The interplay between trust, control and risk in an inter-organizational setting
by
YASHNI BARTOL
University of Groningen
Faculty of Economics and Business
1
ABSTRACT
This study empirically tests Das & Teng’s (2001a) framework about the relation between trust, control and risk and applies it to supplier-buyer relationships in the technology industry in the Netherlands. This study finds evidence that informal control and, both goodwill trust and competence trust, are positively related. Output control is found to be negatively related with goodwill trust. Also, trust as determinant of informal control is found to have positive effects. Finally, goodwill trust is found to be negatively related with relational risk. Hence, Das & Teng (2001a) model partially fits the studied supplier-buyer context.
Keywords: trust, risk, control, supplier-buyer
2
1.
Introduction
Due to globalization, specialization and competition, organizations find themselves in a tough spot to maintain and increase their expertise and skills to compete successfully. To cope with these difficulties and to enhance their competitive position, organizations look for expertise and skills through inter-organizational collaborations (or relationships). (Das & Teng, 2001a; Nooteboom, Berger &
Noorderhaven, 1997). Hence, in the past decade there has been a growing interest in these inter-organizational settings from industries, as well as from academics. Many organizations embraced intra-firm collaborations, e.g. employee teams, to gain synergies. Now organizations breach their intra-firm boundaries to gain this same synergy advantage via inter-organizational settings, e.g. strategic alliances and joint ventures. (Coletti, Sedatole & Towry, 2005)
Engaging in inter-organizational collaborations means engaging in a situation with economic actors outside the firms’ boundaries. Because the actors are outside the firms’ boundaries, the
organization will face uncertainties and potential opportunistic behavior. A major role for management accounting is to manage uncertain and opportunistic behaviors of these actors, and as such, control plays a vital role in inter-organizational settings (Emsley & Kidon, 2007). Following Coletti et al (2005: p. 478),the inter-organizational collaboration ‘‘presents a managerial control challenge, as they attempt to identify optimal levels of control mechanisms, while simultaneously considering the impact of such mechanisms on interpersonal trust’’. This control trust relationship has attracted a lot of attention from scholars. Some researchers indicated that trust is a key determinant for inter-organizational
collaboration success. (Spekman, Spear & Kamauff, 2001; Zaheer & Venkatraman, 1995; Zaheer, McEvily & Perrone, 1998). Also, governance structure is often recognized as being a critical component in the success of inter-organizational collaboration. (Ittner, Larcker, Nagar & Rajan, 1999; Osborn & Baughn, 1990) Following Dyer & Singh (1998), effective governance could even lead to inter-organizational competitive advantage. These governance structures include management control systems (MCS) and trust development which could help to reduce risks and failure rates in these collaborations. (Das & Teng, 2001a; Speklé, 2001)
3 (Costa &Bijlsma-Frankema, 2007; Bachmann, 2001; van der Meer-Kooistra & Vosselman, 2000,). Coletti et al. (2005) on the other hand found a positive relationship between control and trust, i.e. an increase in control results in an increase in trust. Other academics argue that there is a more complex and dynamic relationship (Dekker, 2004; Vélez, Sánchez & Álvarez-Dardet, 2008; Vosselman & Van der Meer-Kooistra, 2009). Because of the contrasting views and findings, the relationship is still under discussion.
This paper will add to this debate, by discussing and testing the interplay between control, trust and risk, as proposed by Das & Teng (2001a), in supplier-buyer relationships in the technology industry in the Netherland. As such, the research question leading this paper is: “what is the relationship between control, trust and risk in a supplier-buyer relationship”.
4
2.
Theoretical background
2.1 Inter-organizational forms
There are various theories that try to explain (inter)organizational governance. These are the most influential: transaction cost economics (TCE), the relational view and the resource based view (RBV). Although it is beyond the scope of this paper to investigate these theories in detail, a small introduction is appropriate.
TCE argues that inter-organizational collaboration is appropriate when there is a situation of high asset specificity, uncertainty and there is a small number of bargaining partners. (Williamson, 1991). In contrast, the relational view proposed by Dyer & Singh (1998) argues that relational rents are the basis for collaboration. Relational rents are defined as: ‘‘a supernormal profit jointly generated in an exchange relationship that cannot be generated by either firm in isolation and can only be created through the joint idiosyncratic contributions of the specific alliance partners’’ (Dyer & Singh, 1998: p. 622).
RBV on the other hand argues that alliance structure is determined by resource profiles of organizations. If a firm is lacking certain resources (e.g. skills or competences), the firm will try to access these
resources through collaboration. As such, existing resources profiles determine (potential alliance) structures. (Chen & Chen, 2003). It appears there is no single answer to which theory should explain the existence of inter-organizational collaboration.
Before moving on, it is important to distinguish between the different governance forms for an organization. Building on TCE, transactions are either governed via market, hierarchy or hybrid
5
Table 1: Governance modes (adapted from Sydow, 1992)
Whether the form is more market or hybrid depends on the potential opportunism by the other party, and hence, the degree of control the organization needs over the relationship. (Dekker, 2004). Although TCE has received critiques about its ability to predict governance structures in
inter-organizational settings (Larson, 1992; Osborn & Hagedoorn, 1997), the used terminology to distinguish between governance structures is helpful. And, the discussion itself is beyond the scope of this paper.
If we talk about an inter-organizational collaboration, we are in the domain of the hybrid-firm. As the organization disengages from discrete market exchange by engaging in collaboration, (Dekker, 2004) and removes itself from hierarchy mode by crossing the organizations’ boundaries. As mentioned, there are a variety of inter-organizational forms; ’‘joint-ventures, buyer-supplier relationships, franchising and licensing agreements, inter-organizational networks’’, (Dekker, 2004: p. 28), which Choe (2008) calls strategic alliances. Following Anderson (1995) a supplier-buyer relationship is formed to either add value or reduce costs of the exchange for both companies. And following Ganesan (1994: p. 1), “firms overlook the sustainable competitive advantage that can be created through long-term relationships with their supplier”. Evidently, firms have much to gain through a solid supplier-buyer relationship. Hence, the focus of this paper is on those supplier-buyer relationships that remove themselves from discrete market exchange. In next section, this supplier-buyer relationship is further elaborated upon.
2.2 Supplier-buyer relationship
6 of as a hybrid culture that develops between the buying and selling firms and reflects elements of both firms’ cultures but is different from either firm’s culture.” (Wilson, 1995: p. 337) This is what Sydow (1992) calls relational contracting. In contrast, discrete market transactions are considered “short-term bargaining relationships between highly autonomous buyers and sellers.” (Choe, 2008: p. 448) Following Sydow (1992), this concerns spot contracting. This study focuses on the long(er)-term supplier-buyer relationships .Referring to the useful model provided by Sydow (1992), this entails the domain of relational (sub-)contracting.
A sometimes confusing term used in supplier-buyer relationship is outsourcing. In essence, outsourcing implies: “the contracting of any service or activity to a third party” (Langfield-Smith & Smith, 2003: p. 282). Hence, outsourcing can be seen as a make-or-buy decision as defined by Williamson (1991). In his view, it pays off to outsource services when asset specificity and uncertainty are low. Nowadays however, it is common knowledge that outsourcing can also be valuable under high asset specificity (Nooteboom, 1999) and/or high uncertainty (Gulati, 1995). Mol (2004) argues that
partnerships with external suppliers (outsourcing) can be a substitute for vertical integration, which is the reason that outsourcing is so widely used. But how does outsourcing fit within the hybrid modes as illustrated by Sydow (1992) and what is the difference with more traditional supplier-buyer
relationships?
