From erosion to adaptation? How collective job searches undermine cooperation before, but not after COVID-19

Western workplaces have become more volatile and are increasingly characterized by high rates of employees searching for and switching jobs (Bidwell, 2013; Jung et al., 2022). Trends of deregulation, de-unionization, increased emphasis on individual interests and the COVID-19 pandemic that alienated employees from physical workspaces together fueled the flexibilization of work and made many people considering other jobs (Bidwell, 2013; de Lucas Ancillo et al., 2023; Dekker, 2017; Sargent and Domberger, 2007). While these trends have been documented, existing research has largely focused on individual job movements, leaving the broader implications of collective job searches underexplored.

Despite such trends, some existing research on collective job search rates mainly focuses on antecedents and less is known about its consequences (Heavey et al., 2013; Hom et al., 2017). Existing research on the consequences of collective job searches almost exclusively studies material performance indicators such as reduced productivity and service quality (e.g. Nienhüser and Matiaske, 2006; Park and Shaw, 2013). Yet, its social repercussions remain poorly understood. This is problematic, as workplace relationships and communication networks are difficult to establish when people are switching jobs on a regular basis (Dess and Shaw, 2001; Shaw et al., 2005). This can threaten workplace cooperation, which relies on stable interactions and shared expectations to develop collaborative norms (van Gerwen et al., 2018; Bercovitz et al., 2006; Feldman, 1984; Spagnolo, 1999). When cooperation is weakened, it can hinder the achievement of team goals and negatively affect employee well-being and performance (Duan et al., 2019; Gerards et al., 2018; Rotemberg, 2006).

This article pioneers an examination of whether collective job search rates at team-level undermine employee cooperation, thereby addressing a gap in both organizational and human cooperation literature. As team-based structures dominate modern workplaces, employees are required to collaborate toward shared goals (Koster et al., 2007). With job searches generally being the first instance in turnover processes in which employees start to withdraw from their jobs (Griffeth et al., 2000; Mobley, 1977), high rates of job searches within teams may destabilize interactions and discourage cooperative behavior.

Simultaneously, it is crucial to consider that the COVID-19 pandemic has likely intensified these dynamics, as it brought about radical changes to the way in which many employees work and cooperate (Bailey and Kurland, 2002; Bloom et al., 2015). Extensive research reported further implementation of hybrid working and technological innovation for workplace interactions, with people only limitedly returning to the office and losses in organizational commitment and well-being of employees (see Da et al., 2022; de Lucas Ancillo et al., 2023; Kniffin et al., 2021; Nieuwenhuis and Yerkes, 2021). This has weakened opportunities for communication with (un)familiar co-workers and reduced interdependency and interconnectedness among employees (Bussin and Swart-Opperman, 2021; Ipsen et al., 2021; Porkodi, 2022; Šmite et al., 2023). These changes may alter how collective job searches impact employee cooperation in post-COVID-19 workplaces. Combining data from 2016, 2018 and 2024, this article provides a rare comparison of pre- and post-COVID-19 workplaces.

To provide further depth, this study compares six sectors (financial services, health care, higher education, ICT, manufacturing and transport). As the nature of work differs across sectors (Sappleton and Lourenço, 2016; van Hoorn, 2018), in some sectors employees extensively count on each other's expertise, skills and materials to solve problems and to develop collective output, e.g. surgeons in the health care sector, whereas in other sectors employees operate relatively autonomously, e.g. truck drivers in transport and assembly line workers in manufacturing (Courtright et al., 2015; Frenkel and Sanders, 2007; Sloan, 2012; van der Vegt et al., 2001; Wageman, 1995). The more jobs require complex and nonroutinized work, the more employees are interdependent for cooperation and it could be that collective job searches are then also more problematic. Moreover, as major changes resulting from the COVID-19 pandemic, such as working from home and teleworking are more prevalent in occupations in the service than in industrial sectors (Barrero et al., 2023; Zwan et al., 2024), it could be that evolutions in the relation between collective job search rates and employee cooperation are sector-dependent. Altogether, we formulate the following research question:

