https://doi.org/10.1177/0149206319880957

Journal of Management Vol. 47 No. 4, April 2021 993 –1023

DOI: 10.1177/0149206319880957 © The Author(s) 2019

Article reuse guidelines: sagepub.com/journals-permissions

993

The Role of Firm Size and Knowledge Intensity in the Performance Effects of Collective Turnover

Kim De Meulenaere Sophie De Winne

KU Leuven

Elise Marescaux IESEG School of Management

LEM-CNRS 9221

Stijn Vanormelingen KU Leuven

As employees are among firms’ most important resources and labor markets are facing serious labor shortages, firm-level collective turnover is one of the most important challenges facing organizations. Context-emergent turnover theory provides a theoretical framework for the per- formance implications of collective turnover and argues that context, and in particular, firm size, plays a crucial role in the collective turnover–performance relationship. Yet, the moderat- ing role of firm size remains undertheorized, empirically understudied, and thus, unclear. Based on the resource-based view of the firm, we develop a theoretical framework for two competing perspectives (a negative and a positive one) on the role of firm size and put forward the firm’s knowledge intensity as a crucial additional moderator. The main premise is that whereas firm size determines what resources firms have to successfully cope with turnover, knowledge inten- sity determines the resources firms need to do so. We propose a three-way interaction, suggest- ing that firm size reinforces the harmful effect of turnover in highly knowledge-intensive firms and buffers it in firms with low levels of knowledge intensity. Using a unique multi-industry and longitudinal administrative data set of 6,913 Belgian firms (2012–2016), we find support for

Acknowledgments: This research was supported by the Research Foundation–Flanders, Grant Numbers G.0661.10N (Research Project), G.0697.16N (Research Project), and 12Z7519N (Postdoctoral Research Fellowship). We are deeply grateful to the editor, David Allen, and the anonymous reviewers of Journal of Management for taking the time to provide highly valuable comments and suggestions to improve our manuscript. We also thank Prof. Dewael- heyns for providing us valuable information on accounting differences between large and small firms.

Supplemental material for this article is available with the manuscript on the JOM website.

Corresponding author: Kim De Meulenaere, KU Leuven, Korte Nieuwstraat 33, 2000 Antwerpen, 2000, Belgium.

E-mail: [email protected]

880957 JOMXXX10.1177/0149206319880957Journal of ManagementDe Meulenaere et al. / Collective Turnover, Firm Size, and Knowledge Intensity research-article2019

994 Journal of Management / April 2021

these assumptions. This study highlights the importance of the context in which firms have to deal with turnover, and it spurs researchers to go beyond studying turnover in narrow study contexts, to take into account the interplay among different but intertwined organizational con- tingencies, and to acknowledge both the quantitative (how many employees leave) and qualita- tive components (who leaves) of turnover.

Keywords: collective turnover; context-emergent turnover theory; firm size; knowledge inten- sity; firm performance

Firm-level collective turnover, or “the aggregate levels of departures that occur within . . . organizations” (Hausknecht & Trevor, 2011: 353), creates major challenges for the function- ing of firms. Coping effectively with employee departures is ever more important for firms’ success and survival in this era in which labor markets are confronted with labor shortages (Hancock, Allen, Bosco, McDaniel, & Pierce, 2013). As a result, the number of empirical studies on the consequences of collective turnover for firm performance has grown consider- ably (for recent overviews, see Hancock et al., 2013; Hancock, Allen, & Soelberg, 2017; Heavey, Holwerda, & Hausknecht, 2013; Park & Shaw, 2013). These studies generally con- clude that high collective turnover is detrimental. It can cause multiple disruptions for firms’ operations due to the depletion of human capital, the loss of social capital, the difficulties facing firm members with newcomer socialization, and the high costs of replacements (Hancock et al., 2017; Hom, Lee, Shaw, & Hausknecht, 2017).

Yet, as supported by the wide variety in effect sizes in empirical research, similar col- lective turnover rates are not equally harmful to all firms (Allen, Hancock, Vardaman, & Mckee, 2014; Hancock et al., 2013; Hom et al., 2017). This is also the basic premise of context-emergent turnover theory (CETT), stating that contextual factors may be at play (Hausknecht & Holwerda, 2013; Nyberg & Ployhart, 2013). Turnover scholars have repeat- edly mentioned firm size as one of the key organizational contingencies, as it determines the resources firms have available to manage the disruptions caused by turnover (Hancock et al., 2017; Hausknecht, Trevor, & Howard, 2009). However, its actual role remains unclear as two competing, yet undertheorized, perspectives can be considered. Accordingly, the limited empirical literature has reported evidence for a negative moderation (Hancock et al., 2013), a positive one (Park & Shaw, 2013), and even no moderation at all (Hancock et al., 2017).

With the present study, we want to unravel the role of firm size. We therefore rely on a resource-based perspective and argue that a distinction should be made between the resources firms have and need to successfully deal with turnover. Firm size relates to the amount and type of resources (i.e., human, financial, social, and organizational) that firms have and that they can deploy to deal efficiently with the disruptions caused by turnover. However, the nature of the turnover disruptions and, thus, the resources needed to manage them depend on firms’ knowledge intensity, that is, the extent to which firms rely deeply upon an extensive body of complex human resources (i.e., knowledge) for realizing their core value-creation activities (Von Nordenflycht, 2010). It reflects the quality of the human resources lost with turnover, the complexity of employee interactions within a firm, and the depth and intricate- ness of the products or services provided by the firm (Datta, Guthrie, & Wright, 2005; Dess

De Meulenaere et al. / Collective Turnover, Firm Size, and Knowledge Intensity 995

& Shaw, 2001). Consequently, it determines the specific implications of turnover for the firm’s human and social capital and its operations (Nyberg & Ployhart, 2013) and therefore the resources needed to cope with them. Thus, to fully understand the role of firm size, we put forward knowledge intensity as a crucial additional moderator.

Throughout this article, we first provide theoretical grounding for two competing perspec- tives on the role of firm size, a positive and a negative one, that have thus far been addressed only in a limited and intuitive way (e.g., Hancock et al., 2013, 2017; Hausknecht et al., 2009; Shaw, Park, & Kim, 2013). The positive perspective implies that larger firms can buffer the negative turnover effects because they have more human, financial, and organizational resources to realize quick and effective replacements (Hancock et al., 2013; Park & Shaw, 2013). The negative perspective argues that larger firms lack a high-quality and dense struc- ture of social resources, which thwarts the socialization and adaptation of the altered work- force (Hausknecht et al., 2009). We then argue that firm size can have both a buffering and reinforcing influence on the harmful performance effects of collective turnover, depending on the firm’s knowledge intensity. It determines whether resources for quick and effective replacements or for successful socialization and adaptation are most relevant to cope suc- cessfully with collective turnover and, therefore, whether the positive or the negative per- spective of firm size prevails. We thus propose a three-way interaction, suggesting that firm size buffers the harmful effect of turnover in firms with a low level of knowledge intensity and reinforces it in highly knowledge-intensive firms. Using a unique multi-industry and longitudinal administrative data set of 6,913 Belgian firms (2012–2016), we find support for these assumptions.

This study contributes to the literature in several ways. First, while most turnover studies have controlled for the number of employees, we show that firm size has a more fundamental meaning as a moderator in the collective turnover–performance relationship. Understanding the role of firm size is essential as (a) it is one of the most crucial contextual features of the firm that shapes the basic context in which organizations have to deal with turnover (Hancock et al., 2013; Park & Shaw, 2013), and (b) firm-level moderators in the collective turnover– performance relationship have been largely neglected, though considered highly important, in previous studies (Brymer & Sirmon, 2018; Shaw, 2011). Further in support of CETT, we show that the quantity of turnover alone cannot explain its consequences (Nyberg & Ployhart, 2013) but that the quality (in terms of the knowledge intensity of the workforce and, thus, of the departing employees) needs to be taken into account. As such, this study shows that the interplay of contextual moderators is essential to understanding the implications of collective turnover, which goes beyond the tradition of examining separate effects of individual mod- erators (Hausknecht & Holwerda, 2013).

Theory and Hypotheses

The Effect of Firm-Level Collective Turnover on Firm Performance

Throughout the years, an increasing number of studies on the performance consequences of firm-level collective turnover has been published (e.g., Call, Nyberg, Ployhart, & Weekley, 2015; Guthrie, 2001; Hale, Ployhart, & Shepherd, 2016; Kacmar, Andrews, Van Rooy, Steilberg, & Cerrone, 2006; Shaw, Duffy, Johnson, & Lockhart, 2005). To conclude on the main pattern of findings, several authors executed reviews and meta-analyses (Hancock

996 Journal of Management / April 2021

et al., 2013, 2017; Hausknecht & Trevor, 2011; Heavey et al., 2013; Park & Shaw, 2013; Shaw, 2011). They predominantly document a negative effect. In line with CETT, they estab- lish that the key mechanisms through which this relationship holds go beyond the simple loss of human capital that may well explain the effects of an individual exit (Hausknecht & Holwerda, 2013; Nyberg & Ployhart, 2013). Specifically, collective turnover also induces the depletion of social capital and the disruption of organizational processes caused by the depar- ture of employees (Hancock et al., 2013; Heavey et al., 2013; Mawdsley & Somaya, 2016). We briefly discuss these mechanisms next.

First, human capital refers to the variety of the set of knowledge, skills, and abilities (KSAs) that individual firm members have acquired through their work experience, training, and education (G. Becker, 1975; Mawdsley & Somaya, 2016). Employees’ human capital can add significantly to organizations’ competitive advantage, particularly if it is difficult to imitate and to transfer to other firms (Barney, 1991; Coff & Kryscynski, 2011; Hausknecht & Trevor, 2011; Lepak & Snell, 1999; Nyberg & Ployhart, 2013; Shaw et al., 2013). From a human capital perspective, turnover can therefore be detrimental to workforce productivity by causing the loss of highly valuable human capital (Dess & Shaw, 2001; Hancock et al., 2013; Heavey et al., 2013; Nyberg & Ployhart, 2013; Shaw, 2011; Shaw et al., 2013).

Second, turnover may also cause the loss of valuable social or relational capital—that is, the capital embedded in employees’ social relationships developed over time, in which human capital resources, such as knowledge and expertise, are shared and integrated (Dess & Shaw, 2001; Hancock et al., 2013; Mawdsley & Somaya, 2016; Nahapiet & Ghoshal, 1998; Nyberg & Ployhart, 2013; Shaw, Duffy, et al., 2005). These social relationships can be built both within a firm through close collaborations between employees on the work floor (i.e., internal social capital) and across the boundaries of the firm between firm members and external agents, such as clients, experts, partners, and professional networks (i.e., external social capital; Mawdsley & Somaya, 2016). A number of benefits have been associated with the accumulation of social capital in organizations, such as increased knowledge sharing and learning (Mawdsley & Somaya, 2016), enhanced intellectual capital, fostered communica- tion efficiency, employee trust and commitment (Leana & Van Buren, 1999), and improved investments by employees in their coworkers (Nyberg & Ployhart, 2013), all of which con- tribute to workforce performance (Dess & Shaw, 2001). As turnover harms current social relationships with employees who exit the firm—that is, “when employees leave a unit, all of their contributions leave, including their relationships with other employees” (Nyberg & Ployhart, 2013: 117)—it undermines these benefits at the expense of workforce performance (Dess & Shaw, 2001; Hancock et al., 2013; Shaw, Duffy, et al., 2005).

Finally, turnover may also cause operational disruptions that complicate the functioning of the workforce (Heavey et al., 2013). Turnover can disrupt organizational operations directly—for example, by increasing the volume of unfinished work or by creating a flux in the coordination, knowledge exchange, and socialization of employees—simply because the work needs to be performed by fewer employees or because replacement employees cannot operate efficiently as they may have little experience (Dess & Shaw, 2001; Mawdsley & Somaya, 2016; Nyberg & Ployhart, 2013; Summers, Humphrey, & Ferris, 2012). Turnover can also disrupt firm procedures indirectly because the increased time (and money) spent on redirecting incumbent firm members and on hiring, training, and socializing replacement workers implies that less attention is diverted to the key activities that contribute to the

De Meulenaere et al. / Collective Turnover, Firm Size, and Knowledge Intensity 997

productivity of the firm (Allen, Bryan, & Vardaman, 2010; Hausknecht et al., 2009; Heavey et al., 2013; Messersmith, Lee, Guthrie, & Ji, 2014; Watrous, Huffman, & Pritchard, 2006).

