Master of Science in International Business and Management
Master Thesis
The association of supply chain flexibility and supply chain agility
with performance – A meta-analysis
Author: Vanessa Sutterlüti Student Number: S3492931 Email: v.j.sutterluti@student.rug.nl
Supervisor: Dr. C. Schlägel Co-Assessor: Dr. R. De Vries Date of submission: 17th of June, 2019
Word count: 14,820 (excluding abstract, tables, figures, and references)
Faculty of Economics and Business University of Groningen
Duisenberg Building, Nettelbosje 2, 9747 AE Groningen, The Netherlands P.O. Box 800, 9700 AV Groningen, The Netherlands
ABSTRACT
In the last two decades the interest in supply chain flexibility (SCF) and supply chain agility (SCA) and their respective relationships with performance have increased as the business environment has become more globalized, volatile and fast changing. Several empirical studies on these two relationships have been conducted and have contributed to a better understanding of the concepts. However gaps in our knowledge still remain as there are inconsistencies in strengths and directions of the associations. Drawing on the resource-based view, dynamic capabilities view, relational view and contingency theory, the present study advances the understanding of these relationships. In order to do so, this paper (1) meta-analyzes 38 studies for the relationship between SCF and performance, and 33 studies for the relationship between SCA and performance, (2) divides performance into four performance dimensions and meta-analyzes the relationships between SCF/SCA and each performance dimension, (3) examines the boundary conditions influencing each of these relationships, and (4) examines the unique and common effects of SCF and SCA with performance and each performance dimension. The results of this meta-analysis show a positive association of SCF/SCA with performance and with the performance dimensions. Additionally, several boundary conditions influencing these relationships are revealed. In this way this study contributes to theory and practice and reveals future research directions.
TABLE OF CONTENTS
1. INTRODUCTION ... 1
2. LITERATURE REVIEW AND THEORETICAL BACKGROUND ... 4
2.1. Literature review ... 4
2.2. Theoretical background ... 12
3. METHODOLOGY ... 22
3.1. Literature search ... 22
3.2. Sample and coding ... 23
3.3. Meta-analytic procedures ... 30
4. RESULTS ... 31
4.1. Results of the meta-analysis SCF-performance ... 31
4.2. Results of the moderator analysis SCF-performance ... 32
4.3. Results of the meta-analysis SCA-performance ... 40
4.4. Results of the moderator analysis SCA-performance ... 41
4.5. Results of the commonality analysis ... 47
5. DISCUSSION ... 49
5.1. Theoretical contributions and implications ... 49
5.2. Practical implications ... 52
5.3. Limitations and future research ... 54
6. CONCLUSION ... 56
REFERENCES ... i
i
LIST OF FIGURES
FIGURE 1 - Unique and common effects of SCF/SCA in the explained variance of
performance (own composition) ... 3
FIGURE 2 - Conceptual model (own composition) ... 13
FIGURE 3 - Overview of results of statistically significant moderators – SCF-performance 35 FIGURE 4 - Overview of results of statistically significant moderators – SCA-performance 43
LIST OF TABLES
TABLE 1 - Definition of operational flexibilities ... 6TABLE 2 - Dimensions of SCF on the network level ... 7
TABLE 3 - Overview of SCF dimensions used in the literature... 8
TABLE 4 - Capabilities and characteristics of SCA ... 10
TABLE 5 - Overview of SCA dimensions used in the literature ... 11
TABLE 6 - Overview of complete sample ... 23
TABLE 7 - Studies included in the analysis of SCF on performance ... 26
TABLE 8 - Studies included in the analysis of SCA on performance ... 28
TABLE 9 - Results of bivariate meta-analysis - SCF on performance ... 36
TABLE 10 - Results of moderator analysis - SCF on performance ... 37
TABLE 11 - Results of bivariate meta-analysis - SCA on performance ... 44
TABLE 12 - Results of moderator analysis - SCA on performance ... 44
TABLE 13a – Overview of results of commonality analysis - FP ... 47
TABLE 13b - Results of commonality analysis - FP ... 47
TABLE 14a – Overview of results of commonality analysis - OPP/MP ... 48
TABLE 14b - Results of commonality analysis - OPP/MP ... 48
TABLE 15a – Overview of results commonality analysis - OP ... 48
TABLE 15b - Results commonality analysis - OP ... 48
TABLE A1 - Journals used in this meta-analysis ... xiii
TABLE A2 - Overview of some items used by empirical research to measure SCF/SCA .... xiii
LIST OF ABBREVIATIONS
CT Contingency theory DCV Dynamic capabilities view FP Financial performance MP Market performance OP Overall performance OPP Operational performance RBV Resource-based view ROI Return on investment RV Relational view SBU Strategic business unit SC Supply chain
SCA Supply chain agility SCF Supply chain flexibility
1.
INTRODUCTION
In the last decades, an increasingly volatile and uncertain business environment has led researchers to investigate the ways of how companies can adapt and react to these uncertainties and unpredictable circumstances in an effective and efficient way (Christopher, 2000; McCann, 2004). Reasons for these changes are, among others, globalization, decreasing product life cycles, innovation, uncertain supplies, increasing competition and growing customer expectations (Gligor & Holcomb, 2012a; Sharma, Sahay, Shankar, & Sarma, 2017). Agility and flexibility were first recognized to be important for competitiveness in manufacturing in the 1990s (Bernandes & Hanna, 2009). The concepts then disseminated to organizational agility and flexibility, from which supply chain agility (SCA) and supply chain flexibility (SCF) were derived as reactions to the uncertainties within markets (Dove, 1996; Fawcett, Calantone, & Smith, 1996; Fayezi, Zutshi, & O’Loughlin, 2016).
However, there is no consensus on the definition of agility and flexibility. According to Fayezi et al. (2016), SCF relates to the operational ability of an organization to adjust internally and/or externally within the supply chain to changes in the market and increasing uncertainties. SCA on the other hand relates to the strategic ability of an organization to quickly respond to changes and uncertainties in the market via supply chain relationship integration. Wadhwa and Rao (2003) claim that some researchers consider agility as a phenomenon, which is constructed by a number of elements, one of which is flexibility. In the last two decades a lot of research has been conducted on antecedents and enablers of SCF and SCA (e.g. Swafford, Ghosh, & Murthy, 2006; Gligor & Holcomb, 2012a). Additionally, the terms agility, flexibility and responsiveness have been used interchangeably which has led to confusion (Bernandes & Hanna, 2009; Fayezi et al., 2016).
have been heavily researched. In empirical literature, performance is divided into different dimensions: financial, operational, market, and overall (Sharma et al., 2017). Sharma et al. (2017) provide an overview of studies investigating SCA, as well as of studies which looked at the impact of SCA on performance.
