- 1 -
R&D Investment and Interfirm Absorptive Capacity
Master Thesis, Marketing Management
University of Groningen, Faculty of Economics and Business
January 16, 2016
Contact details:
Neptunusstraat 15, 9742JK, Groningen, The Netherlands.
Tel: + 31 684029718,
e-mail: gabimarulea@me.com
student number: s2962160.
First Supervisor: H. J. Berger, e-mail: j.berger@rug.nl.
- 2 -
MANAGEMENT SUMMARY
In the present research paper we have studied 166 buyer-supplier dyadic relationships. The independent firms operate in three different industries.
Firms aim to improve their innovation performance, thus they invest in learning activities in order to sustain their competitive advantage. Therefore the scope of the research is focused on the R&D investment of the firm. Given the field of the industries, we infer their necessity of research and adoption of new and innovative ideas. Moreover, a firm’s objective is to expand the boundaries of knowledge acquisition and to implement activities that would engender inter-organizational learning. Partners in interfirm relationships rely on their partnership for such exchange benefits as increased efficiency, flexibility and absorptive capacity (Cannon et al.2000).
Thus, in this paper we argue that the capability of a firm to “recognize new external knowledge, assimilate it and apply it to commercial ends” (Cohen and Levinthal,1990) is a function of R&D spending.
Moreover, in an inter-organizational relationship one of the challenges is to set a governance mechanism which will promote the expansion of the benefits and will safeguard the exchanges between the two parties. Therefore we study the moderating effects of contractual versus relational forms of governance.
Lastly, we want to advance our understanding on how absorptive capacity impacts relationship performance. To fully grasp the construct of absorptive capacity, we distinguish between two dimensions, potential absorptive capacity and realized absorptive capacity. Our results are analyzed from two perspectives, from the buyer-side and the supplier-side.
- 3 -
Table of Contents
1. Introduction
... 52. Theoretical Framework
... 82.1
The multidimensional construct of ACAP
... 82.2 Role of R&D in developing ACAP
... 92.3 The role of governance mechanism in appropriation of ACAP
...112.4 Relational governance
...122.5 Contractual governance
...142.6 Relationship Performance
...152.7 Conceptual Model
...173. Method
...183.1 Data Collection
...183.2 Measurements
...193.2.1 Independent variables
...193.2.2 Moderator variables
...203.2.3 Dependent variables
...213.3 Statistical Procedure
...224. Analysis and Results
...234.1 Measurement model
...234.2 Outer model
...244.2.1 Reflective Constructs ...24
- 4 -
4.3 Structural Model
...294.4 Hypothesis Testing
...31 4.4.1 Buyer Database ...31 4.4.2 Supplier Data ...32 4.4.3 Distribution of Benefits ...335. Discussion
... 355.1 Direct effects ...35 5.2 Moderation affects ...36
5.3 Scientific and managerial implications ...37
5.4 Limitations and future research ...38
References
...40- 5 -
1. INTRODUCTION
In the era of continually growing competitiveness and rapid pace of technological change, firms are looking to increase their intellectual assets by exploiting knowledge both from internal and external sources (Huggins and Johnston, 2012). In line with the knowledge-based theory of the firm, knowledge is considered one of the central factors contributing to competitive advantage (Grant, 1996). Thus, firms view relationships with other companies as a way of creating competitive advantage by capitalizing on the resources and valuable information possessed by their alliance partner.
According to Cohen and Levinthal (1990), the organization’s possibility to reach competitive advantage and to innovate is determined by the firm’s ability to recognize the valuable external knowledge, assimilate it and apply it for commercial purposes. This particular capability has been defined by Cohen and Levinthal (1990) as “absorptive capacity”.
In this paper we take a closer look on how firms build their absorptive capacity (ACAP), particularly in the context of buyer-supplier interfirm relationships.
In order to build up absorptive capacity and develop competitive advantage, organizations tend to make substantial investments of resources in research and development (R&D). The primary purpose of firms is to generate innovation and learning, which represents the main outcome of R&D spending (Cohen and Levinthal, 1989). In their publications on absorptive capacity (1989, 1990), Cohen and Levinthal have stressed the importance of R&D, stating that the capacity of utilizing external knowledge is generally a byproduct of R&D investment. Therefore, it is necessary to invest in R&D in order to create absorptive capacity. However, to have the incentive to initiate research and development, the organization must be capable to derive benefits, which are satisfactory enough to make the investment profitable (Levin et al., 1987).
- 6 - 1993; Moenart and Souder, 1990; Nonaka, 1994,). Particularly routines and structures tailored within the relationship play a significant role in facilitating the flow of new external knowledge (ibid.).
Likewise, knowledge flows at different rates and different degrees (Fey and Birkinshaw, 2005; Kogut & Zander, 1992), and is subject to the governance form and the knowledge characteristics. Different mechanisms which govern a relationship have different approaches and implications on achieving external knowledge, therefore the flow of knowledge into the organization and out are distinct from one organization to another (Fey & Birkinshaw, 2005). Therefore since the R&D investment plays a decisive role in absorption of new knowledge (ACAP), it can be inferred that governance can influence this relationship.
Respectively, scholars have emphasized the importance of two types of governance, contractual and relational. Contractual governance refers to the degree to which an interfirm relationship is governed by formal contracts (Li et al., 2010); these contracts highlight the importance of formal agreements to safeguard against opportunism (Cao and Lumineau, 2014). On the other hand, the relational governance regards trust and relational norms as the main characteristics of the relationship, where partners have expectations about each other’s behaviors (ibid.) and the relationship exchange is regulated by social relations.
Expanding upon these ideas, the aim of this paper is to advance the understanding of interfirm absorptive capacity. First, with respect to ACAP the present study focuses on the R&D investment of the firm and its contribution to the creation and exploitation of knowledge. Second, the objective of this paper is, to increase the understanding on how different governance mechanisms in an interfirm relationship contribute to exchanges between partners and lead to the creation of innovative information. The research setting involves the empirical examination of both partners within an interfirm collaboration; therefore this paper provides a separate view on how the buyer and supplier act with respect to their individual R&D investment and how this activity is influenced by the alliance governance. Despite its importance, none have focused on determining whether governance has a moderating effect on the relationship between resource allocation and absorptive capacity. In contrast, the interest in solely R&D spending and external knowledge acquisition has been widely addressed in the literature (Cohen and Levinthal, 1990; Veugelers and Cassiman, 1998). Consequently, this paper invokes the following research question:
- 7 - The contribution of this paper to the existing literature is twofold. First, the study investigates the relationship of R&D and ACAP in the context of interfirm relationships, particularly by analyzing dyadic responses gathered from buyers and suppliers within the relationship. The objective is to understand the extent to which the interfirm relationship benefits from parties’ individual resource allocation. In this way this paper stands out from the existing literature where the relationship between R&D and ACAP has been mainly assessed in the setting of separate buyer and supplier groups (Vuegelers and Cassiman, 1999; Matusik and Heeley, 2005; Deeds 2001).
