Beyond HR Practices: Interacting HR Policies and

BEYOND HR PRACTICES: INTERACTING HR POLICIES AND FIRM PERFORMANCE 1

DRAFT AUGUST 2009, PLEASE DO NOT QUOTE Prepared for the Research Seminar, Australian Centre for Research in Employment and Work (ACREW) at Monash University, August 5 2009 ERIK POUTSMA & PAUL E.M. LIGTHART University of Nijmegen Nijmegen School of Management, P.O. box 9108, 6500 HK Nijmegen, the Netherlands JOHN DELERY Department of Management Sam M. Walton College of Business, University of Arkansas Business Building, Room 402 Fayetteville, AR 72701 BART DIETZ Erasmus University Rotterdam School of Management, P.O. Box 1738, NL-3000 Rotterdam, the Netherlands

Any comments? Æ [email protected]

1

The authors thank Chris Brewster, Ed Carberry, Mark Huselid, Wolfgang Mayerhofer, Lisa Nishi and Patrick Wright for helpful suggestions on earlier versions of this manuscript.

1

HR SYSTEMS AND FIRM PERFORMANCE:

ABSTRACT

To do Based on data from 1,492 firms across 12 European countries, the authors demonstrate that an HR system of control and commitment HR policies has a positive synergic effect on firm performance. However, the positive interaction effect only holds when the control HR policy is geared towards individual employees. The results are robust across countries and industries.

2

INTRODUCTION An insight into the relationship between HR and performance is essential for both practitioners and researchers (Rynes, Giluk & Brown, 2007). Scholars exploring this strategic HR area of “performance management” seek to guide managers in managing employee efforts and measuring performance outcomes, such that organizational performance is maximized (den Hartog, Boselie & Paauwe, 2004). Over a decade ago, performance management research arrived at a point where “bestpractice” or “universalistic” performance effects of HR practices were well established but also criticized, and a call to move beyond a best-practice mode of theorizing was made (e.g. Delery & Doty, 1996). This generated a massive sub-stream in the literature on performance management, proposing systems of HR practices to have synergic effects on organizational performance (e.g. Arthur, 1994; Huselid, 1995; Becker & Gerhardt, 1996; Delaney & Huselid, 1996). Scholars from the latter research stream propose numerous “ideal” systems of HR practices (i.e. bestsystems), or so called “High Performance Work Systems,” to lead to substantial improvements in firm performance over and beyond best practice HR approaches (Becker & Huselid, 1998). Taking stock of this field today, after more than a decade of theoretical and empirical work and despite some empirical evidence for the synergic performance effects of HR systems, significant conceptual and methodological issues remain and as a result the literature has to cope with “deficient empirical support” (Martín-Alcázar, Romero-Fernández & Sánchez-Gardey, 2005: 645). We have two arguments for this observation: firstly, while HR systems have typically been composed of HR practices (e.g. Ichniowski et al., 1997; Delery & Doty, 1996), the attributes of HR systems have recently been proposed as being conceptualized from additional levels of abstraction. Beyond the level of practices, HR attributes have recently been

3

theoretically classified into practices, policies and principles (Colbert, 2004); however , empirical studies on the performance effects of HR systems have merely been performed on practice-level HR systems. Consequently, empirical investigation of HR systems on the level of policies and principles remains unexplored. Scholars have called for research that takes this suue into account (Delery & Shaw, 2001; Arthur & Boyles, 2007). Secondly, while scholars have indeed proposed that HR attributes, when aligned in a coherent system, have synergic effects on firm performance, the theoretical rationale of these systems remains “under-explored” (e.g. Kinnie, Swart & Pyrcell, 2005). Delery & Doty (1996) have called upon HR scholars to adopt a “configurational mode of theorizing” and sparked a research stream in search of ideal HR systems (Becker & Huselid, 1998). Hence, while authors have argued that HR systems “more accurately reflect the multiple paths through which HR policies will influence successful strategy implementation” (Becker & Huselid, 1998: 55), and that HR systems should be “aligned,” (Becker & Huselid, 1997), represent “fit” (Wright & Snell, 1998) and need to be “configurational” (MacDuffie, 1995); a theory-driven rationale for the composition of HR systems is called upon (Delery & Shaw, 2001: 189). In exploring this theoretical rationale of HR systems, we investigate the different theoretical approaches to the management of the workforce. We draw on principle agency work to ‘individual performance focused policy’ consisting of several control and monitoring practices. We discuss resource dependency and commitment theory to discover a ‘policy focused on participation in decision making’ consisting of several information sharing practices. In addition we investigated incentive contract theory and property rights theory to denote a ‘collective organizational performance focused policy’ consisting of profit sharing and share ownership practices. This latest policy is also inspired by the recently developed shared

