Exploring the Influence of the Propensity to Trust on Control and Information Sharing Within Inter-Organizational Relationships

(1)

Master’s Thesis

Exploring the Influence of the Propensity to Trust on Control and Information

Sharing Within Inter-Organizational Relationships

by

W.H. VAN DER PLOEG

S3016390

MSc BA – Organizational & Management Control

Faculty of Economics and Business

University of Groningen

Supervisor

A. Rehman Abbasi Co-assessor

dr. M.P. van der Steen

19-06-2017

(2)

2

Abstract

Organizational studies rarely investigate the influence of personal characteristics on the effectiveness of inter-organizational relationships. This study examines how the propensity to trust influences control mechanisms and information sharing amongst partners. Based on motivation crowding theory hypotheses were built on the relationship between control mechanisms and information sharing, as well as on the moderating effect of the propensity to trust on this relationship. The hypotheses are tested using survey data gathered from 87 inter-organizational relationships. The findings indicate that the propensity to trust positively moderates the relationship between social control mechanisms and information sharing. The effectiveness of social controls in increasing information sharing thus depends on the propensity to trust. Due to this positive moderating effect social controls have a positive influence on information sharing. Furthermore, the findings show that the propensity to trust itself also directly contributes to the information exchange between partners. These findings indicate the importance of having people managing inter-organizational relationships with a strong propensity to trust in practice.

(3)

3

1. Introduction

In recent years there has been a dramatic increase in the formation rate of inter-organizational relationships (IORs) (Mellewigt, Madhok, & Weibel, 2007; Nicolaou, Sedatole, & Lankton, 2011). Firms establish IORs to share knowledge and increase their competitiveness. However, there is a high failure rate in such alliances, and scholars often cite a lack of cooperation as one of the main causes (Das & Teng, 1998). Academic researchers therefore consider cooperation to be crucial for successful IOR performance (Chen, Chen, & Meindl, 1998; Faerman, McCaffrey, & Van Slyke, 2001).

Subsequently, the exchange of information between partners is considered vital for cooperative behaviour and successfully managing IORs (Das & Teng, 1998; Mahama, 2006). However, there are many factors that can influence information sharing within IORs (Yang & Maxwell, 2011). Two of the most discussed factors in literature are trust and control. According to Das and Teng (1998) trust and control together create confidence between the partners which is key for cooperative behaviour such as information sharing.

The key role of trust within IORs is acknowledged by many researchers which argue that interfirm trust is necessary for partners to cooperate and even for the IOR in general to function (Das & Teng, 1998; Duarte & Davies, 2004; Payan & Svensson, 2007). However, there is one aspect of trust which is not considered in these studies. This gap in previous studies was recently addressed by Lu & Yan (2016) who argue that there has been little research on the effects of personal characteristics on active trust development between organizations. A personal characteristic that can influence trust development and cooperation between partners is the propensity to trust. The propensity to trust can be defined as “a stable within-party factor that influences the degree of trust that a trustor will have on the trustee without any prior information or history of exchanges being available”

(Nambudiri, 2012, p.978). There have been several organizational studies in regards to the

propensity of trust (Bernerth & Walker, 2009; Nambudiri, 2012; Chiu & Ng, 2015). However, these were mostly intra-organizational studies focused on the relationship between managers and

employees. Lu & Yan (2016) therefore argue that advanced research should integrate the propensity to trust with organizational control to increase understanding of trust between organizations. In contrast to trust, the role of control in IORs is more complex and often discussed in literature (Dekker, 2004; Vélez, Sànchez, & Álvarez-Dardet, 2008). Control is induced through various policies, procedures and governing mechanisms. Most of the research focused on control debate whether it has a positive or negative influence on trust building and cooperation within IORs. A longitudinal case study performed by Vélez et al. (2008) showed that even when trust is well established, management control systems enable conditions that directly build trust and contribute to cooperation. However, there are also studies that argue a negative influence. Tomkins (2001) discussed that the introduction of new management control systems in certain stages of an IOR can cause harm. Thus, it is still unclear how control mechanisms eventually affect cooperative behaviour, such as information sharing, between partners. Lu and Yan (2016) argued that different levels of the propensity to trust among individuals could explain these different results in previous studies on the relationship between trust and control. Therefore, this study will look into the effects of the

propensity to trust on control mechanisms.

Furthermore, theories used to explain control in IORs, such as Transaction Cost Economics (TCE), have a too narrow focus to fully explain the effects of control mechanisms (Dekker, 2004). According to Dekker (2004), the prediction made by TCE is insufficient to adequately explain the management and control of IORs. One of these reasons is that TCE takes too little account of the social mechanisms of governance, due to a lack of dynamism. It is important to take these mechanisms into account because IORs often are developed in a social context (Dekker, 2004). A theory that potentially takes these social mechanisms in account is the Motivation Crowding Theory (MCT). MCT originates from the field of psychology and sociology and suggests that external

(5)

5

Based on MCT, Christoffersen and Robson (2017) concluded that firms need to be careful when using external interventions, and think of the consequences on trust-building when using these within IORs. However, they also adressed that future work is necessary to identify how IORs may be disrupted by external interventions (Christoffersen & Robson, 2017). Additionally, Festré and Garrouste (2015) emphasized the importance of understanding the effects of external interventions on intrinsic motivation for managing such interventions, and argued that the debate concerning MCT is still open. This study will therefore draw on MCT to study the influence of control mechanisms on the exchange of information between partners. By examining control mechanisms as external interventions, which possibly influences the motivation of a partner to exchange information, knowledge is increased on the effects of control mechanisms within IORs.Furthermore, the

interaction effect of the propensity to trust on this relationship will be examined to address the gap noted by Lu & Yan (2016), and to contribute to the current debate on the relationship between trust and control. All in all, this leads to the following research question:

RQ: “What is the relationship between control mechanisms and information sharing in inter-organizational relationships, and how does the propensity to trust moderate this relationship?” The contribution of this study is twofold. First, it contributes to organizational trust literature by specifically examining the influence of the propensity to trust. Second, the social perspective this study takes on the use of controls in IORs will give a better understanding about the effectiveness of control mechanisms on information sharing with partners. This is important for organizations since increasing the information exchange with partners eventually increases the chance that an IOR will be a succesful partnership.

