Felt or Thought: Distinct Mechanisms Underlying Exploitative Leadership and Abusive Supervision

Abstract

The last two decades have seen a mounting fascination with unethical and destructive forms of leadership. Yet, do we know what all encapsulates this “dark” side of leadership? Despite initial evidence that exploitation is a notable addition to the unethical leadership scene, our understanding of its distinctiveness as well as of how and why it exerts its negative effects is limited. We speak to this gap by testing the distinct mechanisms through which exploitative leadership—relative to the more popular counterpart, abusive supervision—affects followers. Borrowing from the aggression literature, we describe exploitative leadership and abusive supervision as varying forms of aggression that undermine followers’ satisfaction with the leader via altered experiences of their social exchange relationship. Our theoretical model proposes that abusive supervision, as an inherently interpersonal provocation, primarily implicates followers’ emotional experiences within the social exchange process. By contrast, given its inherent focus on self-interest, exploitative leadership is assumed to affect followers primarily through the cognitive understanding of the social exchange. Results from multiple studies using different samples, measures, and research designs provide general support for our predictions. In sum, the evidence emerging from our data shows that exploitative leadership is not a symptom of construct proliferation but rather, adds cumulative knowledge to the field of unethical and destructive leadership.

Introduction

Leaders within organizations have a substantial impact on their followers through the way they allocate resources, assign tasks, and how they shape and manage their interpersonal interactions. While leadership can be a source of meaning and recognition, there is abundant evidence for the negative and costly effects of unethical and destructive leadership on key workplace outcomes (Hassan et al., 2022; Mackey et al., 2020; Schyns & Schilling, 2013). The bulk of theorizing and empirical research in this field has expanded on highly intense and inherently hostile forms of destructive leadership (Almeida et al., 2021). One of which most notably includes abusive supervision (Tepper et al., 2017), with active interpersonal mistreatment, such as ridiculing subordinates or putting them down verbally, at its core (Mitchell & Ambrose, 2007). In contrast, far less attention has been given to unethical leader behaviors that are more subtle, yet nonetheless detrimental in their effects on followers (Almeida et al., 2021). One behavior that particularly stands out among less intense and more indirect behaviors is exploitation—that is, leaders taking advantage of followers for personal gain. Recent polls indicate that nearly 80% of employees at some point feel exploited at work, causing well-being and job satisfaction to plummet (Paychex, 2019). It is, therefore, surprising that unethical and destructive leadership researchers have only recently begun to more thoroughly investigate the nature and consequences of exploitative leadership in organizations (Schmid et al., 2019).

Initial findings support exploitation as a notable addition to the unethical leadership scene. Not only does it appear to meaningfully supplement the existing range of unethical behaviors, but even more, the construct may predict unique variation in important employee outcomes, such as lower well-being and commitment and higher workplace deviance and intentions to leave the organization (Lyu et al., 2022; Schmid et al., 2019). However, with increasing differentiation of (un)ethical leader behaviors come questions concerning construct novelty and redundancy (Banks et al., 2018; Fischer & Sitkin, 2023; Lemoine et al., 2019; Mackey et al., 2020). When new leadership facets are introduced or further refined, it is important to likewise gain a thorough understanding of their unique contribution beyond conceptually similar. Otherwise, we risk undermining the development and maintenance of parsimonious theories and cumulative knowledge in the field of leadership ethics (Babalola et al., 2022; Kaiser & Craig, 2014). Not merely theoretical in nature, this concern poses important practical implications, as we must know which distinctions among unethical leadership constructs and their mechanisms are (or are not) useful in assessing and preventing destructive leadership and its detrimental consequences in organizations.

That being said, in order to add value to a more differentiated yet meaningful understanding of the unethical leadership domain, three major shortcomings in the emerging evidence on exploitative leadership must be addressed. First, the position of exploitative leadership within the nomological network of destructive leadership—namely, distinguishing it from high-intensity (i.e., abusive) forms of harmful leader behavior—remains insufficiently understood. The few existing empirical studies indicate high correlations and have, so far, not gone beyond showing that exploitative leadership explains unique variance in job outcomes and that measures of exploitative versus interpersonally abusive leader behaviors can be discriminated via factor analysis (Schmid et al., 2018, 2019). Second, we have a rather limited understanding of how and why exploitative leadership exerts its negative effects. Even more, relative to the overt interpersonal mistreatment covered by abusive supervision, we do not know whether the underlying mechanisms are unique to exploitative leader behaviors. Third, and building on this ambiguity, the problem is exacerbated by what research on construct redundancy has referred to as common theory; that is, the same theoretical mechanism is used to explain the way in which supposedly different forms of leadership are related with particular outcomes (Lemoine et al., 2019). Specifically, empirical research on both exploitative leadership and abusive supervision commonly employs social exchange theory as the primary framework to explain negative follower effects (Choi et al., 2019; Schmid et al., 2019; Wu et al., 2021; Xu et al., 2012). Yet, such generic theoretical applications leave us with limited insight into the differences between leadership forms and even blur our understanding of the underlying processes of these phenomena. Accordingly, scholars are increasingly urging the field to “apply general theories in more specific ways to better elucidate the theoretical differences among forms of leadership (moral or otherwise)” (Lemoine et al., 2019, p. 155).

In order to overcome these shortcomings, the main purpose of the present research is to investigate the unique underlying mechanisms through which exploitative leadership affects followers relative to interpersonally abusive leader behaviors. We theorize that breaking down exploitative and abusive behaviors via different forms of aggressiveness will uncover distinct theoretical mechanisms. To this end, we build upon the call for sharper theoretical specification (Lemoine et al., 2019) and map on a more fine-grained use of social exchange. Specifically, our theoretical model posits that abusive supervision is an emotional, high-arousal type of aggression (Neuman & Baron, 2005; Wrangham, 2018), and thus, primarily “felt” by followers, that is, implicates their emotional experiences within the social exchange process. By contrast, with its inherent focus on self-interest and deliberate efforts to accomplish strategic objectives, exploitative leadership represents a more calculated, low-arousal type of aggression. It is, therefore, assumed to be “thought” by followers, meaning that it primarily affects their cognitive understanding of the social exchange, reflected in perceptions of impaired leader–member social exchange (LMSX, Bernerth et al., 2007).

We test these differential mechanisms as a dual pathway model (i.e., emotional versus cognitive responses) in a series of three studies. While they involve different samples, measures, and research designs, all three studies link exploitative leadership and abusive supervision to followers’ satisfaction with the leader as the main outcome. Followers’ leader satisfaction fits particularly well into our theoretical framework for two reasons. First, it represents a summary evaluation of the leader with affect and cognition as distinct influences, rather than dimensions of it (Weiss, 2002; Weiss & Cropanzano, 1996), which fits well with distinguishing social exchange in terms of cognition versus affect. Second, multifoci research (Lavelle et al., 2007; Rupp et al., 2014) shows that employees’ perceptions that a particular party has violated ethical norms (e.g., a leader engaging in unethical leadership) are most proximately related to reactions specifically directed at that party (i.e., dissatisfaction with the leader). Finally, from a practical perspective, satisfaction with the leader is a critical intermediary between leadership perceptions and employee well-being and retention and has important implications for organizational functioning and performance (Faragher et al., 2013; Rubenstein et al., 2018; Scarpello & Vandenberg, 1987).