Looking at the empirical studies on the relationships of SCF, and SCA with performance, it can be seen that the magnitude of the results is not consistent. While some found a weak association between SCF and performance (e.g. Malhotra & Mackelprang, 2012, or Qrunfleh, 2010), others found a moderate association (e.g. Lio et al., 2010;Zhang, Vorderembse, & Lim, 2005; Vickery, Calantone, & Dröge, 1999), while other authors found a strong association (e.g. Amoako- Gyampah, Gyasi Boakye, Adaku, & Famiyeh, 2019; Ku, Wu, & Chen, 2016; or Kumar, Verma, Sharma, & Khan, 2017). The same can be observed for the relationship of SCA and performance. Some studies found a weak association between SCA and performance (e.g. Alfalla-Luque, Machuca, & Marin-Garcia, 2018; Qi et al., 2009), others found a moderate association (e.g. Gosh & Tan, 2005; Swafford, Gosh, Murthy, 2008; or Hwang & Kim, 2018), while other authors found a strong association (e.g. Liu et al., 2018; Chan et al. 2017; or Gligor & Holcomb, 2012a). Additionally, some studies even found a negative relationship between SCF and performance, and SCA and performance (e.g. Wagner, Grosse-Ruyken, & Erhun, 2018). Thus, it is important to research where the differences in magnitude and direction come from.
In order to measure the effect size of these relationships and reduce the limitations of the results of a single study, a meta-analysis provides researchers and managers with valuable information. Therefore, in this study a bi-variate meta-analysis with the following overall research question will be conducted:
RQ1: What are the average effect sizes for the relations between SCF and performance, and SCA and performance?
In order to gain a deeper understanding of the boundary conditions of these relationships, and specifically, by which moderators the effect sizes of these relationships are influenced, methodological and theoretical moderators will be included in the analysis. Therefore, research question 2 is formulated:
RQ2: Which methodological and theoretical moderators enhance or mitigate these relationships?
Some authors highlighted SCF as an antecedent of SCA and empirically investigated mediating effects of SCA in the SCF–performance relationship (e.g. Rouis, 2010; Swafford et al., 2008). However, to the best of my knowledge, no research has been conducted on the unique and common effects of SCF and SCA in the explained variance of performance and its dimension (see Figure 1). Hence, research question 3 is formulated:
RQ3: What is the respective unique and common effect of SCF and SCA on performance?
It is important to gain a better understanding of these relationships as it impacts theory and practice. Theoretically, this study reviews the current state of the literature and fills the gap of a meta-analysis on this topic. It measures the unique and combined effect sizes of the relationship of SCF on performance, and SCA on performance. By investigating methodological and theoretical moderating effects, it helps to understand why there are differences in the effect sizes between single studies. Additionally, it provides directions for what topics need further investigation in the future.
For practitioners and managers, it is important to know how they can improve the company’s performance. Conducting a meta-analysis helps to understand why and under which conditions investments in SCF and SCA have a positive association with performance and under which conditions they don’t. This is especially important in fast-changing and uncertain market environments. Depending on the results of this meta-analysis managers can take decisions regarding these investments based on a deeper understanding of the circumstances and potential outcomes.
2.
LITERATURE REVIEW AND THEORETICAL
BACKGROUND
This section provides an overview of the existing literature on SCF and SCA and their respective influence on performance. In order to do so, first a literature review of SCF and SCA is provided. Second, the SCF-performance and SCA-performance relationships are explained and hypotheses developed. Third, the potential methodological and theoretical moderators are explained.
2.1. Literature review
Supply chain flexibility
Evolution and definition. Flexibility can be defined as “an organization’s ability to
Vokurka and O’Leary-Kelly (2000) found in their investigation of empirical research on manufacturing flexibility the following dimensions: Machine, material handling, operations, automation, labor, process, routing, product, new design, delivery, volume, expansion, program, production and market. They suggest that manufacturing flexibility can lead to a competitive advantage (Vokurka & O’Leary-Kelly, 2000). However, due to increasing uncertainty and changes in demands and the environment, as well as the internationalization of markets, it is not enough to only focus on internal manufacturing flexibility, but the whole supply chain has to be taken into account (Duclos, Vokurka, & Lummus, 2003; Prater, Biehl, & Smith 2001; Tiwari, et al., 2013). Lummus, Duclos, and Vokurka (2003) highlight that in the context of supply chains flexibility refers to all actors of the supply chain, including the internal departments of an organization, like manufacturing, as well as flexibility between all externally involved partners like suppliers or information system providers, etc. The literature on SCF emerged in the late 1990s and takes on a customer-oriented perspective (Vickery et al., 1999; Abdelilah, El Korchi, & Balambo, 2018).
In the literature, there is no consensus on how to define SCF (Tiwari et al., 2013). Vickery et al. (1999) were one of the first authors defining SCF. In their point of view it is composed by flexibilities which are directly influencing the firm’s customers and satisfying their needs. It is the responsibility not only of the firm internally, but also of external supply chain members. Kumar, Fantazy, Kumar and Boyle (2006), for example, highlight that SCF enables firms and their supply chain partners to adjust their operations and strategies according to the environment and customers’ demands. They also highlight the shared responsibility of the firm and its SC partners in producing products in line with customers’ expectations, while achieving high performance. Gosain, Malhotra and El Sawy (2005) highlight that SCF is derived from the internal company’s flexibility as well as from the external flexibility of these supply chain linkages (Gosain et al., 2005.). Tiwari et al. (2013) conducted a literature review on SCF and introduced the following definition:
“A SC is said to be flexible if it can ensure smooth undisrupted supply of the products from supplier to the end user under all risks and uncertainties in the environments, with the least variation in the difference between the demand and supply at every demand-supply node, and without much penalty or impact on the SC resources and the costs incurred.” (Tiwari et al., 2013: 771)
Dimensions of SCF. SCF has many different components at different levels. As SCF is
flexibility components that are internal to a company as well as flexibility components which relate to external partners of the supply chain (Sánchez & Pérez, 2005; Abdelilah et al. 2018). Stevenson and Spring (2007) highlight that SCF consists of flexibility dimensions on the resource and shop-floor level, the plant-level, the firm-level as well as the whole supply chain network level.
Operational flexibilities: These flexibilities can be achieved on the level of resources and the
shop-floor and they relate to machines, material handling, operations, automation, labor, process, and routing (see Table 1). These flexibilities refer to the manufacturing process and are internal to the firm (Stevenson & Spring, 2007; Vokurka & O’Leary-Kelly, 2000).
TABLE 1 - Definition of operational flexibilities
Dimension Definition
Machine Ability of the equipment to complete a wide range of operations without changes in the configuration
Material handling Capability of the manufacturing system to move parts and components via an effective material handling process.
Operations Number of alternative ways in which a product or part can be manufactured.
Automation Degree to which the automation of the manufacturing processes is responsible for flexibility
Labor Ability of a worker to perform a wide range of tasks across the manufacturing system Process Ability of the manufacturing system and processes to produce a large number of
different parts and components without changing the configuration.
Routing Ability of the system to provide multiple alternative production paths for parts and components
Source: Vokurka & O'Leary-Kelly, 2000
Tactical flexibilities: Tactical flexibilities are flexibility types on the plant-level. They include
product and modification, volume, delivery, and production flexibility (Stevenson & Spring, 2007). According to Vickery et al. (1999), product and modification flexibility refers to the customization of the product and the ability of the firm to deal with difficult orders, which do not conform to the standard. Volume flexibility is explained by Vickery et al. (1999) to be the ability of the firm to increase or reduce production in order to meet customer demands. Delivery flexibility refers to the ability of the firm to change lead times according to customer demands (Sánchez & Pérez, 2005). Production flexibility captures the scope of products the plant can make without changing the manufacturing equipment (Vokurka & O’Leary-Kelly, 2000).