Arnold’s (2000) model helps to clarify:
Table 2: Types of outsourcing (adapted from Arnold, 2000)
7 between internal and external outsourcing? Following Arnold (2000: p. 25), ”internal outsourcing refers to a higher degree of hierarchical steering by forming independent profit centers”. This is clearly beyond the scope of this paper. Excluding spot transactions, external outsourcing on the other hand is within the scope of this paper: “external outsourcing means spot transactions or long-term relationships with the supplier” (Arnold, 2000: p. 25). So, external outsourcing (Arnold, 2000) complies with relational sub-contracting (Sydow, 1992). An important difference between outsourcing and relational sub-contracting is that outsourcing involves the supplying and buying of services or activities. In contrast, traditional supplier-buyer relationships involve the supplying and buying of products. As this paper is about the interplay between control, trust and risk in inter-organizational collaborations, both relational
contracting and external outsourcing as previously described are within the scope of this paper. So in this paper, a supplier-buyer relationship may involve both relational contracting and external outsourcing. From here on forward, the term supplier-buyer relationship will be used.
As highlighted in the introduction, inter-organizational collaborations and thus supplier-buyer relationships, cause risk. The different parties might have different objectives; hence there arises the potential for opportunistic behavior. (Das & Teng, 2001a) This high-risk problem also arises from the fact that nowadays not only supporting services are outsourced, but even some essential core activities which are crucial for the firm’s continuity. Especially when the outsourcing concerns such delicate matters, close forms of cooperation between the parties can be observed (Langfield-Smith & Smith, 2003). Where free competition and authority are controlling devices for markets and hierarchies respectively, hybrid governance generally entail long-term contracts (relational contracting) as control vehicle. (Langfield-Smith & Smith, 2003; Williamson, 1996). Faulkner (1995), mentions that attitudes and personal relationship between the collaborators is essential. This is acknowledged by Das & Teng (2001a) and Speklé (2001), who argue that the development of trust can also act as control vehicle and safeguard for risk, and even reduce opportunism (Nooteboom et al., 1997; van der Meer-Kooistra & Vosselman, 2000) . But which way are inter-organizational collaborations managed best?
2.3 Managing inter-organizational collaborations
8 van der Meer & Vosselman, 2000). Clearly, control and trust are important to act as safeguard of risk. Next section further elaborates on control, trust and risk.
2.4 Control
Management control systems have received extensive attention by academics (Chenhall, 2003; Malmi & Brown, 2008; Ouchi, 1979; Williamson, 1979). Because of this extensive research, several definitions of control were introduced. Das &Teng (2001a) mentions that control is broadly defined as a process regulating behavior in reaching organizational goals. Coletti et al’s (2005: p. 480) definition is: ‘’the various policies and procedures that firms use to mitigate different types of risks’’. Both definitions of control are useful for this paper. It is the essence of control to align employee behaviors with
organizational goals. However, by entering in a supplier-buyer relationship, the extent to which
employee behaviors of the external parties can be controlled, i.e. aligned with the own company goals, is limited. It is limited because the external employees involved follow the organizational goals of their own company. As described in the previous section, this induces risk to the organization. Therefore, the definition of control as stated by Coletti et al (2005) comes in handy, as it portrays control as a vehicle to mitigate risk.
Following Ouchi & Maguire (1975) control can be broken down to two types of control: formal and informal (or social) control. Formal control can comprise of controls measuring behavior (behavior control) or output (output control). As Das & Teng (2001a: p. 259) note: “measuring behavior (or
behavior control) is to ensure that the process is appropriate, while measuring outcomes only (or output controls) is to rely on an accurate and reliable assessment of members’ performance.” Or, in Ouchi & Maguire’s (1975: p. 559) words: “studies have distinguished between control based on direct personal surveillance (behavior control) and control based on the measurement of outputs (output control).”
9 goals among members so that appropriate behaviours will be reinforced and rewarded”. From now on, the term informal control will be used and this thus entails social control as defined by Das & Teng (2001a) and clan control (Ouchi, 1979).The distinction between the different types of control is important, as each type of control can have a different influence on trust and risk. This will become clearer in section 2.7.
To summarize, the definitions of Ouchi & Maguire (1975), Das & Teng (2001a) and Coletti et al (2005) are combined and thus control consists of those policies and procedures, both informal control and formal control (comprising of behavior and output control), that are used to mitigate risk in inter-organizational collaborations.
2.5 Trust
Beside the extensive literature on control, there is also extensive literature on trust and its definitions. In general it is accepted that trust is a psychological state of mind that is reflected in specific behavior towards a certain person or group (Kramer, 1999). Coletti et al (2005: p. 481) make a distinction in their paper between trustworthiness and trust. Trustworthiness is defined as: ´´an innate personal characteristic reflecting one’s preference for upholding some social norm of behavior, regardless of economic incentives´´. Trust on the other hand is defined as ‘‘one’s perception of another’s
trustworthiness”.
Coletti et al’s (2005) definition reflects the general accepted notion of trust as mentioned by Kramer (1999). Mayer, Davis & Schoorman (1995: p. 712) provide the definition of trust as: “the
10 another’s ability and/or competence, where goodwill trust reflects trust in another’s benevolence, goodwill, integrity and/or willingness, as described by Mayer et al (1995).
For this research, the definition of Mayer et al (1995), but the terminology as noted by Nooteboom (1996) will be used to categorize the types of trust. Hence, we will use the terms competence trust and goodwill trust from now on. Note that the notion of Mayer et al (1995) about vulnerability is important, as this implies a source of risk.
2.6 Risk
As mentioned, control systems and trust can work as a safeguard for risk in hybrid governance situations, i.e. they can act as safeguard for risk in supplier-buyer relationships. (Das & Teng, 2001a; Speklé, 2001). Before moving on to determine the relationships between control, trust and risk, it is important to elaborate on risk itself. Following Das & Teng (2001a: p. 252), “risk is often conceptualized as variances of outcomes of importance to the risk-taking subject”. These variances of outcomes can stem from multiple sources.
Following Das & Teng (1996: p. 830), alliance success is influenced by ‘’a good working
relationship among the partners.’’ This good working relationship is a source of risk. I.e., an organization has no perfect knowledge about the quality of the future relationship. Hence, when entering a
relationship, there is relational risk (Coletti et al, 2005; Das &Teng, 1996; Nooteboom et al, 1997). Relational risk stems from potential opportunistic behavior from one of the collaborators, with all its consequences. Or, following Coletti et al (2005: p. 480) relational risk is ‘‘the probability that
11 Another source of risk involves performance risk, which encompasses all types of factors that could influence the performance of the alliance, ranging from lack of competence to bad luck (Coletti et al, 2005; Das & Teng, 1996, 2001a). Following Das & Teng (1996), the sources of relational risk and performance risk come from the domains of inter-firm cooperation and competitive environment respectively. Hence, if market factors are the source of risk, it is termed performance risk and if the risk arises from the specific collaboration relationship, it is termed relational risk.