Overall, our intended contributions to the business ethics literature are threefold. First, we seek to continue the momentum of exploitative leadership research and to expand our understanding of what makes exploitative leadership uniquely important—notably, relative to the most studied high-intensity form of destructive leadership, abusive supervision. We thereby respond to the ongoing call in the destructive leadership literature to more thoroughly “position the elements within the destructive leadership domain relative to each other,” seeking to “reduce redundancy and highlight unaddressed areas” (Kaiser & Craig, 2014, pp. 276, see also Babalola et al., 2022). Second, linking exploitative leadership and abusive supervision to different forms of aggression and disentangling their differential effects on distinct social exchange processes allows a better understanding of how and why these leadership behaviors, especially exploitative leadership, are uniquely destructive. We thus extend earlier research in the field of exploitative leadership, which juxtaposed it with other forms of destructive leadership, while disregarding the examination of differential mechanisms (Schmid et al., 2018, 2019). As such, we help attenuate common theory by offering theoretical perspectives and mechanisms that are more specifically attuned to each form of destructive leadership (Lemoine et al., 2019). Finally, from a practical perspective, putting the distinct characteristics and mechanisms of different types of destructive leadership on an organization’s radar will engender a more nuanced understanding of their leadership culture and in turn, inform the development and implementation of efficient remedies and interventions tailored to specific types of destructive leader behaviors.

Theoretical Background and Hypothesis Development

Abusive Supervision and Exploitative Leadership

Abusive supervision represents, by far, the most widely used lens through which researchers look at destructive leader behaviors (Almeida et al., 2021; Mackey et al., 2020). In his seminal work, Tepper (2000, p. 178) defined abusive supervision as “subordinates' perceptions of the extent to which supervisors engage in the sustained display of hostile verbal and nonverbal behaviors, excluding physical contact.” On this basis, it is commonly framed and measured around overt interpersonal affronts, including harsh criticism, ridicule, or the silent treatment (Barnes et al., 2015; Johnson et al., 2012; Mitchell & Ambrose, 2007; Priesemuth et al., 2014).

Exploitative leadership was introduced to address a more subtle facet of destructive leadership that was often implied in the destructive leadership literature yet had previously not been explicitly conceptualized and distinctly measured (Schmid et al., 2019). While leader self-interest per se is not necessarily directed against others (Williams, 2014), exploitation refers to behaviors that serve the interests of the leader at the expense of followers (Kim et al., 2020). On this basis, Schmid et al. (2019) developed and validated a measure of exploitative leadership, or “leadership with the primary intention to further the leader’s self-interest by exploiting others, reflected in five dimensions: genuine egoistic behaviors, taking credit, exerting pressure, undermining development, and manipulating” (p. 4126). Simply put, as their own goals have priority, exploitative leaders expect that the work of followers can be used for their personal benefit. They claim credit for follower achievements, increase the workload of followers, and manipulate employees to ensure their own interests are met. They also undermine followers’ efforts to develop professionally, for instance, by constantly giving them tedious tasks, as they prioritize their own goals over the goals of their followers.

Previous work has argued that abusive supervision and exploitative leadership are conceptually distinct. Schmid et al. (2018) suggested that destructive leadership forms differ along a continuum of hostility, that is, the overt expression of intense animosity and/or antagonism. While abusive supervision, by definition, includes high levels of hostility, exploitative leadership is framed as more subtle and low in hostility. In a similar vein, Almeida et al. (2021) argued that an important distinction between types of destructive leadership is the level of intensity and whether the destructive behavior is task- versus people-focused. On this basis, they describe abusive supervision as a high-intensity destructive leadership form that is directly targeted at people. By contrast, exploitative leadership is less direct and intense, as exploitative leaders tend to use soft tactics to influence and carefully navigate social dynamics to achieve their desired outcomes (Schmid et al., 2019). Moreover, the focus of exploitative leadership is more on tasks and how they can be organized to prioritize leaders’ goals.

Nonetheless, while abusive supervision and exploitative leadership can be distinguished conceptually, their practical manifestations appear to be intertwined. As destructive leader behaviors vary both between and within leaders (Aasland et al., 2010), both forms may, to varying degrees, appear together yet differ in terms of intensity and frequency. Precisely because of this proximity, we need a thorough understanding of the way in which interpersonally abusive and exploitative behaviors may be uniquely destructive.

Differential Mechanisms of Abusive Supervision and Exploitative Leadership

To disentangle the effects of abusive supervision versus exploitative leadership, we build upon the well-established notion that destructive leadership decreases followers’ satisfaction with their leader (Schyns & Schilling, 2013). However, we theorize that abusive supervision and exploitative leadership obtain this outcome via different mechanisms. Specifically, we posit that abusive supervision and exploitative leadership deplete satisfaction with the leader by differentially affecting how followers perceive and experience distinct qualities of the social exchange relationship with their leader. This is in marked contrast to previous approaches in this field, which have treated social exchange as an overall relational quality, typically reflected in leader-member exchange (LMX) measures (Graen & Uhl-Bien, 1995; Liden et al., 1997). In this tradition, poor LMX relationships are described in rather general terms, referring to low levels of mutual trust, respect, and influence. Not surprisingly, abusive supervision and exploitative leadership have both been empirically associated with such poor relationship quality (Choi et al., 2019; Wu et al., 2021; Xu et al., 2012). Underscoring criticisms that this approach too broadly captures beliefs and feelings about relationships (Bernerth et al., 2007; Colquitt et al., 2014; Sheer, 2014), extant studies finding abusive supervision and exploitative leadership equally related to such broad exchange qualities do little to enhance our understanding of how these phenomena may actually work differentially.

Therefore, we herein adopt a more fine-grained approach to social exchange theory by explicitly distinguishing its affective (i.e., mutual affection) from its cognitive (i.e., mutual obligation) elements (Colquitt et al., 2014). This distinction implies that followers’ perceptions and experiences regarding the social exchange relationship with their leader can be differentiated according to whether they reflect affective (i.e., emotional responses towards the relationship with the leader) or cognitive (i.e., the rational assessment of reciprocity) responses. In what follows, we provide a theoretical rationale for differentially linking abusive supervision and exploitative leadership to dissatisfaction with leadership via distinct affective versus cognitive exchange relationship pathways. Figure 1 summarizes our theoretical model. While abusive supervision is theorized to work primarily through negative affective responses, exploitative leadership is thought to operate mainly through followers’ cognitive assessment of the social exchange relationship with the leader.

Fig. 1

Theoretical model

The Affective Route: Followers’ Negative Emotional Experience in Social Exchange

It is well-established in leadership research that emotions play a pivotal role in followers’ assessment of social exchange relationships with their leaders (Tse et al., 2015). Based on affect-based theories, most notably affective events theory (Weiss & Cropanzano, 1996), research has emphasized that leaders display emotions and simultaneously evoke emotions in their followers (Dasborough, 2006; Gooty et al., 2010). On this basis, both theoretical and empirical research has consistently indicated that abusive supervision represents an intense, interpersonal provocation which threatens a person’s sense of respect and dignity and elicits negative affect toward the perpetrator (Lian et al., 2014; Michel et al., 2016; Oh & Farh, 2017). This is in line with early research relating abusive supervision to perceptions of organizational (in)justice (Tepper, 2000), which is a well-established predictor of negative emotional reactions (Cropanzano et al., 2011).