Functional flexibilities: These flexibilities include flexibility types related to the different
manufacturing flexibility includes the flexibility types of the lower levels. Hence, manufacturing flexibility includes tactical and operational flexibilities.
Strategic flexibilities: Strategic flexibilities are those on the firm-level. They include new
design, expansion and market flexibility (Stevenson & Spring, 2007). According to Stevenson and Springs (2007), new design flexibility refers to the design and introduction of new products and how fast and cost effective a firm can perform this task. Expansion flexibility is explained by the “ease with which the firm can add long-term capacity to the system” (Vokurka & O’Leary-Kelly, 2000: 486). Market flexibility refers to the internal ability to deal with changes and uncertainties in the market environment (Stevenson & Spring, 2007).
Network / supply chain flexibilities: According to Stevenson and Spring (2007), supply chain
flexibilities are those types of flexibility that go beyond the firm level and include robustness, re-configuration, relationship, logistics, and inter-organizational information systems. According to Lummus, Duclos, and Vokurka (2003), supply chain flexibilities additionally include operation systems, supply networks, and organizational design. Sánchez and Pérez (2005), as well as Swafford et al. (2006) added procurement/sourcing flexibility to the list of components of SCF. See Table 2 for definitions and Table 3 for an overview of SCF dimensions used in the literature.
TABLE 2 - Dimensions of SCF on the network level
Dimension of network flexibility
Definition
Robustness Ability of the supply chain (SC) configuration to cope with a wide range of market changes
Re-configuration Ability to adjust and reconfigure the SC to market changes
Relationship Ability to establish collaborative, upstream and downstream relationships with SC partners
Logistics Ability to rapidly and cost effectively adjust logistics, like transportation carriers and warehouse space, to sources of supply and customer needs
Organizational Ability to adjust available skills to satisfy present needs in the SC Inter-organizational
information systems
Ability to align and integrate information systems with SC participants in response to changes in information needs
Sourcing flexibility / supply network
Ability source raw materials, parts and components from different suppliers if necessary
Operations systems Ability to reconfigure assets, change processes and adjust capability in a dynamic way.
Launch flexibility Ability to quickly introduce new products and product varieties by integrating activities across the SC
TABLE 3 - Overview of SCF dimensions used in the literature Dimensions/components Vickery et al. 1999 Swafford 2006 Duclos et al., 2003 Zhang et al. 2005
Sánchez & Perez 2005 Kumar et al. 2006 Stevensen & Spring, 2007 Product flex /
Product development flex / product mix flex
x x x x x x Volume flex x x x
Launch flex / new product flex x x x x
logistics flex / access / distribution / delivery x x x x x x x
Responsiveness flex x x x x
Sourcing flex / supply (network) flex x x x x x x
Information systems flex x x x
Manufacturing flex x x
Market flex x x
Organizational flex x x
Operations systems flex
Supply chain agility
Evolution and definition. SCA has its origins in agile manufacturing, which emerged in the
1990s (Katayama & Bennet, 1999; Fayezi, Zutshi, & O’Loughlin, 2015). Agility in manufacturing was introduced in a report by the Iaccoca Institute at the Lehigh University in 1991, and was considered to improve firm performance (Yusuf, Saradi, & Gunasekaran, 1999). As agility was found to have positive implications on firm performance, it spread into other areas of the organization like supply chain management. Dove (1996) first introduced the concept of agile supply chain management, which has become a crucial topic in supply chain management.
There is also no consensus on a common definition of SCA, which has resulted in the concept of agility in supply chain management remaining ambiguous (Gligor & Holcomb, 2012b). Several authors consider SCA as way of adapting the supply chain to the needs of the market which are constantly changing (Sharp, Irani, Desai, 1999; Christopher, 2000; Jain, Benyoucef, Deshmukh, 2008, Swafford et al., 2008). Li, Chung, Goldsby and Holsapple (2008) highlight the importance of being alert to internal and external changes, both opportunities and challenges, as well as being capable of using resources to rapidly and flexibly respond to these changes. Sharma et al. (2017) conducted a review on SCA to provide conceptual clarity. Based on their review, they introduced the following definition for SCA:
“SCA is the strategic capability of a supply chain to quickly sense and respond to internal and external changes, either proactively or reactively, leveraging intra- and inter-organizational capabilities in an effective manner that ensures profitability.“ (Sharma et
al., 2017: 543)
Enablers and dimensions of SCA. Agility in the supply chain has different enablers and
dimensions, which will be discussed in this section. Jain et al. (2008) developed a framework of SCA including enablers, capabilities and characteristics based on the findings of several other authors.
developing products together and sharing information. Furthermore, Lin et al. (2006) add process and information integration to the enablers of SCA. Process integration means that the actors of the supply chain constitute a confederated network. Information integration relates to the use of information technology to easily exchange information within the supply chain. Lin et al. (2006) also included customer and marketing sensitivity to the enablers of SCA. It encompasses the pillar of mastering change and uncertainty by Goldman et al. (1995) and it relates to the ability to sense customer demands and needs. Gligor and Holcomb (2012a) base their study on the relational view by Singh and Dyer (1998) and highlight that coordination, cooperation, and communication are important elements to attain agility in the supply chain. The goal of SCA is customer enrichment and satisfaction while dealing with uncertainties in the market and environment. In order to fulfill these goals and be agile, a supply chain needs to have certain capabilities and characteristics (Lin et al., 2006). These capabilities and characteristics are presented in Table 4.
TABLE 4 - Capabilities and characteristics of SCA
Dimension Definition
Responsiveness Ability to detect changes, and respond to them in a fast, reactive or proactive way, and to recuperate from them.
Competency Ability to achieve company goals in an efficient and effective way
Flexibility/adaptability Ability to implement and configure different processes and facilities to attain the same objectives
Quickness/ speed Ability to execute a task as fast as possible Source: Sharp et al., 1999; Christopher, 2000; Lin et al., 2006; Jain et al. 2008
TABLE 5 - Overview of SCA dimensions used in the literature Dimensions/ components Christopher (2000) Swafford et al. (2006) Lin, et al (2006) Swafford et al. (2008) Bernandes & Hanna (2009) Gligor & Holcomb (2012a) Eckstein et al. (2015) Li et al (2015) Tse et al. (2016) Fayezi et al. (2016) Strategic x Relational x Ability/capability x x x x x x x x x Short Term x x Internal Changes x x x x External Changes x x x x x x x Proactive x Reactive x Sense x x Alert x Adapt x x Respond x x x x x x x Reconfigure x Quick x x x x x x x x x Effective x Flexible x x x Integrated x
Relationship and differentiation of the concepts
The term agility in supply chain has frequently been used interchangeably with the terms flexibility and responsiveness (Bernandes & Hanna, 2009). However, there is a notable difference between the concepts. As described above, flexibility is often considered an important element of agility (Fayezi et al. 2016; Sharma et al., 2017). Other researchers consider “agility as an extension of flexibility” (Fayezi et al., 2016: 1). Again, other researchers consider flexibility as an enabler of agility (e.g. Swafford et al. 2006). Swafford et al. (2006) explain flexibility and agility according to the relationship between competency and capability. In their understanding, flexibility constitutes a competency while agility constitutes a capability (Swafford et al., 2006; Fayezi et al., 2015). Goldsby, Griffis and Roath (2006), for example, state that the key for agility is flexibility in the whole supply chain. In their literature review, Fayezi et al. (2016) describe SCA as a strategic ability while SCF is considered to be an operational ability, both dealing with uncertainties in the market or internally in the organization.