A further distinction made by Das & Teng (2001a), is that perceived risk by collaborators is not the same as the real risk. This could be because of bounded rationality, or because of other factors, but perceived risk is by definition subjective. It is therefore important to note that in this paper the aim is on perceived risk and not on real risk.
To the reader, all these definitions can be confusing. Therefore, the following table was made to summarize the definitions mentioned above.
Construct Definition Authors
Control a process regulating behavior in reaching organizational goals Das & Teng (2001a)
‘’the various policies and procedures that firms use to mitigate different types of risks’’
Colletti et al (2005)
Output control "the measurement of outputs" Ouchi & Maguire (1975)
"measuring outcomes by relying on an accurate and reliable
assessment of members' performance" Das & Teng (2001a)
Behavior control "Control based on direct, personal surveillance" Ouchi & Maguire (1975) "behavior control is to ensure that the process is appropriate" Das & Teng (2001a)
Social /clan control
"developing shared values, beliefs and goals among members so that appropriate behaviours will be reinforced and
rewarded."
Das & Teng (2001a)
"it relies for its control upon deep level of common agreement between members on what constitutes proper behavior, and it requires a high level of commitment on the part of each individual to those socially prescribed behaviors"
12 Table 3: Summary of definitions
Now that there is a clear definition on the various types of control, trust and risk perception, the interplay between these three is addressed.
Trust ‘‘one’s perception of another’s trustworthiness”. Colletti et al (2005)
"the willingness of a party to be vulnerable to the actions of another party based on the expectation that the other party will perform a particular action important to the trustor,
irrespective of the ability to monitor or control that other party.”
Mayer et al (1995)
Competence trust "a partner's ability to perform according to agreements" Nooteboom (1996)
Trust in another's ability and/or competence Mayer et al (1995)
Goodwill trust intentions to perform according to agreements Nooteboom (1996)
Trust in another's benevolence, goodwill, integrity and/or
willingness Mayer et al (1995)
Risk "variances in outcomes of importance to the risk-taking subject" Das & Teng (2001a)
Relational risk "the probability that collaboration objectives will not be
achieved because of a lack of cooperation"
Colletti et al (2005)
Risk arising from inter-firm cooperation Das & Teng (2001a)
Performance risk
All types of factors that could influence the performance of the alliance, ranging from lack of competence to bad luck
13 2.7 Control, trust & risk interplay
Das & Teng (2001a) developed a useful risk, trust, control framework. In their framework, they made the same distinctions between the various forms of risk, trust and control as mentioned in the previous section. Their framework, and thus their propositions will be the departure for this paper. It is important to note here that when spoken of risk and trust, this implies risk perception and trust perception. As such, as discussed earlier, it does not imply real risk or real trust.
2.7.1 Control & risk perception
Following the distinction between formal control and informal (social) control, Das & Teng (2001a) provide arguments for several relationships between control and risk perception. As described in the previous section, relational risk is about potential opportunistic behavior of the partner.
Behavior control and relational risk
Following Das & Teng (2001a), behavior control and relational risk go hand in hand as relational risk in essence is “about the likelihood of underhanded and surreptitious activities” and therefore “it is difficult, ordinarily, to measure the outcomes of opportunistic behavior and relational conflicts in a precise and objective manner.” (Das & Teng, 2001a: p. 260). This complies with Ouchi (1979), who argues that when it is difficult to measure output, behavioral control is more appropriate. Following Dekker (2004), these behavior controls can be set ex ante through planning, procedures, rules and regulations and ex post through behavior monitoring and rewarding. It is important to remember what relational risk actually entails: "the probability that collaboration objectives will not be achieved because of a lack of cooperation" (Coletti et al, 2005). By making use of behavior controls as described by Dekker (2004), desirable behavior of the partner can be achieved. (Coletti et al, 2005; Das & Teng, 1998).
14 non-equity alliances.” Non-equity alliances, like relational contracting, are based on pre-specified
agreements and therefore rely more on output control (Das & Teng, 2001a; Sengün & Wasti, 2007). These pre-specified agreements are in themselves however ex ante behavior control mechanisms (Dekker, 2004). Hence, although output control is assumed to play a more important role than behavior control, behavior control is considered to affect relational risk more than output control. As such:
Hypothesis 1 (H1): In supplier-buyer relationships, the use of behavior controls reduces the level of relational risk.
Output control and performance risk
In contrast, performance risk is result oriented. Following Sengün & Wasti (2007: p. 436), “performance risk is the probability that the alliance objectives are not satisfied despite satisfactory cooperation among partner firms”. As this definition shows, performance risk is about meeting alliance objectives. As it involves meeting objectives, by definition the relevant control is output control. (Das & Teng, 2001a) Following Yan & Gray (1994), the objectives and performance indicators are the result of the bargaining process between the partners. Hence, these performance objectives are operationalized and made measureable before the actual partnership is formed. Following Dekker (2004), the output control can be ex ante through goal setting and reward / incentive systems and ex post through performance monitoring and rewarding. These control mechanisms are determined in the bargaining process of which the outcome depends on each partner’s bargaining power. (Yan & Gray, 1994).
Another necessary condition for output control to be effective is that there should at least be some knowledge of the transformation process (Eisenhardt, 1985; Ouchi, 1979). Following Das & Teng (2001a) there is limited knowledge about the transformation process concerning performance risk. It is known that performance risk emerges from market forces and/or a lack of competence (Coletti et al, 2005), but “how precisely these factors affect alliance functioning is difficult to comprehend (Das & Teng, 2001a: p. 261). The problem with behavior control is that it is not known what kind of process or behavior would lead to better performance. As such, output control fits with performance risk. Via output control, the key performance measures can be monitored closely and enables managers to react to high levels of performance risks. (Das & Teng, 2001a) Hence, the following hypothesis:
15 Informal control and risk
Das & Teng (2001a) argue in their paper that informal control would reduce relational risk and performance risk. Following Das & Teng (2001a: p. 262), in informal control: “the goal-setting process is decentralized and evolving in nature” and “in strategic alliances, social control is often used, because agreed upon goals do not exist.” This is exactly what is meant by Yan & Gray (1994) above, when they mentioned that the goals and performance measures are the result of a negotiation process. The argument is that during this consensus-making process, i.e. the negotiation, the parties become committed to the cause and its goals, because of socialization and agreed upon goals. Following Dekker (2004), an ex ante social control mechanisms is partner selection. As such, a supplier that has a
reputation of ‘being fair’ can mitigate relational risk by providing a certain promise and expectation of a truthful and fair commitment to the alliance. So, if a supplier has the reputation of being fair, this reduces relational risk. This reputation is a form of informal control, hence, common culture, shared values and a shared vision can mitigate relational risk. (Das &Teng, 2001; Dekker, 2004; Sengün & Wasti, 2007).
