The aim of this study was to investigate the development of job insecurity and employee well-being in response to short-time work as a disruption of employment, taking into account previous unemployment as a potential moderator. The findings contribute to a growing body of studies on the long-term temporal mechanisms between job insecurity, its antecedents and outcomes (e.g., Kinnunen et al.,
2014; Klug et al.,
2019; Watson & Osberg,
2017), as well as the impact of past labor market experiences on later job insecurity and well-being (Barrech et al.,
2011; De Witte,
2016; Ellonen & Nätti,
2015; Erlinghagen,
2008). Investigating the long-term repercussions for employee well-being, our findings also complement existing studies on immediate employee reactions to short-time work (Rauvola et al.,
2022; Wels et al.,
2022), and support job insecurity as an important explanatory mechanism underlying lingering effects on well-being.
4.1 Trajectories of Job Insecurity and Well-being Following Short-time Work
Our findings show that despite serving to prevent job loss, short-time work can have the paradoxical effect of triggering job insecurity, with negative repercussions for employee well-being: Short-time work predicted higher levels (H1), as well as an immediate increase in perceived job insecurity, followed by a decrease over time (H2). With regard to employee well-being, both levels (H3) and changes (H4) in job insecurity perceptions were negatively related to levels and changes in well-being, with the exception that changes in job insecurity were not related to changes in life satisfaction. Overall, short-time work was related to less favorable trajectories of well-being over time via job insecurity (H5). These findings are in line with previous research on organizational change and career disruptions as antecedents of job insecurity (Hofer et al.,
2021; Keim et al.,
2014), adding short-time work as a specific new determinant of insecurity perceptions. The findings are also in line with stress theories (Hobfoll,
1989; Lazarus & Folkman,
1984) and previous studies on the consequences of job insecurity (Jiang & Lavaysse,
2018).
What our study adds is a detailed analysis of the nonlinear temporal dynamics between antecedents, insecurity, and outcomes: Compared to the control group, employees affected by short-time work not only showed an immediate increase in job insecurity and a decrease in well-being, but these negative effects persisted 2 to 3 years after short-time work before they receded. This pattern suggests that short-time work may have long-lasting effects on employees’ perceptions of security, comparable to effects in the aftermath of actual layoffs (Dlouhy & Casper,
2021; Maertz et al.,
2010). Looking at longer time frames and considering job insecurity as an explanatory mechanism may also explain why previous research has found that restorative effects of short-time work on well-being were rather short-lived (Rauvola et al.,
2022).
Contrary to our expectations (H6), previous unemployment experiences did not seem to heighten employees’ susceptibility to job insecurity in response to objective antecedents (i.e., short-time work). In line with previous research (Ellonen & Nätti,
2015; Erlinghagen,
2008), past unemployment predicted present insecurity perceptions, but it does not seem to affect employees’ appraisals of the same situation as more threatening, at least not in the case of short-time work. Although these results require replication, they suggest that employee reactions to short-time work may be irrespective of previous labor market experiences.
4.2 Theoretical and Practical Implications
This study has several implications for theory and future research regarding the dynamics between job insecurity, its antecedents and consequences for employee well-being. First, regarding our research question how employee reactions to short-time work develop over time, our findings indeed suggest a nonlinear response in which employees’ job insecurity perceptions increased, then decreased, but remained elevated even after short-time work ended, only receding 3 years later (2 years when we took repeated short-time work into account). We also found different temporal dynamics between job insecurity and its antecedents compared to between job insecurity and well-being: As described above, short-time work had lingering effects on job insecurity perceptions. As insecurity perceptions decreased eventually, our results resemble a prolonged stress reaction model (Frese & Zapf,
1988), in which stress appraisals decrease only slowly over the years after the objective source of insecurity subsides: Employees showed elevated threat appraisals even after the objective stressor had ceased, but in contrast to an accumulation model, these reactions decreased eventually and did not persist throughout the whole study period. Regarding the consequences for well-being, the results resemble a more immediate stressor–strain reaction model (Frese & Zapf,
1988). That is, well-being responds immediately with synchronous decreases to increasing insecurity, but also synchronous improvements once perceived job insecurity is reduced (see also Bernhard‐Oettel et al.,
2020; Kinnunen et al.,
2014).
Second, our findings are generally in line with both appraisal theory (Lazarus & Folkman,
1984) and COR theory (Hobfoll,
1989), such that short-time work signaled a threat that triggered perceptions of job insecurity, which, in turn, were consistently linked to employee well-being over time. Considering the prolonged effect of short-time work on job insecurity, our findings are by comparison more supportive of COR theory, which states that resource loss tends to affect people more strongly than resource (re-)gain (i.e., the primacy of loss principle; Hobfoll et al.,
2018).