2.2. Theoretical background
SCF and SCA are considered to be positively associated with firm performance. This section first looks at the relationship of SCF and performance, followed by the relationship of SCA and performance. In the last part of this section, the methodological and theoretical moderators used for this meta-analysis are presented. The conceptual model can be seen in Figure 1.
The relationship of SCF and performance
times, adjusting production capacity as well as the product mix while meeting customer demands. Therefore, it enables the firm to deal with uncertainties and changes in demand and the environment, and improves its competitive business performance (Swafford et al., 2008).
Different empirical studies used different theories to explain the relationship between SCF and performance. They mainly used the RBV, DCV, as well as CT. According to the RBV, a sustained competitive advantage of a firm depends on the inherent resources the firm owns and controls. A firm is able to obtain a competitive advantage if its resources can be considered to be valuable, rare, inimitable, and non-substitutable (VRIN) (Barney, 1991). Hence, a company has higher chances to improve its competitive position and performance when it is able to employ its VRIN resources in an effective way (Chan et al. 2017). Flexibility in the supply chain is considered a VRIN resource (Chan et al. 2017; Ying, 2010).
The DCV is an extension of the RBV, as dynamic capabilities are considered to be valuable, rare, in-imitable, and non-substitutable and therefore constitute a source of competitive advantage of the firm with the ultimate goal of improving performance (Teece et al., 1997). Dynamic capabilities are defined as “the firm’s ability to integrate, build, and reconfigure
Supply chain flexibility Performance Supply chain agility Methodological moderators - Publication status (H3) - Eigenfactor (H4) - Unit of analysis (H5) H1+ H2+ Theoretical moderators
- State of the economy (H6)
- Region (H7) - Uncertainty avoidance (H8) - Long-term orientation (H9) - Firm size (H10) - Industry (H11) - Type of firm (H12)
- Year of data collection (H13)
internal and external competences to address rapidly changing environments” (Teece et al., 1997: 516). This is in accordance with the definition of SCF. Thus, SCF is considered a dynamic capability.
According to Donaldson (2001), the CT takes the environment in which the company operates, the organizational size, as well as the organizational strategy as contingencies into account. The fit and interaction between the firm and these contingencies influences the company’s performance outcomes (Donaldson, 2001). Especially in markets with rapidly changing demands and with high uncertainties, which indicates a change in contingencies, it is important for the firm to align its structure and characteristics with these changing contingencies. Flexibility in the supply chain is a way to align the company with the environment which is why it is considered to be positively associated with performance (Merschmann & Thonemann, 2011).
The empirical studies on the SCF-performance relationship differ in SCF dimensions as well as performance dimensions under investigation. Performance measures which have been taken into consideration by different authors within this relationship are financial performance (FP) measures (e.g. Vickery et al, 1999; Fantazy, Kumar & Kumar, 2009; Sánchez & Pérez, 2005), operational performance (OPP) measures (e.g. Malhotra & Mackelprang, 2012; Merschmann, Thonemann, 2007; Um, Lyons, Lam, Cheng, Dominguez-Pery; 2017; ), market performance (MP) measures (e.g. Vickery et al., 1999; Sánchez & Pérez, 2005), as well as overall performance (OP) measures (e.g. Chang, Chen, Lin, Tien, & Sheu, 2006; Swafford et al., 2008; Qi, Huo, Wang, & Yeung, 2017). Even within these dimensions, the effect sizes vary. Based on these arguments Hypothesis 1 is formulated:
Hypothesis 1: SCF is positively associated with performance.
The relationship between SCA and performance
as it enables the firm to deal with fast changing environments in which the firm operates (Gligor & Holcomb, 2012a; Chiang, Kocabasoglu-Hillmer, Suresh, 2012).
Other authors use the RV to describe the positive association of SCA with performance. The RV states that “idiosyncratic interfirm linkages” as well as “knowledge-sharing routines” between firms may be a source of competitive advantage (Singh & Dyer, 1998: 660). Hence, the RV highlights the importance of a firm’s external inter-firm relationships to increase competitiveness. As collaborative relationships as well as process and information integration within the whole supply chain are important antecedents of SCA, SCA can be considered a source of competitive advantage based on the RV (Gligor & Holcomb, 2012a; Hwang & Kim, 2018).
In the existing empirical literature on the SCA-performance relationship several SCA dimensions as well as performance measures are investigated. Performance measures which have been taken into consideration by different authors within this relationship are FP measures (e.g. McCann, 2009; Vickery, Dröge, Setia, & Sambamurthy, 2010; Li, Wu, Holsapple, & Goldsby, 2017), OPP measures (e.g. Saeed, Malhotra, & Abdinnour, 2019; Eckstein, Goellner, Blome, & Henke, 2015), MP measures (Yusuf, Gunasekaran, Musa, Dauda, El-Berishy, Cang, 2014), as well as OP measures (e.g. Blome, Schoenherr, & Rexhausen, 2013 Liu, Ke, Wei, & Hua 2013; Hwang & Kim, 2018). ). Even within these dimensions the effect sizes vary. Based on these arguments Hypothesis 2 is developed:
Hypothesis 2: SCA is positively associated with performance.
Methodological and theoretical moderators
Methodological moderators
Publication status. The first methodological moderator taken into consideration is publication
status. In this way it is examined if the study was published or unpublished and if this has an impact on the effect size (Schmidt & Hunter, 2015). Due to the argumentation above, it is expected that flexibility and agility in the supply chain are positively related to performance. In order to provide managers with a useful solution for uncertainties in the environment, strongly positive, significant relationships are more likely to be published in scientific journal than negative relationships. Journals might not be willing to publish negative or insignificant relationships. If it is observable that there is a significant difference in the effects of published and unpublished studies, it would result in a publication bias. In case of publication bias, the published literature is not representative of the whole population of studies conducted on the relationship, meaning that statements and conclusions which are only based on the published literature might not be correct and generalizable (Rothstein, Sutton, & Borenstein, 2005). Therefore, I expect the relationship between SCF/SCA and performance to be stronger in published studies than in unpublished studies. Thus, I formulate Hypothesis 3:
Hypothesis 3: Studies which are published in scientific journals will report stronger positive relationships between SCF/SCA and performance, than studies which are not published in scientific journals.
Eigenfactor. The next methodological moderator which I am taking into account is the
Eigenfactor of the journal in which the study was published. Eigenfactor is an indicator for the quality and importance of a journal and is based on the number of citations of the studies published in these journals. It is expected that Eigenfactor has an positive impact on the SCF/SCA-performance relationship. Therefore, Hypothesis 4 is formulated:
H4: Studies published in journals with a high Eigenfactor score will report stronger SCF/SCA-performance relationships than studies published in journals with a lower Eigenfactor score.
