UvA-DARE (Digital Academic Repository) Job preservation efforts: When does job insecurity prompt performance?

Purpose – While job insecurity generally impedes performance, there may be circumstances under which it can prompt performance. The purpose of this paper is to examine a specific situation (reorganization) in which job insecurity may prompt task and contextual performance. The authors propose that performance can represent a job preservation strategy, to which employees may only resort when supervisor-issued ratings of performance are instrumental toward securing one ’ s job. The authors hypothesize that because of this instrumentality, job insecurity will motivate employees ’ performance only when they have low intrinsic motivation, and only when they perceive high distributive justice. Design/methodology/approach – In a survey study among 103 permanent employees of a company in reorganization, the authors assessed perceived job insecurity, intrinsic motivation and perceived distributive justice. Supervisors rated employees ’ overall performance (task performance and organizational citizenship behaviors). Findings – Multilevel analyses showed that job insecurity was only positively related to supervisor-rated overall performance among employees with low intrinsic motivation and, unexpectedly, among employees who experienced low distributive justice. Results were cross-validated using employees ’ self-rated performance, replicating the findings on distributive justice but not the findings on intrinsic motivation. Research limitations/implications – The results can inform future research on the specific situations in which job insecurity may prompt job preservation efforts, and call for research to uncover the mechanisms underlying employees ’ negative and positive responses to job insecurity. The results and associated implications of this study are largely based on conceptual evidence. In addition, the cross-sectional design warrants precaution about drawing causal inferences from the data. Originality/value – By combining insights from coping responses and threat foci, this study advances the understanding of when and why job insecurity may prompt performance. 2: p ¼ 0.05; Model 3: p ¼ 0.09). Employee-rated contextual performance To cross-validate our findings, we explored whether the relationships with supervisor-rated performance would be to of contextual performance. we calculated the correlations between supervisor-rated OCB-I and OCB-O and employee-rated OCB-I and OCB-O.

significant stressor that results in poorer performance (e.g. Cheng and Chan, 2008;Gilboa et al., 2008). Yet, there are also a few studies that support the latter reaction, indicating that job insecurity can create a motive to secure one's job, resulting in better performance (e.g. Probst et al., 2007). According to this job preservation perspective (cf. Shoss, 2017), employees may attempt to demonstrate their worth to their employer by devoting extra effort toward behaviors that will be noticed and valued, such as task performance and organizational citizenship behaviors (OCB) (Shoss, 2017;Shoss and Probst, 2012;Huang et al., 2013).
Yet, the cases in which job insecurity prompts performance are rare, suggesting that the potentially motivating effect of job insecurity may only emerge in specific situations and/or under certain conditions. Arguably, employees may particularly resort to job preservation strategies in the form of performance when their performance is evaluated for dismissal decisions. For example, during mergers or organizational downsizings, supervisors' performance ratings may determine which employees will "make the cut" (Borman, 1991). In such situations, job insecurity may motivate employees to engage in behaviors that could result in better supervisor-issued ratings of their overall performance, such as in-role and extra-role behaviors (Huang et al., 2013;Shoss, 2017). In this study, we include these different facets of performance (i.e. task and contextual performance) to reflect employees' overall performance in response to job insecurity. We aim to advance the understanding of conditions under which job insecurity may prompt rather than impede overall performance. As such, we will examine the potentially positive effects of job insecurity on performance in the context of an organizational reorganization. Examining the job insecurity-performance relationship in this specific context can help us to better demarcate and understand the deviant positive findings from the dominant negative findings, in ways that cannot be anticipated from extrapolations of the existing literature.
Using both Lazarus and Folkman's (1984) stress and coping framework and job preservation theory (cf. Shoss, 2017) as a theoretical backdrop, we regard changes in performance in response to job insecurity as a behavioral, problem-focused coping strategy, rather than as an outcome that is negatively affected by its strain ( Jordan et al., 2002;Lazarus and Folkman, 1984;Selenko et al., 2013). Inherent to this view, however, is the assumption that job insecure employees believe that their performance will influence their chance of job continuance (Brockner et al., 1992;Lam et al., 2015). That is, the assumption that job performance represents a job preservation strategy hinges on the belief that performance is instrumental toward securing one's job, and, hence, implies that the motivating effects of job insecurity on performance will depend on its instrumentality.
Here, we examine when job insecurity motivates employees to engage in job preservation strategies in the form of performance. We hypothesize that the instrumental nature of performance during a reorganization implies that job insecurity may only motivate employees' performance when their performance is typically not driven by internal motives (i.e. low intrinsic motivation), and when their performance is typically fairly rewarded within the organization (i.e. high distributive justice). We thus take into account that employees' behavioral reactions to job insecurity are not only determined by individual factors but also depend on how organizations treat their employees (Sverke et al., 2002;Wang et al., 2015).