The remainder of the paper is structured as follows. In section 2 previous IOR-related literature on information sharing and the concepts of control and the propensity to trust will be discussed and used to develop hypotheses. Section 3 lays out details on the research methods. In section 4 the results of the empirical analysis are reported. Finally, section 5 provides theoretical and practical implications, as well as limitations and suggestions for future research.

2. Literature Review

There are many factors that influence the exchange of information between partners in IORs. This section will first elaborate on these factors and the role that information sharing has within IORs. Second, MCT will be discussed and used as a basis to form hypotheses on the relationship between different forms of control and information sharing. Finally, previous literature and models concerning the propensity to trust will be examined to develop hypotheses about its moderator effect on the aforementioned relationship.

2.1 Information Sharing

According to Tomkins (2001) there are two types of information in IORs. Type 1 concerns information on competence and integrity in action and communication, which relates to the amount of trust placed in a partner. Type 2 information is concerned with creating a collaborative mastery of events and planning what each party is going to do within the IOR (Tomkins, 2001). This research focuses on the latter where information sharing concerns the willingness to share important, possibly

proprietary, information with each other (Dekker, Sakaguchi, & Kawai, 2013). Dekker, Ding and Groot (2016) recently summarized the benefits of information sharing as follows: “Sharing

(6)

6

Thus, the exchange of information is crucial to coordinate and cooperate with a partner (Nicolaou & Christ, 2016). However, effectively exchanging information within IORs is often considered a

challenge. Researchers argue that there can be complex interactions involved in information sharing between partners due to their difference in origins, values and cultures (Yang & Maxwell, 2011). Additionally, firms may be reluctant to share information because of other reasons such as concerns with security, privacy and intellectual property (Li, Sikora, Shaw, & Tan, 2006). Therefore, the amount and effectiveness of information sharing within IORs has often been linked to the amount of trust between the partners (Tomkins, 2001). The trust developed between partners to share information comes from different sources and can decrease easily when there are for instance concerns with autonomy loss or when information is misused by a partner (Yang & Maxwell, 2011). In literature there are certain types of trust discerned which play an important role in inter-organizational information sharing (Rousseau, Sitkin, Burt, & Camerer, 1998; Yang & Maxwell, 2011). However, these types of trust are all based on prior information about the partner, or based on previous interactions with the partner, like calculus-based trust and relational trust. So, the propensity to trust of individuals involved in the IOR is not mentioned in literature concerning information sharing. This is a gap since it potentially can have impact on the willingness to rely on trust and subsequently on exchanging information. This is also noted by Lu and Yan (2016) who found that managers’

propensity to trust is positively related to trust in their partners. Therefore, this study will add to previous literature by specifically investigating the effect of the propensity to trust on the

relationship between control mechanisms and information sharing. However, before elaborating on the potential effect of the propensity to trust on this relationship, the following section will first discuss MCT and how different control mechanisms possibly influence information sharing. 2.2 Motivation Crowding Theory

MCT argues that external interventions can weaken (crowding-out), or strengthen (crowding-in) intrinsic motivation under certain circumstances. The origin of these theorized effects stems from the the distinction between intrinsic and extrinsic motivation. According to Festré & Garrouste (2015) intrinsic motivation refers to the natural, self-initiating process which comes from inside someone. Whereas extrinsic motivation refers to the process where motivation is fostered through incentives and rewards. The Cognitive Evaluation Theory (CET) proposes that the notions of locus of control of reinforcement and locus of causality are at the core of the distinction between intrinsic- and extrinsic motivation (Festré & Garrouste, 2015). CET argues that intrinsic motivation can be affected

negatively by the two notions named above under the following circumstances: (1) Intrinsic motivation can decrease when someone receives extrinsic rewards for intrinsically motivated activities (locus of causality), and (2) when someones feelings of competence and self-determination are diminished (locus of control of reinforcement) (Boal & Cummings, 1981). In social psychology there have been many experiments on these effects. The meta-study by Deci, Koestner and Ryan (1999) of 128 experiments concluded that tangible rewards have a significant negative effect on intrinsic motivation, and that these effects are consistent.

In economic studies the principal-agent theory is most often used as a basis for analyzing the motivation crowding effects. The principal-agent model assumes that incentives, like increasing payments, always increase the efforts of an individual (Festré & Garrouste, 2015). According to Frey & Jegen (2001) there are three situations that illustrate the effect of external interventions on performance in a principal-agent setting: (1) Following the basic models, external intervention raises performance by imposing higher marginal benefits for performing. The outcome is the same when the effect of the external intervention is strengthened by the ‘crowding-in effect’. (2) The opposite occurs when external intervention negatively affects the agent’s marginal benefit from performing. (3) In general, both effects are active, so whether an intervention works from the principal’s point depends on the relative size of the two countervailing effects (Frey & Jegen, 2001). So, in economic terms, an incentive or controlling policy normally stimulates performance, except when the

(7)

7

Frey and Jegen (2001) adopted the notions of CET and identified two processes when the crowding-effects appear: (1) Crowding-out: Intrinsic motivation is lowered by external interventions when the interventions are perceived to be controlling. When this happens, self-determination and self-esteem are hurt, and the individuals react by acting less cooperative and reduce their intrinsic motivation in the activity controlled. (2) Crowding-in: Intrinsic motivation is increased by external interventions when the interventions are perceived as supportive. When this happens, self-esteem grows and individuals feel that they have more freedom to act, which positively influences self-determination. Organizational literature increasingly draws on MCT to study the effects of management controls and incentives on employee motivation (Sliwka, 2007; Mikkelsen, Jacobsen, & Andersen, 2017). However, these studies are on an intra-organizational level. Christoffersen and Robson (2017) were recently the first to extend MCT studies to an inter-organizational level. In line with the

crowding-out effect they found that the amount of (financial) support negatively associates with partner trust. The main reason for this effect was that the support reduced the incentives of

managers to act freely within the alliance development (Christoffersen & Robson, 2017). Concerning the effects of control mechanisms it can then be argued that a positive or negative effect on

information sharing, depends on if the controls are perceived as controlling or as supportive by the partner. The next section will first distinguish different types of control mechanisms before

constructing hypotheses about their relationship with information sharing. 2.3 Control Mechanisms