Although it is clear that in most alliances there is a consensus-making process, it must be said this could be less the case in traditional supplier-buyer relationships that approach the collaboration as spot transaction/contracting (Arnold, 2000; Sydow, 1992; Williamson, 1991) .However, as Sengün & Wasti (2007: p. 445) showed, organizations can also approach the collaborations more hierarchical, in which goal incongruence is solved by informal communications and creating “a family-like environment”. This again can be viewed as relational contracting (Sydow, 1992). In this way, informal controls do matter and have an impact on risk perception, as the established norms and values are becoming shared norms and values. Hence, the incentive for opportunistic behavior decreases (relational risk). Also, performance risk will tend to decrease because performance measures and goals are set together, rather than that the company with the strongest bargaining position presses for a specific measure or goal. (Das & Teng, 2001a; Sengün & Wasti, 2007). Following Das & Teng (2001a: p. 262) informal control “encourages partner firms to lay out reasonable and achievable goals for the alliance.” And if the firm’s objectives are reasonable and achievable, meeting these objectives will be more likely and as such, performance risk is reduced.
16
In supplier-buyer relationships:
Hypothesis 3a (H3a): Informal control reduces the level of performance risk. Hypothesis 3b (H3b): Informal control reduces the level of relational risk.
2.7.2 Trust and risk
Recall that trust comprises of two types of trust: goodwill trust and competence trust. Competence trust is based on the capability of the partner to meet the objectives and goodwill trust involves the partner’s intentions to perform according to the agreements and rules of the partnership (Das & Teng, 1998, 2001a; Nootenboom, 1996; Sengün & Wasti, 2007; van der Meer-Kooistra & Vosselman, 2000). Competence trust and performance risk
As competence trust involves the capability to meet the partnership’s objectives, it is quite logical that this should go hand in hand with performance risk. Recall that performance risk involves risk about meeting the partnership’s goals and objectives. More importantly, this performance risk stems from forces out of the competitive environment and lack of competence. (Das & Teng, 2001a, 2004) In
essence, competence trust thus involves the trust in one’s partner to meet the objectives by dealing with competitive forces or market pressures.
Following Das & Teng (2001a: p. 258),”because competence trust gives a firm a sense of confidence that the partner is capable of accomplishing given tasks in the alliance, performance risk of the alliance will be perceived as relatively low”. It is therefore not surprising that if a company has a reputation for being very competent, this also reduces performance risk. Important to note is that competence trust does not affect perceived relational risk, as relational risk entails the intention to perform according to agreement, not the ability per se. In literature there are some synonyms used for competence trust (Das & Teng, 2001a). For example, it is the same as ability trust (Mayer et al, 1995). Following Mayer et al (1995), ability trust affects the level of trust and risk perception. Mayer et al (1995) however does not distinguish between different types of risk perception. The reasoning of Das & Teng (2001a) and Mayer et al (1995) make sense however, hence, the following hypothesis:
17 Goodwill trust and relational risk
Following Das & Teng (2001a) and Nooteboom (1996), goodwill trust is expected to reduce relational risk, i.e. the perceived probability of the other party to behave opportunistically, as it involves the intention of the collaborator to perform according to agreement. In a sense, goodwill trust thus implies that partners can assume that employees in both firms will cooperate when a critical situation arises (e.g. when there is need for imminent supply due to shortage), and not try to take advantage of the situation (Sengün & Wasti, 2007). Lui & Ngo (2004) found in their study about the role of trust on cooperation that there is a negative relationship between goodwill trust and relational risk. This supports Das & Teng’s (2001a) notion.
Goodwill trust further corresponds with Mayer et al’s (1995: p. 718) benevolence construct: “benevolence is the extent to which a trustee is believed to want do good to the trustor, aside from an egocentric profit motive”. This implies that even when there is an economic incentive to take advantage of the partner, one believes the partner will not make use of this advantage. Especially in case of a critical situation, this is a moment that cooperation of the supplier is necessary. And by definition, relational risk is “the probability that collaboration objectives will not be achieved because of a lack of cooperation” (Coletti et al, 2005: p. 480) As such, goodwill trust affects the relational risk perception. Following Das & Teng (2001a), goodwill trust does not affect performance risk, because these two variables involve separate dimensions. Goodwill trust is about intention, and not about ability of the collaborator. Hence the following hypothesis:
Hypothesis 5 (H5): In supplier-buyer relationships, goodwill trust reduces the level of perceived relational risk.
2.7.3 Control and trust
18 Formal control and trust
Das & Teng (1998) argue that the effect of control on trust (or vice versa) is contingent on particular situations. Formal control can have a negative effect on trust because formal controls, like rules and regulations, undermine autonomy in decision making. It is argued that this creates “an atmosphere of mistrust”, which affects goodwill trust in a negative way. (Das & Teng, 2001a: p. 263). Also following Lau & Tan (1998), mistrust and resentment is created through extensive performance evaluation. If there is a high level of goodwill trust, there would be no need to set specific rules and regulations to control the relationship as one would believe that the partner has all the good intentions and will not behave opportunistically.
Also, following van der Meer-Kooistra & Vosselman (2000), trust replaces control in transactional relationships. Sengün & Wasti (2007) found in their study that indeed output control was negatively related with both goodwill and competence trust. Malhotra & Murnighan (2002) also argued that formal controls reduce trust. However, the study of Malhotra & Lumineau (2011) showed that although formal control reduces the level of goodwill trust it also increased competence trust. Das & Teng (2001a) on contrast argue that higher levels of output controls indicate that a firm does not fully trust the partner’s competence and therefore should be frequently and closely monitored. This implies that competence trust decreases when formal controls are increased.
The studies of Malhotra & Lumineau (2011), Malhotra & Murnighan (2002) and Sengün & Wasti (2007) focused on formal controls in general and did not make the distinction between output controls and behavior controls. Das & Teng (2001a) do make this distinction. They argue that behavior controls also reduce trust as it demonstrates that the partner cannot be trusted to do their part of the job in their own way to achieve the best possible outcome for the alliance. As such “since it is safe to assume that few firms like to be directed by other, close monitoring of behavior tends to generate tension that makes partners more skeptical of each other’s approaches. Competence trust will therefore be compromised” (Das & Teng, 2001a; p. 263) Hence, the following hypotheses:
Hypothesis 6a (H6a) : In supplier-buyer relationships, formal controls will decrease the level of goodwill trust.
19 Informal control and trust
Social controls on the other hand are argued to enhance trust as it increases mutual
understanding through the creation of shared norms, values and goals. With social controls, the partners rely less on outcome and behavior measures and more on socialization and frequent meetings. (Das & Teng, 1998; 2001a; Sengün & Wasti, 2007) Following Das & Teng (2001a: p. 264) “a sense of confidence in each other’s goodwill is implied in social control”, because there are no or limited pre-specified rules and procedures, both partners have more freedom. Following Ouchi (1979), social control is most appropriate in high trust situations, where it also prospers. This notion is supported in literature by Aulakh, Kotabe & Sahay (1996) as they found a positive relationship between informal control and goodwill trust. Following the findings from Sengün &Wasti (2007), social control indeed seems to be positively related with both goodwill and competence trust.
The same relationship is argued to apply for informal control and competence trust. The idea is that because there are no or limited pre-specified rules and procedures, the partners are given the autonomy to decide on appropriate behaviors and outcome measures. (Das & Teng, 2001a; Sengün & Wasti, 2007) Also, informal controls imply that there are shared values and norms that are enacted reciprocally. Through this mechanism, consensus is reached about goals and hence competence trust is increased. (Das & Teng, 2001a). Hence the following hypothesis:
Hypothesis 7a: In supplier-buyer relationships, informal control increases the level of goodwill trust. Hypothesis 7b: In supplier-buyer relationships, informal control increases the level of competence trust.