Second, given that in this study we investigate the nature of ACAP from the perspective of interfirm relationships, we would like to shed light on the fact that the interchange of knowledge and the way it is being deployed by partners in the relationship, is being controlled by the governance mode. Thus, apart from the R&D investment which in the previous studies has been identified as an antecedent of ACAP (Cohen and Levinthal ,1990), distinct configurations of governance lead to different routines of knowledge-exchange (Selnes and Sallis, 2003). Hence, this paper intends to analyze the moderating effect of governance mechanism on the relationship between R&D and ACAP.
- 8 -
2. THEORETICAL FRAMEWORK
A large number of academics and practitioners have shown interest in the topic of absorptive capacity in the context of interfirm alliances (Berger, 2015; Lane and Lubatkin, 1998; Lane, Salk, and Lyles, 2001). Nevertheless, few of them have empirically investigated both sides of the buyer-supplier relationship, and how partners individually allocate resources for R&D activities which implicitly realize communal ACAP. Current literature also emphasizes the effect governance mechanism has on the exchanges between two partners in an alliance (Rowley et al. 2000; Lee and Cavusgil, 2006). Therefore, in this paper we aim to fill in the gap in the existing literature by determining whether governance structure has a moderating effect on the relationship between R&D and absorptive capacity, by empirically examining this relationship in the context of buyer-supplier matched-pair relationships. For this purpose, in the following section the variables used in this study will be depicted and explained.
2.1 The multidimensional construct of ACAP
Cohen and Levinthal (1989; 1990) have stressed the importance of external knowledge in the conceptualization of ACAP. They emphasize three main processes a firm must tackle in building absorptive capacity. First, one should be able to recognize the value of new information, assimilate the information and apply it to commercial ends (ibid.). Furthermore, they accentuate that ACAP is primarily a function of prior individual knowledge accumulated. In order to create firm-level ACAP, the individual possessing the new knowledge must exploit it at the organizational-level (ibid).
In another relevant study, Zahra and George (2002) have elaborated on Cohen and Levinthal's concept of ACAP and redefined it as a dynamic capability. They argue that ACAP is subject to multiple learning levels. According to their reconceptualization of the construct, the systems, processes, routines and structures, within an organization, determine the four stages of learning, namely, the underlying dimensions of ACAP. The four dimensions are structured as follows: 1) acquisition, which implies the capability of a firm to identify and capture the valuable knowledge resided in the external environment, which is critical to its operations; 2) assimilation, refers to the firm's routines and processes that allow it to analyze, process, interpret and understand the information obtained from external sources. The first and second dimensions encompass the firm's potential absorptive capacity (herewith as: PACAP) and denote the explorative learning.
- 9 - assimilated knowledge; and 4) exploitation, which is based on the firm's abilities to deploy and consistently leverage the new acquired knowledge and to create new competencies by using the acquired knowledge and integrating it into its operations. Thus, the third and fourth dimensions constitute the realized absorptive capacity (herewith as: RACAP), which refers to exploitative learning (ibid.).
Balancing explorative and exploitative learning capabilities is essential for fully capitalizing on the benefits of created knowledge (Levinthal and March, 1993; Berger, 2015). For the reason that in the fast-paced environment, firms must be flexible in matching the demands from the market and therefore they need to implement the worthwhile knowledge to increase the organization’s performance (Teece, Pisano, and Shuen, 1997; Berger, 2015). Thus we can assume that firms with underdeveloped dynamic capabilities are likely to lag behind.
2.2 Role of R&D in developing ACAP
- 10 - For that reason, the importance of investing in R&D has been highly regarded by scholars and practitioners. Top management of companies are paying a lot of interest in enhancing organization’s innovation performance. Organizations are exploring the innovation process by merging internal R&D and activities related to external knowledge acquisition (Cassiman& Veugelers, 2006).
The organization engages in R&D activities in order to accumulate knowledge either by appropriating it from the industry spillover, or by developing new knowledge (Cohen and Levinthal, 1990). Many studies regard the concept of ACAP as an antecedent and at the same time as an outcome of learning. In fact, Veugelers (1997) argues that the company's existing absorptive capacity allows it to learn from its external R&D collaboration. As a consequence, a firm deploys this learning by engaging in in-house R&D activities, thus continuously building the firm's ACAP. Finally, this process prompts the company to further invest in external research and development and learning.
Therefore, it might be inferred that R&D intensity results in firm’s ACAP. However, according to a study of Lane and Lubatkin (1998), R&D effort is essential but not sufficient for building absorptive capacity. To generate the desired result in explorative learning performance such as new product development, the flow of R&D spending needs to be consistent (Deeds, 2001).Therefore, in this study we would like to test whether the separate R&D of the buyer and the supplier, correlates with the creation of communal ACAP for the relationship. Hence, we would like to investigate this relationship by empirically testing it on 166 dyadic buyer-supplier relationships.
- 11 -
H1.a: We posit that the suppliers’ R&D expenditures are stronger correlated with PACAP than the buyers’ ones.
Normally suppliers are involved in technology transfer which implies that they share and/or implement technology on behalf of their partner, which assumes knowledge transfer and close collaboration (Hult et al. 2004). One advantage of the knowledge accumulation activities is that these activities improve the supply chain’s knowledge flows. According to the knowledge-based view, the flow of information will enrich each supply chain connection and its ability to efficiently accomplish its task (Hult et al., 2004; Grant, 1996). Namely, the outcome derived from knowledge acquisition activities is improved. Furthermore, suppliers also engage in more refinements for meeting the buyers’ requirements (Oosterhuis et al. 2013). Thus, since the supplier is more pressured than the buyer for the refinement of his offer, the suppliers tend to perceive their performance better than it is (Chen and Paulraj, 2004b). However, R&D output measured by the promptness of developing knowledge is expected to be higher from the buyer’s side, since the domain of the buyer’s activity in the area of manufacturing is directed particularly to the employment of newly stored knowledge to the process of development and creation of new products and expertise. Given that the associates have diverse interests and different needs in the relationship, it is equitable to question whether the buyer allocates more resources to the development of new knowledge. On the basis of this reasoning, the next hypothesis is conveyed as follows:
H1.b: We posit that the buyers’ R&D expenditures are stronger correlated with RACAP than the suppliers’ ones.
