Wellbeing typically refers to an employee's work satisfaction and state of mental and physical health (Danna and Griffin, 1999). Research on the job demand-control (JDC) model (Karasek, 1979) shows that employee wellbeing is affected by the joint effects of two work elements: job demands and job control (Karasek, 1979; Theorell and Karasek, 1996). Job demands refer to stressors in the job and work environment, such as a high workload or workplace conflicts, that require employees to exert effort and engage in action to respond to these demands (Karasek, 1979). When employees experience high job demands, they are more likely to experience stress and strain, which negatively impacts their wellbeing (Gonzalez-Mulé et al., 2021). Job autonomy or job control (e.g., Karasek, 1979) refers to the extent to which employees have freedom, independence, and personal discretion to carry out their work (Breaugh, 1985).

The central tenet of the JDC model is that the combination of job demands and job control predicts wellbeing outcomes. One important joint effect is the buffer effect (Gonzalez-Mulé et al., 2021); job control buffers against the negative effects of job demands on wellbeing. Poor wellbeing is more likely when employees experience high demands in combination with low job control. That is, having low control decreases an employee's ability to cope with job demands (Karasek, 1979). However, when employees have more control over the way they respond to stressors they are, overall, better equipped to resolve job demands.

We note here that the literature on organizational justice proposes a similar process with respect to how people respond to experiences of distributive and procedural justice (Brockner, 2011). Specifically, similar to how appropriate levels of job control buffer against high job demands or other stressors (Karasek, 1979), so can procedural justice buffer against the stressor of distributive injustice. As explained by equity theory (Adams, 1965), unfair outcomes constitute a direct personal loss to the person who is disadvantaged and creates tension and dissatisfaction. This claim is supported by research showing that unfair outcomes invoke discrete negative emotions, such as anger (Weiss et al., 1999), and also, more broadly, negative affect (Colquitt et al., 2013). Distributive injustice can therefore be viewed as a stressor that may lead to poor wellbeing, especially when control or autonomy is low (for a meta-analysis, see Robbins et al., 2012).

One way in which procedural justice may buffer against the stressor of distributive injustice is by supporting a feeling of control (Thibaut and Walker, 1975; Judge and Colquitt, 2004). In their instrumental model of procedural justice, Thibaut and Walker (1975) argued that process control is key to procedural justice. That is, people desire to have control over the processes by which decisions are made, and processes are seen as more fair when some control can be exercised. This instrumental model of procedural justice, therefore, suggests that when processes are fair, employees are less affected by the stressor of unfair outcomes as they perceive some control over future outcomes.

A few studies show that distributive justice and procedural justice interact to affect or predict different indicators of employee wellbeing in the direction that is hypothesized by the JDC model (for a review, see Brockner, 2011). For instance, Tepper (2001) found that employees experiencing unfair outcomes reported the highest levels of depression, anxiety, and emotional exhaustion when procedures were also unfair. In other words, experiencing a job stressor (such as low levels of distributive justice) combined with perceiving little control (such as is the case when people experience low levels of procedural justice) is likely to lead to the lowest levels of employee wellbeing. In contrast, perceiving some control (such as higher levels of procedural justice) may well buffer against the stressor of unfair outcomes. Fields et al. (2000) indeed showed that the combination of low procedural and distributive justice predicted the lowest levels of job satisfaction, while high procedural justice weakened the negative effect of low distributive justice on job satisfaction. In the present paper we build on these insights to propose that the Distributive Justice × Procedural Justice interaction on wellbeing-related outcomes is more pronounced among employees with low (vs. high) job autonomy.

Following the logic of the buffering effect in the JDC model (Karasek, 1979; Gonzalez-Mulé et al., 2021), we argue that job autonomy mitigates the effects of stressors (i.e., demands), such as distributive injustice, and compensates for low process control, such as resulting from procedural injustice. Previous research showing that high (vs. low) job autonomy protects against the effect of low (vs. high) distributive justice (Haar and Spell, 2009), and low (vs. high) procedural justice (Rousseau et al., 2009), supports this expectation.

We extend this prior research that has looked at the role of job autonomy in effects of distributive or procedural justice in isolation by arguing that low job autonomy increases the importance of procedural justice when distributive justice is low. Employees with low job autonomy have less control over the way they respond to stressors encountered at work (Bakker et al., 2005), such as unfair outcomes. As procedural justice makes outcomes feel more controllable and predictable (Thibaut and Walker, 1975), employees with low job autonomy (i.e., with low job control) may rely more on procedural justice to cope with low distributive justice. Work by Van Prooijen (2009) supports this expectation by showing that people with low levels of autonomy rely more on procedural justice judgments than those with high levels of autonomy. In other words, when employees with low job autonomy are confronted with unfair outcomes, high procedural justice leads them to perceive some control that helps them cope with the stressor of such outcomes. Employees with high job autonomy need to rely less on procedural justice when they are confronted with unfair outcomes because job autonomy provides them more control over the way that they respond to and cope with these demands (Karasek, 1979; Bakker et al., 2005). Thus, we expect that the Distributive Justice × Procedural Justice interaction emerges for employees with low levels of job autonomy and not for employees with high levels of job autonomy.