Unit of analysis. The next methodological moderator included in this study is the unit of
involved in the provision of the product. With this background, SCF and SCA become more important on the firm-level unit of analysis because more actors are involved, resulting in flexibility and agility within supply chain partners gain in importance. In accordance with Jain et al. (2008), Stevenson and Spring (2007), and Tiwari et al. (2013), I argue that on the unit of analysis of the SBU, plant or lower, manufacturing flexibility, organizational flexibility, or other operational and tactical flexibilities (as explained in section 2.1.) have a stronger association with performance than SCF or SCA. Therefore, I suggest that the relationship between SCF/SCA and performance is stronger for studies on the firm level of analysis than on the plant or SBU level. Hypothesis 5 is formulated:
Hypothesis 5: Studies using samples captured on the firm level unit of analysis will report stronger SCF/SCA-performance relationships than studies which are using samples captured on lower level unit of analyses.
Theoretical moderators
State of economy. This moderator takes into consideration in which country the study was
conducted and if this country is considered an emerging or advanced market. Emerging markets are characterized by underdeveloped, yet rapidly growing economies, low incomes and they usually have undergone major reforms or liberalizations in the market (Kumar et al., 2017). Rapid growth leads to changes in demands and consumer preferences, which results in increased market volatility. Additionally, emerging markets show higher levels of financial, political and social risks (Marquis & Raynard, 2015). Hence, companies in these markets face high uncertainties for which they need to find solutions. As explained above SCF and SCA are considered to be solutions for uncertainties in the environment and markets. Advanced markets are characterized by a stable economy, lower risks but also a lower growth rate (Marquis & Raynard, 2015). As the environment in advanced markets is more stable and poses lower risks and uncertainties, SCF and SCA are expected to have a weaker association with performance. Therefore, I suggest that the association of SCF and SCA with performance is stronger in emerging markets than in advanced markets. Hypothesis 6 is formulated:
Region. In addition, it will be investigated if the region influences the relationship under
investigation. Empirical research has been conducted in Africa, Asia, the Middle East, North America and Europe. It is interesting to see whether studies conducted in different regions of the world show different results. The previous moderator (state of the economy) does not take the different regions of the world into account and we cannot draw conclusions about the impact of the location of the country from this moderator. The regions mentioned differ according to cultural values, economic development, political and institutional frameworks as well as social and topographical aspects (UNSD, 2019). Many Asian or Middle Eastern countries are considered emerging economies, while North American or European countries are considered to be advanced economies. It is interesting to investigate the differences in effect sizes of the relationship between regions. I suggest that companies located in regions with more stable governance and political systems, as well as social and economic structures show a weaker relationship of agility and flexibility in the supply chain with performance. Therefore, Hypothesis 7 is formulated:
Hypothesis 7: Studies using samples captured in North America or Europe will report lower SCF/SCA-performance relationships than studies using samples captured in Asia or Africa.
Culture – Uncertainty avoidance (UA). In order to understand the impact of culture on the
societies with low levels of UA. Because of this rationale, I argue that companies implementing flexibility and/or agility in their supply chain have higher performance outcomes in societies with low levels of uncertainty avoidance, compared to societies with high levels of uncertainty avoidance. Thus, I formulate Hypothesis 8:
Hypothesis 8: UA moderates the SCF/SCA-performance relationships in such a way that for studies captured in countries with lower UA scores stronger SCF/SCA– performance relationships will be reported and for studies captured in countries with higher UA scores weaker SCF/SCA-performance relationships will be reported.
Culture – Long-term orientation (LTO). This dimension relates to how people of a society
link their past with the present and future. Short-term oriented societies focus on those events taking place in the present or near future, while long-term oriented societies focus on those events taking place in the future. Additionally, short-term oriented cultures maintain traditions and are reluctant to change. They have a closer linkage to their past, which is why they are often called normative societies in the business context. Contrastingly, long-term oriented societies are more pragmatic. They are willing to invest in education and accept change in order to be equipped for the future (Hofstede Insights, 2019). Therefore, I argue that societies with a long-term orientation are more willing to invest in SCF and SCA in order to be able to deal with changes and uncertainties in the future. Short-term oriented societies might not be willing to change their processes. As they are not focused on the future, they are not looking for ways to deal with potential changes or shifts in the future. Thus, I suggest that in societies which score high on the LTO dimension SCF and SCA have a stronger positive association with performance than in societies with low scores on this dimension. Hence, I formulate Hypothesis 9.
Hypothesis 9: LTO moderates the SCF/SCA-performance relationships in such a way that studies using samples captured in countries which score high on the LTO dimension will report stronger SCF/SCA-performance relationships than studies which are using samples captured in countries with low scores on the LTO dimension. Firm size. Additionally, firm size will be considered a theoretical moderator influencing the
be more bureaucratic and it is harder for them to change rapidly (Donaldson, 2001). Smaller companies have restricted resources and are more dependent on their supply chain relationships, which enable the firms to align its resources with those of its suppliers according to customer needs (Tokman, 2013). At the same time, smaller firms have less bureaucratic layers and are therefore able to respond quicker to the market and customer needs (Donaldson, 2001). Ko, Liu, Ngugi and Chapleo (2018) also include the age of a company as influencing performance. Older firms are usually bigger and are less innovative. Younger firms are usually smaller and have higher incentives to increase performance and customer satisfaction in order to gain a sustained competitive advantage (Ko et al., 2018). Hence, it is expected that the relationship between SCF/SCA and performance is stronger in smaller firms than in larger firms. Therefore, Hypothesis 10 is formulated:
Hypothesis 10: Studies using samples from smaller firms will report stronger SCF/SCA-performance relationships than studies using samples from larger firms. Industry. The next potential moderator which can influence the relationships between
SCF/SCA and performance is the variety of industries under investigation. It is argued that studies which focus on a single industry will have a greater SCF/SCA-performance relationship than studies which take multiple industries into account. Researchers take multiple industries into consideration in order to increase generalizability of their results (Wowak, Craighed, Ketchen Jr, & Hult 2013). In this way, however, they do not entirely capture the relation of SCF/SCA with performance as it may vary across different industries. More specifically, strong SCF/SCA-performance relationships in a certain industry might be compensated by weak SCF/SCA-performance relationships in another industry. Hence, by conducting studies captured in multiple industries, researchers might not be able to entirely capture the specificities and contextual conditions of single industries. Studies only focusing on a single industry in turn may be able to capture the true association of SCF/SCA and performance because the study can focus on the specific aspects of the industry (Wowak et al. 2013). Therefore, I formulate Hypothesis 11:
Hypothesis 11: Studies which are using samples captured from a single industry will report stronger SCF/SCA-performance relationships than studies which are using samples captured from multiple industries.