2.3 The role of governance mechanism in appropriation of ACAP
The firm’s incentive to innovate is contingent on the level of appropriation of the outcomes from innovation activities, and whether these can be easily appropriated and distributed within the firm (Veugelers and Cassiman, 1999). There has been found an adverse consequence of a low level of appropriation. A low level of appropriation might result in a disincentive to invest, thus firms diminish their internal R&D investment, since they are unable to capture the benefit of their entire investment (Arrow, 1962; Spence, 1984; Veugelers and Cassiman, 1998).
- 12 - External know-how is expected to encourage the incentive of individual R&D activities, so far as internal R&D management groups are ready to effectively absorb it (Veugelers, 1997). Moreover, absorptive capacity acts as an instrument for capitalizing on the complementarity of internal and external know-how. However, if a company’s strategy doesn’t accentuate the importance of the link between internal R&D and external knowledge-acquisition activities, the firm’s R&D groups might hinder the potential external linkages (ibid.). Therefore, the strategy adopted by a firm influences the appropriation of knowledge.
Likewise, as stated by Fey & Birkinshaw (2005), the diverse types of governance mechanisms have distinct implications and affect the way firms’ access knowledge and its flow into the organization. Therefore, it might be expected, that the firm’s strategy to appropriate new external information might be impacted by the governance mode of the interfirm relationship.
In his study, Oxley (1999) adopts the transaction cost economics (TCE) view in examining the governance connotation in “appropriatability hazards”. Oxley (1999) suggests that in an interfirm alliance in which technology transfers and intellectual property are subject to the relationship, governance mechanism plays a role in appropriability hazards.
Therefore, it is expected that the mechanism applied to govern the buyer-supplier relationship can impact the appropriatability of R&D spending on accessing absorptive capacity. In the following chapters the types of governance mechanisms will be reviewed for getting a better understanding whether these might affect the relationship between allocation of resources of an individual firm and the joint ACAP.
2.4 Relational governance
- 13 - Relational norms denote that partners shape expectations about the behavior of each other (Cao and Lumineau, 2015). Flexibility, solidarity and information exchange are the norms which increase the effective interfirm coordination, facilitate the access to resources and enhance the company’s opportunities. This on the other hand would be more challenging in an arm’s length tie (Uzzi, 1997; Poppe and Zegner, 2002). The enumerated above norms, drive firms to perform in expected ways. Consequently, trust and relational norms can decrease opportunism (ibid.). So, if partners involved in an exchange rely that they will not be impaired, exploited or risk opportunistic behavior from their partner, they are more willing to share information (Jap 1999, 2001; Selnes and Sallis, 2003; Morgan and Hunt, 1994). Accordingly, we can expect that relational governance will positively affect the appropriation of external knowledge from R&D investment
H2.a: Relational governance positively moderates the relationship between buyer’s R&D spending and PACAP.
H2.b: Relational governance positively moderates the relationship between supplier’s R&D spending and PACAP.
According to Uzzi (1997), when the two partners in an interfirm alliance are governed by relational norms and trust, the exchange is characterized by embedded ties. Embeddedness promotes economies of time and allocative efficiency, due to the reason that in relationally-governed relationships the cost of contracting is avoided, since the partners trust that the relational rents will be distributed impartially. It implies that information is transferred at a faster pace (economies of time) and firms assist each other in understanding production methods for accelerating the decision-making process (allocative efficiency). Moreover, allocative efficiency facilitates the firms in matching the production and demand, particularly under circumstances or rapid product innovation.
As a result, it can be expected that relational governance enhances the transformation and exploitation capabilities. Thus, the partners will expect greater appropriation as an outcome of these capabilities being successfully deployed. Accordingly, the following hypotheses are proposed:
H3.a. Relational governance positively moderates the relationship between buyer’s R&D spending and RACAP.
- 14 - However, according to Uzzi (1997), embeddedness in an interfirm network can create also inhibitions for the firm. Trust might also be the key to the ‘’dark side ‘’ of the relationship. The high level of trust between two partners can lead to opportunistic behavior since one may rely heavily on the other, and the mechanism created to increase the pie of benefits is ruined by the partners’ incentive to “cheat” on another (Anderson and Jap, 2005). Thus, the free-riding problem is common in interfirm relationships, therefore we might expect that contractual governance will mitigate this issue and there might be room for contract as well.
2.5 Contractual governance
In contractual governance, the use of formal contracts is emphasized in the relationship (Lee and Cavusgil, 2006). Formal contracts highlight the importance of rules and procedures as a means to protect the relationship against opportunistic behavior. By stipulating each aspect regarding responsibilities and obligations, and rights and duties, the partner safeguards the relationship and reduces possible opportunism (Williamson, 1985; Cao and Lumineau, 2015). The exchange of the information between partners of each party takes place according to the written agreements. Besides, in a relationship in which formal contracting governs the relationship formalization is increased (Carson, Madhok and Wu, 2006). Formalization refers to the extent to which specified instructions and procedures are formalized (Jansen et al. 2005; Khandwalla, 1977). Accordingly, partners are less flexible and are less likely to exceed the contract agreement. Moreover, when partners are connected by arm’s length tie, they are more likely to act in their own self-interest and the relationship is characterized by lack of reciprocity (Uzzi, 1997). The shortage of flexibility hinders a partner to convey the fine-grained information without potential loss of valuable know-how (Uzzi, 1997).
According to Polanyi (1962) firms which make the exchange via arm’s length tie, come across difficulties in transferring tacit know-how, because by definition tacit knowledge is hard to transfer if the partners are not engaged in personal contact with each other.
Thus, it can be expected that contractual governance negatively impacts the assimilation and acquisition of new information, therefore the firm wouldn’t be able to appropriate the investment made by R&D activities to explore new knowledge and this will represent a disincentive to invest.
H4.a: Contractual governance negatively moderates the relationship between buyer’s R&D spending and PACAP.
- 15 - However, formalization implies that actors of a relationship will not deviate from their specified tasks. Generally, formal contracts regulate what outputs must be provided (Poppo and Zenger, 2002). In consequence actors will tend to effectively retrieve the already generated knowledge due to the fact that formalization assists to reinforcing the application of existing information (Jansen et al. 2005). This in turn, will represent an incentive to invest due to the fact that exploitation of knowledge will be enhanced. For that reason, we expect the following relationships.