Exploitative leadership can also be linked to followers’ negative emotional responses. People usually regard egoistic behaviors and exploitation as unfair (Mikula et al., 1990), and individuals are generally averse to unequitable treatment (Allen & Leary, 2010). Unfair treatment can also be seen as a negative affective event and accordingly, the experience of inequity has been linked to negative affect toward the perpetrator (Cropanzano et al., 2011). Exploitative leadership additionally involves behaviors that followers may perceive, at least to some degree, as interpersonal transgressions (e.g., manipulation and deception). Similarly, such impressions are likely to promote negative affect (Vohs et al., 2007). On this basis and considering the well-established evidence that affective reactions of employees are an important precursor of work-related attitudes, including satisfaction with the leader (Wagner & Ilies, 2008), we specify the following predictions:

Hypothesis 1

Abusive supervision is negatively related to followers’ satisfaction with the leader through the mediating effect of followers’ negative emotional responses towards the relationship with the leader.

Hypothesis 2

Exploitative leadership is negatively related to followers’ satisfaction with the leader through the mediating effect of followers’ negative emotional responses towards the relationship with the leader.

Although both forms of destructive leadership are expected to evoke negative affective responses among followers, we argue that abusive supervision will be more proximately associated with and particularly damaging to followers’ affective responses within the social exchange relationship. In contrast, negative affective responses are less impactful in cases of exploitative leadership. To support this line of reasoning, we draw from the psychological aggression literature and explicate abusive supervision and exploitative leadership via two forms of aggression—emotional (or “hot”) versus instrumental (or “cold”) aggression (Neuman & Baron, 2005; Wrangham, 2018). In this way, we are able to consider how each form of destructive leadership, given their unique concordance of aggression type, differentially implicates emotional reactions.

Abusive supervision can be viewed as emotional or “hot” aggression, describing aggression as a primarily affective response to a threat or frustrating event, reflected in high autonomic arousal (i.e., anger and loss of control). Accordingly, this type of aggression is also commonly referred to as reactive or hostile aggression (Wrangham, 2018). In fact, research has indicated that abusive supervision is often evoked by intense negative emotions among the leader, typically resulting from perceived identity threats (e.g., perceived provocative follower behavior or unfair treatment from the organization) and impaired self-regulation (e.g., cognitive or emotional overload) (Tepper et al., 2017; Zhang & Bednall, 2016). Furthermore, there is consistent evidence that leaders’ behaviors, including abusive supervision, can transfer the negative affect of leaders onto followers (Clarkson et al., 2020; van Knippenberg & van Kleef, 2016). As the overtly hostile behaviors that typify abusive supervision (e.g., yelling and putting subordinates down verbally) represent a direct and substantial affront to followers’ sense of self and social standing (Burton & Hoobler, 2006; Vogel & Mitchell, 2015), they are particularly salient to followers and can consequently fuel this emotional contagion effect. Empirical support for this notion comes from data reported by Hoobler and Hu (2013), suggesting that leaders’ perceptions of unfairness trigger negative affect, which, in turn, are induced and transferred to followers through abusive supervision. Against this backdrop, we contend that negative affect is the dominating mechanism through which the aroused behaviors reflected in abusive supervision affect followers. Notably, we argue that this is also the case when leaders use such affective displays strategically to force compliance or to intimidate or punish followers (Tepper et al., 2012), as they still represent an inherent ego threat that is typically followed by strong negative affect (Vogel & Mitchell, 2015).

In contrast to abusive supervision, exploitative leadership is a more purposeful and planned effort of leaders to increase personal gain (Schmid et al., 2019). As such, it can be viewed as instrumental or “cold” rather than emotional aggression. This type of aggression is motivated by perceived benefits, thus representing an instrumental means to achieve goals and secure goods from others (Neuman & Baron, 2005; Wrangham, 2018). Accordingly, researchers have called this form of aggression proactive, controlled, or premediated aggression (Chichinadze et al., 2011). Importantly, instrumental aggression is characterized by positive outcome expectancies and self-efficacy whilst involving low emotional arousal (Wrangham, 2018). On this basis, exploitative leadership is thought to result more strongly from cognitive, that is, deliberative and intendedly rational processes, as opposed to reduced emotion regulation and executive control. Accordingly, due to its lower emotional arousal, there is far less room for the transfer of negative affect (Hoobler & Hu, 2013). In fact, exploitative leaders have even been associated with the use of positive affect to influence followers, as they may push their agenda in “an overtly friendly way, for example, by being exceedingly pleasant to ensure their interests are met” (Schmid et al., 2019, p. 1404). That said, instead of directly precipitating negative affect, the experience of exploitative leadership is more likely to induce cognitive reactions, e.g., in the form of reflection, as followers consider the meaning of their leaders’ behavior. Taken together, we made the following prediction:

Hypothesis 3

Exploitative leadership exerts a weaker indirect effect on followers’ satisfaction with the leader through their negative emotional responses towards the relationship with the leader, as compared to abusive supervision.

The Cognitive Route: Followers’ Perceived Imbalance in Social Exchange

Thus far, we have argued that abusive supervision is a direct and emotionally charged form of aggression and, therefore, expected to prompt affective contagion effects. Exploitative leadership, in turn, is more subtle and consequently, its negative effects are likely driven by a cognitive interpretation process, that is, followers’ cognitive assessment of the social exchange relationship with the leader. To specify this cognitive response, we draw on the concept of leader-member social exchange (LMSX). Initially described by Bernerth et al. (2007) and further advocated by Colquitt et al. (2014), LMSX frames social exchange around the expectation that one’s own efforts will be returned by the leader, that is, reciprocity. In so doing, it explicitly captures the cognitive understanding of the social exchange (Jian et al., 2014).

Exploitative leadership, with its inherent focus on instrumental aggressiveness, is highly likely to undermine the perceived balance between inputs and outputs. That is, as an exploitative leader uses their power in exceedingly self-serving ways, they go beyond what can be seen as legitimate within the natural power differential in typical leader–follower relationships (Williams, 2014). When a leader is perceived to repeatedly prioritize their goals over the needs of others, followers will likely come to see the exchange relationship as one-sided. Even more, by taking credit for followers’ accomplishments, using pressure to get tasks done, and undermining personal development, such a leader does not provide the inputs that lead to expected outcomes—rather, they may even reap the benefits of the inputs provided by employees. No matter how friendly and pleasant an exploitative leader presents themselves, being subjected to exploitation is likely to stimulate strong beliefs among followers that the leader is unreliable and that they choose expediency over principles of reciprocity.

Abusive supervision can also be related to impaired LMSX. Drawing from organizational justice theory, Tepper (2000) argued that employees may conclude that an abusive leader spends more time berating them instead of providing mentoring and support. Moreover, they may have to spend more time and effort to overcome the obstacles that come with abusive supervision, such as conflicts, lack of information, or more complicated communication. Overall, this may lead employees to conclude that expectations of reciprocity are violated.