Theoretical background and hypotheses
To uncover the conditions under which job insecurity may represent a job preservation strategy that prompts performance, we draw on Lazarus and Folkman's (1984) stress and coping framework. In this framework, coping is defined as the cognitive and behavioral efforts to manage or reduce the demands created by the stressful situation (Lazarus and Folkman, 1984). In general, coping strategies can be divided into problem-focused 288 CDI 25,3 strategies and emotion-focused strategies. Problem-focused coping strategies are directed at altering or changing the stressor, whereas emotion-focused coping strategies are directed at regulating or managing one's emotional reactions to the stressor. Research shows that especially active, problem-focused coping strategies, aimed at changing the situation, can buffer the negative effects of job insecurity on indicators of well-being (e.g. Cheng et al., 2014).
Yet, people are not necessarily inclined to resort to this active, problem-focused type of coping. In fact, research into coping behaviors suggests that employees under stress are most likely to adopt passive coping behaviors, rather problem focused or emotion focused (Catalano et al., 1986). This suggests that the dominant reaction to job insecurity is a passive one, which is supported by the majority of research showing that job insecurity forms a stressor that interferes with employees' effort, performance and satisfaction (cf. Cheng and Chan, 2008;De Witte et al., 2015;Jiang and Lavaysse, 2018;Sverke et al., 2002;Piccoli et al., in press;Fischmann et al., 2018). Still, there are some exceptions to this rule. For example, job insecure employees have been found to put more effort in their work than less job insecure employees during organizational restructuring (Brockner et al., 1992;Galup et al., 1997), and there is evidence that especially moderate to high levels of job insecurity canagain, under certain circumstanceslead to enhanced efforts at work in the form of task performance (Probst et al., 2007) and OCB (Lam et al., 2015).
Past research has sought to explain why employees sometimes respond to job insecurity by increasing their performance. At the core of these explanations lies the assumption that job insecurity may motivate employees to work harder in order to secure their positions (i.e. job preservation motivation, Shoss, 2017). That is, when facing job insecurity, creating a positive workplace image can be highly relevant for maintaining one's job. As such, employees may want supervisors to see that they are valuable to the organization (Huang et al., 2013). Job insecurity, then, becomes a motive to engage in impression management tactics, reflected in performance-related behaviors such as intensifying work efforts (i.e. task performance) and helping out colleagues or contributing to the organization (i.e. contextual performance). Engaging in these behaviors can give the supervisor a more positive impression of the employee and may result in better supervisor-issued performance ratings (Huang et al., 2013). Given the crucial role of supervisors' evaluations of overall performance for personnel decisions (Borman, 1991), these employees may believe that their chance on actual job loss might be reduced if they demonstrate these performance behaviors (Huang et al., 2013;Lam et al., 2015). Yet, it remains unclear under which circumstances job insecurity becomes such a job preservation strategy that prompts performance, rather than a stressor that impedes performance.
We propose that job insecurity may motivate employees to increase their overall performance when such performance is, or is assumed to be, instrumental toward securing one's job. This instrumentality can both illuminate specific circumstances under which job insecurity may prompt performance, and helps to understand why job insecurity, in general, rarely prompts performance. That is, in most situations, performance may not be directly associated with employees' chance of job continuance, or employees may not believe that their performance will influence their chance of job continuance. However, when performance is associated with job continuance, employees may recognize that by improving their task and contextual performance they will be seen as assets to the organization, thereby enhancing their chance to secure their job. Arguably, this may particularly be the case in situations such as mergers and organizational restructuring, where employee performance is often directly linked to layoff decisions (Borman, 1991). Indeed, when a positive link between job insecurity and performance is found, it is mostly found in contexts where participants are threatened by layoffs or have just experienced one (e.g. Brockner, 1988;Probst et al., 2007). Of course, for performance to represent a job 289 Job preservation efforts preservation strategy, employees also need to believe that higher supervisor-issued performance ratings imply a lower risk of dismissal, and thus that putting in more effort at work is a functional impression management strategy for them to counteract job loss (Staufenbiel and König, 2010). While there can be other reasons for employees to put in more effort at work during reorganizationssuch as enhanced workload or improving the organization's position and therewith the security of its membersemployees generally view individual output as the best safeguard against involuntary job loss (Van Vuuren et al., 1991). Thus, employees likely believe that enhancing their performance during reorganization is a fruitful and instrumental strategy to preserve their job. Yet, even in a situation where the instrumentality of performance in response to job insecurity is relatively straightforward, there may be additional requirements for job insecurity to translate into active, problem-focused coping in the form of performance. That is, our assumption that job insecurity will prompt performance in such situations is predicated on the idea that performance will be driven by instrumental rather than by intrinsic motives, and that employees believe that their performance will be rewarded by their organization. Thus, we propose that the motivating effect of job insecurity on performance during a reorganization may only emerge when employees' motivation to perform well does not arise from within themselves (i.e. low intrinsic motivation), and when their organization tends to reward performance in a fair way (i.e. high distributive justice).