H5.a: Contractual governance positively moderates the relationship between buyer’s R&D spending and RACAP.
H5.b: Contractual governance positively moderates the relationship between supplier’s R&D spending and RACAP.
2.6 Relationship performance
In this study the relationship performance denotes the explorative and exploitative learning of the firm in an alliance. We refer to learning performance as the extent to which the alliance co-operation has resulted in joint creation of new expertise, namely, explorative learning performance illustrates the amount of new generated knowledge, while exploitative performance refers particularly to the realized outcomes from the co-operation, for instance, new product development, lower costs, flexibility, better product quality, etc.
In this paper the dyadic data from the buyer-supplier relationship makes it possible to appreciate the extent to which each partner benefits from the relationship. In many studies researchers have examined how firms that are engaged in collaborations generate economic rents and particularly benefits associated with partnerships, such as learning, lower transaction costs or maximization of resources (Dore, 1983; Dyer, 1996; Hamel, 1991; Dyer and Singh, 1998). In their study, Dyer and Singh (1998) refer to the benefits accumulated in a relationship as relational rents. Relational rents1 denote the “supernormal benefits generated jointly in the relationship, which cannot be created by a firm in isolation and requires joint idiosyncratic investments and specific alliance partners “(ibid.). Therefore, in our research setting we expect that the explorative and exploitative capabilities will impact the relationship performance, and particularly the interaction between partners will result in the joint creation of new expertise and additional benefits.
1
- 16 - The aim of this paper is to investigate the relationship between R&D expenses and absorptive capacity, and thus how it implicitly affects relationship performance. With this intention, we would like to evaluate the relationship between PACAP and RACAP and learning performance. Likewise, there exists a wide array of literature which demonstrates that absorptive capacity is central to the learning processes (Cohen and Levinthal, 1990; Tsai, 2001; Zahra and George, 2002 ;).
Above all, the relationship between ACAP and performance is of interest in our study since we investigate how a firm can enhance its innovative knowledge and competitive advantage by means of allocating resources to R&D activities. Thus, the expertise achieved by explorative learning, like new product development expertise and new technological expertise is a byproduct of ACAP and were proven to increase competitive advantage (Cohen and Levinthal, 1990; Grant, 1996a).
In line with Zahra and George’s study (2002), the distinguished multidimensional construct of absorptive capacity refers to PACAP as to the capability of assimilating and acquiring new knowledge, consequently, it can be expected that it will be positively related to the creation of new expertise, and as a result influence explorative learning. Accordingly, we propose the following hypothesis:
H6.a: PACAP has a positive influence on joint explorative learning performance.
However, as a company shifts from exploration phase to exploitation, the accumulated information needs to be grasped, refined and applied for commercial purposes. Namely, RACAP which refers to transformation and exploitation (Zahra and George, 2002) is expected to reflect on exploitative learning performance. Therefore, next hypothesis is developed:
H6.b: RACAP has a positive influence on joint exploitative learning performance.
In their study Ambrose et al. (2010) have found that buyers and suppliers have different perceptions of their relationship. Hence the authors argue that there are significantly different drivers of a relationship performance for buyers and suppliers in the same relationship. Moreover, another study on the differences in perspectives by Oosterhuis et al. (2013), confirms that on many attributes buyers and suppliers share different views, and in order to get better insights about the relationship, the differences in perceptions from both buyers and suppliers should be assessed. Therefore, suppliers and buyers might have discrepancies about the perceived benefits created by the collaboration, namely the learning performances generated by the explorative and exploitative capabilities.
- 17 - and suppliers. As a result, it is of interest to investigate whether the perceptual differences among the two partners will have any consequences for the model outcomes. Subsequently, we would like to test the following relationships:
H7.a. PACAP has a positive influence on the distribution of benefits generated by explorative learning.
H
7.b. RACAP has a positive influence on the distribution of benefits generated by exploitative learning.2.7 Conceptual model
By incorporating the proposed hypotheses, the conceptual model was formed:
- 18 -
3. METHOD
In this section the data collection, measures and the statistical procedure used for the data analysis will be depicted.
3.1 Data Collection
The objective of this study is analyzing the moderating effect of governance on the relationship between the R&D spending of the buyer and supplier. To conduct this study an existing database will be used. The data on 166 dyadic buyer-supplier relationships was gathered by Berger (2015) for his previous study on ACAP. Given the fact that relationships are not always observable, there is the probability of a measurement error (Berger, 2015; Selness & Sallis, 2003). Moreover, according to other empirical studies on interfirm relationships, there might be potential discrepancies in the perceptions of different aspects of the relationship among the two partners (Ambrose et al.2010). For this purpose, data was collected from key informants from both sides of the dyad (suppliers n=166, buyers n=166), by means of a detailed survey. The questionnaire encompasses questions regarding the matters of the collaboration between the firms, and in particular it investigates the influence of relationship´s explorative and exploitative capabilities on relationship performance.
For the empirical research the following industries have been analyzed: automotive, machinery, chemicals, pharmaceuticals, semiconductors, and electronics industry. To support variance, the firms selected for the research activate in different industries, given that according to a prior study it was found that these industries illustrate different knowledge strategies (Lichtenthaler and Ernst, 2007; Berger, 2015). General questions determined the working positions of the key informants, duration of participation in the relationship, and the percentage of time spent on the relationship. The answers to these questions revealed the degree to which the respondents were involved in the relationship and their knowledgeability (Campbell, 1955; Berger, 2015).
- 19 -
3.2 Measurements
The scales selected for this study have been established previously in other studies, in the context of intrafirm, interfirm and firm learning. These have been adapted and further extended by Berger (2015) to correspond to vertical, interfirm relationships. The questionnaire is comprised of first-and second-order constructs. Thus, first-second-order construct is a latent construct and its indicators depict the observable variables, while the second-order construct has first-order constructs as their indicators. Furthermore, in the present study we distinguish between reflective and formative measurement approaches. In case of the reflective measurement model, it’s the latent construct that causes the change in the value of the indicators. In case of the formative model, it´s the measurement items (indicators) which cause the change in the latent construct. In the reflective measurement model, the correlation between the indicators will be high. On the other hand, in case of the formative model, the construct validity is of greater importance, and as suggested by researchers, the correlation among the indicators in the formative model should be as low as possible, since they are assumed to measure different aspects of the same construct (Bollen and Lennox, 1991; Diamantopoulos and Winklhofer, 2001; Berger, 2015).
