The unavoidable balance between trust and control How the interaction between outcome, behavior and social control and goodwill and competence trust differ in inter-organizational relationships

(1)

The unavoidable balance between trust and control

How the interaction between outcome, behavior and social control and

goodwill and competence trust differ in inter-organizational relationships

Master thesis Business Administration Organizational and Management Control

University of Groningen Faculty of Economics and Business

Department of Accounting January 2015 Aukeline Wind S1873016 Supervisors: A. Rehman Abbasi Dr. H.J. van Elten

(2)

A.C. Wind – The unavoidable balance between trust and control | 2

ABSTRACT

To date, there is much controversy in literature about the relationship between trust and control. Some researchers suggest that trust is negatively affected by control, while others affirm the reverse. Still others claim that trust and control are substitutes, while they can also be complements. Therefore, this study investigates the interactions between goodwill and competence trust on the one hand, and outcome, behavior and social control on the other hand. Using survey data of 104 respondents, the interactions between trust and control are estimated. It was found that trust and the different types of control vary in their interactions. Furthermore, these interaction are context-specific as well.

(3)

INTRODUCTION

Recently, a growing number of companies has been collaborating with one or multiple organization(s) due to the challenges of globalization and increasing competition (Aulakh, Kotabe, & Sahay, 1996; Lai, Tian, & Huo, 2012; Langfield-Smith & Smith, 2003; Munoz & Medina-Munoz, 2004; Ring & van de Ven, 1992), i.e. they struggle to develop and maintain the competences necessary by themselves to compete successfully (Langfield-Smith & Smith, 2003). Simultaneously, there is an increasing attention from academics regarding inter-organizational relationships (Dekker, 2004; Nooteboom, Berger, & Noorderhaven, 1997; Ring, & van de Ven, 1992). Existing literature already found several reasons that can be imagined for engaging in such a collaboration such as creating competitive advantages (Ding, Dekker, & Groot, 2013; Dyer & Singh, 1998), transferring knowledge (Dyer & Singh, 1998) and creating relational rents (Dyer & Singh, 1998) which will be higher than the sum of the individual rents (Eisenhardt & Schoonhoven, 1996). Although these benefits sound attractive, some risks may be involved in such relationships. On the one hand, performance risk arises by unachieved objectives while there is a satisfactory cooperation (Das & Teng, 2001b). On the other hand, relational risk is caused by unsatisfactory relationships (Das & Teng, 2001b), such as opportunistic behavior which is the search for self-interest by exceeding implicit and/or explicit promises through cheating (Morgan & Hunt, 1994) . Since these risks may negatively influence those relationships and even cause high failure rates (Das & Teng, 1998, 2001b; Ireland, Hitt, & Vaidyanath, 2002; Langfield-Smith & Smith, 2003; Lunnan & Haugland, 2008) of 30 to 70 percent (Lunnan & Haugland, 2008), a good basis between the parties involved is necessary (Doz, 1996; Kanter, 1994). A lot of literature has already been written about reducing risks such as (perceived) opportunistic behavior by creating the opposite: a cooperative relationship which is 'the willingness of a partner firm to pursue mutually compatible interests in the alliance rather than act opportunistically' (Das & Teng, 1998, p. 492) and which can be based on formal structures as regulations and contracts (D. Malhotra & Murnighan, 2002) as well as informal structures as trust and commitment (Dyer & Singh, 1998). This partner cooperation is crucial for the success of collaboration, but it is difficult to achieve (Das & Teng, 1998).

(5)

A.C. Wind – The unavoidable balance between trust and control | 5 whereas others suggest that control positively affects trust (Coletti, Sedatole, & Towry, 2005). Some argued that trust and control are substitutes of each other (Costa & Bijlsma-Frankema, 2007; Das & Teng, 1998; Zaheer & Venkatraman, 1995), while others argued that they complement one another (Coletti et al., 2005; Sanchez, Velez, & Araujo, 2012; Tomkins, 2001). Thus, existing literature disagrees about the connection between trust and control. However, these studies have only partly investigated the relationship. For instance, Malhotra and Murnighan (2002) investigated the role of a specific form of formal control (i.e. contracts) on interpersonal trust. Furthermore, Coletti et al. (2005) examined the relationship between trust and control, the mediating role of control-induced cooperation included. In addition, although Sanchez et al. (2012) argued that the relationship depends on, among others, the type of control and the specific control tools, they did not consider the different types of trust, i.e. goodwill and competence trust. While all this research has considered the relationship between trust and control, they have inadequately determined how and why they are substitutes and/or complements (Klein Woolthuis, Hillebrand, & Nooteboom, 2005). In addition, they looked at different aspects of the relationship between trust and control, not considering all the types of trust and control (Langfield-Smith & Smith, 2003), which might explain the conflicting results. In 2001b, Das and Teng already recognized this limitation and they proposed an integrated framework about the relationship between trust, control and risk in strategic alliances, including both; the three types of control (e.g. behavior control, outcome control and social control as proposed by Ouchi (1977, 1979)) as well as the two often mentioned types of trust (e.g. goodwill trust (Ring & van de Ven, 1992) and competence trust (Das & Teng, 2001b; Dekker, 2004; Langfield-Smith & Langfield-Smith, 2003)). However, to the best of my knowledge this framework has not been tested as a whole.

(6)

A.C. Wind – The unavoidable balance between trust and control | 6 to reduce opportunistic behavior, they do not necessarily have similar effects on each other. For instance, the interaction between behavior control and goodwill trust might be different than the interaction between behavior control and competence trust. Therefore, it is interesting for academics to know whether there is any difference in effect of those control mechanisms and trust types, what that difference is and what consequences that difference has. Third, it tries to understand the effects of the different purposes of control, as suggested by Dekker (2004), on the interaction between control and trust in inter-organizational relationships.

From a practical point of view, this study is interesting because it provides more detailed insights in the interaction between trust and control types. These insights can be used to anticipate on the risks of collaboration and they might reduce the high failure rates that accompany such relationships.

The remainder of this paper is organized as follow: first, existing literature about the key concepts (i.e. inter-organizational relationships, trust, control and their interactions) is reviewed leading to the development of a framework for understanding the relationships between trust and control. Next, the methodology and data are described. Then the data will be analyzed and interpreted. The paper concludes with a discussion, theoretical and managerial implications and suggestions for further research.

LITERATURE REVIEW

This section discusses existing literature regarding inter-organizational relationships, trust and control. Once those concepts have been clarified, their interrelationships will be explained resulting in the development of hypotheses. This section summarizes with a conceptual model in which all relationships are shown.

Inter-organizational relationships

(7)

A.C. Wind – The unavoidable balance between trust and control | 7 negotiation, monitoring, etc. – should be chosen (Williamson, 1979) in order to maximize firm performance (Crook, Combs, Ketchen, & Aguinis, 2013). However, this theory is recently criticized as being a one-sided perspective in explaining inter-organizational collaborations (Dekker, 2004; Langfield-Smith & Smith, 2003). Although transaction cost theory is useful in explaining the desired form of governance structure in inter-organizational relationships, relying heavily on this prediction might be insufficient due to its lack of dynamics, its lack of identification of the diversity of relationship types and goals (Dekker, 2004) and its lack of consideration of the social context (Dekker, 2004; Langfield-Smith & Smith, 2003).