Third, it is an unresolved question in the literature why past unemployment may affect future job insecurity perceptions. Our findings did not support the explanation that previously unemployed people react to the same objective cue (i.e., short-time work) with more subjective insecurity than their colleagues, suggesting that selection into less secure jobs upon re-employment may be more likely (see Shoss,
2017). The lack of an interaction between previous unemployment and short-time work also challenges the assumption in both COR and appraisal theory that previous experiences of loss would affect people’s susceptibility to future loss or threat. However, according to appraisal theory, the effect of unemployment may additionally depend on how successfully or unsuccessfully people have coped with it (for example, if they quickly found a new, stable full-time job, the threat of job loss might be less severe in the future). More research is needed for a more differentiated analyses of these mechanisms, for example considering the length, frequency and recency of unemployment spells (see Erlinghagen,
2008). To reach a conclusion, more research is needed to systematically test these mechanisms in a comparative approach and for different antecedents of job insecurity (e.g., restructuring, temporary employment). Additionally, the scope could be expanded to predictors and moderators of job insecurity perceptions beyond unemployment, such as employability (De Cuyper et al.,
2012). It may also play a role how a previous job loss was experienced or attributed (e.g., to external circumstances or personal failure, see Winefield et al.,
1992).
In terms of practical implications, our study underlines the importance of reducing employment uncertainty as an important determinant of employee well-being, as many others have already noted (Cheng & Chan,
2008; Jiang & Lavaysse,
2018; Sverke et al.,
2002). Our analysis from the previous recession (2008–2009) can be informative regarding potential long-term effects of job retention schemes in the current crisis, which have been popular measures to buffer the economic repercussions of the pandemic (Drahokoupil & Müller,
2021).
We want to stress that our findings should not be read as advice to refrain from implementing short-time work. Obviously, it is a useful policy instrument during crises and preferable to mass layoffs, for employees, organizations, and societies alike (OECD,
2020b). Analyses of short-time work (furlough) in the UK during the pandemic indeed suggest that employees in job retention schemes reported worse mental health compared to secure employees, but were still better off than those who lost their jobs completely (Wels et al.,
2022). Nevertheless, employers should be aware of the side effects for employee well-being and strive to alleviate experiences of job insecurity, for example by improving organizational communication to reduce rumors, re-organizing work efficiently to avoid ambiguity, and developing such measures in a participatory approach with employees (Abildgaard et al.,
2018; Keim et al.,
2014).
4.3 Strengths and Limitations
The study has some notable strengths: The large-scale longitudinal data allowed us to delve into within-person processes of job insecurity as it unfolds over time, and investigate long-term, nonlinear trajectories, addressing questions regarding the onset and reversibility of effects (see Klug et al.,
2020,
2024a; Watson & Osberg,
2017). Keeping in mind that no single study can establish perfect causality, our study provides comparably strong evidence, because a) short-time work as our independent variable reflects an event with an identifiable onset, b) the repeated measures design allowed tracking the process unfolding upon the onset of short-time work, and c) the matched control group design accounted for selection effects into short-time work based on sociodemographic characteristics. As for effects of short-time work on employee perceptions, reverse causality (i.e., previous insecurity or well-being predicting short-time work) or self-selection based on other psychological dispositions and personality characteristics is unlikely, because short-time work is implemented for whole organizations or departments, not for individual employees. As for job insecurity and well-being, we analyzed synchronous relationships where reverse and reciprocal causation are possible (De Cuyper et al.,
2012). However, the parallel process models allowed separating baseline levels from within-person changes in the variables over time. All in all, we can therefore conclude that short-time work likely triggered the rise in job insecurity and, in turn, a decline in subjective well-being.
Some limitations should be kept in mind as well when interpreting the results. The first limitation concerns the operationalization of constructs: All variables were measured as self-reports such that we cannot rule out common method bias (Podsakoff et al.,
2003). However, our measurement models suggested that the variables differentiated sufficiently between one another, and it was unlikely that a single method factor explained all of the shared variance. Moreover, concepts such as perceived job insecurity, satisfaction and affective well-being require an individual’s introspection and are difficult to measure otherwise.
A related second limitation was due to the secondary analysis of existing data instead of a primary data collection specifically tailored to our study. This strategy was necessary to observe a large, representative sample over a long time frame, but limited the measurement of some constructs: Job insecurity and job satisfaction were each measured with single items which may have reduced their reliability (Gnambs & Buntins,
2017). However, regarding job satisfaction and job insecurity, single items have been supported as valid and reliable measures (Fisher et al.,
2016; Matthews et al.,
2022), and our results give no reason to suspect that these measures performed worse than the multi-item scales for life satisfaction and affective well-being. Regarding job insecurity, the item that we used did not allow differentiating affective from cognitive insecurity (see Shoss,
2017). However, specific mechanisms of cognitive versus affective job insecurity were not central to our research questions—we were interested in the overall experience of job insecurity, which includes both cognition and affect (Klug et al.,
2024a,
2024b). This seemed appropriate in light of evidence showing that cognitive job insecurity tends to show substantial correlations with affective job insecurity (Chirumbolo et al.,
2020; Jiang & Lavaysse,
2018). The correlations with short-time work as an objective indicator of insecurity also support this item’s validity in the study context. Additionally, single items of job insecurity have shown meaningful relationships with employee well-being, but with lower effect sizes than multi-item scales (Sverke et al.,
2002). Therefore, the effects in this study may be underestimated, rendering Type I errors less likely (see also Gnambs & Buntins,
2017). Nevertheless, future research could replicate our findings with multidimensional job insecurity measures and physiological well-being indicators. Such an approach could also include qualitative job insecurity measures to explore whether short-time work may relate to concerns about deteriorating job quality, in addition to threats of losing the job as such.