Type of firm. The next potential moderator included in this study, which is argued to have an
type, such as service or manufacturing firms, different performance outcomes can be expected (Jack & Raturi, 2002). Companies involved in manufacturing heavily depend on SCF and SCA in order to meet customer needs and improve performance. Historically, flexibility and agility in the SC arose from manufacturing and are directly linked to it (see section 1 and 2). By definition, SCF and SCA include dimensions of manufacturing. Hence, the relationship between SCF and SCA and performance is expected to be high in manufacturing companies. The relationships of SCF/SCA and performance for the service industry have received less attention (Ivens, 2005). However, Rouis (2010) and Ivens (2005) argue that SCF has a positive impact on customer satisfaction in service environments. Due to the characteristics of services, flexibility in the provision of services is important to respond to customer needs (Ivens, 2005; Rouis, 2010). These characteristics of services include intangibility, simultaneous production and consumption, as well as the integration of the customer into the production process (Grönroos, 2000). Rouis (2010) also shows that SCA is considered not to be as important to customers as the quality of the provided service in the sense of meeting their needs. Wowak et al. (2013) highlight that supply chain knowledge and knowledge exchange is especially important to generate performance in the service industry. Based on this argumentation it is predicted that SCF and SCA have a stronger association with performance in service companies than in manufacturing companies. Therefore, Hypothesis 12 is formulated:
Hypothesis 12: Studies which use samples captured from service firms will report stronger SCF/SCA-performance relationships than studies which use samples from manufacturing companies.
Year of data collection. The last theoretical moderator which will be included in this study is
phenomena. Therefore, the importance of SCF and SCA has increased in recent years. I predict that the studies for which data was collected in more recent years will have stronger SCF/SCA-performance relationships than studies conducted earlier. Therefore, Hypothesis 13 is formulated:
Hypothesis 13: Studies for which data was collected more recently will report stronger SCF/SCA-performance relationships than studies for which data was collected earlier in time.
3.
METHODOLOGY
In the following the methods of the literature search, the sample and coding procedures as well as the meta-analytical procedures are described.
3.1. Literature search
In order to conduct this meta-analysis, empirical studies which examined the relationships between SCA and performance, and SCF and performance were included. In order to identify as many relevant studies as possible, a computerized literature search was conducted. Combinations of the keywords supply chain agility, supply chain flexibility, agile, flexible,
supply chain, Triple-A, correlation, performance, outcome, profitability, competitiveness, impact and the abbreviations SCF and SCA were used in the literature search. Electronic
journal databases like EBSChost, JSTOR and Elsevier as well as Google Scholar and online archives of the different journals were revised. Additionally, the reference sections of the articles were checked for relevant studies. This procedure was frequently repeated until no further relevant empirical research studies were found. Though the literature search has been conducted with maximum carefulness, it cannot be entirely guaranteed that the sample is complete and includes all studies written on the topic. Studies published in the journals listed in Table A1 of the Appendix were included in this meta-analysis.
calculations. Secondly, two studies could not be based on the same sample, as the inclusion of the same sample twice would distort the results. Thirdly, the empirical studies included needed to measure the same constructs as explained in section 2, as well as the same direction of impact. Finally, attention had to be paid to the operationalization of the constructs in the sense of conformity to the constructs explained in section 2.
3.2. Sample and coding
The final sample of this meta-analysis includes 71 studies – 38 on the relationship of SCF-performance relationship and 33 on the of SCA-SCF-performance relationship. This Master thesis constitutes an update of an existing analysis of these relationships and extends it by five years from 2015 until April 2019. Additionally, other studies from the time period of the existing data set (1999-2014) will be included. The final sample covers a period from 1999 until April 2019. Detailed information about the final sample as well as the single studies included can be found in Tables 6, 7, and 8 respectively.
TABLE 6 - Overview of complete sample
SCF-P SCA-P Total
Studies (k) 38 33 71
Independent Samples
38 28 (Due to multiple uses of five studies the amount of studies for each
construct do not sum up to the total number of studies. These five studies are: Chan, Ngai, & Moo (2015, N = 141), Rouis (2010, N=60), Swafford et al. (2008, N=131), Um et al. (2017, N=363)
and Kumar et al. (2017, N=227))
66
Observations (N)
7,146 6,142
(922 are not added to total due to use in both relationships)
12,366
Several studies were excluded a priori as their effect sizes were around twice the size of the other effect sizes, and above the critical threshold of .70 and therefore constituted outliers (e.g. Dhiaf, Benabdelhafid, & Jaoua, 2012; Jin, Vonderembse, Ragu-Nathan, Turnheim Smith, 2014; and Vivek, Sen, Savitskie, Ranganathan, & Ravindran, 2011). In the occurrence that multiple studies used the same sample, only one of the studies was included in this meta-analysis. Studies which were excluded because of this reason are, for example, Qi, Zhao, and Sheu (2011), and Qrunfleh and Tarafdar (2013).
in ROI), (2) operational (cost, customer related measures, delivery and lead time, speed, etc.), (3) market (market share, market share growth, sales, sales growth) and (4) overall, and recorded the correlations. If an article investigated several performance measures, I coded them into each of the respective categories. Six studies were included in the SCF-FP relationship; 20 in the SCF-OPP relationship; 14 studies were included in the SCF-MP relationship, and 13 studies were included in the SCF-OP relationship.
The same procedure was conducted to code for H2 – the relationship between SCA and performance. Ten studies were included in the SCA-FP relationship; 17 studies were included in the SCA-OPP relationship; six studies were included in the SCA-MP relationship, and 13 studies were included in the SCA-OP relationship.
The moderator analysis was then conducted for each SCF-performance dimension, and SCA-performance dimension relationship. The methods to code for the methodological moderators are as follows. To code for H3, I grouped the studies into two groups according to their publication status - (1) published, or (2) unpublished. To code for the methodological moderator proposed in H4, I retrieved the Eigenfactor score of the respective journals from the website www.eigenfactor.org. Studies published in journals which are not recorded on this website were not included in the analysis (e.g. Lehnert, Zentes, & Schramm-Klein 2013; Bag et al., 2018). In order to code for H5, I grouped the articles into two categories of unit of analysis – (1) firm level, and (2) others, which include lower level unit of analyses, like SBU or plant.