In general, control is viewed as a process of regulation and monitoring for the achievement of organizational goals. In IORs control has been found to facilitate coordination and learning, and is therefore important for satisfactory IOR performance (Das & Teng, 2001). Emsley & Kidon (2007) define control in IORs as a means through which the partner’s behaviour can be made more predictable. Predictability decreases the uncertainty of the partner’s behaviour, resulting in the increased chance that the goals of the IOR are achieved. However, besides these advantages, control mechanisms also have considerable negative effects. Especially the monitoring aspect of controls is considered as damaging to trust because it assumes opportunistic behaviour of a partner which creates doubt and suspicion between the partners (Vélez et al., 2008). Consequently, there has been many debate in literature on the effects of control mechanisms on IORs. It is still unclear how

controls affect partner trust and eventually cooperative behaviour like information sharing (Nicolaou et al., 2011). This study aims on reducing this unclarity by studying the effects of different types of control mechanisms on information sharing. These different types are commonly classified in literature as formal control mechanisms and social control mechanisms (Dekker, 2004). Both classifications will now be elaborated on further.

2.3.1 Formal Control Mechanisms

(8)

8

Moreover, the study of Christoffersen and Robson (2017) showed that interventions aimed to support can lower trust development due to decreasing the partners’ sense of freedom to act. Since trust is considered crucial for information sharing, it can be argued that formal controls can be perceived as controlling by the partner to a level were self-determination and autonomy suffers. Referring to MCT, this would mean that formal controls crowd out the motivation and willingness to share information. Therefore, in line with MCT and previous literature, the following is hypothesized:

H1A:Formal control mechanisms have a negative relationship with information sharing.

2.3.2 Social Control Mechanisms

Social control mechanisms have their origin in the clan mechanism of control proposed by Ouchi (1979), who argued that creating a group (clan) with the same values can serve as the basis of control in organizations. Clan control is exercised when task-related behaviours and outputs are not specified by the organization. Instead of rules and procedures, social controls focus on developing shared values, beliefs, and goals within the clan to create a sense of unity and strengthen suitable behaviours and reward such behaviour (Das & Teng, 2001). Members of the clan internalize organizational goals, which increases their commitment and motivation to achieve these goals. A large advantage of this type of control is that they consume less resources than controlling through formal rules and procedures, which consumes administrative overhead (Ouchi, 1979).

Social controls have a large conceptual overlap with trust, but they are not the same. Shared values and norms result into trust development, but trust itself is not about influencing behaviour of others, it is the belief that the other will act on the IORs behalf (Velez et al., 2008). Social control mechanisms are therefore defined as: “Structural arrangements that foster socialization and interaction between partners, which result in social control based on mutual understanding and the development of shared values and norms” (Vélez et al., 2008, p.973). Emsley & Kidon (2007) further argue that through social controls “influence is extended by developing shared values and beliefs about the future through a sense of commitment to joint action” (p. 832). The difference with formal control mechanisms is then that they do not actively try to regulate behaviour through strict rules and procedures. Therefore, social controls are more likely to be perceived as supportive, instead of controlling. Where formal controls reduce the sense of freedom of partners, social controls provide more freedom to act, which positively influences self-determination. Consequently, it is expected that a crowding-in effect would appear, where social controls lead to an increase in intrinsic motivation to share information (Frey & Jegen, 2001). Dekker (2004) acknowledges this by arguing that social control mechanisms can result in informal coordination and monitoring, and high trust between partners. Since trust is an important basis for the exchange of information between

partners it can be expected that social control mechanisms result in a greater amount of information sharing. Therefore, the following hypothesis is formed:

H1B: Social control mechanisms have a positive relationship with information sharing. 2.4 Propensity to Trust

Before considering the propensity to trust, first the concept of trust itself will be defined. There are many definitions of trust in organizational literature. This study adopts the more social definition of trust, which is quite commonly accepted in organizational literature, and defines trust as “the willingness to accept vulnerability based on positive expectations about the trustee’s intentions or behaviour” (Bidault, de La Torre, de Rham, & Sisto, 2007, p.318; Lu & Yan, 2016, p.461). Thus, trust is about accepting vulnerability, because it is expected that the advantages of having trust are greater than putting control mechanisms in place. However, decisions about trusting a partner must often be made before enough data has been gathered on the trustworthiness of the partner (Colquitt, Scott, & LePine, 2007). In such situations, the propensity to trust becomes of high importance. The

(9)

9

So, without having experience with the trustee, some level of vulnerability is already accepted. In research, there are in general two views regarding the propensity to trust; ‘embeddedness’ and ‘calculativeness’ (Bidault et al., 2007). Embeddedness has its origins in sociology, and emphasizes the social context of developing trust, which is called the embeddedness of economic action. According to sociologists a propensity to trust is determined by factors such as ethical values and assumptions about human nature (Bidault et al., 2007). However, for economics the propensity to trust is

developed from an assessment of the economic conditions of the business transaction (Bidault et al., 2007). The view of economics on trust development can therefore be seen as more objective. The propensity to trust is a calculation of the risks for both parties. So, if both parties are rational they should come up with the same evaluation of the risks and claim identical contractual safeguards or controls. However, this view is rather simplistic. The empirical study of Bidault et al. (2007) showed this by concluding that the propensity to trust was not determined by the objective (economic) conditions but by personal characteristics, which is consistent with the sociological approach to trust adopted in this study. Furthermore, the “calculative perspective may also seem to ignore differing personal propensities to trust which obviously exist” (Tomkins, 2001, p. 167). This study therefore adopts the social perspective on the propensity to trust.