The RBV argues that relationships are driven by resource needs and opportunities (Eisenhardt & Schoonhoven, 1996). The choice of structure is determined by resource profiles (Chen & Chen, 2003; Das & Teng, 2000) since organizations will not only acquire their partners' valuable resources but also protect their own (Chen & Chen, 2003). This need for managing and protecting valuable resources represents two purposes of control, namely appropriation concerns and coordination requirements (Dekker, 2004).

According to Medina-Munoz and Medina-Munoz (2004), relationships have become an indispensable part of the business environment due to increasing globalization, competition and uncertainty. Partially as a consequence of these developments it became more difficult to supervise these relationships (Costa & Bijlsma-Frankema, 2007), thus resulting in a downside for these relationships: the existence of performance risk – e.g. unachieved objectives in a satisfactory cooperation – and relational risk – e.g. unsatisfactory relationships such as opportunistic behavior (Das & Teng, 2001b) which can cause high failure rates of inter-organizational relationships (Das & Teng, 1998, 2001b; Ireland et al., 2002; Langfield-Smith & Smith, 2003; Lunnan & Haugland, 2008). As a result, risk management is very important for the success of those inter-organizational relationships (Das & Teng, 2001b; Ring & van de Ven, 1992).

Opportunistic behavior

(8)

A.C. Wind – The unavoidable balance between trust and control | 8 that the higher the potential financial benefits will be, the more a self-interested partner is inclined to behave opportunistically (Das & Rahman, 2010).

Cooperation

In contrast, confidence in partner cooperation is 'the willingness of a partner firm to pursue mutually compatible interests in the alliance rather than act opportunistically' (Das & Teng, 1998, p. 492) and can preclude opportunistic behavior. They argue that it is based on either trust or control or a combination of them. It should be noted, however, that although the importance of trust in relationships is acknowledged by many researchers (Altinay & Brookes, 2012; Costa & Bijlsma-Frankema, 2007; Das & Teng, 1998, 2001b; Sengün & Wasti, 2011; Vlaar, Van den Bosch, & Volberda, 2007), a certain amount of control is required as well (Klein Woolthuis et al., 2005). Since both appear to be important during cooperation, it might be insufficient to base a relationship solely on trust or control. Thus, this will result in an unavoidable balance between control and trust (Inkpen & Currall, 2004). Therefore, both trust and control will be considered in this paper.

Trust

A lot of literature in many disciplines has considered the concept of trust. However, as noted by Mayer, Davis and Schoorman (1995) and Das and Teng (2001) studying trust encompasses several problems such as the absence of a clear definition. Although there is no uniform definition of trust, some common characteristics are available; for instance a certain degree of dependence (Knemeyer & Murphy, 2005), accepting vulnerability (Mayer et al., 1995; Rousseau, Sitkin, Burt, & Camerer, 1998), positive expectations (Coletti et al., 2005; Costa & Bijlsma-Frankema, 2007; Das & Teng, 2001b; D. Malhotra & Murnighan, 2002; Mayer et al., 1995) and belief in another party (Emsley & Kidon, 2007; Knemeyer & Murphy, 2005). These elements are summarized in the often cited definition of Mayer et al. (1995, p. 712), who defined trust as ‘the willingness of a party to be vulnerable to the actions of another party based on the expectation that the other will perform a particular action important to the trustor, irrespective of the ability to monitor or control that other party’. From a similar point of view, Rousseau et al. (1998, p. 395) defined trust as ‘a psychological state comprising the intention to accept vulnerability based upon positive expectations of the intentions or behavior of another’.

(9)

A.C. Wind – The unavoidable balance between trust and control | 9 for breach of trust exceeds any potential benefits from opportunistic behavior’ (Rousseau et al., 1998, p. 398). Calculus-based trust is ‘based on rational choice – characteristic of interactions based upon economic exchange’ (Rousseau et al., 1998, p. 399). Third, relational trust ‘derives from repeated interactions over time between trustor and trustee’ (Rousseau et al., 1998, p. 399). Relational trust can be compared with the affect-based trust and cognition-based trust of McAllister (1995) which consists of emotional relationships between individuals which are based on insights in the partners’ motivations and choices about who we want to trust, respectively (McAllister, 1995). Finally, institutional trust ‘can ease the way to formulating both calculus-based trust, and relational trust’ (Rousseau et al., 1998, p. 400).

Others suggested that trust is a two-dimensional concept consisting of competence-based trust on the one hand, and goodwill trust on the other hand (Costa & Bijlsma-Frankema, 2007; Emsley & Kidon, 2007; Nooteboom et al., 1997; Sengün & Wasti, 2011). Competence-based trust – also referred to by ability (Mayer et al., 1995) – is the confidence that a partner is able to perform as expected (Nooteboom et al., 1997; Xie & Peng, 2011). On the other hand, goodwill trust – also denoted as integrity (e.g. Mayer et al., 1995) – is the confidence in the others’ intentions and integrity (Das & Teng, 2001b). According to Nooteboom (1996, p. 990) ‘trust may concern a partner’s ability to perform according to agreements (competence trust), or his intentions to do so (goodwill trust)’. This two-dimensional view of trust will be used in this paper.

Control

Like trust, the concept of control is commonly investigated. According to Ouchi (1977), a control system includes a number of conditions regarding reliability and validity that should be met on the one hand, and the process of monitoring and evaluating performance on the other hand. Thus, performances are compared to certain desired standards. As argued by Costa and Bijlsma-Frankema (2007), control refers to a regulatory process concerning the behavior of others to achieve a desired purpose and thereby making their behavior more predictable (Das & Teng, 2001b). More generally, control is a process of monitoring another party in order to achieve a certain purpose.

(10)

A.C. Wind – The unavoidable balance between trust and control | 10 monitoring the achievement of the specified desired results (Emsley & Kidon, 2007) in which reliable measurements are required (Ouchi, 1977). Both types of formal control can be applied in a market or bureaucracy as suggested by Ouchi (1979), which require high levels of information, e.g. explicit rules about behavior or output levels, and low levels of social requirements, i.e. a norm of reciprocity is sufficient.

On the contrary, informal controls are unwritten control mechanisms that are not necessarily supporting organizational objectives (Medina-Munoz & Medina-Munoz, 2004). Informal controls are also referred to as social control, which are the shared norms, values and believes (Costa & Bijlsma-Frankema, 2007; Emsley & Kidon, 2007; Medina-Munoz & Medina-Munoz, 2004). Informal controls can be implemented in a clan type of control (Ouchi, 1979) resulting in lower levels of information, but higher levels of social requirements relative to the market and bureaucracy since it is (almost) impossible to evaluate and monitor those attributes. This means that clans are especially based on commitment and consensus about appropriate behavior (Ouchi, 1979). This three-dimensional perspective of control – i.e. behavioral, outcome and social control – will be used in this paper.

