Enhancing Service-Oriented Behaviors in an Asian Business Context: Lessons From a Pakistani Bank

This article examines how service organizations can enhance employees’ customer orientation, often exhibited through the display of service-oriented citizenship behaviors. The study, in this respect, quantitatively analyses the relationships between organizational distributive justice, leader-member exchange (LMX), and team-member exchange (TMX) on customer orientation. Data were gathered through a survey of 658 middle managers working in a Pakistani bank and were analyzed via full structural equation modelling. The findings reveal that organizational distributive justice, LMX, and TMX are key predictors of customer orientation. Specifically, the analysis suggests that the relationships between organizational distributive justice and LMX with customer orientation are mediated partially and fully, respectively, by TMX. Simultaneously, TMX partially mediates the relationship between organizational tenure and customer orientation. Our study contributes to both theory and practice of service organization functioning by signifying the importance of the organization’s fair distribution of rewards as well as it’s leader’s and co-worker’s behaviors in affecting organizationally desired employee behaviors and thereby, arguably, enabling positive organizational outcomes. Service organizations can, based on our findings, create a culture of service excellence by placing emphasis on specific elements at the organizational, leadership, and team level.

The paper is organized in five parts. The first part presents the theoretical framework and hypotheses. The second part discusses the methods used for data collection. The third and fourth parts constitute the results and discussion. The fifth part concludes the paper by summarizing the key contributions of the research and presenting avenues for future research.