Effects trust on control
20 “reduces the level of resistance and brings harmony to the controller-controllee relationship” (Das & Teng, 2001a: p. 265). Also, following Vryza and Fryxell (1997), trust makes control mechanisms more effective. Das & Teng (2001a: p. 265) further argue that in absence of trust, a partner may not be willing to accept “outcome measurements, to follow specified behavior patterns and to share values. In this paper however the focus is on the use of controls, not the effectiveness per se. As such, Costa & Bijlsma-Frankema (2007) are followed, who argue that trust decreases the level controls and Sengün & Wasti (2007), who found partial support for this hypothesis:
Hypothesis 8a: In supplier-buyer relationships, goodwill trust and competence trust increase the level of informal control
Hypothesis 8b: In supplier-buyer relationships, goodwill trust and competence trust decrease the level of formal controls.
Mediation
The provided hypotheses suggest several potential mediating variables. Following Baron & Kenny (1986: p. 1176), a variable acts as a mediator if three conditions are met:
“(a) variations in levels of the independent variable significantly account for variations in the presumed mediator (i.e. Path a), (b) variations in the mediator significantly account for variations in the dependent variable (i.e. Path b), and (c) when Paths a and b are controlled, a previously significant relation between the independent and the dependent variables are no longer significant”.
Relational risk is considered to be affected primarily through goodwill trust. However, the relation may be mediated through behavior control:
H9a: The effect of goodwill trust on relational risk is mediated by behavior control
The other way around is also a possibility, as goodwill trust and behavior control are hypothesized to affect each other in both directions. Hence:
H9b: The effect of behavior control on relational risk is mediated by goodwill trust.
21
H9c: The effect of competence trust on performance risk is mediated by output control.
However, as competence trust is also dependent of output control, the other way around also applies. Hence:
H9d: The effect of output control on performance risk is mediated by competence trust.
Performance risk and relational risk are hypothesized to be dependent of Informal control, but also, respectively, of competence trust and goodwill trust. As such influences may be mediated:
H9e: The effect of informal control on performance risk is mediated by competence trust H9f: The effect of informal control on relational risk is mediated through goodwill trust.
2.8 Conceptual model
It is important to emphasize again that these hypotheses are derived from Das & Teng (2001a). To help the reader, the table summarizes the hypotheses and it is followed by an elaborate conceptual model.
Hypothesis Constructs Expected relationship
H1 Behavior control --> relational risk Negative
H2 Output control --> performance risk Negative
H3a Informal control --> performance risk Negative
H3b Informal control --> relational risk Negative
H4 Competence trust --> performance risk Negative
H5 Goodwill trust --> relational risk Negative
H6a Formal control --> goodwill trust Negative
H6b Formal control --> competence trust Negative
H7a Informal control --> goodwill trust Positive
H7b Informal control --> competence trust Positive
H8a Trust --> informal control Positive
H8b Trust --> Formal control Negative
H9a: Goodwill Trust --> relational risk, mediated by behavior control H9b Behavior control --> relational risk, mediated by goodwill trust H9c Competence trust --> performance risk, mediated by output control H9d Output control --> performance risk, mediated by competence trust H9e Informal control --> performance risk, mediated by competence trust H9f Informal control --> relational risk, mediated by goodwill trust
22 Note: control , trust and risk constructs
Figure 1: conceptual model Output control
Informal control
Behavior control Relational risk
23
3.
Methodology
As discussed in previous sections, there has been much research about the control-trust interplay. (Bachmann, 2001; Costa & Bijlsma-Frankema, 2007; Das & Teng, 2001a; Dekker, 2004; Van der Meer-Kooistra & Vosselman, 2000, Vélez et al, 2008; Vosselman & Van der Meer-Meer-Kooistra, 2009).Also, research investigating the control – risk relationship is numerous (Das & Teng, 2001a; Sengün & Wasti, 2007).
These theories have proposed certain relationships between these constructs. As such, following Eisenhardt (1989), these prior studies were involved in the first part of the empirical cycle, i.e. they were involved in inductive research. Aken, Berends & van der Bij (2012: 2.2 p. 3), describe that the second part of the empirical cycle follows the deductive research approach, and here “‘the most promising ideas of the induction step are transformed into hypotheses”. Following Aken et al (2012), the main reason for theory testing is a certain business phenomenon that is faced by organizations of which existing literature and its evidence is still inconclusive.
Following this line of reasoning, a quantitative deductive approach is appropriate as there is not yet consensus about the control-trust interplay, and because this involves a relevant business phenomenon. That is, if the relationship between control and trust would be conclusive, organizations could use this theory in practice. This paper extends the control-trust interplay by also looking at the interaction with risk perception, as proposed by Das & Teng (2001a). Hence, this paper adopts a quantitative
methodology to analyze the proposed hypotheses.
Pretest
For this research a survey was developed: questions are based on surveys from prior studies. To ensure comprehensibility, determine the time needed for the survey and to hear about potential
comments on the survey, a pretest was conducted. Also, the accompanying “introduction letter”, written following the guidelines provided by Dillman (2007), was pretested to check whether it was
24 Sample
This study is targeted at the technology/production industry. It has been argued that this industry and its technology have become so complex that firms cannot handle technology development and production on their own anymore (Bougrain & Haudeville, 2002). Following Fusfeld (1986: p. 144) firms have been touched by “the declining technical self-sufficiency”. As such, firms enter inter-firm collaborations to overcome these problems. This drive for inter-firm collaborations is further pushed by public authorities of OECD (Organisation for Economic Co-operation and Development) countries who “have laid increasing emphasis on inter-firm collaboration in advanced technology. Most of the time these programmes urge large firms located in the same geographical area to co-operate” (Bougrain & Haudeville, 2002: p. 736). Also, following Teece (1992) the technology sector has high uncertainty and therefore the need for coordination is great. As such, inter-firm collaborations are used to a broad extent. Because of the above mentioned reasons, the technology/production industry is a fruitful industry for this research as it involves a lot of inter-firm collaborations.
The target companies were derived mainly from the website of FME (http://www.fme.nl/). FME is a Dutch association representing the Dutch technology/production industry, with 2400 companies tied to it. The companies represented by FME are all based in The Netherlands, covering all provinces. As such, there is not made a distinction between small, large or international companies. From the website of FME, the websites of the companies were derived, and consequently the contact information. If there were not contact names available, this information was derived via LinkedIn. The targeted contacts all were in jobs that are concerned with inter-firm collaborations. As such, Owners, Chief Executive Officers, Chief Operating Officers, General Managers, Operational Managers, Project Managers and Purchasing Managers were contacted. By contacting multiple sources of informants that have knowledge about the subject, this likelihood of single-informant bias was decreased. (Zaheer et al., 1998) There was no distinction made by either age or gender. Only the job title was relevant. If both the company website and LinkedIn did not provide contact information, then the mail was sent without a specific addressee. Data collection
25 Also a reminder was sent to the companies. In total 1617 surveys were distributed, 84 were filled in, resulting in a response rate of 5.2%.