Third, in our analysis of trajectories of job insecurity and well-being following short-time work, we were not able to account for time-varying predictors to rule out alternative explanations for the observed changes in the variables of interest (such as changes in task structures and workload, or turnover). We also did not differentiate effects depending on the duration of short-time work or the varying extent of reductions in working hours and earnings (Drahokoupil & Müller,
2021; Tušl et al.,
2021). Accounting for the duration would have complicated comparisons with the control group, whereas information about the intensity was not available.
Fourth, the yearly time lag between observations may have been too long to detect more complex dynamics between short-time work and its consequences. On the other hand, a strength of our design was that with short-time work, the onset of the stressor was clearly identified—in such cases, longer time lags are preferable, because exposure time is comparable for all respondents (Dormann & Van de Ven,
2016). Still, additional studies could provide a more nuanced analysis across shorter intervals and consider the duration and extent of short-time work. Future studies could also consider additional, potentially opposing mediators, such as recovery from job demands through reduced hours, which might explain why the total effects in the mediation models across time points were often non-significant (Möhring et al.,
2021; Rauvola et al.,
2022; see also Rucker et al.,
2011).
Regarding the generalizability of our findings, it should be noted that our study was situated in the specific context of Germany in the aftermath of the 2008 recession and focused on individual well-being. Due to the matching procedure, the sample was no longer random or representative of the whole active labor force in Germany, but rather of the subpopulation likely to encounter short-time work in their career. Typically, short-time work is more prevalent among labor market insiders (i.e., prime age, qualified, in permanent full-time employment), and in male-dominated sectors such as manufacturing and construction (Brenke et al.,
2013; Tarullo & Desiere,
2023). At large, this is what we also observe in our data, with the subpopulation in short-time work being more likely to be male, manual workers with vocational degrees whereas, for example, age was comparable to the general population. Nevertheless, the findings are in line with meta-analyses across a range of countries and employees showing positive effects of organizational change on perceived job insecurity (Keim et al.,
2014) and negative effects of job insecurity on employee well-being (Cheng & Chan,
2008; Jiang & Lavaysse,
2018; Llosa et al.,
2018; Sverke et al.,
2002). Since effect sizes between job insecurity, its antecedents and outcomes have been found to vary depending on the social safety net and economic climate (Fullerton et al.,
2020; Probst & Jiang,
2017; Sverke et al.,
2019), future studies should also test and replicate our findings in different countries and time periods. Comparative studies could also explicitly consider benefits and drawbacks of job retention schemes on labor market flexibility and the resulting effects for well-being on the population level.
Generalizing across different subgroups in the workforce also warrants some reflections: Our use of propensity score matching aimed at holding constant sociodemographic and job-related differences between employees with and without short-time work, to find the average pattern of effects irrespective of such influences. This approach and the necessary specifications for subgroup analyses in propensity score matching (Green & Stuart,
2014) limited our ability to test more moderators than previous unemployment. But some of the matching variables, such as age or contract type, could affect employee reactions to short-time work or job insecurity as additional moderators (see Cheng & Chan,
2008; De Cuyper & De Witte,
2007; Lübke & Erlinghagen,
2014). An employees’ family and financial situation may also influence the degree to which they feel dependent on their current job and thus influence the magnitude of effects (Richter et al.,
2014). Whereas we tested moderating effects of previous unemployment experience, future studies could take different moderators into account to investigate the generalizability across sociodemographic groups, job- and household contexts.
Finally, this paper focused on the population of employees and their job insecurity perceptions, because only people who have a job can report such perceptions. Therefore, the study may have been influenced by a healthy worker effect (Dahl,
1993). That is, individuals with very strong strain reactions may have dropped out of employment, as well as job-insecure individuals who may have actually become unemployed (see also Garst et al.,
2000). This may partly explain why some effects in the mediation analyses were rather small. However, the magnitude of relationships between objective antecedents such as short-time work, subjective stressors such as job insecurity, and outcomes of mental well-being is unlikely to be large for other reasons as well: First, especially in correlational studies, the complexities regarding measurement errors and potential unobserved influences render large effects unlikely. Second, the work environment is only one of many social, psychological and biological factors influencing an individual’s well-being trajectory, so that one particular stressor cannot reasonably be expected to explain a large share of the variance (Frese & Zapf,
1988).