Intrinsic motivation
We propose that the extent to which an employee is driven by internal rewards to perform well in his/her job, i.e., their intrinsic motivation (Warr et al., 1979), may be one of the essential factors that determines whether job insecurity can prompt performance during a reorganization. In general, motivational theory (Pinder, 2011;Ryan and Deci, 2000) suggests that performance can be determined by intrinsic motivationin which the motivation to perform well arises from within, because it is naturally satisfyingas well as by extrinsic motivationin which the motivation to perform well arises from external factors, because it is instrumental to a different outcome. In that sense, a job preservation strategy in the form of performance represents extrinsically motivated behaviorbehavior that is instrumental toward securing one's job.
Given the instrumental nature of performance as a job preservation strategy, we propose that job insecurity will only motivate those employees who are not intrinsically motivated to perform well. This assumption builds on the literature on performance, in which a distinction can be made between typical and maximum performance (e.g. Klehe and Anderson, 2007;Sackett et al., 1988). Typical performance situations represent enduring work situations, whereas maximum performance situations represent short and evaluative work situations during which the implicit or explicit instruction to invest effort is quite apparent (Klehe and Anderson, 2007). In a typical performance situation, employees are not aware that their performance is being evaluated, are not consciously attempting to perform to the best of their ability and are loosely monitored over an extended period of time. In a maximum performance situation, however, employees are aware that their performance is being evaluated, accept implicit or explicit instructions to maximize their effort and are evaluated for a relatively short time. In such situations, employees usually show higher effort and higher contextual performance compared to their typical, day-to-day, performance.
Yet, research on self-determination theory (Ryan and Deci, 2000) suggests that employees with high intrinsic motivation may work as hard under typical performance conditions as they do under maximum performance conditions, while employees with low intrinsic motivation may work harder under maximum performance conditions than under typical performance conditions (Klehe and Anderson, 2007). Put differently, employees with 290 CDI 25,3 high intrinsic motivation are inclined to perform well regardless of the external situation, while those with low intrinsic motivation may only do so when they believe that their performance is instrumental to a different outcome. Here, we consider a reorganization as a situation similar to a maximum performance situation, in the sense that employees may implicitly recognize that their overall performance is being evaluated for personnel decisions for a relatively short timethat is, until supervisors determine which employees will "make the cut." Following Klehe and Anderson (2007), we expect that during a reorganization, employees with low intrinsic motivation are likely to put more effort into performance-related behaviors that will be noticed and valued by their supervisors (i.e. impression management tactics), while employees with high intrinsic motivation may put the same amount of effort into these behaviors. Engaging in higher task and contextual performance is, then, a strategic way for low intrinsically motivated employees to bolster their reputation as a "good actor" (Bolino, 1999), driven by instrumental motives to keep their job (Lam et al., 2015;Schreurs et al., 2012).
Thus, we propose that low intrinsically motivated employees will respond to job insecurity through impression management tactics, reflected in higher supervisor-issued ratings of performance. Yet, we expect no such responses for high intrinsically motivated employees: H1. The relationship between employee perceived job insecurity and supervisor-rated overall performance is moderated by employee intrinsic motivation, in such a way that perceived job insecurity is only positively related to performance when intrinsic motivation is low.

Perceived distributive justice
We propose that employees' perceived distributive justice within the organization may be another essential factor that determines whether job insecurity can prompt performance.
In fact, we expect that the positive effects of job insecurity on performance during reorganization are likely to emerge only when the employee perceives that performance is typically fairly rewarded within the organization. Distributive justice isnext to procedural and interactional justiceone of the components of organizational justice that refers to the extent to which employees are treated fairly by the organization (Colquitt et al., 2013). Employees form distributive justice perceptions by comparing the ratio between their efforts (time, energy and training) and the rewards they receive (pay, support and security) to the ratios of others. In this study, we focus specifically on distributive justice as it reflects how performance is generally rewarded within an organization, which provides an important guide for employees to direct their behaviors needed to deal with job insecure situations (cf. Wang et al., 2015). In addition, when people make overall fairness judgments, perceptions of distributive justice tend to be more salient and influential than other forms of justice (Colquitt, 2001). High distributive justice may particularly prompt performance in response to job insecurity during reorganization. When performance is typically fairly rewarded with valued outcomes within an organization, employees are more likely to believe that enhancing their efforts and contributions to the organization may also result in a higher chance of job continuance. Thus, when facing job insecurity, distributive justice may provide an indication for employees to establish the likelihood of fair treatment in dismissal decisions (Lind, 2001). Put differently, high distributive justice may function as a proxy and, hence, support employees' belief that putting more effort in their job can result in more security (Colquitt et al., 2006). We therefore expect that employees who perceive high distributive justice will exert extra effort and engage in more OCB in response to job insecurity during reorganization, in the hope that they will obligate theirfairorganization to provide job continuance. In contrast, employees who perceive low distributive justice may feel that their job continuity 291 Job preservation efforts is less predictable (De Witte, 2005), regardless of their performance levels. In such an unfair work environment, they may believe that increasing their performance is fruitless because the organization tends to be unfair when it comes to rewarding efforts (Wang et al., 2015). Employees who perceive low distributive justice are therefore unlikely to resort to active coping in the form of performance: H2. The relationship between employee perceived job insecurity and supervisor-rated overall performance is moderated by employee perceived distributive justice, in such a way that perceived job insecurity is only positively related to performance when perceived distributed justice is high.