Literature Review and Hypotheses Development 2.1 Organizational Distributive Justice
Organizational distributive justice is concerned with an individual's perception of whether s/he receives fair treatment by the organization not only in terms of whether rewards are impartially distributed, but also in terms of punishment (Chou & Lopez-Rodriguez, 2013). Different scholars have highlighted the positive effects of organizational distributive justice on employee commitment, employee satisfaction with rewards, job attitudes, work quality, job satisfaction, cooperation with co-workers, and labor turnovers, among others (DeConinck & Stilwell, 2004;Hornung, Glaser, & Rouseau, 2010). Low levels of organizational distributive justice, on the other hand, have been associated with opportunistic employee behaviors against the interest of the organization (Mahajan & Benson, 2013).
Employees' justice perceptions are rooted in social exchanges (Cardona, Lawrence, & Bentler, 2004) as employees who perceive the distribution of rewards and resources to be fair and equitable respond positively by being more cognitively, physically, and emotionally engaged in the workplace (Biswas, Varma, & Ramaswami, 2013) leading, among other things, to innovative behaviors at work (Pundt, Martins, & Nerdinger, 2010). We, therefore, hypothesize that if organizations want to become customer oriented by embracing organizational policies, practices, and procedures intended to support and reward service-giving behaviors that create and deliver service excellence (Lytle et al., 1998), they require a high level of cognitive, physical, and emotional engagement by employees. Organizational distributive justice may be a critical mechanism in creating such engagement. On the basis of this logic, we hypothesize the following: Leaders in organizations form a variety of exchange relationships with their subordinates. The quality of this relationship is based on the degree of support (Paillé, 2009;Steffens, Haslam, Kerschreiter, Schuh, & van Dick, 2014) and exchange of valued resources between these two parties and can lead to either low-quality leader-member relations characterized by mere contractual exchanges, or to high-quality relations supported through social exchange mechanisms. This latter type of relationship typically extends beyond the mere contractual requirements often evident in a work context (Brunetto, Farr-Wharton, & Shacklock Brunetto, 2010). Because of this, LMX has been examined as a factor which can affect organizational identification. Employees in high LMX relationships are more likely to develop a personalized customer orientation and to have access to supervisor support to help them with customer interactions (Tangirala, Green, & Ramanujam, 2007). The importance of leaders and of leadership style in creating SOCBs and customer orientation has widely been discussed in the literature (Carter, Mossholder, Field, & Armenakis, 2014;Chiaburu, Lorinkova & Van Dyne, 2013;González & Garazo, 2006;Menguc & Auh, 2008;Martin & Bush, 2003), in particular how a transformational leadership style can enable customer orientation (MacKenzie, Podsakoff & Rich, 2001;Harris & Ogbanna, 2001). A positive relationship between LMX and customer orientation has been established (Paparoidamis & Guenzi, 2009), as well as a strong LMX-organizational identification relationship and a significant LMX-customer orientation relationship (Farrell & Oczkowski, 2012). Accordingly, we hypothesize the following: TMX relates to the reciprocity between an individual with his/her team members with respect to contribution of ideas, feedback, and assistance, in return for information, help, and recognition (Seers, 1989;Seers, Petty, & Cashman, 1995). Liu, Loi and Lam (2011) suggested that high-quality TMX enables team members to receive important psychological and socioemotional resources. They further discuss that this can facilitate organizational identification with positive effects on employee in-role and organizational citizenship behavior (OCB), whereas low-quality TMX would only have a minimal effect on the team member's willingness to share information with peers and provide assistance and recognition to them, therefore limiting exchanges to the standard requirement of job completion. To be more precise, OCB tends to be excluded from formal organizational expectations such as job descriptions and relates to discretionary individual behaviors that are not explicitly recognized by the formal reward system (Jiao, Richards, & Hackett, 2013). Closely related to this concept is the notion of SOCB, which refers to customer focused discretionary behaviors that extend beyond the employees' formal role requirements when servicing customers (Chou & Lopez-Rodriguez, 2013). SOCB can lead to increased employee loyalty and participation and overall higher-quality service delivery (Bettencourt et al., 2001). The fact that the desired outcome of SOCB is higher-quality service delivery suggests that SOCB incorporates the concept of customer orientation which focuses on customer loyalty and satisfaction (Lytle & Timmerman, 2006;Homburg et al., 2002;Lytle et al., 1998).
When considering the relationship between organizational justice and SOCB, Chou and Lopez-Rodriguez (2013) argued that from a social exchange perspective, one can expect that service employees are likely to engage in service-related discretionary behaviors as an exchange for fair treatment received by the organization, and that service employees' perceptions of fair treatment are expected to have an important impact on their SOCB. As it was argued above, SOCB is similar to customer orientation, and thus, Chou and Lopez-Rodriguez's (2013) arguments can be extended to include a positive link between organizational justice and customer orientation.
Further to this argument, Ohana (2012) found that group commitment moderated the relationship between perceived organizational support (POS) and job satisfaction and that this interaction effect, in turn, mediates the distributive justice-job satisfaction relationship. Additional literature on POS (Byrne & Hochwarter, 2008;Rhoades & Eisenberger, 2002) presents the view that employees form beliefs concerning the extent to which the organization values their contributions and cares about their well-being and suggest that employee behaviors, such as service orientation towards customers, can be affected by such beliefs. One aspect of POS relates to perceived co-worker support which refers to the degree to which employees perceive that their co-workers care about their well-being, respect their contributions to the organization, help them acquire new skills and knowledge, and support them in times of organizational change (Fuchs & Prouska, 2014). This is related to the notion of TMX which refers to reciprocity between a member and his/her team which in turn may facilitate organizational identification and has positive effects on OCB and SOCB (Liu et al., 2011). Furthermore, one important aspect of POS is co-worker support which is closely linked to TMX (Seers et al., 1995), and because POS can be influential in affecting employee behaviors (Byrne & Hochwarter, 2008) such as customer orientation, we hypothesize the following:

Hypothesis 3: Team-member Exchange Mediates the Relationship Between Organizational Distributive Justice and Customer Orientation
The role of leadership in creating high-quality LMX that extends beyond the mere contractual requirements has been widely discussed in the literature (Carter et al., 2014;Kahrobaei & Mortazavi, 2016;Menguc & Auh, 2008;Martin & Bush, 2003;MacKenzie et al., 2001;Harris & Ogbanna, 2001). Oczkowski (2012) and Paparoidamis and Guenzi (2009), for instance, have established a positive relationship between LMX and customer orientation. Leadership styles fostering organizational support as discussed in the POS literature (Byrne & Hochwarter, 2008;Rhoades & Eisenberger, 2002) can be strong enough to influence TMX and enable individuals to exhibit citizenship behaviors. In more detail, LMX is linked to citizenship behaviors and typically high-quality LMX is characterized by high levels of trust, interaction, support, and formal/informal rewards, including an exchange of material and non-material goods that extend beyond what the formal job description specifies (Ilies, Nahrgang, & Morgeson, 2007). In order to reciprocate high LMX relationships, subordinates will go beyond their specified role and engage in citizenship behaviors to maintain an equitable exchange (Kuvaas & Dysvik, 2010). Due to the established positive relationship between LMX and customer orientation (Oczkowski, 2012;Paparoidamis & Guenzi, 2009) and given that leadership styles in LMX can affect TMX (Byrne & Hochwarter, 2008), and further may in turn affect customer orientation, we hypothesize the following:

Context
The study followed a quantitative approach in order to test the above hypotheses. The investigation occurred in one of the largest financial institutions in Pakistan, which has operations across the entire country. Pakistan is an emerging market that has undergone and continues to undergo significant structural transformations based on liberal market reforms that attempt to develop a promising economy despite its still volatile and weak business system (Jormanainen & Koveshnikov, 2012). The banking sector in Pakistan is on the whole underresearched, even though more attention has recently been drawn on exploring various issues in conventional and Islamic banking in emerging markets (Rammal & Parker, 2012). The banking sector is one of the fastest growing service sectors in Pakistan, and the with this associated increase in competition poses critical pressures to organizations to become more customerfocused by fostering a customer-oriented culture (Qayyum, Sukirno, & Mahmood, 2011). It is, therefore, important to explore how SOCBs, such as customer orientation, can be developed within this context, since this may provide useful insights for banks operating in Pakistan and potentially in other emerging economies.

Sample
The organization employs more than 80,000 people in Pakistan. Our survey focused on middle managers due to their important role in changing the culture of an organization (Psychogios, Alexandris, & Onofrei, 2008) and creating a strategic orientation towards service-oriented culture (Balogun & Johnson, 2004;Huy, 2011). For the purpose of this study, we adopted a broad definition of middle managers suggested by Dopson, Risk and Stewart (1992), including Human Resource Research ISSN 2329-9150 2018 all those managers just below the group of senior managers (CEOs, CFOs, etc.). In particular, the whole population of these managers that were approached included 660 individuals holding a regional head position in one of the functional areas in the bank; audit, financial control, commercial/consumer/wholesale banking, operations, treasury and foreign exchange investment, special asset management, Islamic banking, strategic planning and capital markets, risk management, business development, compliance, human resources, and IT. In total, 658 responses were received, representing a 99.7% response rate. Such a high response rate is not untypical for quantitative studies conducted in high power distance cultures such as Pakistan (House, Hanges, Javidan, Dorfman & Gupta, 2004), as filling in a survey distributed via top management may be something that is seen as a duty rather than a voluntary activity. Each respondent received an email introducing him/her to the purpose of this research and an invitation to fill in an electronic version of the questionnaire.
The questionnaire was written in English and Urdu, so that all participants could understand and fill in the survey. A one-pager accompanied the questionnaire signed by the President and the Group Head HR to ensure a high response rate. From the 658 respondents 88% were male. In terms of their age group, 14% were 20-30 years old, 30% 31-40 years, 19% 41-50 years, and 37% were 51 years old or older. Most respondents (31%) had a long length of service at the company of 30 or more years, 15% of 21-30 years, 12% of 11-20 years, 9% of 4-10 years, 19% of 1-3 years, and 14% of less than 1 year. The majority of respondents (54%) were holders of a postgraduate degrees (M.A., M.B.A., M.Sc., etc.), 37% of an undergraduate degree (e.g. B.A., B.Sc., etc.), and 0.3% were holders of a Ph.D.

Measures
All study variables were measured on a five-point Likert-type scale asking for respondents' agreement, disagreement, or indifference (1=agree, 2=somewhat agree, 3=neutral 4= somewhat disagree, 5= disagree). All scores were reversed prior to the analyses so that higher values reflect stronger agreement and lower values weaker agreement with the questionnaire items. Organizational distributive justice was measured with the scale suggested by Byrne and Cropanzano's (2000) scale. Sample items include "My compensation as compared to the industry average (top 4 local banks) is fair and equitable" and "I believe that my workplace rewards are given fairly". LMX was measured with the scale proposed by Wang, Law and Chen (2008). The scale includes items such as "My manager/supervisor is available to me when I have questions or need help" and "My manager/supervisor effectively resolves employee problems and issues". The TMX scale was adopted form Seers ' (1989) work. Sample items are "Employees at [the organization's name] regularly share and exchange ideas" and "In [the organization's name], teamwork is encouraged and recognized". Customer orientation was measured with items adapted from the SOCO scale developed by Saxe and Weitz (1982), such as "I try to help customers achieve their goals", "I try to get customers to discuss their needs with me", "I try to figure out what a customer's needs are", and "I try to find out what kind of product would be most helpful to a customer".