As indirect support that job autonomy may moderate the Distributive Justice × Procedural Justice interaction, previous work shows that this interaction effect is especially pronounced among organization members in lower power and status positions (Chen et al., 2003; Blader and Chen, 2012; Bianchi et al., 2015; Van Dijke et al., 2019), who may also experience lower autonomy. Yet, job autonomy and power are distinct constructs. Job autonomy refers to the extent to which employees have control over their work and therefore the ability to respond to job demands (Karasek, 1979), while power refers to controlling the outcomes of others (Lammers et al., 2016). Furthermore, the effect of power is often explained in terms of trust: those in lower power positions rely on justice to judge whether those in higher power positions can be trusted to not abuse their power (e.g., Bianchi et al., 2015; Van Dijke et al., 2019). In contrast, the current study focuses on employees' control over how and when to respond to the demands of injustice. Taken together, our reasoning leads to the following hypothesis:

Study 1 supports our hypothesis that the Distributive Justice × Procedural justice interaction on the two wellbeing outcomes, emotional exhaustion and job satisfaction, is pronounced most among employees who have low (vs. high) job autonomy.

To test the robustness of our findings, we conducted a second study, in which we introduced some changes. First, some results in Study 1 were affected by one influential outlier. As larger sample sizes reduce the influence of individual points (i.e., outliers; Belsley et al., 1980), we increased the sample size in Study 2 to decrease the influence of individual cases. Second, we wanted to test if the results obtained on the single-item job satisfaction measure in Study 1 can be replicated with a multi-item measure of the same construct, which we included in Study 2. Finally, in Study 2 we also used more extensive measures of job autonomy and emotional exhaustion.

The JDC model predicts that job control buffers against the effects of job demands (i.e., stressors) on employee wellbeing (Karasek, 1979). Similarly, previous work on organizational justice has argued that procedural justice buffers the negative effects of distributive injustice on employee wellbeing (Tepper, 2001; Vermunt and Steensma, 2005; Brockner, 2011). One way in which procedural justice acts as a buffer is that fair processes give employees (a feeling of) control (Thibaut and Walker, 1975; Judge and Colquitt, 2004). Following this work, we expected that the Distributive Justice × Procedural Justice interaction is more likely to emerge for employees that have lower job autonomy (i.e., lower control) than for those with higher job autonomy. We expected this as employees in low autonomous jobs should be more likely to rely on other resources that may provide them with some (feeling of) control, such as procedural justice (Van Prooijen, 2009). In line with this prediction, we show that the buffering effect of procedural justice depends on the level of job autonomy. We thus identified job autonomy as a novel and theoretically relevant boundary condition to the Distributive Justice × Procedural Justice interaction effect on wellbeing outcomes.

The present findings thus may contribute to explaining previous inconsistent findings regarding the Distributive Justice × Procedural Justice interaction effect for wellbeing outcomes. For instance, Tepper (2001) found that the interactive relationship predicted emotional exhaustion, depression, and anxiety. However, other studies did not find this interaction effect on related measures of wellbeing (McFarlin and Sweeney, 1992; Fischer et al., 2014). Our study suggests that one reason for the different findings between studies could be variations in the level of job autonomy between the studies involved.

For instance, Fischer et al. (2014), who did not find a significant interaction effect between distributive and procedural justice, excluded shop floor workers and employees doing manual labor. As these jobs often involve low levels of autonomy (Vidal, 2013), the resulting sample might primarily consist of employees with higher levels of autonomy. Tepper (2001) who, in contrast, did find a significant interaction effect in two studies, included participants from a wide range of jobs that includes low levels of autonomy, such as construction workers and clerical workers (Vidal, 2013). In addition, average levels of job autonomy have increased in the past few decades (Wegman et al., 2018). Differences between older papers with significant findings (Fields et al., 2000; Tepper, 2001) and more recent papers with non-significant findings (Fischer et al., 2014; cf. McFarlin and Sweeney, 1992) might therefore be due to different levels of job autonomy in the study samples.