The measurement scale applied from previous study is annexed in the Appendix 1. The scales used for the form questions were 7-point-likert-scales. Thus, labelled from 1= strongly disagree to 7= strongly agree. Excluding the answers given to the questions which measured explorative learning performance, these were labelled: 1= to no extent; 7= to a great extent. The English version of the survey is offered in the Appendix 1. As follows, the measures of the variables identified in this study are depicted.
The variables of interest in this research paper are R&D spending from both sides of the dyad, contractual governance, relational governance, potential absorptive capacity (PACAP), realized absorptive capacity (RACAP), exploitative learning performance, explorative learning performance and the distribution of benefits. The constructs were adopted from previous studies. The following section outlines how constructs were measured on the survey.
3.2.1 Independent variables
- 20 - to take into consideration the approach of how to deal with the missing values. A common strategy to deal with the missing data is the deletion method, the list wise deletion and pairwise deletion. The list wise deletion approach implies deleting from the data set any observations that have missing data. However in our case, we assume that this approach can yield biased results since this method supposes a reduction of the sample size and we risk a loss of information and reduction of the power of the estimate. The second approach is the pairwise deletion, which tries to minimize the loss which might occur by list wise deletion. Nevertheless, according to Little (1992) this method is a source of impossible values in covariance matrices which can cause estimation issues in multivariate analyses such as Structural Equation Modeling. Another method for treating the missing values is the mean replacement. According to this approach, the arithmetic mean of the variable is calculated and the missing values are substituted with the mean. It will not change the sample size nor the sample mean of the variables. Therefore, this method allows maximizing the usefulness of the collected data (Little, 1988). We analyze the descriptive statistics in SPSS in order to see how these missing values are distributed among the two groups. The table below indicates descriptive statistics of the R&D variable. Consequently, we continue with the path analysis and the existing missing values which measure the variable in the data set will replaced by mean substitution.
Table 1: Descriptive statistics of the variable R&D (buyer-supplier).
3.2.2 Moderator variables
Contractual governance is denoted by ”contracting” in the measurement construct. It indicates the degree to which formal agreements control the buyer-supplier relationships (Geringer and Hebert, 1989). For this construct the measures were collected by Deshpande and Zaltman (1982), Buvik and Reve (2002), Cannon and Perreault (1999), and Jansen et al. (2006).
Relational governance is associated to ”relational norms” which is a second order construct. It comprises the measures developed previously by Heide and John (1992). Namely, relational norms are described by flexibility, information exchange and solidarity norms. These represent first-order constructs.
N Minimum Maximum Mean Std. Deviation
%R&D (buyer) 131 ,00 20,00 5,6527 5,40752
- 21 -
3.2.3 Dependent variables
PACAP and RACAP. This study takes into account the multidimensional construct of ACAP defined by Zahra and George (2002), for explaining its dynamic capability. Zahra and George (2002) conceptualized four stages of building ACAP and grouped them into two separate constructs. Thus, PACAP and RACAP represent a second-order construct.
The first-order constructs which measure PACAP are: acquisition, which refers to the “relationship’s capability to identify and acquire externally generated knowledge that is critical to its operations” (Zahra & George,2002;Berger,2015); and assimilation, which denotes “the firm’s routines and processes that allow it to analyze, process, interpret and understand information obtained from external sources” (Zahra & George,2002;Berger, 2015).
Consequently, RACAP is measured by the first-order constructs, transformation and exploitation. Exploitation is based on: “routines that allow firms to refine, extend, and leverage existing competencies or to create new ones by incorporating acquired and transformed knowledge into its operations” (Berger, 2015). Along with transformation which stands for “a firm’s capability to develop and refine the routines that facilitate the combination of existing knowledge as well as the newly acquired and assimilated knowledge” (Berger, 2015).
Explorative learning performance. The measure for this variable was previously identified by Lane et al. (2001), as “learning”. It refers to the extent to which the firm’s collaboration benefited from “joint creation” of ” new expertise ” in several areas, new product development, manufacturing and production, technology, etc.
Exploitative learning performance. Likewise, the measure for this variable was earlier depicted by Selnes and Sallis (2003). It regards the firm’s capabilities to make sense of new acquired knowledge, integrating this information into relationship domain-specific activities (Selnes and Sallis, 2003).
- 22 -
3.3 Statistical procedure
One of the essential elements of the conceptual model depicted in this research paper is the presence of a moderating effect of the governance structure. A moderation occurs when the variable X is supposed to cause variable Y, and the moderator variable M is assumed to be a variable that modifies the strength and/or direction of the causal relationship between X and Y (Baron & Kenny, 1986). Normally, moderator effects are denoted by the interaction of X and M in explanation of Y.
In the presented conceptual model, the dependent variable (PACAP/RACAP) is a function of the predictor variable (R&D buyer/supplier), the moderator variable (governance structure) and the interaction between moderator variable and the predictor variable. The below figure denotes the visual representation of a moderation effect.
Figure 2. Moderator model.
The coefficient c measures the moderation effect
;
a measures the simple effect of X, also referred to as the main effect of X, when M is zero. To assess the moderation effect, the differential effect of the independent variable on the dependent variable as a function of the moderator must be measured and tested (Baron & Kenny, 1986). The multiple regression analysis must be run to test for these effects.- 23 -
4. ANALYSIS AND RESULTS
In the present chapter, the results of the data analysis will be discussed. Once we have denoted the measurement model, we would like to precede the data analysis by evaluating the outer and inner model. Therefore, first we address the outer model. The aim is to assess the existing relationships between the latent constructs and the associated measurement items. Subsequently, if the outer model will show to be strong enough, the inner model will be estimated. The structural model, (referred to as the inner model) estimates the relationships between the latent constructs. We advance the data analysis by estimating the outer model for reliability, convergent validity, discriminant validity. A soon as the validity and reliability have been established, we continue with the inner model and path analysis which allows us to evaluate the hypothesized relationships. The data analysis will be performed in SMART-PLS.
4.1 Measurement model
- 24 -
4.2 Outer model
In order to precede the analysis of the inner model, we first test for validity and reliability the measures items associated to the latent constructs. In order to do this, we need to distinguish between formative and reflective measurement models when evaluating the reliability and validity. There are two different approaches for evaluating the reliability and validity depending on the type of measurement, reflective or formative.
4.2.1 Reflective constructs
Reliability is employed to measure for internal consistency of a construct. There are two indexes which denote reliability: Composite Reliability and Cronbach’s alpha. In order to assess the reliability of the reflective scales we inspect these coefficients. The estimated Cronbach’s alpha coefficient shows whether all the indicators contribute correspondingly to the total reliability. Generally, the Cronbach’s alpha is said to underestimate the internal consistency reliability, therefore we also consider the Composite Reliability coefficient (Hair et al.2014). The reliability index must be > 0.7. Furthermore, we examine the outer loadings of the indicators, which have to be equal or greater than 0.5.