Finally, this article contributes to the broader study of human cooperation. While research has largely examined individual-level determinants of cooperative behavior (Balliet and van Lange, 2013; Pinto et al., 1993; Fu et al., 2019), little is known about how the workplace context and the structure of group compositions affects individuals' willingness to cooperate (Otten et al., 2024). Only recently, some experiments investigated how newcomers in groups affect contributions in public good games (Grund et al., 2015; McCarter and Sheremeta, 2013; Otten et al., 2021). Although turnover in group members was not found to directly affect cooperation in public good games, it did negatively affect tie formation, group identification and cooperative norms. Different from when actors are artificially put together in lab experiments, in workplace settings co-worker job searches might drive employees to cease investing in current employees and in getting acquainted and developing ties with new employees (Call et al., 2015; De Stefano et al., 2019). Unlike lab experiments on turnover and cooperation, this study investigates real-world workplaces, examining whether collective job searches disrupt employees' willingness to cooperate beyond controlled experimental settings. In doing so, we extend research on cooperation in post-COVID-19 work contexts by showing that the relationship between collective job searches and cooperation may depend on how work is organized and carried out, with the COVID-19 pandemic providing a particularly relevant context to examine this.

Extensive prior research has outlined when employees are likely to cooperate. Generally, workplace cooperation is largely determined by the extent to which employees engage in repeated interactions (Acedo and Gomila, 2013; Balliet and van Lange, 2013; Bottom et al., 2006; Ferrin et al., 2008). More recent empirical work reinforces this, as the stability of ties among employees is essential for buffering against withdrawal (Peltokorpi and Allen, 2024). Employees are more likely to interact when their investment is likely to be reciprocated, which fosters trust, interdependency and stable cooperation (Bó, 2005; Van Lange et al., 2011; Molm et al., 2000; Bernerth and Walker, 2009). Recent evidence even shows that this is largely shaped by the organizational context, as co-worker support and satisfaction with internal communication enhance relational stability, reduce withdrawal and enable cooperative behaviors (Su et al., 2023). Cooperative interactions then aggregate into stocks of cooperative employee networks (Brehm and Rahn, 1997; Leana and van Buren, 1999; Shah, 1998) and norms that further govern individual propensities to cooperate (Bercovitz et al., 2006; Feldman, 1984; Spagnolo, 1999).

Yet, in teams where many employees search for different jobs, the conditions for employees to engage in cooperative interactions are violated for three reasons. First, job searches are accompanied by withdrawal behaviors (Johns, 2002; Kaplan et al., 2009; Zimmerman et al., 2016), characterizing teams with high collective job search rates by high absenteeism rates and lower job involvement. In such teams, employees will have less fruitful interactions and limited interpersonal ties, which negatively impacts their inclination to engage in cooperative interactions. Moreover, fewer cooperative interactions among employees also hamper the development of social norms of cooperation (Titlestad et al., 2019; van Kleef and Côte, 2018), leaving such teams with weaker norms that are more difficult to maintain (Friedkin, 2001). Second, as job searches turn into departures, remaining team members will continuously have to develop relations with newcomers (Call et al., 2015; De Stefano et al., 2019), as opposed to teams in which most employees have worked for a long time and developed close relationships (Gonzalez-Mulé et al., 2020). Third, high rates of co-workers looking for different jobs also create an environment with reduced prospects of long employment relations. This diminishes the likelihood of future interactions, i.e. the shadow of the future (Poppo et al., 2008), which, given the long-term perspectives of employees and their need for reciprocity (see, e.g. Axelrod, 1984), is likely to negatively affect employees' propensities to invest in cooperation (Bó, 2005; van Lange et al., 2011):

The extent to which collective job search effects are associated with employee cooperation likely varies between pre and post-COVID-19 workplaces, as new work practices have reshaped employee interactions and interdependencies in cooperation for several reasons. First, workplaces experienced a dramatic increase in working from home (Kniffin et al., 2021; Vyas, 2022). Although this can have benefits such as efficiency and better work-life balance (see Ipsen et al., 2021), it was also associated with less co-worker contact and feedback, reduced visibility at work and an increased risk of social isolation (Crandall and Gao, 2005; Felstead et al., 2003; Kossen and Van den Berg, 2022; Soga et al., 2022). With employees only limitedly returning to workplaces, both workplace cohesion and organizational commitment of employees are under pressure (Bussin and Swart-Opperman, 2021; Malik et al., 2020), which widens social distances between employees. Second, virtual interactions and meetings became standard after the pandemic (Karl et al., 2022; Standaert et al., 2022). While established relationships can be maintained virtually, forming new relationships and sustaining informal interactions remains challenging, reducing small talk and social support (Axtell et al., 2004; Harris, 2003; Lal et al., 2023; Bleakley et al., 2022). Put together, it can be argued that new ways of working have at least partially stripped cooperation of its interactive and social practice. Workplace interactions became less common (Van der Lippe and Lippényi, 2020), which reduced employee interconnectedness.