Interactions between trust and control

Existing literature about trust and control in (inter-organizational) relationships is contradictory and paradoxical. Some researchers suggest that trust is just a control mechanism (Bradach & Eccles, 1989), being the main aspect of social control (Ring & van de Ven, 1992) or being seen as self-control (Merchant, 1985). However, as argued by Das and Teng (1998), trust cannot be a control mechanism in this way since the behavior of others is not monitored.

(11)

A.C. Wind – The unavoidable balance between trust and control | 11 Still others suggest that control and trust are complements (Coletti et al., 2005; Fryxell, Dooley, & Vryza, 2002; Poppo & Zenger, 2002; Sanchez et al., 2012; Tomkins, 2001) meaning that control not necessarily affect trust in a negative way (Sanchez et al., 2012), and might even be positive (Coletti et al., 2005), because ‘objective rules and clear measures help to institute a ‘track record’ for people who do their jobs well’ (Sitkin, 1995). In addition, Sanchez et al. (2012) suggest that control will help organizations to construct trust. Vice versa, trust enhances the use of control mechanisms through the provision of timely and accurate information (Edelenbos & Eshuis, 2011). ‘In cases where people do not trust each other, they will try to minimize their vulnerability vis-à-vis the other and evade control by the other’ (Edelenbos & Eshuis, 2011, p. 655).

Recently, a fourth stream purposes that the relationship between trust and control is not static and more complex in nature than suggested by the previous research streams (Das & Teng, 1998, 2001b; Dekker, 2004; Inkpen & Currall, 2004; Klein Woolthuis et al., 2005; Vlaar et al., 2007; Vosselman & Meer-Kooistra, 2009). According to Vosselman and Meer-Kooistra (2009) this relationship is an interaction meaning that control influences trust, while simultaneously trust influences control. Whether they substitute or complement and harm or strengthen each other might depend on the level of transaction risks, whether it is a direct or moderating effect and the aim of the control mechanisms (Dekker, 2004). Following the reasoning of this fourth stream of researchers, and especially Das and Teng (1998, 2001), this paper suggests that the interaction of control and trust can differ across situations. In the following paragraphs, the interactions between trust and control will be described more specifically using the framework of Das and Teng (2001) as a starting point. However, only the interactions between control and trust will be considered meaning that their relationship to risk is beyond the scope of this research. Based on the findings of Sengün and Wasti (2007) who partly tested this framework in the Turkish drug distribution sector as well as based on other literature, some adjustments in the hypotheses were made. Furthermore, a time element was included due to assumption that this relationship is dynamic in nature (Das & Teng, 1998).

Influence of control on trust

(12)

A.C. Wind – The unavoidable balance between trust and control | 12 task in the best manner possible. In line with this proposition, Sengün and Wasti (2007) found that outcome controls harm goodwill trust and competence trust. However, this conclusion should be interpreted with caution, because it is limited to a single industry in a single country and a single type of inter-organizational relationships.

From a similar point of view, Aulakh et al. (1996) suggest that the use of outcome controls moves the collaboration risk to the partner firm, because these standards are established for them (Eisenhardt, 1989). However, when they rely too much on those quantitative performance standards, conflicts and tensions may arise leading to a deterioration of the relationship between the parties involved (Hopwood, 1972). Thus, the use of outcome controls may harm the long-term objectives of the relationship meaning outcome control and goodwill and competence trust are more or less substitutes. While Aulakh et al. (1996) did not find a significant effect in their study, this assumption is confirmed by other researchers (for instance Ring & van de Ven, 1992; Zaheer & Venkatraman, 1995).

Regarding behavior control, Das and Teng (2001) proposed that both goodwill and competence trust will be weakened by the use of behavior control mechanisms using the same logic as used for outcome control. However, this proposition has not been tested empirically. From a different perspective, Aulakh et al. (1996) have suggested that the use of behavior controls requires active participation whereby uncertainty will be reduced through higher formalization resulting in a positive effect of control on trust. In accordance with this logic, Poppo and Zenger (2002) argued that both partners should mutually establish and commit to processes in formal control (i.e. behavior and outcome control) to cope with unforeseen situations, punishment for noncompliance etc. Thus, this process of generating contracts positively influences future performances through specification of certain boundaries. While Poppo and Zenger (2002) found support for their logic that formal control enhances trust, Aulakh et al. (1996) did not find a significant effect in their study. Thus, the actual impact of behavior control is still uncertain.

(13)

A.C. Wind – The unavoidable balance between trust and control | 13 insufficient level of trust available between the parties (Doz, 1996). Thus, trust building might be enhanced by rules and objectives which are apparent to and accepted by the parties involved (Inkpen & Currall, 2004). This may lead to an increase in competence trust since they built insurance in each others’ performance, so that they can check each other. In addition, goodwill trust will be increased as well, because partners’ behavior becomes more predictable through legalization (Costa & Bijlsma-Frankema, 2007). In line with this argumentation, it will be tested if the assumption that both behavior and outcome control mechanisms enhance trust-building in the beginning of a relationship and in short relationships.

However, its effects might be diminished in more enduring relationships. As noted by Fryxell et al. (2002), the benefits of formal controls that appear in the beginning of a relationship decline rapidly due to changing needs and expectations. In that case, partners may become more uncertain about the others’ intentions meaning a decrease in goodwill trust. Furthermore, competence trust might be reduced as well since both are uncertain about each other’s required skills for the changing needs. Accordingly, it will be hypothesized that:

H1a: Outcome control will strengthen goodwill trust in inter-organizational relationship, but its

effect will become negative over time.

H1b: Outcome control will strengthen competence trust in inter-organizational relationships,

but its effect will become negative over time.

H1c: Behavior control will strengthen goodwill trust in inter-organizational relationship, but its

effect will become negative over time.

H1d: Behavior control will strengthen competence trust in inter-organizational relationship, but

its effect will become negative over time.

In contrast, social control is not based on explicit rules and objectives, but on shared goals, norms and believes (Costa & Bijlsma-Frankema, 2007; Emsley & Kidon, 2007; Munoz & Medina-Munoz, 2004). These fundaments do not require rules of conduct or outcome targets, but create a state of mutual understanding (Das & Teng, 2001b) which are useful in partner cooperation (Fryxell et al., 2002). Therefore, trust (Aulakh et al., 1996; Ouchi, 1979), and especially goodwill trust, will be increased due to interpersonal, repeated interactions (Aulakh et al., 1996) such as frequent meetings and communications, cultural blending and socialization (Das & Teng, 1998). Empirically, the positive relationship between social control and goodwill trust is confirmed by Aulakh et al. (1996) and Sengün and Wasti (2007). However, both did not control for the length of the relationship.