Having comparable exchanges of giving and taking reduces uncertainty and promotes trustworthiness of the exchange partner (Molm et al., 2007). Thus, low levels of LMSX likely undermine followers’ satisfaction with their leader. Empirical support for this notion stems from many studies linking proxies of positive reciprocity, such as equity or psychological contract fulfillment, to followers’ satisfaction with leadership (Cohen-Charash & Spector, 2001; Coyle-Shapiro et al., 2019). On this basis, we offer the following predictions:

Hypothesis 4

Exploitative leadership is negatively related to followers’ satisfaction with the leader through the mediating influence of followers’ perceived imbalance in LMSX.

Hypothesis 5

Abusive supervision is negatively related to followers’ satisfaction with the leader through the mediating influence of followers’ perceived imbalance in LMSX.

Both abusive supervision and exploitative leadership are thought to undermine LMSX beliefs. We posit, however, that in contrast to abusive supervision, this effect is substantially stronger and remarkedly detrimental in cases of exploitative leadership. In fact, leaders may show abusive behaviors like yelling, humiliating, or ridiculing employees while still reciprocating their efforts in terms of, for instance, rewards, assignments, or promotions. Lending empirical support for this notion, meta-analytical evidence has linked abusive supervision to different justice perceptions (Mackey et al., 2017). Of all justice types, abusive supervision appears the least associated with perceptions of distributive injustice (i.e., unfair outcome allocation). In contrast, the relationship with procedural (i.e., unfair decision processes) and interpersonal (i.e., unfair and disrespectful interpersonal treatment) injustice is much stronger. It is, therefore, plausible that followers attribute abusive supervisory behavior mainly to other reasons, such as a lack of self-control or being upset (Tepper & Almeda, 2012), but not easily to being exceedingly egoistic at the expense of others. Taken together, we theorize that impaired LMSX is the prime mechanism through which exploitative leadership affects leadership satisfaction, whereas abusive supervision (given it primarily works through affective reactions) is assumed to be less detrimental to LMSX:

Hypothesis 6

Abusive supervision exerts a weaker indirect effect on followers’ satisfaction with the leader through their perceived imbalance in LMSX, as compared to exploitative leadership.

To test the hypotheses under investigation, we conducted three studies. Study 1 is a two-wave field study, while Studies 2 and 3 use the power of experimental design to more thoroughly assess the causal nature of the proposed differential mechanisms¹.

Study 1

Sample and Procedures

Participants for this study were employees from various industries and occupations in Germany. Data were gathered via a German research panel provider at two points in time separated by approximately two weeks. At Time1, respondents were asked to rate their immediate leader in terms of exploitative leadership and abusive supervision and to provide personal information. At Time 2, respondents provided self-reports on the focal outcome variables. Overall, data of 166 respondents were successfully matched. The average tenure with the leader was 6.97 years (SD = 6.39). The mean age was 46.39 years (SD = 13.10), and 57.2% of the respondents were male. In terms of education, 44% had vocational training or similar, 19.9% had a high school degree, and 31.1% held a university degree.

Measures

We assessed exploitative leadership behavior using the 15-item measure developed by Schmid et al. (2019). A sample item was “My leader takes it for granted that my work can be used for his or her personal benefit” (the full list of items is presented as supplementary material). To capture employees’ perceptions of abusive supervision, we employed Mitchell and Ambrose’s (2007) shortened five-item version of Tepper’s (2000) measure, explicitly focusing on actively hostile and aggressive behaviors towards followers (e.g., “My supervisor tells me my thoughts and feelings are stupid”). For both leadership measures, respondents were instructed to indicate how frequently each item fits their leader, using a five-point continuum (1 = not at all to 5 = frequently if not always).

To assess respondents’ emotional responses towards the social exchanges with the leader, the five-item negative affect subscale taken from the Positive and Negative Affect Schedule (PANAS) Short Form (Thompson, 2007) was used. Respondents were presented with items about negative affect (e.g., afraid, upset) and then instructed to indicate the extent to which they feel this way in relation to the exchanges with their leader. The response choices ranged from 1 (very slightly) to 5 (extremely).

To capture respondents’ cognitions of social exchange with their leader, the eight-item LMSX scale developed by Bernerth et al. (2007) was used. A sample item was “My relationship with my leader is composed of comparable exchanges of giving and taking.” Items were rated on a five-point continuum, ranging from 1 (strongly disagree) to 5 (strongly agree). Since our theoretical focus was on perceptions of imbalance in social exchange, all item scores were reversed. Thus, high scores on the composite scale reflect a high level of perceived imbalance in social exchange (i.e., low reciprocity).

To measure respondents’ satisfaction with their leader, four items were included: one item adopted from Cicero et al. (2010) (“I receive great satisfaction from the relationship with my leader”), two items adopted from Bass and Avolio’s (1995) Multifactor Leadership Questionnaire (“My leader uses methods of leadership that are satisfying” and “My leader works with me in a satisfactory way”), and another general satisfaction item (“Overall, I am very satisfied with my leader”). Responses were given on a 5-point response scale (1 = strongly disagree to 5 = strongly agree).

Common Method Bias

We followed the recommendations by Conway and Lance (2010) to address concerns about common method bias. Firstly, we justify the use of self-reports for our focal variables. While both abusive supervision and exploitative leadership are in the eye of the beholder, followers are also best suited to self-report their responses in terms of affect, LMSX, and satisfaction. Secondly, common method bias was proactively considered in the design of the study. Respondents were guaranteed anonymity and assured that there were no right or wrong answers in the survey, thus minimizing both evaluation apprehension and social desirability. Furthermore, the survey for the self-reports was administered in two waves, with a lag of two weeks. Thirdly, we made sure that items for different constructs did not overlap, and we provided thorough evidence for construct validity among our measures.

Results

Descriptive Statistics and Validity of Measures

Table 1 presents the means, standard deviations, and correlations among the key study variables.

Table 1 Means, standard deviations, and intercorrelations (study 1)

Prior to testing our hypotheses, we examined whether our measurement model was appropriate. Therefore, we conducted confirmatory factor analysis (CFA) using the lavaan package in R (Rosseel, 2012) and tested a series of nested models. Given the relatively small sample size relative to the large number of estimated parameters, we used item parcels as manifest indicators for latent variables. For exploitative leadership, five parcels were built based on the five dimensions (i.e., genuine egoistic behaviors, taking credit, exerting pressure, undermining development, and manipulating; Schmid et al., 2019). Two parcels were created for abusive supervision, negative affect, as well as satisfaction, and three parcels were built for perceived imbalance by using the factorial algorithm. In this procedure, each parcel sequentially includes the items with the highest to the lowest factor loadings, alternating the direction of item selection (Matsunaga, 2008). On this basis, we tested a series of theoretically viable models. To evaluate differences in model fit, we focused on the change in chi-square and the change in CFI (Comparative Fit Index). The results of this procedure are reported in Table 2, showing that the fit of all alternative models was inferior to that of the predicted measurement model. Overall, these results indicate that our measures captured distinct constructs and represented valid tools to measure the target constructs.

Table 2 Comparison of measurement models (study 1)

Hypothesis Tests

Having established the integrity of our measures, we next tested the proposed structural linkages. Perceived LMSX imbalance and negative affect were allowed to correlate in this model². The estimated coefficients of this analysis are summarized in Fig. 2, providing an adequate fit with our data (χ² = 120.815, df = 67, p < 0.001, χ²/df = 1.803, CFI = 0.979, RMSEA = 0.070, SRMR = 0.03).

