Strategic HRM as process: how HR system and organizational climate strength influence Chinese employee attitudes

Xiaobei Lia*, Stephen J. Frenkela and Karin Sandersb

aSchool of Organization and Management, Australian School of Business, The University of New South Wales, Sydney, Australia; bOrganizational Psychology and Human Resource Development,

Faculty of Behavioral Sciences, Twente University, Enschede, The Netherlands

In contrast to the high-performance work systems literature that focuses on HR practices, we follow Bowen and Ostroff in examining human resource management (HRM) processes, specifically the strength of an HR system (its distinctiveness, consistency, and consensus) and its contribution to the organizational climate (employees’ shared perceptions of the HR system). Based on 810 employees within 64 units in three Chinese hotels, we examine how employee perceptions of HRM system strength and organizational climate are associated with employees’ work satisfaction, vigor, and intention to quit. The distinctiveness of an HRM system was found to be related to the three employee work attitudes, and high climate strength increases both the positive relationship between consensus and work satisfaction, and the negative relationship between consensus and intention to quit. We draw on aspects of Chinese society to interpret these findings. Several important research and HR practice implications are highlighted and discussed.

Keywords: China; high-performance work system; human resource practices; organizational climate; strategic HRM; work attitudes

Introduction

An interest in the effects of high-performance work systems (HPWS) on employees in

service industries has been growing in recent years (Batt 2002; Boxall and Macky 2007).

HPWS is usually defined as a set of human resource (HR) practices aimed at increasing

employees’ abilities, motivation, and opportunity to participate in decision making (Tsui

and Wang 2002; Guest 2007; Sun, Aryee and Law 2007). HPWS, like high-commitment

Human Resource Management (HRM) (Benkhoff 1997; Agarwala 2003) and high-

involvement work systems (Xiao and Bjorkman 2006; Macky and Boxall 2008), assumes

that various types of HR practices interact to improve employees’ work attitudes,

ultimately contributing to positive employee behaviors and organizational effectiveness

(Boxall and Macky 2009). Despite some skepticism (Wood and Wall 2007), the weight of

empirical evidence favors this assumption (Hailey, Farndale and Truss 2005; Combs, Liu,

Hall and Ketchen 2006; Boxall and Macky 2009).

Rather than focusing on HR practices or the content of HRM, some academics (Bowen

and Ostroff 2004; see also Patterson, Warr and West 2004; Neal, West and Patterson 2005;

ISSN 0958-5192 print/ISSN 1466-4399 online

q 2011 Taylor & Francis

DOI: 10.1080/09585192.2011.573965

http://www.informaworld.com

Dr. Xiaobei Li is now a research fellow at Guanghua Leadership Institute, Guanghua School of Management, Peking University, China.

*Corresponding author. Email: [email protected]

The International Journal of Human Resource Management,

Vol. 22, No. 9, May 2011, 1825–1842

Nishii, Lepak and Schneider 2008) have recently turned their attention to HRM process, in

particular, the way HR policies and practices are communicated to employees. Bowen and

Ostroff (2004) identify a strong HRM system as comprising three features: distinctiveness,

consistency, and consensus. They suggest that these features contribute to a strong

organizational climate defined as the shared perceptions of the organization in terms of

practices, policies, procedures, routines, and rewards (Bowen and Ostroff 2004: 205). The

establishment of a strong organizational climate builds on an individual’s psychological

climate, defined as an experientially based perception of what people ‘see’ and report

happening to them as they make sense of their environment (pp. 205).

When HR practices are perceived by employees as distinctive, consistent with each

other, and applied by key policy makers in a similar way, individual perceptions are likely

to converge and will tend to be reinforced by the collectivity. In other words, feelings of

well-being will lead to higher performance through ‘motivating employees to adopt

desired attitudes and behaviors that in the collective, help achieve the organization’s

strategic goals’. (Bowen and Ostroff 2004, p. 204)

In this article, we test Bowen and Ostroff’s key ideas by examining the effects of the three

features of an HRM system and shared perceptions regarding HPWS on three commonly

used measures of employee work attitudes: work satisfaction, vigor, and intention to quit.1 In

addition, we examine the moderating effects of organizational climate on the relationships

between features of an HRM system and employee work attitudes. Our contribution is to

subject a leading theory of strategic HRM process to empirical test and by so doing advance

our understanding of the mechanisms linking HR systems to employee attitudes. In pursuing

this path, we improve on a previous study by Sanders, Dorenbosch and de Reuver (2008) by

using three dependent variables rather than a single measure (affective commitment), and by

employing a more convincing measure of consensus based on employee perceptions of HR

policy and practice implementation rather than the extent of agreement as reported by line

and HR managers (Fiske and Taylor 1984; Boxall and Macky 2007; see also Sanders et al.