26
TABLE 7 - Studies included in the analysis of SCF on performance
Study N Year Country Region Firm size Industry Type of firm Performance
measure Theory
Direct/ indirect effect
Agus (2011) 250 2008 Malaysia Asia - - manufacturer FP Program Theory Med
Amoako-Gyampah, Gyasi Boakye, Adaku, & Famiyeh (2019)
149 2016 Ghana Africa SML multiple manufacturer OP
CT, resource dependency theory; relational view
(RV)
Med
Bag, Gupta, & Telukdarie (2018) 175 2015 South Africa Africa SML multiple - OPP RBV, Institutional theory Mod
Banchuen, Sadler, & Shee (2017) 184 2008 Thailand Asia multiple manufacturer OPP Theory of strategic choice direct
Chan, Ngai, & Moon (2017) 141 2014 China and other
Asian countries Asia SML single manufacturer OP RBV Med
Chang, Chen, Lin, Tien, & Sheu
(2006) 105 2003 Taiwan Asia SML single manufacturer FP, OPP, OP - direct
Chavez, Yu, Jacobs, &
Feng(2017) 329 2014 China Asia SML multiple manufacturer OPP RBV direct
Danesh, Danaei, & Bazghal'eh
(2012) 199 2009 Iran Middle East - single - OPP - direct
Elmuti, Minnis, & Abebe (2008) 81 2003 U.S. North America L multiple manufacturer FP, OPP, MP, OP - direct
Fantazy, Kumar, & Kumar
(2009) 175 2006 Canada North America SME multiple manufacturer FP, OPP, OP - direct
Huo, Gu, & Wang (2018) 216 2015 China Asia SML multiple manufacturer OPP, OP Organisational capability
perspective direct
Ivens (2005) 206 2002 Germany Europe - single service OPP Relational-contracting
theory direct
Ko, Liu, Ngugi, & Chapleo
(2018) 236 2015 UK* Europe SME multiple manufacturer OPP external RBV Mod
Ku, Wu, & Chen (2016) 166 2013 Taiwan Asia SME food service service OPP RBV Mod
Kumar, Verma, Sharma and
Khan (2017) 227 2014 India Asia - multiple manufacturer OPP - direct
Lehnert, Zentes, &
Schramm-Klein (2012) 147 2000 - - - multiple multiple OPP CT, DCV direct
Liao, Hong, & Rao (2010) 201 2007 U.S. North America SML multiple manufacturer OPP RBV direct
Luo & Yu (2016) 212 2010 China Asia SML multiple manufacturer OPP CT, information
processing theory direct
Malhotra & Mackelprang (2012) 158 2009 U.S. North America SME multiple manufacturer OPP theory of complementarity direct
Mandal (2015) 163 2014 India Asia SML multiple OPP DCV, RBV Mod
TABLE 7 - Studies included in the analysis of SCF on performance (continued)
Study N Year Country Region Firm size Industry Type of firm Performance
measure Theory
Direct/ indirect
effect
Merschmann & Thonemann
(2007) 85 2006 Germany Europe SML multiple manufacturer OPP CT + process-based view Mod
Morash (2001) 111 1998 U.S., Canada North America - - multiple OP - direct
Obayi, Koh, Aglethrope, &
Ebrahimi (2017) 211 2014 UK* Europe SML multiple retailers OPP RV Med
Qi, Huo, Wang, & Yeung (2017) 604 2014 China Asia ML multiple manufacturer OP DCV Med
Qrunfleh (2010) 205 2007 U.S. North America ML multiple manufacturer OP CT direct
Rouis (2010) 60 2007 Tunisia Africa SME single service OP RBV, cognitive load
theory Med
Sánchez & Pérez (2005) 126 2003 Spain Europe SML multiple manufacturer FP, MP, OP - direct
Sánchez, Jiménez, Pérez, &
de-Luis-Carnicer (2009) 156 2005 Spain Europe SML multiple multiple OPP DCV direct
Sharma & Shah (2011) 82 2008 India Asia - single manufacturer OPP - direct
Song & Song (2009) 126 2006 China Asia SML multiple multiple OP - Med
Sreedevi & Saranga (2017) 91 2013 India Asia SML multiple manufacturer OPP - Mod
Swafford, Gosh, & Murthy
(2008) 131 2005 U.S. North America - multiple manufacturer OP RBV Med
Tokman, Richey Jr, Morgan,
Marino, & Dickson (2013) 209 2008
Finland, Sweden,
Norway Europe SME multiple - OP
RBV,strategic behavior
theory Mod
Um, Lyons, Lam, Cheng, &
Dominguez-Pery (2017) 363 2014
UK & South
Korea Europe & Asia SML multiple manufacturer OPP DCV Med
Vickery, Calantone, &
Dröge(1999) 65 1996 U.S. North America - single - FP, MP, OP - direct
Wagner, Grosse-Ruyken, &
Erhun (2018) 336 2015
UK, France, Germany, Austria,
Switzerland and US
Europe & North
America SML multiple manufacturer OPP, OP
Information processing
theory direct
Ying (2010) 192 2007 China Asia SML single multiple OPP
RBV, culture theory of organizational
effectiveness
Med Zhang, Vorderembse, & Lim
(2005) 273 2002 U.S. North America ML multiple manufacturer OPP
Competence and capability
theory direct
28
TABLE 8 - Studies included in the analysis of SCA on performance
Study N Year Country Region Firm size Industry Type of firm Performance
measure Theory
Direct/ indirect
effect
Alfalla-Luque, Machuca, &
Marin-Garcia (2018) 151 2016 - Europe ML multiple manufacturer FP, OPP DCV, RBV direct
Al-Shboul ( 2017) 113 2014 - Middle East L multiple manufacturer OP - direct
Attia (2015) 153 2012 Egypt * Africa - single - OPP - direct
Blome, Schoenherr, & Rexhausen
(2013) 121 2010 Germany Europe L multiple manufacturer OPP DCV, RBV direct
Chan, Ngai, & Moon (2017) 141 2014 China and other
Asian countries Asia SML single manufacturer OP RBV Med
Chen (2018) 204 2015 Taiwan Asia SML multiple manufacturer MP DCV direct
DeGroote & Marx (2013) 193 2010 U.S. North America L multiple manufacturer OPP, MP - direct
Dhaigude & Kapoor (2017) 122 2014 India Asia SML multiple manufacturer OPP DCV Med
Eckstein, Goellner, Blome, &
Henke (2015) 143 2012 Germany Europe SML multiple multiple OPP CT, DCV Mod
Ghosh (2007) 115 2004 - - ML multiple manufacturer OP Process-based view of SC Med
Gligor & Holcomb (2012) 151 2009 - North America ML multiple multiple OPP DCV, RV direct
Gligor (2016) 242 2013 - L multiple manufacturer FP DCV Mod
Gligor, Esmark, & Holcomb
(2015) 283 2012 US North America ML multiple manufacturer FP, OPP DCV Mod
Hwang & Kim (2018) 279 2015 Korea Asia SML multiple manufacturer OP RBV, RV direct
Kumar, Verma, Sharma, & Khan
(2017) 227 2014 India Asia - multiple manufacturer OPP - direct
Li, Wu, Holsapple, & Goldsby
(2017) 77 2014 USA North America SML multiple manufacturer FP DCV direct
Liu, Ke, Wei, & Hua (2013) 286 2010 China Asia SML multiple multiple OP DCV Med
Liu, Shang, Lirn, Lai, & Lun
(2018) 112 2014 Taiwan Asia - single service OP RBV Med
McCann, Selsky, & Lee (2009) 471 2006 Canada, U.S.,
Mexico North America ML multiple - MP - Med
Qi, Boyer, & Zhao (2009) 604 2006 China Asia ML multiple manufacturer FP, OPP, OP - direct
Qrunfleh & Tarafdar (2013) 205 2007 U.S. North America ML multiple manufacturer OPP, OP RBV, strategic-choice
theory Med
Rouis (2010) 60 2007 Tunisia Africa SME single service MP RBV, cognitive load theory Med
Saeed, Malhotra, & Abdinnour
(2019) 103 2016 US North America SML multiple manufacturer OPP
DCV, co-specialization
perspective direct
Sahin, Cemberci, Civelek. & Uca
(2017) 247 2014 Turkey Europe SML multiple multiple OP - direct
Sukwadi, Wee, &Yang (2013) 160 2010 Indonesia Asia SME single - OPP - direct
TABLE 8 - Studies included in the analysis of SCA on performance (continued)
Study N Year Country Region Firm size Industry Type of firm Performance
measure Theory
Direct/ indirect
effect
Swafford, Gosh, & Murthy
(2008) 131 2005 U.S. North America - multiple manufacturer OP RBV Med
Tse, Zhang, Akhtar, & MacBryde
(2016) 266 2013 China Asia SML single manufacturer OP - Med
Um, Lyons, Lam, Cheng, &
Dominguez-Pery (2017) 363 2014
UK & South
Korea - SML multiple manufacturer OPP DCV Med
Vickery, Dröge, Setia, &
Sambamurthy (2010) 57 2007 U.S. North America L single manufacturer FP
Theory of resource
complementarities Med
