Controlling service delivery in service triads: the impact of buyer’s HRM practices

(1)

Controlling service delivery in service triads: the impact of

buyer’s HRM practices

Master Thesis

MSc. Supply Chain Management

Supervisor:

prof. dr. D.P. van Donk

(2)

2

Abstract

In a service triad, the buying organization is heavily dependent on the interaction between supplier’s employee and the customer during the service delivery. Regarding ways to monitor behaviour of supplier’s employees, this thesis investigates the role of buyer’s HRM practices in a service triad. The effects of buyer’s HRM practices on supplier’s employee’s organizational identification and employee-customer identification are investigated. Therefore, survey data of 589 employees from four suppliers are analysed. Findings show that commitment-focused HRM practices positively affect employee-customer identification. This relationship is fully mediated by organizational identification with the buyer. Besides that, findings show that control-focused HRM practices positively affect employee-customer identification. This relationship is partially mediated by organizational identification with the buyer. These results give managers of the buying organization insights about preferred HRM practices with respect to supplier's employee's customer identification. This study recommends managers of the buying organization to mainly focus on commitment-focused HRM practices.

(3)

3

1. Introduction

Service outsourcing becomes more and more popular. The revenue of global service outsourcing increased from $45,6 billion dollars (U.S.) in 2000 to $88,9 billion dollars in 2015 (Feng, Ren & Zhan, 2019). Firms tend to focus more on performing what they can do the best, and outsource other tasks to outside providers. Outsourcing where the buyer (outsourcing organization) contracts a supplier who delivers services directly to the buyer’s customers, is called a service triad (Wynstra, Spring & Schoenherr, 2015). In a service triad, the performance of the buyer becomes heavily dependent on the service delivery of the supplier, because supplier’s employees directly interact with buyer’s customers (Uenk & Telgen, 2019). This arises the following question: how can the buyer efficiently monitor service performance if it is not involved in the actual service delivery, especially when this service delivery depends on the interaction between supplier’s employees and the customer? This paper focuses on means to control supplier’s employee’s service performance from the buyer’s perspective.

Existing literature mentions various ways to control service delivery in a service triad. Standardizing processes, separating low and high contact services, and the use of contracts and service level agreements increase buyer’s service delivery control in a service triad (Van der Valk & Iwaarden, 2011; Van Iwaarden & Van der Valk, 2013). Providing output-based incentives to suppliers is another way of controlling service delivery (Uenk & Telgen, 2019). However, the literature review of Wynstra, Spring & Schoenherr (2015) indicates that very few studies have dealt with OM and SCM topics such as service delivery control at the operational level in service triads. Service delivery at the operational level can be controlled by influencing employee’s behaviour and job performance (Van Iwaarden & Van der Valk, 2013). There is a need for a better understanding of how supplier’s employee’s behaviour can be influenced in a service triad context (Cappelli and Keller, 2013).

Concerning service triads, existing research is mainly focused on ‘hard’, objective ways to monitor behaviour (Van Iwaarden & Van der Valk, 2013). ‘Softer’ ways, such as management practices related to controlling employee behaviour need more attention (Uenk & Telgen, 2019; Capelli and Keller, 2013). An effective way to do this is by exercising HRM practices. HRM practices can contribute to service delivery control by motivating employees to adopt desired behaviours and attitudes (Bowen and Ostroff, 2004). However, it is unknown how this actually works when the buyer performs HRM practices to influence supplier’s employees’ behaviour in a service triad. This gap is the focus of this thesis. More specifically, this thesis investigates which types of the buyer's HRM practices are preferred when the goal is to achieve desired behaviour among supplier’s employees.

(5)

5 with their customers. Employee-customer identification (ECID) is the ‘extent to which employees self-identify with customers based on their interdependent role relationships’ (Anaza and Rutherford, 2012). According to Anaza and Rutherford (2012) and Anaza (2015), ECID is strongly influenced by employee’s organizational identification (OI): The extent to which the individual defines him or herself in terms of an organization’ (Mael and Ashforth, 1992). There is little knowledge about the OI of employees who are engaged in performing activities for one organization (the buyer) while being employed by another organization (George and Chattopadhyay, 2005). The issue of dual organizational identification is interesting: “How workers see themselves and which organization they identify with when they are employed by one but working on behalf of another is an important and unexamined question” (Cappelli & Keller, 2013, p. 591). This issue is relevant to this research.

This thesis looks into the role of buyer’s HRM practices to increase customer-oriented behaviour among supplier’s employees. Customer-oriented behaviour is investigated by focusing on ECID. The question is how buyer’s HRM practices affect supplier’s employee’s organizational identification with the buyer and eventually ECID. This thesis investigates the research gap from the supplier’s employee’s perspective; recent HRM approaches underpinned that HRM practices should be measured by the way employees interpret these HRM practices (Bowen and Ostroff, 2004; Nishii, Lepak & Schneider, 2008). The following research question will be investigated:

What are the effects of buyer’s HRM practices on supplier’s employee’s OI with buyer and ECID in a service triad?

A survey research method is used to investigate the research question. Questionnaires are distributed to 621 employees of four different suppliers, who all serve customers for the same buyer in a service triad context. The distribution resulted in a total of 589 returned questionnaires.

(6)

6 ECID contributed to the development of hypotheses in a service triad context. This thesis investigates whether certain relationships between these concepts also occur in a service triad.

Providing insights to these gaps in literature can also offer practical value. This research provides recommendations about which practices work well in relation to OI and ECID, so that supplier's employees are likely to exhibit more customer-oriented behaviour. Buyer’s managers get to know which HRM practices should be performed to achieve more customer-oriented behaviour among supplier’s employees. This should contribute to more service delivery control and eventually this should lead to better service performance and customer satisfaction in a service triad.

(7)

7

2. Theory and hypotheses

In this section, concepts related to the research question are explained by providing a theoretical background. The theoretical background starts with explaining buyer’s relatively low service delivery control in a service triad. At the same time, that part discusses why supplier’s employee’s identification with buyer’s customers is important when focusing on buyer’s service delivery control. After that, employee-customer identification (ECID) is further introduced and relationships with organizational identification (OI) and HRM attributions are explained. Initially, this is done by using existing literature that takes a traditional employment perspective (only one organization involved). After that, the horizon is expanded to the service triad context. This section ends with hypotheses that are related to the service triad context and a conceptual model (figure 1).

