The replacement effect: compensating for organizational culture flaws with HRM practices in lean manufacturing implementation

(1)

1

The replacement effect: compensating for organizational

culture flaws with HRM practices in lean manufacturing

implementation

MSc Technology & Operations Management

Rijksuniversiteit Groningen Faculteit Economie en Bedrijfskunde

Zernike Gebouw, Nettelbosje 2 Groningen, Groningen, 9747 AE Tel: (050) 363 3741 Fax: (050) 363 7970 S.A. Antonissen S1969323 S.A.Antonissen@student.rug.nl Supervisor N. Ziengs, MSc Co-assessor Prof. dr. ir. J. Slomp

(2)

2

Abstract

Purpose – Many plants fail to implement lean manufacturing (LM) practices successfully. Important to success is human behavior, which is influenced by Human Resource Management (HRM) practices and organizational culture (OC). The aim of this study is to investigate whether HRM practices can compensate for OC flaws at plants to achieve high manufacturing performance: a phenomenon called the replacement effect.

Design/methodology/approach – The study uses data obtained from a survey of 133 manufacturing plants and from interviews with consultants experienced in LM implementation. Findings – The outcomes of the study show that HRM practices can compensate for OC flaws to mitigate the negative impact on manufacturing performance, as HRM practices can steer the behavior of employees by means of, among others, employee empowerment and structures. Furthermore, the outcomes show that HRM practices strengthen the impact of Just-In-Time (JIT) and Total Quality Management (TQM) implementation on manufacturing performance, as HRM practices guide the behavior of employees in using JIT&TQM tools with the use of, among others, coaching and structures.

Practical implications – A major implication of this study is that practitioners have to be aware of OC, as OC highly impacts manufacturing performance. Furthermore, the findings assist organizations by showing that HRM practices can be used to compensate for OC dimensions with regard to its impact on manufacturing performance, implying that practitioners can focus on developing HRM practices instead of focusing on both OC and HRM development.

Originality/value – This paper uses both quantitative and qualitative data to explore the replacement effect. Next to showing the existence of the replacement effect between OC and HRM practices, this paper explains why and how HRM practices can compensate for OC dimensions. Furthermore, this paper shows and explains that HRM practices increase the effectiveness of JIT&TQM implementation.

Keywords Replacement Effect, Lean Manufacturing, Organizational Culture, Human Resource Management Practices

(3)

3

1. INTRODUCTION

In today’s organizations turbulent environments, organizations are continuously attempting to improve (Belekoukias, Garza-Reyes, & Kumar, 2014). A widely recognized and powerful tool for improvement is lean manufacturing (LM) (Belekoukias et al., 2014; Bortolotti, Boscari, & Danese, 2015; Demeter & Matyusz, 2011; Shah & Ward, 2003). Despite lean’s popularity, many organizations that applied lean failed to achieve the desired manufacturing performance (Bhasin & Burcher, 2006; Bortolotti et al., 2015; Naor, Goldstein, Linderman, & Schroeder, 2008).

It can be stated that factors related to human behavior are among the most likely reasons as to why LM implementations fail. Marodin and Saurin (2013) conducted a literature review on, among others, factors that influence the implementation of LM and classified these factors as human, work organization, external environment, and technological. Especially the human elements, like problem solving, teamwork, and leadership, are “vital for success and is resolute that people and cultural change are predominant reasons for lean failures” (Bhasin & Burcher, 2006, p. 65). Next to the importance of these factors separately, Marodin and Saurin (2013) emphasize that interrelationships between factors impacting LM implementation should be understood, since understanding their interactions provides a broader view on why organizations fails or succeed in implementing LM successfully.

As a consequence of the importance of human behavior, the current study focusses on two important success factors for LM implementation: organizational culture (OC) and human resource management (HRM) lean practices. The focus on OC is valuable, as OC is argued by several researchers to be an explanation of the low effectiveness of LM (Bortolotti et al., 2015; Naor et al., 2008). Bortolotti et al. (2015) conducted research on the impact of OC on LM and suggest that further research is required as the role of OC in lean initiatives is still poorly understood. Specifically, this role of OC relates to whether OC is an antecedent of LM or whether there exists a recursive relationship between OC and LM.

(5)

5

successful lean organizations with unsuccessful lean organizations (Bortolotti et al., 2015). It has to be noted that even though the successful lean organization is differentiated by soft practices, the hard practices are still important for LM (Bortolotti et al., 2015). In other words, it can be stated that the soft practices are the order winners and the hard practices are the order qualifiers.

Consistent with suggestions for future research by prior studies, this study investigates the interrelationships between OC and HRM practices (Bortolotti et al., 2015; Marodin & Saurin, 2013). Bortolotti et al. (2015) suggest further research on how soft lean practices (HRM practices) and OC relate to each other and mutually add to a successful LM implementation. A preliminary view on the kind of interrelationship between OC and HRM practices is provided by Bortolotti et al. (2015), who state that it might be the case that, among others, HRM practices can replace cultural dimensions for high performance, which they call the replacement effect.

Two views can be used to explore the relationship between OC and HRM practices: the compensatory view and the additive view. The compensatory view holds that lack of presence of certain practices can be offset by a high degree of presence of other practices (Wu, Melnyk, & Swink, 2012). Contrary, the additive view states that certain practices must be present and cannot be compensated by other practices (Wu, Melnyk, & Swink, 2012). In the latter view, trade-offs are impossible. Relating this to OC and HRM, it could be the case that the investments in the development of OC or HRM practices are compensatory, meaning that a lack of development in OC might be compensated for by investments the HRM bundle. Whether HRM practices can compensate for OC flaws (OC aspects in the organization that negatively impact manufacturing performance) is not researched before, making it the focus of this research to fill that knowledge gap.

(6)

6

compensate for OC flaws in the LM implementation at manufacturing plants? In order to answer this research question, this research will use surveys and conduct a case study with interviews.

This study allows managers to tailor the LM implementation to their OC to improve the chances of it being successful. By examining whether HRM bundle implementation can counter the negative effects of certain aspects of OC, this study provides a guideline in the choices in LM implementation. This study shows practitioners whether practitioners have to focus on both OC and HRM development, or can focus on HRM development as it compensates for OC dimensions.

This paper is structured as follows. First, the theoretical framework is provided. Then, the methodology that is applied in this research is explained. Next, the results of the research are presented, followed by a discussion of the results. The final section of this paper consists of a conclusion, including the limitations and suggestions for future research.

