تمایز نیروی کار: اثرات عملکرد در استفاده از نیروی کار مشروط در زمینه سیستم کار با عملکرد بالا

In this study we focus on the impact of contingent labor on the outcomes of high-performance work systems (HPWS). Building on the emerging research on the social mechanisms linking HRM to organizational effectiveness, we argue that a higher incidence of contingent labor diminishes the productivity payoff associated with the use of HPWS for managing standard employees. We test these arguments using a sample of 229 British firms of different industries. The results support our arguments and help develop a more holistic and critical viewpoint in the analysis of workforce differentiation.

One of the most common ways to differentiate the workforce is by distinguishing between “standard” and “contingent” employees (Broschak & Davis-Blake, 2006). The differences between the two types of employees reside in two fundamental criteria: the duration of their employment relationship and the quantity of HR investments the firm directs to them (Kalleberg, 2001). While standard employees have an ongoing employment relationship and receive a number of HR investments (e.g., economic incentives, training, empowerment opportunities), contingent ones have employment relationships of limited duration and generally receive only minor HR investments from the employer. The most common types of contingent labor are fixed-term, temporary, and agency workers (Polivka & Nardone, 1989); these are the types we focus on in this paper. Both standard and contingent employees deliver benefits to the firm. Through the use of standard employment arrangements and HR investments, firms may boost firm productivity and profitability by fostering the commitment, skill level and flexibility of the employees in these arrangements (Lepak & Snell, 1999). Contingent employment, on the other hand, provides more numerical flexibility (Kalleberg, 2001) and access to new knowledge (Vogus & Welbourne, 2003), at the same time as it helps to achieve a range of labor cost savings (Cappelli & Neumark, 2004). That being said, the advantages associated with these two employee groups should not be analyzed solely in isolation. Rather, it is also essential to analyze whether these groups may be mixed, as well as the extent to which their mixing is beneficial for firms (Cappelli & Neumark, 2004). Following this direction of analysis and focusing on the simultaneous use of standard and contingent employees, it has been shown that such a combination is not without costs. On the contrary, it may reduce trust (Pearce, 1993), loyalty (Davis-Blake, Broschak, & George, 2003), commitment (George, 2003), and helping behaviors (Broschak & Davis-Blake, 2006) among the standard workforce. Moreover, some authors found that the combination of standard and contingent labor is not necessarily beneficial for firms' final performance (Roca-Puig, Beltrán-Martín, Bou-Llusar, & Escrig-Tena, 2008). In the present study we aim at expanding on this research to explore the impact of contingent labor on the effectiveness of the HR practices used to manage standard employees. Specifically, based on the abundant literature considering so-called high-performance work systems (HPWS) as the most effective initiative for managing the standard workforce (e.g., Huselid, 1995), we pose the following research question: Are HPWS more effective when used in contexts where a contingent workforce is also deployed, or in contexts of a uniform culture where the entire workforce consists of standard employees? To date, empirical research has not sufficiently addressed this question. Yet, for organizations wishing to improve their productivity, such a question is certainly relevant. Indeed, by exploring this question one can assess the extent to which the highly acclaimed tendency to differentiate the workforce (e.g., Becker et al., 2009 and Lepak et al., 2003) may have downsides and counterweights. Based on emerging research on the social mediators of the HRM–organizational effectiveness relationship (e.g., Collins and Smith, 2006, Evans and Davis, 2005, Paré and Tremblay, 2007 and Takeuchi et al., 2007) and using data from the Workplace Employment Relations Survey 2004 (WERS2004), in this work we aim at identifying and describing some of these downsides by presenting empirical evidence from British firms. Thus, the first contribution of our study lies in extending the research on workforce differentiation with an empirical examination of some of its potential drawbacks. While previous studies on this topic have highlighted the consequences of contingent labor deployment for standard employees' attitudes and behaviors, we move a step further by focusing on how the use of contingent labor interacts with important strategic HR practices (i.e., HPWS) for standard employees. Second, in so doing we are also able to gain further insights into the HPWS–performance relationship and add to literature exploring from a contingency perspective the organizational circumstances under which such systems are valuable.