2.1 Service delivery in a service triad.

Service delivery depends on employee’s interaction with the customer. In a service triad, this interaction is directly provided by the supplier’s employees (Bastl, Johnson & Finne, 2019). Therefore, this can be seen as service outsourcing: ‘The conscious choice of replacing internal service functions with the use of external agents to perform one or more service activities’ (Li and Choi, 2009, p.28). Outsourcing of services in a service triad faces a paradox: on the one hand, supplier’s employees become very important for the buyer because of high dependency and on the other hand, those employees are still not considered as important as ‘core’ employees (Johnson & Ashforth, 2008). So, buyer’s management desires supplier’s employees to act in a customer-oriented way and at the same time management does not show that those employees are just as valuable as core employees. However, low service performance of supplier’s employees has a direct influence on service quality and customer satisfaction (Van Iwaarden & van der Valk, 2013). Whether supplier’s employees identify with buyer’s customers, is an essential aspect regarding employee’s customer-oriented behaviour (Anaza and Rutherford, 2012). Anaza and Rutherford (2012) argued that ECID may contribute to behaviours that are beneficial for customer satisfaction, service quality, and fulfilling customer needs. This thesis focuses on ECID: supplier’s employee’s identification with buyer’s customers. In the next section, ECID is further introduced.

2.2 The role of ECID and relationships with OI and HRM attributions

(8)

8 it helps the employee to form his/her own role-based identity as service provider. ECID may positively influence service quality, customer satisfaction and the fulfillment of customer needs (Anaza and Rutherford, 2012). Besides that, Anaza and Rutherford (2012) found that ECID positively relates to customer orientation and employee’s commitment. ECID occurs to a certain extent when employees adopt the defining role characteristics of the customer for themselves. ECID contributes to an environment in which service providers and customers are motivated to benefit from each other given their roles in the relationship. This should provide economic returns for the service provider and the customer both (Anaza & Rutherford, 2012).

2.2.1 Organizational identification

The concept of organizational identification (OI) involves a psychological state concerning that the individual perceives himself/herself to be part of the organization (Ashforth and Mael, 1989). Mael and Ashforth (1992) define OI as: ‘The extent to which the individual defines him or herself in terms of an organization’ (p.106). The stronger an employee identifies with an organization, the more likely it is that he or she is committed to organization’s interests (Van Knippenberg & Van Schie, 2000). Dutton, Dukerich and Harquail (1994) argued that stronger organizational identification positively affects the way employees look at the organization. Therefore, employees are more attracted to the organization. Strong identification with the organization means that employees are likely to focus on tasks that contribute to organizational interests instead of purely self-interested ones.

According to Sluss and Ashforth (2007), employees may develop deeper bonds with customers when they identify strongly with an organization. Identifying favourable organizational values and goals helps the employee to act in a customer-oriented way when these values and goals are in line with serving the needs of customers. Anaza and Rutherford (2012) found that OI is positively related to ECID. They argue that an increase of OI results in an increased desire to bond with customers. The next section discusses how organization’s HRM practices can contribute to employee’s organizational identification.

2.2.2 HRM attributions

(9)

9 (as part of HRM practices) is positively related to OI and that OI is positively related to oriented behaviour. As explained in the introduction, ECID is an important predictor of customer-oriented behaviour (Anaza & Rutherford, 2012). Recent HRM approaches underpinned that HRM practices should be measured by the way employees interpret these HRM practices (Bowen and Ostroff, 2004; Nishii et al., 2008). Bowen and Ostroff (2004) argued that the effect of HRM practices depends on the shared interpretation of the aim of HRM practices by employees. Nishii et al. (2008) indicate that employees’ attributions are based on the explanations they make regarding the underlying motivation for using HRM practices. These explanations may differ among employees. One employee may explain that a certain HRM practice is implemented with the aim of increasing the quality of processes, another employee may experience a different intention.

This research builds on the framework of Nishii et al. (2008) regarding the interpretation of HRM practices by employees. Whether employees interpret HRM practices as ‘commitment-focused’ or ‘control-focused’ affect their committed behaviour to organizations (Nishii et al., 2008). This has impact on individual outcomes such as commitment to the organization. This research looks into how supplier’s employees interpret buying organization’s HRM practices in a service triad and how this affects OI with the buyer and eventually ECID. HRM attributions perceived as ‘commitment-focused’ relate positively to affective commitment to the organization, HRM attributions perceived as ‘control-focused’ relate negatively to affective commitment to the organization (Nishii et al., 2008). Commitment-focused HRM attributes include a quality HRM strategy and considers employee well-being. With regard to the quality HRM strategy, employees are viewed as an important asset. Control-focused HRM attributes include a cost reduction HRM strategy and are focused on exploiting employees. With regard to the cost reduction HRM strategy, employees are viewed as a cost that has to be controlled. Exploiting employees means getting the most out of employees, without always taking their interests into account (Nishii et al., 2008).

2.3 OI with the buyer and buyer’s HRM practices in service triads

This part of the theory explains how organizational identification works when two organizations are involved, like in a service triad. By doing this, it becomes clear that supplier’s employee can identify with the buyer even though the employee is not contracted by the buyer. After that, it is explained why the buyer is likely to perform HRM practices aimed at supplier’s employees. This helps to get a better understanding of buyer’s role regarding OI and HRM practices before moving to the development of hypotheses.

2.3.1 OI with the buyer

(10)

10 Ashforth, Johnson, Hogg & Terry 2001). Other researchers have demonstrated that identification with two organizations is possible (Barker & Tomkins, 1994; Scott, 1997; Van Knippenberg & Van Schie, 2000). George and Chattopadhyay (2005) found that employees can identify with two organizations simultaneously and this is likely to happen when organizations show similarities. An employee will identify stronger with only one organization if he or she perceives two organizations to be very different. Vora and Kostova (2007) developed a theoretical model of the construct of dual organizational identification (DOI). Therefore, they used the term ‘relative magnitude’: “The relative strength of the sense of identification with each of the two entities” (p.331). From the context of multinational enterprises and subsidiaries, we know that an employee can have higher identification with the multinational enterprise compared to the subsidiary and the other way around (Vora and Kostova, 2007). Besides that, comparable levels of OI regarding the two organizations can exist. It can be suggested that this also applies in a service triad context.

George and Chattopadhyay (2005) and Vora and Kostova (2007) have shown that identification with two organizations certainly is possible. The way this works regarding supplier’s employee’s OI with the buyer in a service triad needs more attention. How does supplier’s employee’s OI with the buyer relate to ECID? And what is the influence of buyer’s HRM practices on employee’s organizational identification? The role of buyer’s HRM practices is discussed in the next part.

2.3.2 Buyer’s HRM practices

The involvement of two organizations has also consequences for the concept of HRM attributions. Employees are employed by the supplier, but perform activities for the buyer. The fact that supplier’s employees have to deal with two organizations makes understanding their organizational commitment more complex compared to regular employees. Studies have shown that they may be committed to both organizations (Liden, Wayne, Kraimer & Sparrowe, 2003; Leveson et al., 2010). Both organizations strive for affective commitment of employees to their organization. Therefore, a so-called dual HRM involvement is likely to exist. Employee’s commitment to the supplier is important because it is related to lower turnover, higher employee satisfaction and higher performance. Employees’ commitment to the buyer is important because their commitment is strongly related to desired behaviour and higher performance (Fontinha, Chambel & De Cuyper, 2012). This explains why the buyer is likely to exercise its own HRM practices. However, it is unknown how buyer’s HRM practices relate to employee’s OI with the buyer and how OI with the buyer impacts ECID. We focus on this gap in this thesis.