In sum, because high collective turnover rates engender human and social capital deple- tions and operational disruptions, they are considered detrimental to firm performance. However, as put forward by CETT, the extent to which collective turnover has negative consequences for firm performance depends on the organizational context (Brymer & Sirmon, 2018; Nyberg & Ployhart, 2013). In this study, we propose firm size as a key orga- nizational contextual moderator.

Firm Size

To delineate the potential moderating role of firm size, we systematically and theoretically derive two opposing hypotheses using the resource-based theory of the firm. This theory defines resources as (in)tangible assets controlled by the organization, which can be deployed to improve the firm’s efficiency and effectiveness and to achieve a competitive advantage (Barney, 1991; Guthrie, 2001; Shaw et al., 2013). We argue that large and small firms have different amounts and types of human, social, financial, and organizational resources (Nyberg, Moliterno, Hale, & Lepak, 2014; Nyberg & Ployhart, 2013), which can either help or hinder firms in successfully alleviating collective turnover repercussions.

Positive role of firm size. On the one hand, we can hypothesize that because larger firms have more resources, they have a competitive advantage over smaller firms when they are faced with turnover as they can withstand the loss of human capital and its associated dis- ruptions better (Barney, 1991; Hancock et al., 2013; Josefy, Kuban, Ireland, & Hitt, 2015; Kozlowski & Bell, 2013).

First, to the extent that firm size is conceptualized as the number of employees (Hancock et al., 2013), larger firms have a greater absolute pool of human resources (HR). While one could argue that similar turnover rates cause comparable human capital losses for both large and small firms, the abundancy of human capital makes larger firms more resilient to the departures of employees. The more employees, the more likely several workers will perform the same job (Barron, Black, & Loewenstein, 1989), implying that multiple employees possess similar gen- eral and firm-specific human capital. For the same turnover rate, larger firms will thus experi- ence fewer operational disruptions because they will have fewer difficulties in replacing the lost human capital in the short run, for example, by others taking over the extra work.

Second, larger firms generally have more financial resources and more potential to attract further monetary means (Mitchell, 1994); for example, they can more easily raise capital and secure loans because they face better institutional regulations (Audia & Greve, 2006; Brüderl & Schüssler, 1990). Accordingly, studies have found that this provides large firms with a larger absolute pool of financial slack resources, or the monetary means that firms either have available (e.g., excess liquidity) or can accrue in the short run (e.g., through loans) (Daniel, Lohrke, Fornaciari, & Turner, 2004). Such a flexible form of slack creates a finan- cial buffer that may help firms to deal quickly and efficiently with such organizational phe- nomena as turnover (Audia & Greve, 2006; Fuentelsaz, Gomez, & Polo, 2002; Josefy et al., 2015; Sharfman, Wolf, Chase, & Tansik, 1988).1 More specifically, it enables them to man- age operational disruptions caused by turnover—for example, by outsourcing specific tasks

998 Journal of Management / April 2021

to flexible, temporary workers—and/or to invest additionally in the recruitment of well- qualified replacement employees who can quickly take over the work of leaving employees (Fuentelsaz et al., 2002; Kozlowski & Bell, 2013; Park & Shaw, 2013).

To achieve quick and effective replacements, larger firms can additionally draw from their stronger employer brand as the great amount of financial resources enables them to offer more attractive working conditions (e.g., higher wages, employee benefits, and formal devel- opment opportunities) than can smaller firms (Cardon & Stevens, 2004; Cosic, 2018; O’Connell & Byrne, 2010). This gives them more prestige (Hannan & Freeman, 1984), power (Boone, Carroll, & van Witteloostuijn, 2004), and legitimacy in the labor market to attract high-quality replacement employees quickly, compensating for the human resource losses incurred (Williamson, Cable, & Aldrich, 2002).

Third, larger firms also benefit from more organizational resources, residing in the pro- cesses, systems, routines, practices, and structures of the organization (Greene, Brush, & Brown, 1997). Compared to small firms, large firms can draw from a larger pool of HR-related knowledge and expertise (Cardon & Stevens, 2004; Josefy et al., 2015). As a result, they make more use of sophisticated and formal HR practices and policies than smaller firms (Guthrie, 2001; Jackson & Schuler, 1995). These practices typically reflect a wide range of formalized routines, rules, and procedures, which can help organizations deal with turnover in a systematic and efficient way (Barber, Wesson, Roberson, & Taylor, 1999). For example, it enables them to create routines on who takes up the work when employees leave the firm (Mowday, 1984), to develop procedures for maintaining the knowledge that departing employees have developed (Call et al., 2015), to set up a continuous recruitment process, and to standardize the training of newcomers (Allen, 2006).

Organizational resources also reside into well-developed internal labor markets (Guthrie, 2001; Pfeffer & Cohen, 1984; Williamson et al., 2002). This implies a strong division of labor in which employees are responsible for a specialized task with clear boundaries (Cardon & Stevens, 2004; Carley, 1992), which differs substantially from smaller firms, where employees generally have broader responsibilities over multiple tasks (Meijaard, Brand, & Mosselman, 2005). This gives larger firms a competitive advantage over smaller firms when employees leave because it is less harmful to lose employees when they have a single respon- sibility and/or perform a single task, and it is easier to recruit new hires for specific tasks. Moreover, internal labor markets can supply (temporary) internal replacements, implying that the work of the exiting employees can be covered and potential operational disruptions are limited (Groothuis, 1994; Pfeffer & Cohen, 1984).

In sum, the larger pool of human, financial, and organizational resources may help large organizations to cope more successfully with collective firm-level turnover, in particular by coping with operational disruptions and by enabling quick and effective replacements for the departing employees:

Hypothesis 1a: The negative relationship between collective firm-level turnover and firm perfor- mance is weaker for large firms than for small firms.

Negative role of firm size. Whereas a larger pool of human, financial, and organizational resources may help larger firms dealing with turnover consequences, the structure and qual- ity of their social and organizational resources may impose barriers and disadvantages for

De Meulenaere et al. / Collective Turnover, Firm Size, and Knowledge Intensity 999

employees’ socialization and adaptation when firm members exit the organization. In this view, large firms have a competitive disadvantage as compared to smaller firms.

First, large and small firms differ in terms of the structure of their social resources, more specifically in the pattern of their internal social networks (Josefy et al., 2015). As larger firms have more employees and working units, social relationships among employees and units become more complex (Hannan & Freeman, 1984; Josefy et al., 2015). As a result, it is more difficult to create a dense social network in a large firm, that is, a network with a large relative number of relationships compared to the number of possible ones (Scott, 2000). This high complexity and low density of internal social resources hampers information sharing within the workforce, implying that employees are less aware of each other’s roles and responsibili- ties in large firms (Shaw, Gupta, & Delery, 2005; Sparrowe, Liden, Wayne, & Kraimer, 2001). This complicates the socialization of newcomers and the quick and easy adaptation of the remaining employees to the altered working situation and the associated operational disrup- tions after turnover (Fang, Duffy, & Shaw, 2011; Shaw, Gupta, et al., 2005).

Second, the quality of the social resources decreases with firm size. Research emphasizes that as firm size increases, relationships between employees become less genuine and more superficial, firm members interact and participate less, and they become less satisfied and committed to the firm and their coworkers (Hausknecht et al., 2009; Josefy et al., 2015; Scott, 2000; Shaw, Gupta, et al., 2005). This might negatively influence the amount of trust present in the relationships (i.e., the relational dimension of social resources; Granovetter, 1984) and the shared values, interpretations, and systems of meaning that provide sense and facilitate knowledge transfer and learning (i.e., the cognitive dimension of social resources; Nahapiet & Ghoshal, 1998), thereby hindering or complicating the socialization and adapta- tion of “old” and potential new firm members after turnover (Hausknecht et al., 2009; Park & Shaw, 2013).

Third, the organizational resources in large firms are characterized by a hierarchical structure of work and a bureaucratic management approach, reflected in formal norms and rules (Josefy et al., 2015; Meijaard et al., 2005). As mentioned earlier, such resources can provide benefits—for example, they can lead to organizational routines for dealing system- atically with employee turnover. Yet, the bureaucracy and formalization also makes large firms more rigid and impersonal (Groothuis, 1994; Meijaard et al., 2005). This may damage firms’ success in managing the adaptation and socialization needs triggered by turnover (Hausknecht et al., 2009). Specifically, decisions (e.g., related to the redistribution of tasks or to the hiring of new employees) are made more slowly as they need to travel through dif- ferent layers in the hierarchy and follow strict procedural guidelines (Audia & Greve, 2006; Cardon & Stevens, 2004; Josefy et al., 2015; Shaw, Gupta, et al., 2005). Moreover, the greater mental and physical distance between managers and employees (Meijaard et al., 2005) limits managers’ personal support and guidance toward employees, which further complicates employees’ adaptation to the altered working situation and the socialization (e.g., training, mentoring) of new employees. In smaller firms, the level of bureaucracy is lower and the distance between managers and employees is smaller (Hayton, 2003), which allows for a more flexible and informal management approach (Groothuis, 1994; Meijaard et al., 2005). For example, it allows managers to open up to and effectively deal with employee initiatives and suggestions related to the reorganization of work after turnover (Cardon & Stevens, 2004).

1000 Journal of Management / April 2021

In sum, while—as argued earlier—large firms have relatively more access to human, financial, and organizational resources, the quality and structure of their social and organiza- tional resources may also aggravate operational disruptions and hamper firms in dealing with the socialization and adaptation needs brought on by turnover. As such, we hypothesize the following:

Hypothesis 1b: The negative relationship between collective firm-level turnover and firm perfor- mance is stronger for large firms than for small firms.

Firm-Level Knowledge Intensity

We argue that a firm’s knowledge intensity determines whether the positive or negative perspective prevails. Firm-level knowledge intensity is the extent to which an extensive body of complex knowledge is needed for realizing the output of the firm (Von Nordenflycht, 2010). Firms with high and low levels of knowledge intensity thus differ in the type of human and social resources they rely upon (Datta et al., 2005; Dess & Shaw, 2001; Messersmith et al., 2014) and lose when confronted with high turnover (Nyberg & Ployhart, 2013). We argue that this influences whether quick and effective replacements (Hypothesis 1a) or effi- cient adaptation and socialization processes (Hypothesis 1b) are more important for firms when dealing with turnover.

First, higher firm-level knowledge intensity implies more work of an intellectual nature. Therefore, highly knowledge-intensive firms (e.g., universities and legal and accounting firms) rely to a large extent on a vast pool of complex knowledge and skills embedded in an intellectually skilled workforce (Campbell, Ganco, Franco, & Agarwal, 2012; Starbuck, 1992; Von Nordenflycht, 2010). In other words, intellectual human capital constitutes the key ingredient for organizational success (Alvesson, 2000; Campbell et al., 2012; Starbuck, 1992). This human capital is typically tacit and nonstandardized, and research has empha- sized that the human capital in highly knowledge-intensive firms can incite a competitive advantage only when it is shared intensely by employees with other firm members and when it is integrated into the firm’s memory (Alvesson, 2000; Hitt, Biermant, Shimizu, & Kochhar, 2001; Shalley, Gilson, & Blum, 2009; Subramaniam & Youndt, 2005). This makes social resources, or the social interactions and relationships within the workforce, essential as they leverage the human resources to achieve higher performance (Dess & Shaw, 2001; Hitt et al., 2001; Käpylä, Laihonen, Lönnqvist, & Carlucci, 2011; Mawdsley & Somaya, 2016). The so-called bundles of human resources created by combining different intellectual inputs are difficult to imitate and, so, generate competitive advantage for knowledge-intensive firms (Barney, 1991; Von Nordenflycht, 2010; Wernerfelt, 1984).