Next, we evaluate the reflective scales by testing for convergent and discriminant validity. Convergent validity is frequently used to determine the correlation of the dimension’s indicators. It is satisfactory if the subsequent conditions are met: a) the indicator loadings must be statistically significant, with a p-value lower than 0.5; indicator loadings equal to or greater than 0.5; b) reliability index >0.7; c) average variance extracted (AVE) is greater than 0.5.
The aim is to examine whether the indicators used for measuring the construct are linked to each other, specifically we want to show a convergence among related constructs (Trochim and Donnelly, 2006). Therefore, given that the indicators of the reflective scales are assumed to be correlated to each other, we test the measurements items of the model for convergent validly.
- 25 - Table 2: Construct reliability and convergent validity of reflective scales.
Constructs/
Indicators:
Complete Supplier Buyer Outer Loadings Cronbach’s alpha Composite Reliability Outer Loadings Cronbach’s alpha Composite Reliability
- 26 -
4.2.2 Formative constructs
In case of the formative scales, the measurement items are not supposed to be related among them due to the fact that the indicators cause the change in the latent construct. Therefore, a different approach for assessing the validity will be selected (Jarvis et al., 2003). Hence, we evaluate the convergent validity of the formative scales by examining the VIF scores of the indicator weights. The VIF score addresses the issue of multicollinearity. Given that the formative construct indicators represent different aspects of the construct, high collinearity between indicators might hamper the results (Hair et al. 2014). The VIF score should be below the threshold value of 3.3 subsequently, we check the indicator weights for statistical significance by looking at the p-values, a value lower than 0.05 is expected for meeting the requirements of validity (Hair et al. 2014). For the reflective construct Acquisition, the indicator Acquis.2 showed an insignificant p-value, therefore it doesn’t meet the necessary requirements for validity and we removed it from the measurement model. Likewise,for the latent construct Transformation, the indicators Transf.1 Transf. 3, Transf. 5 were deleted as they have revealed insignificant p-values. Removing these indicators in this case wouldn’t harm the conceptual domain of the construct, since other measurement items of the construct capture the same common theme (Coltman et al. 2007).
Constructs/ indicators:
Complete Supplier Buyer
Indicator weights
p-value: VIF: Indicator
weights:
- 27 - Table 3: Convergent validity of formative scales.
The following step is assessing the reflective and formative measurement scales for discriminant validity. The test for discriminant validity calculates the squared roots of average variance extracted (AVE) of the latent construct and depicts the correlation matrix of the latent construct with other existing constructs. The coefficient of AVE denotes the ”grand mean value of the squared loadings of a set of indicators“ (Hair et al., 2014). The squared root of average variance extracted (AVE) is expected to be > 0.5.; the values of AVE are indicated on the diagonal of the below table. The coefficients on the off-diagonal denote the correlations of the latent constructs. For adequate discriminant validity, the values on the diagonal are estimated to be higher than any of the correlations (Hair et al. 2014). Accordingly, each construct shares more variance with its own indicators than with other latent constructs. The method which assesses for discriminant validity is Fornell and Larcker (1981) criterion. This method is applied for both formative and reflective models.
The correlation coefficients represented in the Table 4a and 4b are all above the threshold valueof 0.5
and the square root of the average variance extracted shows values higher than the correlations of the specific construct. Table 4b hence depicts the discriminant validity for the second-order constructs
while Table 4a assembles the results for the first-order construct.
The results in the following Tables 4a and 4b are shown for the complete dataset given that the results
- 28 -
Table 4a: Squared roots of the average variance extracted (AVE) and correlation matrix (first-order constructs).
In the tables 4a and 4b, we inspect the squared roots of the average variance extracted and the
correlation matrix for first order-constructs and second-order constructs.
Table 4b: Squared roots of the average variance extracted (AVE) and correlation matrix (second-order constructs).
Accordingly, the values meet the prerequisites for satisfactory convergent and discriminant validity. Therefore, the evaluated measurement scales have shown to be strong enough in order to proceed further with the interpretation of the inner model. The measurement models have been examined for both datasets, “buyer” and “supplier”.
1 2 3 4 5 6 7 8 9 10 1. Acquisition 0,726 2. Assimilation 0,646 0,739 3. Contracting 0,121 0,064 0,707 4. Exploitation 0,544 0,652 0,102 0,723 5. Exploitative Learning Performance 0,424 0,531 0,092 0,524 0,637 6. Explorative Learning Performance 0,266 0,315 0,120 0,269 0,543 0,778 7. Flexibility 0,182 0,346 0,008 0,351 0,340 0,097 0,811 8. Information exchange 0,246 0,380 0,074 0,365 0,337 0,220 0,534 0,866 9. Solidarity 0,303 0,413 0,022 0,379 0,340 0,214 0,504 0,673 0,777 10. Transformation 0,550 0,658 0,076 0,687 0,528 0,282 0,305 0,394 0,489 0,749
1.PACAP: (Acquisition + Assimilation)
1
0,666
2 3
2.RACAP: (Transformation + Exploitation) 0,505 0,663
- 29 -
4.3 Structural model
In the previous section we have assessed the reliability and validity of the outer model for data sets, buyer and supplier. In order to evaluate the proposed hypotheses of the current research paper, we advance the analysis by examining the inner model. Namely, the structural model, which according to a structural equation modeling (SEM) analysis, entails the evaluation of the relationships between the latent constructs. In order to continue with the estimation of the relationships among constructs, we first assess the quality of the inner model. The criteria which evaluate the model’s quality are the following: Coefficient of determination ( R2), Predictive validity (Stone-Geiser Q-square), Collinearity values (VIF).
The value of the R2 indicates the explanatory power of the endogenous latent variable. The value of R2 canrange from 0 to 1, closer to 1 shows a considerable predictive accuracy (Hair et al. 2014). The principal measure of predictability is the Stone-Geiser Q-square (Henseler et al., 2009). The value of Q2 denotes the predictive validity of the model. In the case the coefficient Q2 shows a value, of 0.02, 0.15, and 0.35, the values denote a small, medium, or large predictive relevance of a certain latent variable and the model’s predictive accuracy is satisfactory (Henseler et al., 2009; Hair et al. 2014; Berger, 2015). VIF values indicate whether there is a collinearity issue among the latent construct. A value below 3.3 is expected.