2008). Employee surveys in three five star-hotels located in comparable urban areas in China

provide the data for our study. Focusing on a single industry segment helps to minimize the

influence of labor and product markets, and other environmental characteristics (Baron and

Kreps 1999; Wright and Haggerty 2005). HPWS is relevant to these workplaces, because,

compared to other privately-owned organizations in China (Zhu 2005, Zhu, Thomson and

Cieri 2008), five-star Chinese hotels have been eager to adopt advanced HRM practices

including extensive training and formalized performance appraisal systems (Sun et al. 2007).

The remainder of this article is organized in four sections. First, we outline our key

concepts and theoretical framework followed by a discussion of motivating hypotheses.

Second, we describe the data and methodology. Third, we report our results that are

discussed in a fourth section that identifies some of the limitations of the study and

considers the implications for further research and HR policy and practice.

The research framework

As mentioned above, the key concepts in our study are features of an HRM system and

employee attitudes. The research framework is summarized in Figure 1, followed by an

outline of our hypotheses.

Relationships between HRM system features and employee attitudes

Based on attribution theory (Kelley 1973), the process view of HRM explains how HR

practices shape an individual’s psychological climate (Ostroff and Bowen 2000; Bowen

X. Li et al.1826

and Ostroff 2004). Employees use HRM messages as communication signals from

management to make sense of their work situation (Guzzo and Noonan 1994;

Schneider 2000). This sense-making process is facilitated by individual attributions about

cause–effect relationships (Nishii et al. 2008). When employees become increasingly

confident in making such cause–effect inferences, a strong psychological climate is likely

to emerge. As noted earlier, three features of an HRM system that contribute to a strong

climate are distinctiveness, consistency, and consensus (Bowen and Ostroff 2004).

Distinctiveness refers to an HRM system being visible, understandable, legitimate, and

relevant to employees’ goals (Kelley 1973; Bowen and Ostroff 2004; Sanders et al. 2008).

When the HRM process clearly captures attention, employees are more likely to attribute HR

messages to a purposeful management. Consistency refers to the features of an HRM system

being internally aligned. This means that HR practices reinforce one another synergistically

and are more likely to be viewed as a causal bundle having distinctive effects ultimately

attributable to management across contexts and time (Sanders et al. 2008, p. 414). Consensus

refers to the extent to which there is agreement among policy makers – typically HR and line

managers – in the way HR practices are implemented. Thus, when HRM policy

implementation, including procedures, are seen as highly consensual among decision makers,

employees are more likely to agree that these emanate from management, i.e. that there is a

cause–effect relationship. According to Bowen and Ostroff (2004), when an HRM system is

high in the three features referred to above, employees will tend to have a clearer view of

cause (HRM)–effect (a purposeful management) relationships and are likely to be strongly

influenced by these system properties, especially where it conveys positive messages. This

conjecture is largely but not entirely supported by the only empirical study we are aware of

that has tested Bowen and Ostroff’s key ideas. Sanders et al. (2008) used multi-actor data

(671 employees, 67 line-managers, and 32 HR managers) from 18 departments in four Dutch

hospitals to analyze the relationships between HRM system features and employees’

affective commitment. Distinctiveness and consistency were found to be positively related to

affective commitment, but consensus (measured as the deviance score of the perceptions of

line and HR managers concerning HR practices) did not predict affective commitment.

As noted above, in this study, we focus on three employee attitudes: work satisfaction,

vigor, and intention to quit rather than affective commitment as our dependent variables

and following Bowen and Ostroff (2004) we hypothesize that:

Employees’ perceptions of the distinctiveness, consistency, and consensus of the HRM system are positively related to their work satisfaction (H1) and vigor (H2), and are negatively related to intention to quit (H3).