2.4 HRM attributions, OI and ECID in a service triad: hypotheses.

(11)

11 hypothesis is about OI in relation with ECID. It is expected that this relationship works in the same way in a service triad context compared to a traditional employment context. In both cases, it is about employee’s OI with the organization for which the employees serve customers. Hypotheses two to five are developed by using argumentations from literature where two organizations are involved.

2.4.1 Hypothesis 1

Chen, Chi and Friedmann (2013) looked into the dual organizational identification of sales representatives who are employed by supplier, but work at a department store (buyer in a service triad context). They found a positive relationship between sales representatives’ OI with the supplier and customer-oriented behaviour. Customer-oriented behaviour was even higher when sales representatives’ OI with the department store increased. So, they found that customer-orientated behaviour of the sales representatives was strongest in case of OI with the employing company and the department store both. Customer-oriented behaviour is not exactly the same as ECID. Although, ECID is an important predictor of customer-oriented behaviour (Anaza and Rutherford, 2012). This indicates the importance of supplier’s employee’s OI with the buyer. Based on this line of reasoning, we assume that OI with the buyer is important in relation to ECID. Therefore, we propose the following hypothesis regarding a service triad:

H1. Organizational identification with the buyer is positively related to ECID.

2.4.2 Hypotheses 2-5

According to Leveson et al. (2010), the key function of HRM practices is fostering perceptions of organizational support because perceptions of support are associated with employee’s commitment to the organization. They found that perceived organizational support is positively associated with commitment to the client organization when an employee is outsourced by an employment agency. This perceived organizational support can be increased by client organization’s HRM practices. In the field of temporary work, Liden et al. (2003) also found that perceived organizational support is associated with commitment to client organization. As explained above, HRM plays a critical role in fostering perceived organizational support.

(12)

12 employees positively relate to OI (Newman et al., 2016; Chih and Lin, 2019). That’s why we expect that buyer’s HRM practices initially impact employee’s OI with the buyer, before impacting other employee outcomes such as organizational commitment and desired behaviour. This thesis investigates whether buyer’s HRM practices indeed affect supplier’s employee’s OI with the buyer.

Regarding HRM practices, we use the framework of Nishii et al. (2008). ‘Commitment-focused’ HRM practices concerning employee well-being and reflecting a quality HRM address the concerns of employees. ‘Control-focused’ HRM practices concerning employee exploiting and reflecting a cost reduction connote low levels of concern for employees (Nishii et al., 2008). Nishii et al. (2008) found that HRMquality and HRMwell-being positively relate to employee’s satisfaction and commitment. We assume that these HRM practices initially affect OI with the buyer before affecting employee outcomes such as employee’s commitment and ECID. However, we expect the same effects of HRMquality and HRMwell-being on OI with the buyer. As explained before, we expect OI with the buyer to positively relate to ECID (H1). Therefore, we propose the following hypotheses in a service triad context:

H2. OI with the buyer fully mediates the relationship between HRMquality and ECID, such that HRMquality is positively related to OI which is, in turn, positively related to ECID.

H3. OI with the buyer fully mediates the relationship between HRMwell-being and ECID, such that HRMwell-being is positively related to OI which is, in turn, positively related to ECID. Nishii et al. (2008) found that HRMcost-reduction and HRMemployee-exploitation negatively relate to employee’s satisfaction and commitment. Again, we assume that these HRM practices initially affect OI with the buyer before affecting employee outcomes such as employee’s commitment and ECID. However, we expect the same effects of HRMcost-reduction and HRMemployee-exploitation on OI (with the buyer). Once again, this thesis uses ECID as a final employee outcome. Therefore, we propose the following hypotheses:

H4. OI with the buyer fully mediates the relationship between HRMcost-reduction ECID, such that HRMcost-reduction is negatively related to OI which is, on the contrary, positively related to ECID.

(13)

13

(14)

14

3. Methodology

The purpose of this study is to investigate the effects of buyer’s HRM involvement on ECID via employee’s OI with the buyer. To do so, this research uses a survey design as a research method. Survey design is most suitable because relationships within a service triad context are tested with well-defined constructs.

3.1 Research design

This research aims to investigate the impact of buyer’s HRM involvement by focusing on the relationships with employee’s OI with the buyer and ECID. This thesis uses a quantitative approach to identify what the effects are of buyer’s HRM practices on supplier’s employee’s OI and ECID. In the theoretical section, hypotheses are formulated. These hypotheses are tested by using a survey design as research method. Survey research is a preferred method when testing whether certain relationships also occur in contexts other than the context where these relationships initially were discovered (Karlsson, 2016). Regarding the development of hypotheses, findings from an employment agency context and a traditional employment context were used to come up with hypotheses. In this case, survey research examines how these findings relate to a service triad context. Survey research is most suitable for this research because it aims to test hypotheses that are based on well-defined theoretical concepts, models and propositions in relation to a specific context, in this case a service triad context (Karlsson, 2016).

3.2 Sample and data collection

(15)

15 Therefore, these employees are suitable for this research and match the description of the unit of analysis.

The questionnaire is distributed to the employees of the concerned organizations. This is done with the cooperation of managers of the suppliers. Firstly, the questionnaire was evaluated by managers of each organization to ensure that there are no irrelevant questions included in the questionnaire. Secondly, work teams within the organization are selected for the research. This is based on clustered random sampling. The work teams differ in terms of shifts. Together with the supervisors of the selected work teams, employees were taken to a room to explain the context of the research. The questionnaire was physically distributed to the employees. After that, the introduction to the questionnaire was read to the employees to make sure that they understood the context of the research. It was explicitly addressed that participation is voluntary. In terms of time, employees had one shift to complete the questionnaire. The questionnaires of the employees are collected from the supervisors.

The questionnaires are distributed to 621 employees in total, spread over four suppliers.

Customer service Contact center Cleaning service Aviation security Total Questionnaire distributed 70 16 250 285 621 Questionnaire returned 61 12 240 276 589 Response rate 87.1% 75% 96% 96.8% 94.8%

Table 1. Response rates employees.