(14)

A.C. Wind – The unavoidable balance between trust and control | 14 The time aspect suggested above could also be relevant for the social control aspect. Based on social control, some shared goals, norms and believes (Costa & Bijlsma-Frankema, 2007; Emsley & Kidon, 2007; Medina-Munoz & Medina-Munoz, 2004) will be created already at the beginning of a relationship. However, it should be noted that the use of social control separate from other constructs might be insufficient. Social control mechanisms are only effective when they are complemented by trust (Fryxell et al., 2002). When some trust is built and partners learn more from each other over time, their mutual understanding might be extended. As noted by Inkpen and Currall (2004), a partner who learned more about the other increases its dependence on him (e.g. the opportunity costs of leaving the relationship increases) resulting in more performance risks. In turn, the desire for social controls will increase. In addition, Fryxell et al. (2002) suggested that the reduced benefits of formal control over time makes the development of social control more desirable. Thus, social control might be important across time. Therefore, it will be hypothesized that:

H2a: Social control will strengthen goodwill trust in inter-organizational relationships

throughout the relationship.

H2b: Social control will strengthen competence trust in inter-organizational relationships

Influence of trust on control

(15)

A.C. Wind – The unavoidable balance between trust and control | 15 Goodwill trust will increase the effectiveness of all types of control (Das & Teng, 2001b; Langfield-Smith & Langfield-Smith, 2003), because partners have confidence in each others’ honest motivations to enter the relationship, e.g. they believe the other party is not self-interested. Thus, the organization looks what is the best for the collaboration instead of trying to achieve its own maximum profits (Morgan & Hunt, 1994). As noted by Poppo and Zenger (2002), customized contracts (i.e. formal control) decrease the possibility to behave opportunistically and have a more open-ended character (Edelenbos & Eshuis, 2011). Poppo and Zenger (2002) also argue that those contracts might not be sufficient. When conflicts arise, additional relational governance mechanisms (i.e. trust) is needed to ensure mutuality, bilateralism and continuance. This is in concordance with the finding of Langfield-Smith and Langfield-Smith (2003) that trust allows continuing the relationship when specified standards are absent.

Competence trust will increase the effectiveness of all types of control, because they both believe these control mechanisms are appropriate (Das & Teng, 2001b). It will be difficult to decide which control mechanisms should be used, what rules and procedures will be taken, etc. without a certain trust level (Das & Teng, 1998). In addition, determining those rules, mechanisms, etc. will costs less time in situations with a high trust level (Sengün & Wasti, 2007). Empirically, Sengün and Wasti (2007) already found a positive association between competence trust and social control. However, relying solely on a combination of competence trust and social control will not be sufficient in an inter-organizational relationship since competence trust can enhance the risk of opportunistic behavior (Lui & Ngo, 2004). Thus, when the trust between parties is mainly based on competences additional (formal) control mechanisms are needed (Lui & Ngo, 2004). Therefore, it will be expected that competence trust will strengthen social control, whereas simultaneously it requires a higher level of behavior and outcome control.

(16)

A.C. Wind – The unavoidable balance between trust and control | 16 high level of trust, decreases the level of control needed. However, Dekker (2004, p.46) argued that control ‘mechanisms were nevertheless extensively used’ for coordination purposes, although they were not necessary for safeguarding. Thus, it seems that partners still use control mechanisms when there are high levels of trust. Therefore, it will be hypothesized that

H3a: Goodwill trust will strengthen behavior and outcome control in inter-organizational

relationships throughout the relationship.

H3b: Goodwill trust will strengthen social control in inter-organizational relationships

H3c: Competence trust will strengthen behavior and outcome control in inter-organizational

relationships throughout the relationship.

H3d: Competence trust will strengthen social control in inter-organizational relationships

Figure 1: Conceptual model

(17)

METHODOLOGY

Although a lot of research has been written about control and trust, there are still ambiguities about the interaction between control and trust. To get better insights in this interaction, a theory testing approach was followed which includes deduction, testing and evaluation (Aken, Berends, & Van der Bij, 2012; Blumberg, Cooper, & Schindler, 2011; Saunders, Lewis, & Thornhill, 2009). This method is appropriate since a theory testing approach is useful (1) when evidence about the investigated business phenomenon is still inconclusive (Aken et al., 2012), and (2) when a study tries to discover the causal relationships between the concepts (Saunders et al., 2009). Therefore, this method is used to get evidence about the phenomenon that can be put into practice.

Sampling and data collection

The hypotheses established in the previous section were tested using a survey. An internet-based (e)mail survey was chosen since internet-based surveys and especially those administered in combination with email, have a greater likelihood that the right person is answering the questionnaire compared to their postal counterparts because most people check their own mail (Witmer, Colman, & Katzman, 1999). In addition, data could be collected more quickly and cost efficiently. However, a main disadvantage is that the response rate is often very low (11% or lower, e.g. Saunders et al., 2009).

(18)

A.C. Wind – The unavoidable balance between trust and control | 18 therefore, excluded from further analysis. In addition, some responses were from different people of the same organization regarding different collaborations. As such, 104 responses about 104 separate collaborations were appropriate representing 94 different organizations. Descriptive data for the respondents are provided in table 1.

Pretest

A pilot test of the questionnaire was conducted to check the clarity of the cover letter, comprehensibility and interpretability of the measurement items and the appropriateness of its length. Seven persons with different backgrounds and unrelated to potential participating organizations filled out the questionnaire and commented on it. Based on these comments, the cover letter and the questionnaire were slightly modified. The measurement items are described below.

Table 1: Characteristics of the respondents (n = 104)

N Mean ± sd Length of relationship (in years)

0-5 years 5-10 years 10-15 years 15-20 years 20-25 years > 25 years 20 17 25 14 9 19 13.87 ± 10.10

(19)

Measurement

An in-depth literature review yielded potential measurement items for all types of trust and all types of control used in this study. Multi-item scales will be used and they will be measured on a seven-point Likert-type scale ranging from 1 = "Strongly disagree" to 7 = "Strongly agree" unless otherwise specified. These items will be shown in Appendix A.

Goodwill trust. As mentioned before, goodwill trust is defined as the confidence in others' intentions and integrity (Das & Teng, 2001). After eliminating the redundant items, a seven-item scale remained based on Johnson, Cullen, Sakano, and Takenouchi (1996), Sengün and Wasti (2007, 2009, 2011), and Zaheer et al. (1998). These and similar items were also used by other researchers such as Lai et al. (2012) and Lui and Ngo (2004).

Competence trust. In this study, competence trust is defined as the confidence that a partner is able to perform as expected (Nooteboom et al., 1997; Xie & Peng, 2011). Based on measurement items of Fryxell et al. (2002), Johnson et al. (1996) and Sengün and Wasti (2007, 2009, 2011), a six-item scale for competence trust is composed.

Outcome control. Outcome control refers to monitoring the achievement of the specified desired results (Emsley & Kidon, 2007) in which reliable measurements are required (Ouchi, 1977). According to Sengün and Wasti (2009) the operationalization of outcome control should include ex ante efforts for goal setting and clarifications of the partners' expectations on the one hand, and on the other hand, ex post information processing and performance evaluation during the relationship to manage unresolved control problems. Furthermore, Challagalla and Shervani (1997) underline the importance of both information and reinforcement in measuring all types of control. Thus, measurements should be based on goal setting, performance measures, performance evaluations and reinforcements. In this study, items of Aulakh et al. (1996), Sengün and Wasti (2007), and Wallenburg and Schäffler (2014) were used with some minor modifications resulting in an eight-item scale for outcome control.