Thus, when faced with turnover, firms do not merely lose highly valuable knowledge but also experience a gap in their social resources, causing disruptions in the knowledge-sharing and knowledge-generating activities of the workforce (Dess & Shaw, 2001). To guarantee the continuity of the knowledge-based activities, it is crucial that the remaining employees adapt efficiently to the altered social landscape of the workforce due to the departures of col- leagues. At the same time, the socialization process of replacement knowledge workers needs to go smoothly in order to fill gaps efficiently and to facilitate the further accumulation of human resources and social resources in such a way that it can contribute to the competitive advantage of the firm. While we acknowledge that replacements are also important for

De Meulenaere et al. / Collective Turnover, Firm Size, and Knowledge Intensity 1001

knowledge-intensive firms to guarantee a continuous knowledge creation process, quick replacements are not a priority simply because employees’ tacit knowledge and their particu- lar role in social networks is difficult to replace in the short run—that is, it takes time for new employees to adapt, to learn the organizational routines (Nyberg & Ployhart, 2013), and to develop social capital (Coleman, 1988).

From this point of view, we argue that under high turnover rates, firms that are more knowledge intensive benefit more from (a) carefully selecting and hiring new employees (versus quickly replacing employees) with knowledge that is complementary to the existing knowledge and (b) a high network density, high trust, and a shared mind-set that will help existing employees to adapt quickly and that will also help new employees to socialize quickly. These adaptation and socialization processes will, in turn, encourage new knowl- edge creation and transfer. As postulated by the negative perspective of firm size, adaptation and socialization can be managed less efficiently in larger firms (Hausknecht et al., 2009). For these reasons, we argue that the negative perspective of the role of firm size prevails in highly knowledge-intensive firms.

In contrast, firms with lower levels of knowledge intensity rely less on the intellectual input and social capital of their workforce (Dess & Shaw, 2001; Shaw, Duffy, et al., 2005). We acknowledge that less knowledge-intensive firms (e.g., retail, food- and beverage-pro- ducing firms) also rely on certain types of knowledge, such as those embedded in routines and information systems (Coff, 1999). Yet this involves more explicit and standardized knowledge that is easier to develop and transfer (Campbell et al., 2012; Hancock et al., 2013; Hitt et al., 2001; Shaw, Duffy, et al., 2005; Von Nordenflycht, 2010). Accordingly, we want to argue not that social interactions and collaborations among employees are not important but that the information exchanged in low knowledge-intensive settings is more standardized and less complex than in more knowledge-intensive firms (Alvesson, 2000; Campbell et al., 2012; Starbuck, 1992). This also implies that employees’ human capital and their positions in social networks are more easily replaceable in low-level than in highly knowledge-inten- sive firms (Nyberg & Ployhart, 2013). Hence, when facing high levels of turnover, the effi- cient adaptation and socialization of employees is less important than quick and effective solutions for the human capital losses (e.g., other employees taking over or hiring replace- ment employees). As argued by the positive perspective, larger firms have more resources to provide such solutions (Hancock et al., 2013; Josefy et al., 2015; Kozlowski & Bell, 2013). Therefore, we postulate that firm size acts as a buffer for the negative consequences of turn- over in firms with low levels of knowledge intensity. Thus, we hypothesize the following:

Hypothesis 2: The higher a firm’s knowledge intensity, the more the negative perspective on firm size will prevail. The lower a firm’s knowledge intensity, the more the positive perspective on firm size will prevail.

Method

Sample

To test our hypotheses, we use the Bel-first data warehouse, containing annual account reports submitted by Belgian firms to the National Bank of Belgium. The database registers detailed information on the financials and the social balance of the companies, which enables

1002 Journal of Management / April 2021

us to extract information on the size, number of employees leaving in a given year, knowl- edge intensity, and firm performance of all Belgian firms included in the warehouse.

We selected firms that have at least 50 employees in a minimum of one of the observation years (2012–2016). Firms with 50 or more employees are obliged to hire a commissioner for auditing their accounts and controlling all information that is reported to the National Bank of Belgium. As these commissioners control the accounts for multiple years—thus, also for the years in which firms do not reach the number of 50 employees—we can guarantee that we have accurate firm-level collective turnover and accounting information for these firms. Accordingly, our selection of firms includes both observations of firms with more than 50 employees (80% of the observations) and observations of firms with fewer than 50 employ- ees for which the accounts are controlled by a commissioner (20% of the observations). We excluded firms from the “employment activities” sector as they are confronted with excep- tionally high turnover rates because they work with daily and weekly contracts. This resulted in a final sample of 6,913 organizations over a time span of 5 consecutive years (2012–2016), covering a total of 32,294 firm-year observations. In our sample, firm size ranges from one to 28,198 full-time-equivalent employees. Given that European and U.S. standards define small and medium-sized enterprises (SMEs) as firms with fewer than 250 and 500 employ- ees, respectively (and large organizations as firms with more employees than these thresh- olds; Eurostat, 2018; Office of the U.S. Trade Representative [USTR], 2018), the range of sizes in our sample is highly suitable for investigating the role of firm size.

Measures

Labor productivity. In accordance with previous studies (e.g., Glebbeek & Bax, 2004; Heavey et al., 2013; Shaw, 2011), we focus on labor productivity to capture firm perfor- mance. This outcome reflects the gross added value of the firm per full-time-equivalent employee (see also Glebbeek & Bax, 2004; Shaw, Duffy, et al., 2005; Siebert & Zubanov, 2009), rendering it well suited to evaluating the functioning of the workforce after turnover. We took the natural logarithm of labor productivity to reduce the influence of outliers.

Collective firm-level turnover rate. Building on previous turnover studies (e.g., Allen et al., 2010; Hausknecht & Trevor, 2011; Kacmar et al., 2006), our measure of firm-level col- lective turnover reflects the number of all workers leaving the organization in a given year divided by the number of full-time-equivalent employees in that year. In our sample of firms, a small percentage of organizations (7.81%) was characterized by turnover rates exceeding 1. The highest turnover rate was 356. Further investigation of the data revealed that organi- zations with such a high outflow of workers were also characterized by an extremely high inflow. Hence, we know that firms with very high turnover rates are generally organizations that frequently replace their (entire) workforce, which is the case for firms employing sea- sonal workers, student workers, and other types of short-term interim employees. To avoid our results being biased by these extremely high turnover rates for such a small group of firms, we winsorized our turnover variable. We did so by setting all turnover rates larger than one equal to one. We chose this cutoff value because all values equal to and higher than one imply that the entire workforce has been replaced. Note that our results did not change when we used another cutoff value (e.g., two; analyses available in the online supplement).

De Meulenaere et al. / Collective Turnover, Firm Size, and Knowledge Intensity 1003

We acknowledge that the most common approach to deal with highly skewed distributed variables is by taking the logarithm, but this transformation is not suitable for our turnover variable. A log-transformation would manipulate the data in such a way that two firms with turnover rates of 1% and 2% equally differ from each another to the same degree as two firms having turnover rates of 10% and 20%. Winsorizing enables us to preserve the smaller dif- ferences in turnover while alleviating the impact of extreme differences between firm-level turnover rates. This is particularly relevant for our sample in which more than 25% of the firms have small turnover rates (here defined as a turnover rate below 10%). Note, however, that the results remained unchanged by taking the log-transformed turnover rate (analyses available in the online supplement).

Firm size. We measured firm size as the number of full-time-equivalent employees work- ing in the organization each year. We took the logarithm of this variable because it was extremely skewed to the right (see also Messersmith et al., 2014; Park & Shaw, 2013).

Firm-level knowledge intensity. In line with previous studies (e.g., Alvesson, 2000; Käpylä et al., 2011), we measured firms’ knowledge intensity by calculating the share of highly educated employees (i.e., employees with a bachelor/master degree), ranging from 0 (not one employee is highly educated) to 1 (all employees are highly educated). In our data set, the average organizational knowledge intensity is 0.28. Note that we observed an increase in the average knowledge intensity over the observation years, ranging from 0.27 (in 2012) to 0.29 (in 2016), which is in line with the general understanding that Western econo- mies are becoming increasingly knowledge based (Hancock et al., 2013).

Control variables. Building on previous studies that have examined turnover and/or used labor productivity as an outcome variable (e.g., Call et al., 2015; Hancock et al., 2013; Mess- ersmith et al., 2014), we include the following control variables: firm age, capital intensity, and the year of observation. Firm age is included as the number of years since the firm’s incorporation (Call et al., 2015). Capital intensity is defined as the ratio of real tangible fixed assets over the number of full-time-equivalent employees and is associated with higher labor productivity (e.g., Van Ark, O’Mahony, & Timmer, 2008). We took the natural logarithm of firm age and capital intensity as these variables were highly skewed (Huselid, 1995; Mess- ersmith et al., 2014). As our data cover firm-level observations over 5 years (2012–2016), we controlled for potential year effects using year dummy variables. This way we minimize autocorrelation and heteroscedasticity issues that can result from our panel data structure (Brymer & Sirmon, 2018; Hitt et al., 2001).

Data Analysis Approach

As our data comprise repeated firm observations over 5 years (2012–2016), we used a panel data estimation technique to analyze the link between turnover and organizational labor productivity (the “xtreg” command in STATA). We adopted a fixed-effects model design because it takes into account the assumption that one or more variables that are not included in the model affect the relationship between turnover and productivity (Bell & Jones, 2015). Management policy, for example, might influence the turnover–performance

1004 Journal of Management / April 2021

link (Park & Shaw, 2013). If so, the fixed-effects model specification accounts for this (and other) omitted but important time-invariant variable(s). The Hausman test (χ² = 2260.22, p < .0005) also corroborated that a fixed-model design should be preferred over a random- effects design. However, our results are robust when running our analyses using the random- effects specification (analyses available in the online supplement). We clustered our standard errors at the firm level to account for serial correlation and possible heteroscedasticity (Call et al., 2015).

As panel data techniques complicate the use and interpretation of the R-squared goodness- of-fit test statistic (Nakagawa & Schielzeth, 2013), we provide an alternative informative statistic to compare the fit between our models. In particular, by rerunning each model with maximum likelihood estimators, we are able to calculate the likelihood ratio (LR) test statistic that indicates whether the estimated coefficients add significantly to the prediction of labor productivity. We report the LR statistics together with the fixed-effects estimation results.

Results

Table 1 shows the descriptive statistics and correlations of all variables used in this study. We mean-centered the variables that form the interactions (i.e., collective turnover rate, firm size, and knowledge intensity) to avoid multicollinearity between these cross-products. We computed the variance inflation factors based on an ordinary least squares regression of labor productivity on all our variables, including the interaction terms. They were all below 1.61 (findings available in the online supplement), thus indicating no multicollinearity problem (O’Brien, 2007).

Table 2 reveals the regression coefficient estimates. Model 1 includes the control vari- ables and moderators and reveals that firm age (B = .29, p < .0005) and capital intensity (B = .05, p < .0005) are positively associated with labor productivity, that firm size is nega- tively related to labor productivity (B = –.37, p < .0005), and that knowledge intensity is not related to productivity (B = 03, p = .113). Model 2 reveals a nonsignificant association between turnover and labor productivity (B = .01, p = .637).

In Model 3, we test whether and how firm size moderates the effect of firm-level collec- tive turnover on labor productivity. Hypothesis 1a proposed a positive moderation, whereas Hypothesis 1b suggested a negative interaction effect. We find a positive interaction effect between turnover and firm size (B = .08, p = .084) in line with Hypothesis 1a. However, it is only marginally significant, which might indicate either that firm size plays only a minor (positive) role or that an additional moderator, such as knowledge intensity, is at play. In Model 4, we test Hypothesis 2, proposing that the positive (negative) moderating impact of size is more likely to prevail as knowledge intensity decreases (increases). Our estimates support this expectation. The three-way interaction of turnover, firm size, and knowledge intensity is negative and significant (B = –.29, p = .009), meaning that the positive moderat- ing effect of firm size (B = .08, p = .093) in the firm-level turnover–performance relation- ship reduces with increasing knowledge intensity.

Interpreting the Three-Way Interaction: Post Hoc Exploratory Data Analyses

To interpret the three-way interaction, we probe our findings by performing two additional post hoc analyses: a simple-slopes analysis and a split-sample analysis (Dawson & Richter,

1005

T a b

le 1

M ea

n s,

S ta

n d

a rd

D ev

ia ti

o n

s, a

n d

C o rr

el a ti

o n

s o f

A ll

V a ri

a b

le s

U se

d i

n t

h e

A n

a ly

se s

V ar

ia b le

M in

. M

ax .