Table 5: Full collinearity VIF, explained variance and predictive validity measures (PLS-SEM).
Full collinearity VIFs: Explained variance R2 Predictive validity:
Stone-Geiser Q-square
Buyer supplier Buyer Supplier Buyer Supplier
- 30 - Accordingly, in the table represented above the multicollinearity scores are below the cut-off point of 3.3, which denotes that there are no multicollinearity issues among the variables in the data sets, buyer and supplier. This leads to the conclusion that the structural model meets the specified requirements (Cenfetelli and Bassellier, 2009; Berger, 2015). The values which refer to the variance explained and the predictive validity of the endogenous latent variables range from 0.04 to 0.44. Therefore, the values are sufficient to continue with further interpretation of the results.
In the next step, we perform the Multi-group analysis (MGA) procedure in SMART PLS. This procedure allows us to identify the significant differences among the two groups of buyer and supplier. The estimated parameters indicate the relationships among constructs (paths) (Hair et al. 2014). In order to test for significance of the path coefficients the bootstrapping procedure is performed (ibid.) The previous studies on interfirm relationships reveal differences in perceptions among buyers and suppliers (Ambrose et al. 2010; Oosterhuis et al. 2013; Berger, 2015). One approach for assessing for differences between the two parties is the Welch-Satterthwait Test. It is one of the methods of the multi-group analysis which allows for testing the dissimilarities of group-specific PLS-SEM results, by assuming unequal variances across groups. The path coefficients given as a result of the PLS structural model can be interpreted as standardized betas (Henseler et al., 2009).
Table 6: Standardized path coefficients and multi-group analysis results (PLS-SEM). *** p ≤ 0, 01; ** p ≤ 0, 05; *
Independent Dependent Buyer
- 31 - In the MGA analysis, the beta coefficients are used for the interpretation of the direct and moderating effects.Namely, to interpret the path coefficients (betas) we have to be sure that the p-values are below the specified levels: p ≤ 0, 01; p ≤ 0, 05; p ≤ 0, 1; TheWelch-Satterthwait Test, assumes the variances are not homogeneous among groups. We obtained p-value smaller than 0.05, for the relationship between both dimensions of ACAP and the distribution of benefits. We can conclude that the means of the two groups are significantly different. For the remainder of the relationships, the interpretation of the t-test shows that only there are no significant differences among buyers and suppliers. Thus, given these estimates we can proceed by evaluating the hypothesized relationships.
4.4 Hypothesis testing
In the following section, we would like to test whether the hypotheses proposed in the research model are accepted or rejected. Accordingly, we analyze the structural model and the results presented in Table 6. The estimated parameters refer to both “buyer only” and “supplier only” separately. An important step in testing the hypotheses is to evaluate the direct and indirect relationships. In our research model, we discuss the potential occurrence of a moderating effect.
Table 7. Overview of (supported and rejected) hypotheses (PLS-SEM).
Independent => Dependent Buyer Hypothesis Supplier Hypothesis
R&D => PACAP
R&D x Relational Norms => PACAP R&D x Contracting=> PACAP
H2a H4a Rejected Rejected H1a H2b H4b Rejected Rejected Rejected R&D => RACAP
R&D x Relational Norms=> RACAP R&D x Contracting => RACAP
H1b H3a H5a Rejected Rejected Rejected H3b H5b Rejected Rejected
PACAP => Explorative Learning Performance PACAP => Distribution of Benefits
H6a H7a Accepted Accepted H6a H7a Accepted Rejected
RACAP => Exploitative Learning Performance RACAP => Distribution of Benefits
- 32 - 4.4.1 Buyer database
With respect to the buyer database the hypothesis
H
1b, tests whether buyer’s R&D has a strongercorrelation with RACAP than the supplier’s, (=0.042, p=0.434), the values show an insignificant relationship, therefore we reject the hypothesis. The next hypothesis H2a tests for the positive indirect
effect of R&D and relational norms on PACAP (x=-0.105, p=0.108), the values show that the moderating effect is insignificant, thus the hypothesis is not being supported. As follows, the H3a tests
for the analogous moderation on RACAP (x=-0.055, p=0.336), according to the results, there was no evidence for a positive moderation effect of R&D and relational norms on RACAP, hence H3a was left
unsupported. The hypothesis H4a tested whether there occurs a negative indirect effect of R&D
interacting with contracting on PACAP (x=0.110, p=0.106), the results of the moderation show to be insignificant therefore we reject the proposed hypothesis. Likewise, the hypothesis H5a aims to assess
whether there is a positive indirect effect of R&D and contracting on RACAP (x= -0.094, p= 0.142), according to the values this effect has been found to be insignificant and H5a unsupported.
Thus, contrary to expectations, there hasn’t been found any moderating effect of the governance mechanism on the relationship between R&D and ACAP. Given that there wasn’t found a direct effect of R&D on ACAP, it is admissible that the moderation effects were found to be insignificant.
Next, we evaluate the hypotheses regarding ACAP and relationship performance. Accordingly, PACAP has found to have a significant positive effect on explorative learning performance (=0.332, p=0.000), thus H6a has been supported. H6b suggests a positive relationship between RACAP and exploitative
learning performance (=0.675, p=0.000), thus the data found evidence for the suggested positive relationship. Furthermore, we test the impact of PACAP on the distribution of explorative learning performances, the values (=0.181, p=0.040) imply a positive, significant relationship, hence H7a is supported. The las hypothesis H7b expects RACAP to impact the distribution of exploitative learning performances, (=0.323, p=0.000), the values given by the results specify support for the hypothesis. 4.4.2 Supplier database
With regards to the supplier-side, the effect of R&D on PACAP has not been found to be statistically significant (=0.015, p=0.828), thus H1a is rejected. Next, we evaluate the assumed moderation effects.
- 33 - assumed to be the contractual form of governance, thus two hypotheses with respect to contracting have been depicted. According to H4b
we expect a negative indirect effect of R&D and
contractingon PACAP, hence the data doesn’t indicate support for the proposed hypothesis (=-0.053, p=0.485). H5bassumes there is a positive indirect effect of R&D and contracting on RACAP (= -0.005, p=0.950),
the results reject the suggested relationship.
Furthermore, the hypotheses H6a and H6b expect a positive relationship between PACAP and
explorative learning performance (=0.368, p=0.000), and RACAP and exploitative learning performance (=0.546, p=0.000), the values provide support for both hypotheses. Lastly, we assess the influence of PACAP on the distribution of benefits generated by explorative learning, the values (=0.086, p=0.664) imply an insignificant relationship, hence H7a is left unsupported. H7b aims to test whether RACAP impacts the distribution of exploitative learning performances, (=0.170, p=0.562), hence results don’t indicate support for the hypothesis.