Fig. 2

SEM results (study 1)

In line with Hypothesis 1, abusive supervision predicted negative affect toward the relationship with the leader (β = 0.45, SE = 0.11, z = 4.733, p < 0.001), which, in turn, predicted satisfaction with the leader (β = −0.38, SE = 0.13, z = −4.843, p < 0.001). The unstandardized indirect effect of abusive supervision on satisfaction with the leader through negative affect was ab = −0.34, 95%, SE = 0.18, CI [−0.85, −0.11]. Exploitative leadership was also related to negative affect (β = 0.32, SE = 0.07, z = 3.507, p < 0.001), and its unstandardized indirect effect on satisfaction ratings through negative affect was ab = −0.16, SE = 0.08, 95% CI [−0.34, −0.02]. Thus, Hypothesis 2 was supported. To see whether the difference between the two indirect effects was significant, we tested a model in which the two indirect effects (i.e., abusive supervision → negative affect → satisfaction; exploitative leadership → negative affect → satisfaction) were constrained to be equal against a model in which the indirect effects were freely estimated. The obtained difference was significant (Δχ²₍₁₎ = 109.150, p < 0.001, ΔCFI = 0.042). Thus, Hypothesis 3 was supported.

Next, exploitative leadership was related to LMSX imbalance (β = 0.49, SE = 0.12, z = 4.960, p < 0.001), which predicted satisfaction with the leader (β = −0.46, SE = 0.06, z = −7.982, p < 0.001). Exploitative leadership exerted a direct effect on satisfaction (β = −0.22, SE = 0.09, z = −3.289, p < 0.001). The unstandardized indirect effect through perceived LMSX imbalance was ab = −0.30, SE = 0.09, 95% CI [−0.50, −0.15]. Thus, Hypothesis 4 was supported. Abusive supervision did not predict LMSX imbalance. Consequently, the indirect effect on satisfaction ratings was also found to be non-significant [ab = −0.08, SE = 0.11, 95% CI (−0.30, 0.13)]. Hence, Hypothesis 5 was rejected. In turn, we obtained evidence supporting Hypothesis 6. Specifically, we found that a model with the two indirect effects constrained to be equal performed significantly worse compared to an unconstrained model (Δχ²₍₁₎ = 85.008, p < 0.001, ΔCFI = 0.035).

These results provide initial support for the predicted differential mechanisms. Whereas abusive supervision primarily worked through negative affective responses, exploitative leadership exerted its negative influence through impaired reciprocity beliefs.

Study 2

To shed further light on the proposed differential mechanisms of exploitative leadership and abusive supervision, we designed a second study using an experimental design to assess the direction of causality between the focal study variables.

Sample and Procedures

Following an experimental vignette approach (Aguinis & Bradley, 2014), we created three hypothetical scenarios to manipulate leader behavior. To facilitate experimental realism and, thus, the generalizability of results, we incorporated critical incidents, mostly about exploitative leadership, obtained from 30 semi-structured interviews about destructive leadership that we conducted with working individuals³. The interviewees describe, among other things, situations in which exploitative behavior is evident, for example, when the leader demands excessive sacrifices, withholds information, or claims credit for the accomplishments and contributions of team members. All participants of the experimental study were presented with the same baseline information to ensure contextual grounding. Specifically, the scenarios described a situation in which a leader assigned the participant with a project for an external customer. The first scenario covered exploitative leader behaviors (i.e., egoism, taking credit, exerting pressure, underchallenging, and manipulating), while the second scenario described abusive leader behavior in terms of interpersonal acts of aggression and hostility (e.g., behaving in a nasty or rude manner, yelling at followers, and belittling them). The third scenario did not contain any information about exploitative or abusive leader behaviors and served as a neutral control condition (see supplementary material). Because of time constraints and concerns about potential carryover effects between conditions, we applied a between-person design.

Snowball sampling, starting from the networks of participants of a Part-Time executive MBA program at a large business school in Germany, was used to recruit participants for this study. The final sample consisted of 319 working adults from diverse industries and jobs in Germany, who were randomly assigned to one of the three conditions. The mean age of participants was 35.89 (SD = 10.09) and 62% were male. Their average work experience was 12.21 years (SD = 9.12). In terms of education, 2.5% had vocational training, 6.6% had a high school degree, 34.8% had a Bachelor’s degree, 48.7% had a Master’s degree, and 7% held a PhD or doctoral degree. These sociodemographic characteristics did not differ by experimental condition.

Measures

For the manipulation check, participants were instructed to rate the presented leader behaviors in terms of exploitative leadership and abusive supervision. Because of constraints in the time we had available for working adults to read a scenario and to answer the survey questions, we had to be concise. As a result, we measured exploitative leadership using only six of the 15 items of the exploitative leadership measure (Schmid et al., 2019). We selected these six items based on relevancy and the size of factor loadings. Using data from Study 1, we found a correlation of 0.97 (p < 0.001) between this 6-item measure and the full scale, indicating that the selected items covered the core variance of exploitative leadership. Previous research has successfully applied similar procedures with shortened instruments in experimental vignette studies (Van Dierendonck et al., 2014). For abusive supervision, we used the five-item measure reported by Mitchell and Ambrose (2007) as described in Study 1. All leadership-related measures were anchored on a five-point response scale, ranging from 1 (strongly disagree) to 5 (strongly agree).

To assess participants’ cognitive (i.e., LMSX imbalance) and affective (i.e., negative affect) responses relative to the exchange relationship with the leader, the same measures and response formats as in Study 1 were used. Because of the time restrictions, satisfaction with the leader was assessed with only two items adopted from the previous study (“I receive great satisfaction from the relationship with this leader” and “This leader uses methods of leadership that are satisfying “). Importantly, respondents were thoroughly instructed to refer their responses toward the leader described in the scenario.

Results

Table 3 shows the reliability estimates and correlations among the study variables by condition.

Table 3 Correlations across experimental conditions (study 2)

Manipulation Tests

An initial multivariate analysis of variance (MANOVA) revealed a statistically significant difference in respondents’ ratings of the three leadership scenarios in terms of abusive supervision and exploitative leadership, F (4, 630) = 332.79, p < 0.001; Wilk's Λ = 0.10, partial η² = 0.68. Separate one-way analyses of variance (ANOVA) with post-hoc comparisons via Tukey’s HSD test revealed that all mean differences were in the predicted direction, showing that the manipulations were successful (see Table 4). In the exploitative leadership condition, participants scored significantly higher on the exploitative leadership measure than in the abusive supervision condition and the neutral condition (F(2, 316) = 224.76, p < 0.001, partial η² = 0.59). Likewise, in the abusive supervision condition, respondents scored significantly higher on the abusive supervision scale as compared to the exploitative leadership condition and the neutral condition (F(2, 316) = 481.24, p < 0.001, partial η² = 0.75).