2. THEORETICAL FRAMEWORK

2.1 Lean manufacturing

Lean manufacturing (LM) can be seen as a management philosophy to manufacturing and is implemented by using a set of socio-technical practices focusing on the elimination of waste and increasing value along the value chain within and between organizations (Belekoukias et al., 2014; Dal Pont et al., 2008). According to Dal Pont et al. (2008), the lean philosophy in a manufacturing environment is based on “doing the simple things well, gradually doing them better and squeezing out waste at every step of the value chain” (p.150). On a more operational level, this lean philosophy is translated into a set of techniques and practices, i.e. lean bundles, which implement and facilitate the lean philosophy aiming at the elimination of waste and continuous improvement (Demeter & Matyusz, 2011; Shah & Ward, 2003, 2007). The lean practices are often classified in three bundles: TQM bundle, JIT bundle and HRM bundle.

2.2 Human resource management bundle

(7)

7

thus avoiding traditional barriers that hamper continuous improvement (Dal Pont et al., 2008; Shah & Ward, 2003).

An appropriate level of HRM bundle implementation is often argued to be a prerequisite for implementing other lean practices (Bortolotti, Danese, Flynn, & Romano, 2014; Dal Pont, Furlan, & Vinelli, 2008; Naor, Goldstein, Linderman, & Schroeder, 2008). Failing to start with an appropriate level of HRM bundle implementation (highly related to what other studies call organizational infrastructure, fitness bundle, quality management infrastructure) hampers the bundles JIT and TQM in achieving their potential (Bortolotti et al., 2014). Various studies support the importance of HRM practices (Naor et al., 2008; Sousa & Voss, 2002), stressing its impact on the technical lean practices and/or the manufacturing performance. Therefore, it can be stated that the HRM bundle is an important subject of inquiry. The HRM bundle is consisting of, among others, the following practices: management leadership, multifunctional employees, teamwork, employee involvement, employee empowerment, and training (table 1) (Bortolotti et al., 2015, Shah & Ward, 2003).

As the HRM bundle is often discussed in the literature in relation to technical lean bundles (Dal Pont et al., 2008; Furlan et al., 2011; Shah & Ward, 2003), the most notable lean bundles, JIT and TQM, have to be briefly elaborated upon. TQM is a manufacturing approach with the purpose of continuously enhancing and supporting the quality of products and processes in order to meet or exceed customer expectations (Cua, McKone, & Schroeder, 2001). The practices in this bundle include statistical process control, proprietary equipment development, cleanliness and organization, process management, product design and management, and quality data analysis (Cua et al., 2001; Shah & Ward, 2003).

(8)

8

Table 1 HRM bundle practices: definitions and examples of prior studies that used these concepts

HRM bundle practice Definition Example of studies on LM who used this concept

Management leadership Encouraging the practices and behaviors that result in increased manufacturing performance throughout the organization

Bortolotti et al. (2015), Cua et al. (2001), Flynn et al. (1994; 1995) Multi-functional

employees

Individuals capable of doing various activities instead of focusing on one

Bortolotti et al. (2015, 2014), Cua et al. (2001), Dal Pont et al. (2008), Furlan et al. (2011), Swink et al. (2005)

Teamwork A coordinated effort by a group of individuals to reach a common goal

Bortolotti et al. (2014), Cua et al. (2001), Dal Pont et al. (2008), Furlan et al. (2011)

Employee involvement Employee participation in decision-making and implementation in the plant (Amah & Ahiauzu, 2013)

Bortolotti et al. (2014), Cua et al. (2001), Furlan et al. (2011), Hofer et al. (2012), Shah and Ward (2007)

Employee empowerment

The actual capacity by employees to make decisions over a given area by themselves without the consent of the manager (Verhoest et al., 2004)

Swink et al. (2005)

Training The instruction and education to individuals or groups to increase their knowledge and skills related to their tasks

Bortolotti et al. (2015), Cua et al. (2001), Dal Pont et al. (2008), Furlan et al. (2011)

2.3 Organizational culture

Generally, OC represents a collection of shared assumptions, values and beliefs among members of an organization (Detert, Schroeder, & Mauriel, 2000; Schein, 2004; Zu, Robbins, & Fredendall, 2010). Specifically, OC is defined as “a combination of artifacts, values and beliefs, and underlying assumptions that organizational members share about appropriate behavior” (Detert et al., 2000, p. 851). A comparable definition is provided by Liu et al. (2010): “A collection of shared assumptions, values, and beliefs” (p.375), which is reflected in organizational behavior and goals “and that helps its members to understand organizational functioning” (p.375). From these definitions, this paper defines OC as follows: A collection of shared assumptions, values, and beliefs that organization’s members help to understand how the organization functions and that affect the way that organization’s members behave. Schein (2004) posits that culture steers and constrains the behavior of the group members by the shared values, beliefs, and underlying assumptions that are held in in that group (Schein, 2004).

(9)

9

tangible. This means that whereas HRM practices are quite easy to describe and observe, OC is more difficult to observe since it is woven in the assumptions and inner thoughts of the people in the organization. Consequently, often OC is overlooked since it is embedded into the organization.

Next to the previous mentioned distinction, OC differs from HRM practices in that OC is hard, or even impossible, to change in a short time period (Burnes, 2009). According to Schein (2004) culture develops over time as “A pattern of shared basic assumptions that was learned by a group as it solved its problems of external adaption and internal integration that worked well enough to be considered as valid.” (p.17) It takes a long time to change these assumptions. On the other hand, HRM practices can be more easily implemented.

In the current study, six dimensions of OC are included: power distance, institutional collectivism, in-group collectivism, performance orientation, assertiveness, and humane orientation (table 2).

Table 2 Dimensions of OC: definitions

Adopted from Bortolotti et al. (2015, p. 184), derived originally from House et al. (2004)

OC dimension Definition

Power distance The degree to which members of an organization expect and agree that power should be stratified and concentrated at higher levels of an organization

In-group collectivism The degree to which individuals express pride, loyalty, and cohesiveness in their organizations

Performance orientation The degree to which an organization encourages and rewards group members for performance improvement and excellence

Institutional collectivism The degree to which organizational institutional practices encourage and reward collective distribution of resources and collective action

Assertiveness The degree to which individuals in organizations are assertive, confrontational, and aggressive in social relationships

Humane orientation The degree to which individuals in organizations encourage and reward individuals for being fair, altruistic, friendly, generous, caring, and kind to others

2.4 Replacement effect: Compensatory relationship between OC and HRM practices

(10)

10

replace, and there are significant costs involved in attracting and training new people (Bortolotti et al., 2015). The problem of low IG can be reduced by multi-skilled trained employees, as this “decreases the inefficiencies due to sudden absences and high turnover.” (Bortolotti et al., 2015: p. 194)

Whether this compensatory effect holds for OC and HRM practices in organizations affects the way managers deal with choices in the LM implementation regarding the development of OC and HRM in the LM implementation. When HRM practices do compensate for OC dimensions, managers could focus on the development of HRM practices. In case HRM practices cannot compensate for OC dimensions, managers should focus on the development of both OC and HRM practices.

