0
University of Amsterdam
How knowledge sharing mediates the relationship between learning
mechanisms and competitive advantage
A Quantitative Study of Real Estate Agencies in the Netherlands
Author: M.J.J. Holterman
Student number: 10475435
Date of Submission: June 27, 2017
Final version
University of Amsterdam
Faculty of Business and Economics Amsterdam Business School Executive Programme in Management Studies
Track: Strategy
1
Statement of Originality
This document is written by student M.J.J. Holterman who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this
document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
2
ABSTRACT
This study aims to examine to what extent the relation between learning mechanisms and a competitive advantage is mediated by knowledge sharing. Learning mechanisms help shaping dynamic capabilities which in turn could lead to a competitive advantage. Dynamic capabilities are essential for firms to be successful in dynamic markets with fierce competition. Based on the literature (Teece, Pisano & Shuen, 1997; Zollo & Winter, 2002) it can be assumed that the step from learning mechanisms, as the underlying construct of dynamic capabilities, to a competitive advantage seems too simplistic and is a black box. To address this research gap, it is expected that there is a mediating effect of knowledge sharing on the relation between learning mechanisms and competitive advantage.
An online survey was sent to 480 agencies, active in the segment housing in real estate. Findings illustrate that (1) experience accumulation, (2) knowledge articulation and (3) knowledge codification (three learning mechanisms) are positively related to competitive advantage and knowledge sharing. This research also provides evidence that knowledge sharing is positively related to competitive advantage. Based on this research it can be concluded that the expected mediating effect on the relation between the three learning mechanisms and competitive advantage is absent. Implications and future avenues for research are also offered in this study.
Keywords: Dynamic capabilities; learning mechanisms; knowledge sharing; competitive
3 LIST OF FIGURES
FIGURE 1: Conceptual model………..12
FIGURE 2: Mediator……… 34
LIST OF TABLES TABLE 1: Distribution of provinces……… 23
TABLE 2: Distribution of years work experience……… 23
TABLE 3: Distribution of number of employees………. 24
TABLE 4: Cronbach’s Alpha………... 24
TABLE 5: Mean, Standard deviation, minimum and maximum for all variables………... 27
TABLE 6: Normal distribution all variables……… 29
TABLE 7: Correlations……… 30
TABLE 8: The influence of the control variables on competitive advantage (ANOVA)…………... 31
TABLE 9: Regression results hypotheses 1 and 3………... 33
TABLE 10:Regression results hypotheses 2………... 34
TABLE 11: Regression results mediating effect……….. 35
TABLE OF CONTENTS STATEMENT OF ORIGINALITY………...1 ABSTRACT………... 2 1. INTRODUCTION……….. 4 2. LITERATURE REVIEW………... 7 2.1 MAIN CONSTRUCTS……… 7 2.2 CONCEPTUAL MODEL……… 12 2.3 HYPOTHESES………..13
3. DATA AND METHOD………. 18
3.1 DESCRIPTION OF THE SURVEY……….…...…...18
3.2 POPULATION AND SAMPLE………...18
3.3 CONSTRUCTING THE SURVEY AND DATA COLLECTION………....19
3.4 MEASUREMENTS OF VARIABLES……….…...20
3.5 RELIABILITY AND VALIDITY………..24
3.6 STATISTICAL STRATEGY……….25 4. RESULTS……….. 27 4.1 UNIVARIATE ANALYSIS………. 27 4.2 BIVARIATE ANALYSIS………29 4.3 REGRESSION ANALYSIS……….30 5. DISCUSSION……… 36
5.1 FINDINGS RELATED TO THE LITERATURE……….….36
5.2 RESEARCH CONTRIBUTIONS..……… 39
5.3 LIMITATIONS AND FUTURE RESEARCH………. 40
6. CONCLUSION……….. 42
7. REFERENCES………... 43
8. APPENDICES 8.1 SURVEY………. 48
8.2 MEAN AND STANDARD DEVIATION FOR THE ITEMS OF THE SURVEY…... 49
4
1. INTRODUCTION
Firms are operating in increasingly complex and dynamic markets (Teece, Pisano & Shuen, 1997; Powell & Snellman, 2004; Sirmon, Hitt, Ireland & Gilbert, 2011). Competition becomes fiercer, new technological developments follow each other in rapid succession, customers become more demanding and market boundaries are fading. It is difficult for firms to obtain a competitive advantage in such dynamic and discontinuous environments and to
maintain acompetitive advantage (Teece, Pisano & Shuen, 1997; Eisenhardt & Martin, 2000).
Obtaining a competitive advantage is a central theme for firms, but why? It is important because it is connected to value creation (Rumelt & Kunin, 2003) and performance (Barney, 2001; Porter, 1991). Performance in turn relates to outperforming the competition (Chan Kim & Mauborgne, 2005). In order to build and maintain a competitive advantage in such discontinuous environments a firm could deploy dynamic capabilities (Teece, Pisano & Shuen,
1997; Eisenhardt & Martin, 2000). Because the article of Teece, Pisano and Shuen (1997)1 is
one of the most influential articles about dynamic capabilities this will be the starting point for this Research.
Various literature acknowledge there is a relation between dynamic capabilities and
competitive advantage (Teece, Pisano & Shuen, 1997; Eisenhardt & Martin, 2000; Teece,
2007). Since the article of Teece, Pisano and Shuen (1997) much has been written and published about the concept of dynamic capabilities. Does this flow of information makes the concept crystal clear and not open to discussion? Not at all. Zahra, Sapienza and Davidsson (2006) argue that the emergent literature on dynamic capabilities and their role in relation to value creation is fulfilled with inconsistencies, overlapping definitions and outright contradictions.
There are also similarities in the literature about the concept of dynamic capabilities. These similarities can be bundled under the denominator learning. Zollo and Winter (2002)
5
state that dynamic capabilities are shaped by the coevolution of learning mechanisms. According to Eisenhardt and Martin (2000) well-known learning mechanisms guide the evolution of dynamic capabilities. Zollo and Winter (2002) argue that learning mechanisms consist of experience accumulation, knowledge articulation and knowledge codification. This organizational learning could lead to knowledge (Bierly, Kessler & Christensen, 2000). Important to note is that having knowledge alone is not enough to create a competitive advantage. Teece (2007) argues that a sustainable competitive advantage requires more than the ownership of difficult to replicate (knowledge) assets. In order to create a competitive advantage knowledge must be unleashed and shared (Wang & Noe, 2010).
The earlier mentioned literature acknowledge a connection between dynamic capabilities and a competitive advantage (Teece, Pisano & Shuen, 1997; Eisenhardt & Martin, 2000; Teece, 2007) and state that dynamic capabilities are shaped by the coevolution of learning mechanisms (Zollo & Winter, 2002). At the same time the literature is not clear how dynamic capabilities and, to be more specific learning mechanisms, lead to competitive advantage. Based on previous, the conclusion can be drawn that the step from learning mechanisms to competitive advantage can be considered as a black box.