Data Analysis
Human Resource Research ISSN 2329-9150 2018 Our analysis consists of two key elements based on DeVellis's (2003) and Long's (1983) recommendations and is performed in LISREL 8.80 (SSI, Scientific Software International, Lincolnwood, IL). Firstly, we tested our measurement scales using confirmatory factor analysis to identify their psychometric properties. After this, we conducted independence and the Harman (1976) test to determine the independence and empirical distinctiveness of our variables. Due to violations of non-normality assumptions associated with five-point Likerttype questionnaire items, the data is treated as ordinal, and polychoric correlation matrices with associated asymptotic covariance matrices are created as input into the analysis (Jöreskog, 2005). The independence and Harman (1976) tests involve a number of steps. Firstly, all items of each study variable are loaded onto their respective latent construct in combination with the items of another latent construct (e.g. organizational distributive justice and LMX). Secondly, all of these items are loaded onto a single latent factor before their fit to the data is compared with the two-factor model. If the two-factor model fits the data better than the one-factor model, both constructs are empirically distinct from each other. Thirdly, for the Harman (1976) test, all items are loaded onto their respective latent construct for all four study variables, and subsequently onto a single factor. A data fit comparison of both models then indicates the distinctiveness of all study variables. In line with the recommendations put forward by Jöreskog (2005) and Satorra and Bentler (2010), we also performed the Satorra-Bentler (2010) scaled chi square difference test (SB χ2 difference) to determine the statistical difference of our measurement scales.
Secondly, when testing our structural models, we treated the multi-item constructs as continuous data and use covariance matrices in line with Jöreskog's (2005) recommended approach for full structural equation modelling. Further, and in line with the analytic convention for mediation analysis presented by Baron and Kenny (1986) and Kelloway (1998), we firstly tested a fully mediated version of our structural model as supported by Hypotheses 3 and 4, then a partially mediated version of the model, and finally a non-mediated version of the model. The model with the best fit to the data subsequently constituted our final and supported structural model. We considered four key control variables (age, gender, organizational tenure, and educational level) as they have been linked to a number of our study variables in similar research (e.g. Halstead, Jones, Lesseig, & Smythe, 2008).
To determine the adequacy of our measurement and structural models, we adhere to a number of fit indices commonly applied to this type of analysis. Specifically, and in line with Hu and Bentler (1999) and Jöreskog (2005), the chi square (χ2) statistic used in our study is the SB χ2 as it adjusts for non-normality with ordinal data in confirmatory modelling (i.e. the confirmatory factor analyses). For our structural models (and the respective difference tests) which treat the data as continuous, however, we have used the normal theory χ2 as suggested by Jöreskog (2005). Regarding the χ2/degrees of freedom (df) ratio, we rely on Bollen (1989) and Kelloway's (1998) recommendation and consider a ratio below two as a good fitting model, a ratio between two and three as an acceptable fit, and a ratio between three and five as approaching an acceptable level of fit. For the standardized root mean square residual (SRMR) and the root mean square error of approximation (RMSEA), we have used a cut-off of 0.05 for good fitting models, a cut-off between 0.05 and 0.08 for acceptably fitting models, and a cutoff between 0.08 and 0.10 for models that approach an acceptable fit (Hu & Bentler, 1999;Human Resource Research ISSN 2329-9150 2018, Vol. 2, No. 1 Steiger, 2000. Additionally, we have presented the comparative fit index (CFI) and the goodness of fit index (GFI). For these indices, and in line with Bollen (1989), we have treated values between 1.00 and 0.95 as indicators of good-fitting models, values between 0.95 and 0.90 as indicators of satisfactorily-fitting models, and values below 0.90 as indicators of models that approach acceptable levels fit. The expected cross-validation index (ECVI) is used to determine the badness of fit, with higher values indicating a weaker fit to the data (Jöreskog, 2005). In addition to these fit statistics, we have also provided results from the Sobel (1982) tests for the best-fitting structural model to gain further confidence in our final model.