In order to observe this compensatory effect, HRM bundle implementation can be viewed as a moderator for the relationship between OC and manufacturing performance, as the compensatory effect is confirmed when the impact of OC on manufacturing performance weakens in case of high HRM bundle implementation. In this way, the HRM bundle might compensate for OC flaws. In previous research on LM implementation, HRM bundle implementation is also positioned in a moderator role (Ahmad, Schroeder, & Sinha, 2003). Ahmad et al. (2003) found that the impact of JIT practices on plants competitiveness becomes stronger in a situation with high HRM bundle implementation. The HRM bundle is related to creating an environment that is conductive for technical lean practices to be effective (Ahmad et al., 2003). This points to a contextual role of HRM bundle implementation, making a moderator role most appropriate for HRM bundle implementation.

In the following section, hypotheses are developed regarding whether a lack on certain dimensions of OC can be compensated by a high degree of implementation of HRM practices.

2.5 Hypothesis development

(11)

11

Impact OC on manufacturing performance. Several studies investigate the impact of OC on manufacturing performance (Bortolotti et al., 2015; Naor et al., 2008, 2010). These studies show that different cultural traits result in varying degrees of effectiveness in quality management or LM organizations (Bortolotti et al., 2015; Naor et al., 2008, 2010). The outcomes of the study of Bortolotti et al. (2015) present that successful LM organizations were characterized by higher institutionalism, future orientation, humane orientation, and assertiveness. The insights gained from the research by Naor et al. (2010) show that plants having a supportive OC have a higher manufacturing performance. Specifically, an OC that increases manufacturing performance contains low power distance and assertiveness and high institutional collectivism, in-group collectivism, humane orientation.

To investigate the possible moderating effect of HRM on the impact of OC on JIT & TQM implementation and manufacturing performance, the relationship of (manifestations of) HRM practices with each of the OC dimensions has to be discoursed.

Power distance and HRM practices. The outcomes of the study by Bortolotti et al. (2015) show that several OC dimensions are distinguishing features of successful non-LM organizations, as they are not for successful LM organizations. This outcome might be clarified with that specific lean practices might compensate for or replace the cultural dimensions. As HRM practices are, among the lean practices, the most related to OC dimensions, it might be the case that these practices replace the need of certain cultural dimensions. One of these OC dimensions that came forward in this way in Bortolotti et al.’s (2015) study is power distance. Normally, in non-LM organizations, high power distance is argued to hamper the communication within the organization and impede innovation and effective collaboration between different layers (Bortolotti et al., 2015). In LM organization, though, this negative impact is mitigated by practices that decrease the need for face-to-face communication at moments that problems emerge. Due to high empowerment in successful LM organizations, which is a HRM practice, employees are given a lot autonomy, making it less necessary to communicate all problems with upper levels.

(12)

12

cohesiveness in their organizations (Bortolotti et al., 2015). For non-LM organizations, a high degree of in-group collectivism was distinguishing successful from unsuccessful organizations, whereas this was not a distinguishing dimension for LM organizations. A possible clarification here is also that certain lean practices replace the need for a high degree of in-group collectivism in successful LM organizations. According to Bortolotti et al. (2015), “low in-group collectivism could lead to high levels of personnel turnover due to low loyalty to their organization” (p.194). In turn, this would affect the organization negatively, since in non-LM organizations workers are very specialized and hard to replace (Bortolotti et al., 2015). This problem is mitigated in LM organizations, since employees are multi-skilled and can in turn be used for multiple tasks within the organization. Multi-skilled employees can mitigate the negative effects from a higher turnover.

Performance orientation and HRM practices. A similar interpretation of the results in the study of Bortolotti et al. (2015) was stated for performance orientation. For non-LM organizations, performance orientation is of indispensable value, since it directs employees towards the performance goals and to look for ways for performance improvements by means of rewarding and appreciation (Bortolotti et al., 2015). Contrary, for successful LM organizations, performance orientation was no distinguishing element. This might suggest that lean practices mitigate the adverse effects of a low degree of performance orientation. One possibility may be that the successful LM organizations mitigate this by HRM practices. It might be that employees are more intrinsically motivated to search for improvements and innovative behavior as a results of the fact they are more involved and have more empowerment (Scott & Bruce, 1994).

(13)

13

Assertiveness and HRM practices. As was mentioned in section 2.3, assertiveness lies in the way organization’s individuals are confrontational and aggressive in social relationship (Bortolotti et al., 2015). This can be related to the extent and the way in which employees communicate with each other. Organizations that possess low amounts of assertiveness tend “to seek consensus and solve conflict among individuals” (Bortolotti et al., 2015, p. 187; Naor et al., 2010), whereas high assertiveness can be associated with individualism (Morrison, Chen, & Salgado, 2004). HRM practices might also cover this dimension of OC, by encouraging interaction between employees, teamwork, and group problem solving. These practices are associated with collectivism. On the other hand, some HRM practices encourage employees to make independent decisions and give them a lot of autonomy, which is associated with individualism (Hui & Villareal, 1989). Depending on the implemented HRM practices, the impact of the assertiveness dimension of OC on manufacturing performance might be moderated. For example, a low score on assertiveness for enhancing manufacturing performance might be replaced by high encouragement for group problem solving.

Humane orientation and HRM practices. Humane orientation focuses on acknowledging contributions of employees to achieve operational excellence, instead of punishing employees when something goes wrong (Bortolotti et al., 2015). In such an OC, employees try to make the organization successful, instead of pursuing personal benefits (Bortolotti et al., 2015). The negative consequences of a lack of a humane orientation might be mitigated by HRM practices like problem solving, employee suggestions, and teamwork. By promoting interaction and actively asking employees for input, employees feel more valuable and become aware that their input is important for the continuous improvement of the organization.

It can be stated from the preceding that HRM practices might be able to compensate for certain OC dimensions to reach high manufacturing performance. Therefore, the following hypotheses can be proposed:

H1a: OC is positively related to manufacturing performance.