This research assumes that there is a mediating effect on the relation between learning mechanisms and competitive advantage. This claim can be substantiated on the work of various authors. According to Zahra, Sapienza and Davidsson (2006) the relation between dynamic capabilities and performance is mediated by the quality of substantive capabilities. This research, however, focuses on learning mechanisms and not dynamic capabilities themselves or its different appearances (Teece, 2007).
Important to note is that in the literature performance is defined differently than competitive advantage. Performance could be seen as the performance of the firm itself and is strongly related to return on investment and profitability (Calantone, Tamer Cavusgil & Zhao,
6
2002). Competitive advantage is a wider concept and compares the own firm performance with that of competitors and its actions in relation to the market (Li, Ragu Nathan, Ragu Nathan & Subba Rao, 2006). Grant (1996) and Argote and Ingram (2000) also give ground for assuming a mediating effect on the relation between learning mechanisms and competitive advantage. These authors argue that the creation and transfer of knowledge in a firm could lead to a competitive advantage.
Having taken note of all the forgoing the following research question is formulated: “To
what extent does knowledge sharing mediate the relationship between (1) experience accumulation, (2) knowledge articulation and (3) knowledge codification (three learning mechanisms), and competitive advantage?
By itself the constructs are not new. What makes this research distinctive is that these constructs are placed in one model. In order to answer the research question a quantitative cross-sectional survey is carried out under 480 (N) Real Estate Agencies in the Netherlands active in the segment housing and who are connected to the NVM (Nederlandse Vereniging van Makelaars).
Moving forward, this study consists of five additional chapters. In the next chapter an overview of the relevant literature and the hypotheses that are grounded on this literature are given. Chapter three gives a description of the survey, population and sample, the construction of the survey and data collection, measurement of the variables, reliability and validity and outlines the statistical strategy. In chapter four the univariate, bivariate and regression analyses are demonstrated. Chapter five is about the findings related to the literature, the contributions of the research, limitations and suggestions for future research. The last chapter offers a conclusion about this research and the key findings.
7
2. LITERATURE REVIEW
The first section of this chapter starts with an overview of some relevant constructs which are central to the research. The next section connects the different constructs and leads to the hypotheses. This chapter ends with a visualization of the conceptual model.
2.1 Main constructs
The literature review starts with a closer look at the constructs and theories that are relevant for the research. These constructs and theories are: dynamic capabilities, learning mechanisms, competitive advantage, knowledge and knowledge sharing.
Dynamic capabilities
It is known that the dynamic capabilities view is an extension of the resource based view. Clear is, what dynamic capabilities are (Eisenhardt & Martin, 2000; Teece, Pisano and Shuen, 1997; Helfat & Peteraf, 2003; Winter, 2003; Teece, 2007) and how they have evolved (Zollo & Winter, 2002). According to Teece (2007) dynamic capabilities could be seen as the capacity to sense and shape opportunities and threats, seize opportunities and maintain competitiveness through enhancing, combining, protecting and when necessary, reconfiguring the firm’s intangible and tangible assets.
Zahra, Sapienza and Davidsson (2006) see dynamic capabilities as abilities to reconfigure a company’s resources and routines in a manner envisioned and deemed appropriate by a company’s principal decision maker(s). Winter (2003) state that dynamic capabilities are those that operate to extend, modify or create ordinary capabilities. With his definition Winter (2003) is connecting with Eisenhardt and Martin (2000) because both built their definition around reconfiguration and modification of resources. On the basis of the written above, dynamic capabilities could be related to an analytical framework (Teece, Pisano & Shuen, 1997) but also to processes (Eisenhardt & Martin, 2000) or routines (Zollo & Winter, 2002).
8
Even though we know all the previous, the concept of dynamic capabilities is still surrounded with inconsistencies, complexity and contradictions (Zahra, Sapienza & Davidsson, 2006; Barreto, 2010; Bromiley & Arend, 2009; Winter, 2003). In order to make the concept of dynamic capabilities more specific we follow Zollo and Winter (2002). Zollo and Winter (2002) argue that learning mechanisms could lead to dynamic capabilities and these dynamic capabilities could lead to a competitive advantage (Teece, Pisano & Shuen, 1997; Eisenhardt & Martin, 2000; Teece, 2007).
Learning mechanisms
Zollo and Winter (2002) state that dynamic capabilities are shaped by the coevolution of learning mechanisms. According to Zollo and Winter (2002) learning mechanisms are divided into experience accumulation, knowledge articulation and knowledge codification.
Zollo and Winter (2002) state that experience accumulation is about repeated execution of tasks and learn from another. Thinking of experience accumulation Prencipe and Tell (2001) speak about learning by doing and using. Levinthal and March (1993) state that learning based on experience is the case when it is difficult to draw or ignore inferences to causality.
According to Zollo and Winter (2002) knowledge articulation is a process through which implicit knowledge is articulated on the basis of collective discussions, debriefing sessions and performance and evaluation sessions. Each individual contributes to an improved level of understanding of the causal mechanisms intervening between a certain task and a performance outcome.
A higher level of cognitive effort is required when individuals codify their understandings of the implications of performance (Zollo & Winter, 2002). One could think of internal routines in written tools, manuals, blueprints, spreadsheets, decision support systems, etc. That is to say that the process of codification will emerge with a more clear definition of what works, what does not work and why. According to Zollo and Winter (2002) codification
9
is potentially important as an underlying construct for the entire knowledge evolution process. These authors come up with the suggestion that codification facilitates the generation of new proposals in order to change the currently available routines. Cohendet and Steinmueller (2000) argue that knowledge that is codified into informational messages can be reconstituted at a later time, in a different place or by a different group of individuals with varying degrees of effectiveness depending upon their cognitive framework. Prencipe and Tell (2001) state that the outcome of the codification process are codified manuals and procedures and that the economic benefits are economics of information (diffusion, replication and reuse of information).
Competitive advantage
Considering that the dynamic capability view is an extension of the resource based view, it is valuable to have a first look at what is said about competitive advantage from a resource based view. According to Barney (1991) a resource must have four characteristics in order to be a source of a sustainable competitive advantage. A resource must be valuable, rare, inimitable, and non-substitutable. Barney (1991) further states a firm has competitive advantage when it is implementing a value creating strategy that is not implemented by a competitor now or in the future. Barney (1991) also states that competitive advantage is sustainable when this advantage resists erosion by competitor behavior.
From a dynamic capabilities perspective a competitive advantage is approached in another way. Teece, Pisano and Shuen (1997) argue that competences can provide a competitive advantage and generate rents only if these competences are difficult to imitate. The authors also use the term emulation which equals substitutability. However, the authors do not refer to valuableness or rareness. Dierickx and Cool (1989) also see sustainability of a firms asset position depending on inimitability and non-substitutability.