Structural Models
Our approach to testing the structural models composes of three steps. Firstly, we present a model which tests for full mediation between the independent variables of organizational distributive justice and LMX, the mediator TMX, and the dependent variable customer orientation.  ISSN 2329-9150 2018 to either TMX or to customer orientation (age: beta=-0.02, p>0.05; gender: beta=0.01, p>0.05; educational level: beta=-0.05, p>0.05). The path between organizational tenure and customer orientation is also non-significant (beta=0.17, p>0.05) in this model.
In addition to these structural tests, we perform χ2 difference tests to determine whether the three models are significantly different from each other. The best-fitting partially mediated model, in fact, is different from the fully mediated model (χ2 difference=15.26, df=2, p<0.001), and from the non-mediated model (χ2 difference=90.73, df=35, p<0.001). The fully mediated model is also different from the non-mediated model (χ2 difference=105.99, df=37, p<0.001). We perform Sobel (1982) tests on the best-fitting partially mediated model to gain further confidence in the structural relationships. This additional testing indicates that TMX mediates the relationships between organizational distributive justice and customer orientation (Sobel=2.08, p<0.05), between LMX and customer orientation (Sobel=6.43, p<0.001), and between organizational tenure and customer orientation (Sobel=2.54, p<0.05). Hypotheses 3 and 4 are supported by the data. Figure 2 depicts all significant relationships from this bestfitting partially mediated structural model. Non-significant paths, however, are not displayed due to graphical constraints. ISSN 2329-9150 2018 Figure 2. The best-fitting partially mediated model Note: ***p<0.001, **p<0.01, *p<0.05.

Discussion
Overall, our results extend the existing work that looks at the links between organizational distributive justice, LMX, TMX, and customer orientation. Specifically, the results of our study indicate a best-fitting partially mediated model (see Figure 2) among the chosen study variables. In total, there are four key findings. The first finding is that organizational distributive justice has a positive relationship with customer orientation. The second finding is that LMX has a positive relationship with customer orientation. The third finding is that TMX partially mediates the relationship between organizational distributive justice and customer orientation. The fourth finding is that TMX mediates the relationship between LMX and customer orientation. To further elaborate and reflect on these findings, the following discussion reviews them in greater detail and against the current literature.
Firstly, our study shows that organizational distributive justice is positively related to customer orientation. We can, based on this, argue that distributive justice is a crucial mechanism which can create engagement through organizational policies, practices, and procedures which are intended to support and reward behaviors focused on service excellence (Lytle et al., 1998). Our work adds to that of Chou and Lopez-Rodriguez (2013) who argue that service employees are likely to engage in service-related discretionary behaviors as an exchange for fair treatment received by the organization, and that service employee' perceptions of fair treatment are expected to have a positive impact on their SOCBs, such as customer orientation. Secondly, our study indicates that LMX is positively related to customer orientation. This is an important finding because it demonstrates the significance of leaders and of leadership behaviors for creating customer orientation (Menguc & Auh, 2008;Martin & Bush, 2003). More specifically, our study adds to the work of Paparoidamis and Guenzi (2009) and Farrell and Oczkowski (2012) who also found a significant LMX-customer orientation relationship.
Thirdly, we found that TMX partially mediates the relationship between organizational distributive justice and customer orientation. This means that social interactions in the workplace can be affected by the perceived levels of fairness and that the fairer employees perceive the organization to be, the more positive their relationships with their co-workers are. Our findings are in accordance with prior research which has explored the positive relationship between justice perceptions and TMX (e.g. Murphy, Wayne, Liden & Erdogan, 2003). Such positive relationships can lead to desirable organizational outcomes, such as excellent customer support through customer orientation.
Fourthly, our study shows that TMX mediates the relationship between LMX and customer orientation. We provide additional support for the idea that leaders and leadership styles are important for influencing TMX, as previously discussed in the literature (Menguc & Auh, 2008;Martin & Bush, 2003), within the Pakistani banking context. Psychological and socioemotional support between colleagues can facilitate organizational identification with positive effects on employee in-role and organizational citizenship behavior (Kwan et al., 2015;Liu et al., 2011) and, thus, positively affect behaviors such as customer orientation. The SOCB literature (Chou & Lopez-Rodriguez, 2013) discusses how such behaviors can lead to increased employee loyalty and participation and overall higher-quality service delivery (Bettencourt et al., 2001). Our study adds to this body of research and suggests that co-worker interactions in the workplace are crucial in creating customer loyalty and satisfaction (Lytle & Timmerman, 2006;Homburg et al., 2002;Lytle et al., 1998). The mediating effects of TMX between organizational distributive justice and customer orientation and LMX and customer orientation may be explained as TMX alone does not account significantly enough for customer orientation. In practical terms, good co-worker interactions and relationships on their own may not be enough to create organizational identification, employee commitment, and drive individual employees to exhibit desirable organizational citizenship behaviors, such as customer-oriented behaviors. Indeed, other factors such as psychological ownership for the organization and the work environment have also been found to affect citizenship behaviors (O'Driscoll, Pierce & Coghlan, 2006). The results indicate that organizational distributive justice and LMX are expected to encourage employees to more positively interact with coworkers, and once individuals develop effective team relationships, they become more likely to behaviorally support SOCBs such as customer orientation.
Our results also indicate that organizational tenure plays a significant role in the presented model as it relates positively to TMX and customer orientation. This is in line with Fuchs and Edwards (2012) and Riketta (2005) who suggest that tenure is positively related to organizational identification through a gradual convergence of values and goals between employees and the organization. This, importantly, is a rather interesting finding in itself that ought to be further explored, as there is the possibility that organizational tenure may be a direct antecedent of TMX and LMX, rather than LMX being an antecedent of TMX.