Methods
Participants and procedure A sample of 125 employees from a technical maintenance department of a privatized public transportation company in the Netherlands was invited to fill out a paper-andpencil questionnaire about their job experiences. At the time of study, the company was in the initial phase of a reorganization process to meet severe government cuts, which were publicly announced a few months earlier. Because of these cuts, employees were already notified that the work package of their department was to be greatly reduced. Participants (n ¼ 103, response rate 82.4 percent) were aged between 22 and 64 years (M ¼ 46.23, SD ¼ 12.19) and a majority of them were male (98.1 percent). Most participants (61.2 percent) were electrotechnical technicians, 30.1 percent of the participants were mechanical technicians and 8.7 percent had administrative or managerial jobs. Participants had high school or vocational education (89.3 percent) or a bachelor or master degree (10.7 percent). The supervisors of five teams were asked to rate the overall performance of the employees in their own team. Supervisors rated a minimum of 12 and a maximum of 29 employees.

Measures
Unless indicated otherwise, employees were asked to indicate their level of agreement with all items on a five-point Likert scale ranging from (1) strongly disagree to (5) strongly agree. Employee task and contextual performance was assessed by their supervisors. Job insecurity. We measured job insecurity using three items reflecting the quantitative dimension of job insecurity (perceived threats to the continuity of the job itself ) derived from Hellgren et al. (1999). An example item is: "There is a risk that I will have to leave my present job in the year to come." Intrinsic motivation. We measured intrinsic motivation using six items (Warr et al., 1979), such as "I like to look back on the day's work with a sense of a job well done." One item in which a word was missing ("I feel a sense of personal satisfaction when I do this job"; missing word: well) was removed.
Perceived distributive justice. We measured perceived distributive justice using three items (Colquitt, 2001). An example item is: "The reward I receive from my company reflects the effort I put into my work." Supervisor-rated overall performance. Supervisors were asked to rate the current overall performance of each employee with three items on a ten-point scale ranging from (1) extremely poor to (10) outstanding. This scale includes task and contextual performance (OCB) items. The supervisors rated: the task performance of the employee, an employee's willingness to support peers (OCB-I) and the effort an employee puts into the organization (OCB-O). The average correlation among the three items was 0.66 and the Cronbach's α of the three items together was 0.85.

CDI 25,3
Employee self-rated contextual performance. Although supervisors are generally able to provide accurate and complete pictures of an employee's task and contextual performance (cf. Williams and Anderson, 1991), not all employee work behaviors may be within the purview of the supervisor. Furthermore, employees may have a different view on their work behaviors than their supervisor. We therefore explored whether the relationships with supervisor-rated overall performance would be comparable to those with employee self-rated contextual performance [1]. Specifically, we measured employees' OCB-I (e.g. I am willing to offer my time to help others who have work-related problems) and their OCB-O (e.g. I defend the organization when other employees criticize it), each with eight items (Lee and Allen, 2002), ranging from (1) never to (5) always.
Control variables: demographics, perceived organizational support and job satisfaction. We included three demographic covariates (age, education and tenure) because they have been found to affect employee performance assessments Feldman, 2009, 2010). In addition to demographics, we also included organizational support and job satisfaction as covariates because they have been found to relate to task and contextual performance (Alessandri et al., 2017;Williams and Anderson, 1991). More specifically, based on reciprocity, job performance tends to increase when employees perceive high organizational support (Eisenberger et al., 2001). Perceived organizational support was assessed with seven items, such as "The organization really cares about my well-being" (Eisenberger et al., 1997). Additionally, positive attitudesin particular job satisfactionare key influencers of job performance (Ajzen and Fishbein, 2005;Judge et al., 2001). Job satisfaction was assessed with four items from Judge et al. (2001), such as "Most days I am enthusiastic about my work." Table I presents the means, standard deviations, internal consistencies and correlations between all variables in this study.