HRM system features: Distinctiveness,

Consistency, and Consensus

HPWS Climate Strength

Employee attitudes: Work satisfaction, Vigor

and Intention to Quit

Individual Level

Unit Level

Figure 1. Hypothesized relationships linking HRM system features, HPWS climate strength and employee outcomes.

The International Journal of Human Resource Management 1827

Shared perceptions: the moderating effects of HPWS climate strength

Bowen and Ostroff (2004, p. 204) propose organizational climate as a mediator in the

relationship between HRM system strength and organizational performance. Sanders et al.

(2008) suggest that the concept of strong organizational climate used by Bowen and

Ostroff (2004) refers to climate strength rather than climate level. Although climate level

represents the convergent ratings of perceptions of a specific facet of the work situation

such as safety, service, or HRM (Schneider 1990, 2000; Klein, Conn, Smith and Sorra

2001; Schneider, Salvaggio and Subrirats 2002) and is usually measured by the mean of

individual perception scores, climate strength refers to the extent of agreement about the

climate. It is measured by homogeneity statistics relating to the aggregation of members’

perceptions, such as standard deviation and within-group correlations (Klein et al. 2001;

Luria 2008). Thus, the concept of climate strength more closely represents Bowen and

Ostroff’s (2004) concept of organizational climate as employees’ shared perceptions.

In this study, we define HPWS climate strength as the extent of shared perceptions of

HPWS in an organization. Where this is high, established norms induce conformity in

terms of responses and foster skills that facilitate appropriate attitudes and behavior

(Mischel 1973, 1977; Mischel and Peake 1982; Bowen and Ostroff 2004; Johns 2006).

According to organizational climate research, climate strength usually has a moderating

effect on outcomes (Gonzalez-Roma, Peiro and Tordera 2002; Schneider et al. 2002;

Ehrhart 2004). Specifically, the relationship between antecedents and outcomes is stronger

in a strong situation than in a weak one. Reflecting the convergence of group members’

perceptions regarding climate level, strong climate strength implies that associated

relationships of antecedents and outcomes are inclined to be interpreted in a similar way

by group members (Mossholder, Bennett and Martin 1998; Schneider et al. 2002; Yang,

Mossholder and Peng 2007). Moreover, contra Bowen and Ostroff (2004), Sanders et al.

(2008) found that organizational climate moderated rather than mediated the relationship

between consistency and affective commitment, this relationship being stronger when

employees had more similar perceptions concerning the existence of high commitment

work systems within their department. Accordingly, we expect that HPWS climate

strength has a moderating rather than mediating effect on the relationship between the

features of the HRM system and employee attitudes. In a situation where HPWS climate

strength is high, implying that employees share perceptions regarding HPWS (HRM

content), employees will be more confident about attributing this as having benign effects

on their work experience. Thus, it can be hypothesized that:

HPWS climate strength moderates the relationships between key features of an HRM system (distinctiveness, consistency and consensus) and work satisfaction (H4), vigor (H5), and intention to quit (H6) such that these relationships are stronger when HPWS climate strength is high.

Method

Sample and procedures

Data were collected from three five-star hotels, located in three urban cities (Shanghai,

Ningbo and Dongguan) in China. Each hotel has at least 200 rooms and is more than 4

years old. Each is privately owned and one is managed by an international hotel

group. Management was approached through personal contacts, which is useful in doing

research in China (Easterby-Smith and Malina 1999). Surveys were distributed to each

participating hotel. Sealed completed questionnaires were returned first to the hotel’s HR

manager and then to a researcher. For all three hotels, 810 valid responses of frontline

employees (90% response rate) were collected. This high response rate has been observed

X. Li et al.1828

in several Chinese management studies (see Cooke 2009). The dataset included 484

(59.8%) female and 326 (40.2%) male employees, with an average of 25.5 (SD ¼ 7.9)

years of age and an average tenure in the organization of 26.6 months (SD ¼ 37.15). Over

two-thirds of employees (68%) had obtained qualifications from vocational or high

schools and earned higher salaries than their counterparts in other local hotels.2

Each hotel consists of several service departments, such as catering, reception, and

security. Within each department, there are several work units. For example, the catering

department of one hotel includes banqueting, beverage, restaurants, and room service

units. Our dataset comprising the three hotels included 64 units.