(14)

14

Impact OC on JIT & TQM implementation. Naor et al. (2008) investigated the relationship between culture and core quality practices, whereby core quality practices are referred to as quality information, process management, and product design. The results of their study show no significant relationship between OC and core quality practices. As a possible explanation, Naor et al. (2008) state that OC might be more related to behavioral and social aspects and core quality practices are more technically focused, depending more on mathematical skills and automation.

Contrary to the findings of Naor et al. (2008), findings of the work by Baird, Hu and Reeve (2011) indicate that cultural dimensions related to group feeling and respect to people are of significant importance in enhancing the use of technical TQM practices. Hence, organizations that value and respect individuals in the organization and that encourage collaborations between individuals, teams, and divisions, will to a greater extent use technical TQM practices (Baird et al., 2011). Next to the study by Baird et al. (2011), the results of the study by Lagrosen and Lagrosen (2005) suggest that values of continuous improvement, management by facts, and participation positively influence the functioning of technical quality management efforts in organizations. Lagrosen and Lagrosen (2005) state that for successful implementation of TQM, an OC is required that is permeated by these values. Although the research findings are not unanimous, there is a tendency to expect that OC positively influences the technical part of JIT & TQM bundle implementation.

OC elements like respect for people, collaboration, and participation are argued to be vital OC aspects that positively influence JIT & TQM bundle implementation. To mitigate the negative impact on JIT & TQM bundle implementation in case of absence of these values, HRM practices aimed at increasing the elements like respect for people, collaboration, and others can be implemented. Examples of HRM practices that could increase these aspects are teamwork, group problem solving, employee involvement, and empowerment.

From the previous, the following hypotheses are composed: H2a: OC is positively related to JIT & TQM bundle implementation.

(15)

15

Impact JIT & TQM bundle implementation on manufacturing performance. The impact of JIT & TQM bundle implementation on manufacturing performance is widely discussed in literature (Cua et al., 2001; Dal Pont et al., 2008; Furlan et al., 2011; Shah & Ward, 2003). The results of the study by Cua et al. (2001) show that the joint implementation of technical-oriented practices, next to socially-oriented practices, explained differences in performance measurements significantly. Dal Pont et al. (2008), in line with the results of the results of Cua et al (2001), found a direct and positive effect of JIT & TQM bundles on manufacturing performance.

Although the positive impact of JIT & TQM bundles on manufacturing performance is widely agreed upon in existing literature, it are the soft, HRM related, lean practices that distinguish high performing LM organizations from low performing LM organizations (Bortolotti et al., 2015). This might imply that the effectiveness of JIT&TQM implementation is increased in situations where HRM bundle implementation is high. Therefore, it can be stated that the relationship between JIT & TQM bundle implementation and manufacturing performance might be moderated by HRM practices, in a way that the positive effect of TQM & JIT bundles on manufacturing performance is increased when HRM practices are implemented extensively.

From the preceding, the following hypotheses are composed:

H3a: JIT & TQM bundle implementation is positively related to manufacturing performance.

H3b: The effect of JIT & TQM bundle implementation on manufacturing performance is moderated by HRM bundle implementation, such that the relationship becomes stronger when

HRM bundle implementation is high rather than low.

(16)

16

Figure 1 Conceptual framework

3. METHODOLOGY

(17)

17

Figure 2 Display of methodological organization of the study (adapted from Jabbour et al.(2014))

3.1 Methodology quantitative phase in the research

Data content. The data content that was aimed to be collected was different in the two phases of the research. In the first phase, the quantitative phase, data intended to be collected were related to the “what” relationships. So, “what” is relationship between OC, JIT&TQM, and manufacturing performance, and “what” is the impact of HRM practices on the aforementioned relationships.

(18)

18

country. In order to ensure accurate translation, the translation was checked by another researcher.

Table 3 Survey sample distribution.

* Industries (manufacture of...) 1=food products and beverages, 2=fabricated metal products, except machinery and equipment, 3=machinery and equipment, 4=electrical machinery and apparatus, 5=medical, precision and optical instruments, watches and clocks, 6=motor vehicles, trailers and semi-trailers, 7=other transport equipment, 8=other

Industry* Total plants per country 1 2 3 4 5 6 7 8 Country CH ₄ ₅ ₃₅ ₁₆ ₆ ₁ ₃ ₂₂ ₉₂ DE 0 0 0 0 0 0 0 1 1 NL ₃ ₈ ₁₂ ₄ ₁ ₄ ₁ ₇ ₄₀

Total plants per

industry 7 13 47 20 7 5 4 30 133

The questions and scales used in the questionnaires are based on the literature and earlier used measurement scales. Even though OC is intangible and often invisible, there are various models proposed to measure OC, with many dimensions associated with OC (Jung, Bower, & Mannion, 2009). The measurement of the multidimensional construct OC is grounded on the operationalization and scales used by Bortolotti et al. (2015) and Naor et al. (2010) and originates from the GLOBE model (House et al., 2004), of which the following dimensions are included: power distance, in-group collectivism, performance orientation, institutional collectivism, assertiveness, humane orientation. The respondents were requested to deliver their perceptions of the amount of presence of the different dimensions by Likert scales perceptual items, with values varying from 1 (strongly disagree) to 5 (strongly agree). It has to be stated that OC is considered formative construct meaning that it can score high on one dimension and low on another dimension.

The HRM, TQM, and JIT lean bundles were measured grounded on measurement items and scales used by Dal Pont et al. (2008), Furlan et al. (2011) and Shah and Ward (2007). The multi-items concerning the lean bundles were obtained by using a 5-point Likert scale where the respondents were asked about their perception on the actual implementation of each practice.

(19)

19

manufacturing performance. The respondents had to provide their perceptions of the performance of their plant compared to three years ago on several items of the four dimensions by means of a Likert scale, with values ranging from 1 (poor, low) to 5 (superior). The choice has been made for manufacturing performance compared to three years ago, as many organizations put extensive effort in LM the last three years and we are aiming for the impact of, next to OC, LM implementation on manufacturing performance. It has to be noted, though, that the difference in manufacturing performance compared to three years ago may also be caused by other reasons than LM.

