. Agency theory: an assessment and review. Acad Manage Rev 1989;14(1):57–74. Fornell Claes, Larcker David F. Evaluating structural equation models with unobservable variables and measurement error. J Mark Res 1981;18(1):39–50. Gerbing David W, Anderson James C. The effects of sampling error and model characteristics on parameter estimation for maximum likelihoodconfirmatory factor analysis. Multivar Behav Res 1985;20(3):255–71. Ghoshal Sumantra, Moran Peter. Bad for practice: a critique of the transaction cost theory. Acad Manage Rev 1996;21(1):13–47. Heide Jan B. Interorganizational governance in marketing channels. J Mark 1994;58(1):71–85.

Heide Jan B, John George. Do norms matter in marketing relationships? J Mark 1992;56(2):32–44. Hewett Kelly, Bearden William O. Dependence, trust, and relational behavior on the part of foreign subsidiary marketing operations: implications for managing global marketing operations. J Mark 2001;65(4):51–66. House Robert J. A path-goal theory of leader effectiveness. Admin. Sci. Quar. 1971;16(3):321–39. Hunt Shelby D, Morgan Robert M. The comparative advantage theory of competition. J Mark 1994;59(2):1–15. Jap Sandy D. Pie-expansion efforts: collaboration processes in buyer–supplier relationships. J Mark Res 1999;36(4):461–75. Jarvis Cheryl Burke, Mackenzie Scott B, Podsakoff Philip M. A critical review of construct indicators and measurement model misspecification in marketing and consumer research. J Consumer Res 2003;30 (2):199–218. Jaworski Bernard J. Toward a theory of marketing control: environmental context, control types, and consequences. J Mark 1988;52(3):23–39. Jaworski Bernard J, MacInnes Deborah J. Marketing jobs and management controls: toward a framework. J Mark Res 1989;26(4):406–19. Jöreskog Karl G, Sörbom Dag. LISREL 8: user's reference guide. Chicago: Scientific Software International; 1996. Kenny David, Judd Charles M. Estimating the nonlinear and interactive effects of latent variables. Psychol Bull 1984;96(4):201–10. Kumar Nirmalya, Stern Louis W, Anderson James C. Conducting interorganizational research using key informants. Acad Manage J 1993;36 (6):1633–51. Landis Ronald S, Beal Daniel J, Tesluk Paul E. A comparison of approaches to forming composite measures in structural equation models. Org Res Methods 2000;3(2):186–207. Lusch Robert F, Brown James R. Interdependency, contracting, and relational behavior in marketing channels. J Mark 1996;60(3):19–38. Macneil Ian. Contracts: adjustment of long-term economic relations under classical, neoclassical, and relational contract law. Northwest Univ Law Rev 1978;72(6):854–905. Macneil Ian. The new social contract: an inquiry into modern contractual relations. New Haven: Yale University Press; 1980. Mishra Debi Prasad, Heide Jan B, Cort Stanton G. Information asymmetry and levels of agency relationships. J Mark Res 1998;35(3):277–95. Morgan Robert M, Hunt Shelby D. The commitment–trust theory of relationship marketing. J Mark 1994;58(3):20–38. Netemeyer Richard G, Boles James S, McKee Daryl O, McMurrian Robert. An investigation into the antecedents of organizational citizenship behaviors in a personal selling context. J Mark 1997;58 (3):20–38. Nevin John R. Relationship marketing and distribution channels: exploring fundamental issues. J. Acad. Mark. Sci. 1995;23(4):327–34. Noordewier Thomas G, John George, Nevin John R. Performance outcomes of purchasing arrangements in industrial buyer–vendor relationships. J Mark 1990;51(4):80–93. Nunnally Jum C. Psychometric theory. New York: McGraw–Hill; 1978. Oliver Richard L, Anderson Erin. An empirical test of the consequences of behavior- and outcome-based sales control systems. J Mark 1994;58 (4):53–67. Ouchi William G. A conceptual framework for the design of organizational control mechanisms. Manage Sci 1979;25(10):833–48. Ping Jr Robert A. A parsimonious estimating technique for interaction and quadratic latent variables. J Mark Res 1995;32(3):336–47. Rodríguez Carlos M, Wilson David T. Relationship bonding and trust as a foundation for commitment in U.S.–mexican strategic alliances: a structural equation modeling approach. J Int Mark 2002;10(4):53–76. Siguaw Judy A, Baker Thomas L, Simpson Penny M. Preliminary evidence on the composition of relational exchange and its outcomes: the distributor perspective. J Busi Res 2003;56(4):311–22. Stinchcombe Arthur L. Norms of exchange. In: Stinchcombe AL, editor. Stratification and organization: selected articles. Cambridge: Cambridge University Press; 1986. p. 231–67. Thibaut John W. The development of contractual norms in bargaining: replication and variation. J Confl Resolut 1968;12(1):102–12.

A.T. Stephen, L.V. Coote / Journal of Business Research 60 (2007) 285–295 Wathne Kenneth H, Heide Jan B. Relationship governance in a supply chain network. J Mark 2004;68(1):73–89. Werts CE, Rock DR, Linn RL, Jöreskog KG. A general method of estimating the reliability of a composite. Educ. Psychol. Meas. 1978;38 (4):933–8.

295

Willer David, Anderson Bo, editors. Networks, exchange, and coercion: the elementary theory and its applications. New York: Elsevier; 1981. Wright Peter, Mukherji Ananda, Kroll Mark J. A reexamination of agency theory assumptions: extensions and extrapolations. J Socio-Econ 2001;30 (5):413–29.