Preliminary analyses
Construct validity. To examine whether our scales represented separate constructs, we conducted confirmatory factor analyses involving different combinations of the five employee measures (job insecurity, intrinsic motivation, distributive justice, organizational support and job satisfaction). The hypothesized five-factor model provided a reasonable fit ( χ 2 (179) ¼ 265.68, CFI ¼ 0.914, TLI ¼ 0.900, RMSEA ¼ 0.069) and explained the data better than alternative measurement models (e.g. one-factor model, χ 2 (189) ¼ 822.97, CFI ¼ 0.363, TLI ¼ 0.221, RMSEA ¼ 0.181; three-factor model with intrinsic motivation and job satisfaction together, and justice and organizational support together, χ 2 (186) ¼ 506.44, Multilevel data structure. Because employees can be considered as nested within work units and supervisors may systematically differ in their performance ratings, the data collected within work units were not independent from each other. Non-interdependence among observational data violates a basic assumption of traditional linear model analyses and results in α-error inflation (Raudenbush and Bryk, 2002). By means of multilevel analyses, it is possible to control for the dependence of data stemming from the same units and to keep the α-error level constant. We therefore first assessed if there was between work group variance within our data, which warrants multilevel regression analyses rather than ordinary linear regression analyses. Prior to analyses, all the predictor variables were centered at the group mean (see Kreft et al., 1995). An intercept-only model in a mixed model analysis showed that the ICC was 0.15, indicating differences in supervisor-rated overall performance between work units. Hence, we used multilevel regression analyses to test our hypotheses.  Hypotheses testing First, we estimated a model with fixed intercepts that explored the association between the demographic variables and performance and compared this model with the intercept-only model. The −2 log-likelihood of the overall model fit did not improve (−2 log-likelihood: showing that the demographic control variables did not relate to performance. Therefore and to save power, we removed the demographic control variables from the further analyses. To test H1, in which we proposed that intrinsic motivation would moderate the relationship between job insecurity and performance, we estimated three fixed intercept models with supervisor-rated performance as the dependent variable (see Table II): Model 1 included job insecurity, intrinsic motivation and the interaction of job insecurity and intrinsic motivation as independent variables; Model 2 extended Model 1 by adding the control variables perceived organizational support and job satisfaction; Model 3 additionally included perceived distributive justice. Results showed that the interaction of job insecurity and intrinsic motivation was significantly related to performance (Table II; Figure 1). To further examine the moderating role of intrinsic motivation, we conducted simple slope analyses. These analyses revealed that both the slope for low intrinsically motivated employees and for high intrinsically motivated employees was unrelated to performance at −1SD and +1 SD, and at −1.5 SD and +1.5 SD, respectively. Yet, at −2SD and +2 SD, the slope for low intrinsically motivated employees was positively related to performance    Job preservation efforts (t ¼ 2.045, p ¼ 0.04), while the slope for high intrinsically motivated employees remained unrelatedor marginally negatively relatedto performance (t ¼ −1.809, p ¼ 0.07). The results described above provided initial support for H1. Note, however, that after controlling for perceived organizational support and job satisfaction in Model 2, and for distributive justice in Model 3, the interaction term was no longer significant. This implies that the inclusion of these variables not only reduced available degrees of freedom and statistical power, but also the amount of explainable variance available in the dependent variable (Becker, 2005). Thus, the direct and moderating role of intrinsic motivation regarding supervisor-rated performance (see Model 1, Table II) could not explain incremental variance in performance above and beyond organizational support, job satisfaction and distributive justice.
H2, in which we proposed that perceived distributive justice would moderate the relationship between job insecurity and performance, was tested in a similar way as H1. Results showed that the interaction between job insecurity and perceived distributive justice was significantly related to performance (Table III; Figure 2). Simple slope analyses were used to further examine the moderating role of distributive justice. At −1SD and +1 SD, these analyses revealed no significant relationship between job insecurity and performance for employees with low or high distributive justice perceptions. Yet, at −1.5 SD and +1.5 SD, job insecurity was positively related to performance among employees with low distributive justice perceptions (t ¼ 2.379, p ¼ 0.02) and negatively related to performance among employees with high distributive justice perceptions (t ¼ −2.0142, p ¼ 0.05 at +1.5 SD).   Also at −2SD and +2 SD, job insecurity was positively related to performance among employees with low distributive justice perceptions (t ¼ 2.574, p ¼ 0.01) and negatively related to performance among employees with high distributive justice perceptions (t ¼ −2.211, p ¼ 0.03). The findings described above contradict H2: job insecurity was associated with higher performance for employees who perceive low distributive justice, and with lower performance for employees who perceive high distributive justice. The findings remained stable when perceived organizational support and job satisfaction (Model 2) and intrinsic motivation (Model 3) were entered into the regression: at −1SD and +1 SD, respectively, the slope for low distributive justice was positively related to performance (t ¼ 2.368, p ¼ 0.02 and t ¼ 1.997, p ¼ 0.05, respectively), while the slope for high distributive justice was unrelatedor marginally negatively relatedto performance (t ¼ −1.982, p ¼ 0.051 and t ¼ −1.867, p ¼ 0.07).
Additional analyses on performance outcomes Supervisor-rated task and contextual performance. To deepen the understanding of the pattern of findings in our study, we also tested our hypotheses with the three supervisor-rated performance measures separately (task performance, OCB-I and OCB-O) instead of with one overall supervisor-rated performance measure. Results and interaction patterns were largely similar to the results described earlier: the interaction between job insecurity and intrinsic motivation was significant in Model 1 for task performance ( p ¼ 0.01) and OCB-O ( p ¼ 0.05), although not for OCB-I. The interaction between job insecurity and distributive justice was (marginally) significant in all Models for task performance (Model 1: p ¼ 0.06; Model 2: p ¼ 0.04; Model 3: p ¼ 0.07), OCB-I (Model 1: p ¼ 0.00; Model 2: p ¼ 0.00; Model 3: p ¼ 0.00) and OCB-O (Model 1: p ¼ 0.06; Model 2: p ¼ 0.05; Model 3: p ¼ 0.09).
Employee-rated contextual performance. To cross-validate our findings, we explored whether the relationships with supervisor-rated performance would be comparable to employee-rated measures of contextual performance. First, we calculated the correlations between supervisor-rated OCB-I and OCB-O and employee-rated OCB-I and OCB-O. Supervisor-rated OCB-I correlated significantly with employee-rated OCB-I (r ¼ 0.49, p ¼ 0.00) and supervisor-rated OCB-O correlated significantly with employee-rated OCB-O (r ¼ 0.53, p ¼ 0.00). Although these correlations show substantial agreement among supervisors and employees, they also point to some differences in OCB assessments. We therefore ran our analyses again estimated with employee-rated OCB-I and OCB-O as dependent variables. This time, results showed no moderating effect of intrinsic motivation in any of the Models, while distributive justice did consistently moderate the relationship between job insecurity and employee-rated contextual performance, with the exception of OCB-O in Model 3 (Table IV ). The direction of the moderating role of distributive justice was similar to that in the main analyses: positive for employees with low distributive justice perceptions (e.g. Model 2: OCB-I: t ¼ 4.697, p ¼ 0.00; OCB-O: t ¼ 2.8169, p ¼ 0.01 at −1SD), while not significant for employees with high distributive justice perceptions (e.g. Model 2: OCB-I : t ¼ −0.396, p ¼ 0.69; OCB-O: t ¼ −0.4523, p ¼ 0.65 at +1SD). Together, our main analyses and additional analyses show that job insecurity is associated with higher performance outcomes for employees with low distributive justice perceptions.