Measures

The questionnaire was administered in Mandarin after initially being developed in

English. Two bilingual researchers back-translated the survey independently (Brislin

1980). In addition, a pilot study was conducted on a group of frontline employees; these

were subsequently excluded from the final dataset. The questionnaire was finalized with a

few changes in wording.

For the items of all scales, we used six-point rather than five-point Likert scales. This

was done in order to address Chinese people’s tendency to conceal positive emotions and

hence select midpoints of a range (Lee, Jones, Mineyama and Zhang 2002). Response

items ranged from 1 ¼ strongly disagree to 6 ¼ strongly agree.

Work satisfaction (Cammann, Fichman, Jenkins and Klesh 1983) was measured by a

three-item scale (Cronbach’s a ¼ 0.81). Two illustrative items were: ‘All in all, I am

satisfied with my job’ and ‘In general, I like working here’. Vigor (Schaufeli and Bakker

2004) was measured by a five-item scale (Cronbach’s a ¼ 0.74). Example items included

‘At my work, I feel bursting with energy’ and ‘When I get up in the morning, I feel like

going to work’. Intention to quit (Firth, Mellor, Moore and Loquet 2004) was measured by

a three-item scale (Cronbach’s a ¼ 0.84). For example, ‘I often think about quitting my

job’ and ‘I am starting to ask my friends/contacts about other job possibilities’.

High-performance HR practices was measured by a 17-item scale, modified from the

scale specifically developed by Sun et al. (2007) to study Chinese hotel employees. This

covered five HR practices related to training, internal promotion, employee participation,

results-oriented pay, and job security. Items included ‘I have had sufficient job-related

training’ and ‘My job allows me to make decisions on my own’. Each HR practice

demonstrated good reliability (Cronbach’s a ranged from 0.70 to 0.88). Assuming that the

system of HR practices rather than a single practice reflects an organization’s investment

in employees and influences the organization’s performance beyond the sum of such

practices (Rousseau 1995; Delery and Doty 1996; Allen, Shore and Griffeth 2003;

Whicker and Andrews 2004; Guest 2007), a HPWS index was developed along lines

similar to other scholars (Ramsay, Scholarios and Harley 2000; Batt 2002; Beugelsdijk

2008; Doellgast 2008). The sum of the item scores for each of the five HR practices

mentioned earlier was averaged and then an average was calculated across the five

practices (Cronbach’s a ¼ 0.72). Confirmatory factor analysis suggested a good fit with

the data as indicated by the fit statistics (x 2(109) ¼ 424.30; x 2/df ¼ 3.89; p , 0.001;

TLI ¼ 0.92; CFI ¼ 0.94; RMSEA ¼ 0.06).

Following several climate studies (Luria 2008; Sanders et al. 2008), climate strength of

HPWS was calculated as the inverse standard deviation of the HPWS index at the unit level.

Regarding the HRM system, distinctiveness was measured by a shortened five-item

scale developed by Frenkel, Li and Restubog (in press) with good reliability (Cronbach’s

The International Journal of Human Resource Management 1829

a ¼ 0.84). Example items included ‘HR practices here help me to achieve the company’s

goals’ and ‘HR practices here make me feel much more confident in my ability to do my

job well’. Consistency (Sanders et al. 2008) was assessed by within-respondent agreement

in relation to the HPWS index, operationalized as the inverse average deviation for each

HR practice for each respondent (consistency-based approach, Burke, Finkelstein and

Dusig 1999). Consensus (Delmotte, Winne, Gilbert and Sels 2007) was measured by a

modified four-item scale (Cronbach’s a ¼ 0.86), with items such as ‘HR practices are

delivered by mutual agreement between HR management and line management’ and

‘Management unanimously supports HR policies’. Confirmatory factor analysis

demonstrated that a two-factor (distinctiveness and consensus) structure

(x 2(26) ¼ 183.33; x 2/df ¼ 7.05; p , 0.001; TLI ¼ 0.94; CFI ¼ 0.97; RMSEA ¼ 0.08)

fits the data better than a one-factor structure (x 2(27) ¼ 322.63; x 2/df ¼ 11.95;

p , 0.001; TLI ¼ 0.89; CFI ¼ 0.93; RMSEA ¼ 0.11).3 Therefore, although distinctive-

ness and consensus were strongly related (see later), they were analyzed as two variables.