Measurements
Control
The measurements in this study follow Das & Teng’s (2001a) framework, which entails three categories: control, trust and risk. In accordance with Eisenhardt (1985) and Ouchi (1979), control was divided in output control, behavior control (both constituting formal control) and informal control . Following Dekker (2004) and Sengün & Wasti (2007: p.449), output control is operationalized as: “(a) the effort required in setting objectives and performance measures for clarifying expectations during the initial development of the relationship and (b) ex post information processing and performance evaluation during the relationship to manage unresolved control problem”. Based on this
operationalization, six survey items were adapted from Buvik & Reve (2002) and Sengün & Wasti (2007) to measure output control.
Behavior control is concerned with (a) pre-specified structural specifications like planning, procedures, rules & regulations and (b) behavior monitoring and evaluation (Das & Teng, 2001a; Dekker, 2004). As such, 15 survey items that covered this domain of controls were derived from Buvik & Reve (2002) and Desphande & Zaltman (1982).
The last variable is informal (social) control. Following Dekker (2004), social control is comprised of reputation, joint decision making and joint problem solving, which is acknowledged by Das & Teng (2001a) and Sengün & Wasti (2007). Hence, 18 survey items that covered this domain were adapted from Jonhson, McCutcheon, Stuart & Kerwood (2004), Johnson, Cullen, Sakano & Takenouchi (1996) and Sengün & Wasti (2007).
Risk
26 Performance risk was defined as the risk arising from market factors (Das & Teng, 2001a). To measure performance risk as result of market factors, the survey addresses questions regarding market uncertainty and stability. Hence, there were 8 measurements adapted from prior studies. (Buvik, 2002; Buvik & Gronhaug, 2000; Jap, 1999; Skarmeas, Katsikeas, Spyropoulou & Salehi-Sangari, 2008 ;
Trust
Following the distinction made by Das & Teng (2001a), trust comprises of goodwill trust and competence trust. Following Mayer et al (1995) , Nooteboom (1996) and Zaheer et al (1998) goodwill trust is about trust in one’s intentions and willingness to adhere to made agreements. Also, goodwill trust involves to not take advantage of critical situations (Sengün & Wasti, 2007). With this in mind, 11 survey items were derived from prior studies (Mayer et al, 1995)
The second form of trust is competence trust. Instead of intentions, this form of trust is about one’s trust in the partner’s ability and competence to perform to standard as agreed upon in the agreement. (Mayer et al, 1995; Nooteboom, 1996) So, the survey items should measure the degree of trust in the partner’s competence. From Mayer et al (1995), 6 items were derived.
27 Control variables
Three control variables are included in the study: the length of the relationship, dependence on the relationship and trust propensity. The notion behind relationship length is that the longer the relationship, the better both seller and buyer know each other and as such, are able to better evaluate and estimate potential opportunistic behavior. Hence, risk assessment is influenced. Also, the longer the relationship, the more trust can develop and the need for certain controls can be influenced. (Sengün & Wasti, 2007)
Following Izquierdo & CIllian (2004), interdependence and trust influence the functioning of a relationship between partners. If a partner is highly dependent of another partner in terms of firm survival, it influences the necessity to foster the relationship. This could entail having stricter controls, but it could also mean building a more informal relationship to create a bond. Specifically, in high trust and high dependence situations, partners feel the need for continuity and fostering of the relationship (Izquiredo & Cillian, 2004). In any case, the relationship is affected. As such, dependence is added as control variable: 9 items were drawn from Handfield & Bechtel (2002), Heide & John (1988) and Noorderhaven, Nooteboom & Berger (1998), who incorporated dependence in their studies.
Finally, trust propensity is a factor that influences risk perception and trust perception. As such, it also affects the need for certain controls. Following Colquitt, Scott & LePine (2007: p. 910), trust propensity is defined as : “a stable individual difference that affects the likelihood that a person will trust”. As such, trust propensity may vary per individual and therefore, not everyone will place the same amount of trust in a particular person, or situation. This is important, as this can influence trust and risk perception. If these are affected, controls are affected as well. Hence, 8 survey items were derived from Mayer & Davis (1999) to control for this effect.
Analysis
28
4.
Results
Factor analysis
The trust construct was analyzed using principle component analysis (PCA) with Varimax
rotation. Kaiser-Meyer-Olkin Measure of Sampling Adequacy (KMO) and Bartlett’s Test of Sphericity was performed to assess whether the data was suitable for factor analysis. KMO was above 0.5 with
Bartlett’s significant at p < .01. As such, the data is suitable for factor analysis. (Williams, Brown & Onsman, 2010)
PCA demonstrated that the trust construct loaded on two separate factors. An eight-item measurement is loaded on goodwill trust, with a Cronbach Alpha of 0.848. A four-item measurement loaded on competence trust, with a Cronbach Alpha of 0. 836.Also a PCA with Varimax rotation was done for the construct risk perception. This resulted in two factors: eleven measurement items loaded on relational risk with a Cronbach Alpha of 0.880; and a four-item measurement was associated with performance risk, with a Cronbach Alpha of 0.702.
The control construct loaded on three factors using PCA with Varimax rotation. Output control has a five-item measurement, with a Cronbach Alpha of 0.818. Five measurement items were associated with behavior control and have a Cronbach Alpha of 0.740.
Finally, informal control was associated with eight measurement items and a corresponding Cronbach Alpha of 0.806. The Cronbach Alpha and associated measurement items can be found in Appendix II. In Appendix I, the factor analyses for the constructs can be found. In table 2 the descriptive statistics of the constructs can be found. Here can be seen that there are numerous bivariate
29 Table 3: Descriptive statistics and construct correlations
Note: * = significant at p < 0,05 ; ** = significant at p < 0,01. N= 84
DESCRIPTIVE STATISTICS AND CONSTRUCT CORRELATIONS
30 Hypothesis testing
Using a hierarchical multiple regression the hypotheses were tested. The control variables were included in the ‘first block’ and the other variables were added to the ‘second block’. As such, the change in R2 and its significance can be analyzed. This allows for a check of the effect of the control variables.
Relational risk as dependent variable
Table 4: Relationship behavior control, informal control, goodwill trust and relational Risk
B SE β Sig. Tolerance VIF
Behavior Control -,043 ,086 -,058 ,616 ,924 1,083 Informal Control -,297 ,181 -,227 ,106 ,641 1,560 Goodwill Trust -,302 ,167 -,249 ,076 ,641 1,559 Dependence -,113 ,115 -,155 ,193 ,781 1,280 Trust Propensity -,180 ,137 -,123 ,330 ,884 1,131 Collaboration Length -,003 ,007 -,049 ,668 ,938 1,066
Dependent variable: Relational Risk
31 Performance risk as dependent variable
Table 5: Relationships output control, informal control, competence trust and performance risk
B SE β Sig. Tolerance VIF
Output Control -,011 ,114 -,012 ,921 ,967 1,034 Informal Control ,119 ,229 ,076 ,605 ,696 1,436 Competence Trust -,141 ,194 -,103 ,472 ,744 1,343 Dependence ,105 ,139 ,098 ,452 ,898 1,114 Trust Propensity ,247 ,167 ,190 ,145 ,903 1,108 Collaboration Length ,002 ,009 ,029 ,140 ,931 1,074
Dependent variable: Performance Risk
The regression for output control, informal control, competence trust and performance risk (See table 5) shows that the model is not a good predictor of performance risk (R2 = .053). As such, the model only predicts 5.3 % of the variance. The change in R2 is 0.9%, which is minimal and is also insignificant. Although competence trust and output control appear to be negative related (respectively: β = -.103 and β = 0.012), the relations are both insignificant ( p > 0.05), thus H2 and H4 are rejected. Informal control shows a slight positive relation with performance risk (β = .076) but is also insignificant (p > 0.05). Hence, H3a is rejected.