Discussion
In the beginning of this paper, we questioned whether employees would perform better or worse when the continuity and stability of their job is at risk. Our findings suggest that employees show higher overall performance in response to job insecurity when they are not intrinsically motivated andsurprisinglywhen they feel that they cannot rely on their 297 Job preservation efforts organization to reward their performance fairly. The latter results on distributive justice were cross-validated when using employees' self-rated performance instead of their supervisors' ratings, and remained the same when controlling for perceived organizational support and job satisfaction. The results on intrinsic motivation, however, were less robust and did not hold when using self-rated performance, nor after controlling for perceived organizational support and job satisfaction. As such, the moderating role of intrinsic motivation should be interpreted with care. The results of this study do not clearly support our initial assumption that the motivating effect of job insecurity on performance only occurs when such performance is instrumental toward securing one's job. While we expected that job insecurity could become a motive to engage in performance-related impression management tactics for employees with low intrinsic motivation and with high distributive justice perceptions, we found that job insecurity prompted such performance behaviors for employees with low distributive justice perceptions and, albeit results were unstable, for employees with low intrinsic motivation. These employees may have realized that their position in the reorganization process was less than suboptimal, and, hence, may have put in extra effort as a last attempt to keep their job. Below, we will discuss how our findings give rise to the alternative hypothesis that employees may choose to exert extra effort to secure their job when they feel that the potential for job loss is greatest.