M S D

1 2

3 4

5 6

7 8

9 1 0

1 . L

ab o r

p ro

d u ct

iv it

y (

lo g )

– 6 .9

1 9 .5

3 4 .2

6 0 .6

6 —

2 . F

ir m

a g e

(l o g )

0 4 .7

6 3 .2

3 0 .7

3 .1

4 *

—

3 . C

ap it

al i

n te

n si

ty (

lo g )

– 4 .6

0 1 4 .8

5 3 .3

7 1 .9

7 .5

0 *

.2 3 *

—

4 . F

ir m

s iz

e (l

o g )

– 4 .4

9 8 .1

0 0

0 .9

7 .0

1 .1

4 *

.0 6 *

—

5 . K

n o w

le d g e

in te

n si

ty (

K I)

– .2

8 .7

2 0

0 .2

9 .3

9 *

.0 7 *

.1 7 *

.0 9 *

—

6 . T

u rn

o v er

– .2

9 .7

1 0

0 .2

8 – .1

9 *

– .1

3 *

– .0

8 *

– .1

6 *

– .1

4 *

—

7 . T

u rn

o v er

× F

ir m

S iz

e – 3 .2

0 3 .4

8 – 0 .0

4 0 .3

0 – .0

2 *

.0 2 *

– .0

5 *

– .0

0 .0

1 *

– .1

9 *

—

8 . T

u rn

o v er

× K

I – .2

1 .5

1 – 0 .0

1 0 .0

7 – .0

6 *

.0 5 *

– .0

2 *

.0 2 *

– .1

0 *

– .2

7 *

.1 3 *

—

9 . F

ir m

S iz

e ×

K I

– 3 .2

3 2 .1

9 0 .0

3 0 .2

9 .0

6 *

.0 0

.0 6 *

.0 4 *

– .0

9 *

.0 2 *

– .2

2 *

– .1

2 *

—

1 0 . T

u rn

o v er

× F

ir m

S iz

e ×

K I

– 2 .3

0 1 .0

6 0

0 .0

9 – .0

8 *

– .0

6 *

– .0

3 *

– .2

4 *

– .1

1 *

.1 4 *

– .3

3 *

– .2

7 *

– .1

1 *

—

N o te

: S

o u rc

e d at

a o f

B el

-f ir

st (

d at

ab as

e o f

an n u al

f in

an ci

al r

ep o rt

s o f

B el

g ia

n f

ir m

s) .

T h e

v ar

ia b le

s th

at c

o n st

it u te

t h e

in te

ra ct

io n e

ff ec

ts (

tu rn

o v er

r at

e, f

ir m

s iz

e, a

n d

K I)

a re

m ea

n -c

en te

re d . N

= 3

2 ,2

9 4 .

* p <

. 0 1 .

1006 Journal of Management / April 2021

2006; Hollenbeck & Wright, 2017). First, we used the “margins” command in STATA to com- pute the simple slopes of the impact of turnover on labor productivity for different combina- tions of firm size and knowledge intensity values (Dawson & Richter, 2006). For knowledge intensity, we work with the (centered) conditional values that correspond with actual (noncen- tered) knowledge intensity rates of 0, 0.25, 0.50, 0.75, and 1, thus covering the entire range of values in our database. For our firm size variable, we work with the (centered and log-trans- formed) values that correspond with actual firm sizes of 10, 50, 250, and 500 employees. These firm size levels are widely used to categorize firms into micro (<10 employees), small (between 10 and 49), medium (between 50 and 250 in Europe and between 50 and 500 in the United States), and large firms (>250 in Europe and >500 in the United States; Eurostat, 2018; USTR, 2018). In Table 3, we report the results. In Figure 1, we provide a graphical representation. It shows the moderating influence of firm size on the effects of turnover on labor productivity for different conditional values of knowledge intensity. In Figure 2, we display the distribution of firm sizes across the range of these conditional values.

Table 2

The Combined Impact of Collective Firm-Level Turnover, Firm Size (Log), and Knowledge Intensity on Labor Productivity (Log)

Model 1 Model 2 Model 3 Model 4

Predictor B (SE) (t; p)

B (SE) (t; p)

B (SE) (t; p)

B (SE) (t; p)

Intercept 3.13 (0.16) (19.38; .000)

3.13 (0.16) (19.38; .000)

3.16 (0.15) (20.89; .000)

3.17 (0.15) (21.52; .000)

Firm age 0.29 (0.05) (5.91; .000)

0.29 (0.05) (5.92; .000)

0.28 (0.05) (6.11; .000)

0.28 (0.04) (6.23; .000)

Capital intensity 0.05 (0.01) (6.98; .000)

0.05 (0.01) (6.99; .000)

0.05 (0.01) (7.05; .000)

0.05 (0.01) (7.17; .000)

Firm size –0.37 (0.05) (–7.82; .000)

–0.37 (0.05) (–7.78; .000)

–0.37 (0.05) (–7.78; .000)

–0.37 (0.04) (–8.59; .000)

Knowledge intensity (KI) 0.03 (0.02) (1.58; .113)

0.04 (0.02) (1.62; .106)

0.03 (0.02) (1.32; .187)

0.08 (0.02) (3.30; .001)

Turnover 0.01 (0.02) (.47; .637)

0.06 (0.03) (2.07; .039)

0.05 (0.03) (1.96; .050)

Turnover × Firm Size 0.08 (.05) (1.73; .084)

0.08 (0.05) (1.68; .093)

Turnover × KI –0.18 (0.07) (–2.63; .009)

Firm Size × KI 0.13 (0.06) (2.37; .018)

Turnover × Firm Size × KI –0.29 (0.11) (–2.62; .009)

F 31.26 28.27 27.44 23.07 LR test (model compared to) 15.49 (1) 10.97 (2) 547.66 (2)

563.69 (3)

Note: Dependent variable: logarithm of labor productivity. Source data of Bel-first (database of annual financial reports of Belgian firms). Robust standard errors are clustered for firm ID. Significant results are in bold. All observation years are controlled for. N = 6,913.

De Meulenaere et al. / Collective Turnover, Firm Size, and Knowledge Intensity 1007

We pinpoint three important observations. First, for firms with a knowledge intensity rate of 0%, 25%, and 100%, we observe that the effect of turnover on labor productivity is signifi- cantly different across firms of different sizes (p = .015, p = .070, and p = .076, respec- tively). This indicates that firm size has a moderating impact on the turnover–performance relationship in firms with (very) low or very high knowledge intensity. Second, firm size appears to have a positive moderating influence on the turnover effect for firms with the low- est rates of knowledge intensity (0% and 25%), whereas firm size negatively influences the impact of turnover for firms with the highest rate of knowledge intensity (100%). This is fully in line with Hypothesis 2. Third, turnover seems to have a positive effect for large, low knowledge-intensive firms. For example, based on the regression coefficient of turnover for firms with 250 employees and with a knowledge intensity rate of 25% (B = .14, p = .037), we calculated that a 10-percentage-point increase in turnover was associated with an increase in labor productivity of 1.4% (= exp[.1*.14] − 1). For firms of 250 employees with a knowl- edge intensity rate of 0% (B = .25, p = .006), a 10-percentage-point increase in turnover generated a productivity increase of 2.5%. These findings reveal that the hypothesized

Table 3

Simple Slopes of the Impact of Collective Firm-Level Turnover on Labor Productivity (Log) for Multiple Combinations of Firm Size and Knowledge Intensity Values

Knowledge Intensity Firm Size B (SE) z (p)

90% Confidence Interval

Significance of Difference

0% 10 –.26 (0.12) –2.10 (.035) [–.47, –.06] p = .015 50 –.01 (0.03) –0.21 (.837) [–.05, .04]

250 .25 (0.09) 2.76 (.006) [.10, .40] 500 .36 (0.13) 2.67 (.008) [.14, .58]

25% 10 –.14 (0.09) −1.53 (.125) [–.29, .01] p = .070 50 –.00 (0.02) –0.11 (.911) [–.04, .04]

250 .14 (0.07) 2.08 (.037) [.03, .24] 500 .20 (0.10) 2.01 (.044) [.04, .36]

50% 10 –.02 (0.09) –0.24 (.809) [–.16, .12] p = .750 50 .00 (0.03) 0.03 (.979) [–.05, .05]

250 .02 (0.06) 0.39 (.696) [–.07, .12] 500 .03 (0.09) 0.37 (.708) [–.11, .17]

75% 10 .10 (0.11) 0.88 (.381) [–.09, .29] p = .271 50 .00 (0.04) 0.10 (.919) [–.06, .07]

250 –.09 (0.07) −1.24 (.214) [–.21, .03] 500 –.13 (0.11) −1.23 (.220) [–.31, .05]

100% 10 .22 (0.16) 1.41 (.157) [–.04, .48] p = .076 50 .00 (0.05) 0.14 (.892) [–.08, .10]

250 –.20 (0.10) –2.01 (.045) [–.37, –.04] 500 –.30 (0.15) −1.98 (.048) [–.54, –.05]

Note: Source data of Bel-first (database of annual financial reports of Belgian firms). Robust standard errors are clustered for firm ID. Significant results are in bold. All observation years are controlled for. N = 6,913. Note that we ran our simple-slopes analyses with the centered and log-transformed values of knowledge intensity and firm size. For example, a firm size of 10 corresponds to a value of −1.63 [= log(10) – mean(log firm size) = log(10) – 4.59 =]. To facilitate the interpretation of this table, we chose to report the corresponding real values for firm size and knowledge intensity.

1008 Journal of Management / April 2021

positive moderating role of firm size in firms with low levels of knowledge intensity is more impactful than we anticipated. We will elaborate on this unexpected finding in the Discussion.

As a second post hoc analysis, we performed a split-sample analysis of the interaction between turnover and firm size for two subgroups of firms, one representing low and one representing high knowledge intensity. Therefore, we needed to carefully consider which of our firm observations belong to the low- and the high knowledge-intensive subgroup. We constructed two subgroups composed of the 25% lowest- and 25% highest knowledge-inten- sive firms in our sample, thus excluding 50% of the intermediate values.2 Knowledge intensity ranged from 0% to 4% in the first subgroup (this small range follows naturally from the fact that a large share of firms employ either no or only a few highly educated workers) and from 44% to 100% in the second subgroup (all rates in this subgroup are equally represented). Note

Figure 1 Simple-Slope Effects of Collective Firm-Level Turnover on Labor Productivity (Log):

The Moderating Influence of Firm Size (Log) for Different Values of Knowledge Intensity

Note: We ran our simple-slopes analyses using the centered and log-transformed values of knowledge intensity and firm size. For example, a firm size of 10 corresponds to a value of –.2.28 [= log(10) – mean(log firm size) = log(10) – 4.59]. To facilitate the interpretation of this figure, we chose to report the corresponding real values for firm size and knowledge intensity and put the log and centered values in parentheses. Also note that not all marginal effect sizes displayed are statistically significant. For example, the positive effects for smaller firms (10 or 50 employees) with knowledge intensity values of 0.75 and 1 are not significant. Table 3 reveals the significance of all marginal effect sizes displayed in Figure 1. Note that because we worked with centered values for the variables that constitute the interaction (turnover, firm size, and knowledge intensity), the simple slopes all go through the same point, that is, the mean value for firm size (which is zero as this variable is mean-centered).

De Meulenaere et al. / Collective Turnover, Firm Size, and Knowledge Intensity 1009

that our analyses are robust for other relevant splits, too (e.g., firms with knowledge intensity ranging from 0% to 25% vs. 75% to 100%; analyses available in the online supplement). Tables 4 and 5 report the regression results for the low and high knowledge-intensive sub- groups (descriptive statistics and correlations of all variables used in these split sample analy- ses are available in the online supplement).

Model 1 in both tables tests for the effects of the control variables. We observe significant differences between the two subgroups, but these are outside the scope of our article. The main effect of turnover is tested in Model 2. We observe that the impact is positive in the low knowledge-intensive subgroup (B = .04, p = .237; Table 4) and negative in the high knowl- edge-intensive subgroup (B = –.05, p = .287; Table 5), but the effects lack statistical signifi- cance. In Model 3, we test the interaction effect of turnover and firm size. Table 4 reveals a positive and significant moderating role of firm size for the low knowledge-intensive sub- group (B = .23, p = .003). Table 5 reveals a negative and significant moderating role in the high knowledge-intensive subgroup (B = –.07, p = .039). These findings provide further support for Hypothesis 2.