There is a construct validity issue regarding whether manufacturing performance is a unidimensional, reflective construct. Two contradicting conceptual views on manufacturing performance are the tradeoff view and the production competence view (Shah & Ward, 2003). The researcher argues in favor of the production competence view for three reasons. First, the researcher is convinced that organizations are able to develop multiple competences, which impact multiple manufacturing performance dimensions. Second, as organizations currently operate in a highly competitive global world, it is necessary to possess multiple competences to succeed and survive (Shah & Ward, 2003). Lastly, the underlying manufacturing performance dimensions are highly related, especially for plants that put effort in LM implementation. As an example, the philosophy behind JIT is that processes are quicker and more reliable, in order to serve in the first time the right quantity in the right quality at the right time. Serving a product in the right quality in the first time also implies less costly (Shah & Ward, 2003). Therefore, the researcher argues that the dimensions of manufacturing performance are related and represent a single, reflective, construct, which is in line with other research in Operations Management (e.g. Bortolotti et al., 2015; Furlan et al., 2011; Naor et al., 2010; Shah & Ward, 2003).

(20)

20

series mean. The mean substitution missing data handling technique is chosen, because this is an appropriate technique when less than 10% of the data is missing and it preserves the data (Tsikriktsis, 2005).

In order to assess the reliability of the reflective constructs (JIT&TQM and manufacturing performance), the Cronbach’s alpha and Composite reliability are determined (Hair, Black, Babin, & Anderson, 2009). The Cronbach’s alpha of all the main constructs exceed the threshold of 0.6 (Hair et al., 2009) and the composite reliability threshold of 0.6 for exploratory research (Bagozzi & Yi, 1988) (table 4). Therefore, it can be concluded that the reliability of the constructs is high. Though, it has to be noted that the OC dimensions PD does not exceed the threshold for the Cronbach’s alpha of 0.6 (0.504 is the actual α). The composite reliability of the dimension PD is 0.722, exceeding the threshold of 0.7.

(21)

21

Table 4 Reliability and validity results of the variables

* α=Cronbach's alpha, CR=Composite reliability, AVE=Average Variance Extracted, Perf=Manufacturing performance

Variable OC dimension variable α* CR* AVE* HRM 0.958 0.494 JIT&TQM 0.957 0.961 0.398 Perf 0.895 0.909 0.306 AS 0.695 0.814 0.525 HO 0.644 0.804 0.578 IC 0.717 0.817 0.475 IG 0.844 0.895 0.681 PD 0.504 0.722 0.402 PO 0.874 0.922 0.799

Table 5 Discriminant validity test based on Fornell and Lacker (1981) * Square root of the AVE (bold and on the diagonal)

Variable AS HO HRM IC IG JIT&TQM OC PD PO Perf

AS 0.725* HO -0.449 0.760* HRM -0.464 0.450 0.703* IC -0.621 0.529 0.480 0.689* IG -0.518 0.670 0.426 0.508 0.825* JIT&TQM -0.470 0.235 0.595 0.607 0.219 0.631* OC -0.835 0.700 0.582 0.848 0.747 0.588 X PD 0.532 -0.536 -0.385 -0.451 -0.567 -0.247 -0.683 0.634* PO -0.680 0.308 0.469 0.641 0.311 0.672 0.775 -0.390 0.894* Perf -0.297 0.289 0.582 0.369 0.365 0.362 0.416 -0.228 0.326 0.554*

For the formative construct, the outer model weight and significance are checked (table 6). The weights are all significant.

Table 6 Outer model weights and significance of formative construct OC

OC Dimension Weight T Statistic P value

PD -0.125 5.734 0.000 IC 0.282 10.796 0.000 IG 0.242 8.238 0.000 PO 0.278 9.729 0.000 AS -0.219 10.818 0.000 HO 0.137 6.917 0.000

(22)

22

All items belonging to the formative construct OC score below five on the collinearity statistic (VIF). Therefore, it can be concluded that there is a tolerable amount of multi-collinearity present (Hair et al., 2009).

On the whole, the results of the reliability and validity assessments show that most of the indicators are tolerably appropriate measures of the constructs. The construct validity, consisting of the convergent and discriminant validity is rather low, which should be kept in mind when interpreting the results of the study.

Data Analysis. The data analysis is performed in two steps. Hypotheses H1a, H2a and H3a are investigated by step one and hypotheses H1b, H2b and H3b are investigated by step two.

Step 1 Structural equation modeling using Partial Least-Squares (PLS)

In order to investigate hypotheses H1a, H2a and H3a, the regression-based model PLS modelling is used, which made it able to assess the impact of the independent variable on the dependent variable. PLS modelling has the advantage that it has minimal assumptions regarding data’s characteristics, e.g. that it is not required that the data is normal distributed (Reinartz, Haenlein, & Henseler, 2009). Additionally, PLS handles reflective as well formative constructs (Hair et al., 2009). Furthermore, PLS is less subtle to sample size considerations, making it able to still conduct analysis with rather small sample sizes (Hair et al., 2009). Since in the current study the data of most of the indicators is not distributed normally, both formative and reflective measures are involved, and the sample size is not that large, PLS modelling is appropriate for this study. Step 2: Multi-group analysis using Partial Least-Squares (PLS)

To make the moderating effect of HRM bundle implementation evident, the dataset had to be split in two groups: high HRM bundle implementation and low HRM bundle implementation. For each of these groups, the analysis in step one is performed. By comparing the results of these analyses, it becomes clear whether the relationships become stronger or weaker with high/low HRM bundle implementation. As a result, hypotheses H1b, H2b and H3b can be investigated.

(23)

23

Preacher, 2015) that are presented in the limitation section, the median-split is a frequently used method in Operations Management research (e.g. Bortolotti et al., 2015; Zhang, Linderman, & Schroeder, 2012). An advantage of the median-split technique is that all participants stay included in the analysis (Aziz et al., 2010). Furthermore, it is argued by Cohen (1983) and DeCoster, Iselin and Galucci (2009) that it makes analyses, especially for interaction effects, easier to perform and interpret, as it provides a clear overview of the impact of the groups by comparing the effects in the models between the groups. Also, many statistical tests require normal distributions of data. The proofs that Cohen (1983) and MacCallum et al. (2002) provide on their statements that dichotomization results in misleading statistical tests is based on the assumption that the dichotomized variable has a normal distribution (DeCoster, Iselin, & Gallucci, 2009). It might be, though, that in situations in which data is not normally distributed (as in the current research), “dichotomized indicators provide better representations of the underlying constructs” (DeCoster et al., 2009, p. 352) than the observed indicators in the Likert scale measures.

Table 7 Group characteristics for HRM moderation * M=mean, SD=standard deviation

M* SD* HRM low (<3.92, N=65) 3.372 0.440

HRM high (≥3.92, N=68) 4.165 0.258

3.2 Methodology qualitative phase in the research

Data content. In the second phase, the qualitative phase, the “why” behind the different relationships was collected. Specifically, information was required to obtain answers related to questions like “why can certain HRM practices compensate for flaws in the OC?” In this phase, the data content that was aimed to be collected were experiences related to the implementation of LM at plants. More specifically, the goal of the researcher was to collect information about incidents where choices were made whether or not to implement lean practices and how and why these choices differed per situation. Also, the researcher desired to collect information on how the OC of a plant impacted the choices of lean practices to implement and whether HRM practices can replace absence of certain OC bases at a plant. In Appendix III, the data content aimed to gather per hypothesis is presented (see Appendix II for the interview protocol).