4

capitalism approach (Freeman et al., 2004). Following the debate on an overarching HR system versus different HR systems for different groups, we substantiate the debate by investigating the interactions of the different policies as possible HR subsystems adding another level of internal consistency of the HR system. We theoretically argue and show that HR individual performance policy and HR participation in decision making policy have a combined impact on organizational performance. The present study aims to advance the field by (a) providing a theoretical underpinning of HR as a system of policy-level attributes, and (b) empirically exploring the direct and synergic performance implications of control and commitment HR policies. The paper is structured as follows: firstly, we will provide a conceptual background on HR systems literature. Secondly, we will propose “system of attributes” theory as a theoretical foundation for HR systems. Thirdly, we will organize the HR systems literature into practice-level and policy-level systems. Finally we will develop hypotheses, describe our method, present results and discuss implications for managers and scholars.

CONCEPTUAL BACKGROUND In the mid 1990s, several HR scholars argued that HR is more likely to influence firm performance when conceptualized as a “system,” or “configuration” (e.g. MacDuffie, 1995, Delery & Doty, 1996). As a consequence, authors have since been developing numerous HR systems (for an overview, see: Becker & Huselid, 1998) and until the present, the extant body of literature on HR Systems was growing rapidly (e.g. Kintana, Alonso & Olaverri, 2006; Evans & Davis, 2005). A common theme in the HR systems literature is the exploration of performance effects of “ideal type” HR systems. Delery & Doty (1996) called this perspective the

5

configurational mode of theorizing and advocated “horizontal fit” between practices. The central idea is that some ideal type HR configurations lead to better performance then others. In their quest for empirical evidence for firm performance in these configurations, a broad variety of ideal designs have emerged, such as: “control” and “commitment”, “traditional” and “innovative,” “calculative” and “participative,” “internal” and “market type” HR systems (Arthur, 1994; Ichniowski, et al., 1997; Gooderham, Nordhaug & Ringdal, 1999; Delery & Doty, 1996). Apparently there is no consensus about the ideal types and its constituent practices. The often argued existence of two opposing ideal types may also be challenged. One of the early writers on this subject, Arthur (1994), suggested the existence of mixed systems where his discovered “control” and “commitment” systems may interact in generating positive performance effects (ibid.: 684). Whereas HR systems are typically hypothesized to generate positive effects on firm performance, and scholars have argued that “firms realize the largest gains in productivity by adopting clusters of complementary practices” (Ichniowski et al: 295), the broad stream of HR systems literature has yielded mixed and conflicting results (Delaney & Huselid, 1996; Delery & Doty, 1996; Ahmad & Schroeder, 2003). Scholars have been proposing multiple arguments for these unsatisfactory performance effects, such as the level of analysis or the measurement methods in operationalizing HR systems (Delery & Shaw, 2001). There are a number of reasons for this drawback. (1) A common theme in these lines of research, is that HR systems should be “aligned” so that synergy, fit, or complementarity can blossom (Delery & Doty, 1996). However, a theoretical framework guiding this coherence is lacking and hence, the alignment process is concentrated on each HR attribute separately. Consequently, HR variables are aligned sequentially and a “set of best practices”, instead of a coherent “best system” is the end result. (2) Few studies provide a theoretically