(20)

A.C. Wind – The unavoidable balance between trust and control | 20 conducted based on measurement items of Aulakh et al. (1996) and Wallenburg and Schäffler (2014) with some minor modifications.

Social control. Social control refers to the shared norms, values and believes (Costa & Bijlsma-Frankema, 2007; Emsley & Kidon, 2007; Medina-Munoz & Medina-Munoz, 2004). Social control consists of reputation, interaction and social networks on the one hand, and the building of trust through risk taking, joint decision making and problem solving on the other hand (Dekker, 2004). As noted by Das and Teng (2001), socialization and consensus-making processes are appropriate. Based on measurement items of Arranz and de Arroyabe (2012), Sengün and Wasti (2009) and Srivastava and Teo (2012), a seven-item scale for social control was composed.

Control variables. In this study, five control variables were included: relationship length, control purpose, trust propensity, interdependence and relational risk. The length of the relationship is chosen, because firstly the likelihood of opportunistic behavior decreases when there is a more enduring relationship between the organizations (Lai et al., 2012). This is due to the fact that a shared history makes behavior more predictable (Selnes, 1998). In addition, a longer shared history can make relationships more long-term-oriented (Lai et al., 2012). Secondly, previous experiences between the organizations can affect the effectiveness of control mechanisms (Fryxell et al., 2002).

Purpose of control. Dekker (2004) mentioned three purposes of control mechanisms in inter-organizational relationships. One is the establishment of a state that will motivate both parties to gain certain preferred outcomes (e.g. safeguarding). The other is the coordination of tasks that are interdependent and require mutual modifications. The need for coordination increases when tasks heavily depend on each other (Dekker, 2004; Gulati & Singh, 1998). Both, safeguarding and coordination emphasize the need for managing and protecting the creation and appropriation of value (Dekker, 2004). A third purpose of control, as suggested by Dekker (2004), is providing stability and continuity of the relationship in the future. As such, three types of control purposes are yet identified: safeguarding, coordination and stabilization and will be covered in this research. Respondents can choose between appropriation concerns, coordination requirements, stability/durability and none of these. This variable will be treated as a dummy variable.

(21)

A.C. Wind – The unavoidable balance between trust and control | 21 of human nature in general’ (Yamagishi & Yamagishi, 1994, p. 139) and it is important when there is an insufficient level of knowledge of the partners. In an inter-organizational context, people with high general trust will notice fewer risks in relationships with others. Consequently, those people will be more likely to quit committed relationships due to exchange value uncertainty which is ‘a situation in which there is uncertainty about the value of future exchanges with a certain actor, whether or not there is a propensity to act opportunistically’ (Lazzarini, Miller, & Zenger, 2008, p. 711). When this exchange value uncertainty is high, partners will be more inclined to look for alternatives. Furthermore, whether contract enforcement is present or not will be less important for them, even if there is social uncertainty (Lazzarini et al., 2008). Since the level of trust propensity differs between persons and affects their perception about control (Lazzarini et al., 2008), it is important to include this variable in this research. A five-item scale is adapted from Mayer and Davis (1999).

Interdependence. Furthermore, interdependence is chosen ‘because high interdependence coupled with high levels of trust will reinforce the desire for maintaining the continuity and the correct functioning of the relationship between exchange partners’ (Sengün & Wasti, 2007, p. 451). Following Gulati and Sytch (2007) and Kumar, Scheer, and Steenkamp (1998), dependence concerns two main aspects: magnitude of exchange and the irreplaceability. Measurement items of Gulati and Sytch (2007) and Sengün and Wasti (2007) were used to conduct a six-item scale for interdependence including both aspects of dependence.

Relational risk. Relational risk consists of ‘the possibility and the consequence that the partners in inter-firm alliances do not fully commit themselves to joint efforts’ (Das & Teng, 1996, p. 831). For instance, this means that a partner is behaving opportunistically. Based on the measurement items of Das and Teng (2001a), an eleven-item scale for relational risk was composed.

Analysis

(22)

A.C. Wind – The unavoidable balance between trust and control | 22 Anderson, 2014, p. 603). This means that instead of allowing the factor analysis to determine the number of factors and loadings as in an exploratory factor analysis (EFA), a CFA tests how the theory match with the data (Hair et al., 2014).

The initial model did not suggest that the conceptual model fits the data well (2 (832) = 1507.952; GFI = .637 ; CFI = .761 ; NFI = .595 ; RMR = .189 ; RMSEA = .089). After making some adjustments, the model fit of the CFA has improved (2 (821) = 1325.922; GFI = .679 ; CFI = .821 ; NFI = .644 ; RMR = .178 ; RMSEA = .077), but it is still does not meet the rules of thumb of a CFI > .92 (Hair et al., 2014). A possible explanation for the lack of fit concerning the goodness of fit (GOF) indices is that the fit might increase with larger sample sizes.For instance, Jackson (2001) found that with a sample size of 50, the mean GFI was .745, a RMSEA of .061 and a CFI of .911, whereas the GFI was . 923, the RMSEA was .013 and the CFI was .992 for a sample size of 200. Furthermore, the standard deviation of those fit indices decreases with as a function of sample size (Jackson, 2001). Therefore, the results of this CFA were used for further analysis which are shown in Appendix B and C. Although this CFA shows a reasonable convergent validity, the discriminant validity was problematic since goodwill trust and competence trust were hard to distinguish (e.g. correlations of .899). Additional collinearity diagnostics were conducted that show high variance inflation factors (VIFs) for goodwill and competence trust (VIF = 8.860 and 10.110 respectively). This can be very problematic since high collinearity makes regression coefficients untrustworthy and the importance of predictors become difficult to assess (Field, 2009; N. K. Malhotra, 2010). To avoid those problems, a new CFA was conducted with trust as a single variable. Next, the factors were tested using structural equation modeling (SEM) in SPSS Amos. First, a multidirectional model was estimated which includes all reciprocal causal relationships and subsequently two unidirectional models were conducted which estimates one direction of the relationship; thus, either trust on control or control on trust. All models will be treated in the next section.

RESULTS

Factor analysis

The new CFA shows a significant chi-square statistic (2 (839) = 1547.399) and a moderate model fit (GFI = .629 ; CFI = .749 ; NFI = .585 ; RMR = .192 ; RMSEA = .091). After making some adjustments, the model fit of the CFA has improved (2 (827) = 1319.323; GFI = .680 ; CFI = .826 ; NFI = .646 ; RMR = .185 ; RMSEA = .076). Table 2 shows the standardized factor loadings for each construct and, table 3 shows the mean, standard deviation and (squared) correlations between the factors.

(23)

A.C. Wind – The unavoidable balance between trust and control | 23 AVEs did not exceed the 50 percent rule of thumb, they are quite close to it. Construct reliabilities range from .923 for OC to .971 for Trust exceeding the .7 threshold (Hair et al., 2014) which suggest an adequate reliability. Taken together, the evidence shows a reasonable convergent validity. Discriminant validity is exhibited in that all squared interconstruct correlations between the factors are less than each factor’s AVE (Hair et al., 2014).