(24)

24

better, since they allow the questions of why and how to be answered to gain more in-depth insights of a phenomenon in its natural setting (Karlsson, 2009). Interviews are used to collect information, as this qualitative method is appropriate to use to gain insights on the explanatory mechanism of a relationship (Eisenhardt, 1989). In order to collect data regarding questions like “why is the relationship between OC and JIT&TQM and manufacturing performance different or the same with varying levels of HRM bundle implementation?”, semi-structured interviews are conducted with five consultants, as this kind of interview allows the interviewees to explain their answers and enable certain thoughts to be questioned in depth (Humphrey & Lee, 2004).

Consultants are interviewed, because they have experience at multiple organizations and they take an independent angle. Specifically, five consultants specialized in LM implementation were interviewed, since they possess a lot experience in implementing LM in different organizations and are expected to be able to compare the different OCs at the different organizations and its implications for the implementation of LM.

The consultants were asked prior to the interview to think about two plants in which they are involved. The two plants had to be polar types in terms of OC in the beginning of the project. This kind of polar types sampling has the advantage “to more easily observe contrasting patterns in the data” (Eisenhardt & Graebner, 2007, p. 27). Theoretical sampling increases the likelihood that the interviews will offer theoretical insight (Eisenhardt & Graebner, 2007). In this way, the LM implementations at various plants were evaluated, with the advantage that these multiple cases enable comparisons and generate more generalizable outcomes (Eisenhardt & Graebner, 2007). To make the interviews more concrete, the consultants were asked to fill in a questionnaire prior to the interview with questions related to OC, JIT&TQM, and HRM practices.

Five consultants are interviewed, with in total nine plants (one plant was chosen by two consultants), as Eisenhardt (1989) argued that the number of cases between the four and ten often works well with regard to complexity and volume of the data and related to convincing empirical grounding. Table 8 presents the characteristics of the nine plants, whereby each plant is represented as a letter in order to ensure anonymity.

(25)

25

conducted whereby the consultant uses two cases illustratively. By using these polar type cases, consultants could provide more illustrations on how the different OCs of the plants impacted the decisions in the LM implementation and if and how they tried to compensate for flaws in the OC of plants.

The collected data was documented per interview, as the interviews were taped (saved as sound files) and subsequently transcribed (saved as Word files).

Table 8 Characteristics of the plants that are mentioned in the interviews

Attributes A B C D E F G H I

Industry Medical Food Food Food Stair lifts Food Food Food Food Size

(employees)

300/400 150 300 200 150 125 130 250 250

Time with lean (serious begin)

1 year 1 year 3 year 1 year 2 year 2.5 year _{5 year} 8 year 2.5 year

Consultant C1 & C4 C4 C2 C2 C5 C5 C3 C3 C1

Reliability & Validity. A case-study protocol and an interview protocol are used with explicit procedures, increasing standardization and in turn researcher reliability (Van Aken et al., 2012). Instrument reliability is increased by using multiple instruments: the interviews and the questionnaires prior to the interviews, providing triangulation (Van Aken et al., 2012). The researcher always has to be aware of the interviewee bias, especially with positive answers to questions (Van Aken et al., 2012). As consultants are independent and they describe an organization where they do not work themselves, this bias will be less present.

To check the extent to which conclusions regarding the relationships between the factors derived from the surveys are certifiable, it is important to assure that cause and effect identified are because of each other, instead of some other relationship (Karlsson, 2009). To increase this internal validity, the interviews with the qualitative data increase understanding of the relationships identified (Eisenhardt, 1989).

Analysis. Interview segments in the transcripts were coded by using Atlas.ti. Most of the coding is based on deductive codes.

(26)

26

researcher to observe patterns within and between the interviews. These patterns relate, among others, to OC dimensions that are frequently stated by interviewees to be compensated for and certain HRM practices that can compensate for the flaws in OC dimensions.

Next, per hypothesis, the insights per plant were analyzed and related to each other, complemented with general remarks of the interviewees, followed by an overview of the insights of the plants per hypothesis.

4. RESULTS

Section 4.1 presents the results of the quantitative phase, based on surveys, and section 4.2 shows the results of the qualitative phase, based on the interviews.

4.1 Results of the study’s quantitative phase

The results of the study’s quantitative phase are derived from the survey data provided by 133 manufacturing plants. The goal of this section is to show the relationships between OC, JIT&TQM implementation and manufacturing performance and the impact of HRM bundle implementation on these relationships.

Descriptive statistics. The means, standard deviations and correlations between the variables are presented in table 9.

Table 9 Descriptive statistics and correlations among the variables *=95% significance level, **=99% significance level

(27)

27

Outcomes of step 1: results of the structural model. Figure 3 graphically displays the empirically tested model. The intent is to research the relationship between OC and JIT&TQM, OC and manufacturing performance and JIT&TQM and manufacturing performance.

The results show a positive (β=0.310) and significant relationship of OC on manufacturing performance, which corresponds with hypothesis H1a that OC is positively related to manufacturing performance. Next to the direct impact of OC on manufacturing performance, it has also an indirect impact via JIT&TQM: β=0.127 (95% significant). Furthermore, the results present a strong (β=0.582) and significant relationship of OC on JIT&TQM bundle implementation, which is in line with hypothesis H2a. Lastly, JIT&TQM is as hypothesized (H3a) positively (β=0.219) and significantly impacting manufacturing performance, though the relationship is rather weak. It can be concluded that all the results of step 1 are in line with the hypotheses. The standard error, T-statistic and P-values can be observed in table 14 in Appendix IV.

Next to the direct effects that are shown above, also the indirect effects of the individual OC dimensions on JIT&TQM and manufacturing performance are notable (table 10). In table 10 is shown that PD and AS negatively (and significantly) impact JIT&TQM and manufacturing performance and that all the other dimensions positively (and significantly) influence the dependent variables.