6

driven approach to the rationale behind the system (i.e. what constitutes “fit”?) and why specific practices are selected for certain HR systems. (3) The relationship between HR practices is weakly specified (e.g. substitutes or complements). Consequently, a theoretical framework linking HR variables to each other and predicting combined effects, such as possible substitutions and synergies, is lacking. (4) There has been little consistency in the literature concerning the measurement of HR practices in general and even less concerning the level of certain HRM practices in the architecture of the organization. A distinction between levels is called for (Kaarsemaker & Poutsma, 2006; Arthur & Boyles, 2007). As Becker and Gerhart (1996) pointed out, there are likely very few practices at a low level in the HR system architecture (e.g. a specific appraisal technique or a specific reward technique) that are universally effective, but higher level policies may have much more predictive power across all types of organizations and situations. (5) Lepak and Snell (1999) argued that, although intuitively appealing, it may be inappropriate to simplify the HR system and suggest that there exists a single “optimal” HR architecture for managing all employees. They showed that there may be different HR systems for different categories of personnel. Lepak, Takeuchi and Snell (2003) found also important interaction effects of different HR systems on performance. In what follows, we address these issues in more depth by substantiating the configurational mode and adding an additional level of abstraction (i.e. the HR policy level) to the debate on systems of practices.

HR AS A SYSTEM OF ATTRIBUTES In economic literature, organizations have been argued to be systems, composed of attributes (Milgrom & Roberts, 1990; Holmstrom & Milgrom, 1994; Ichniowski et al., 1997).

7

Within the system of attributes (SOA) paradigm, an attribute represents an organizational aspect. The attributes form a system together, because the payoffs associated with the level of one attribute depend on the levels of all the other attributes. Attributes are therefore interdependent. Applied to HR, an HR practice such as “selection” is an attribute of an HR system. The SOA view postulates, therefore, that HR systems are built up out of interdependent attributes that – in combination – form a system. The central premise of this perspective is that the payoffs of attribute choices (i.e. the firm performance effects of alternative HR actions) are dependent on the choices of other attributes. For instance, payoffs of HR strategies of the compensation and benefit “attribute” (e.g. team-based versus individual) are interdependent on the strategies of the staffing system (e.g. strong versus moderate screening for team skills). Recognizing that an HR system as a whole can yield varying payoffs, depending on the choice of attribute values, SOA literature argues that successful organizations have coherent systems of attributes (i.e. such that the attributes are complementary), whereas organizations typically fail when there is a misfit between the values of the attributes (Milgrom & Roberts, 1990). From an HR perspective, the challenge is to select attribute values that optimize firm performance. Following this line of reasoning, a fundamental principle in the perception of HR as a system of attributes is that coordination of this system adds to organizational effectiveness. HR practices (i.e. attributes) have been described to co-exist in four ways in an HR system (Delery, 1998): (1) additive: when the effectiveness of the HR system is the simple accumulation of the effectiveness of all HR practices; (2) substitutable: when the effectiveness of one HR practice can also be achieved by another HR practice; (3) negatively synergic: when the combined effectiveness of HR practices is less than the simple accumulation of the effectiveness of all HR practices; and (4) positively synergic: when the combined effectiveness of HR practices

8

is more then the simple accumulation of the effectiveness of all HR practices. The latter structure illustrates what HR researchers have been calling “horizontal fit” (Delery & Doty, 1996). The horizontal fit hypothesis proposes that coordination of HR practices can lead to complementarity in terms of the effects on firm performance (Milgrom & Roberts, 1990; Ichniowski & Shaw, 2003). For coordination towards a coherent system, it is important to establish the “objects” of coordination (i.e. HR system attributes). In line with this, researchers have been proposing to take an architectural approach to these HR system and its attributes (e.g. Wright & Snell, 1998; Colbert, 2004). BEYOND PRACTICES In their pursuit of “ideal” HR systems or configurations, researchers have generally focused on HR practices as system-attributes. Whereas HR systems have been formulated in terms of “many different ideal possibilities” (Martín-Alcázar, Romero-Fernández & SánchezGardey, 2005: 637), the level of abstraction of these systems has systematically been on the level of practices. Recently, however, scholars have suggested that HR systems can be theorized on multiple levels of abstraction, beyond HR practices (e.g. Colbert, 2004; Kaarsemaker & Poutsma, 2006). More specifically, Colbert (2004) distinguished between: (1) HR practices, at the lowest level of abstraction (e.g. variable compensation schemes, quality circles or newsletters), followed by (2) HR policies (e.g. team-based work systems, incentive pay or suggestion systems), and finally (3) HR principles, at the highest level of abstraction (e.g. the explicit communication of the necessity of employee participation in all aspects of a business). 2 Using these levels of abstraction as a lens for HR systems literature, the potential performance 2

In this paper, we only focus on HR practices and HR policies. HR principles are guidelines on how to treat and value people and function as higher-order guiding principles for the coordination of HR policies.