10
Eisenhardt and Martin (2000), however, argue that it is never explained why some firms have competitive advantage in situations of rapid and unpredictable change. Eisenhardt and Martin (2002) further argue that the functionality of dynamic capabilities could be duplicated across firms. Hence, the value of dynamic capabilities for competitive advantage is not in the capabilities themselves, rather in the configurations a firm is able to make. At this point, we find connection with Winter (2003) and Zahra, Sapienza and Davidsson (2006). These authors argue that competitive advantage does not come from dynamic capabilities themselves but comes from new configurations of resources and operational routines resulting from them.
Important for this research is to know how competitive advantage can be defined. According to Volberda, Van der Weerdt, Verwaal, Stienstra and Verdu (2012) competitive advantage is divided into measurable items related to profitability, doing well in comparison
with other organizations and jealousy of competitors about performance. According to Li,
Ragu-Nathan, Ragu-Nathan and Subba Rao (2006) competitive advantage is the extent to which a firm is able to create a defensible position over its competitors. Li, Nathan, Ragu-Nathan and Subba Rao (2006) state that the empirical literature is quite consistent in identifying the aspects of competitive advantage. On the basis of their findings they come up with the following dimensions of competitive advantage: price/cost, quality, delivery dependability, product innovation, and time to market.
Knowledge
Knowledge is defined in different ways in the literature. Nonaka (1994) sees knowledge as justified beliefs. According to Starbuck (1992) knowledge is a stock of experience. Elliot and O’Dell (1999) define knowledge as information leading to action. Knowledge is a resource that is embedded in different forms in different organizational members (Grant, 1996; Zollo & Winter, 2002). Boland jr. and Tenkasi (1995) state that the creation of knowledge within a firm asks for a process of mutual perspective where different individual knowledge is exchanged,
11
evaluated and integrated with other firm members. Goldstein and Ford (2001) argues that knowledge is an adequate understanding of facts, concepts, and their relationship, and the basis for information a person needs to perform a task.
Bartol and Svrivasta (2002) are connecting with Goldstein and Ford (2001) by stating that knowledge includes information, ideas, and expertise relevant for tasks performed by individuals, teams, work units and the organization as a whole. According to Nonaka and Takeuchi (1995) tacit knowledge is personal, context-specific and difficult to formalize and communicate. Nonaka and Takeuchi (1995) describe explicit knowledge as knowledge that is transmittable in formal, systematic language. This thesis will not make a distinction between tacit and explicit knowledge. The reason for this is that most studies on knowledge sharing do not make this distinction either (Lee, 2000; Bartol & Svrivastava, 2002; Ipe, 2003, Chow & Chan, 2008).
Knowledge sharing
Wang and Noe (2010) state that the success of knowledge management activities depends on knowledge sharing. According to Lee (2000) knowledge sharing relates to activities of transferring or disseminating knowledge from one person, group or firm to another. Lee (2000) measures both (explicit and tacit knowledge) when measuring knowledge sharing. This is also the case for Bartol and Svrivastava (2002). These authors see knowledge sharing as individuals sharing firm relevant information, ideas, and expertise with one another.
As stated by Ipe (2003) knowledge sharing relates to activities of transferring or disseminating knowledge from one person or group to another within the firm. This research will follow Bartol and Svrivastava (2002) and Ipe (2003) when speaking about knowledge sharing. These authors have agreement about the fact that knowledge sharing relates to sharing knowledge between individuals within a firm. Additional, Ipe (2003) states that knowledge
12
sharing is a process by which knowledge from an individual is converted into a form that can be understood, absorbed, and used by other individuals.
Chow and Chan (2008) in their article come up with an extension to the traditional view on sharing knowledge. Chow and Chan (2008) state that the social network, social trust and shared goals influence organizational members’ intention to share knowledge. This research will focus on how knowledge could be shared, not on the conditions for sharing knowledge.
2.2 Conceptual model
Figure 1. Conceptual model
The conceptual model rests on a number of assumptions. First, it only applies to firms active in dynamic environments. Or as said by Eisenhardt and Martin (2000), high velocity markets. High velocity markets are markets were boundaries are blurred, successful business models are not clear and were the different market players are inconclusive. In high velocity markets, change becomes nonlinear and less predictable. In these markets firms rely on new (quick) created knowledge. Learning mechanisms support quickly created new knowledge because exchange and sharing knowledge is a central theme of the learning mechanisms (Zollo & Winter, 2002). Static markets are more predictable and the evolution of such markets takes
Learning mechanisms Competitive advantage Knowledge sharing Experience accumulation Knowledge articulation Knowledge codification H1a, H1b, H1c H2a, H2b, H2c H3 H4a, H4b, H4c
13
place in a more structural way. The dynamic capabilities approach is just developed to allow a firm to integrate, built and reconfigure internal and external competences to deal with dynamic environments.
Second, it is assumed that members of a firm think about learning in the same way. However, this is a complicated point. An important reason for the failure of knowledge sharing is the lack of consideration of how organizational and personal characteristics influence knowledge sharing (Carter & Scarbrough, 2001; Voelpel, Dous, & Davenport, 2005). When not everyone assigns the same positive value to knowledge, the effect is less stronger.
Third, it is assumed that the ultimate goal for firms is to obtain and maintain a competitive advantage as defined in the literature review and that firms are highly competitive. According to Porter and Millar (1985) and Porter (1991) a competitive advantage is at the heart of a firm’s performance in competitive markets. Some firms also strive for creating social value which can be at the expense of the pursuit of profit or market share.
At last, it is assumed that firms are intensely entrepreneurial. According to Teece (2007) this type of firms have strong dynamic capabilities and are able to adapt to their environment, but are also able to shape the environment through innovation and collaboration.
2.2 Hypotheses
This section will connect the different constructs that are relevant to the research and are part of the conceptual model. The different subsections in their turn lead to the hypotheses.
Learning mechanisms and competitive advantage
It is already mentioned that dynamic capabilities could lead to a competitive advantage (Teece, Pisano & Shuen, 1997; Eisenhardt & Martin, 2000; Teece, 2007) and that dynamic capabilities are shaped by the coevolution of learning mechanisms (Zollo & Winter, 2002).
Proceeding on the previous and the arguments already put forward it is admissible that the deployment of learning mechanisms could lead to a competitive advantage. This statement
14
is built on arguments that can be found in articles related to dynamic capabilities (Teec, Pisano and Shuen, 1997; Eisenhardt & Martin, 2000; Teece, 2007; Zollo & Winter, 2002). Looking at learning as a stand-alone construct, learning itself could also be a source of a competitive advantage (Hatch & Dyer, 2004).