McKnight, Cummings & Chervany (1998) developed a model which predicts the level of the propensity to trust a party under certain conditions. According to the model a high propensity to trust can be explained by three factors; (1) personality, (2) institutional trust and (3) cognitive processes (McKnight et al., 1998). The first two relate to the faith in humanity and the believe that the situation is perceived as normal and that contextual conditions are in place such as regulations. The cognitive processes relate to categorization processes and illusions of control. The categorization processes develop a propensity to trust from putting oneself in the same group as the other in terms of common goals/values (unit grouping), from the reputation the other has and from stereotyping. Illusions of control generate initial trust through token control efforts. Token control efforts are tests to see to see if one can deal with the other successfully. According to McKnight et al. (1998) these efforts give a person the illusion that his/her positive faith in humanity can apply to the other party. If all these named factors are at high levels, one’s propensity to trust is high. The model of McKnight et al. (1998) concerns trust developed in people. However, Tomkins (2001) argued that organisations are a group of people that act as one to the outside world. A group then can decide to place trust in another group, and therefore organizations can place trust in each other (Tomkins, 2001). Moreover, Christoffersen and Robson (2017) assumed in their study that IOR managers deal with organizational issues as if they were personal. So, taking the model McKnight et al. (1998) into account, the level of propensity to trust on an inter-organizational level depends on the same three factors as on an individual level.

2.5 Interaction Effect of the Propensity to Trust

Trust is important as a basis for effective information exchange between partners (Tomkins, 2001; Yang & Maxwell, 2011). However, as noted before, the propensity to trust is not specifically mentioned in previous literature about this relationship (Rousseau et al., 1998; Yang & Maxwell, 2011). The study of Colquitt et al. (2007) shows that the propensity to trust is a significant predictor of trust, which could mean that the propensity to trust also could be an important basis for

(10)

10

The same relationship could be argued for the propensity to trust. When a partner has a propensity to trust they are already willing to take more risk and accept vulnerability, before having credible information of the partner (Colquitt et al., 2007). Therefore, a strong propensity to trust reduces the need for formal coordination and control mechanisms (Tomkins, 2001; Dekker, 2004). Furthermore, based on the model of McKnight et al (1998), when organizations have a strong propensity to trust they see their partners as competent based on their reputation, and deem them as similar in terms of common goals and values. The organization then will accept vulnerability instead of implementing strict regulations and procedures deemed necessary for coordination and cooperation. This is confirmed by Bernerth and Walker (2009) who argue that the extent to which managers in general release control is largely determined by their willingness to trust others to do a good job. So, when one of the partners has a high propensity to trust, while the other partner implements formal control mechanisms to regulate behaviour and outcomes, the former will deem it as unnecessary and overly controlling. Following the crowding-out effect the partner will then start to become less intrinsically motivated to share information. So, higher levels of the propensity to trust strengthen the negative effect of formal control mechanisms on the willingness to share information. This can be termed as a negative moderator effect, which results in the following hypothesis:

H2A: The propensity to trust has a negative moderating effect on the relationship between formal control mechanisms and information sharing.

Recently, Jiang and Lu (2017) found that trust based on long-term interactions between the partners has a strong positive influence on social control. Furthermore, Dekker (2004) argued that trust is an important component of social control. Thus, in general trust seems to be important for the

effectiveness of social controls. However, the propensity to trust was not examined in these studies. The model of McKnight et al. (1998) shows that the propensity to trust is partly created through cognitive processes where the partner is deemed similar in terms of common goals and values. Social control mechanisms are developed to implement shared values and norms to create mutual

understanding, and to further develop such similar goals (Vélez et al., 2008). So, when individuals managing an IOR already have a propensity to trust the partner, the main foundation for strong social control is already in place. Subsequently, social control mechanisms will only be seen as more supportive as they already where. Therefore, a strong propensity to trust will strengthen the positive effect of social control mechanisms on information sharing between partners. Thus, the following can be hypothesized:

H2B: The propensity to trust has a positive moderating effect on the relationship between social control mechanisms and information sharing.

Collectively, the four hypotheses result in the following conceptual model in Figure 1.

(11)

11

3. Methodology

3.1 Data Collection

To test the generated hypotheses, data were collected through a survey instrument. Previous research on control and trust within IORs often focused on one specific type, industry, or performed a case-study within one organization (Mahama, 2006; Vélez et al., 2008). This study did not focus on certain industries or IOR-types. Requirements for selecting respondents were that a contract was signed with the partner firm. Furthermore, the partner firm had to be a business organization. To ensure that respondents sufficiently could answer questions, regarding the role of control and trust between the partners, selected respondents had to be knowledgeable about the arrangements with the partner firm and how the alliance was operated day to day. Most of the time these were

employees who had a management position within the firm. Data were collected together with six other researchers, and the survey was conducted in a face-to-face setting. This has several

advantages, of which the main advantage is that possible ambiguous questions could be made clear instantly (Brace, 2008). Target respondents were selected from the personal network of the

researchers and were mostly approached through e-mail or telephone. This recruitment approach was chosen because personalising recruitment can aid in increasing the response rate (Moutou & Greaves, 2014). Since the survey contained sensitive questions about the relationship with the partner firm, complete confidentiality was assured to all respondents. Regarding this specific

research 49 possible respondents were approached, out of which 22 respondents filled in the survey completely. Thus, the response rate was 44,9% which is consistent with the overall response rate in business and management journals (Mellahi & Harris, 2016).

3.2 Variables

Variables in this study are measured using multi-item scales ranging from 1 to 7 (1 = completely disagree; 7 = completely agree). The survey was designed and delivered by the supervisor of this thesis. First, a Confirmatory Factor Analysis (CFA) was performed, because this method allows to explicitly test a model were the number of factors are already known (Vyas, Jain, & Roy, 2016). To determine the sampling adequacy of the CFA, Bartlett’s sphericity test and the Kaiser-Meyer-Olkin (KMO) measure were assessed. The Bartlett’s test was significant (p < .001) and the KMO had a level of 0.678 which significantly exceeds the cut-off level of 0.5 (Budaev, 2010). Furthermore, Harman’s single-factor test was used to assess common method bias. The analysis showed that the first factor explained 23% of the total variance, which suggests that common method bias is not an issue (Fuller, Simmering, Atinc, Atinc, & Babin, 2016). Next, construct validity of the variables was assessed through Cronbach’s alpha (Aken, Berends, & Bij, 2012). The Cronbach’s alpha values all ranged between 0.762 and 0.849, which shows satisfactory reliability. More details on the reliability of the variables are showed in Appendix A.