Goodwill trust as dependent variable
Table 6: Relationships among output control, behavior control and goodwill trust
B SE β Sig. Tolerance VIF
Ouput Control -,167 ,083 -,251 ,048 ,841 1,190
Behavior Control ,051 ,082 ,079 ,537 ,801 1,248
Dependence ,200 ,083 ,278 ,019 ,979 1,022
Trust Propensity ,159 ,115 ,163 ,174 ,930 1,076
Collaboration Length -,005 ,006 -,096 ,419 ,941 1,063
Dependent variable: Goodwill Trust
32 predictor of goodwill trust. Also, output control is a significant predictor of goodwill trust (β = -.251, p < .05). Behavior control on the other hand is not significantly related with goodwill trust (β = .079, p > 0.05). So, as output control is significant, but behavior control is not, there is partial support for H6a.
Table 7: Relationship informal control and goodwill trust
B SE β Sig. Tolerance VIF
Informal Control ,577 ,116 ,519 ,000 ,906 1,104
Dependence ,107 ,075 ,147 ,159 ,918 1,089
Trust Propensity ,172 ,093 ,186 ,070 ,962 1,040
Collaboration length ,001 ,005 ,020 ,848 ,945 1,059
Dependent variable: Goodwill Trust
The regression model for informal control and goodwill trust (table 7) is significant with relatively strong predictability (R2 = .594, p < 0.01). The R2 implies that 59.4 % of the variance in goodwill trust can be explained by the model. Adding informal control to the model results in a significant (p < 0.01) change in R2 of 24.4 %. As such, informal control should account for 24.4% of the variance in goodwill trust. Informal control is significantly positive related to goodwill trust (β = .519, p < 0.01), hence we accept hypothesis H7a.
Competence trust as dependent variable
Table 8: Relationships among output control, behavior control and competence trust
B SE β Sig. Tolerance VIF
Output Control -,027 ,086 -,041 ,750 ,827 1,209
Behavior Control -,047 ,084 -,072 ,582 ,806 1,240
Dependence ,129 ,087 ,174 ,145 ,981 1,020
Trust Propensity ,212 ,120 ,214 ,081 ,932 1,072
Collaboration Length ,001 ,006 ,010 ,934 ,941 1,063
Dependent variable: Competence Trust
33
Table 9: Relationship informal control and competence trust
B SE β Sig. Tolerance VIF
Informal control ,540 ,124 ,472 ,000 ,913 1,095
Dependence ,038 ,081 ,051 ,639 ,916 1,092
Trust propensity ,191 ,101 ,200 ,062 ,959 1,043
Collaboration length ,004 ,006 ,075 ,483 ,946 1,057
Dependent variable: Competence Trust
The prediction model for competence trust is significant (p < .01) and accounts for 27.5 % of variance in competence trust (R2 = .275). There is significant change in R2 (p < .01) of 20.4 %. Informal control is significantly and positively related with Competence trust (β = .472<, p < .01). (See table 9). As such, H7b is supported: informal control is positively related with competence trust.
Informal control as dependent variable
Table 10: Relationship trust and informal control
B SE β Sig. Tolerance VIF
Competence Trust ,540 ,124 ,472 ,000 ,913 1,095
Goodwill Trust ,375 ,097 ,417 ,007 ,771 1,297
Dependence ,066 ,065 ,102 ,313 ,904 1,106
Trust propensity -,144 ,082 -,174 ,083 ,928 1,077 Collaboration length -,005 ,004 -,113 ,250 ,960 1,042
Dependent variable: Informal Control
When Informal control is entered as dependent variable with Goodwill trust and Competence trust as predictors, the regression shows a significant model (R2 = .412, p < .01). Adding Goodwill trust and Competence trust results in a change in R2 (p < .01) of 31.8 %. Goodwill trust is significantly positively related with Informal control (β = .417 , p < .01). Competence trust is also significantly positively related with informal control (β = .293 , p < .01). As such, H8a is fully accepted (See table 10).
Formal control as dependent variable
Table 11: Relationship trust and output control
B SE β Sig. Tolerance VIF
Competence Trust ,031 ,198 ,021 ,876 ,830 1,205
Goodwill Trust -,363 ,205 -,240 ,082 ,787 1,271
Dependence ,099 ,137 ,091 ,471 ,912 1,096
Trust propensity ,011 ,177 ,007 ,952 ,928 1,077
Collaboration length -,008 ,009 -,111 ,370 ,962 1,040
34 The regression with output control as dependent variable showed that the model is insignificant (R2 = .060, p > .05). Not surprisingly therefore, competence trust and goodwill trust are insignificantly related with output control (both p > .05). Goodwill trust is, as expected, negatively related with output control (β = -,240). Competence trust on the other hand shows a minimal positive relationship (β = .021). (table 11).
Table 12: Relationship trust and behavior control
B SE β Sig. Tolerance VIF
Competence Trust -,102 ,204 -,067 ,617 ,808 1,238
Goodwill Trust ,001 ,214 ,001 ,995 ,767 1,304
Dependence ,158 ,143 ,140 ,274 ,902 1,108
Trust Propensity ,203 ,188 ,136 ,284 ,917 1,090
Collaboration length ,004 ,010 ,057 ,647 ,957 1,045
Dependent variable: Behavior Control
The regression with behavior control as dependent variable shows similar results. The model is insignificant (R2 = .044, p > .05) and good will trust and competence trust are both insignificant related (p > 0.05), with respectively a minimal positive relationship (β = .001) and a weak negative relationship (β = -.067). (Appendix VI, table 9). As such, H8b is rejected.
Mediation
Recall, following Baron & Kenny (1986: p. 1176), a variable acts as a mediator if three conditions are met: “(a) variations in levels of the independent variable significantly account for variations in the presumed mediator (i.e. Path a), (b) variations in the mediator significantly account for variations in the dependent variable (i.e. Path b), and (c) when Paths a and b are controlled, a previously significant relation between the independent and the dependent variables are no longer significant”.
35 When output control was tested as a dependent of competence trust, it resulted in an insignificant relation between competence trust and output control. (β = .021, p > .05) As output control here is de potential mediation variable, condition (a) is violated and therefore H9c is rejected. When competence trust is the dependent variable and output control the independent, there is also an insignificant relationship (β = -.041, p > .05) and therefore, again condition (a) is violated, hence H9d is rejected. H9e involves competence trust as mediator and informal control as independent. The test showed a significant relationship between informal control and competence trust (β =.472, p < .01), hence condition (a) is satisfied. Condition (b) however is violated, taking performance risk as dependent
variable and competence trust as independent variable results in an insignificant relationship (β =-.103, p > .05). Therefore, competence trust does not mediate the relationship between informal control and performance risk. Hence, H9e is rejected. The same applies for H9f: when goodwill trust (mediator) is taken as dependent variable of informal control, this yields a significant relationship (β =.519, p < .05), hence condition (a) is satisfied. When relational risk is taken as dependent variable of goodwill trust however, this yields an insignificant relationship (β =-.249, p > .05). Therefore, condition (b) is not met and H9e is rejected.