Table 4 Mean Scores in the three experimental conditions (study 2)

Hypothesis Tests

A MANOVA revealed there was a statistically significant difference in respondents’ reactions (i.e., perceived LMSX imbalance, negative affect, and satisfaction with the leader) based on the leadership manipulation, F (6, 628) = 144.84, p < 0.001; Wilk’s Λ = 0.18, partial η² = 0.58. Results of separate ANOVAs for each outcome variable are reported in Table 4. As expected, both abusive supervision and exploitative leadership triggered similarly low scores of satisfaction with the leader. With regard to negative affect toward the relationship with the leader, the mean score for the abusive supervision condition was significantly higher than for the exploitative leadership condition and the neutral condition. In a similar vein, LMSX imbalance scores were highest in the exploitative leadership condition and significantly different from the other conditions.

To test the proposed mediating effects, we analyzed our data within a path analysis framework. Two dummy variables were created to model the experimental conditions. That is, in the abusive supervision condition, abusive supervision was coded as 1, and the two other conditions were coded as 0. In the exploitative leadership condition, exploitative leadership was coded as 1 and the two other conditions were coded as 0. Figure 3 shows the results of this analysis. In line with the ANOVA results, abusive supervision predicted negative affect toward the relationship with the leader (β = 0.86, SE = 0.08, z = 21.395, p = 0.008), which, in turn, predicted satisfaction with the leader (β = −0.17, SE = 0.05, z =−4.496, p < 0.001). The unstandardized indirect effect of abusive supervision on satisfaction with the leader through negative affect was ab = −0.42, 95%, SE = 0.10, CI [−0.61, −0.23]. Thus, Hypothesis 1 was supported. Also, exploitative leadership was related to negative affect (β = 0.60, SE = 0.09, z = 14.843, p = 0.008). The unstandardized indirect effect on satisfaction ratings via negative affect was ab = −0.30, SE = 0.07, 95% CI [−0.44, −0.17], supporting Hypothesis 2. A comparison of the two indirect effects revealed a significant difference (Δχ²₍₁₎ = 23.524, p < 0.001, ΔCFI = 0.02), thereby supporting a weaker effect for exploitative leadership and corroborating Hypothesis 3.

Fig. 3

Mediation model (study 2)

Next, exploitative leadership predicted perceived LMSX imbalance (β = 0.91, SE = 0.09, z = 20.135, p < 0.001), which was negatively related to satisfaction with the leader (β = –0.40, SE = 0.08, z = −7.456, p < 0.001). The unstandardized indirect effect of exploitative leadership on satisfaction through perceived LMSX imbalance was ab = −0.1.10, SE = 0.15, 95% CI [−1.43, −0.81]. Hence, Hypothesis 4 was supported. Abusive supervision also caused LMSX imbalance (β = 0.63, SE = 0.09, z = 14.126, p < 0.001). The unstandardized indirect effect on satisfaction ratings was ab = −0.74, SE = 0.11, 95% CI [−0.97, −0.53], lending support to Hypothesis 5. The difference between the two indirect effects was significant (Δχ²₍₁₎ = 52.195, p < 0.001, ΔCFI = 0.041), indicating a weaker effect for abusive supervision. Thus, Hypothesis 6 was again supported.

Overall, the causal pattern that emerged from this study supports our core predictions. While abusive supervision appears to more strongly evoke negative affective reactions, exploitative leadership is a stronger source of perceived LMSX imbalance. Nonetheless, a limitation of this study is the use of a single-factor design, investigating abusive supervision and exploitative leadership in isolation. While the manipulations we used effectively encompass the fundamental aspects of these underlying concepts, there is a possibility of potential confounding due to the inclusion of supplementary information within each condition beyond merely indicating the presence, absence, or extent of a single construct. To overcome these shortcomings, we designed an additional study using a fully crossed design.

Study 3

Study 2 applied a full factorial experimental vignette approach to more thoroughly disentangle the distinct causal effects of exploitative leadership and abusive supervision. This methodological choice allowed us to systematically assess the impact of each destructive leadership type across all possible combinations of levels (high vs. low) alongside its counterpart. This reduces the potential for uncontrolled confounding variables and provides a more comprehensive understanding of how the leadership manipulations influence the response variables (Podsakoff & Podsakoff, 2019). This approach resonates with prior research efforts that delved into the distinct impacts of various leadership behaviors via controlled experimental manipulations (Howell & Frost, 1989; Podsakoff et al., 2011).

Sample and Procedures

We adapted the materials we used in Study 2 and created four hypothetical scenarios, combining different levels of abusive supervision and exploitative leadership (i.e., high versus low), thus resulting in a 2 × 2 experimental design. The adjustments of the scenarios aimed at optimizing the differentiation between high and low values of the two focal leadership constructs. Again, a between-person design was applied. To avoid order effects, we randomly varied the order in which the descriptions of (high versus low) abusive supervision or exploitative leadership were presented within each condition.

Data were gathered via a professional research panel provider from Germany. The final sample consisted of 285 working adults from a diverse set of industries and jobs in Germany, who were randomly assigned to the experimental conditions. The mean age of participants was 47.95 (SD = 12.34) and 44% were male. Their average work experience was 24.49 years (SD = 13.12). In terms of education, 44.2% had vocational training or similar, 25.6% had a high school degree, and 30.2% held a university degree. These sociodemographic characteristics did not differ by condition.

Measures

For the manipulation check, participants were instructed to rate the presented leader behaviors in terms of exploitative leadership and abusive supervision. For exploitative leadership, the full 15-item measure by Schmid et al. (2019) was used. Abusive supervision was measured with the five-item scale by Mitchell and Ambrose’s (2007), as described in Studies 1 and 2. Both leadership measures were anchored on a five-point response scale, ranging from 1 (strongly disagree) to 5 (strongly agree).

Participants’ cognitive (LMSX imbalance) and affective (negative affect) responses relative to the exchange relationship with the leader were assessed with the same measures and response formats as described in the previous study. Again, respondents were thoroughly instructed to refer their responses toward the leader described in the scenario.

In this study, we had the advantage of being able to include a greater number of items. Therefore, we adopted all four satisfaction items from Study 1 to more comprehensively assess satisfaction with the leader (e.g., “I receive great satisfaction from the relationship with this leader” and “This leader uses methods of leadership that are satisfying”). Responses were given on a 5-point response scale (1 = strongly disagree to 5 = strongly agree).

Results

Table 5 shows the reliability estimates and correlations among the study variables across the four conditions.

Table 5 Correlations across experimental conditions (study 3)

Manipulation Tests

We conducted a two-way ANOVA having two levels of exploitative leadership (low, high) and two levels of abusive supervision (low, high). Results are shown in the upper part of Table 6. With regard to the abusive supervision manipulation check, the interaction effect was non-significant, F (1, 281) = 0.18, p = 0.67, partial η² = 0.001. While we found a non-significant main effect of exploitative leadership, F (1, 281) = 0.007, p = 0.93, partial η² = 0.00, the main effect pertaining to abusive supervision was significant, F (1, 281) = 339.84, p < 0.001, partial η² = 0.56, indicating that the mean abusive supervision scores were significantly greater for the high abusive supervision conditions. Regarding the exploitative leadership manipulation check, we found a non-significant interaction term, F (1, 281) = 3.03, p = 0.08, partial η² = 0.01. The main effects were significant for both the exploitative leadership manipulation, F (1, 281) = 107.601, p < 0.001, partial η² = 0.27, and the abusive supervision manipulation, F (1, 281) = 37.92, p < 0.001, partial η² = 0.12. Importantly, post-hoc comparisons via the Tukey HSD test revealed that the exploitative leadership mean scores in the four conditions of interest differed in the predicted direction with significantly higher mean scores in the high exploitative leadership conditions. This indicates that our exploitative leadership manipulation worked as intended.