3.3 Control Variables

Three control variables are used in this study. These are variables directly related to information sharing between partners. All the used control variables are adopted from previous research. Heide and Miner (1992) used dependency on the partner as control variable because the ‘weaker’

dependent firm may share information because the partner demands it, but this is compliance rather than cooperation. The other two control variables are adopted from the study of Nicolaou et al. (2011). They used the length of the relationship since “an increase in social knowledge can result from longer-term alliance relationships that in turn can increase information sharing and trust among partners” (Nicolaou et al., 2011, p.1030). Furthermore, this study controls for size because larger IORs can lead to more complexity and a need for more information use (Nicolaou et al., 2011). Thus, the control variables included in the model are: (1) dependency on the partner; (2) length of the relationship, and (3) total size of the alliance. The length of the relationship is measured in years and the total size of the alliance through the number of partners. Dependency on the partner is

(12)

12 3.4 Analytical Models

To test the main effects of formal- and social control mechanisms on information sharing, a multiple regression analysis will be performed. Subsequently, to test the moderator effect of the propensity to trust the moderated multiple regression (MMR) model will be used, which is the most used statistical model for conducting moderation analysis (Zhang & Wang, 2015). A common problem within MMR is multicollinearity (Jaccard & Turrisi, 2003). This problem arises when there is a high level of interdependence between the predictors in a regression model (Thompson, Kim, Aloe, & Becker, 2017). To address this problem the variables will be standardized before conducting MMR, and a multicollinearity diagnostic test will be performed to check for contamination of results.

4. Results

4.1 Sample Characteristics

There are several assumptions about the data that need to hold prior to conducting regression analysis. First, normal distribution of the residuals was tested through an examination of the P-P plot (Osborne & Waters, 2002). The P-P plot showed that the residuals generally followed the normality line, so it can be assumed that the residuals are normally distributed. Osborne and Waters (2002) further argue that multiple regression can only estimate the relationship between the dependent and independent variables when the relationship between them is linear in nature and when there is homoscedasticity. Both assumptions were checked through an examination of the scatter plot of the residuals. Examination of the scatter plot of the data in this research showed that residuals were randomly scattered across the horizontal line, which indicates both a linear relationship and homoscedasticity. To check for multicollinearity, the variance inflation factor (VIF) was evaluated. Literature suggests that a VIF value between 5 and 10 indicates moderate to strong multicollinearity between variables (Midi & Bagheri, 2013). The VIFs all have a value below 2, which shows that multicollinearity is not an issue in the data. Another assumption of regression is that the observations are independent. The Durbin-Watson statistic is commonly used to assess this assumption, and should be between 1.5 and 2.5 (Karadimitriou & Marshall, 2017). The

Durbin-Watson for the regression model is 2.022 which shows independence of the observations. Finally, the presence of outliers can lead to a biased estimation of the parameters and misinterpretation of the model (Wang & Li, 2017). To make sure that outliers were detected properly, multiple forms of analysis were used. First, possible outliers were detected by examining the residual plot and looking for values that were more than three standard deviations from the mean. Subsequently, the

observations were evaluated on their leverage and influence through centered leverage, Cook’s distance and Mahalanobis distance (Cook & Hawkins, 1990; Nurunnabi, Hadi, & Imon, 2014). Four observations had a value of more than three standard deviations from the mean, and showed values above the cut-off value of one or more of the distance indicators.

4.2 Descriptive Statistics

(13)

13

This could be more important when the dependency on the partner is higher, hence the positive correlation. Furthermore, social control mechanisms negatively correlate with the length of the IOR (-0.255, p < 0.05). Which could mean that as the length of the IOR increases, social control

mechanisms are less necessary because the organizations already developed shared values over time. Furthermore, when there are more partners involved, IORs seem to be significantly longer. (0.340, p < 0.01).

Table 1

Descriptive statistics and correlation matrix

Correlation matrix (Pearson) Mean Std. Deviation 1. 2. 3. 4. 5. 6. 7.

1. Information sharing 5.33 1.057 1,00

2. Formal control mechanisms 5.19 1.497 0,069 1,00

3. Social control mechanisms 5.29 1.203 0,121 -0,005 1,00

4. Propensity to Trust 4.70 0.870 0,112 0,025 0,229* 1,00

5. Dependency on Partner 3.70 1.449 0,067 -0,102 0,315** 0,071 1,00

6. Length of IOR (in years) 8.98 8.837 -0,012 -0,148 -0,225* -0,150 0,009 1,00

7. Number of partners in IOR 9.41 52.093 -0,037 -0,071 0,107 0,007 0,119 0,340** 1,00

*Correlation is significant at the 0,05 level (2-tailed) ** Correlation is significant at the 0,01 level (2-tailed)

Table 2

Regression results Dependent variable: Information Sharing

Model 1 Model 2 Model 3 β (SE) β (SE) β (SE) Intercept 5.10 (0.36) 3.11 (0.94) 2.02 (1.01) Control variables IOR length -.002 (0.016) .008 (0.016) .012 (0.016) Size of IOR .001 (0.012) .003 (0.012) -.001 (0.012) Dependency of Partner .089 (0.084) .066 (0.087) .087 (0.086) Main effects

Formal control mechanisms .109 (0.078) .094 (0.076) Social control mechanisms .139 (0.109) .224** (0.111) Propensity to trust .148 (0.141) .268* (0.145)

Interaction effect

Formal control mechanisms x Propensity to trust

.060 (0.111) Social control mechanisms x

(14)