Reduced employee interactions and interconnectedness are subsequently likely to determine the disruptiveness of collective job searches. In environments where interdependency is high, employees are pushed to communicate, share information and help each other in order to achieve collective goals (Wageman, 1995), which makes many employees searching for jobs costly. In workplaces where the necessity of interconnectedness in cooperation is lowered, employees are more distant and less dependent on others. This makes high collective job search rates less disruptive for cooperative processes, and it is expected that collective job searches have a milder influence on employees' propensities to cooperate in post-COVID-19 workplaces:

The effect of collective job searches in teams on employees' propensities to cooperate may also be sector-dependent. The extent to which employees are interdependent for cooperative interactions is governed by how tasks, rewards and standardized procedures are structured (Balliet and Lindström, 2023; Courtright et al., 2015; Tjosvold and Tsao, 1989; Wageman, 1999). This is largely determined by the sector in which organizations operate, as institutional theory suggests that they are generally subject to different contingencies and sector-specific institutions such as shared norms, rules and beliefs (Donaldson, 2006; Scott, 2013; van Hoorn et al., 2018). For example, the public sector is typically much more exposed to accountability demands than private sector organizations (Almquist et al., 2013). Sectoral contingencies shape the degree of interdependence in cooperation by influencing how employees interact to achieve overarching goals. Sectors vary in the complexity and routinization of their operations, which affects the communication and collaboration required among workers. Service sectors such as higher education, health care, financial services and ICT often require high coordination to handle uncertainty, innovation and problem-solving (Fernández-Macías and Bisello, 2022; Fernández-Macías et al., 2023). These sectors also more heavily rely on the tacit knowledge of employees, as expertise cannot be easily standardized or documented, e.g. teachers engaging university students (van Hoorn, 2018). In contrast, industrial sectors like manufacturing and transport are typically characterized by a focus on efficiency, precision and scalability, often relying on formalized knowledge and standard procedures, which results in routine interdependence to achieve their objectives.

While work in sectors too encompasses different roles, ranging from administration to more strategic functions, and interdependence among employees thus also varies within sectors, the somewhat simplified distinction outlined above remains relevant for illustrating how the impact of collective job search rates varies between sectors. Many co-workers considering other jobs and the prospect of their imminent leaves is more disruptive in service sectors where innovation, problem-solving and adaptability are more often prioritized. Employees have to develop extensive cognitive and interpersonal skills to address ambiguity and unstructured challenges effectively (Chung-Yan, 2010; Valcour, 2007) and the lowered job involvement of many employees then puts the collective effort that is required to achieve these goals under pressure. In contrast, in industrial sectors cooperation more often follows standardized procedures, so even if many co-workers are less involved and consider other jobs, employees will continue to cooperate by adhering to the established lines of these procedures:

Moreover, the degree to which COVID-19 pandemic reshaped workplaces is not equally distributed across sectors, as working from home only becomes feasible when work is computerized and requires little physical presence (Barrero et al., 2023). Obviously, this is difficult in manufacturing and transport sectors, and rather easy in financial services and ICT sectors. Accordingly, these sectors respectively report very low and very high levels of remote work (Barrero et al., 2023). Yet, even in the sectors of higher education and healthcare, remote work increased. While in higher education some tasks continue to require physical presence, i.e. offline teaching, others do not, i.e. grading and preparing classes-. Moreover, research reports that combinations of offline and online teaching are likely to stay (González et al., 2023; Imran et al., 2023). For the health care sector, much of the work remains physical, but some of it shifted to remote work as well, including online clinical duties, remote monitoring of patients and writing reports at home (Garavand et al., 2022; Jones et al., 2023). Put generally, the COVID-19 pandemic altered the social nature of cooperation mainly in service sectors. Therefore, it is plausible that cooperation in service sectors has started to resemble cooperation in industrial sectors, in which employee distance in cooperation was already large before the pandemic. Accordingly, we expect sectoral differences in the effects of collective job searches in teams on employee cooperation to be smaller after COVID-19 than before. We formulate:

This study combines all three cross-sectional waves from the European Sustainable Workforce Survey [ESWS] (Van der Lippe et al., 2016, 2018; Van der Put et al., 2024). The first wave was conducted in 2015/16, the second in 2017/18 and the third wave after the COVID-19 pandemic in 2023/24. To answer our research questions and to keep the waves compatible with each other, this study only relies on the data from the Netherlands. This is particularly relevant as the Dutch response to the COVID-19 pandemic was relatively strict, including workplace closures and a strong emphasis on working from home where possible (see, e.g. Oude Groeniger et al., 2021; Van den Broek-Altenburg and Atherly, 2021). This widespread shift to remote work brought significant technical implications and has had a lasting impact on Dutch workplaces. This makes the Netherlands a suitable context for examining how the relationship between collective job searches and employee cooperation evolved in workplaces that changed post-COVID-19.