The article of Zollo and Winter (2002) will be the leading article for formulating the hypotheses. This means the separate relations between experience accumulation, knowledge articulation and knowledge codification and competitive advantage are tested. Drucker (1992), Lam (2000) and Von Krogh (1998) state that learning within an organization is crucial for firms in order to be successful and to obtain a competitive advantage.
It is expected that firms who deploy learning mechanisms (predictor) as defined by Zollo and Winter (2002) as (1) experience accumulation, (2) knowledge articulation and (3) knowledge codification have competitive advantage (dependent variable). This wil lead to the following hypotheses:
H1a. Experience accumulation is positively related to competitive advantage. H1b. Knowledge articulation is positively related to competitive advantage. H1c. Knowledge codification is positively related to competitive advantage. Learning mechanisms and knowledge sharing
Organizational learning could lead to knowledge (Bierly, Kessler & Christensen, 2000; Popper & Lipshitz, 2000). In this research it is expected that the learning mechanisms as divided by Zollo and Winter (2002) are positively related to knowledge sharing. According to Lee (2000) knowledge sharing relates to activities of transferring or disseminating knowledge from one person, group or firm to another. Previous statement strongly relate to the description as given by Zollo and Winter (2002) about knowledge articulation and knowledge codification. The articulation and codification of knowledge, as learning mechanisms, are strongly linked to the exchange of knowledge in different forms within a firm.
15
This research sees knowledge sharing as sharing knowledge (obtained with a learning mechanism) from individual members among other organizational members. This means that knowledge sharing, is making knowledge available to others within the firm (Ipe, 2003). Or as said by Cohendet and Steinmueller (2000), sharing knowledge is the process by which organizational members could obtain codified knowledge at a later time and in a different place.
The research assumes that learning mechanisms contribute to identifying, capturing and creating knowledge in a structural manner. This leads to a better and more effective way of knowledge sharing.
H.2a Experience accumulation is positively related to knowledge sharing. H.2b Knowledge articulation is positively related to knowledge sharing. H.2c Knowledge codification is positively related to knowledge sharing. Knowledge sharing and competitive advantage
Knowledge is an important resource that gives firms a sustainable competitive advantage in a dynamic and competitive environment (Davenport & Prusak, 1998; Foss & Pedersen, 2002; Grant, 1996). However, having knowledge alone is not enough to obtain a competitive advantage. According to Teece (2007) competitive advantage requires more than the ownership of knowledge. Knowledge must be revealed and shared to be valuable. Unleashing internal knowledge could lead to new knowledge and better performance (Wang & Noe, 2010). Wang and Noe (2010) further argue that knowledge sharing is the fundament means through which members of a firm can contribute to a competitive advantage.
According to Chow and Chan (2008) members of a firm must share their knowledge because it a necessity for having a competitive advantage. The authors argue also that sharing knowledge is difficult to ensure, because knowledge is generated and stored within the members of a firm. To encourage this sharing Chow and Chan (2008) come up with social capital factors (social network, social trust, shared goals).
16
Based on earlier research (Mesmer-Magnus & Dechurch, 2009; Collins & Smith, 2006) the conclusion can be drawn that sharing knowledge is positively related to firm performance. As an extension of the previous it could be expected that knowledge sharing has a positive influence on a competitive advantage.
H.3 Knowledge sharing is positively related to a competitive advantage.
The mediating effect of knowledge sharing on the relation between learning mechanisms and competitive advantage
In the introduction the conclusion is drawn that the step from learning mechanisms (dynamic capabilities) to a competitive advantage is a black box. This gives rise for the idea that there must be “something” mediating the relation between learning mechanisms and a competitive advantage. This research assumes that learning mechanisms lead to a competitive advantage through knowledge sharing.
First of all, it can be concluded that learning leads to knowledge (Bierly, Kessler & Christensen, 2000; Popper & Lipshitz, 2000) and that having knowledge alone is not enough (Teece, 2007; Wang & Noe, 2010). Knowledge must be released and shared in order to obtain a competitive advantage (Chow and Chan, 2008). Hatch and Dyer (2004) state that learning as a stand-alone construct could be a source for a competitive advantage. However, in order to be a source of a competitive advantage this learning must have an outcome and should give further direction too.
The research builds further on the previous authors and the work of Grant (1996). According to Grant (1996) knowledge integration is the basis for a competitive advantage in dynamic markets. Grant further argues that knowledge could be seen as the most important resource of the firm and that the essence of a firm is knowledge integration. So, the integration and management of knowledge enables companies to compete in (dynamic) markets.
17
Von Krogh (1998) argues that the process of identifying, capturing and levering the collective knowledge refers to knowledge management and enables a firm to compete. The process of identifying, capturing and levering knowledge is strongly related to the learning mechanisms as described by Zollo and Winter (2002). According to Hendriks (1999) the knowledge at the individual level could be converted into economic and competitive value for the firm as a whole. In order to make this individual knowledge valuable it must be shared.
On the basis of the work of Grant (1996) it is admissible to believe that a competitive advantage spawned by an organization capability depends, partly, upon the efficiency of knowledge sharing.
H.4a Knowledge sharing mediates the positive relationship between experience accumulation and a competitive advantage.
H.4b Knowledge sharing mediates the positive relationship between knowledge articulation and a competitive advantage.
H.4c Knowledge sharing mediates the positive relationship between knowledge codification and a competitive advantage.
18
3. DATA AND METHOD
In this chapter the empirical setup of this research is described. The first paragraph gives a description of the survey. The second paragraph describes the population and sample. The third paragraph gives insight in the construction of the survey and the collection of data. After this paragraph the measurement of the variables is being described. The fifth paragraph describes the reliability and validity. At last, we take notice of the analytical strategy.
3.1 Description of the survey
In order to show whether there is a connection between the variables a cross-sectional survey is carried out. Using a survey allow us to collect data from a large amount of people in an effective manner and to generate findings that are representative for the whole population. A drawback of a survey is that the data collected is unlikely to be as detailed when choosing another research strategy. The data obtained from a survey is also strongly influenced by the people who fill in the questions and the perceptions they have about the items. To ensure a proper interpretation, it is important that the items are formulated carefully and unambiguously. Another point to note is that the respondent cannot be influenced or be given direction during the survey.
3.2 Population and sample
The population in this research is being formed by all NVM real estate agencies active
in the housing segment in the Netherlands (total number of 2689)2. To focus this research on
the sector real estate is interesting for several reasons. The sector has taken several steps in further professionalization in recent years. Learning and knowledge is a central theme. Members of a real estate agency should follow education to be able to enroll in various registers, like the “Nederlands Register Vastgoed Taxateurs”. The NVM also carries out audits among its
19
members, whereby having knowledge about the diverse aspects of the real estate sector plays a major role.