Furthermore, Harman’s single-factor test is used to test the existence of common method bias (CMB). The CMB should be tested since it can affect the results substantially (Podsakoff, MacKenzie, Lee, & Podsakoff, 2003). The data indicates that no single factor accounts for the majority of variances among the constructs, since the exploratory analysis shows a 27.6 percent total variance for one factor. Therefore, the results are not vulnerable to CMB.

Multidirectional model testing

Next, the structural model is conducted. The model shows a non significant chi-square statistic (2 (19) = 25.792, 2/d.f. = 1.357, p = .136) with good model fit indices (GFI = .958, RMR = .032, CFI = .981, NFI = .938, RMSEA = .059). As such, there is evidence for an adequate fit allowing the relationships to be analyzed. ‘Short’ consist of relationships between organizations less than five years (n=20), whereas ‘long’ consists of relationships between organizations of five years and longer (n=84).

The effects of control on trust

Table 4 summarizes the effects of control on trust. It was found that, in general, outcome and behavior control have a significant effect on trust. Whereas it was positive for outcome control ( = 1.006, p<.01), it was negative for behavior control ( =-.876, p<.01). This implies that a higher level of trust is achieved in relationships with a higher level of outcome control mechanisms. In addition, outcome control is negative in short-term relationships ( = -.960, p>.1), although this effect is not significant, whereas it is positive in long-term relationships ( = 1.007, p<.01). Therefore, hypotheses H1a and H1b are not supported by the multidirectional model. In contrast, behavior control is positive

in short relationships ( = 1.085, p>.1), although this effect is not significant as well, whereas it is negative over time ( = -.897, p<.01). Although goodwill and competence trust cannot be distinguished, the general impact of trust suggest that H1c and H1d are partly supported by the

(24)

A.C. Wind – The unavoidable balance between trust and control | 24 Table 2: Factor loadings, average variance explained (AVE) and construct reliability (CR)

Trust Outcome Control Behavior Control Social Control Inter-dependence Trust Propensity Relational Risk GT1 .633 GT2 .736 GT3 .624 GT4 .758 GT5 .622 GT6 .802 GT7 .843 CT1 .796 CT2 .793 CT3 .723 CT4 .732 CT6 .674 OC1 .814 OC2 .799 OC4 .721 OC5 .769 OC6 .453a OC8 .502a BC3 .980 BC4 .702 BC5 .742 SC1 .549 SC4 .760 SC5 .905 SC6 .579 SC7 .812 ID1 .615 ID2 .912 ID3 .572 ID5 .641 TP2 .711 TP3 .842 TP4 .837 TP5 .724 RR1 .429a RR3 .575 RR4 .759 RR6 .340a RR8 .375a RR9 .666 RR10 .821 RR12 .847 RR14 .826 AVE .535 .478 .668 .538 .487 .611 .455 CR .971 .923 .954 .965 .937 .959 .968 a

(25)

A.C. Wind – The unavoidable balance between trust and control | 25 Table 3: Mean, standard deviation and (squared) interconstruct correlations

Mean sd CT OC BC SC ID TP RR Trust 3.673 .591 1.000 .121 .003 .533 .027 .026 .299 Outcome Control 3.180 .799 .348 1.000 .228 .179 .058 .005 .097 Behavior Control 2.776 1.372 .055 .477 1.000 .023 .060 .001 .025 Social Control 4.909 .755 .730 .423 .152 1.000 .049 .023 .446 Interdependence 3.393 1.066 .164 .240 .245 .221 1.000 .031 .000 Trust Propensity 3.042 1.006 .162 .070 .027 .150 .177 1.000 .034 Relational Risk 1.073 .962 -.547 -.311 -.159 -.668 -.021 -.185 1.000

Values below the diagonal are interconstruct correlations, construct variances are shown on the diagonal and the values above the diagonal are the squared interconstruct correlations.

The effect of trust on control

Vice versa, trust has only a significant effect on social control ( = .570, p<.01) meaning that trust strengthens social control in a relationship. When looking at the short-term vs. long-term relationships, trust only affects social control in long-term relationships significantly ( = .540, p<.01). Therefore, H3a and H3c are not supported, while H3b and H3d are supported by the multidirectional

model although the data cannot distinguish goodwill and competence trust.

The effects of the control variables

First of all, trust propensity as well as relationship length have no significant effects on either trust or control. Interdependence has, in general, a significant positive effect on outcome and social control ( = .329, p<.01;  = .143 p<.01 respectively). Although they are far from significant for short-term relationships, they are significant for long-term relationships ( = .355, p<.01;  = .213 p<.01 respectively). This suggests that relationships with higher dependency require a higher level of outcome and social control.

(26)

A.C. Wind – The unavoidable balance between trust and control | 26 Table 4: Parameter estimates of the multidirectional model

Dependent variable

Trust Outcome Control Behavior Control Social Control

Variable  p-value  p-value  p-value  p-value

(27)

A.C. Wind – The unavoidable balance between trust and control | 27 Behavior control is, in general, mostly affected by stability ( = .518, p<.01), followed by coordination ( = .375, p<.05) and safeguarding ( = .373, p<.05). Although the effects are not significant in short-term relationships, they are significant in long-short-term relationships ( = .572, p<.01;  = .391, p<.05;  = .416, p<.05 respectively). This suggest that all control purposes require a higher level of behavior control mechanisms.

Finally, perceived relational risk has, in general, a significant negative influence on trust, outcome control and social control ( = -.660, p<.01;  =-.640, p<.05;  = -.395, p<.01 respectively). This implies that trust and the level of outcome and social control decrease when the partners perceive that there are more relational risks present. In addition, the effect of perceived relational risks on social control is stronger (e.g. more negative) for shorter relationships ( = -.753, p<.01) than for longer relationships ( = -.395, p<.01).

Figure 2: Results multidirectional model

Note: trust propensity is excluded from this figure since it has no significant effects on either trust and all types of control. In addition, the different purposes of control has no significant effect on social control. Therefore, only their effects on outcome and behavior control are shown in order to make Figure 2 less cluttered.

* p <.05

** p < .01

Unidirectional model testing

(28)

A.C. Wind – The unavoidable balance between trust and control | 28 control and vice versa separately. With these two additional models the reverse causation in the multidirectional model are avoided. The first unidirectional model estimates the effects of control on trust (2 (38) = 80.893, 2/d.f. = 2.129, p = .000; GFI = .874, RMR = .229, CFI = .882, NFI = .807, RMSEA = .105), and the second unidirectional model estimates the effects of trust on control (2 (24) = 32.444, 2/d.f. = 1.352, p = .116; GFI = .944, RMR = .055, CFI = .977, NFI = .922, RMSEA = .058). Since almost all the goodness of fit indices are satisfactory, the relationships can be analyzed. Again, ‘short’ consist of relationships between organizations less than five years (n=20), whereas ‘long’ consists of relationships between organizations of five years and longer (n=84).