The data is further stratified based on sector, including health care, higher education, financial services, ICT, manufacturing and transport, and size of the organization, i.e. organizations with less than 100 employees, between 100 and 250 employees, and more than 250 employees. The sectors were selected based on the challenges they face to the sustainability of their workforce. These include severe labor shortages, a lack of demographic diversity, i.e. unequal distribution of personnel in terms of gender, age and educational level, and pressures brought by technological innovation. Particularly in these sectors, it is important to investigate how to effectively address these challenges.

The ESWS has as its main goal to contribute to a better understanding of what works for realizing a sustainable (future) workforce. It is a unique data collection project as it contains information at three levels. At the individual level, employees were asked a range of questions, resulting in information on, for example, employee productivity, satisfaction, cooperation and work-life balance. At the team level, the data includes information on both team characteristics such as team cohesion and team productivity, and information on team managers such as their well-being. At the organizational level, the data includes information on characteristics of the establishment, with topics such as employer investments in employees, organizational policies and organizational culture. The multilevel structure of the ESWS enables the examination of how high collective job search rates in teams affect individual employees' propensities to engage in cooperation with other employees.

After setting the criteria for the strata, organizations were randomly selected from the Dutch Commercial Register, which includes all Dutch legal entities and businesses. Organizations were approached via e-mail or telephone, after which management and/or the board of the organization had to give permission for participation. Surveys were generally distributed digitally, and in some cases on request with pencil and paper. The survey was filled out at organizational level by HR managers or CEOs, at the team level by team managers and at individual level by employees. Organizations were asked to participate with at least three teams. Participating organizations were rewarded with a benchmark report, in which they received information on the performance of their organization. Within the benchmark report, organizations were also compared to other organizations in the same sector and to the total sample.

A total of 112 organizations, 348 teams and 5372 employees participated in at least one of three ESWS waves collected within the Netherlands. Between waves, response rates for organizations ranged across waves from 89% to 98%, for teams from 65% to 75% and for employees from 48% to 54%. For several reasons, only a subset of this sample qualifies for our study. First, we excluded respondents for whom managers and/or CEOs did not fill out the survey, as for these respondents information is missing on both team and workplace level. By doing this, we excluded 1,308 employees, 6 teams and 7 organizations. Second, as we examined employee cooperation in team settings and not among dyads, we excluded teams with fewer than three employees or teams for which we sampled fewer than three employees. With this step, a total of 56 employees from 36 teams and 6 organizations were excluded. Remaining missing values were dealt with by applying multiple imputation (MI) techniques. Five datasets were imputed in SPSS 27, using fully conditional specification and iterative Markov Chain Monte Carlo methods. Missing values in variables ranged from 0% to 22.9%, for which an overview can be found in Appendix A. The MI procedure assumes that data are missing at random, meaning that missing values can be inferred from the observed data. The five imputed datasets were subsequently pooled according to Rubin's rules (Rubin, 2004). Our final sample consists of 4,008 employees, 306 teams and 99 organizations.

Employee Cooperation. Following previous research (van Dyne and LePine, 1998), employees' propensities to cooperate with others – i.e. horizontal cooperation – were measured by asking them four items on whether they frequently assisted other employees with work tasks, whether they helped co-workers to adapt to a new working environment and whether they found it important to work together and cooperate well with others. Respondents were asked on a 5-point scale ranging from “strongly disagree” to “strongly agree.” Internal consistency of our measure for employee cooperation was good, with α = 0.71 (M = 3.95, SD = 0.50).

Collective Job Search Rates. Our measure for collective job searches in teams was based on the proportion of employees in teams that are searching for a different job. Prior research showed that job searches reflect employee withdrawal and that high collective job searches reflect high turnover contexts as searching for a different job is an essential step in the job switch process (Bartunek et al., 2008; Mobley, 1977; Griffeth et al., 2000). Employees were asked whether or not they actively searched for a different job in the past six months, where a score of “1” indicated that they were searching for a different job and a score of “0” indicated that they were not. Subsequently, the number of job searches per team was divided by the total number of members per team, which constitutes our measure for collective job searches. On average, the proportion of job searches in teams was 0.19, with SD = 0.13.