3.3 Measures

The measurement items for investigating the constructs are based on evidence from existing literature. So, measurement items of previous studies are used. These measurement items are adapted when the original measurement items did not directly fit into the aviation industry context of this research. This research contains three main constructs: HRM attributions, employee’s OI with the buyer and ECID. As discussed before, ECID (dependent variable) is a relatively new construct. Anaza and Rutherford (2012) adapted items from organizational identification (Mael and Ashforth, 1992) and customer-company identification (Homburg, Wieseke & Hoyer, 2009). This resulted in five items that measure the concept of ECID and these items are also used in this research. For example, one of the questions that are asked to the employees is: ‘When someone praises my relationship with my customer, it feels like a personal compliment’. This item measures to what extent the employee of the supplier identifies with buyer’s customer. Items related to ECID are measured by using a 7-point Likert scale varying from 1 “strongly disagree” to 7 “strongly agree”.

(16)

16 involved (Chattopadhyay, 2005). In this thesis, we apply the measurement items of Mael and Ashforth (1992) for identification with the buyer. This means that the related questions are asked to find out to what degree the employee identifies with the buyer. The accompanying items, including factor loadings, are presented in table 2. For example, one of the questions that are asked to the employees is: ‘When I talk about the airport, I usually say ‘we’ rather than ‘they’. Items related to OI are measured by using a 7-point Likert scale varying from 1 “strongly disagree” to 7 “strongly agree”.

The third construct, HRM interpretation, is based on the framework of Nishii et al. (2008). According to the literature review of Wang, Kim, Rafferty and Sanders (2020), the framework of Nishii et al. (2008) is the most widely used measurement tool regarding employee’s HRM interpretation. It looks at the effects of HRM practices from the employee perspective by focusing on the employee’s interpretation of HRM practices. This research borrows items from the framework of Nishii et al. (2008). The items measure to what extent the buyer (in this case the airport) exercises HRM practices according to the employees. There are five main HRM practice areas. These HRM practice areas, originating from Nishii et al. (2008), are adapted to the specific context of the airport and the suppliers: scheduling, determining qualifications, daily training, performance monitoring and performance incentives (Nishii et al., 2008). To investigate how the employees interpret the airport’s involvement in those five HRM practice areas, the same four questions are asked for every HRM practice area. These questions were to asked the employees to indicate the degree to which the HRM practice was implemented to (1) enhance service quality; (2) reduce costs, (3) promote employee well-being and (4) get the most work out of employees (Nishii et al., 2008). The items are measured using a five-point Likert scale varying from 1 “strongly disagree” to 5 “strongly agree”. In consultation with managers from the cleaning supplier, it was decided to exclude the first two HRM practice areas (scheduling & determining qualifications) from the questionnaire for this supplier specifically. They indicated that the airport was not involved in these two practice areas.

3.4 Data reliability, validity and common method bias

Firstly, we detected outliers in the observations by using a rule of thumb for labelling outliers. In terms of the lower quartile (FL) and the upper quartile (FU), an observation is labelled as an outlier if it falls outside (Hoaglin & Iglewicz, 1987):

FL – 2.2(FU - FL) and FU + 2.2(FU – FL)

According to Hoaglin and Iglewicz (1987), it is appropriate to use 2.2 as a multiplier. With 1.5 as multiplier (Tukey, 1977), nearly 25% of the observations can be labelled as outliers while not all these observations are outliers in fact. Therefore, we used k = 2.2 to detect and remove outliers.

(17)

17 sample 2. We made this distinction because of the differences in airport’s involvement in HRM practice areas: regarding the cleaning service, the airport is not involved in ‘scheduling’ and ‘determining qualifications’. This influences the exploratory factor analysis, for the cleaning supplier there are fewer items that participate in the exploratory factor analysis. Therefore, we conducted a separate exploratory factor analysis for the cleaning supplier. These exploratory factor analyses are conducted to assess the unidimensionality of the constructs. We looked into convergent and discriminant validity of the items that belong to HRM attributions and OI buyer together and assessed the convergent validity of ECID separately. Rotation varimax is used. Regarding the first sample (supplier 1, 2 &3), the exploratory factor analysis resulted in the removal of four items (table 2, second column): three items that initially belonged to the HRM attributions and one item that initially belonged to OIbuyer. Regarding the second sample (supplier 4), the exploratory factor analysis resulted in the removal of five items (table 2, third column): four items that initially belonged to the HRM performance incentives area and one item that initially belonged to OIbuyer.

For the first sample, the eigenvalues-greater-than-one rule suggested that there are five constructs in addition to the dependent variable. These five extracted constructs explain 68.33% of the total variance. The lowest factor loading of an item on a construct is .652. The highest cross-loading of an item on a different construct is .355. We used the criterium of .40 as maximum for evaluating discriminant validity and .60 as minimum for evaluating convergent validity. Regarding employee-customer identification, the lowest factor loading is .833.

For the second sample, the eigenvalues-greater-than-one rule suggested that there are three constructs in addition to the dependent variable. Items that theoretically belong to HRMquality and HRMwell-being load on one construct. From now we call this construct HRMcommitment. Items that theoretically belong to HRMcost-reduction and HRMemployee-exploitation also load on one construct. From now we call this construct HRMcontrol. These three extracted constructs explain 58.45% of the total variance. The lowest factor loading of an item on a construct is .631. The highest cross-loading of an item on a different construct is .381. Regarding employee-customer identification, the lowest factor loading is .786. These loadings fit the criteria mentioned above, therefore the validity of constructs is acceptable and this research continues using these constructs.

(18)

18

Sample 1 Sample 2

Measurement item Factor loadings Factor loadings

HRMquality α = .82

HRMcommitment α = .78

The airport involves in determining the number of [supplier name] personnel in service that it does …

The airport involves in determining the qualifications of [supplier name] employees that it does …

The airport provides [supplier name] employees the daily refreshment training that it does…

The airport involves in the performance monitoring of [supplier name] employees that it does…

The airport gives [supplier name] employees the performance incentive that it does…

… in order to in order to help [supplier name]’s employees deliver quality service to customers.

.652 .726 .780 .706 Deleted - - .715 .687 Deleted HRMwell-being α = .80

The airport involves in determining the number of [supplier name] personnel in service that it does …

… so that employees will feel valued and respected – to promote employee well-being. .717 .742 Deleted .720 .706 - - .793 .799 Deleted HRMcost-reduction α = .88 HRMcontrol α = .82 The airport involves in determining the number of [supplier name]

personnel in service that it does …

…to try to keep costs down.

(19)

19

HRMemployee-exploitation α = .84 The airport involves in determining the number of [supplier name]

personnel in service that it does …

… in order to get the most work out of [supplier name]’s employees. .832 .747 .698 .697 Deleted - - .779 .844 Deleted OIbuyer α = .92 OIbuyer α = .86 When someone criticizes the airport, it feels like a personal insult.

I am interested in what others think about the airport.

When I talk about the airport, I usually say ‘we’ rather than ‘they’. The airport’s successes are my successes.

When someone praises the airport, it feels like a personal compliment.