In this sector having a competitive advantage is of great importance. The sector is highly dynamic and competition is fierce. Also the need for sharing knowledge increases. Members of real estate agencies face a variety of issues whereby knowledge sharing could lead to solutions.
We choose NVM agencies because of their professionalism and the fact that these agencies must meet high standards for a membership. The survey is carried out under 480 (N) real estate agencies equally divided over the 12 provinces (40 agencies per province). There were 480 agencies that received the survey. In total there were 99 respondents. However, five of these respondents only fill in the first five till 10 items and did not complete the survey. In order to increase the reliability these five respondents are not part of the data analysis. As a consequence, the total amount of respondents is 94. This leads to a response rate of 20%. In
order to calculate the minimal size of a representative sample the following formula3 is used:
Using the formula for the minimum recommended size for a representative sample of the research population leads to a number 336 respondents. This means that based on the number of respondents in this study it is not possible to generalize the results for the entire population. From the 94 respondents, 32 respondents are male (34%) and 62 respondents are female (66%).
3.3 Constructing the survey and data collection
The survey consists of 40 closed questions and four open questions. Three questions are used as control variables. The respondents were asked about years working experience, number of employees and the province wherein the agency is located. The fourth open question about
20
gender is used as a describing statistic to give insight in the distribution male/female. The survey can be found in appendix 1 and 2.
For the research a quantitative survey is developed on the basis of the various literature and constructs as presented in the literature review. Some items are used from existing scales, some existing items are supplemented with findings based on literature and other items are fully constructed on the base of the various literature.
For the respondents the whole survey is translated from English to Dutch. A five point Likert scale is used, (1 = strongly disagree, 2 = disagree, 3 =neutral, 4 = agree, 5 = strongly agree) since respondents are less likely to choose for extremes in such a scale. The survey is developed with Qualtrics and pré-tested by two persons.
The website of the NVM gives the possibility to obtain an overview of all registered agencies per province. When the name of a province is entered, a random overview of the names of all affiliated offices per province is offered. The first 40 agencies per province are selected. After that, all contact information must be manually searched. Data collection has taken place by sending an accompanying e-mail (Appendix 3) to the real estate agencies, connected to the NVM, wherein the aim of the research is emphasized. The survey was carried out from March
28, 2017 till May 12, 2017. A reminder was send on May 5th. Sending the reminder led to 12
additional respondents. From an ethical perspective the response is processed anonymously.
3.4 Measurement of variables
Core concepts are translated into variables within the research question. These variables are the operationalized features of the central terms and are converted into measurable terms. As discussed, this research makes use of existing constructs and of own questions based on the theoretical framework.
21
Predictor variable: learning Mechanisms
It is expected that firms who deploy learning mechanisms (predictor) as defined by Zollo and Winter (2002) as (1) experience accumulation, (2) knowledge articulation and (3) knowledge codification have a competitive advantage (dependent variable). In order to measure experience accumulation, knowledge articulation and knowledge codification a five point Likert scale is used, (1 = strongly disagree, 2 = disagree, 3 =neutral, 4 = agree, 5 = strongly agree). This Likert scale is also used for the other variables.
To test the hypotheses one and two, some items of Marsick and Watkins (2003) are used. There are also some items based on earlier items used in theories of Chang Lee, Lee and Won Kang (2004), Marsick and Watkins (2003), Prencipe and Tell (2001) and Kogut and Zander (1993). Looking at Marsick and Watkins (2003) these authors used a six point Likert scale. Chang Lee, Lee and Won Kang (2004) and Kogut and Zander (1993) used a seven point Likert Scale. Prencipe and Tell come up with a theorectical framework providing examples of the different learning mechanisms. At last, five own items are constructed based on the theories of the mentioned authors.
A complete overview of the items for experience accumulation, knowledge articulation and knowledge codification as learning mechanisms can be found in appendix 1.
Dependent variable: competitive Advantage
As argued, it is expected that firms who deploy learning mechanisms have a competitive advantage. To measure competitive advantage this research makes use of items of Jaworski and Kohli (1993) in Volberda, Van der Weerdt, Verwaal, Stienstra and Verdu (2012) and items of Li, Ragu-Nathan, Ragu-Nathan and Subba Rao (2004). Three items were self-constructed. A complete overview of the items for measuring competitive advantage measures can be found in appendix 1 and 2. The next paragraph is about knowledge sharing as a mediator. However, when testing hypothesis 2 and 3 it is also a dependent variable.
22
Mediator variable: knowledge sharing
In this thesis knowledge sharing will be seen as a mediator. This means it is investigated whether the variable knowledge sharing has a mediating effect on the relation between learning mechanisms and having a competitive advantage. It is expected that knowledge sharing mediates the relationship between the learning mechanisms: (1) experience accumulation, (2) knowledge articulation, (3) knowledge codification and the dependent variable competitive advantage.
If the mediating effect is significant, this means that the basic relationships between the independent variable (1) experience accumulation and competitive advantage, (2) knowledge codification and competitive advantage and (3) knowledge codification and competitive advantage becomes weaker through the variable knowledge sharing.
In order to measure knowledge sharing items of Gold, Maholtra and Segars (2001) and Lee (2000) are modified and used in the survey. This modification had to be done in order to bring more unity in the several items. For knowledge sharing two items are self-constructed. A complete overview of the knowledge sharing measures can be found in appendix 1 and 2.
Control Variables
The following control variables are used: years work experience of the respondent, the province in which the organization operates and the number of employees. Table 1 shows a distribution of the respondents over the provinces in the Netherlands. From this table the conclusion can be drawn that the respondents are not equally divided over the twelve province. The variable province has been converted into a dummy variable with two categories, namely Randstad (Noord Holland, Zuid Holland and Utrecht, 35.1%) and No Randstad (other Provinces, 64.9%). For the variable Provinces the encoding is No Randstad = 0 and Randstad = 1. The scores of the respondents within the Randstad are compared with the dummy variable of respondents located outside the Randstad.
23
Table 1
Provinces in which the real estate agency operates
Province Respondents 1. Groningen 4 2. Friesland 6 3. Drenthe 3 4. Noord - Holland 11 5. Flevoland 2 6. Overijssel 15 7. Gelderland 11 8. Utrecht 7 9. Zuid - Holland 15 10. Noord - Brabant 11 11. Limburg 6 12. Zeeland 3
The variables years of experience as a broker and number of employees are ratio-level variables and are included as control variables. Table 2 give an overview of the distribution of years of work experience. Most of the respondents are having between the 11 – 20 years work experience (41,5%). De next category is work experience between the 21 – 30 years (31,9%). Table 2
The distribution of years’ work experience
Category Frequency % Between 0 – 10 years 19 20.2 Between 11 – 20 years 39 41.5 Between 21 – 30 years 30 31.9 Between 31 – 40 years 5 5.38 Between 41 – 50 years 1 1.1
24
Table 3 gives an overview of the variable “number of employees”. Table three shows that 67% of the respondents works within an organization with one till five employees. Within this 67%, 19% of the respondents work within an organization with one or two employees, 16% with three employees, 16% with four employees and 16% with five employees. The largest organization a respondent is working at, has 99 employees.