Sectors. Our measure for sectors is derived from the Dutch Commercial Register. Sectors in which organizations operate were labeled based on what they indicated to be their main activities. The sectors of health care (33%), higher education (14%), financial services (8%) and ICT (5%) were labeled service sectors (60%), whereas the sectors manufacturing (27%) and transport (13%) were labeled industrial sectors (40%).

Pre- and Post-COVID-19 Workplaces. Our measure for comparing pre- and post-COVID-19 workplaces was based on the year in which the survey was conducted. Organizations that participated in one of the first two waves in 2015/16 and 2017/18 were labeled as pre-COVID-19 workplaces (71%). Organizations that participated in the third wave in 2023/24 were labeled as post-COVID-19 workplaces (29%).

Controls. At the individual level, we control for several factors. First, we control for whether employees themselves are withdrawn from work, as this could already contribute to frustrated cooperation (Goodman and Atkin, 1984; Pauly et al., 2002). It could be that withdrawn employees are also the ones who have lower propensities to cooperate, which would confound the collective job search effect. To account for this, we included the individual measure for employee job searches as a control variable. Second, given that the frequency of workplace interactions determines cooperation (Acedo and Gomila, 2013; Bottom et al., 2006; Ferrin et al., 2008), we control for the extent to which employees have opportunities to meet other team members. Models were controlled for actual hours worked and the number of days employees work from home. Moreover, we control for how demanding employees' jobs were, as this can negatively influence the amount of employee interaction (Goodboy et al., 2017). Third, we controlled for demographic characteristics by including age, education, occupational groups, i.e. based on collar type and skill level, gender, migrant status, tenure, having a partner and number of children.

We also control for several factors at the workplace level. First, to isolate collective job search effects, we control for another indicator of collective withdrawal by including team absenteeism rates, measured by asking managers on a 5-point scale how often their team encounters high absenteeism rates. Second, we controlled for team characteristics as these might govern the extent to which employees are willing to engage in cooperation with others (Bandiera et al., 2004). Models controlled for team atmosphere, measured by asking the manager about the general work climate in the team on a 5-point scale and team productivity, measured by asking the manager how productive their team is on a 5-point scale. Third, we control for other indicators of how work is structured, as this might affect the degree of cooperation in teams (Balliet and Lindström, 2023; Courtright et al., 2015). Our models contain four measures on a five-point scale for interdependency in work tasks, which include the frequency of working on long-term projects and working in project teams, and the degree of dependence in both work tasks and decision-making. Fourth, to account for within-sector role variation, ranging from administration to more strategic functions, we control for whether teams mainly have supportive or core tasks and for the size of the team.

Given that our analyses include both team- and individual-level variables, we estimated two-level mixed-effects models, with employees nested within teams. To assess the extent of clustering in the data, we first estimated a null model with the dependent variable employees' propensities to cooperate. The intraclass correlation indicated that 5.6% of the variance was attributable to the team level and 3.4% to the organizational level. The larger proportion of variance at the team level supports accounting for team-level clustering in the analyses and therefore the use of mixed-effects models rather than ordinary least squares regression. We focused the multilevel specification on the team level because team membership represents the primary clustering structure relevant to our research questions and the focal independent variable is measured at the team level.

Our analyses consisted of three key building blocks. First, we tested hypothesis one by estimating a model with the independent variable collective job search rates and all control variables. Second, to test Hypothesis 2 and 3, both interaction terms for pre- and post-COVID-19 workplaces and sectors were included. These were added in iterative fashion but presented in one model. By including these interaction terms, we were able to assess whether the association between collective job search rates and employees' propensities to cooperate differed before and after COVID-19 and across sectors. After doing so, we tested our fourth hypothesis by including a three-way interaction effect of collective job searches, sectors and pre- and post-COVID-19 workplaces. This interaction tested whether the effect of the COVID-19 period on the relationship between collective job search rates and employees' propensities to cooperate varied across sectors.

For ease of interpretation, results are presented for the full sample and separately for pre- and post-COVID-19 workplaces. These split-sample analyses are intended to facilitate the interpretation of the interaction effects by showing the estimated relationships within each period. The formal tests of differences between periods are based on the interaction terms estimated in the full-sample models.

Given that we imputed our missing data, all coefficients, p-values and standard errors are pooled statistics, which were calculated based on Rubin's rules (Rubin, 2004). Pooled descriptive statistics are presented in Table 1. An overview of bivariate analyses and correlations between all variables can be found in Appendix B.