If a story in the media criticized the airport, I would feel embarrassed.

Working for the airport reflects my personality well.

Sometimes I would rather not say that I work for the airport (R). I am actively involved in activities organized by the airport.

.756 .818 .776 .870 .870 .860 .858 Deleted .725 .674 .631 .632 .808 .810 .752 .751 Deleted .641 ECID α = .92 ECID α = .86 When someone praises my relationship with my customer, it feels

like a personal compliment. I feel attached to my customers.

I am interested in what others including my co-workers think about my customer.

I identify with my customers.

I feel good to be of service to my customers.

.869 .868 .880 .833 .885 .786 .800 .834 .792 .825

Table 2. Measurement items.

Table 3 and table 4 represent the descriptive statistics and the correlations between the variables of sample 1 and 2. Correlations between variables show discriminant validity because these correlations are relatively low. The highest correlation between two variables is .461.

N Mean Std. dev. 1 2 3 4 5 6 1 HRMquality 330 4.22 .47 1

(20)

20

3 HRMcost-reduction 349 3.11 .99 .123* .092 1

4 HRMemp.exploitation 346 3.66 .88 .237** .190** .461** 1

5 OIbuyer 316 5.03 1.05 .350** .407** .116* .270** 1

6 ECID 322 5.73 .88 .272** .153** .203** .273** .376** 1

Table 3. Descriptive statistics and correlations among variables (sample 1). Notes: N = sample size. mean = sample mean. Std. dev. = standard deviation. * p < .05, ** p < .01

N Mean Std. dev. 1 2 3 4 1 HRMcommitment 236 4.37 .50 1

2 HRMcontrol 238 3.31 .93 .302** 1

3 OIbuyer 234 5.47 .90 .351** .157* 1

4 ECID 228 5.77 .81 .109 .040 .479** 1

Table 4. Descriptive statistics and correlations among variables (sample 2). Notes: N = sample size. mean = sample mean. Std. dev. = standard deviation. * p < .05, ** p < .01

3.5 Data analysis

Firstly, we performed a multilevel analysis to investigate whether differences between suppliers influence the dependent variable ECID. As explained earlier, data is collected from 589 employees (level 1) who belong to four different suppliers (level 2). To ensure that supplier (level 2) does not or minimally affect the dependent level 1 variable, ECID, a multilevel analysis with the effect of ‘supplier’ on ECID is performed. We estimated a null-model (only includes the dependent variable ECID) with no predictor at level 1 or level 2. The estimated regression coefficient of the intercept is 5.8520 and is significant at a 95% confidence level. The variance partition coefficient (VPC) = .0573, this indicates that 5.73% of the variance in ECID can be attributed to differences between suppliers. The effect of supplier is significant at the 5% point of a chi-squared distribution on one degree of freedom. However, it was decided in consultation that this influence is excluded from further analyses because of (1) the relatively small VPC, (2) the absence of level 2 variables and (3) the boundaries of a thesis research in terms of time and complexity.

Secondly, we performed a simple regression to test H1. This regression should show whether employee’s OI with the buyer positively relates to ECID. Data from both samples are used together during this analysis because there are no differences between the two samples in terms of how OI and ECID are measured.

(21)

21 effect with the total effect, without even analysing the indirect effect (a * b) itself. With the causal steps approach, the mediation stops when certain criteria are not been met. However, theoretically, there can still be a significant indirect effect when the total effect is not significant or when the separate effects and are not significant (Hayes, 2017).

Taking the advantages of the bootstrap confidence interval approach and the limitations of the causal steps approach into account, this thesis uses Hayes’ PROCESS macro (v3.4) in SPSS that follows the bootstrap confidence approach. We have used the 95% bootstrap confidence interval with the standard 5,000 bootstrap samples for estimating the mediation effects. The indirect effect is significant when the confidence interval does not include zero. There is a significant positive indirect effect when the entire interval is higher than zero and there is a significant negative indirect effect when the entire interval is lower than zero. This bootstrap confidence interval approach is applied four times for sample 1 and two times for sample 2 in order to get results that give answers to H2-H5.

3.6 Control variable ‘gender’

(22)

22

4. Results

This section starts with the results of a simple regression. These results give insights into hypotheses 1. After that, the results of mediation analyses regarding both samples are reported. Overviews of the results that are obtained by performing mediation with the confidence bootstrap interval approach are presented in table 5.

4.1 Hypothesis 1. OIbuyer is positively related to ECID

In order to analyse whether OI with the buyer has influence on ECID, a regression of OIbuyer on ECID is performed. This regression is significant, R2_{= .132, F(2,518) = 39.424, p <.001. OI positively relates} to ECID, B = .327, t = 8,821, p <.001. The influence of control variable ‘Gender’ is not significant, B = .006, t = .079, p = .94. H1 is supported.

4.2 Hypotheses 2-5

Table 5 represents the mediation results of the bootstrap confidence interval approach. The last three columns represent the results of the indirect effect (ab). Furthermore, the effects of the HRM attributions on OI with the buyer (a), the effect of OIbuyer on ECID when the HRM attribution and OIbuyer both predict ECID (b), the total effects of the HRM attributions on ECID (c) and the direct effects of the HRM attributions on ECID (c’) are pictured in table 5.

Point of estimate Indirect effect

(23)

23 Sample 2 HRMcommitment - Gender b R2 .3038*** .2462* .1231 .3594*** .1102ns .1928 .0701ns .1987ns .0252 -.0391ns _.1092 _.0588 _.1741 HRMcontrol - Gender b R2 .1313* .3212* .0392 .3724*** .1154ns .2183 .0375ns .2351* .0221 -.0114ns _.0489ns_{-.0009 .1038}

Table 5.Indirect effects of HRM attributions (X) on employee-customer identification through organizational identification with the buyer. Note: * p < .05, ** p < .01, *** p < .001.

4.2.1 H2. OIbuyer fully mediates the relationship between HRMquality (+) and ECID.

Sample 1

According to hypothesis two, HRMquality positively influences OI with the buyer and OI with the buyer positively influences ECID. OI with the buyer should fully mediate the relationship between HRMquality and ECID. Results show indeed that HRMquality positively relates to OI with the buyer (t(280) = 6.556, p <.001) and that OI with the buyer positively relates to ECID (t(279) = 6.185, p <.001). While the total effect of HRMquality on ECID is significant (t(280) = 3.966, p <.001), the direct effect of HRMquality on ECID is not (t(279) = 1.674, p = .10). The results on the indirect effect suggest that OI with the buyer fully mediates the relationship between HRMquality and ECID. The indirect effect of HRMquality on ECID through OI with the buyer is .1158 at a 95% confidence interval. Hypothesis two is supported.