To evaluate the effect sizes, we used STATA to compute the simple slopes of the impact of collective firm-level turnover on labor productivity for different levels of firm size (note that we once again use the [log] centered values that correspond to the firm sizes of 10, 50, 250, and 500) for the low and high knowledge-intensive subgroups. Table 6 reports the simple slopes for the low knowledge-intensive subgroup. Based on the coefficients, we calculated that an increase

Figure 2 Distribution of Firm Sizes Across the Range of the Conditional Values Used for the

Simple-Slopes Analysis

Note: N = 29,922 (covering 93% of the total firm observations).

1010 Journal of Management / April 2021

Table 4

Split-Sample Analysis: The Moderating Role of Firm Size (Log) on the Impact of Collective Firm-Level Turnover on Labor Productivity (Log) for the 25% Lowest

Knowledge-Intensive Firms

Model 1 Model 2 Model 3

Predictor B (SE) (t; p)

B (SE) (t; p)

B (SE) (t; p)

Intercept 2.27 (0.33) (6.79; .000)

2.26 (0.34) (6.73; .000)

2.39 (0.30) (7.94; .000)

Firm age 0.49 (0.11) (4.48; .000)

0.49 (0.11) (4.49; .000)

0.45 (0.10) (4.57; .000)

Capital intensity 0.08 (0.02) (5.59; .000)

0.09 (0.02) (5.64; .000)

0.08 (0.01) (5.62; .000)

Firm size –0.55 (0.07) (–7.42; .000)

–0.54 (0.07) (–7.34; .000)

–0.54 (0.06) (–8.91; .003)

Turnover rate 0.04 (0.04) (1.18; .237)

0.15 (0.04) (3.79; .000)

Turnover Rate × Firm Size 0.23 (0.08) (2.94; .003)

F 14.19 13.94 15.27 LR test (model compared to) 0.09 (1) 212.84 (1)

212.74 (2)

Note: Dependent variable: logarithm of labor productivity. Source data of Bel-first (database of annual financial reports of Belgian firms). Robust standard errors are clustered for firm ID. Significant results are in bold. All observation years are controlled for. N = 2,398.

Table 5

Split-Sample Analysis: The Moderating Role of Firm Size (Log) on the Impact of Collective Firm-Level Turnover on Labor Productivity (Log) for the 25% Highest

Knowledge-Intensive Firms

Model 1 Model 2 Model 3

Predictor B (SE) (t; p)

B (SE) (t; p)

B (SE) (t; p)

Intercept 4.07 (0.25) (16.02; .000)

4.06 (0.26) (15.90; .000)

4.06 (0.26) (15.82; .000)

Firm age 0.15 (0.08) (1.90; .057)

0.15 (0.08) (1.95; .052)

0.15 (0.08) (1.94; .053)

Capital intensity 0.01 (0.01) (1.08; .282)

0.01 (0.01) (1.09; .277)

0.01 (0.01) (1.12; .264)

Firm size –0.19 (0.04) (–4.47; .000)

–0.19 (0.04) (–4.50; .000)

–0.19 (0.04) (–4.30; .000)

Turnover rate –0.05 (0.05) (−1.06; .287)

–0.10 (0.05) (−1.97; .049)

Turnover Rate × Firm Size –.07 (0.04) (–2.07; .039)

F 13.94 12.53 12.75 LR test (model compared to) 36.08 (1) 71.19 (1)

35.11 (2)

Note: Dependent variable: logarithm of labor productivity. Source data of Bel-first (database of annual financial reports of Belgian firms). Robust standard errors are clustered for firm ID. Significant results are in bold. All observation years are controlled for. N = 2,275.

De Meulenaere et al. / Collective Turnover, Firm Size, and Knowledge Intensity 1011

in turnover of 10 percentage points decreases labor productivity by 2.86% in firms with 10 employees but increases productivity by 4.50% in firms with 250 employees and by 6.18% in organizations with 500 members. Table 7 shows the simple slopes for the high knowledge- intensive subgroup. We observe that firms with 10 and 50 employees appear to experience no significant impact of turnover, but this changes as firm size increases. More specifically, we find that a 10-percentage-point increase in turnover decreases labor productivity by 1.59% in firms with 250 employees and by 2.08% in firms with 500 employees.

Robustness Analyses

We performed several additional analyses to check for the robustness of our three-way interaction. For each robustness check, we reported the parameter estimates for the

Table 6

Simple Slopes of the Impact of Collective Firm-Level Turnover on Labor Productivity (Log) for Different Values of Firm Size: Subgroup of the 25% Lowest Knowledge-

Intensive Firms

Firm Size B (SE) (z; p) 90% Confidence Interval Significance of Difference

10 –.29 (0.13) (–2.15; .031)

[–.51, –.07] p = .003

50 .08 (0.03) (2.17; .030)

[.02, .13]

250 .44 (0.12) (3.57; .000)

[.24, .65]

500 .60 (0.18) (3.41; .001)

[.31, .89]

Note: Source data of BEL-first (database of annual financial reports of Belgian firms). Robust standard errors are clustered for firm ID. Significant results are in bold. All observation years are controlled for. N = 2,398.

Table 7

Simple Slopes of the Impact of Collective Firm-Level Turnover on Labor Productivity (Log) for Different Values of Firm Size: Subgroup of the 25% Highest Knowledge-

Intensive Firms

Firm Size B (SE) (z; p) 90% Confidence Interval Significance of Difference

10 .07 (.09) (.87; .385)

[–.07, .22] p = .039

50 –.04 (.05) (–.87; .387)

[–.13, .04]

250 –.16 (0.06) (–2.50; .012)

[–.27, –.06]

500 –.21 (0.08) (–2.56; .010)

[–.35, –.08]

Note: Source data of Bel-first (database of annual financial reports of Belgian firms). Robust standard errors are clustered for firm ID. Significant results are in bold. All observation years are controlled for. N = 2,275.

1012 Journal of Management / April 2021

three-way interaction effect (Model 4) in Table 8 (full tables are available in the online supplement).

First, our fixed-effects design may well rule out the endogeneity concern for stable firm attributes, yet more time-varying variables, like unobserved managerial effectiveness (above and beyond what has been picked up by firm size in accordance with our hypotheses), may still influence both turnover and labor productivity. For this reason, we tested for endogene- ity in our model using the Olley and Pakes (1996) regression technique. This has been devel- oped because firm-level instrumental variables for inputs of the production function are hard to find—that is, a variable that determines turnover is typically also an explanatory variable for labor productivity (Ackerberg, Benkard, Berry, & Pakes, 2007)—rendering a standard endogeneity check, like one based on two-stage least squares regression, impossible to apply

Table 8

Robustness Checks of the Three-Way Interaction Effect of Turnover, Firm Size (Log), and Knowledge Intensity on Labor Productivity (Log)

Model 4aa Model 4bb Model 4cc Model 4dd

Predictor B (SE) (t; p)

B (SE) (t; p)

B (SE) (t; p)

B (SE) (t; p)

Intercept −1.33 (1.16) (−1.14; .253)

7.82 (0.14) (57.47; .000)

6.50 (0.17) (37.32; .000)

0.39 (0.99) (.39; .69)

Firm age 0.021 (0.01) (2.34; .019)

0.26 (0.01) (7.03; .000)

–0.50 (0.05) (–9.5; .000)

0.13 (0.31) (4.88; .000)

Capital intensity −1.19 (0.28) (–4.26; .000)

0.05 (0.01) (7.03; .000)

–0.12 (0.01) (–11.98; .000)

0.08 (0.06) (1.46; .144)

Firm size −1.36 (0.28) (–4.85; .000)

0.69 (0.03) (24.45; .000)

0.06 (0.03) (1.99; .046)

–0.38 (0.24) (−1.57; .115)

Knowledge intensity (KI) 0.56 (0.03) (19.23; .03)

0.10 (0.02) (4.06; .000)

0.16 (0.03) (4.73; .000)

–0.11 (0.12) (–.93; .351)

Turnover –0.11 (0.02) (–5.17; .000)

0.03 (0.01) (5.03; .000)

–0.08 (0.03) (–3.19; .001)

–0.89 (0.21) (–4.19; .000)

Turnover × Firm Size 0.03 (0.03) (1.00; .315)

0.03 (0.01) (3.83; .000)

0.03 (0.03) (.79; .431)

–0.32 (0.16) (–2.04; .041)

Turnover × KI –0.26 (0.10) (–2.67; .008)

–0.05 (0.01) (–3.27; .001)

–0.21 (0.08) (–2.83; .005)

–0.45 (0.65) (–.69; .488)

Firm Size × KI 0.04 (0.04) (1.04; .298)

0.08 (0.05) (1.71; .088)

0.00 (0.05) (.02; .983)

–0.34 (0.23) (−1.51; .132)

Turnover × Firm Size × KI –0.19 (0.09) (–2.01; .044)

–0.04 (0.01) (–3.57; .000)

–0.20 (0.09) (–2.15; .032)

–0.90 (0.45) (–2.01; .045)

F/χ² 76.08 154.12 32.09 10.40 R²/LR test (model compared to) .520 1,828.86 (2)

581.65 (3) 70.67 (2) 56.32 (3)

24.72 (2) 19.30 (3)

Note: Source data of Bel-first (database of annual financial reports of Belgian firms). Robust standard errors are clustered for firm ID. Significant results are in bold. All observation years are controlled for. Unless mentioned otherwise, N = 6,913. Note that Models 4a through 4d represented in this table are compared to Model 4 in Table 2. aOlley and Pakes endogeneity check. bLabor productivity and turnover variables without firm size as denominator. cDependent variable = logarithm of added value/total fixed assets; N = 6,335. dDependent variable = logarithm of earnings before interest, taxes, depreciation, and amortization.

De Meulenaere et al. / Collective Turnover, Firm Size, and Knowledge Intensity 1013

in our study (Semadeni, Whithers, & Trevis Certo, 2014). The Olley and Pakes estimation technique builds on the idea that, under a number of assumptions, unobserved productivity can be written as a function of observables, that is, firm capital and investments. As our data comprise information on these observables, this allowed us to control for it. The parameter estimates (Model 4a, Table 8) are similar to those of our baseline analyses (Model 4, Table 2). Although this methodology rests on strong assumptions, this is a reassuring result.

Second, firm size (i.e., the number of employees) enters into our analyses both as the denominator in our measures of turnover and labor productivity and as a moderator. As argued by Wiseman (2009), this may overstretch the statistical power of firm size, creating an unstable model that may result in spurious findings. To deal with this issue, we followed Wiseman’s suggestions and reran our analyses without firm size as a denominator in our outcome and turnover variables, leaving us with the gross added value as an indicator of firm performance and the outflow of workers in absolute numbers as turnover variable. As the pattern of findings (Model 4b, Table 8) is similar to our baseline results (Model 4, Table 2), this indicates that the three-way interaction effect is not a spurious result and that our find- ings are robust.

Finally, it is valuable to test our model based on alternative indicators of organizational performance. First, our outcome variable has firm size in the denominator. Some firms may be able to outsource and/or use temporary workers as a result of turnover, which would arti- ficially inflate a firm’s added value per employee. Hence, we need to check if our findings are robust for other indicators of firm performance that do not include the number of employ- ees. Second, we acknowledge that organizational performance is a multilevel construct (Richard, Devinney, Yip, & Johnson, 2009), implying that the test of our three-way interac- tion may show different results for different performance indicators. For these reasons, we reran our analyses using two alternative indicators of firm performance: one based on the added value of the workforce but without firm size included—that is, the added value to total fixed assets (logged)—and one financial indicator of organizational performance—that is, earnings before interest, tax, depreciation, and amortization (logged; Richard et al., 2009). In Table 8, we report the parameter estimates for the three-way interaction effect for each of these performance indicators (Models 4c and 4d). As our findings remain the same for both alternative measures, we consider them robust for different performance indicators.

Discussion

Using a large longitudinal sample (2012–2016) of 6,913 firms, we reveal that firm size can both buffer and strengthen the detrimental effects of collective turnover on labor produc- tivity, depending on the firm’s knowledge intensity. Firm size negatively influenced the impact of turnover in highly knowledge-intensive firms but mitigated it in organizations with low levels of knowledge intensity. This is in line with our arguments that firms’ knowledge intensity determines the type of human and social capital that firms lose when they are con- fronted with high turnover rates. This, in turn, determines the best way to mitigate or over- come the turnover disruptions. Quick and effective replacements, for which large firms have the appropriate resources, seem to matter most for low knowledge-intensive firms in which human and social capital are easy replaceable. Efficient socialization and adaptation, for which smaller firms have the ideal resources, appear to be a priority in highly knowledge- intensive firms in which complex human capital and dense social networks are key to firm

1014 Journal of Management / April 2021

performance. In sum, whereas firm size determines the resources that firms have at their disposal, knowledge intensity determines the resources that firms need when they have to deal with turnover.