Contingent employment is a “low-cost” HR strategy that enables a firm to rapidly access a cheaper workforce, reduce its fixed costs, and manage its capacity more efficiently. In this work we have shown that such a strategy is not without downsides. More concretely, we have analyzed the impact that the deployment of contingent labor has on the effectiveness of HPWS for standard employees. We posited that the presence of contingent workers may hinder the deployment of important mechanisms through which HPWS operate, namely OCBs, cooperation, and shared mental models and languages. Our findings indicate that the effect of HPWS on productivity is lower—if not negative—in workplaces that use higher levels of contingent labor, thus providing empirical support for our arguments. These results are consistent for both of the productivity measures considered. The study thus adds to previous research that explored other potential downsides of contingent labor deployment. Overall, our findings call into question the arguments of those that advocate for mixed work arrangements as a way to improve firm efficiency (e.g., Kalleberg, 2001 and Lepak et al., 2003). More generally, our findings suggest that the so-called “high-road” HR practices, which emphasize continual investment in the (standard) workforce and of which HPWS are the quintessential expression, may be jeopardized if used in conjunction with “low-road” HR practices such as the use of contingent employment. The study contributes in parallel also to the literature which aims to model the link between HPWS and their intended outcomes. Thus far two main views, a universal approach and a contingency approach, have been developed in this regard ( Datta et al., 2005). The universal perspective implies a direct relationship between HPWS and performance, whereas the contingency perspective posits that an organization's context moderates the effect of such systems on performance. Our results provide support for the contingency perspective by suggesting that the proportion of contingent workers in the company significantly influences the degree of HPWS impact on labor productivity. Our results also shed light on the effects of contingent employment on productivity. Overall, previous research has been inconclusive regarding the performance effects of contingent labor, since it has highlighted its potential risks along with its benefits (Blyton and Morris, 1992 and Cardon, 2003). Some authors have suggested that the performance effects of contingent employment may be non-linear, so that there is an optimum temporality level (Broschak and Davis-Blake, 2006 and Cardon, 2003). Consistent with these findings, our results show no significant linear effects (models 1 and 2). We also extend previous research by showing that the productivity effects of contingent workers are contingent on the level of HPWS. Applying the simple slope tests proposed by Dawson and Richter (2006) as a post hoc analysis, we found that the average effect of temporary work agreements on value added per employee (model 4) was positive and significant when HPWS < − 0.1, non-significant when − 0.1 < HPWS < 1.4 and negative and significant when HPWS > 1.4. In other words, contingent employment contributes positively to productivity for under-average levels of high-performance work practices, and negatively just for very high levels of HPWS (nearly one and a half standard deviations over the sample mean). Although our findings highlight a counterweight of a workforce differentiation strategy, we can point to two reasons why they do not delegitimize this strategy. First, our analysis shows that the positive association between HPWS and productivity still holds, if to a lesser extent, in the presence of a moderate incidence of contingent employment; it is at high levels of contingent labor that the effect becomes negative. Second, HPWS and contingent labor may positively interact to improve other outcomes not considered in our study. For example, as both HPWS and contingent employment may help reduce labor expense (Guthrie et al., 2009), it is possible that when they are used in conjunction this outcome will further improve. Moreover, although our findings suggest that fully reaping the productivity pay-off of HPWS may be difficult if contingent employment is present, they do not exclude that at least some firms may be able to do so. Arguably, this depends on several issues: What kind of positions do contingent workers have? How does the firm manage the relationships between contingent and standard employees? How do internal and external contextual factors influence such relationships? The exploration of such questions, however, lies well beyond the scope of our paper. Nonetheless, the implications for managers of our findings seem clear: Even in the presence of sound reasons for deploying contingent labor, its deployment must be considered in relation to the potential negative impact it may have on the effectiveness of HPWS for standard employees. We have shown that such a negative impact appears to be ignored by a number of firms in our sample. The study is not without limitations, most of which arise from the data set used for the empirical analysis. First, the cross-sectional nature of the data calls for a word of caution when interpreting empirical evidence in causal terms, although the relations we tested were based on prior theorizing. In this regard, the main issues are potential simultaneity and reverse causality between HPWS and workplace performance (Huselid, 1995 and Takeuchi et al., 2007). In order to test for possible simultaneity bias, we re-estimated our empirical models by 2SLS and conducted a Hausman–Wu test of endogeneity of the HPWS variable. We could not reject the null hypothesis of exogeneity (χ2 (1) = 0.987, p-value = 0.32). In other terms, we found no conclusive evidence showing that HPWS is endogenous with respect to labor productivity. As Huselid noted, simultaneity bias “is less probable in the case of turnover and productivity, because these variables would be unlikely to widely influence the selection of High Performance Work Practices” (1995: 640). Thus, our conclusions remain valid, even more so since the 2SLS also produced significant evidence supporting the hypothesized relationships, with somewhat larger coefficients. Second, although WERS2004 has the advantage of being validated by the important published works it gave rise to in the field of HRM and Industrial Relations, it has the disadvantage of not including specific measures of those mechanisms described in this study through which HPWS operate. It also does not provide specifics that would allow direct measurement of the effect of contingent labor on such mechanisms. As a result, we were limited in that we could not directly test some of the causal relationships that we addressed. It is worth noting, though, that prior research on which we built our study does provide strong theoretical and empirical support to the arguments we proposed, attenuating the non-measurement issues that may arise. Finally, WERS2004 is limited to the United Kingdom. As the effectiveness of HR practices may depend upon the socio-institutional context in which they are used (Werner, 2011), it may not be appropriate to generalize the present research beyond British workplaces. All of these issues may certainly inspire future research. Future investigations may also extend our study by including other performance indicators, such as quality or innovation. Furthermore, we included fixed-term, temporary and agency workers in our measure of contingent employment, but other forms of contingent labor also exist (e.g., outsourcing arrangements), and their effects on HPWS effectiveness may be different. Accordingly, it may be informative to extend our analysis to them. Moreover, contingent employment may have different impacts on HPWS effectiveness depending on whether the aim of using it is (a) to improve standard employee employment stability or (b) to reduce labor costs. It is possible that these different uses may also change the impact of contingent employment on HPWS effectiveness as productivity enhancement devices; this may be a worthwhile issue to be analyzed in the future. Finally, it may be informative to investigate whether specific HPWS can be designed to cater to the needs of organizations that make use of contingent employees. Clearly much work remains to arrive at a full understanding of how HPWS and contingent employment interact and affect firm performance. Nevertheless, we hope that our study contributes to a better comprehension of the relationships between these variables, as well as suggesting ways in which firms might deploy workforce differentiation strategies more successfully.