4.2.2 H3. OIbuyer fully mediates the relationship between HRMwell-being (+) and ECID.

Sample 1

According to hypothesis three, HRMwell-being positively influences OI with the buyer and OI with the buyer positively influences ECID. OI with the buyer should fully mediate the relationship between HRMquality and ECID. Results show indeed that HRMwell-being positively relates to OI with the buyer (t(291) = 7.815, p <.001) and OI with the buyer positively relates to ECID (t(290) = 6.111, p <.001). While the total effect of HRMwell-being on ECID is significant (t(291) = 2.656, p <.01), the direct effect of HRMwell-being on ECID is not (t(290) = .016, p = .99). The results on the indirect effect suggest that OI with the buyer fully mediates the relationship between HRMwell-being and ECID. The indirect effect of HRMquality on ECID through OI with the buyer is .1342 at a 95% confidence interval. Hypothesis three is supported.

Sample 2

(24)

24 terms of OI with the buyer. Besides that, OI with the buyer is positively related to ECID (t(211) = 6.62,

p <.001). The total effect of HRMcommitment on ECID is not significant (t(212) = 1.339, p = .18) and

the direct effect HRM commitment on ECID is also not significant (t(211) = -.773, p = .44). However, results on the indirect effect suggest that OIbuyer fully mediates the relationship between HRMcommitment and ECID, because the indirect effect is significant (last three columns table 5). The indirect effect is .1092 at a 95% confidence interval. This supports hypotheses two and three.

4.2.3 H4. OIbuyer fully mediates the relationship between HRMcost-reduction (-) and ECID.

Sample 1

According to hypothesis four, HRMcost-reduction negatively influences OI with the buyer and OI with the buyer positively influences ECID. OI with the buyer should fully mediate the relationship between HRMcost-reduction and ECID. Results show that HRMcost-reduction positively relates to OI with the buyer (t(295) = 2.100, p <.05) and OI with the buyer positively relates to ECID (t(294) = 6.293, p <.001). The total effect of HRMcost-reduction on ECID is significant (t(295) = 3.339, p =.001), the direct effect of HRMcost-reduction on ECID is also significant (t(294) = 2.760, p <.01). The results of the indirect effect suggest that OIbuyer partially mediates the relationship between HRMcost-reduction and ECID. Results show a partial mediation because the direct effect of HRMcost-reduction is still significant and is even bigger than the indirect effect (see the direct and indirect effect in table 5). Besides that, the results show that HRMcost-reduction is positively related to OI with the buyer while a negative effect between these two constructs was hypothesized. So, hypothesis four is rejected.

4.2.4 H5. OIbuyer fully mediates the relationship between HRMemployee-exploitation (-) and ECID.

Sample 1

According to hypothesis five, HRMemployee-exploitation negatively influences OI with the buyer and OI with the buyer positively influences ECID. OI with the buyer should fully mediate the relationship between HRMemployee-exploitation and ECID. HRMemployee-exploitation positively relates to OI with the buyer (t(293) = 4.735, p <.001) and OI with the buyer positively relates to ECID (t(292) = 5.358, p <.001). The total effect of HRMemployee-exploitation on ECID is significant (t(293) = 4.649,

p <.001), the direct effect of HRMemployee-exploitation on ECID is also significant (t(292) = 3.259, p

<.01). The results of the indirect effect suggest that OIbuyer partially mediates the relationship between HRMemployee-exploitation and ECID. Results show a partial mediation because the direct effect of HRMemployee-exploitation is still significant and is even bigger than the indirect effect (see the direct and indirect effect in table 5). Besides that, the results show that HRMemployee-exploitation is positively related to OI with the buyer while a negative effect between these two constructs was hypothesized. So, hypothesis five is rejected.

(25)

25 As explained in the methodology, items that initially belonged to HRMcost-reduction and HRMemployee-exploitation are assigned to one construct: HRMcontrol. Therefore, outcomes of the mediation analysis provide insights to both hypotheses (H4 & H5). HRMcontrol is positively related to OI with the buyer (t(213) = 1.971, p =.05) and at the same time, the control variable ‘Gender’ has also a positive significant effect on OI (t(213) = 2.456, p <.05). This suggests that female employees score higher in terms of OI with the buyer. Besides that, OI with the buyer is positively related to ECID (t(212) = 7.295, p <.001). The total effect of HRMcontrol on ECID is not significant (t(213) = .677, p = .68), but the effect of the control variable ‘Gender’ on ECID is significant (t(213) = 2.161, p <.05). This suggests that female employees score higher in terms of ECID. The direct effect of HRMcontrol on ECID is also not significant (t(212) = -.228, p = .82) and the indirect effect of HRMcontrol on ECID is not significant either (last three columns table 5). Therefore, there is no support for hypotheses four and five.

4.3 Post-hoc analysis

Outcomes of the mediation analyses suggest that the control variable ‘gender’ influences OIbuyer and ECID within sample two. Therefore, we performed a post-hoc analysis that looks further into these influences. The descriptive statistics, presented in table 6, show indeed that female score higher on OIbuyer and ECID compared to male. Besides that, female score lower on HRMcontrol compared to male.

Male Female Independent samples t-test N Mean Std. dev N Mean Std. dev t df p HRMcommitment 120 4.35 .41 104 4.37 .58 -.242 222 .809 HRMcontrol 122 3.40 .93 104 3.19 .93 1.714 224 .088 OIbuyer 120 5.35 .92 103 5.62 .83 -2.275 221 .024*

ECID 118 5.66 .88 101 5.87 .67 -1.936 217 .054

Table 6. Descriptive statistics and independent samples t-test. Notes: N = sample size. mean = sample mean. Std .dev. = standard deviation. * p < .05, ** p < .01, *** p < .001.

An independent samples t-test is suitable for testing the significance of differences in means when comparing two groups (Park, 2009). In this case, an independent samples t-test is performed to test whether the means of HRMcommitment, HRMcontrol, OIbuyer and ECIDResults significantly differ between male and female. Results of the independent samples t-test (table 6) show that only the influence of gender on OIbuyer is significant. So, female score significantly higher on OI compared to male.

(26)

26

Sample 2 Point of estimate Indirect effect

Bootstrapping 95% CI X Effect of X on OIbuyer (a) Unique effect of OIbuyer on ECID (b) Total effect (c) Direct effect (c’) Indirect effect (ab) Lower Upper Male HRMcommitment b R2 .3299** .0813 .4055*** .2250 .1932* .0350 .0594ns _.1338 _.0422 _.2568 HRMcontrol b R2 .1377ns .0183 .4665*** .2699 -.0012ns .0000 -.0655ns _.0642ns _{-.0179 .1651} Female HRMcommitment b R2 .2893*** .1383 .2942*** .1402 .0019ns .0000 -.0832ns _.0851 _.0330 _.1559 HRMcontrol b R2 .1239ns .0175 .2473*** .1273 .0813ns .0142 .0506ns _.0306ns_{-.0188 .0847}

Table 7. Indirect effects of HRM attributions (X) on employee-customer identification through organizational identification with the buyer. Sample 2: Male vs. female. Note: * p < .05, ** p < .01, *** p < .001.