Our findings also revealed that turnover may have a positive performance effect in large, low knowledge-intensive firms. Multiple scholars (e.g., Abelson & Baysinger, 1984; Allen et al., 2010; Glebbeek & Bax, 2004; Hancock et al., 2013) have already claimed that it may be a misconception that turnover is inherently dysfunctional because it may gener- ate benefits, such as the depletion of poor performers, decreases in labor costs, the oppor- tunity to reorganize work, and the infusion of new ideas, skills, knowledge, and energy in the workforce.

To determine what is driving the positive effect in this study, it is important to shed light on the Belgian institutional context. Belgium is characterized by strict employment protec- tion legislation and strong labor unions. Consequently, firms are confronted with long notice periods and high severance payments for individual and collective dismissals (Federal Public Service Employment Belgium, 2019). Statistics from the Organisation for Economic Cooperation and Development (OECD) show that Belgium is much stricter regarding dis- missing permanent workers as compared to any other European country and the United States—that is, on a scale of 1 (very loose) to 5 (very strict), Belgium scored 3, whereas the United States had a value of only 1.2 (OECD, 2019). As a result, Belgian firms are seriously discouraged to restructure their labor stock by dismissing and replacing employees.

However, this restructuring through the departure and replacements of employees is of particular importance for large firms with low levels of knowledge intensity, that is, the firms for which we find a positive effect of turnover. The largest and least knowledge-intensive firms in our data are highly recognizable brand name firms operating in widely different industries (e.g., retail, mail delivery, public transportation). They are all operating in con- stantly changing environments, fostering a need to continuously restructure. Most of them aim to become more cost-efficient by realizing the same or higher output with fewer employ- ees in order to cope with their main competitors or because technological advancements have obviated specific low knowledge-intensive jobs. For example, Bpost, the Belgian mail deliv- ery company, has announced that it will install automatic mail-sorting devices in the next years, thereby replacing the jobs of low-skill workers. Another example is Carrefour, one of the largest warehouses in Belgium, which will invest more in e-commerce, thus obviating in-store personnel.

Given that large low knowledge-intensive firms have enough resources to deal with the high dismissal costs (as well as the costs and disruptions associated with other types of turn- over), turnover may help to pursue these pressing restructurings. High levels of turnover enable to adjust the labor stock by not replacing employees leaving the firm or by replacing them with new profiles. Firms that need restructurings and have the resources to deal with them but have only limited turnover rates face a serious competitive disadvantage over firms with high turnover rates. They are more likely to end up with a mismatch in their labor stock in terms of both numbers and skill mix. This may explain why we found a positive effect of turnover for large firms with low levels of knowledge intensity. Note that in the discussion of the contributions and limitations of this study (see further), we provide several suggestions for future research in order to further improve our understanding of the potential positive effect of turnover.

De Meulenaere et al. / Collective Turnover, Firm Size, and Knowledge Intensity 1015

Overall, and in line with CETT, we conclude that context matters and that CETT research can benefit from taking into account the interplay of organizational contingencies as well as institutional factors when investigating the implications of firm-level collective turnover.

Contribution to the Literature

This study contributes to the turnover literature in at least four substantial ways. First, while two competing views on the moderating role of firm size in the collective turnover– performance relationship have sporadically appeared in the turnover literature, theoretical grounding for the perspectives was missing, and the direction in which firm size influences the impact of turnover remained unclear (Park & Shaw, 2013). This study provides an explicit theoretical framework for the positive and negative perspectives and introduces knowledge intensity as an additional moderator to reconcile these competing viewpoints.

Specifically, in our theoretical framework, we both follow and extend CETT (Nyberg & Ployhart, 2013) by building on the resource-based view of the firm. CETT focuses particu- larly on human and social resources and considers these valuable, scarce, noninimitable and nonsubstitutable resources as a source of competitive advantage (Nyberg & Ployhart, 2013). Taking a resource-based perspective, CETT then argues that because turnover depletes the pool of human and social resources, it undermines firms’ competitive advantage and, thus, harms organizational performance. We used these arguments in Hypothesis 2. In the present study, however, we also extend this resource-based perspective of CETT by introducing financial and organizational resources (in addition to human and social resources) to the theoretical framework. More specifically, we consider the firm’s pool of human, social, financial, and organizational resources as a means to cope with turnover consequences. As the pool of resources depends upon firm size (cf. Hypotheses 1a and 1b), our extended resource-based view of collective turnover helps to further emphasize the importance of context in CETT. Moreover, in terms of turnover consequences, we focus not only on the depletion of human and social capital but also on operational disruptions following turnover. The latter received less explicit theoretical attention in CETT (scholars like Dess and Shaw, 2001, and Brymer and Sirmon, 2018, do take it into account) but can become an integral part of the CETT framework via our extended resource-based view.

Second, our findings broaden the current understanding of the firm-level context in which turnover effects take place. Previous scholars have studied the effects of firm-level turnover in narrow study contexts with respect to firm size and knowledge intensity. They focused almost exclusively on large firms, ignoring the small and medium-sized firms that are con- sidered the backbone of the U.S. and European economies (Williamson et al., 2002)—that is, in the United States, they account for almost two thirds of the new jobs in the private sector (USTR, 2018), and in Europe, they represent 99% of the population of enterprises (Eurostat, 2018). They also focused primarily on one specific industry, such as the banking industry (Hale et al., 2016), restaurant chains (Kacmar et al., 2006; Shaw, Duffy, et al., 2005), or the employment activity sector (Glebbeek & Bax, 2004), implying minimal variance in firm- level knowledge intensity. Ignoring small and medium-sized firms and the variance in knowl- edge intensity has undeniably limited the generalizability of previous research findings (Allen et al., 2014), which is evidenced by the wide variety in effect sizes that has been found over the years (Hancock et al., 2013). This has led to claims that organizational context

1016 Journal of Management / April 2021

should be given more theoretical concern in turnover research (Allen et al., 2014; Hancock et al., 2013; Hom et al., 2017; Nyberg & Ployhart, 2013; Park & Shaw, 2013; Shaw, 2011). In fact, CETT has been created as an answer to these claims. However, to date, organiza- tional-level moderating contingencies remain seriously understudied in CETT research (Brymer & Sirmon, 2018; Shaw, 2011). This is probably because it is difficult to obtain data for a large sample of firms covering a wide variety of contexts in terms of firm size, knowl- edge intensity, industry, and so on (Allen et al., 2014). By means of our large multiorganiza- tion sample of firms of all sizes (N = 6,941), we were able to guarantee the generalizability of our findings and to take such an urgently needed organizational-level contextual approach.

Third, one of the main premises of CETT is that the impact of the quantity of turnover (i.e., turnover rate) cannot be understood without taking into account the quality of turnover in terms of the KSAs firms lose when they are confronted with turnover (Nyberg & Ployhart, 2013). By studying knowledge intensity as an additional moderator, we provide theoretical and empirical support for this proposition. While the turnover rate refers to the quantity of turnover, our knowledge-intensity moderator embodies the qualitative dimension of turnover as it determines the type of human and social capital lost with employee departures. In accor- dance with CETT, our three-way interaction between turnover, firm size, and knowledge intensity indicates that organizational contextual contingencies can both reinforce and coun- terbalance one another in their influence on collective turnover effects. Thus, while most turnover studies have been displaying these and other firm-level contingencies as indepen- dent influencers of the turnover effects, we show in line with CETT that their interplay explains much more variance in the consequences of collective turnover. Future research might therefore further explore whether and how other important moderators may work together in their influence on turnover effects, further expanding the CETT (Nyberg & Ployhart, 2013) and the research agenda set by recent meta-analytic reviews (e.g., Hancock et al., 2013, 2017; Hausknecht & Trevor, 2011; Heavey et al., 2013, Hom et al., 2017; Park & Shaw, 2013).

Fourth, our findings add to the understanding of the distinction between “functional” and “dysfunctional” turnover (Allen et al., 2010; Dalton, Todor, & Krackhardt, 1982). Several scholars have proposed that although turnover is disruptive, it is not inherently bad for firm performance because it can reduce labor costs, enable the departure of poor performers, and inject new ideas and knowledge in the workforce. So far, most scholars who have studied the idea of functional turnover have therefore assumed that there is an optimal level of turnover, implying that a certain degree of turnover is beneficial, up to a point at which the drawbacks outweigh the benefits and increasing turnover rates harm firm performance (De Winne, Marescaux, Sels, Van Beveren, & Vanormelingen, 2018; Glebbeek & Bax, 2004; Shaw, 2011; Shaw, Gupta, et al., 2005). In addition to this curvilinear approach, our results encourage another perspective of functional turnover: the contextual approach. Our finding that firm- level employee turnover is positively associated with labor productivity in large firms with low levels of knowledge intensity reveals that, in some contexts, the benefits of turnover are more likely to effectuate. This insight extends the current understanding of functional versus dysfunctional turnover, and it may encourage future scholars to further explore how the orga- nizational or institutional context of the firm may influence the impact of employee turnover. In particular, we agree with previous scholars (e.g., Glebbeek & Bax, 2004; Shaw, Gupta, et al., 2005) that there are limits to the functionality of turnover, even in organizational or

De Meulenaere et al. / Collective Turnover, Firm Size, and Knowledge Intensity 1017

institutional contexts that evoke the benefits of turnover. We thus encourage future researchers to integrate the curvilinear and the contextual perspective so as to gain a more complete under- standing of the (potentially positive) implications of collective turnover.

Practical Implications

Our findings point out that firm size matters when organizations have to deal with turn- over, but not in an unambiguous way. While size is a constraint in highly knowledge-inten- sive firms, it appears to be a remedy for organizations with low levels of knowledge intensity. These findings suggest that particularly large highly knowledge-intensive and small low- knowledge-intensive firms would do well to focus on the retention of their employees. Past research on the antecedents of turnover (e.g., Allen, Shore, & Griffeth, 2003; Batt & Colvin, 2011; Combs, Liu, Hall, & Ketchen, 2006; Guthrie, 2001; Heavey et al., 2013; Huselid, 1995) could be inspiring. For example, several authors have found that HR practices (e.g., growth opportunities, decision-making participation, and rewards) can encourage employee retention (Allen et al., 2003; Heavey et al., 2013).

However, small and large firms may need to achieve employee retention in different ways. As small firms have few financial and organizational resources, they have fewer opportuni- ties for developing and implementing a formal retention policy. Moreover, they cannot ben- efit from economies of scale when they implement a new strategy (Sels, De Winne, Maes, Delmotte, Faems, & Forrier, 2006). For these reasons, investments in HR management may create lower returns in smaller than in larger firms. However, smaller firms have other strengths, like the lower degree of formality and the mental and physical proximity of firm leaders to employees (Hayton, 2003). This could imply that employees’ direct supervisors may play an important role in the retention of employees in small low knowledge-intensive firms. Research on the impact of leaders’ behaviors and styles supports this, as Eisenberger, Stinglhamber, Vandenberghe, Sucharski, and Rhoades (2002) found that perceived supervi- sor support positively affected employee retention of retail sales employees.

In contrast, large companies do have the resources to develop and implement formal reten- tion practices. Fostering employees’ workplace identity and their well-being within the firm are the most important actions firms can take for retaining knowledge workers (Jayasingam, Govindasamy, & Singh, 2016; Rothausen, Henderson, Arnold, & Malshe, 2017). However, their retention is a difficult challenge because they have many working opportunities outside of the firm (Allen et al., 2010; Sutherland & Jordaan, 2004), particularly in this era of global- ization, where the war for talent is raging. Thus, apart from focusing on their retention, schol- ars and practitioners should also find ways for large knowledge-intensive firms to cope with the departures of these valuable workers. One way to guard themselves against the loss of the valuable knowledge is by organizing themselves into smaller entities. In the past decades, an increasing number of large, established firms have generated smaller divestments or spin-offs (Parhankangasa & Arenius, 2003). The idea is that venture managers are encouraged to leave the parent firm and start new, smaller companies on their own. While the spin-off benefits from the large parent firm’s resources (e.g., financial resources for R&D investments), the parent firm benefits from the resources inherent to small firms. Most importantly, the knowl- edge transfer between the parent firm and the spin-off is guaranteed such that social capital remains intact and is further developed (Parhankangasa & Arenius, 2003). Research tapping

1018 Journal of Management / April 2021

into how this organizational refocusing may overcome the negative implications of turnover in large knowledge-intensive firms would be highly relevant.