For our first hypothesis, we expected employees in teams with higher proportions of co-workers searching for different jobs to be less inclined to engage in cooperation. As can be derived from Table 2, we find no significant association between collective job search rates in teams and employees' propensities to engage in cooperation and reject our first hypothesis. This means that, across the full sample of workplaces observed between 2015 and 2024, there is no general effect of collective job searches on employee cooperation. This overall null effect, however, does not preclude the possibility that the impact of collective job searches may vary under different contextual conditions, which we explore in subsequent analyses.

For our second hypothesis, we expected the association between collective job searches and cooperation to be less strong in post-COVID-19 workplaces as compared to pre-COVID-19 workplaces. For this interaction term, we find a significant positive effect, with B = 0.44 and p = 0.003. This means that, as compared to pre-COVID-19 workplaces, employees in post-COVID-19 workplaces are less likely to refrain from cooperating in teams with high job search rates and we confirm Hypothesis 2. For ease of interpretation, Figure 1 visually presents this finding. Zooming in, Table 2 also shows that it is informative to interpret this together with the direct effect of collective job search rates on employee cooperation. When splitting the data for pre- and post-COVID-19 workplaces, we find the expected significant negative association between collective job search rates and employee cooperation for pre-COVID-19 workplaces (B = −0.20, p = 0.045), but not for post-COVID-19 workplaces (B = 0.23, p = 0.068). This suggests that prior to the COVID-19 pandemic, employee cooperation was more disrupted in teams with higher job search rates than in teams with lower job search rates, but that this effect disappeared after the pandemic and even became a marginally significant positive effect. This could also explain why the negative association between collective job search rates and employee cooperation is absent when analyzing the full sample.

Moving to our third hypothesis, we expected the negative association of collective job searches with employee cooperation to be less negative in the industrial sectors of manufacturing and transport as compared to the service sectors of finance, health care, higher education and ICT. For this interaction term, Table 2 shows a significant positive interaction effect (B = 0.29, p = 0.049). Contrary to our expectations, this means that in service sectors an increase in the number of co-workers searching for different jobs has a less strong negative impact on employee cooperation than in industrial sectors, and we reject our third hypothesis. This finding suggests that the structure and nature of work in industrial sectors may make cooperation more sensitive to collective job searches than in service sectors. When breaking the analysis down into examining the interaction effects for all six sectors separately, we find that it is only the health care sector in which collective job searches are less disruptive to employee cooperation (B = 0.42, p = 0.019), with the manufacturing sector as reference category. For the other sectors, no significant interaction effects were found.

Finally, for our fourth hypothesis, we tested a three-way interaction to examine how the effects of collective job search rates before and after the COVID-19 pandemic varied across sectors. We expected the differences between sectors to be smaller in post-COVID-19 workplaces. No significant evidence was found for the existence of a three-way interaction effect, with F(7, 4007) = 0.50 and p = 0.84. Accordingly, we reject our fourth hypothesis. An overview of rejected and confirmed hypotheses can be found in Table 3.

To assess the robustness of our findings, we performed several sensitivity analyses for which all coefficients can be found in Appendix C. First, in our sample, 85.7% of employees have a permanent contract, whereas 14.3% work on temporary or flexible contracts. This could affect our results, as employees with temporary contracts need to orient themselves to future steps and thus often search for jobs, and we might overestimate the effects of collective job searches by including these employees. Furthermore, employees on a temporary contract might more often use job searches to review their employability and job searches then do not reflect desires to leave and withdrawal as much as they do for employees on a permanent contract. To account for this, we reran our models keeping only employees with permanent contracts. For several effects, results showed minor differences from our main models. The interaction effect of COVID-19 workplaces was slightly weaker but remained a significant positive association. The interaction effect for sectors and the negative effect of collective job searches on employee cooperation for pre-COVID-19 workplaces jumped to being marginally significant. From this, we deduce that although some effects are slightly weaker, the essence of our findings remains intact.

Second, to make a more refined comparison between pre- and post-COVID-19 workplaces, we conducted an analysis using data from the periods closest to the pandemic. The results of an analysis that only includes data from the second (2018) and third (2024) waves reveal that directions and significance levels were similar to those observed in our original models, but effect sizes of the main effects were even stronger.

Third, one of the items in our dependent variable workplace cooperation presumes that staff turnover occurs, as respondents are asked whether they help new employees adjust to their workplaces. Given that some teams likely experience no turnover at all, or at least have no employees searching for different jobs, and to exclude some overlap between our independent and dependent variables, we conducted an analysis using a dependent variable comprised of the three remaining workplace cooperation items (α = 0.62). Results were highly similar to our main models.