14 4.3 Regression Analysis

The results of the regression analysis are shown in Table 2. Listwise deletion is used to deal with missing values in the database because it is considered as the most honest method for handling missing data (Allison, 2001). Due to the use of this method 23 observations are lost. Furthermore, the four identified outliers are taken out of the sample, which means that the regression analysis is based on 87 responses. The model is build-up in the following way: Model 1 shows the results for the three control variables, and model 2 presents the effects of the social- and formal controls on

information sharing as predicted by H1a and H1b. Finally, model 3 shows the interaction effect of the propensity to trust on the relationship between the control mechanisms and information sharing. Model 1 explains 1,4% of the variance in information sharing, and demonstrates that none of the three control variables have a significant effect on information sharing. Model 2 improves the explanation of the variance in information sharing to 8%, and model 3 to 15,1%. Furthermore, the F-test shows no significance for model 1 and 2, but the F-change in model 3 is significant which

indicates that model 3 significantly adds to a better explanation of the change in information sharing. Hypothesis 1b proposed a positive effect of social control mechanisms on information

sharing. Model 2 already shows this positive effect (β = .14), however the effect is only significant in model 3 (p = .047). This is due to the significant positive moderating effect of the propensity to trust on this relationship, as proposed in hypothesis 2b (p = .013). So, social controls tend to increase information sharing only when the relationship is moderated by the propensity to trust. Opposed to hypothesis 1a, model 2 shows a positive effect of formal control mechanisms on information sharing, however this effect is not significant. The results further indicate no negative moderating effect of the propensity to trust on the relationship between formal control mechanisms and information sharing. Which means that hypothesis 2a is not supported. Finally, the results in model 3 indicate a direct positive effect of the propensity to trust on information sharing (p = .069). Figure 2 provides a visualization of the significant results found in the analysis.

(15)

15

5. Discussion and Conclusion

Individuals within inter-organizational relationships all have their own subjective understanding of the relationship with a partner. These cognitive structures are essential in the decision-making process and are a key driver of successful collaboration (Ryynänen, Henttonen, & Salminen, 2015). Yet, organizational studies seldom examine the effects of individual characteristics on IORs (Lu & Yan, 2016). This is an important research gap because it could lead to an improved understanding of how relationships are built between organizations. Consequently, this knowledge can aid in the ongoing debate on the relationship between control and trust within IORs (Dekker, 2004; Vélez et al., 2008). Therefore, this study sought to answer the question how the propensity to trust influenced the relationship between control mechanisms and information sharing in IORs.

5.1 Research Implications

Throughout the years researchers acknowledged the importance of different types of trust for the effectiveness of controls and information sharing between partners, but the propensity to trust was not included in such studies (Das & Teng, 1998; Dekker, 2004; Yang & Maxwell, 2011; Jiang & Lu, 2017). The results of this empirical study demonstrate that the propensity to trust positively contributes to the effectiveness of social controls in relation to information sharing within IORs (β = .32). Furthermore, the propensity to trust itself has a significant positive effect on information sharing (β = .27). This study adds to the recent research of Lu and Yan (2016), who found that the propensity to trust positively contributes to the trust formation between organizations. The findings in this study demonstrate that the propensity to trust is important for the effectiveness of social controls and information sharing between partners. Since the exchange of information between partners is crucial for cooperative behaviour and successfully managing IORs, a strong propensity to trust is of indirect importance for effectively managing IORs (Das & Teng, 1998; Mahama, 2006). Hence, the findings empirically support and confirm the call for a deeper focus on the influence that the propensity to trust can have on the development of relationships between organizations (Schoorman, Mayer, & Davis, 2007; Lu & Yan, 2016).

Next, due to the positive moderator effect of the propensity to trust, a positive effect (β = .22) of social controls on information sharing was found. Social controls therefore seem to be perceived as supportive to a scale where they increase intrinsic motivation of the partner to share information. Consequently, based on MCT the argument can be made that social controls give the individuals managing the IOR a feeling that they have more freedom to act and positively contribute to self-determination (Frey & Jegen, 2001). Previous research showed the importance of this

increased feeling of autonomy for instance to transfer knowledge in IORs (Paulsen & Hjerto, 2014). For most organizational researchers, the positive relationship between social controls and

information sharing is not surprising. Studies already showed the positive effect of social controls on risk reduction, trust building and cooperative behaviour (Das & Teng, 2001; Vélez et al., 2008). Since information sharing is considered a dimension of cooperation (Mahama, 2006), this result is in line with previous findings.

(16)

16

Formal controls therefore do not seem to hurt the self-determination and sense of autonomy of a partner. These results possibly indicate that crowding out effects are only strong in an individual principal-agent setting, and not in an inter-organizational setting. Since the crowding out effect was mostly experimented with on an individual level, an explanation might then be that MCT is not applicable on an inter-organizational level (Festré & Garrouste, 2015). This would contradict the assumption of Christoffersen and Robson (2017) that IOR managers deal with organizational issues as if they were personal. Although this discussion is beyond the scope of this research, the findings do confirm that formal controls do not negatively influence information sharing, which seemingly contradicts previous arguments that such controls could hurt mutual trust between partners (Das & Teng, 1998; Das & Teng, 2001; Tomkins, 2001). If this where the case, information sharing would also be negatively influenced, since trust is considered as a key basis for information sharing. However, the results also show no significant positive effect, indicating that the role of formal controls in creating a successful partnership is still open for debate.

5.2 Managerial Implications

In practice, the findings of this study indicate the importance of having people managing the IOR with a strong propensity to trust. Besides making social controls more effective, a strong propensity to trust also contributes directly to the information exchange between partners. However, it should be mentioned that a propensity to trust does not make formal controls more effective in terms of information sharing. This is in line with the recent findings of Lu & Yan (2016) that managers with a high level of propensity to trust are less effective for generating trust in IORs with high contractual control. Thus, while managers with a strong propensity to trust seem to benefit cooperative behaviour, it is not equally effective for implementing rules and procedures aimed at controlling behaviour and outcomes within the IOR. Furthermore, the findings show that social controls positively contribute to information sharing between partners, when there is a propensity to trust. So, creating alignment with a partner in terms of values, beliefs and goals increases the exchange of information between partners. Subsequently, increasing information sharing contributes to joint problem solving, a restraint from use of power and to relationship performance, indicating the importance of social controls (Mahama, 2006). Formal control mechanisms seem to have no

negative, nor a positive effect on information sharing between partners. However, it should be noted that the IORs in the sample of this study have an average length of nine years, and are mostly well-established relationships. Inkpen and Currall (2004) argued that formal controls are preferred at the creation of the IOR when decisions about control must be made. Once an IOR is formed, learning processes and trust are more central to decisions about control. Social controls create this

opportunity for firms to learn about their partners. Although more empirical research is needed to explore the effects of controls in certain IOR stages, the results suggest that social controls are more effective in terms of increasing information exchange in later stages of IORs than formal controls. 5.3 Limitations and Further Research