36 -.012 .076 -.227 -.103 -.058 -.249*** -.251** -.072 .519* .472* .412* -.240*** -.067
Figure 2: Conceptualized results
Note: control , trust and risk constructs. (* = p < .01 ; ** = p < .05 ; *** = p < .10) Output control
Informal control
Behavior control Relational risk
Performance risk Goodwill trust
Competence trust Formal control
37
5.
Discussion
The conducted study analyzed propositions of Das & Teng (2001a) about the interplay between trust, control and risk. The analysis focused on supplier-buyer relationships in the technology/production industry in the Netherlands. Data was collected and then analyzed using statistics.
Das & Teng (2001a) proposed a negative relationship between behavior control and relational risk. The analysis only showed a very weak negative relationship, but was far from significant and as such the proposition was not supported. Sengün & Wasti (2007) found in a similar study like this, but then in the pharmacy industry in Turkey, that behavior controls were not applicable to their supplier-buyer
relationships. Das & Teng (2001a) also argued that behavior controls are moderately applicable to non-equity alliances, which also includes relational contracting like supplier-buyer relationships. It is argued that “as long as standard products and services are provided, the cooperation should remain
satisfactory” (Das & Teng, 2001a: p. 269). This implies that relational risk should stay relatively low in these types of alliances. The absence of proof of a relationship between behavior control and relational risk confirm the argument of Das & Teng (2001a) and the findings of Sengün & Wasti (2007). Also Langfield-Smith & Smith (2003), focusing on an outsourcing relationship involving IT, found that control was achieved through outcome controls and social control. Again, behavior controls do not play a role. As such, it seems that indeed behavior controls are not or moderately applicable to supplier-buyer relationships in general. Specifically, in supplier-buyer relationships in the technology industry in the Netherlands, behavior controls seem inappropriate.
38 potential opportunistic behavior (i.e. relational risk). By increasing the sample size, the effect of goodwill trust on relational risk might become clearer in future research.
Next, this study did not find the relationships with performance risk as proposed by Das & Teng (2001a). Competence trust and output control showed the proposed negative relationship, but there was no statistical significance. In contrast to prediction, informal control showed a slight positive relationship with performance risk but also proved insignificant. Clearly, performance risk must be controlled in another way in this setting. Following Das & Teng (1996) engaging in non-equity alliances controls for, and minimizes performance risk in itself. The argument here is that performance risk is shared with other firms. This is acknowledged by Brouthers (1995), who found in his study that if performance risk is high, firms try to minimize and control this risk by moving it to other firms through collaborations. The notion behind this is that performance risk arises from market forces. Hence, entering an inter-firm alliance helps share the present performance risks among two partners (Das & Teng, 1996). In contrast, relational risk would increase by engaging in inter-firm cooperation, as this arises, by definition, from the collaboration. The need for the sharing of performance risk is further stressed by Hagedoorn (1993) and Pisano (1991) who state that firms that are involved with a lot of R&D activities are specifically interested in these alliances because these markets are considered unstable and uncertain. The technology industry involves a lot of R&D and therefore this implies that this industry is interested in sharing performance risks through non-equity alliances. By entering an alliance, it is therefore argued that this reduces performance risks for the collaborators. Following this line of reasoning, one can argue that the alliance itself is a control mechanism for performance risk in the technology sector. If the alliance in itself is considered as a control mechanism for performance risk, this could be an explanation for why the model was not able to predict the influence of control on
performance risk in a significant way: performance risk is already considered controlled for via partner selection and engaging in the alliance itself. Empirical research could validate this in the future.
39 1996; 2001a; Sengün & Wasti, 2007).Aulakh et al (1996) neither found a significant relationship between behavior control and trust. Formal controls were not found to be related with competence trust which can be explained in terms of partner selection: “the initial selection of the outsourcer is based on perceptions of competence trust” (Langfield-Smith & Smith, 2003: p. 287). This implies that if a collaboration is formed, a certain level of competence trust must already be present. As such, formal controls fail to explain competence trust to a significant extent. It might be worthwhile to look at the relationship between trust and partner selection in more detail in future research.
Another interesting finding emerged from this study: control variable dependence appeared to be significantly positively related with goodwill trust. This indicates that if a buyer is relatively dependent on its supplier, the buyer trusts the supplier to not take advantage of this position. As such, goodwill trust is higher in high dependence situations. Indeed, Andaleeb (1995) found that higher levels of dependence lead to greater intentions to co-operate. Also, Svensonn (2004) indicates that trust is influenced by dependence. Dependence was not within the scope of this paper, but it became clear that it could help to explore this relationship further.
Following Das & Teng (2001a), informal control should be positively related with trust. The results show a strong significant positive relationship between informal control and trust. As such, the results support the proposition of Das & Teng (2001a), the ideas of Ouchi (1979) and the study of Sengün & Wasti (2007), who found a similar relationship. These findings also comply with the findings of Aulakh et al (1996), who found that social control helps to build trust in an alliance. It therefore seems that informal control is a way to enhance trust in an alliance setting.
40
6.
Conclusion
Both academics and practitioners stress the importance of inter-organizational relationships to gain competitive advantage and share risk (Coletti et al, 2005; Das & Teng, 1996; Nooteboom et al, 1997). To gain the benefits of inter-organizational relationships, effective governance is essential (Ittner et al, 1999). Trust and control are mechanisms to mitigate risk in these inter-organizational relationships (Das & Teng, 2001a; Speklé, 2001). As such, Das & Teng (2001a) developed a trust, control and risk framework proposing relationships between these constructs. This study empirically tested this model in supplier-buyer relationships in the technology industry in the Netherlands.
Informal control was found to be positively related with trust, implying complementarities between control and trust. This means that informal control is a mechanism to build trust. Although this study did not find conclusive evidence for a negative relationship between control and trust, results do suggest that there is somewhat of a trend towards a negative relationship, which would imply a substitutive relationship. As both types together constitute control, it is too straightforward to say that control either reduces trust or increases trust. There is indeed a more complex relationship between trust and control, like Dekker (2004), Vélez et al (2008) and Vosselman & Van der Meer-Kooistra (2009) argued. As such, this study adds to the existing fields of research examining the control and trust interplay. As proposed by Das & Teng (2001a), there was evidence found that goodwill trust acts as determinant of relational risk. Neither trust nor control acted as determinant of performance risk, which contrasts the view of Das & Teng (2001a). Further research in the determinants of risk perception is however required. This study also finds evidence supporting Sengün & Wasti (2007) that behavior controls do not seem to play a major role in supplier-buyer relationships. This was also suggested by Das & Teng (2001a). Hence, this study finds that the Das & Teng (2001a) framework partially fits in the studied context.