Whitten, Green, & Zelbst (2012) 132 2009 U.S. North America L multiple multiple FP, OPP, MP DCV, complex adaptive systems theory direct
Wieland & Wallenburg (2012) 268 2010 Germany, Austria,
Switzerland Europe - multiple manufacturer FP, OPP - direct
Yang (2014) 137 2011 China Asia - multiple manufacturer OPP, OP Information theory,
transaction cost economics Med: Yusuf, Gunasekaran, Musa,
Dauda, El-Berishy, & Cang (2014)
95 2011
UK *culture scores from Hofstede Website
Europe SML single - FP, MP, OP - direct
30
3.3. Meta-analytic procedures
In this bivariate meta-analysis, I used the procedure suggested by Hunter and Schmidt (2004). In accordance with Geyskens, Krishnan, Steenkamp, & Cunha (2009), I corrected for measurement unreliability in the dependent and independent variables in each relationship. In order to do so, I used the Cronbach’s alpha values of each study. This value is not given in all studies. In these cases, I calculated the average Cronbach alpha’s value across all studies and used it for the studies in which it was missing (Lipsey & Wilson, 2001). Additionally, I corrected for sampling error, which occurs because only a sample rather than the whole population is tested in each study (Wowak et al, 2013). Then, I calculated the sample size weighted and reliability adjusted average correlation coefficients (also combined effect size or corrected correlation coefficient) for each SCF- and SCA-performance relationship. Additionally, I calculated the confidence interval (CI) at 95% for the combined correlation coefficients.
In order to identify outliers, studies with correlations higher than .70 were not included in this meta-analysis. At the same time, outliers can also result from the sample size of the studies included in the meta-analysis. If the sample size of a particular study is significantly smaller or larger than the sample sizes of the other studies included in the meta-analysis, it can be considered an outlier. Therefore, the sample size weighted and reliability adjusted average correlation coefficient were calculated before and after outlier removal (Geyskens et al., 2015).
To estimate the extent of heterogeneity of the effect sizes three variables are calculated – Q-statistic, I2, and Tau. The Q-statistic is an indicator for the heterogeneity of variance, and specifically it is an indicator for “the degree of difference between the observed and expected effect sizes” (Ellis, 2010: 107). If it is significant, the distribution is heterogeneous and the samples probably do not stem from the same population. A moderator analysis needs to be conducted in order to explain the heterogeneity. The I2-value is an indicator for the excess dispersion divided by the total dispersion. A value of I2 = 25% is considered to be low, 50% mediate and 75% is considered to be a high relative variance. If the I2-value is relatively high
influencing the effect sizes are listed in section 2.2. For these moderators the significance and measures of heterogeneity were calculated.
Additionally, for those relationships for which more than ten studies were available, I tested for publication bias. As explained above, publication bias exists if studies are more likely to be published depending on the direction, magnitude or significance of the results of the study (Begg, 1994). In order to do so, I used the trim-and-fill procedure and looked at the fixed effect (Duval & Tweedie, 2000).
4.
RESULTS
In the first part of this section, I report the results of the bi-variate meta-analysis of the SCF-performance relationships followed by the results of their moderator analyses in the second part. In the third part, I report the results of the bi-variate meta-analysis of the SCA-performance relationships, followed by the results of their moderator analyses in the fourth part.
4.1. Results of the meta-analysis SCF-performance
The results are reported in Table 9. H1 predicted that SCF is positively associated with performance. The results show a positive and statistically significant sample-size weighted and reliability adjusted average correlation coefficient for the relationship of SCF and the performance construct (r= .33, k = 38). Looking at the SCF-FP relationship, H1 cannot be supported as the confidence interval of the combined effect size includes zero which renders the effect size statistically insignificant (r = .17†, k = 6). Thus, no moderator analysis is conducted on this relationship however it is included in the commonality analysis. The results show a positive and statistically significant corrected correlation coefficient for the other relationships which are SCF-OPP (r = .34, k = 20), SCF-MP (r = .34, k = 14) and SCF-OP (r = .23, k = 13). Thus, H1 is supported for these three relationships. Cohen (1988) suggests that an effect size r is considered small if it is between|.1| and |.29|, medium if it is between|.3| and |.49|, and strong if it is between |.5| and |1|. Therefore, the effect sizes for the significant relationships between SCF and the different performance dimensions range from small to medium.
32
322.78; I2 = 89%; SCF-OPP: Q = 250.88, I2 = 90%; SCF-MP: Q = 125.22, I2 = 90%, SCF-OP: Q = 30.98, I2 = 61%). These variables indicate that the domains are heterogeneous which
is why a moderator analysis is useful to understand by what factors the heterogeneity is determined. Hence, a moderator analysis was conducted for each statistically significant SCF-performance dimension relationship (for results see 4.2. and Table 10).
In accordance with the recommendations by Geyskens et al. (2009), I checked the SCF-performance relationships for outliers. In order to do so, the modified sample-adjusted meta-analytical deviancy statistic was calculated (Huffcutt & Arthur, 1995). In the SCF-OPP relationship one outlier was detected. Hence, this effect size was removed. This led to an increase of the sample-size weighted and reliability adjusted average correlation coefficient (before outlier removal: r = .37, k = 20; after outlier removal: r = .42, k = 19).
Publication Bias: In order to identify publication bias, the trim and fill method was employed.
In this way, the combined effect sizes were adjusted for the missing studies (Duval & Tweedie, 2000). For the SCF-performance relationship, I found evidence for publication bias. Eight missing study effect sizes were imputed on the right side, increasing the effect size by .06. For the relationships of SCF-OPP, and SCF-MP, no evidence for publication bias was found. For the relationship SCF-OP, I found evidence for publication bias. Four missing study effect sizes were imputed on the right side, increasing the effect size by .06. After outlier removal, I found evidence for publication bias for the SCF-OPP relationship. Three missing study effect sizes were imputed on the right side, increasing the effect size by .05.
4.2. Results of the moderator analysis SCF-performance
For each SCF-performance dimension relationship, a moderator analysis was conducted for the methodological and theoretical moderators explained in sections 2.2 and 3.2. The results are reported in Table 10. For the SCF-MP relationship, all studies were published which is why this moderator was excluded from the analysis. In the following, I will only concentrate on moderators which are statistically significant.