Limitations and Avenues for Future Research

This study has several strengths, such as the large and representative sample of firms, the longitudinal design, and the heterogeneity of firms regarding firm size and knowledge inten- sity. Nevertheless, it also faces some limitations that might generate future research opportunities.

First, as our main objective was to investigate whether and when firm size buffered or reinforced turnover effects, we did not empirically take into account the proposed underlying mechanisms related to the positive and negative perspectives (e.g., quick and effective replacements vs. adaptation and socialization). Although we did build on solid theories and empirical evidence developed in prior studies for the theoretical argumentation behind our hypotheses (e.g., Hancock et al., 2013; Hausknecht et al., 2009; Hom et al., 2017; Josefy et al., 2015; Nyberg & Ployhart, 2013; Park & Shaw, 2013), we suggest that future research might contribute by investigating these mechanisms. This will not only corroborate the rel- evance and reliability of our findings; it will also provide support for the proposed causal relationship between turnover and firm performance.

Second, although we follow CETT in that the influence of context is not specific to a specific type of turnover (Nyberg & Ployhart, 2013), we can still imagine different effects for voluntary and involuntary turnover (Hausknecht & Trevor, 2011; McElroy, Morrow, & Rude, 2001). For example, one could expect that some resources will be more relevant in coping with voluntary rather than with involuntary turnover (i.e., a strong social network in small firms or financial slack in large firms may be more relevant for dealing with unexpected departures than with planned departures). For this reason, we believe it is worthwhile to rep- licate our analyses for different types of organizational turnover in order to generate a more complete understanding of the role of firm size.

Third, our study offers strong evidence that firm size and knowledge intensity are impor- tant moderators, which is particularly interesting because they characterize the basic struc- tural context of the firm (Datta et al., 2005; Hancock et al., 2013; Park & Shaw, 2013). However, we agree with Brymer and Sirmon (2018) and Shaw (2011) that many organiza- tional-level moderators still remain unexplored, such as the strategy of the firm, organiza- tional innovation, and firms’ management of turnover.

Fourth, though there is evidence that organizational turnover rates substantially vary between periods (Hausknecht & Trevor, 2011), we did not tap into the temporal dynamics of turnover. Recently, De Winne et al. (2018) found that the turbulence of turnover rates over time is negatively associated with labor productivity. In a similar vein, Call et al. (2015) showed for a sample of 108,357 employees of 988 retail units that the change in turnover rates is more detrimental to unit performance than the level of turnover and that the negative effect of turnover rate change is mitigated when turnover rates are high. The authors con- clude that “turnover in the real world is embedded within context and time” (Call et al., 2015: 1227). As a future research avenue, it might thus be interesting to combine our contextual approach with the temporal framework of Call et al. and test whether the moderating influ- ence of firm size and knowledge intensity (context) also holds when we take into account the temporal dynamics of turnover (time).

De Meulenaere et al. / Collective Turnover, Firm Size, and Knowledge Intensity 1019

Finally, our study revealed a positive effect of collective turnover for firm performance in some contexts (large, low knowledge-intensive Belgian firms). We further examined the data and elaborated on the Belgian institutional context to explain the potential dynamics behind this effect. We do believe, however, that future theoretical and empirical research may con- tribute by further scrutinizing the potential benefits of turnover, taking into account firms’ organizational and institutional context. Does the positive effect disappear in countries with- out a strict employee protection toward dismissals? Does it arise particularly in certain sec- tors of activities, like the food service and transportation industries, that have developed routines for the typically high turnover rates they have to deal with (Allen, 2006; M. Becker, 2004)? Is more turnover always better in these contexts, or is a more complex curvilinear effect at play? Future CETT research should address these and other questions in order to offer a better understanding of the phenomenon of turnover in general and of functional turn- over in particular.

Conclusion

We theoretically identified and reconciled two competing perspectives on the moderating role of firm size in the firm-level collective turnover–performance relationship. Testing a three- way interaction between turnover, firm size, and knowledge intensity, we found that firm size buffers the harmful performance effects of turnover in low knowledge-intensive settings, whereas firm size exacerbates the effects in highly knowledge-intensive firms. We believe that these findings help to better understand the complex implications of collective firm-level turn- over for firm performance, both for the turnover literature and for firms that must cope with turnover and manage their employees as they face the everyday consequences of turnover.

ORCID iDs

Kim De Meulenaere https://orcid.org/0000-0002-5503-0317 Elise Marescaux https://orcid.org/0000-0001-8516-9996

Notes 1. Although large firms need to spread their total financial slack over a larger number of (potentially leaving)

employees as compared to smaller firms, the former can assign their resources more efficiently as they benefit from economies of scope (Audia & Greve, 2006) and have more organizational resources to efficiently assign their finan- cial means (see further; Black, Noel, & Wang, 1999; Josefy et al., 2015). Therefore we argue that even though the specific financial slack (i.e., slack per employee) is not necessarily greater in larger firms, the greater pool of total financial slack will still provide a competitive advantage to larger firms when they are confronted with turnover.

2. We acknowledge that this is an arbitrary cutoff, such that the results pertaining to statistical significance are also arbitrary to some extent. However, this split-sample analysis is included to facilitate the interpretation of the three-way interaction between turnover, firm size, and knowledge intensity. Moreover, the findings are robust for other cutoffs.

References Abelson, M. A., & Baysinger, B. D. 1984. Optimal and dysfunctional turnover: Toward an organizational level

model. Academy of Management Review, 9: 331-341. Ackerberg, D., Benkard, C. L., Berry, S., & Pakes, A. 2007. Econometric tools for analyzing market outcomes. In

J. Heckman & E. Leamer (Eds.), Handbook of econometrics, Vol. 6: 4171-4276. Amsterdam: North-Holland.

1020 Journal of Management / April 2021

Allen, D. G. 2006. Do organizational socialization tactics influence newcomer embeddedness and turnover? Journal of Management, 32: 237-256.

Allen, D. G., Bryant, P. C., & Vardaman, J. M. 2010. Retaining talent: Replacing misconceptions with evidence- based strategies. Academy of Management Perspectives, 24: 48-64.

Allen, D. G., Hancock, J. I., Vardaman, J. M., & Mckee, D. L. N. 2014. Analytical mindsets in turnover research. Journal of Organizational Behavior, 35(S1): S61-S86.

Allen, D. G., Shore, L. M., & Griffeth, R. W. 2003. The role of perceived organizational support and supportive human resource practices in the turnover process. Journal of Management, 29: 99-118.

Alvesson, M. 2000. Social identity and the problem of loyalty in knowledge-intensive companies. Journal of Management Studies, 37: 1101-1123.

Audia, P. G., & Greve, H. R. 2006. Less likely to fail: Low performance, firm size, and factory expansion in the shipbuilding industry. Management Science, 52: 83-94.

Barber, A. E., Wesson, M. J., Roberson, Q. M., & Taylor, M. S. 1999. A tale of two job markets: Organizational size and its effects on hiring practices and job search behavior. Personnel Psychology, 52: 841-868.

Barney, J. 1991. Firm resources and sustained competitive advantage. Journal of Management, 17: 99-120. Barron, J. M., Black, D. A., & Loewenstein, M. A. 1989. Job matching and on-the-job training. Journal of Labor

Economics, 7: 1-19. Batt, R., & Colvin, A. J. 2011. An employment systems approach to turnover: Human resources practices, quits,

dismissals, and performance. Academy of Management Journal, 54: 695-717. Becker, G. S. 1975. Human capital: A theoretical and empirical analysis, with special reference to education (2nd

ed.). New York: National Bureau of Economic Research/Columbia University Press. Becker, M. C. 2004. Organizational routines: A review of the literature. Industrial and Corporate Change, 13:

643-677. Bell, A., & Jones, K. 2015. Explaining fixed effects: Random effects modeling of time-series cross-sectional and

panel data. Political Science Research and Methods, 3: 133-153. Black, D. A., Noel, B. J., & Wang, Z. 1999. On-the-job training, establishment size, and firm size: Evidence for

economies of scale in the production of human capital. Southern Economic Journal, 66: 82-100. Boone, C., Carroll, G. R., & van Witteloostuijn, A. 2004. Size, differentiation and the performance of Dutch daily

newspapers. Industrial and Corporate Change, 13: 117-148. Brüderl, J., & Schüssler, R. 1990. Organizational mortality: The liabilities of newness and adolescence.

Administrative Science Quarterly, 35: 530-547. Brymer, R. A., & Sirmon, D. G. 2018. Pre-exit bundling, turnover of professionals, and firm performance. Journal

of Management Studies, 55: 146-173. Call, M. L., Nyberg, A. J., Ployhart, R. E., & Weekley, J. 2015. The dynamic nature of collective turnover and unit

performance: The impact of time, quality, and replacements. Academy of Management Journal, 58: 1208-1232. Campbell, B. A., Ganco, M., Franco, A. M., & Agarwal, R. 2012. Who leaves, where to, and why worry? Employee

mobility, entrepreneurship and effects on source firm performance. Strategic Management Journal, 33: 65-87. Cardon, M. S., & Stevens, C. E. 2004. Managing human resources in small organizations: What do we know?

Human Resource Management Review, 14: 295-323. Carley, K. 1992. Organizational learning and personnel turnover. Organization Science, 3: 20-46. Coff, R. W. 1999. How buyers cope with uncertainty when acquiring firms in knowledge-intensive industries:

Caveat emptor. Organization Science, 10: 144-161. Coff, R., & Kryscynski, D. 2011. Invited editorial: Drilling for micro-foundations of human capital–based competi-

tive advantages. Journal of Management, 37: 1429-1443. Coleman, J. S. 1988. Social capital in the creation of human capital. American Journal of Sociology, 94: S95-S120. Combs, J., Liu, Y., Hall, A., & Ketchen, D. 2006. How much do high-performance work practices matter? A meta-

analysis of their effects on organizational performance. Personnel Psychology, 59: 501-528. Cosic, D. 2018. Wage distribution and firm size: The case of the United States. International Labour Review, 157:

357-377. Dalton, D. R., Todor, W. D., & Krackhardt, D. M. 1982. Turnover overstated: A functional taxonomy. Academy of

Management Review, 7: 117-123. Daniel, F., Lohrke, F. T., Fornaciari, C. J., & Turner, R. A., Jr. 2004. Slack resources and firm performance: A meta-

analysis. Journal of Business Research, 57: 565-574. Datta, D. K., Guthrie, J. P., & Wright, P. M. 2005. Human resource management and labor productivity: Does

industry matter? Academy of Management Journal, 48: 135-145.

De Meulenaere et al. / Collective Turnover, Firm Size, and Knowledge Intensity 1021

Dawson, J. F., & Richter, A. W. 2006. Probing three-way interactions in moderated multiple regression: Development and application of a slope difference test. Journal of Applied Psychology, 91: 917-926.

Dess, G. G., & Shaw, J. D. 2001. Voluntary turnover, social capital, and organizational performance. Academy of Management Review, 26: 446-456.

De Winne, S., Marescaux, E., Sels, L., Van Beveren, I., & Vanormelingen, S. 2018. The impact of employee turnover and turnover volatility on labor productivity: A flexible nonlinear approach. International Journal of Human Resource Management. doi:10.1080/09585192.2018.1449129

Eisenberger, R., Stinglhamber, F., Vandenberghe, C., Sucharski, I. L., & Rhoades, L. 2002. Perceived supervi- sor support: Contributions to perceived organizational support and employee retention. Journal of Applied Psychology, 87: 565.

Eurostat. 2018. Small and medium-sized enterprises (SMEs). Retrieved from http://ec.europa.eu/eurostat/web/struc- tural-business-statistics/structural-business-statistics/sme.

Fang, R., Duffy, M. K., & Shaw, J. D. 2011. The organizational socialization process: Review and development of a social capital model. Journal of Management, 37: 127-152.