Table 3
The distribution of variable number of employees
Category Frequency %
Between 1 – 5 employees 63 67.0
Between 6 – 10 employees 19 20.2 Between 11 – 15 employees 6 6.4
16 employees and > 6 6.4
3.5 Reliability and validity
Reliability checks were run for the variables experience accumulation, knowledge articulation, knowledge codification, knowledge sharing and competitive advantage. The Cronbach’s Alpha is displayed in Table 3. As exhibited in Table 4, the five variables have a Cronbach’s Alpha >.70 which indicates a high level of internal consistency and leads to a high level of validity and reliability.
Table 4
Cronbach’s Alpha
Variable Cronbach’s Alpha
Experience accumulation .73
Knowledge articulation .76
Knowledge codification .71
Knowledge sharing .71
25
The external validity is not high as the number of respondents is 94. Based on this drawn sample, no statements can be made about the entire population. With construct validity, the question is whether you measure what you intend to measure. Within this study, the construct validity has been taken into account by using existing validated scales. The existing scales are supplemented with self-constructed items based on the literature. This combination means that the construct validity is likely to be high as each variable is disassembled and supplemented with items missing in existing scales. The internal validity increases by adding the control variables. However, there is no causality when looking at the level of the models.
In addition some actions were taken in order to increase the reliability. First, an accurate survey was constructed. This survey is partly based on validated instruments and partly constructed on the bases of the various theories. Also the respondents are selected careful and divided over the twelve provinces. The survey is pré-tested by two persons and, on the basis of this test, modified. As said, sending a reminder has led to more respondents which increases the reliability.
3.6 Statistical strategy
Analysis of the data was done using the statistical package IBM SPSS Statistics (v24). Once the data was collected, it was checked and prepared for subsequent statistical tests. For the open questions years of work experience and amount of employees the respondents could formulate the answer themselves. This data was initially read by SPSS as text. The data is manually modified and converted into numeric.
When analyzing the results scores <3.0 were seen as low, scores between 3.0 – 4.0 were seen as average and scores >4.0 as high. The Cronbach's Alpha is calculated for all variables. When the Cronbach’s Alpha is >.70, this indicates a high level of internal consistency. It means the specific items can be merged to one scale. In order to do this the function “compute variable” in SPSS is used to create new variables. With all variables except the dichotomous variable No
26
Randstad/Randstad, it is checked whether the score is normally divided. Therefore we used the Kolmogorov Smirnov test. When the output is significant (p = >.05) the distribution is not normal. A non-significant output (p = <.05) tells that the distribution of the sample is not significant different from a normal distribution and so the distribution is normal.
Prior to the regression analysis the correlation between the variables (r = <.05) is checked and therefor a correlation test is executed. Within this research the following subdivision is used: A weak positive relation at r = <.40. A moderately positive relationship applies at r = >.40 - <.70 and there is a strong positive relation with r = >.70.
On the base of the Kolmogorov Smirnov test it is checked if a Spearman- of Pearson correlation test was needed. Only the dependent variable competitive advantage needs to have a normal distribution. To gain insight in the extent that the dependent variable competitive advantage is explained by the predictor, a hierarchical multiple regression analysis has been used. In block 1 the first three control variables (years of experience, number of employees and Randstad / No Randstad) are included. The independent variables experience accumulation, knowledge articulation and knowledge codification are entered in block 2. The mediator variable “knowledge sharing” within the regression analysis is entered in block 3. When testing the hypotheses with a regression analysis the p-value is based on the specific sample outcome >.10. The p-value has to be divided by two by testing the hypotheses because they are tested unilaterally.
Then a regression analysis is conducted with experience accumulation, knowledge articulation and knowledge codification (IV) predicting knowledge sharing (mediator). This test measures if the effect through the mediator significant (p = <.10) deviates from zero. If this is the case, the basic relation between the independent and the dependent variables becomes weaker through the mediating variable.
27
4. Results
In this chapter we discuss the main findings based on the analysis of the statistical tests. First, on the basis of a univariate test the probability is checked that a particular sample result is derived from a particular population. In the next paragraphs the results of the bivariate test and the regression analysis are shown. These result give insight in the connection between the variables accumulation, knowledge accumulation, knowledge codification and competitive advantage in the population based on the sample.
4.1 Univariate analysis
In order to understand the probability that the sample result is representative for the population, the formula for a sample calculation has previously been described. This paragraph will show the mean, standard deviation, the minimum and maximum scores and the normal distribution for each variable (Table 5). For each variable the minimum and maximum score per variable has been calculated. The range varies for each variable between the score one (totally disagree) till five (totally agree).
Table 5
Mean, Standard deviation, minimum and maximum for all variables
Variable Mean SD Minimum Maximum
Experience accumulation 3.83 .59 1.60 5.00 Knowledge articulation 3.60 .55 2.17 4.83 Knowledge codification 3.10 .69 1.40 4.50 Knowledge sharing 3.60 .38 1.88 4.86 Competitive advantage 3.28 .53 2.38 4.69 Number of employees 7.57 12.66 1 99
28
Looking at the mean and the standard deviation (Table 5) the following can be concluded. The respondents have the highest score on the variable experience accumulation (M= 3.83, SD= .59). There lowest scores are on the variable knowledge codification (M= 2.78, SD= .91). On the variable competitive advantage the respondents score on average high (M = 3.28, SD = .53). On the variables knowledge articulation (M= 3.60, SD = .55) and knowledge sharing (M= 3.60, SD = .38) the scores are equal.
The spread in relation to the mean for most of these items is low to average. This means the respondents have scored quite unanimous on all the items. The mean and the standard deviation for the control variables number of employees and years working experience are not measured with a five point Likert scale. These variables have a wider range. The control variable Randstad / No Randstad is not demonstrated in table 5 because this is a dummy variable.
The next step is to perform a Kolmogorov Smirnov test. This test compares the scores in the sample to a normally distributed set of scores with the same mean and standard deviation. This test is carried out for the control variables, independent variables, the mediating variable and the dependent variable (Table 6). The dichotomous variable No Randstad/Randstad has been omitted since a dummy variable is not normally distributed. If the Kolmogorov Smirnov test is non-significant (p = <.05) it tells that the distribution of the sample is not significant different from a normal distribution and so the distribution is normal. The variables experience accumulation (.24, p = >.01) and the number of employees (.31, p = >.01) are significant. This means that these variables are not normally distributed. On the basis of this we can make the choice for Pearson- or Spearman test as can be seen in table 7.