Fourth, our main models include several team-level control variables that measure levels of withdrawal, workplace interactions and interdependence among employees. This may suppress our main effect of collective job search rates on employee cooperation and to account for this, we reran our models without the team-level control variables of team absenteeism, team atmosphere, working on long-term projects, working in a project team, work dependence and decision dependence. The results were very similar to our main analysis.

In recent decades, the Dutch labor market experienced trends of shortened employment relations and increased rates of employees looking for jobs. Yet, little is known about the consequences of collective employee job searches on the day-to-day functioning of organizations. Some initial research showed that it can reduce tie and network formation among co-workers, which may put the social fabric of workplaces under pressure. This article builds on existing knowledge by examining whether high collective job search rates in teams are negatively associated with employee cooperation and whether these associations differ between pre- and post-COVID-19 workplaces. Both analyses accounted for the type of work and degree of co-worker interdependencies by making comparisons across sectors. Three waves of unique employer-employee linked data from before and after the COVID-19 pandemic were combined, which includes 4,008 employees from 306 teams and 99 organizations.

The main conclusion of this article is that high collective job search rates in teams erode employee propensities to engage in cooperation before, but not after, the COVID-19 pandemic. At first glance, it could be seen as positive when new forms of online cooperation reduce physical and social barriers, allowing employees to collaborate even without prior familiarity (Wellman, 2001; Vriens and Van Ingen, 2018). Employees might not have to know each other well to still work together and might easily cooperate with new employees, which makes the job searches of team members less problematic. However, considering evidence of post-COVID employee disconnectedness – reduced cohesion, lower commitment and increased isolation (Bussin and Swart-Opperman, 2021; Malik et al., 2020; Šmite et al., 2023) – it is likely that teams became so disconnected that individual job searches have minimal impact on others, reflecting a weakened social fabric in post-COVID workplaces.

Second, the findings in this study do show that in pre-COVID workplaces, collective job search rates in teams are negatively associated with employees' propensities to engage in cooperative interactions. This shows that beyond the impact on material indicators such as organizational performance and service quality, many employees searching for different jobs at once can affect the social fabric of organizations. This is likely due to the fact that collective employee job searches make co-worker interactions scarce, limit interpersonal ties and reduce the prospect of future interactions, which together weakens norms of cooperation. This has important implications for organizational practice. Although some ways of working might have permanently changed, it could be that we still experience the tail of how COVID-19 endangered social cohesion and relations (Hartz et al., 2023; Larsen et al., 2023). Yet, it is not unthinkable that workplaces and societies bounce back, and can even improve through adaptation, resilience and transforming practices (Coady, 2024; Rockström et al., 2023). This study reveals that it is vital that, in this process, we should also have learned that extra attention has to be paid to the consequences of high rates of collective employee job searches. If not, we run the risk of falling back into pre-COVID-19 patterns in which teams with high rates of job searches have eroded employee cooperation, and therefore both the ability of employees to achieve collective goals, their well-being and performance (Duan et al., 2019; Gerards et al., 2018; Rotemberg, 2006).

Finally, this article concludes that there is little sectoral variation in the effects of collective job search rates on employee cooperation. While we initially find that the effect in service sectors is less strong than in industrial sectors, a breakdown of the analysis shows that this only holds for the health care sector. It could be that although the degree of complexity and routinization in work tasks might generate different types of interdependencies among employees, it does not affect employees' inclination to set up cooperative interactions themselves. Following prior research (Acedo and Gomila, 2013; Balliet and van Lange, 2013), we speculate that helping co-workers is then a rather social affair and determined by both repeated and expected future interactions. Effects in the health care sector might then be divergent because here the work must be continued and accommodated. It could be that searching for another job has less impact on cooperative behavior of healthcare employees, as cooperation is most critical to their work. For example, surgeons searching for other jobs still have to perform complex operations, where they are forced to fully rely on each other. In such contexts, lower job involvement might not translate to diminished propensities to cooperate.

While we aimed to capture heterogeneous experiences of the COVID-19 pandemic by examining sectoral differences, it is important to acknowledge that such heterogeneity may lie less between sectors and more within organizations. Although institutional theory suggests that sectors develop shared norms and cognitive frameworks (see, e.g. Scott, 2013), implying some degree of overlap within sectors, our findings suggest that heterogeneity in COVID-19 experiences and cooperation patterns might be more likely to occur within organizations rather than between sectors. When organizational-level contingencies outweigh sector-level institutional variation, this helps explain why sectoral differences remained limited in our analyses. Future research should therefore investigate whether organizational-level contingencies such as size, technological infrastructure, managerial practices and internal policies produced stronger differences in how employees experienced and responded to the pandemic. Nonetheless, our findings indicate that the pandemic period coincided with observable shifts in cooperation patterns overall, underscoring the importance of continued attention to post-COVID workplace dynamics.