(27)

27

5. Discussion

Building on existing literature about HRM mechanisms as means to influence employees’ service delivery, this paper discusses how buyer’s HRM attributions are interpreted by supplier’s employees in a service triad, and how this affects their OI with the buyer and ECID. Therefore, we distinguished four types of HRM attributions (Nishii et al., 2008): HRMquality, HRMwell-being, HRMcost-reduction and HRMemployee-exploitation. Table 6 gives an overview of the hypotheses and the results.

Hypotheses Results H1 HRMquality (+)→ OIbuyer (+)→ ECID Supported H2 HRMwell-being (+)→ OIbuyer (+)→ ECID Supported

H3 HRMcost.red. (-)→ OIbuyer (+)→ ECID Rejected HRMcost.red. (+)→ OIbuyer. Partial mediation H4 HRMemp.exp. (-)→ OIbuyer (+)→ ECID Rejected HRMemp.exp. (+)→ OIbuyer. Partial mediation

Table 8. Overview of hypotheses testing.

5.1 Theoretical implications

Building on the ECID concept, introduced by Anaza & Rutherford (2012), we argued that organizational identification with the buyer positively relates to ECID. Anaza & Rutherford’s (2012) study showed that OI positively relates to ECID in a traditional employment context where only one organization is involved. In this traditional employment context, the employee is serving customers for the company where he/she contracted by. In a service triad context, the employee is serving customers for the buyer, but he or she is contracted by the supplier. The results of this thesis suggest that this difference doesn’t seem to matter in terms of OIbuyer affecting ECID. This study is particularly helpful in extending the understanding of employee’s identification with an organization and employee’s identification with the customer in a service triad context. Results regarding H1-H5 show that OI with the buyer positively relates to ECID. The results of this thesis suggest that identification with the company for which employees serve customers in a service triad positively relates to employee identification with these customers.

5.1.1 Buyer’s HRM attributions and the effects on supplier’s employee’s OI with the buyer.

(28)

28 positively relate to OI with the buyer. This suggests that buyer’s involvement in commitment-focused HRM practices indeed positively affects supplier’s employees’ OI with the buyer. Findings regarding sample 2 show the same. Surprisingly, results regarding HRMcost-reduction and HRMemployee-exploitation show that these two HRM attributions also positively relate to OI with the buyer. Besides that, HRMcontrol was also positively related to OI in sample 2. This suggests that all four HRM attributions positively relate to OI with the buyer. However, there is a small side note that has to be made: the positive effects of HRMcost-reduction and HRMemployee-exploitation on OI are considerably smaller (.1259 and .2728) than the expected positive effects of HRMquality and HRMwell-being (.3669 and .4222). The positive effect of HRMcontrol on OI is also smaller (.1313) than the positive effect of HRMcommitment (.3038). But still, these results are surprising and need to be discussed. From traditional employment literature, we know that commitment-focused HRM attributions positively relate to OI (Newman et al., 2016; Chih and Lin, 2019) and employee’s commitment to an organization (Nishii et al., 2008). Control-focused HRM attributions negatively relate to employee’s commitment to an organization (Nishii et al., 2008). However, results of this thesis suggest that control-focused HRM attributions positively relate to OI with the buyer in a service triad. This seems to contradict findings from existing traditional employment literature.

So, how is it possible that control-focused HRM attributions positively affect OI with the buyer? According to the literature review of Wang et al. (2020), there is a need for developing a more nuanced understanding of cultural differences among countries and the impact of these differences on HRM attributions. They argue for instance that HRMemployee-exploitation negatively affects employee and organizational outcomes in the US (Nishii et al., 2008), while HRMemployee-exploitation positively affects employee and organizational outcomes in China, the Netherlands and the United Kingdom (Sanders, Jorgensen, Shipton, Van Rossenberg, Cunha, Li, Rodrigues, Wong & Dysvik, 2018). Employees can have different perceptions of HRM practices. Even within the same workgroup, employees can have different understandings of why practices are introduced in an organization (Wang et al., 2020). It makes sense that these differences are even bigger across countries. As mentioned before, data is collected from the aviation industry in Indonesia. Evidently, this thesis suggests that control-focused HRM attributions positively affect employee’s OI with the buyer within this aviation industry in Indonesia. This contradicts findings of Nishii et al. (2008) that are based on gathered data in the US. So, this thesis emphasizes the need for developing a more nuanced understanding of HRM attributions across different national contexts. Different understandings of HRM attributions can arise from various factors, such as cultural differences in job security policies, power distance, employment regulations, and national government regulations (Wang et al., 2020).

(29)

29 countries and industries. They argue that researchers cannot assume that these measurements capture the same kinds of employee perceptions across different countries. For example, with the measurement item of HRMemployee-exploitation: ‘…in order to get the most work out of employees’, it cannot be assumed that this will always be negatively interpreted by employees in different national contexts. Meanings of measurement items may differ across countries because of different regulatory mechanisms (Wang et al., 2020). Regulatory mechanisms are different across countries because of different labour laws, government regulations and sector-wide agreements (Paauwe & Boselie, 2003). This influences how HRM practices are performed at the organizational level and influences the understanding of HRM attributions among employees (Wang et al., 2020). Referring back to the example mentioned above, employee’s interpretation of HRMemployee-exploitation relates positively to employee outcomes in the Netherlands, while it negatively relates to employee outcomes in the US. Different meanings of measurement items across countries could explain the fact that control-focused HRM contributions positively relate to OI with the buyer in this thesis. These HRM attributions may have a positive meaning for employees of the aviation industry in Indonesia.

5.1.2 The indirect effects of HRM attributions on ECID: OI as mediator (H2-H5).

The next theoretical implication is about the role of OI as mediator in the relationship between HRM attributions and ECID. Hypotheses 2 to 5 all assumed indirect effects of the HRM attributions on ECID, with OI as mediator. However, results show that the role of OI as mediator is not as straightforward as expected. Hypotheses 2 and 3 are supported: OI with the buyer fully mediates the relationship between HRMquality and ECID, and HRMwell-being and ECID. This implies that the buyer should apply HRM practices focused on quality and employee well-being in order to increase employee’s OI with the buyer and eventually ECID. These findings fit with lines of reasoning from existing literature regarding a traditional employment context. So, results of this thesis suggest that these mechanisms also apply in a service triad context.

(30)

30 there is no existing literature that explains the directs effects of HRM attributes on employee’s identification with customers or other close related employee outcomes (in a service context).