Table 6 Mean scores in the 4 experimental conditions (study 3)

Hypothesis Testing

We first ran a series of two-way ANOVAs to see whether abusive supervision and exploitative leadership would differentially impact the focal outcomes. The mean scores across the four experimental conditions are shown in the lower part of Table 6, while the obtained main and interaction effects are reported in Table 7. The main effects as well as the mean differences were in line with our predictions. First, the low/low condition produced the highest satisfaction scores while yielding the lowest scores on negative affect and LMSX imbalance, and was significantly different from all other conditions. Second, it was expected that abusive supervision would be the dominant source of affective responses. In support of this prediction, respondents in the high abusive supervision conditions reported significantly higher mean scores of negative affect towards the relationship with the leader than in the conditions where abusive supervision was low, even when exploitative leadership was high. Third, we expected that LMSX imbalance would be more strongly affected by exploitative leadership. This was supported as the high exploitative leadership conditions produced higher mean scores of LMSX imbalance than the conditions where exploitative leadership was low, albeit abusive supervision was high.

Table 7 Results of 2 × 2 ANOVA (study 3)

Subsequently, we tested the predicted mediation model using path modeling (see Fig. 4). Exploitative leadership and abusive supervision were each included as a dummy variable (low vs. high). In accordance with the ANOVA results, abusive supervision predicted negative affect (β = 0.52, SE = 0.10, z = 10.600, p < 0.001), which, in turn, was negatively related to satisfaction with the leader (β = −0.19, SE = 0.07, z = −0.3.847, p < 0.001). The unstandardized indirect effect of abusive supervision on satisfaction with the leader through negative affect was ab = −0.26, 95% CI [−0.43, −0.12], supporting Hypothesis 1. Exploitative leadership also predicted negative affect (β = 0.16, SE = 0.10, z = 3.142, p = 0.002) and the unstandardized indirect effect on satisfaction was ab = −0.08, 95% CI [−0.15, −0.03]. Thus, Hypothesis 2 was supported. The indirect effect of exploitative leadership via affective responses was significantly weaker than the indirect effect of abusive supervision (Δχ²₍₁₎ = 20.485, p < 0.001, ΔCFI = 0.031).Thus, Hypothesis 3 was supported.

Fig. 4

Mediation model (study 3). N = 285. LMSX = Leader-member social exchange. Dashed lines represent nonsignificant paths. ***p < 0.001. **p < 0.01

In line with Hypothesis 4, exploitative leadership was related to LMSX imbalance (β = 0.52, SE = 0.10, z = 11.066, p < 0.001), which predicted satisfaction with the leader (β = −0.65, SE = 0.06, z = −11.587, p < 0.001). The unstandardized indirect effect on satisfaction was ab = −0.89, 95% CI [−1.15, −0.67]. Abusive supervision also was associated with LMSX imbalance (β = 0.35, SE = 0.10, z = 7.549, p = 0.000) and the unstandardized indirect effect on satisfaction was ab = −0.60, 95% CI [−0.79, −0.43]. Hence, Hypothesis 5 was supported. This indirect effect was significantly weaker than the indirect effect of exploitative leadership (Δχ²₍₁₎ = 7.123, p < 0.01, ΔCFI = 0.01), thus corroborating Hypothesis 6.

Overall, the results of this fully crossed experimental design support the general patterns obtained in the previous studies in that abusive supervision was again more strongly associated with negative emotions about the exchanges with the leader, while exploitative leadership more strongly evoked cognitions of impaired reciprocity.

General Discussion

The current research set out to further our understanding of destructive leadership and examine the unique contribution of exploitative leadership. We assessed whether and to what degree exploitative leadership, as compared to abusive supervision, would affect follower satisfaction through distinct facets of social exchange. Across three studies reported here (and three additional studies reported as supplementary material), we found general support for our theoretical predictions. While abusive supervision appears to work primarily through evoking negative affect, exploitative leadership operates mainly through followers’ cognitive understanding of the social exchange (i.e., LMSX imbalance).

Theoretical Implications

As previously noted, emerging research has documented the detrimental effects of exploitative leadership on followers. Far less attention has been given to its position— most notably, relative to abusive supervision—within the destructive leadership domain, as well as its unique theoretical mechanisms. Though prior work suggested that exploitative leadership elicits less emotional responses among followers (Schmid et al., 2018), it has remained unclear as to what instead drives its negative effects. Our research helps to fill this gap by positioning exploitative leadership and abusive supervision as varying (i.e., emotional or “hot” versus instrumental or “cold”) forms of aggression that differentially impact distinct (i.e., affective versus cognitive) elements of the social exchange relationship experienced by employees. By invoking theoretical distinctions from the aggression literature, we offer a useful perspective for mapping the destructive leadership domain, one formerly unavailable or merely incipiently reflected in destructive leadership research (Tepper et al., 2012). Furthermore, our results highlight the importance of “applying general theories in more specific ways to better elucidate the theoretical differences among forms of leadership” (Lemoine et al., 2019, p. 155). The broad utilization of LMX conceptualizations and measures (Liden et al., 1997) would not have allowed us to adequately describe the varying follower social exchange experiences that are evoked by exploitative leadership and abusive supervision. With these distinctions, we enrich our understanding of how and why exploitative leadership and abusive supervision exert negative effects. This is not to say, exploitative leadership only works through the violation of reciprocity expectations, nor that abusive supervision only works through negative affective responses. Nonetheless, these two mechanisms uniquely evince that exploitative leader behaviors are appraised and processed differently than abusive supervision. In this way, we provide new evidence that exploitative leadership adds complementary, rather than redundant, knowledge concerning the nature and mechanisms of destructive leadership. Exploitative leader behavior hence merits distinct theoretical and conceptual treatment. In so doing, theoretical frameworks and corresponding models will offer more appropriate and meaningful examinations.

A remarkably consistent pattern that emerged from our field studies (Study 1 and supplementary Studies 1a and 1b) was the absence of a substantive link between abusive supervision and LMSX imbalance. Conversely, the experimental studies within our research showed a discernible association between the two, albeit notably smaller compared to the effect associated with exploitative leadership. Plausible conjecture points to the stimuli employed in our experimental settings, which likely accentuated the facets of abusive supervision with the potential to breach reciprocity norms. This salience might differ when dealing with naturally occurring covariance in organizational contexts. Here, instances of abusive behavior might be less frequent, and if they do manifest, the dominant response is emotional, overshadowing the cognitive dimension.

Perhaps one of the most challenging findings was the link between exploitative leader behavior and followers’ negative affect toward their social exchange relationship. A few remarks here are warranted. We are aware of only a few efforts that have investigated affective responses to perceptions of exploitation in the work context. The closest research to that presented here is that by Livne-Ofer et al. (2019), who examined exploitation in the employee-organization relationship. They found that, as compared to abusive supervision, employees’ perceptions of organizational exploitation (i.e., the perception that one has been purposefully taken advantage of in one’s relationship with the organization) were more strongly related to negative affect. Yet, they measured negative affect as the way in which one is treated by the organization—not negative affect towards exchanges with their leader. Given that better prediction of outcomes can be expected when the level of specificity of the predictor matches that of the criterion (Rupp et al., 2014), it is likely their findings are driven by the (mis)fit between source and target (i.e., organization-leader), rather than the type of behavior (i.e., exploitative versus abusive).