Federal Public Service Employment Belgium. 2019. Collectief ontslag [Collective layoffs]. Retrieved from http:// www.werk.belgie.be/defaultTab.aspx?id=493

Fuentelsaz, L., Gomez, J., & Polo, Y. 2002. Followers’ entry timing: Evidence from the Spanish banking sector after deregulation. Strategic Management Journal, 23: 245-264.

Glebbeek, A. C., & Bax, E. H. 2004. Is high employee turnover really harmful? An empirical test using company records. Academy of Management Journal, 47: 277-286.

Granovetter, M. 1984. Small is bountiful: Labor markets and establishment size. American Sociological Review, 49: 323-334.

Greene, P. G., Brush, C. G., & Brown, T. E. 1997. Resources in small firms: An exploratory study. Journal of Small Business Strategy, 8: 25-40.

Groothuis, P. A. 1994. Turnover: The implication of establishment size and unionization. Quarterly Journal of Business and Economics, 33: 41-53.

Guthrie, J. P. 2001. High-involvement work practices, turnover and productivity: Evidence from New Zealand. Academy of Management Journal, 44: 180-190.

Hale, D., Ployhart, R., & Shepherd, W. 2016. A two-phase longitudinal model of a turnover event: Disruption, recovery rates, and moderators of collective performance. Academy of Management Journal, 59: 906-929.

Hancock, J. I., Allen, D. G., Bosco, F. A., McDaniel, K. R., & Pierce, C. A. 2013. Meta-analytic review of employee turnover as a predictor of firm performance. Journal of Management, 39: 573-603.

Hancock, J. I., Allen, D. G., & Soelberg, C. 2017. Collective turnover: An expanded meta-analytic exploration and comparison. Human Resource Management Review, 27: 61-86.

Hannan, M. T., & Freeman, J. 1984. Structural inertia and organizational change. American Sociological Review, 49: 149-164.

Hausknecht, J. P., & Holwerda, J. A. 2013. When does employee turnover matter? Dynamic member configurations, productive capacity, and collective performance. Organization Science, 24: 210-225.

Hausknecht, J. P., & Trevor, C. O. 2011. Collective turnover at the group, unit, and organizational levels: Evidence, issues, and implications. Journal of Management, 37: 352-388.

Hausknecht, J. P., Trevor, C. O., & Howard, M. J. 2009. Unit-level voluntary turnover rates and customer service quality: Implications of group cohesiveness, newcomer concentration, and size. Journal of Applied Psychology, 94: 1068-1075.

Hayton, J. C. 2003. Strategic human capital management in SMEs: An empirical study of entrepreneurial perfor- mance. Human Resource Management, 42: 375-391.

Heavey, A. L., Holwerda, J. A., & Hausknecht, J. P. 2013. Causes and consequences of collective turnover: A meta- analytic review. Journal of Applied Psychology, 98: 412-453.

Hitt, M. A., Biermant, L., Shimizu, K., & Kochhar, R. 2001. Direct and moderating effects of human capital on strat- egy and performance in professional service firms: A resource-based perspective. Academy of Management Journal, 44: 13-28.

Hollenbeck, J. R., & Wright, P. M. 2017. Harking, sharking, and tharking: Making the case for post hoc analysis of scientific data. Journal of Management, 43: 5-18.

Hom, P. W., Lee, T. W., Shaw, J. D., & Hausknecht, J. P. 2017. One hundred years of employee turnover theory and research. Journal of Applied Psychology, 102: 530-545.

1022 Journal of Management / April 2021

Huselid, M. 1995. The impact of human resource management practices on turnover, productivity, and corporate financial performance. Academy of Management Journal, 38: 635-672.

Jackson, S. E., & Schuler, R. S. 1995. Understanding human resource management in the context of organizations and their environments. Annual Review of Psychology, 46: 237-264.

Jayasingam, S., Govindasamy, M., & Singh, G. S. K. 2016. Instilling affective commitment: Insights on what makes knowledge workers want to stay. Management Research Review, 39: 266-288.

Josefy, M., Kuban, S., Ireland, R. D., & Hitt, M. A. 2015. All things great and small: Organizational size, boundaries of the firm, and a changing environment. Academy of Management Annals, 9: 715-802.

Kacmar, K. M., Andrews, M. C., Van Rooy, D. L., Steilberg, R. C., & Cerrone, S. 2006. Sure everyone can be replaced . . . but at what cost? Turnover as a predictor of unit-level performance. Academy of Management Journal, 49: 133-144.

Käpylä, J., Laihonen, H., Lönnqvist, A., & Carlucci, D. 2011. Knowledge-intensity as an organisational character- istic. Knowledge Management Research & Practice, 9: 315-326.

Kozlowski, S. W. J., & Bell, B. S. 2013. Work groups and teams in organizations: Review update. Retrieved from http://digitalcommons.ilr.cornell.edu/articles/927

Leana, C. R., & Van Buren, H. J. 1999. Organizational social capital and employment practices. Academy of Management Review, 24: 538-555.

Lepak, D. P., & Snell, S. A. 1999. The human resource architecture: Toward a theory of human capital allocation and development. Academy of Management Review, 24: 31-48.

Mawdsley, J. K., & Somaya, D. 2016. Employee mobility and organizational outcomes: An integrative conceptual framework and research agenda. Journal of Management, 42: 85-113.

McElroy, J. C., Morrow, P. C., & Rude, S. N. 2001. Turnover and organizational performance: A comparative anal- ysis of the effects of voluntary, involuntary, and reduction-in-force turnover. Journal of Applied Psychology, 86: 1294-1299.

Meijaard, J., Brand, M. J., & Mosselman, M. 2005. Organizational structure and performance in Dutch small firms. Small Business Economics, 25: 83-96.

Messersmith, J. G., Lee, J. Y., Guthrie, J. P., & Ji, Y. Y. 2014. Turnover at the top: Executive team departures and firm performance. Organization Science, 25: 776-793.

Mitchell, W. 1994. The dynamics of evolving markets: The effects of business sales and age on dissolutions and divestitures. Administrative Science Quarterly, 39: 575-602.

Mowday, R. T. 1984. Strategies for adapting to high rates of employee turnover. Human Resource Management, 23: 365-380.

Nahapiet, J., & Ghoshal, S. 1998. Social capital, intellectual capital, and the organizational advantage. Academy of Management Review, 23: 242-266.

Nakagawa, S., & Schielzeth, H. 2013. A general and simple method for obtaining R2 from generalized linear mixed- effects models. Methods in Ecology and Evolution, 4: 133-142.

Nyberg, A. J., Moliterno, T. P., Hale, D., Jr., & Lepak, D. P. 2014. Resource-based perspectives on unit-level human capital: A review and integration. Journal of Management, 40: 316-346.

Nyberg, A. J., & Ployhart, R. E. 2013. Context-emergent turnover (CET) theory: A theory of collective turnover. Academy of Management Review, 38: 109-131.

O’Brien, R. M. 2007. A caution regarding rules of thumb for variance inflation factors. Quality & Quantity, 41: 673-690.

O’Connell, P. J., & Byrne, D. 2010. The determinants and effects of training at work: Bringing the workplace back in. European Sociological Review, 28: 283-300.

Office of the U.S. Trade Representative. 2018. Small- and medium-sized enterprises. Retrieved from http://ustr.gov/ trade-agreements/free-trade-agreements/transatlantic-trade-and-investment-partnership-t-tip/t-tip-12

Olley, G. S., & Pakes, A. 1996. The dynamics of productivity in the telecommunications equipment industry. Econometrica, 64: 1263-1297.

Organisation for Economic Co-operation and Development. 2019. Employment protection legislation. Retrieved from https://www1.compareyourcountry.org/employment-protection-legislation/en/0//datatable/.

Parhankangas, A., & Arenius, P. 2003. From a corporate venture to an independent company: A base for a taxonomy for corporate spin-off firms. Research Policy, 32: 463-481.

Park, T. Y., & Shaw, J. D. 2013. Turnover rates and organizational performance: A meta-analysis. Journal of Applied Psychology, 98: 268-309.

De Meulenaere et al. / Collective Turnover, Firm Size, and Knowledge Intensity 1023

Pfeffer, J., & Cohen, Y. 1984. Determinants of internal labor markets in organizations. Administrative Science Quarterly, 29: 550-572.

Richard, P. J., Devinney, T. M., Yip, G. S., & Johnson, G. 2009. Measuring organizational performance: Towards methodological best practice. Journal of Management, 35: 718-804.

Rothausen, T. J., Henderson, K. E., Arnold, J. K., & Malshe, A. 2017. Should I stay or should I go? Identity and well-being in sensemaking about retention and turnover. Journal of Management, 43: 2357-2385.

Scott, J. 2000. Social network analysis: A handbook (2nd ed.). Newbury Park, CA: Sage. Sels, L., De Winne, S., Maes, J., Delmotte, J., Faems, D., & Forrier, A. 2006. Unravelling the HRM–performance

link: Value-creating and cost-increasing effects of small business HRM. Journal of Management Studies, 43: 319-342.

Semadeni, M., Withers, M. C., & Trevis Certo, S. 2014. The perils of endogeneity and instrumental variables in strategy research: Understanding through simulations. Strategic Management Journal, 35: 1070-1079.

Shalley, C. E., Gilson, L. L., & Blum, T. C. 2009. Interactive effects of growth need strength, work context, and job complexity on self-reported creative performance. Academy of Management Journal, 52: 489-505.

Sharfman, M. P., Wolf, G., Chase, R. B., & Tansik, D. A. 1988. Antecedents of organizational slack. Academy of Management Review, 13: 601-614.

Shaw, J. 2011. Turnover rates and organizational performance: Review, critique, and research agenda. Organizational Psychology Review, 1: 187-213.

Shaw, J. D., Duffy, M. K., Johnson, J. L., & Lockhart, D. E. 2005. Turnover, social capital losses, and performance. Academy of Management Journal, 48: 594-606.

Shaw, J. D., Gupta, N., & Delery, J. E. 2005. Alternative conceptualizations of the relationship between voluntary turnover and organizational performance. Academy of Management Journal, 48: 50-68.

Shaw, J. D., Park, T. Y., & Kim, E. 2013. A resource-based perspective on human capital losses, HRM investments, and organizational performance. Strategic Management Journal, 34: 572-589.

Siebert, W. S., & Zubanov, N. 2009. Searching for the optimal level of employee turnover: A study of a large UK retail organization. Academy of Management Journal, 52: 294-313.

Sparrowe, R. T., Liden, R. C., Wayne, S. J., & Kraimer, M. L. 2001. Social networks and the performance of indi- viduals and groups. Academy of Management Journal, 44: 316-325.

Starbuck, W. H. 1992. Learning by knowledge-intensive firms. Journal of Management Studies, 29: 713-740. Subramaniam, M., & Youndt, M. A. 2005. The influence of intellectual capital on the types of innovative capabili-

ties. Academy of Management Journal, 3: 450-463. Summers, J. K., Humphrey, S. E., & Ferris, G. R. 2012. Team member change, flux in coordination, and perfor-

mance: Effects of strategic core roles, information transfer, and cognitive ability. Academy of Management Journal, 55: 314-338.

Sutherland, M., & Jordaan, W. 2004. Factors affecting the retention of knowledge workers. SA Journal of Human Resource Management, 2: 55-64.

Van Ark, B., O’Mahony, M., & Timmer, P. 2008. The productivity gap between Europe and the United States: Trends and causes. Journal of Economic Perspectives, 22: 25-44.

Von Nordenflycht, A. 2010. What is a professional service firm? Toward a theory and taxonomy of knowledge- intensive firms. Academy of Management Review, 35: 155-174.

Watrous, K. M., Huffman, A. H., & Pritchard, R. D. 2006. When coworkers and managers quit: The effects of turn- over and shared values on performance. Journal of Business and Psychology, 21: 103-126.

Wernerfelt, B. 1984. A resource-based view of the firm. Strategic Management Journal, 5: 171-180. Williamson, I. O., Cable, D. M., & Aldrich, H. E. 2002. Smaller but not necessarily weaker: How small businesses

can overcome barriers to recruitment. In J. A. Katz & T. M. Welbourne (Eds.), Managing people in entrepre- neurial organizations: 83-106. Bingley, UK: Emerald Group.

Wiseman, R. M. 2009. On the use and misuse of ratios in strategic management research. In D. J. Ketchen & D. D. Bergh (Eds.), Research methodology in strategy and management: 75-110. Bingley, UK: Emerald JAI.