Based on the proposition that employees’ personal and employment characteristics are

likely to influence the three dependent variables, the following characteristics were included

as controls: age, gender, type of labor contract (permanent vs. temporary), educational level

(from junior middle school to master degree and above), and workplace tenure.

Most measures were based on self-report data collected at one point in time. The

analysis may, therefore, be vulnerable to the problem of common method variance (CMV)

(Podsakoff, MacKenzie, Lee and Podsakoff 2003). Spector (2006) suggests that CMV

caused by a single data source (a single rater rather than multi-raters) should be

distinguished from CMV caused by the same measurement techniques (such as item

formats, data collection procedures, key methods). Regarding the data source, as

addressed by many researchers, information on employee perceptions and attitudes is

difficult to measure accurately using methods other than self-reports (Frese and Zapf 1988;

Spector 2006). One way to confirm the accuracy of the self-report measures is to link them

with data from other sources. In Table 1, we attempted to compare the self-report measures

from the survey with information obtained from hotels and an industry report. As shown in

the table, the ranking of turnover rate for the three hotels accords with the intention to quit

ranking. Assuming a positive relationship between employee attitudes and performance

(Boxall and Macky 2009), employee self-reported work satisfaction and vigor match well

with the hotel performance ranking. Thus, it can be concluded that bias introduced by a

single source is likely to be limited.

Regarding CMV attributed to common method (survey), most of the correlations

between independent and dependent variables were significantly related (20.43 #

r # 0.52). This raises the possibility that the observed relationships were inflated. In order

Table 1. A comparison of self-reported measures and industry data.

Means of measured variables Objective figures

Organization Work satisfaction Vigor

Intention to quit

Turnover in 2008 (%)

Rankings of city-level hotel performance (by average revenue per available room)a

1. Hotel A 4.57 4.28 2.73 21 1 2. Hotel B 4.22 3.80 3.08 30 2 3. Hotel C 4.17 3.83 3.42 42 3

Source: aChina Hotel Industry Study report (CHIS 2008).

X. Li et al.1830

to rule out the possibility that CMV is so large that this alters the key results, we conducted

a method-variance–marker-variable analysis proposed by Lindell and Whitney (2001).

The rationale for this is to compare the original correlations between independent and

dependent variables with those after controlling a theoretically irrelevant marker-variable

obtained by the same method. If the correlations stay significant and non-zero, the original

correlations observed cannot reasonably be accounted for by a common method factor. In

our study, individual prevention self-regulatory focus, defined as the extent to which

individuals use prevention strategies to reach their goals (Kark and Van Dijk 2007), was

used as a marker-variable. A partial correlation analysis, as reported in Table 2, shows that

the relationships between independent and dependent variables continue to have

significant and non-zero coefficients. Hence, it can be concluded that the bias originating

from the same method has limited influence on the relationships in this study.

Data analysis

The data consist of employees (n ¼ 810) nested in units (n ¼ 64), which are situated in

three hotels. As the variance in the three employee attitude measures is only slightly related

to the hotel level (intra class correlations (ICC(1)’s) are below 0.05), this level was not

taken into account (LeBreton and Senter 2008). This means that the data can be

conceptualized at two levels: employee (level 1) and unit (level 2). Level 1 refers to

individual employee information in each unit (work satisfaction, vigor, and intention to quit

and independent variables). Level 2 captures the variance between units (climate strength).

Accordingly, it is appropriate to employ hierarchical two-level modeling, which allows

simultaneous analysis of the effects of both within- and between unit-levels (Raudenbush

and Bryk 2002). Parameter estimates and chi-square information based on this analysis is

analogous to beta coefficients and R-square indicators in regression analysis. The deviance

in chi-square of two models can be used to judge whether there is significant model

improvement. The cross-level interactions needed to test the hypotheses H4–H6 were

calculated

Related Tags