Table 5: Parameter estimates of unidirectional model of control on trust

(29)

A.C. Wind – The unavoidable balance between trust and control | 29 Unidirectional model of the effects of control on trust

Table 5 summarizes the effects of control on trust. It was found that outcome and behavior control have no significant effect on trust at all. This means that no differences are observed in the available data and therefore, H1a-d are not supported by the unidirectional model. On the other hand, social

control has a significant positive effect on trust in general ( = .668, p<.01). In addition, social control plays an important role in short-term relationships ( = .355, p<.1) as well as in long-term relationships ( = .690, p<.01).

Unidirectional model of the effects of trust on control

Table 6 summarizes the effects of trust on control. Generally, trust has a significant positive effect on outcome control ( = .246, p<.05). This effect is stronger in long-term relationships ( = .233, p<.05) compared to short-term relationships ( = .113, p>.1), though the latter is not significant. This implies that trust strengthens outcome control throughout the relationship, and its influence increases over time. Furthermore, no significant effect from trust on behavior control was found. This means that H3a and H3c are partly supported by the unidirectional model. On the other hand, trust has a

significant positive effect on social control ( = .493, p<.01). Such as outcome control, this effect is stronger for longer relationships ( = .501, p<.01) than the shorter ones ( = .298, p<.05). Since trust strengthens social control throughout the relationship, H3b and H3d are supported by the

unidirectional model.

The effects of the control variables in the two unidirectional models

(30)

A.C. Wind – The unavoidable balance between trust and control | 30 Table 6: Parameter estimates of unidirectional model of trust on control

Dependent

Outcome Control Behavior Control Social Control

Variable b s.d.  p-value b s.d.  p-value b s.d.  p-value

(31)

A.C. Wind – The unavoidable balance between trust and control | 31 Subsequently, all purposes of control have no significant effects on social control. They have, however, significant effects on outcome and behavior control. Outcome control is, in general, mostly affected by safeguarding ( = .461, p<.01), followed by coordination ( = .421, p<.01) and stability ( = .363, p<.01). In short-term relationships, coordination is mostly influential for outcome control ( = .914, p<.01), followed by safeguarding ( = .688, p<.01) and stability ( = .598, p<.01), whereas in long-term relationships, outcome control is mostly affected by safeguarding ( = .416, p<.01), subsequently coordination ( = .375, p<.01) and stability ( = .302, p<.1).

Behavior control is mostly affected by stability ( = .452, p<.01), followed by coordination ( = .413, p<.01) and safeguarding ( = .266, p<.05). In short-term relationships, coordination ( = .691, p<.01) is more important than stability ( = .435, p<.1), whereas in long-term relationships stability is most important ( = .512, p<.01), subsequently coordination ( = .409, p<.01) and safeguarding ( = .335, p<.05).

Furthermore, safeguarding and stability have a significant negative effect on trust in shorter relationships ( = -.420, p<.01;  = -.330, p<.05 respectively). This implies that there is less trust in the relationship when controls are used for safeguarding or stability.

Finally, relational risk has a significant negative effect on behavior and social control ( = -.330, p<.05;  = -.433, p<.01 respectively). Regarding behavior control, relational risk has a greater impact during longer relationships ( = -.265, p<.05) than during shorter relationships ( = -.181, p>.1), though the latter is not significant. In contrast, social control is more affected by relational risk in a shorter relationship ( = -.670, p<.01) than in a longer relationship ( = -.406, p<.01). This implies that social control is less effective when the perceived relational risks are higher; especially in shorter relationships.

DISCUSSION

Theoretical implications

(32)

A.C. Wind – The unavoidable balance between trust and control | 32 the reverse unidirectional model testing the effects of trust on control. This resulted in several interesting findings, however, the results should be interpreted with caution since the sample size was small. Furthermore, it is important to note that it was not possible to make a distinction between goodwill and competence trust based on the available data. Therefore, goodwill and competence trust could not be considered as separate constructs as done by other researchers such as Das and Teng (2001b), Dekker (2004), Langfield-Smith and Smith (2003) and Sengün and Wasti (2007).

First, the interaction between trust and outcome control is examined. While Das and Teng (2001b) proposed that trust will strengthen the effectiveness of outcome control in inter-organizational relationship, this study provides mixed evidence about the influence of trust on outcome control. While the multidirectional model contradicts this proposition, though it is not significant, the unidirectional model shows that trust positively influences outcome control. The latter effect contradicts earlier findings of Sengün and Wasti (2007) and Zaheer and Venkatraman (1995), who found that trust and outcome control act as substitutes. In line with Costa and Bijlsma-Frankema (2007), Langfield-Smith and Smith (2003), Poppo and Zenger (2002) and Vryza and Fryxell (1997) trust strengthens outcome control in an inter-organizational relationship. This can be explained by the fact that the implementation and use of control requires a certain minimum level of trust (Das & Teng, 1998) and makes it more effective (Vryza & Fryxell, 1997). In relationships based on high trust, partners are more likely to respect each others’ interests and desires. Therefore, they will be more inclined to make and accept outcome objectives they make together. In contrast, in relationships with a low level of trust, it will be harder to accept these rules, since it is more difficult to believe in each others’ honesty and abilities. However, further research should examine this relationship since mixed effects are found.

(33)

A.C. Wind – The unavoidable balance between trust and control | 33 Since this research found results contradictory to previous research, further research is needed to clarify on this matter.

Second, it was hypothesized that behavior control strengthens trust in short relationships, but that its effect becomes negative in more enduring relationships. Although there was no support for the dynamic relationship between behavior control and trust, the findings show that, in general, behavior control negatively influences trust. This means that the use of a higher level of behavior control decreases the level of trust in a relationship. As such, the control of partner’s behavior can suggest a sense of mistrust (Argyris, 1952 in Das & Teng, 2001b; Edelenbos & Eshuis, 2011; Poppo & Zenger, 2002) about each others’ competence and goodwill (Das & Teng, 2001b). This will be harmful for an inter-organizational relationship since close monitoring can create conflicts because most organizations do not like to be monitored (Das & Teng, 2001b).

Furthermore, the effect of trust on social control has been found positive meaning that trust strengthens social control which confirms the findings of Aulakh et al. (1996) and Sengün and Wasti (2007). This implies that trust facilitates the development of social control, which are the shared goals, norms and believes (Costa & Bijlsma-Frankema, 2007; Emsley & Kidon, 2007; Medina-Munoz & Medina-Munoz, 2004). In addition, Ouchi (1979) already suggested that social control is optimized in a relationship with a relatively high level of trust. Although the multidirectional model does not provide further (significant) insights regarding the length of relationship, the unidirectional model does. It can be seen that trust has a stronger effect on social control in long-term relationships, which might be explained by an increased level of trust over time due to, for instance, past experiences. This learning of partners about each other develops a higher level of mutual understanding as well (Inkpen & Currall, 2004). In addition, it is shown that the influence of trust on outcome and behavior control decreases over time, whereas it increases for social control. This implies that there is an increased desire for social control due to the reduced benefits of formal controls over time (Fryxell et al., 2002).