Theoretical implications and future directions
Integrating our results with the typology of threat foci (Shoss, 2017), one may argue that job insecurity will elicit job preservation efforts among employees who experience both job-at-risk threatthe person-independent perception that the job itself is threatenedand person-at-risk threatthe person-dependent perception that one's position as jobholder is threatened. Job-at-risk-threat refers to job insecurity where the job itself is insecure, occurring, for example, as a result of macro-economic downturns, layoffs and reorganizations. Person-at-risk threat refers to job insecurity that is linked to the particular jobholder, occurring, for example, when employees perform poorly or have a bad relationship with their supervisor. Our findings indicate that the combination of high job-at-risk threat (i.e. perceptions of job insecurity resulting from a reorganization) and high person-at-risk threat (i.e. a perceived lack of fairness within the organization) may elicit job preservation efforts in the form of performance. Findings from Lam et al. (2015) point in the same direction: they showed that positive effects of job insecurity on OCB were especially pronounced among employees with low psychological capital and low guanxior, in other words, those with both high job-at-risk threat (high job insecurity) and with high  person-at-risk threat (low psychological capital and low guanxi). This suggests that job insecurity may only prompt performance when the potential for job loss is greatest (also see Brockner, 1988;Probst, 2002).
The focus on threat foci can help to further uncover why job insecurity only prompts performance for some employees. It may be the case that in high job-at-risk situations such as reorganizations, employees experiencing high person-at-risk threat try to compensate for their negative expectations regarding job loss by enhancing their performance. Employees experiencing low person-at-risk threat, however, may feel optimistic about their chances to be retained, and, hence, feel uninclined to increase their performance. It should be noted that this assumption, similar to our initial assumptions, implies that putting in extra effort at work during a reorganization is mostly an impression management tactic. Yet, such enhanced performance may just as well signal an increased workload or an attempt to obtain a positive reference and increase one's chances on getting hired elsewhere. We believe that empirically examining the performance motives that can be triggered by job insecurity is a promising route for future research to further unpack its positive and negative outcomes.
The notion of threat foci may also help to create consensus regarding the deviant "positive" findings and dominant "negative" findings in the job insecurity-performance relationship. That is, the experience of job-at-risk threat may simply be more prevalent than the experience of person-at-risk threat. Experiencing both high job-at-risk threat and high person-at-risk threat might be rare, which can explain why relatively few studies show that job insecurity prompts performance. This idea is further supported by the fact that we mainly found a positive relationship between insecurity and performance at the endpoints of our scalesi.e., only in relatively extreme situations. Furthermore, assuming that only employees who experience both job-at-risk threat and person-at-risk threat will resort to active coping in the form of enhanced performance, it can be argued that the majority of employees will resort to the more common passive coping behaviors because they only experience job-at-risk threat. Such passive coping behaviors involve strain reactions and withdrawal behaviors, thereby undermining performance (Piccoli et al., in press). Whether different threat foci, or a combination thereof, indeed result in different coping responses is something that deserves further empirical attention. The assumption in this study that performance can be regarded as an active, problem-focused coping strategy remains tentative without explicit measures of coping, and without empirical examinations of other coping responses. We therefore hope that our study inspires other researchers to uncover the joint processes of coping responses and threat foci when examining job insecurity.
On a related note, it is very plausible that the effects of job insecurity on performance depend on time. That is, there is some evidence that coping is a process that changes over time (Kinicki and Latack, 1990). Specifically, in the initial phase of job insecurity, employees may rely on coping in the form of enhanced performance. For example, when organizational downsizing has been announced, employees may engage in OCB to demonstrate their worth to the organization. Likewise, when tenure depends on meeting performance criteria, employees may work harder in order to meet those criteria. However, as the period of job insecurity lengthens, one's resources to engage in OCB and higher work efforts may begin to deteriorate, resulting in lower rather than higher performance. This notion is supported by the fact that we conducted our study in the beginning of a reorganization, implying that we surveyed employees in an initial phase of high job insecurityand, hence, found that insecurity could prompt performance for some of them. We therefore strongly recommend that researchers adopt a temporal focus when examining job insecurity and its outcomes, and further examine the sustainability of different coping responses and how they may change over time.
In future examinations of the job insecurity-performance relationship, it may also be worthwhile to examine different types of performance separately. In this study, we have 299 Job preservation efforts only touched upon possible differences in performance outcomes in our additional analyses, which did not show any meaningful differences. Yet, researchers have suggested that the relationship between job insecurity and performance might vary depending on the measurement of performance (Probst et al., 2007;Selenko et al., 2013). From a job preservation perspective, one may argue that job insecurity only increases the types of performance that are part of the official job description, because employees may believe that such behavior will be rewarded by decision makers and may reduce threats (Shoss, 2018). Contextual performance, in contrast, may decrease in the face of job insecurity (i.e. OCB and voice), because employees may believe that these behaviors are not helpful or even hinder their chance of job continuance. Tentative support for this notion comes from Probst (2002), who showed that the threat of layoffs increased productivity but decreased output quality and safety compliance. Thus, examining different types of performance outcomes separately in future research may contribute to creating more consensus regarding the job insecurity-performance relationship.

Practical implications
In light of prevailing job insecurity and organizational changes such as downsizing and restructuring, the results of this study can support organizations in understanding and monitoring employees' coping responses to job insecurity. That is, it is important to realize that in times of high job insecurity, enhanced task and contextual performance could be impression management by employees who were not as (intrinsically) motivated before (Bolino, 1999;Huang et al., 2013), or attempts from unfairly treated employees to take matters into their own hands. Although such compensatory behavior may seem positive at first sight, we have speculated in the above that it may have a long-term negative impact. Additionally, it is important to realize that enhanced performance does not necessarily indicate higher quality performance. For example, while the threat of layoffs can indeed prompt performance, it also impedes creative performance (Probst et al., 2007) and increases safety violations (Probst, 2002). It is therefore of great importance that managers and organizations are aware of these job preservation efforts when making decisions about which employees should be retained. At the same time, organizations should be careful with employees whose performance does not appear to increase in times of job insecurity. It is likely that these employees are motivated, yet suffer from the maximum performance situation that high job insecurity represents (Klehe and Anderson, 2007). In fact, they may resort to more passive coping strategies that can undermine their performance and well-being. Such employees may thus require support to help them cope. One way to achieve this is via maintenance of a strong norm of fairness: organizational justice not only reduces the negative effects of job insecurity on attitudinal outcomes (Sverke et al., 2002;Wang et al., 2015), but can also prevent potential stressful compensatory responses in the form of job preservation efforts (the current study).