9

effects of HR policies (i.e. a bundled system of practices) and HR principles (i.e. a bundled system of policies) have received less research attention and remained largely unexplored in the extant literature. Practice-level Attributes On the lowest level of abstraction, we find HR attributes to exist as “HR practices.” This conceptualization is quite well-known in the HR literature (Pfeffer, 1994; Becker & Gerhart, 1996). From the perspective of the system of attributes theory, HR practices represent attributes. The wide range of HR practices has been classified by multiple authors. For instance, Stavrou (2005) categorizes HR practices into four domains: (a) Staffing, (b) Training and Development, (c) Compensations and Benefits and (d) Communication and Participation. Also, Schuler & Jackson (1987: 212) suggested a classification of: (a) Planning Choices, (b) Staffing, (c) Appraising, and (d) Compensating and Training. A common characteristic of these classes of HR practices is that they form the “available array to execute policies” (Colbert, 2004: 345). When perceived from the system of attributes perspective, (a) the budgets for training, and (b) intensity of personnel screening are two examples of practice-level attributes of an HR system (e.g. Ichniowski et al., 1997).

Policy-level Attributes HR policies serve as governing mechanisms that drive the selection of HR practices needed to execute these policies. From a cluster of attributes, or “group of activities” (Milgrom & Roberts, 1990) on the level of HR practices, higher order HR attributes can emerge. On this level of abstraction, therefore, a group of HR practices can form an HR policy. An HR policy is a system of practice attributes. On the level of HR policies, one is much more likely to find theoretical

10

rationale for the contemplation, selection and allocation of practices. When HR policies are the attributes of an HR system, coordination of this system (directed by an HR principle) becomes less complex due to the fact that HR policies serve as guidelines for HR practices. From a system perspective HR policies are considered subsystems of the HR system. We distinguish three distinct policies relevant for performance. We draw on principle agency work to distinguish the ‘individual performance focused policy’ consisting of several control and monitoring practices. To ensure optimum performance of workers employers employ performance management to control and to monitor the behavior of workers. To employ this policy a variety of practices can be used, such as performance appraisal and performance related pay. In addition, workers should have the ability to perform and part of the policy is investment in training and assessment of the effectiveness of training. Since the work is becoming more complex workers tend to have tacit knowledge and skills as important resources for the company. This is specially pertinent now because of greater levels of private information which reside with employees (Levine and Tyson, 1990). Also, workers are usually in the position to detect inefficiencies in operations that diminish productivity. They are also likely to acquire important information concerning the actual productive contributions of co-workers. The information derived from such activity is potentially very valuable to the firm as an input to production. Yet such information transfers will not be induced under an individual performance-based reward or appraisal practices. Commitment (Meyer and Allen, 1997) and goal-setting theory (Locke et al., 1981) suggest elaborate communication policy to discuss the important objectives of the company and how workers can contribute. Employers need a ‘policy focused on participation in decision making’ consisting of several information sharing practices.

11

A third policy for performance we label ‘collective organizational performance focused policy’ consisting of profit sharing and share ownership practices. Incentive contract theory (Lazear, 1986) and property rights theory (Milgrom and Roberts, 1992) indicate that the concept of ownership, combined with statutory property rights are the fundamental means to provide an incentive to create and to develop an asset. The two fundamental aspects of ownership include, firstly, the ‘residual rights of control’, which is the right to make decisions concerning the use of an asset, secondly, the ‘right of residual returns’, which is the right to revenues left over after all obligations have been met. Ben-Ner and Jones (1995) develop a theoretical framework which combine these two aspects of ownership, control and return, and suggest possible performance outcomes associated with transferring these rights from owners to non-owner employees. They contend that the greatest efficiency outcomes exist when both these rights are transferred from owners to non-owners. In the following section, we develop hypotheses for the performance implications, and explore: (a) performance effects of HR policies (i.e. direct effects), and (b) complementary performance effects between HR policies (i.e. interaction effects).