Our findings correspond well with insights on the buffer effect proposed within the JDC literature (Häusser et al., 2010). A recent meta-analysis investigating the validity of the buffer effect found that this effect was more strongly related to wellbeing outcomes when demands refer to hindrances instead of challenges (Gonzalez-Mulé et al., 2021). Challenge demands are stressors that are energizing and provide an opportunity for achievement and learning (Cavanaugh et al., 2000; LePine et al., 2005); a few examples are time pressure and a high workload. Hindrance demands are stressors that offer undesirable constraints and thwart personal growth and goal attainment (Cavanaugh et al., 2000; Gonzalez-Mulé et al., 2021). Receiving unfair outcomes is an example of hindrance demands; it is unlikely to be beneficial and provide an opportunity to learn. As there is less research on the interaction between hindrance demands and control (Gonzalez-Mulé et al., 2021), our current study adds to this body of work and supports the view that hindrance demands and job control interact to predict employee wellbeing.

Overall, focusing on the interaction between distributive justice, procedural justice, and job autonomy presents practical implications that could be tested and implemented depending on the needs and resources of organizations. Viewed in this manner, the current study provides several new insights on when employees benefit most from a focus on justice.

Firstly, fair procedures matter a lot to employees when their outcomes involve little autonomy and are perceived by them as negative or unfair, such as is often the case following layoffs and negative promotion decisions. Research has shown that procedural justice is related to positive employee outcomes both for survivors and victims of job layoffs (Brockner et al., 1994). Some examples of procedural justice in these contexts are giving an advanced notice (Brockner et al., 1994), giving voice to employees, and ensuring consistency in decision-making (Brockner et al., 1995).

Secondly, organizations may well be advised to focus on protecting distributive justice when both procedural justice and job autonomy are low. Personnel screening and selection contexts (Cropanzano et al., 2007) provide a good example of such situations. Standardized and structured tests (i.e., personality and cognitive ability tests) are perceived to be procedurally less fair and to allow less autonomy than unstructured interviews (Nolan and Highhouse, 2014). Nonetheless, standardized tests are a better indicator of job performance, while the predictive validity of unstructured interviews is low and can even hurt personnel selection decisions (Kausel et al., 2016). To minimize the detrimental effects of lowered procedural justice and autonomy, organizations could increase distributive justice perceptions by emphasizing and communicating equity in selection decisions (Celani et al., 2008).

Thirdly, organizations could try to increase job autonomy when neither distributive nor procedural justice can be ensured. Organizations are not always successful in securing high justice. This may be due to, for instance, rapid and radical changes within organizations (Kickul et al., 2002) or to employees' characteristics that influence their justice perceptions (Lang et al., 2011). The current study suggests a strategy for protecting employees' wellbeing when justice is low: increasing job autonomy. For instance, job design research has demonstrated how organizations granting more autonomy to employees positively influence employees' job satisfaction and wellbeing (Humphrey et al., 2007). Small changes in job designs that increase job autonomy can have a substantial impact; for example, giving call-center workers more autonomy rather than rules on how to display facial expressions decreased emotional exhaustion (Goldberg and Grandey, 2007).

Like all research, ours has limitations. One of these is that we cannot draw causal conclusions from our two studies because they relied on cross-sectional designs. However, it should be noted that several experimental studies (e.g., Van den Bos et al., 1997) have established the interactive effect of procedural and distributive justice on various outcomes. Thus, the literature makes us quite confident about the causal direction of these effects. This implies that our reliance on cross-sectional designs is less of a problem. Future research could test causal effects by manipulating the degree of autonomy, the fairness of outcomes, and the fairness of procedures, or use longitudinal designs to exclude some alternative causal explanations.

Furthermore, the single source nature of the studies could result in common method variance that undermines our theoretical explanation (Podsakoff et al., 2012). However, studies show that common method variance does not inflate interaction effects (Evans, 1985; Siemsen et al., 2010), which suggests that our results are not affected by method bias.

The current study shows that the effect of organizational justice on employee wellbeing may depend on a job characteristic—job autonomy. In so doing we answer calls for more research on the interplay between justice and organizational structure and job characteristics (Schminke et al., 2015). Organizational characteristics play a role in organizational justice perceptions and effects of organizational justice (for a review, see Schminke et al., 2015). For instance, jobs that are characterized by high degrees of formalization (i.e., work processes being uniformly structured) tend to be perceived as higher in distributive and procedural justice perceptions (Schminke et al., 2002) but lower in autonomy (Langfred and Rockmann, 2016). Future research should investigate effects of organizational characteristics on wellbeing-related outcomes taking into account job autonomy and justice.

Previous work has identified distributive justice and procedural justice as important factors that improve employee wellbeing. The current study set out to investigate when these two types of justice interact to predict wellbeing, identifying job autonomy as a moderator to the interactive relationship. The results of this study indicate that when employees have low job autonomy, fair procedures help them cope with the stressor of unfair outcomes, thus protecting their wellbeing. When employees have high job autonomy, fair procedures do not buffer unfair outcomes. Taken together, the findings in the current study provide new insights on when employees' wellbeing benefits most from a focus on justice, thus showing relevant theoretical and practical implications.