Limitations
The findings and implications discussed here need to be interpreted in light of several limitations. First, our performance measurements may have affected the results. For example, while the use of supervisor-issued performance ratings minimized the threat of common method bias (Podsakoff et al., 2003), these supervisor-issued performance ratings may have simply represented impression management tactics from employees, rather than actual objective performance. Additionally, our design remains correlational and involved self-report measures assessed at one time point, which warrants precaution about any time-lagged or causal inferences from the data. For example, we cannot conclude that employees' performance increased in response to job insecurity, only that their performance 300 CDI 25,3 was higher when they felt more insecure about the continuance of their job. Conceptually, one could argue that performance may reduce job insecurity rather than the other way around. Indeed, Huang et al. (2013) showed that employees' impression management tactics in the form of performance could decrease their job insecurity over time. In our study, however, we did not find such a negative relationship between job insecurity and performance. Methodologically, it may also be possible that supervisors rated employee's past performance, despite our instruction to rate current performance. Generally, it seems better to specify the period of assessment (e.g. the past few days and weeks), but in our case, we aimed to specifically assess employee performance in the context of a reorganization.
Second, we proposed that job insecurity may motivate employees to increase their performance when such performance is instrumental toward securing one's job. However, we did not explicitly measure participants' instrumental beliefs, nor whether management had given clear signs regarding the instrumentality of their performance. Yet, we believe that most employees realized that dismissal decisions were to be made and that management could use performance as a basic criterion for making these decisions. After all, employees generally view individual output as the best safeguard against involuntary job loss (Van Vuuren et al., 1991), and the employees in our sample were very much aware that the work package of their department was to be greatly reduced and that this would imply a reduction in staff numbers. Likewise, our measure of distributive justice served as a proxy for the likelihood of fair treatment in dismissal decisions (Lind, 2001). That is, participants' distributive justice perceptions were based on the distribution of rewards, not on the distribution of job security. A more direct measure of justice perceptions during a reorganization, and of the importance of task and contextual performance for dismissal decisions, would have benefitted the validity of our results. In the current study, however, the results and associated implications were largely based on conceptual evidence and should therefore also be interpreted as such.
Third, our findings are based on a rather specific and small sample, which gives rise to several alternative explanations for our deviant positive findings and the lack of dominant negative findings. For example, it may be that our sample experienced above average levels of job insecurity. That is, in studies where a null or positive relationship between job insecurity and job performance was found, respondents reported moderate to high levels of job insecurity at around the midpoint of the scale, whereas in studies where a negative relationship was found, respondents reported levels below the midpoint (cf. Selenko et al., 2013). Indeed, within our sample, the average job insecurity was somewhat above the midpoint (M ¼ 3.32). Alternatively, the relatively low level of education of participants may have influenced our results. That is, the work of lower educated employees is often more visible and measurable, making performance a preferred method to safeguard their job (Fischmann et al., 2018). Additionally, lower educated workers are often more dependent upon their current job and have a more vulnerable position in the labor market (De Witte et al., 2015), oftentimes lacking employability that higher educated workers do have. As such, they cannot afford to reduce their work efforts and are especially likely to start engaging in impression management strategies, which could result in better supervisorissued performance ratings. Finally, our sample may have been experiencing high job insecurity for a relatively short time, which resulted in directly visible active coping responses, while the negative consequences of job insecuritysuch as decreased job satisfaction and an inescapable drop in performancehad yet to appear. Again, this highlights the importance of adopting a temporal focus within job insecurity research.
Fourth, it is important to note that the interaction effect of intrinsic motivation disappeared after adding control variables to the analyses, and that it could not be replicated when using self-rated performance instead of supervisor-rated performance. The lacking robustness of the moderating effect of intrinsic motivation could be a result of 301 Job preservation efforts insufficient power or because the control variables were relatively stronger predictors of performance. However, it is more plausible that our measure of intrinsic motivation may not have fully aligned with our conceptualization. That is, we conceptualized intrinsic motivation as the extent to which an employee is driven by internal rewards to perform well in his/her job (Warr et al., 1979), while intrinsic motivation can also be conceptualized as doing something because it is inherently interesting or enjoyable (Ryan and Deci, 2000). The latter conceptualizationand operationalizationmay have fitted better with our theoretical reasoning, and may have resulted in more consistent findings regarding the moderating role of intrinsic motivation. Perhaps more importantly, much of our reasoning relied on the assumption that the performance of intrinsically motivated employees would not be driven by external factors such as job insecurity. Whether that is a valid assumption is impossible to establish without examining extrinsic motivation as well. That is, by including extrinsic motivation, we could have examined if job insecurity also prompted the performance of extrinsically motivated employees. Such empirical evidence could have strengthened the idea that maximum performance conditionssuch as a reorganizationmay especially trigger performance among employees who tend to be motivated by external rather than by internal factors (also see Klehe and Anderson, 2007). Additionally, it is possible that the performance of especially those intrinsically motivated employees may suffer from maximum performance conditions, in which job insecurity may "crowd out" the positive effects of intrinsic motivation on performance (cf. Cerasoli et al., 2014). We believe that examining the joint effects of extrinsic and intrinsic motivation in job insecure settings deserves further attention.

Conclusion
How do people react when the continuity and stability of their job is at risk? Our study suggests that in the midst of a reorganization, employees may devote extra effort toward performance behaviors that will be noticed and valued when they feel that the potential for job loss is greatest. Based on our results, we have speculated that the motivating effect of job insecurity on performance may only emerge when both the job and the position as jobholder are threatened. However, the sustainability of enhanced performance in response to job insecurity is questionable for both organizations and employees.

Note
1. Employees were not asked to rate their task performance as to not compromise their trust in the confidentiality of their responses.
Job preservation efforts