Conclusions
Human Resource Research ISSN 2329-9150 2018 Beyond the academic significance of these findings, important implications for managers, leaders, and entire organizations in enabling positive organizational outcomes such as customer orientation, can be drawn. The discovery of organizational distributive justice, LMX and TMX as key predictors of SOCBs, such as customer orientation, signifies that managers, leaders, and also co-workers all play a key role in affecting employee behaviors and enabling positive organizational outcomes. This means that service organizations wishing to create a culture of service excellence (Lytle et al., 1998;Lytle & Timmerman, 2006) need to place emphasis on three organizational levels. On a leadership level, it is important to have appropriate leadership that positively affects individuals and teams at work in relation to the desired outcome (Paparoidamis & Guenzi, 2009;Farrell & Oczkowski, 2012). On a managerial level, the notion of organizational distributive justice as a key predictor of customer orientation signifies the importance of managers in creating positive employee perceptions of fair treatment at work, especially in terms of distributions and rewards. On a co-worker level, our study shows the importance of team dynamics in creating SOCBs and customer orientation.

Limitations
Further to the above highlighted implications, our study comes with some limitations. For example, the exclusive use of self-reported measures, especially for our customer orientation scale, can be seen as one limitation of the study. A more accurate measure would have been to ask customers about their perceptions of customer orientation, which is what we encourage other scholars to do in future research. Moreover, this study focuses on a sample of middle managers and not employees and more senior managers of the organization which would have provided a more holistic view on the tested variables. Further, the context of our investigation is a single organization operating in a specific industry. Testing these relationships in a different context would enhance the overall rigor of the study and the level of generalizability of the findings. The nature of our data is cross-sectional. Longitudinal data which allows for testing across different times would improve the reliability of our data and we henceforth encourage scholars to test the proposed model using a longitudinal research design in the future.

Future Research
Despite these limitations, there are a number of future research avenues arising from our study. Firstly, although current literature discusses the role of leadership in affecting organizational citizenship behaviors, more research is needed in exploring particular leadership styles that may be applied to a service-oriented organizational culture. Secondly, there is more work needed in exploring how employees define fair treatment at work in a services context and how managers can affect their perceptions of this. Thirdly, although we found that TMX affects customer orientation positively, more research is needed in untangling team dynamics and their power over employee behaviors associated with customer orientation as an outcome variable. More organizational behavior and management studies in the Pakistani context can widen both scholars' and practitioners' views on emerging economies, such as Brazil, Russia, India and China, which fall outside the more frequently researched countries.