29
Table 6
Normal distribution of all variables
Variable Kolmogorov Skewness Kurtosis p
Smirnov Experience accumulation .24 -1.05 2.01 .00 Knowledge articulation .09 -.23 .20 .20 Knowledge codification .07 -.19 .26 .20 Knowledge sharing .09 .11 .20 .86 Competitive advantage .08 -.06 .87 .09 Number of employees .31 5.20 32.14 .00 Years working experience .10 .13 -.21 .09
4.2 Bivariate analysis
Before the multiple regression analysis can be performed it is important to see if the correlation between the different variables is significant. For the variables that are normally distributed a Pearson test is used. For the variables that are not normal distributed a Spearman test is used. As can be seen in Table 7 for the variables experience accumulation, number of employees and Randstad/no Randstad the Spearman Rho (s) is used. For the other variables the Pearson coefficient is used. This is mentioned in the footnote.
Beside the correlation coefficient (r) we also have to look at the p-value (<.10). The correlation matrix shows that the correlations between all variables is significant except for the correlations with the control variables No Randstad/Randstad and years working experience (p = >.10). There is a weak significant relation between the control variable number of employees and competitive advantage (r = .24, p = <.05). Table 7 shows a significant moderately positive relation between knowledge articulation (r = .58, p = <.01) and competitive advantage and
30
knowledge sharing and competitive advantage (r = .51, p = <.01). From Table 7 the conclusion can be drawn that there is a weak positive significant relation between knowledge codification and competitive advantage (r = .33, p = <.01) and between experience accumulation and competitive advantage (r = .35, p = <.01). Table 7 Correlations Variable 1 2 3 4 5 6 7 8 1. Experience accumulation (s) 1 .47*** .22** .35*** .49*** -.05 .09 -.08 2. Knowledge articulation .47*** 1 .40*** .58*** .43 .18* -.15 -.02 3. Knowledge codification .22** .40*** 1 .33*** .59*** .14 -.05 -.09 4. Competitive advantage .35*** .43*** .33*** 1 .51*** .24** -.11 .03
(s) = Spearman test, overige variabelen Pearson test *** Correlation is significant at the .01 level ** Correlation is significant at the .05 level * Correlation is significant at the .10 level
4.3 Regression analysis
Within this research, a (multiple) hierarchical regression analysis is used. This test
examines the linear relation between the independent variables experience accumulation, knowledge articulation, knowledge codification and the dependent variable competitive advantage. Within the hierarchical regression analysis one dichotomous control variable (Province and Randstad / No Randstad) and two ratio level variables (years working experience and number of employees) are entered in block 1. It has been examined if these variables have
5. Knowledge sharing .49*** .58*** .59*** .51*** 1 .12 .09 -.05
6. No. of employees (s) -.05 .18* .14 .24** .12 1 -.1 -.07
7. No. of years work exp. .09 -.15 -.05 -.11 .09 -.1 1 .12 8. No Randstad/Randstad (s) -.08 -.02 -.09 .03 -.05 -.07 .12 1
31
influence on the dependent variable competitive advantage and if there is cohesion with the independent variables.
From the ANOVA’s table (Table 8) we conclude that this first model, with the control variables, is not significant (F = 1.88, p = >.10). Previous means that the control variables years working experience, number of employees and Randstad/no Randstad do not have a significant influence on competitive advantage. The adjusted R square indicates for how much percent the model explains the variance on competitive advantage. The F-value tells us how much variability the model can explain, relative to how much it cannot explain.
Table 8
The influence of the control variables on competitive advantage (ANOVA)
Model F adj.R2 p 1. Control variables 1.88 .06 .14
2. Independent variables 7.48 .34 .00***
3. Mediator variable 7.74 .39 .00***
*** Correlation is significant at the .01 level
The independent variables experience accumulation, knowledge articulation and knowledge codification within the regression analysis are entered in block 2. With the ANOVA test it is examined whether the model is significantly good at predicting the outcome. From Table 8 it can be concluded that model two is significant (F = 7.48, R2 = .34, p = <.01). This means that the independent variables experience accumulation, knowledge articulation and knowledge codification have a significant influence on competitive advantage.
The mediator variable knowledge sharing within the regression analysis is entered in block 3. With the ANOVA tests we test whether the model is significantly good at predicting the outcome. From Table 8 we conclude that model three is significant (F = 7.74, R2 = .39, p = <.01). This means that the mediator variable “knowledge sharing” has a significant influence on competitive advantage.
32
Table 9 provides a more detailed overview of the hierarchical regression analysis for
testing the hypotheses presented per model as can be seen in Table 8. The results show a
regression between the independent variables experience accumulation (B = .10, p = <.10), knowledge articulation (B = .23, p = <.01), knowledge codification (B = .19, p = <.05), and the dependent variable competitive advantage.
The control variables are placed in block 1 of the regression analysis. Despite the fact that model 1 does not show a significant effect on competitive advantage, the control variable
number of employees however does have a significant effect on competitive advantage (B =
.01, p = .05).
The B (unstandardized) means that if the independent variable increases by one point, the dependent variable increases by the value of B. The p-value must be halved by unilateral testing, which can be seen in Table 9. This results in a significant value for the relationship between the variables experience accumulation, knowledge articulation, knowledge codification and competitive advantage. There is a significant positive relationship between knowledge sharing and the dependent variable competitive advantage (B = .22, p = <.01).
33
Table 9
Hierarchical Regression Results hypotheses 1a, 1b, 1c and 3
Testing of control and independent variables in relation to competitive advantage
Variable unst. B unst. B unst. B (model 1) (model 2) (model 3)
Constant 3.58*** 2.07*** 1.94***
Randstad/No Randstad .02 .03 .04
Number of years workexp. -.00** -.00* -.01*
Number of employees .01** .01*** .01***
Experience accumulation .10* .07
Knowledge articulation .23*** .15*
Knowledge codification .11** .04
Knowlegde sharing .22***
*** Correlation is significant at the .01 level with one tailed testing ** Correlation is significant at the .05 level with one tailed testing * Correlation is significant at the .10 level with one tailed testing
There is also a regression analysis conducted for hypothesis 2. It examined the linear relation between the independent variables experience accumulation, knowledge articulation, knowledge codification and the dependent variable knowledge sharing (Table 10).
The results show a significant linear relation between the variables knowledge articulation (B = .34, p = <.01), knowledge codification (B = .33, p = <.01) and the dependent variable knowledge sharing. Moreover, there is a weaker significant relation between experience accumulation and knowledge sharing (B = .11, p = >.10). The strongest linear relation exists between knowledge articulation and knowledge sharing.