While this study adds to organizational literature on the influence of the propensity to trust on IORs, future research is still needed to fully understand the influences of different types of controls, and to address the limitations of this study. First, the sample size is rather small, which reduces the

(17)

17

It is therefore possible that the propensity to trust also has different effects on certain types of formal control, which did not surface in the outcomes of this study. Future research should therefore strive to a more thoroughly understanding of the effects of the propensity to trust on specific types of control mechanisms. A third limitation is that the effects of control mechanisms were investigated within IORs on a specific moment of time. Various researchers have argued that the effects of formal- and social controls significantly vary in the different stages of an IOR (Das & Teng, 1998; Inkpen & Currall, 2004; Vélez et al.,2008). Yet, there are little to none empirical studies which test these effects over larger sample sizes, which could be a valuable avenue for further research to focus on. Finally, this study is one of the first to draw on MCT to explain the effects of controls on an inter-organizational level. While the results support the crowding-in effect by showing that social controls are perceived as supportive and increase motivation to share information, no prove was found for a crowding-out effect. It is therefore arguable if the arguments of MCT hold on an inter-organizational level, and not only on an individual principal-agent level. To add to this debate, future

inter-organizational studies could specifically investigate the influence of control mechanisms on managerial behaviour within IORs. After all, finding clarity on effective application of control

mechanisms remains an important challenge for researchers as well as for practitioners within IORs.

(18)

18

References

Aken, V., Berends, & Bij, V. d. (2012). Problem solving in Organizations - A methodological handbook for

business and management studies.

Allison, P. (2001). Missing Data. Thousand Oaks: SAGE Publications, Inc.

Bernerth, J., & Walker, H. (2009). Propensity to Trust and the Impact on Social Exchange. Journal of Leadership

& Organizational Studies, 217-226.

Bidault, F., de La Torre, J., de Rham, C., & Sisto, M. (2007). What Makes Executives Trust Each Other? The Determinants of the Willingness to Rely on Trust in Business Partnerships. Creativity & Innovation

Management, 317-329.

Boal, K., & Cummings, L. (1981). Cognitive Evaluation Theory: An Experimental Test of Processes and Outcomes . Organizational Behavior and Human Performance, 289-310.

Brace, I. (2008). Questionnaire Design: How to Plan, Structure and Write Survey Material for Effective Market

Research. London: Kogan Page Limited.

Budaev, S. (2010). Using Principal Components and Factor Analysis in Animal Behaviour Research: Caveats and Guidelines. International journal of behavioural biology, 472-480.

Chen, C., Chen, X., & Meindl, J. (1998). How can cooperation be fostered?: the cultural effects if individualism– collectivism. Academic Management Review, 285-304.

Chiu, W., & Ng, F. (2015). Enhancement of organizational commitment through propensity to trust.

Engineering, Construction and Architectural Management, 272-294.

Christoffersen, J., & Robson, M. (2017). Money Can’t Buy Me Trust: Evidence of Exogenous Influences Crowding out Process-based Trust in Alliances. British Journal of Management, 135-153.

Colquitt, J., Scott, B., & LePine, J. (2007). Trust, Trustworthiness, and Trust Propensity: A Meta-Analytic Test of Their Unique Relationships With Risk Taking and Job Performance. Journal of Applied Psychology, 909-927. Cook, R., & Hawkins, D. (1990). Unmasking Multivariate Outliers and Leverage Points: Comment. Journal of the

American Statistical Association, 640-644.

Das, T., & Teng, B. (1998). Between trust and control: Developing confidence in partner cooperation alliances.

Academy of Management Review, 491-512.

Das, T., & Teng, B. (2001). Trust, Control, and Risk in Strategic Alliances: An Integrated Framework.

Organization Studies, 251-283.

Deci, E., Koestner, R., & Ryan, R. (1999). A meta-analytic review of experiments examining the effects of extrinsic rewards on intrinsic motivation. Psychological Bulletin, 627-668.

Dekker, H. (2004). Control of inter-organizational relationships: evidence on appropriation concerns and coordination requirements. Accounting, Organizations and Society, 27-49.

Dekker, H., Ding, R., & Groot, T. (2016). Collaborative Performance Management in Interfirm Relationships.

Journal of Management Accounting Research, 25-48.

Dekker, H., Sakaguchi, J., & Kawai, T. (2013). Beyond the contract: Managing risk in supply chain relations.

Management Accounting Research, 122-139.

Duarte, M., & Davies, G. (2004). Trust as a Mediator of Channel Power. Journal of Marketing Channels, 77-102. Emsley, D., & Kidon, F. (2007). The Relationship between Trust and Control in International Joint

(19)

19

Faerman, S., McCaffrey, D., & Van Slyke, D. (2001). Understanding interorganizational cooperation: public– private collaboration. Organization Science, 372-388.

Festré, A., & Garrouste, P. (2015). Theory and Evidence in Psychology and Economics about Motivation Crowding Out: A Possible Convergence. Journal of Economic Surveys, 339-356.

Frey, B., & Jegen, R. (2001). Motivation Crowding Theory. Journal of Economic Surveys, 589-611. Fuller, C., Simmering, M., Atinc, G., Atinc, Y., & Babin, B. (2016). Common methods variance detection in business research. Journal of Business Research, 3192-3198.

Heide, J., & Miner, A. (1992). The shadow of the future: Effects of anticipated interaction and frequency of contact on buyer-seller cooperation. Academy of Management Journal, 265-291.

Inkpen, A., & Curall, S. (2004). The Coevolution of Trust, Control, and Learning in Joint Ventures. Organization

Science, 586-599.

Jaccard, J., & Turrisi, R. (2003). Interaction effects in Multiple Regression. Thousand Oaks: Sage.

Jiang, W., & Lu, Y. (2017). Influence of initial trust on control from client perspective: construction industry in China. Construction and Architectural Management, 326-345.