HYPOTHESES Generic Effects Individual performance focused policy. Performance management aims to improve organizational performance by aligning behaviors of individual employees with organizational goals. Connecting employee rewards with the firm’s output criteria has been noted as a key lever in achieving this (Arthur, 1994). Indeed, performance management policy is efficiency seeking by nature, and intend to supply production activities with the necessary input of human resources

12

at the lowest possible cost (Gooderham et al., 1999: 510). From this perspective, individual human behaviors and investments in HR are strictly monitored. Consequently, deviations from pre-set norms are relatively transparent and simple for management to act upon. A individuakl performance focused HR policy is likely, therefore, to funnel workforce efforts into efficient behaviors, while at the same time behaviors that are undesirable for firm performance are a transparent target for management intervention. We posit that a individual performance focused HR policy will increase a firm’s effective use of individual workers and we offer the following hypothesis:

Hypothesis 1. An individual performance focused HR policy has a positive effect on firm performance.

As we enter an era in which coordination between employees with tacit knowledge and skills becomes more and more important to achieve firm performance, aligning the behaviors of employees collectively becomes more important (Miller & Schuster, 1987; Kidwell & Bennett, 1993) 3 . In addition, there is empirical evidence (in the financial participation and shared capitalism literature) that collective alignment HR policies, such as profit sharing and employee share ownership positively impact firm performance. Indeed, numerous studies report these findings (e.g. Conte & Svejnar, 1990; Kruse, 2002; Kruse & Blasi, 1995; Sesil et al., 2001). When organizations have a collective alignment HR policy, it is likely that employees will be

3

The distinction between an individual performance focused policy and a is also supported by the possible difference in underlying theoretical principles. Individual performance focus policy is related to the principle of valuing and treating employees as providing labor input that should be controlled for opportunistic behavior. Collective control is related to the principle where employees are valued and treated as stakeholders or co-owners of the firm.

13

more inclined to engage in more efficient teamwork behaviors, and will for instance become more reflexive (Schippers et al., 2007). We therefore offer the following hypothesis:

Hypothesis 2. A collective organisational performance focused HR policy has a positive effect on firm performance.

Participation in decision making focused HR Policy. This policy aims to forge psychological links between organizational and employee goals in order to create employee commitment towards organizational goals. We posit that this policy requires clear and consistent communication of organizational goals towards employees: not only to inform employees, but also to signal and communicate a culture of partnership. Hence, within this HR policy, employees are viewed as active partners, core assets and even “participants in a project premised on commitment, communication and collaboration.” (Gooderham et al., 1999). More recently, Ordiz-Fuertes & Fernández-Sánchez (2003) found “high-involvement” practices to be adopted by high performing firms. A participation in decision making policy is based on an emphasis on management-workforce communication, top-down and bottom-up. Consequently, individuals are better informed of the goals of the organization. This is likely to increase commitment to these goals, as well as motivate alignment between organizational and individual goals. In addition, consulting the views of employees increases the effectiveness of decision making and hence contributes to dynamic efficiency 4 . We therefore offer the following hypothesis:

Hypothesis 3. A participation in decision making focused HR policy has a positive effect on firm performance. 4

Commitment policy relates also to culture as a mechanism of social control or normative control (Kunda, 1992).

14

Synergic Effects In strategic management literature it has long been recognized that complementary resources can help a firm attain greater competitive advantage (Barney, 1995). Indeed, it is also argued in HR that “the variation in firm performance explained by an HR system should be significantly greater than that explained by the individual HR practices in that system” (Ahmad & Schroeder, 2003). From the perspective of HR as a system of attributes, we posit that an HR system can be conceived as a system in which HR policies co-exist. Thus, we conjecture that individual performance oriented policy and the HR policy focused on participation in decision making are candidates for complementary effects. We expect coordination between these two attributes of strategic HR to lead to synergic effects. On the one hand, when an HR policy is highly controlling, economic theory argues that people will be more inclined to focus on those behaviors that lead to incentives (Eisenhardt, 1989). In other words, controlling behavior may have cost implications. On the other hand, when an HR policy is focused on participation, the workforce is well-informed about corporate goals and employees have a better insight into how their behaviors can help to realize the objectives. We argue, therefore, that individual performance focused HR policy are more likely to lead to performance when combined with participation in decision making. Building on the concept of complementarity theory (Milgrom & Roberts, 1990), we posit that these two HR policies have interactive effects. First, in an efficient equilibrium situation, both HR policies are complementary, and synergy between the two takes place (Milgrom & Roberts, 1990). Milgrom & Roberts (1995) proposed the notion of complementarities and defined activities as complementary if: “doing more of any one of them increases the returns of doing more of the