34
Table 10
Regression Results hypotheses 2a, 2b and 2c
Testing of independent variables in relation to knowledge sharing
Variable B
Experience accumulation .11*
Knowledge articulation .34 ***
Knowledge codification .33 ***
*** Correlation is significant at the .01 level with one tailed testing * Correlation is significant at the .10 level with one tailed testing
Figure 2. Mediator
Then, the research proceeded to analyze if there is a mediating effect of knowledge sharing on the relation between learning mechanisms and competitive advantage (Figure 2). A regression analysis is conducted with learning mechanisms (IV) predicting knowledge sharing (mediator). This will give a (unstandardized regression coefficient for the relation between IV and mediator) and sa (standard error of a). The next step is to conduct a regressive analyses with learning mechanisms (IV) and knowledge sharing (mediator) predicting competitive advantage (DV). This will give b (unstandardized regression coefficient for the relation between the mediator and the DV) and sb (standard error of b). The values are presented in Table 11 and are appropriate to run a Sobel test.
35
Table 11
Regression Results for testing a mediating effect knowledge sharing
Hypothese a sa b sb H4 a .11 .07 .11 .06
H4 b .34 .09 .10 .10 H4 c .33 .06 .03 .09
The mediating effect is tested with the Sobel test (Sobel, 1982; MacKinnon, Warsi, &
Dwyer, 1995,http://quantpsy.org/sobel/sobel.htm). This test measures if the effect through the
mediator significantly (p = <.10) deviates from zero. The results from the Sobel test shows that
the variable knowledge sharing does not mediate the relationship between experience
accumulation and competitive advantage (Sobel Z = 1.23, p = >.10). Knowledge sharing also does not mediate the relationship between knowledge articulation and competitive advantage (Sobel Z = .10, p = >.10). The same conclusion can be drawn for the relation between knowledge codification and competitive advantage (Sobel Z = 0.33, p = >.10). Summarizing, there is no significant mediating effect of knowledge sharing on the relations between the learning mechanisms (experience accumulation, knowledge articulation, knowledge codification) and competitive advantage.
36
5. Discussion
This chapter consists of several paragraphs. First, the significance of the findings of this research are discussed. Each separate hypothesis and it’s result and explanations for it are discussed on the basis of the literature. Then, the research contributions, limitations and suggestions for future research are discussed.
5.1 Findings related to the literature
The purpose of this research is to see if experience accumulation, knowledge articulation and knowledge codification have a positive influence on competitive advantage and if this relationship is being mediated by knowledge sharing.
Hypotheses 1a, 1b, 1c, 2a, 2b, 2c and 3 are accepted. There is a significant positive linear relation between the independent variables experience accumulation, knowledge articulation, knowledge codification and competitive advantage. The B-value varies from .10 - .23 with a significant p-value between .00 – <.10. There is also a significant positive linear relation between the independent variables experience accumulation, knowledge articulation, knowledge codification and knowledge sharing. The B-value varies from .11 - .34 with a significant p-value between .00 – <.10. At last hypothesis 3 is accepted. This hypothesis is about the positive influence of knowledge sharing on competitive advantage.
The acceptance of hypotheses 1a, 1b and 1c can be grounded on the literature. It is clear that knowledge could lead to a competitive advantage (Argote & Ingram, 2000; Ndlela & Du Toit, 2001). Knowledge must be created and this can be done by learning (Novak, 2010). And, in general, learning is a source for a competitive advantage as stated by Moingeon and Edmondson (1996). Hypothesis 1a is accepted, but this relation is not that strong as it is for hypotheses 1b and 1c. An argument for this weaker relation can be found in Nelson and Winter (1982). According to these authors knowledge obtained by experience accumulation exhibits lacking conscious volition. At this point we could also find connection with Zollo and Winter
37
(2002) when they state that for experience accumulation it is difficult to draw or ignore inference to causality.
Looking at the hypotheses 2a, 2b and 2c, which are accepted, it is argued that the separate learning mechanisms are positively related to knowledge sharing. According to Ipe (2003) knowledge sharing is making knowledge available to others within the firm. Looking at the learning mechanisms, these mechanisms are about interaction, discussion and evaluation (Zollo & Winter, 2002). Although, hypotheses 2a is accepted, there is a weak relation. Again, we find an argument in Nelson and Winter (1982). Organizational members could be unaware that their experience contributes to knowledge sharing because they are not aware of it. It is also common understanding in the literature about experimental learning that is best conceived as a process and must not be seen in terms of outcome (Kolb & Kolb, 2005). It is not difficult to see the result of knowledge articulation and codification as outcomes.
Hypothesis 3 is also accepted. Looking at the literature, knowledge as a resource could give firms a sustainable competitive advantage in a dynamic and competitive environment (Davenport & Prusak, 1998; Foss & Pedersen, 2002; Grant, 1996). Based on the theory it can be assumed that learning leads to knowledge and that knowledge sharing leads to a competitive advantage (Mesmer-Magnus & DeChurch, 2009; Collins & Smith, 2006).
Hypotheses 4a, 4b and 4c are rejected because knowledge sharing does not mediate the relationship between the learning mechanisms: (1) experience accumulation, (2) knowledge articulation, (3) knowledge codification and competitive advantage. This means that the variable knowledge sharing has no significant mediating effect on the previous relations. The basic relation between experience accumulation and competitive advantage, knowledge articulation and competitive advantage and knowledge codification and competitive advantage
38
studies on the mediating effect of knowledge sharing on the relation between learning mechanisms and competitive advantage.
Additional to rejecting of hypotheses 4a, 4b and 4c it can be stated that Real Estate agencies are operating in increasingly dynamic markets. In this of type of markets firms rely heavily on new knowledge that could be used directly (Eisenhardt & Martin, 2000). One could argue that in this type of markets there is no time to share information. Real estate agents work independently and should make choices quickly.
Another reason for the lack of a mediating effect for sharing knowledge could be the lack of an incentive (Yang & Wu, 2008) and the lack of motivation (Quigley, Tesluk, Locke & Bartol, 2007) or employees face information sharing barriers (Singh & Kant, 2008). An example of a barrier that interferes with the dissemination of knowledge is the organization culture Al-Alawi, Al-Marzooqi & Mohammed, 2007).
Another explanation for the lack of a mediating effect could be that most of the respondents are working in a smaller organization. This means that knowledge sharing could be more difficult and mechanisms for sharing knowledge are not present (Foss, Husted & Michailova, 2010).
The absence of a mediating effect connects with Eisenhardt and Martin (2000) who state that it is not always clear why some firms have a competitive advantage in situations of rapid and unpredictable change. This in turn, connects with Zollo and Winter (2002) who argue that in high velocity markets change is unpredictable, market boundaries are vague and successful business models are unclear. This advocates that real estate agencies sometimes just have to act instead of analyzing a particular situation and that there is in particular situations no need to approach competitive advantage in a structural way based on learning and knowledge.