Karadimitriou, S., & Marshall, E. (2017, May 9). Outliers, Durbin-Watson and interactions for regression in SPSS. Opgehaald van Statstutor: www.sheffield.ac.uk/polopoly_fs/1.531431!/file/MASHRegression_Further_SPSS.pdf Li, J., Sikora, R., Shaw, M., & Tan, G. (2006). A strategic analysis of inter organizational information sharing .

Decision Support Systems, 251-266.

Lu, S., & Yan, H. (2016). Contractual control, the propensity to trust, active trust development: construction industry. Journal of Business & Industrial Marketing, 459-471.

Mahama, H. (2006). Management control systems, cooperation and performance in strategic supply relationships: A survey in the mines. School of Accounting, 315-339.

McKnight, D., Cummings, L., & Chervany, N. (1998). Initial trust formation in new organizational relationships.

The Academy of Management Review, 473-490.

Mellahi, K., & Harris, L. (2016). Response Rates in Business and Management Research: An Overview of Current Practice and Suggestions for Future Direction. British Journal of Management, 426-437.

Mellewigt, T., Madhok, A., & Weibel, A. (2007). Trust and Formal Contracts in Interorganizational Relationships - Substitutes and Complements. Managerial and Decision Economics, 833-847.

Midi, H., & Bagheri, A. (2013). Robust multicollinearity diagnostic measure in collinear data set. Laboratory of

Computational Statistics and Operations Research, Institute for Mathematical Research University Putra.

Mikkelsen, M., Jacobsen, C., & Andersen, L. (2017). Managing employee motivation: Exploring the connections between managers' enforcement actions, employee perceptions, and employee intrinsic motivation.

International Public Management Journal, 183-205.

Moutou, C., & Greaves, S. (2014). Determining strategies to increase response rates in transport surveys of small shop-based businesses. Transportation Planning and Technology, 598-622.

Nambudiri, R. (2012). Propensity to trust and organizational commitment: a study in the Indian pharmaceutical sector. The International Journal of Human Resource Management, 977-986.

Nicolaou, A., & Christ, M. (2016). Integrated Information Systems, Alliance Formation, and the Risk of Information Exchange between Partners. Journal of Management Accounting Research, 1-18.

Nicolaou, A., Sedatole, K., & Lankton, N. (2011). Integrated Information Systems and Alliance Partner Trust.

(20)

20

Nurunnabi, A., Hadi, A., & Imon, A. (2014). Procedures for the identification of multiple influential observations in linear regression. Journal of Applied Statistics, 1315-1331.

Osborne, J., & Waters, E. (2002). Four Assumptions Of Multiple Regression That Researchers Should Always Test. Practical Assessment, Research & Evaluation.

Ouchi, W. (1979). A Conceptual Framework for the Design of Organizational Control Mechanisms. Management

Science, 833-848.

Paulsen, J., & Hjerto, K. (2014). Exploring individual-level and group-level levers for inter-organizational knowledge. The Learning Organization, 274-287.

Payan, J., & Svensson, G. (2007). Co-operation, coordination, and specific assets in inter-organisational relationships. Journal of Marketing Management, 797-814.

Rousseau, D., Sitkin, S., Burt, R., & Camerer, C. (1998). Not so different after all: A cross-discipline view of trust.

Academy of Management Review, 393-404.

Ryynänen, H., Henttonen, K., & Salminen, R. (2015). Inter-organizational cognitive structures: network conception in MobileTV case. Journal of Business & Industrial Marketing, 662-676.

Sánchez, J., Vélez, M., & Rámon-Jerónimo, M. (2012). Do suppliers' formal controls damage distributors' trust?

Journal of Business Research, 896-906.

Schoorman, F., Mayer, R., & Davis, J. (2007). Organizational trust: philosophical perspectives and conceptual definitions. Academy of Management Review, 337-340.

Sliwka, D. (2007). Trust as a Signal of a Social Norm and the Hidden Costs of Incentive Schemes. American

Economic Review, 999-1012.

Thompson, C., Kim, R., Aloe, A., & Becker, B. (2017). Extracting the Variance Inflation Factor and Other Multicollinearity Diagnostics from Typical Regression Results. Basic and Applied Social Psychology, 81-90. Tomkins, C. (2001). Interdependencies, trust and information in relationships, alliances and networks. School of

Management, 161-191.

Vélez, M., Sànchez, J., & Álvarez-Dardet, C. (2008). Management control systems as inter-organizational trust builders in envolving relationships: Evidence from a longitudinal case study. Accounting, Organizations and

Society, 968-994.

Vyas, V., Jain, P., & Roy, A. (2016). Construct Validity and Preferential Values: Mutual Fund as Investment Avenue. SCMS Journal of Indian Management, 103-115.

Wang, T., & Li, Z. (2017). Outlier detection in high-dimensional regression model. Communications in Statistics -

Theory and Methods, 6947-6958.

Yang, T., & Maxwell, T. (2011). Information-sharing in public organizations: A literature review of interpersonal, intra-organizational and inter-organizational success factors. Government Information Quarterly, 164-175. Zhang, Q., & Wang, L. (2015). Evaluating Methods for Moderation Analysis with Missing Data in the Predictors.

(21)

21

Appendix: Variable Reliability

Variable Items Component Loading

COOPERATION (Information sharing) Cronbach’s α = 0.762

COOPERATION_1 COOPERATION_2

0,882 0,856 FORMALCONTSYS (Formal control mechanisms)

Cronbach’s α = 0,762

FORMALCONTSYS_1 FORMALCONTSYS_2

0,894 0,882 SOCIALCONTSYS (Social control mechanisms)

Cronbach’s α = 0,849 SOCIALCONTSYS_1 SOCIALCONTSYS_2 SOCIALCONTSYS_3 SOCIALCONTSYS_4 0,827 0,898 0,846 0,743 PROPTRUST (Propensity to trust)

Cronbach’s α = 0,765 PROPTRUST_2 PROPTRUST_3 PROPTRUST_4 PROPTRUST_5 0,670 0,765 0,771 0,853 DEPONPTNR (Dependency on the partner)

Cronbach’s α = 0,787 DEPONPTNR_2 DEPONPTNR_3 DEPONPTNR_4 0,847 0,857 0,785

Exploring the Influence of the Propensity to Trust on Control and Information Sharing Within Inter-Organizational Relationships