Table 10 Indirect effects of OC dimensions on JIT&TQM and Manufacturing Performance

(28)

28

Figure 3 Structural Equation Modelling using PLS

W= Weight, β=Beta-coefficient, HO=Humane Orientation, AS=Assertiveness, PO=Performance Orientation, IG=In-Group Collectivism, IC=Institutional Collectivism, PD=Power Distance. The dashed paths indicate the weights of the OC dimensions. *=95% significance level, **=99% significance level

Outcomes of step 2: results of the multi-group analysis. In this step, the impact of the moderator HRM bundle implementation on the relationships in the structural model are examined. Figure 4 graphically displays the empirically tested model. The standard error, T-statistic and P-values can be observed in table 15 in Appendix IV.

It can be identified that the strength and the nature of the relationships of the variables differs between the two groups, with the remark that half of the relationships are insignificant.

(29)

29

It was hypothesized that the relationship between OC and JIT&TQM implementation would become weaker in case of high HRM bundle implementation. This is not in line with the results of the analysis, because the β increases (from β=0.687 to β=0.329), which shows that the relationship becomes stronger. It has to be noted, though, that the result regarding the relationship between OC and JIT&TQM for low HRM is insignificant. To conclude, hypothesis H2b is rejected by this analysis.

An important remark has to be made to the results presented above. As OC is a formative construct, the meaning of the construct can be different in the two situations: in the situation with low HRM and high HRM. The meaning of the formative construct is built by its weights, of which the differences are shown also in figure 4. It can be stated that the weight of the dimension PD increases largely in high HRM (W=-0.114) compared to low HRM (W=-0.196). Another big difference is for the dimension IG, which decreases largely in high HRM (W=0.233) compared to low HRM (W=0.307). Accordingly, the meaning of the construct OC differs, as the relative impact of the dimensions on the construct differ, and in other words, the degree to which a dimension is represented in the OC construct changes. As an example, the PD dimension, which negatively impacts JIT&TQM implementation and manufacturing performance, become more represented in the meaning of the construct OC. As the meaning of the construct is different in both situations, it is hard to compare the effects of this construct on JIT&TQM implementation and manufacturing performance. See table 16 in Appendix IV for a comprehensive overview of the weights of OC in the initial situation, high, and low HRM situation.

(30)

30

Figure 4 Structural Equation Modelling using PLS: comparison Low HRM bundle implementation and High HRM bundle implementation

(31)

31

Table 11 Comparison of impact of OC dimensions on JIT&TQM and manufacturing performance in initial situation, low HRM situation, and high HRM situation. *=90%, **=95%, ***=99%

ON JIT&TQM On Manufacturing Performance

OC Dimension β Initial β HRM Low β HRM High β Initial β HRM Low β HRM High

PD -0.073*** -0.064 -0.079* -0.055*** -0.087* -0.018 IC 0.164*** 0.086 0.199*** 0.123*** 0.117** 0.047 IG 0.141*** 0.101 0.160*** 0.106*** 0.137** 0.038 PO 0.162*** 0.096 0.182*** 0.122*** 0.130** 0.043 AS -0.127*** -0.074 -0.156*** -0.096*** -0.100 -0.037 HO 0.080*** 0.039 0.086*** 0.060*** 0.052** 0.020

When comparing the β values for the relationship of the OC dimensions on manufacturing performance, it can be stated that the impact of the OC dimensions on manufacturing performance becomes weaker when HRM bundle implementation is high, which is in line with H1b. Specifically, the negative impact of PD (from 0.087 to 0.018) and AS (from β=-0.100 to β= -0.037) on manufacturing performance becomes weaker negative, whereas the positive impact of the OC dimensions IC (from β=0.117 to β=0.047), IG (from β=0.137 to β=0.038), PO (from β=0.130 to β=0.043), and HO (from β=0.052 to β=0.020) become weaker positive in a situation with high HRM. It has to be noted that many of the relationships presented in table 11 are insignificant. As this study is of exploratory nature and this study aims to provide some preliminary insights which have to be tested at future research, it is allowable to look at the β values only (Van Aken et al., 2012). To conclude, H1b is rejected because the relationships are not all statistically significant, but cautiously can be agreed upon with the remark that many of the results are significant.

(32)

32

For the last hypothesis, it was hypothesized that the impact of JIT&TQM implementation on manufacturing performance would become stronger in case of high HRM bundle implementation. This is confirmed by the results of this study, showing that the β is higher (β=0.343 versus β=0.830) for high HRM compared to low HRM. It has to be noted that the impact of JIT&TQM on manufacturing performance for high HRM bundle implementation is insignificant, so hypothesis H3b is rejected. Though, the pattern of the result clearly shows that the impact becomes stronger which is in line with what the expectations.

4.2 Results of the study’s qualitative phase

The results of the study’s qualitative phase are derived from the interviews conducted with consultants regarding nine plants. The goal of this section is to explain and illustrate the relationships presented in the previous section and provide insights on why and how certain relationships take place.

Qualitative results related to step one of the quantitative analysis. The purpose of the interview outcomes in this step is to provide more insights on the role of OC in the LM implementation, by showing how OC impacts the LM implementation and its effectiveness. Table 12 shows a summary of explanations and examples derived from the interviews. Detailed examples from the interviews are provided in Appendix V.

OC -> Manufacturing performance. From the interviews it became clear that OC relates to the manufacturing performance of the plants, which corroborates the results of the quantitative analysis. The illustrations derived from the interviews provide insights on how and why certain OC dimensions impacted manufacturing performance (table 12).

OC -> JIT&TQM Implementation. From the interviews can be stated that OC relates to the JIT&TQM implementation at manufacturing plants, which is in accordance with the results of the quantitative analysis. The relation between OC and JIT&TQM implementation is in a way that it can be a precursor, an enabler, or a barrier (table 12).

(33)

33

JIT&TQM implementation reduces the non-value added time and in turn increases the OEE. The most notable outcomes from the interviews are summarized in table 12.

Qualitative results related to step two of the quantitative analysis.