15

others (p. 181).” Agency theory suggests just that. It prescribes that employees’ behaviors in line with incentives is more likely to occur when the employees are well-informed about the incentives (Eisenhardt, 1989). Kinnie, Hutchinson & Purcell (2000) demonstrated effects of “fun and surveillance” in call centers, a synergic effect of strategies of control and commitment. We therefore posit that implementing a more commitment-oriented HR policy impacts on the returns of a more controlling HR policy. In terms of Agency Theory: an outcome-oriented contract becomes more efficient in combination with information (a “purchasable commodity”) about the outcomes that are expected. We therefore offer the following hypothesis:

Hypothesis 3. The effectiveness of an individual performance focused HR policy on firm performance is increased by a participation in decision making policy.

In the financial participation literature on the impact of collective incentive schemes, the issue of complementarity between financial participation and other forms of participation has received much attention (e.g. Kalmi, Pendleton & Poutsma, 2006). There are several reasons why HR policies that are highly based on financial participation can be complementary to a participation in decision making policy. In themselves, financial participation plans are vulnerable to a freerider effect: each employee may rely on other workers to deliver the enhanced output and performance necessary to bring about the incentive payments. This is likely to be a significant limitation of financial participation plans in all but the smallest work environment (e.g. Oyer, 2004). Participation schemes may mitigate the free-rider problem by encouraging a co-operative corporate culture (Weitzman & Kruse, 1990) and/or mutual monitoring (Kandel & Lazear, 1992; Freeman, Kruse & Blasi, 2004). At the same time, financial participation can provide an

16

incentive for employees to share information, thereby contributing to the effectiveness of work groups, and related activities such as quality circles, (e.g. Ben-Ner & Jones, 1995; MacDuffie, 1995). We therefore formulate the following hypothesis:

Hypothesis 4. The effectiveness of a collective organizational performance focused HR policy on firm performance is increased by a participation in decision making HR policy.

Here the theoretical rationale for the three way interaction.

METHOD Data

The sample of the present study was borrowed from the Euronet-CRANET survey on Human Resource Management (i.e. CRANET), in 1999. The goal of this survey was to draw representative national samples of multiple countries in Europe. The strategy was to mail out a well-translated survey to HR managers of medium and large-scale firms in the private sector (100 or more employees). In the survey, HR managers were asked to respond on items that operationalized organizational HR practices and performances of the firm. Firms participating in this survey were independent single-establishment businesses, as well as foreign subsidiaries (for a detailed description of the sampling procedure, see: Brewster et al., 2004). Respondents were asked to return their completed survey via mail to the coordinating business school in each specific country. Researchers in all countries distributed their surveys to the CRANET coordinating office in the U.K. for data-entry. Following Gooderham (1998), we

17

only included countries that were European Union (EU) member states. For the databases per country, firms that were: (a) public or semi-public, and (b) employed less than 100 employees were excluded from further analysis. While response rates for the individual countries were relatively low and generally varied between 12 and 35 percent, analyses indicated no nonresponse bias. Table 1 presents an overview of the number of firms for the main explanatory factors and their categories within the dataset..

------------------------------Insert Table 1 about here -------------------------------

Operationalization

HR policies.

Firstly, we distinguished three operationalizations of HR policies: the individual performance HR policy, the collective alignment HR policy, and the commitment HR policy (See appendix A for all scale items used in the study). Individual performance HR policy: Based on the principle of monitoring and control of individual performance a “Control” HR system was developed. Following Gooderham et al., (1999) an individual performance HR policy is defined as consisting of the following attributes: (1) performance appraisal, (2) formal evaluation of training, and (3) individual performancerelated pay. This operationalization of the individual performance policy captures the extent to which firms formally monitor and evaluate training efforts across all individuals as well as individual performance management.