(34)

A.C. Wind – The unavoidable balance between trust and control | 34 a higher mutual understanding is developed. However, further research should examine the reciprocal nature of trust and social control since the unidirectional and multidirectional models differ in the effects of social control on trust.

Regarding the control variables, the results show that interdependence strengthens all control types. Higher interdependence means that there are more risks involved in the relationship as well, e.g. the impact of, for instance, opportunistic behavior will be higher for partners that are highly dependent than partners which are only low or moderately dependent. To reduce potential consequences, a higher level of control is required since this makes partners more predictable (Costa & Bijlsma-Frankema, 2007). Partners can rely on each other, because clear goals, either formal or informal, are established and there might be rules about potential penalties if they do not follow the predetermined rules. Important to note is that those risks can never be completely avoidable, however, they can be reduced dramatically by the use of controls.

When interpreting the role of trust propensity, no effect has been found. Therefore, it is likely to assume that someone’s general perception of trust does not significantly influence the interaction between trust and the types of control.

Furthermore, relational risk is found to have a negative effect on trust, outcome and social control. Relational risk is associated with opportunistic behavior and non-collaborative behavior (Das & Teng, 2001b) meaning low output measurability and relatively high task programmability (Das & Teng, 2001b; Eisenhardt, 1989; Ouchi, 1979). Thus, a higher level of perceived relational risk means that there is more potential that a partner behaves opportunistically and/or that there is an ineffective collaboration. Since it is hard to criticize results in such circumstances, outcome controls will be less useful. In addition, the dependence on social control will be reduced as well, since opportunistic behavior hampered the development of a mutual understanding due to doubts about the others’ intentions. Similarly, trust will be reduced since one can doubt the others’ honesty (Das & Teng, 2001b; Nooteboom, 1996) and abilities.

(35)

A.C. Wind – The unavoidable balance between trust and control | 35 control mechanisms negatively influence the level of trust between partners. Further research is needed to investigate this matter.

However, the suggestion of Dekker (2004) that the interaction of trust and control depends on the purpose of control is supported by this research. On the contrary, Gulati, Nohria, and Zaheer (2000) argued that trust can alleviate appropriation concerns and reduce transaction costs by making it less likely to rely on detailed contracts which is not supported by this research since the purposes of control do not have a significant impact on trust. As such, the finding of Dekker (2004) that goodwill trust seems to weaken transaction risks and seems to reduce the need for formal control is not supported as well. In contrast, he noticed that control mechanisms are still used in order to coordinate relationship tasks. As he suggested ‘increasing competence trust in and knowledge of a partner can result in either a reduced need for formal coordination or in improved coordination’ (Dekker, 2004, p. 35) which can explain that outcome control mechanisms are required when controls are used for the coordination of tasks. In addition, Vosselman and Meer-Kooistra (2009) confirmed that trust and control are important for the stability and durability of a relationship. They argued that the legitimate mistrust should be compensated by the governance structure. However, caution is required since exceeding the necessary compensation might harm the stability and durability of the relationship. Taken together, these arguments suggest that the purpose of control really influences the interaction between control and trust.

Managerial implications

(36)

A.C. Wind – The unavoidable balance between trust and control | 36 decreases the level of trust in the relationship and has no other positive side effects. Furthermore, no organization likes to be closely monitored by another meaning that organizations can better make rules that are focused on outcomes or enhancing their inter-organizational contact.

Limitations and suggestions for further research

Besides the suggestions for further research mentioned in the previous sections, the limitations of this study offer opportunities for further research as well. Due to time constraints, this study includes only a small sample size, something that could have affected the size and significance of the results. In addition, this sample size limits the generalizability of the results. Further research should replicate this study with a larger sample size.

Second, while goodwill and competence trust were considered as two distinct concepts, the data does not fit this distinction resulting in a single construct for trust. Consequently, the possible differences in effects of both types of trust could not be investigated. These issues request further investigation through, for instance, other measurement items or two separate models which includes both one type of trust.

Third, the purpose of control was treated as a dummy variable. Therefore, the results do not take the strength of these effects into account, hence they measure solely whether there is an effect or not, and its direction. Further research should measure the size of the effects of the control purpose on the interaction between trust and control.

Conclusion

(37)

REFERENCES

Aken, Berends, & Van der Bij, H. (2012). Problem Solving in Organizations - A methodological handbook for business and management students. Cambridge University Press.

Altinay, L., & Brookes, M. (2012). Factors influencing relationship development in franchise partnerships. Journal of Services Marketing, 26(4), 278–292.

Argyris, C. (1952). The impact of budgets on people. New York: Controllership Foundation.

Arranz, N., & de Arroyabe, J. C. F. (2012). Effect of Formal Contracts, Relational Norms and Trust on Performance of Joint Research and Development Projects. British Journal of Management, 23(4), 575–588.

Aulakh, P. S., Kotabe, M., & Sahay, A. (1996). Trust and performance in cross-border marketing partnerships: A behavioral approach. Journal of International Business Studies., 27(5), 1005– 1032.

Barney, J. (1991). Firm Resources and Sustained Competitive Advantage. Journal of Management, 17(1), 99–120.

Blumberg, Cooper, & Schindler. (2011). Business Research Methods. McGraw-Hill Education.

Bradach, J. L., & Eccles, R. G. (1989). Price, Authority, and Trust: From Ideal Types to Plural Forms. Annual Review of Sociology, 15(1), 97–118.

Challagalla, G. N., & Shervani, T. A. (1997). A Measurement Model of the Dimensions and Types of Output and Behavior Control: An Empirical Test in a Salesforce Context. Journal of Business Research, 39(3), 159–172.

Chen, H., & Chen, T.-J. (2003). Governance structures in strategic alliances: transaction cost versus resource-based perspective. Journal of World Business, 38(1), 1–14.

Coletti, A. L., Sedatole, K. L., & Towry, K. L. (2005). The Effect of Control Systems on Trust and Cooperation in Collaborative Environments. Accounting Review, 80(2), 477–500.

Costa, C. A., & Bijlsma-Frankema, K. (2007). Trust and Control Interrelations: New Perspectives on the Trust Control Nexus. Group & Organization Management, 32(4), 392–406.

Crook, T. R., Combs, J. G., Ketchen, D. J., & Aguinis, H. (2013). Organizing Around Transaction Costs: What Have We Learned and Where Do We Go from Here? Academy of Management Perspectives, 27(1), 63–79.

Das, T. K., & Rahman, N. (2010). Determinants of Partner Opportunism in Strategic Alliances: A Conceptual Framework. Journal of Business and Psychology, 25(1), 55–74.

Das, T. K., & Teng, B.-S. (1996). Risk types and inter-firm alliance structures. Journal of Management Studies, 33(6), 827–843. doi:10.1111/j.1467-6486.1996.tb00174.x

The unavoidable balance between trust and control How the interaction between outcome, behavior and social control and goodwill and competence trust differ in inter-organizational relationships