(34)

34

Table 12 Overview of results qualitative analysis for hypotheses step 1 Plant B, F, and H are not shown because they provided no remarkable explanations Hypothesis Plant A Plant C Plant D Plant E Plant G Plant I General

H1a: OC -> Perf Dimension: Low IC Effect: negative Reason: people don't care if something goes wrong in the next shift

Dimension: High PD Effect: negative Reason: rebel against management Dimension: High PO Effect: positive Reason: increases willingness to change, speed of change, impact of change Dimension: Low IC Effect: negative

Reason: people don't care

if something goes wrong in the next shift

Dimension: High AS Effect: depends on whether

the sound of voice is positive or negative

Reason: if negative, then

hard to control behavior and negative impact on cooperation H2a: OC -> JIT&TQM Almost no impact of OC. At time of LM implementation, OC was not taken into account Enabler: Reactive OC enabled JIT&TQM implementation to increase awareness on importance prevention of problems

Barrier: Clear impact between OC and JIT&TQM

implementation, plant's OC does not fit with JIT&TQM implementation, which reduces the implementation of JIT&TQM Precursor: OC impacts the program that the plant implements, people-oriented culture requires a more people-oriented program, OC has no impact on the extent of LM implementation, however, it has impact on the speed of the implementation H3a: JIT&TQM -> Perf Intervention: Visualization, standardizing work, et cetera

Effect: OEE of the

plant

Intervention: Process

changes like changeovers, measuring efficiency

Effect: the effectiveness of

the plant

(35)

35

IC. C4 comments on how to compensate for low IC at a plant by changing the consequences of certain behavior with structures: “You have to change the structure in the plant. Apparently, there are consequences for the people to work as an individual or as a team. You have to look at the underlying structures, which cause that individualistic behavior. Accordingly, you change the structures. For example, in order to get more team effort, you can provide a reward to a team instead of only to individuals. In this way, the people are more encouraged to cope with activities as a team.”

At plant A and I, soft lean practices are used to compensate for low IC. As an example, C1 explains as follows how teamwork can compensate for low IC: “With this intervention, cross-functional teams are made and are put together to analyze and discuss problems and to make an approach to solve the problems. In this way you automatically force them to work as a team and you will receive the benefits. What also can be done is that when you notice that a team functions bad, you coach them and train them with team sessions, in order to change the behavior in the team and increase the team focus.” As can be noticed, especially from the last part, the team sessions are only performed when a team functions bad, so when IG is low, which points to the replacement effect.

IG. Plants can compensate for IG by HRM practices according to C1. C1 explains this as follows: “Imagine IG is low at an organization, but you are going to implement soft lean practices, then of course you can compensate, because you can involve the people a lot. As an example, we let people themselves have high input in the design of their responsibilities, we let them develop the structure, and it becomes more from them. Therefore I can imagine that it can compensate.” On the short term, this means that people are forced to provide input and are forced to take responsibility, thus at the short term it compensates for the lack of IG. On the long term, this behavior might become natural for the people.

(36)

36

steering on performance, held the operator responsible for his performance, so you discuss the performance with the operator and discuss with the operator why the machine functions good or bad. On the other hand, training and education is used to compensate: the operators are trained for autonomous management. The operators is provided with more autonomy over the machine and has given more responsibility for the machine and the performance of the machine. So, the responsibility is more and more on the floor, which compensates for the power distance.”

At plant E is compensated for high PD differently than is done at plant D. Where at plant D the work floor has more autonomy and less contact with upper layers and therefore the high PD not that negatively impacts the manufacturing performance anymore, at plant E the interaction between the different layers within the organization is increased. At plant E, for the negative consequences is compensated by getting more in touch with the people from the work floor, involve them more in the process. In this way the interaction increases and the negative effects

from high PD decreases.

That empowerment is argued to be able to compensate for high PD is agreed upon by C1. Furthermore, C1 stresses the importance of structures to compensate for high PD. C1 illustrates this as follows: “Performance behavior ensures that people have to take more responsibility per level and that this becomes a lot clearer in the organization. That the plant manager is not allowed to shout and come on the floor, because now there are systems and structures where people have to deal with professionally. In this way, the power distance disappears, because the organization is fixed with structures, and as a result power distance cannot have a negative effect anymore.” The importance of structures to compensate for OC flaws is stressed, as the negative consequences of OC flaws appear less when the organization and its people are forced to behave in a certain way. These structures refer to procedures, regulations, and clear appointments that have been made and that people have to comply with.

(37)

37

At plant H, HRM practices are used to compensate for the lack of PO, as C3 illustrates (summarized quote): “It is important that managers are being coached at the moment that they have contact with their subordinates. An example, a manager is having a conversation with a subordinate, we are standing next to them and train them on the job, and we help in which questions the manager asks. In this way, the management leadership of the management is developed by on the job training. Together with changing the structures together with the employees, more effectiveness is achieved and there can be compensated for the lack of performance orientation.”

That structures and training/coaching are important in compensating for low PO is in accordance with what is done at plant D to compensate for low PO. At plant D they are planning to compensate for PO by two soft practices. C2 explains these soft practices as follows: “The first one is developing a structure for performance steering, so on which goals you have to steer including the goals on work floor level. The second one is, which behavior you need to achieve that performance. Subsequently, you coach and training the people by for example feedback to develop their behavior. As an example, how do you give feedback as a manager?” Additionally, they focused on teamwork to compensate for the lack of PO, aim together, as a team, for increased manufacturing performance.

The importance of structures, next to empowerment and responsibility, to compensate for low PO is stressed by an example of plant A. According to C1, the people who were responsible for discussing the problems did not take their role sufficiently and did not have a performance orientation. As a consequence, a new structure was developed in which the team leader was given much more responsibility for the team and for the performance of the team. As the team doesn’t achieve the performance, the team leader has to explain why, which has to result that the team leader must be more in touch with the performance.

(38)

38

practices in the long term affect the organizational culture, or the other way around, that the organizational culture determines whether the soft lean practices stay effective?”

Replacement effect: OC -> JIT&TQM Implementation

Useful outcomes of the interviews regarding the replacement effect for the relationship between OC and JIT&TQM implementation are scarce. The results of the quantitative analysis showed that the HRM bundle implementation cannot mitigate the impact of OC on JIT&TQM implementation but instead increases the impact. Only one possible explanation is provided from the interviews. This explanation is illustrated in the text below.

C1 states that the absence of OC in the relationship to JIT&TQM implementation cannot be compensated by softer lean practices on its own. As C1 states: “I think it is logically to explain, because when you implement technical lean aspects, without dealing with the behavior and culture of the organization, then that culture determines how the plant uses the technical aspects. So, in this way you are way more dependent on how people uses the technical aspect, which is influenced by how they used to use them, which is, in turn, embedded in the organizational culture. You have to take account of this culture to change this.”

Replacement effect: JIT&TQM -> Manufacturing performance

Regarding the last hypothesis, the outcomes of the quantitative analysis showed that the relationship between JIT&TQM implementation and manufacturing performance becomes stronger with high HRM bundle implementation. This view is corroborated by the interviews. Below the most notable illustrations are presented, which shows why HRM bundle implementation strengthens the relationship between JIT&TQM implementation and manufacturing performance.

The replacement effect: compensating for organizational culture flaws with HRM practices in lean manufacturing implementation