18

Collective alignment HR policy: This is based on the principle that employees have, as stakeholders, control and return rights (Ben Ner & Jones, 1995). Where employees receive collective returns and co-ownership, an “Ownership” HR system is developed. It covers such practices as profit sharing and employee share ownership. Collective alignment policy consists of: (1) employee share option plans, and (2) profit sharing schemes. Commitment HR policy: Based on the principle that human resources are an important asset with tacit knowledge and skills, a “Commitment” HR system was developed, where employees are informed about the business case, express their views and thereby are enabled to make joint decisions. Commitment HR policy consists of: (1) strategy briefings, (2) performance briefings, (3) written mission statement, (4) written employee communication policies, and different ways in which employees express their views to management (5) through regular workforce meetings, (6) team briefings, and (7) attitude surveys. Note that the items on the commitment HR policy scale mainly cover information sharing and do not indicate participation in decision making. We did not include structural work organization features like teamwork and representative participation, as these neither tell us much about the actual participation in decision making, nor are they a necessary precondition. Mutual information sharing is a necessary precondition to cooperation and collaboration. 5 Firm performance. Secondly, we operationalized our dependent variable: firm performance. Following Laursen & Foss (2003), we adopted multiple indicators to define firm performance. We included financial indicators: (1) gross revenue, (2) stock market performance, and (3) profitability, as well as non-financial indicators: (4) innovation rate, (5) productivity, (6) service quality, and (7) time to market. The performance measure is relative to other 5 Unlike Wood and Fenton-O’Creevy (2005) we did not emphasize in our analyses the structures for commitment. We approached the issue of commitment from the angle of the use of commitment policy by management regardless of the channels they may use. However, we checked any association in our dataset between commitment policy on the one hand and the existence of works councils and the recognition of trade unions by management and both relationships were positive.

19

organizations in the firm’s sector. The use of perceptual measures of firm performance is consistent with prior research in Strategic HR (e.g. Delaney & Huselid, 1996; Jap, 2001; Stavrou, 2005). 6 While we employed a subjective operationalization of firm performance, we did not compromise convergent, discriminant and construct validity (Wall, Michie, Patterson, Wood, Sheehan, Clegg & West, 2004). The advantage of this subjective measure is that it is not distorted by taxation differences and earnings management influences. In addition, the overall performance measure better reflects the possible impact of human capital, motivation, commitment and social capital than a single “objective” financial indicator. Control Variables. We used two types of control variables: (a) institutional factors: country and unionization, and (b) non-institutional factors: industry and firm size. The country factor specified twelve member states of the European Union (EU) in 1999. Unionization was measured as the percentage of the workforce that was a member of a labor union (Mean: 40.8%; SD: 33.6%). The industry factor indicated six basic industries: (1) Construction, (2) Transportation, (3) Banking & Finance, (4) Chemicals, (5) Manufacturing, and (6) Other. Firm size was measured as the logarithm of the number of reported employees (Median: 450 employees). Scaling Based on the survey, we constructed scales for each HR-policy, based on multiple indicators using Cronbach’s alpha (for reliability) and the more restrictive scaling procedure of Mokken’s nonparametric latent trait model (Mokken & Lewis, 1982; Molenaar & Sijtsma, 2000). We used the “Mokken Scaling Program” (MSP; Molenaar & Sijtsma, 2000). Mokken’s scaling approach, i.e. a probabilistic version of the deterministic Guttman model, was used to determine the degree to which the prevalence of a set of HR-practices in firms can be explained by a cumulative latent 6

For a detailed argumentation for the use of perceptual measures of firm performance in international HR research, see: Stavrou (2005).

20

trait of these firms. A cumulative latent trait can be inferred when the probability of a positive response to an item increases monotonically as the latent trait of the respondent also increases. That means that firms applying more “difficult” HR-practices are expected to also make use of easier or more common HR-practices. In contrast to reliability analysis that assumes unidimensionality, Mokken’s approach assesses the uni-dimensionality of a pair of items and the scale directly by calculating an internal scaling criterion, or so called: “Loevinger’s Hcoefficient”. The Loevinger’s H-coefficient (H) signifies the deviation of the observed data structure of the scale from the perfect scalogram structure as incorporated in Guttman’s approach. Following Mokken (1971), Molenaar & Sijtsma (2000) considered a set of items as a “weak” scale if 0.3