4.4.3 Distribution of benefits
A focal point of the research is the potential occurrence of differences in perspectives among the two parties in the relationship. In this research paper the dyadic data from buyers and suppliers allows us to assess which party perceived to have benefited most in the interfirm relationship. We discuss in the previous chapter that parties engage in relationships with partnering firms in order to achieve valuable knowledge and resources which are critical to the firm’s competitive advantage (Dyer and Singh, 1998). Thus, the benefits drawn out of collaboration represent an incentive to further invest in the relationship (ibid.). Interfirm relationships have been proved to deliver two distinctive potential benefits, operational efficiency and new knowledge creation (Malhotra, 2005; Dyer, 1997).
In our study we aim to examine the distribution of performances generated by the relationship between buyers and suppliers.
- 34 - where all measurements from 1 to 3 will indicate that the firm “appropriated less” benefits than the partner; the scale 4 refers to “equal distribution“ and the remainder measurements from 5-7 suggest that the firm “appropriated more” than their partner. Apportioning the sample into three groups out of seven existing, assumes a potential loss of information, however to be able to give a connotation to the differences in perceptions, we decided to continue with the approach of 3 ranges.
Thus, once we have rescaled the variables for the supplier side, we can make a distinction between three levels of distribution of benefits. Therefore, in the below tables we can get a visual representation in form of percentages of how the benefits were distributed among buyers and suppliers in the relationship. For the complete representation of each of the items explaining the variable distribution of benefits and the corresponding percentages, see Appendix 2.
Table 8.a. Distribution of benefits (buyer’s perception).
Likert-scale value 1-3(Appropriated less) 4(Equal Benefits) 5-7(Appropriated more) Total
Explorative learning 20,8 % 49,8 % 29,4 % 100%
Exploitative learning 14,28% 45,7% 38,7% 100%
Table 8.b. Distribution of benefits (supplier’s perception).
Likert-scale value 1-3 (Appropriated less) 4 (Equal Benefits) 5-7(Appropriated more) Total
Explorative learning 33% 48% 19% 100%
Exploitative learning 35,8% 47,4% 16,8% 100%
From the Table 8a we can infer that on average buyers believe that the benefits developed by the collaboration were distributed equally. However, 1/3 of the buyers believe they have adopted more expertise from exploitative learning than their partner, and have appropriated more performances by exploitative learning. Table 8b illustrates the results from the supplier’s point of view, we certainly see that the suppliers score higher in the range from 1 to 3 which denotes that about 1/3 think they have appropriated less benefits in the relationship. We can also deduct from the results that approximately half of the suppliers still believe the benefits were apportioned equally. Moreover, fewer suppliers recognize to have acquired more benefits from the relationship than their partner.
- 35 -
5. DISCUSSION
The following chapter is comprised of the summary of the empirical results and discussion. First, we discuss the findings regarding the direct effects of the researched relationships. Second, we discuss the moderating effects of the governance mechanism configurations. Subsequently, we draw inferences on the scientific and managerial implications of the study, and finally to conclude we discuss the limitations of the proposed study and directions for further research.
5.1
Direct effects
The major focus of the present research paper is to determine if the resource allocation of the buyer and the supplier is correlated to absorptive capacity. The conceptual model suggests that there is a direct effect of R&D spending on interfirm absorptive capacity.
Hence, the first two hypothesized relationships suggest there is a correlation between R&D spending and PACAP from the supplier side only, and an expected positive relationship between R&D spending and RACAP, from the buyer’s side. Our results don’t indicate support to the theorized relationships. In consequence, we decided to carry a post-hoc analysis to determine the opposite relationships from the other side of the dyad, and find whether there is a pattern explaining these relationships. The results didn’t prove to be different, and there hasn’t been found any link between R&D and ACAP dimensions. Buyers and suppliers don’t reveal any dissimilarity in the proposed relationships.
The findings of our study are opposing our expectations. Several studies investigated the relationship between R&D spending and knowledge accumulation and there are some mixed results among scholars. In their study Lane and Lubatkin (1998) extended on Cohen and Levinthal’s concept of absorptive capacity. The authors shift their focus from firm-level absorptive capacity and asses a model of inter-organizational learning. Hence, the authors argue that R&D spending doesn’t have great explanatory power in explaining absorptive capacity, which is in contradiction with Cohen and Levinthal’s (1989) view.
- 36 - the findings of Deeds (2001), greater commitment to R&D should result in greater flow of new information into the firm. Hence a distinction between a consistent flow of R&D resources and a low investment of resources is necessary to fully grasp the relationship it has with interfirm learning. Furthermore, in the present paper we have found evidence for the proposed direct effect of absorptive capacity on the relationship performance. Table 6 indicates that the direct effects of PACAP on explorative learning performance and RACAP on exploitative learning performance prove to be positive and statistically significant for both supplier and buyer databases. These effects are according to our expectations and in line with the findings of Lane et al. (2001).
Additionally, with regards to the relationship performance we expect a positive direct effect of PACAP on the distribution of explorative learning performances. Hence, we find evidence for the ‘buyer-only’ database and no evidence for the supplier database. Likewise, we assume that RACAP has a positive impact on the distribution of benefits from exploitative learning. Our results indicate a significant positive relationship for the ‘buyer-only’ dataset, while no effect was found for the supplier side. Our findings are in agreement with the present studies on the differences in perceptions in the supply chain relationships (Ambrose et al. 2010; Oosterhuis et al. 2013). For evaluating this discrepancy among the buyers and suppliers we have calculated the percentages of the benefits distributed among partners, according to the scores on the variable corresponding to the distribution of performances (appendix 1.7; 1.9). In fact, as mentioned in the previous paragraph, we have found that buyers and suppliers scored differently on the variable which intended to capture how the performances from explorative and exploitative learning were apportioned. Therefore, this finding represents a potential argumentation for the negative relationship between the dimensions of ACAP and the distribution of benefits.
5.2
Moderation effects
In accordance with the literature, there are attributes which enhance the potential of the firms to share knowledge and develop new expertise and performances (Malhotra 2005) hence, in the present research paper we focus on the relational and contractual governance mechanisms that have different implications in deriving benefits from partnerships. We argue that governance mechanisms can influence R&D investment in building absorptive capacity, by enabling better exploration and exploitation of information (ibid.).