As mentioned in the introduction, there are various ways to control service delivery in a service triad. Existing literature is mainly focused on ‘hard’ ways to control service delivery, such as standardizing process, using contracts, service level agreements and providing output based incentives to suppliers (Van der Valk & Iwaarden, 2011; Van der Valk & Iwaarden, 2013; Uenk & Telgen, 2019). Regarding softer ways to control service delivery in a service triad, the results of this thesis show that buyer’s HRM practices can also effectively contribute to service delivery control. More specifically, results of this thesis suggest that both commitment-focused and control-focused HRM practices positively influence ECID. The effects of control-focused HRM practices on ECID are largely direct effects and the effects of commitment-focused HRM practices on ECID are mediated by OI with the buyer. Therefore, these results contribute to existing literature by giving insights about which of buyer’s HRM practices effectively affect supplier’s employee’s outcome ECID.

5.1.3 The impact of ‘gender’ on OI with the buyer

Results of the mediation analyses suggested that female employees score higher on OI with the buyer and ECID within sample 2. This sample consists of employees from the ‘cleaning’ supplier. For female employees, descriptive statistics showed indeed higher means regarding OI with the buyer and ECID. Although, the independent samples t-test showed only a significant influence of gender on OI with the buyer. This shows that female significantly score higher on OI with the buyer compared to male within sample 2. Organizational identification, job satisfaction and turnover intention may be influenced by gender because female may experience less flexibility, in terms of finding another suitable job, as consequence of more non-work commitments (De Moura, Abrams, Retter, Gunnarsdottir & Ando, 2009). To our knowledge, men are responsible for most of the income within a household in Indonesia. Everything that women earn besides the income of men could be seen as extra income in many cases. Therefore female employees from the cleaning supplier may be more easily satisfied and identify more with the buyer compared to male employees.

(31)

31 with the buyer has bigger impact on ECID for male compared to female. It is unclear what possible explanations are for this difference in effect size.

5.2 Managerial implications

The results of this thesis have some practical implications for managers of the buying organization in a service triad. Firstly, these managers must be aware of the fact that supplier’s employees identify to a certain extent with the buyer. The extent to which supplier’s employees identify with the buyer impacts their identification with buyer’s customers and possibly also other favourable employee performance outcomes.

Furthermore, this thesis showed that buyer’s HRM practices can affect supplier’s employees OI with the buyer. Managers of the buyer should take into account that HRM practices, interpreted as commitment-focused HRM attributions, are most effective in relationship with OI. Control-focused HRM attributions, especially HRMemployee-exploitation, can also positively affect employees OI with the buyer. Although, the effects of these HRM are attributions smaller. Besides that, the indirect effects of commitment-focused HRM attributions on ECID are larger compared to the indirect effects of control-focused HRM attributions. So, if manager’s goal is to influence employee’s ECID via OI, it is strongly recommended to focus on HRMquality and HRMwell-being practices. This thesis suggested that control-focused HRM attributions can have a direct positive impact on employee’s ECID. However, findings regarding the positive influences of control-focused HRM attributions on OI with the buyer and ECID contradict other findings from other existing literature Nishii et al. (2008). It is recommended to focus on HRMcommitment practices if managers want to be sure of HRM practices that do have positive consequences. But what can managers do to maximize the likelihood that an employee will actually interpret HRM practices as intended HRM practices?

(32)

32

6. Limitations & future research

Along with the findings of this thesis, there are a few limitations that should be considered. The sample was limited to one type of industry: the aviation industry. Generalizability of findings to other industries may be limited, because these other industries may have different characteristics that possibly influence the way in which HRM practices are performed and interpreted. Within the aviation industry, the service triad context included one buyer (the airport) and four different suppliers in this thesis. It can be questioned whether using one organization as buying organization is sufficient in terms of generalizability. But this thesis certainly gives some interesting insights about buyer’s HRM practices within the aviation industry itself, by investigating employees of four different suppliers.

Furthermore, the measurement of HRM attributions needs attention. As mentioned in the discussion section, many studies directly borrow their measurement items from Nishii et al. (2008), while these items can have different meanings in other countries and industries (Wang et al., 2020) . The five HRM practice area’s from Nishii et al. 2008 were adapted to the aviation industry context in consultation with managers from the four suppliers. However, the way measurement items themselves are measured (HRMquality, HRMwell-being, HRMcost-reduction and HRMemployee-exploitation) was not adapted to the context of the aviation industry in Indonesia. Wang et al. (2020) argue that researchers cannot assume that these measurements capture the same kinds of employee perceptions across different countries. This also affects the generalizability of our results, because the impact the HRM attributions on ECID may differ in other countries than Indonesia. Where negative effects of control-focused HRM attributions on OI were expected, it turned out that these HRM attributions relate positively to OI in this thesis. It is difficult to say whether this has to do with the measurement of the items, but this limitation should be taken into account. Future research should focus on the impact of cultural differences across countries and industries in order to develop a more nuanced understanding of HRM attributions. Qualitative research may contribute to a better understanding of employee’s HRM interpretation in different contexts (Wang et al., 2020). Besides that, future quantitative research should develop specific HRM measures based on the empirical context instead of directly borrowing measurement items from Nishii et al. (2008).

(33)

(34)

34

7. Conclusion

This thesis aimed to investigate how buyer’s HRM practices can be used as a way to control service delivery in a service triad. Where previous research was mainly focused on ‘hard’ ways to control service delivery in a service triad, this thesis investigated which of buyer’s HRM practices are effective in influencing supplier’s employee’s identification with buyer’s customers. This was done by investigating how commitment-focused and control-focused HRM attributions relate to supplier’s employee’s OI with the buyer and ECID.

Results of conducted survey research showed that commitment-focused HRM attributions are most effective in indirectly influencing ECID through OI with the buyer in a service triad. The effects of commitment-focused HRM attributions on ECID are fully mediated by OI with the buyer. Moreover, results of this thesis showed that control-focused HRM attributions are most effective in directly influencing ECID. The effects of control-focused HRM attributions on ECID are partially mediated by OI with the buyer. A comparison between the total effects of commitment-focused HRM attributions and control-focused HRM attributions shows that these total effects do not differ substantially from each other. Therefore, it can be concluded that commitment-focused and control-focused HRM attributions are both effective in influencing ECID, with as side note that commitment-focused HRM attributions influence ECID indirectly through OI with the buyer. Results regarding commitment-focused HRM attributions are in line with findings from traditional employment literature. That’s why this thesis recommends managers of the buyer to focus initially on performing commitment-focused HRM attributions in order to increase ECID among supplier’s employees. Commitment-focused HRM attributions possibly also affect other employee outcomes in a service triad, because existing literature shows that OI has a positive influence on multiple employee and organizational outcomes.

Controlling service delivery in service triads: the impact of buyer’s HRM practices