To conclude, our findings have several theoretical and practical implications. We contribute to theory building by demonstrating that the relationship between collective job searches and employee cooperation is not universal but context-dependent. Where previous research primarily considered the impact of high job search rates on organizations as being uniform across workplaces, our results show that this changed after the COVID-19 pandemic. This pattern suggests that major structural changes in work arrangements, such as hybrid and remote collaboration, can alter the social mechanisms through which turnover intentions affect cooperative behavior. Our findings therefore extend research on cooperation in post-COVID-19 and hybrid work contexts by showing that the consequences of collective job searches may depend on how work is organized and on the social interdependence it creates. This extends existing organizational and cooperation theory by emphasizing that cooperative norms are embedded in transforming work contexts rather than fixed social structures. This contextual sensitivity calls for more dynamic models of employee withdrawal and cooperation that consider how changes in technology, physical proximity and social connectedness reshape behavior at work.

From a practical perspective, our findings suggest that organizations can proactively shape the conditions under which cooperation is sustained, even when collective job search rates are high. Assuming that the post-COVID disappearance of the effects reflects workplaces becoming more loosely connected -so that individual withdrawal matters less-it becomes crucial for managers to focus on maintaining relational stability rather than merely enforcing physical or social presence. Designing cohesive hybrid structures, for example through regular team rituals, structural collaboration platforms, mentoring relationships and cross-team projects (Friedman and Holtom, 2002; Orazani et al., 2023), can reinforce trust and cooperation, even when employees work apart. Yet, at the same time, our findings underline the importance of addressing the sources of volatility that lead employees to search for jobs in the first place. Extensive research has shown that efforts to strengthen job embeddedness, for example through stable co-worker relations, transparent internal mobility pathways and opportunity for personal growth, can prevent the job searches of many employees at once (see Mitchell et al., 2001). Enhancing employees' sense of stability and belonging may prevent the erosion of cooperative relationships before it begins. Overall, these implications suggest that cooperation is not only maintainable under volatility but also safeguarded by reducing volatility itself. Organizations that simultaneously support connection and stability are more resilient to turnover shocks and better equipped to preserve cooperative norms in the long term.

This study ought to be interpreted in light of its limitations. First, it relies on cross-sectional data and is therefore unable to make causal inferences. Instead of collective job search rates eroding employee cooperation, it could be that low employee cooperation drives job search rates instead. This was partially addressed by creating temporal separation between our measures for collective job search rates and employee cooperation, which is a necessary condition for being able to conclude that our independent variable precedes the dependent variable (Wunsch et al., 2010). While this clearly does not fully overcome our limitation of not being able to account for reversed causation bias and future research ought to invest in collecting panel data, this article still gives a first insight into how high job search rates are associated with reduced organizational social fabric, while comparing between pre- and post-COVID workplaces.

Second, this article relies on three waves of cross-sectional data, which presents issues for wave compatibility. For example, the health care sector is overrepresented in the third wave and different levels of cooperation between sectors could make comparisons between waves less accurate. Future research could attempt to follow employees and organizations over time, but the compensation should simultaneously match the notable time investments of employees and organizations.

Third, this article was unable to distinguish between other societal-level trends that might have occurred between 2016 and 2024. While the COVID-19 pandemic undoubtedly posits the biggest shock to employment and workplaces in this period, other major trends could also have had an effect on workplaces and therefore drive our findings. For example, it could be that in times of increased labor shortages and elevated turnover rates, collective employee job searches are especially likely to affect employee cooperation. Low job involvement of co-workers and the prospect of imminent departures might then further increase the workloads for already overburdened employees that remain (Kelly and Moen, 2020). This could then especially limit opportunities for establishing ties with co-workers and thus cooperation. This all the more makes the call for extended research on the impact of societal trends and shocks on the consequences of collective job searches relevant.

All in all, our study demonstrates that high collective job search rates have the potential to erode the social fabric of organizations through reductions in employees' propensities to cooperate. It also showed that this effect is contingent upon the type of workplaces, as it disappeared after COVID-19. This underscores the importance of investigating the consequences of collective job search within the context of current social and economic climates. It also highlights the need to pay attention to how workplaces can be restructured to mitigate and prevent the further erosion of their social fabric. Otherwise, we risk that the damage to workplaces and employee relationships as a result of trends of increased turnover and reduced interconnectedness because of COVID-19 takes on a lasting character.

The supplementary material for this article can be found online.