Be that as it may, our finding that abusive supervision is primarily an affective event that affects followers through the experience of negative emotions is consistent with extant research regarding interpersonal stressors, such as emotional conflict with coworkers or interpersonal injustice. Several researchers have found that interpersonal conflict is one of the most powerful stressors in the workplace, followed by negative affect (e.g., Ilies et al., 2011). In a similar vein, research on organizational injustice has consistently supported the observation of Bies (2001) that interpersonal injustice—characterized by a lack of politeness, respect, and dignity—is an inherently emotional, “hot and burning” experience (Bembenek et al., 2007, p. 90). In turn, our finding that the effect of exploitative leadership is primarily carried by LMSX imbalance, as opposed to negative affect, resonates with the theorizing in the field of distributive justice (i.e., employees’ perceptions about the fair allocation of outcomes). While justice theorists have stressed that inequitable treatment causes tension and distress (Cropanzano et al., 2011), such theorists have likewise argued that distributive discrepancies are rarely at the root of strong emotional reactions. Instead, the affect aroused by violations of distributive justice seems more appropriately conceptualized as a lack of happiness and a sense of disappointment (Bembenek et al., 2007). As such, these diverse concepts of justice can enhance our comprehension of the distinct responses to different destructive leader behaviors. Abusive supervision may represent a source of interpersonal injustice, evoking strong emotional reactions. Exploitative leadership more strongly undermines followers' perceptions of fair outcomes, which are thought to be less interpersonally offensive and emotionally charged (Bembenek et al., 2007; Greenberg & Geanegoda, 2007).

In conclusion, by differentially linking exploitative leadership and abusive supervision to distinct theoretical mechanisms, our research deepens our understanding of what destructive leaders actually do and how and why this affects followers in a unique way. Through critical disentangling unethical leadership forms (Babalola et al., 2022), we strengthen theoretical clarity and foster agreement as to what we mean when we describe leader behaviors as “unethical” or “destructive,” ensuring that our theories not only articulate what a construct is but also what it is not. Otherwise, generally considering abusive supervision and exploitative leadership as “destructive” obscures potentially important variance in both the constructs themselves and their correlates, and oversimplifies much more complex phenomena” (Lemoine et al., 2019, p. 4), and is, simply put, misleading.

Practical Implications

Far from being a sole theoretical exercise, fleshing out the distinct mechanisms of exploitative leadership and abusive supervision offers important implications on a practical level. The first implication is rather straightforward, emphasizing that the prevention of leader abuse and exploitation is an important strategy to control employee satisfaction. While destructive leader behaviors in general are often shown “covertly, unseen by observers or in a manner that perpetrators can shrug off as innocent misunderstandings” (Tepper et al., 2017, p. 126), this tendency may be particularly true for exploitative leadership. Given exploitative leaders use more clandestine and manipulative behaviors, it will be more likely to remain off the radar for organizational decision-makers. Consequently, finding ways to better identify the prevalence of exploitative leadership in organizations, such as incorporating it in 360-degree feedback, is an important first step to addressing and reducing exploitative leadership. Doing so may help reveal the organization’s true leadership culture and prevent the followers’ development of broader perceptions of organizational exploitation (Livne-Ofer et al., 2019). By pinpointing the type of destructive behaviors reported by employees, organizations can learn of their implicit acceptance or even permission for leaders to behave in certain ways. Organizations’ explicit communication of clear standards and normative expectations can also make it more difficult for exploitative leaders to hide behind excuses or rationalizations. Such open communication prevents employees from attributing exploitative leadership as an intentional or structurally-anchored organizational mode.

While reducing destructive leadership should remain the priority, influencing and controlling leaders’ behavior is difficult, and meaningful change takes time. Organizational interventions may additionally consider how to directly impact the key mediators of destructive leadership effects we identified in this research. Importantly, our results indicate that these interventions must be carefully tailored. Concerning exploitative leadership, tackling perceptions of reciprocity, as opposed to negative affect, would be a more effective approach. Further, organizations may seek to attenuate the effects of exploitative leaders by ensuring that exchanges at other levels (e.g., pay, promotion prospects, and continuing education opportunities) are perceived as fair and balanced. Such tactics may be less efficient in buffering against abusive supervision (Thau & Mitchell, 2010), where it will instead be more important to address the emotional responses, as opposed to perceived LMSX imbalance. Organizations might consider investing in initiatives (e.g., coaching) that focus directly on promoting followers’ self-resources, such as resilience and emotional management.

In order to effectively implement the suggestions and ideas mentioned above, organizations are encouraged to consider the implementation of a psychological safety officer role as part of their internal control framework. This role would serve as an internal control point, ensuring that the organization has robust measures in place to identify, assess, and manage the risks associated with destructive behaviors (coming from multiple sources, including colleagues, costumers, and leaders). The psychological safety officer would play a pivotal role in offering employees suffering from abusive and/or exploitative leaders a platform to voice their concerns and file complaints. Empowering followers in this manner not only provides them with an avenue to seek support but also restores a sense of control that may have been compromised by their experiences with destructive leadership. In a similar vein, research on coping with negative emotions suggests that seeking social support can effectively reduce negative affect (Dewe et al., 2010). Thus, encouraging targets of destructive leaders to engage in open discussions with colleagues, share their feelings, and seek advice from individuals who have faced similar situations can be highly beneficial.

Limitations and Future Research

In this research, we tested our hypotheses with several independent samples while employing different designs and measures. This approach aligns with the principle of constructive replication (Köhler & Cortina, 2019), and the consistent results across studies bode well for both the robustness of our findings and the predictive ability of distinct social exchange facets. While we acknowledge that each of our studies has limitations, we argue that the strengths of each compensate for weaknesses in the others. Study 1 used a relatively small sample size yet demonstrated high levels of external validity associated with field research. Studies 2 and 3 used larger samples and feature high levels of internal validity that are associated with experiments.

Although our studies complement one another, some limitations merit further attention. Most notably, whereas our results from the experimental studies suggest that the proposed effects of the two leadership variables on key mediators (i.e., negative affect and perceived LMSX imbalance) are causal, this interpretation does not hold for other parts of the proposed model. Our design does not allow causal inferences with regard to the effects of our mediators (i.e., perceived LMSX imbalance and negative affect) on satisfaction ratings as our focal dependent variable. Nonetheless, given that the proposed relations have been consistently evidenced in prior research, we consider our proposed causal chain plausible. Future research will benefit from testing these predictions within longitudinal or experimental designs.

Next, we recognize that the relationship between cognitive and affective responses deserves further consideration. While abusive supervision appears to directly trigger negative affect through transfer and overtly denigrating followers, it is possible that the effect of exploitative leadership on negative affect may be delayed and partially arise from cognition, including perceptions of impaired LMSX. That is, as exploitation and perceptions of reciprocity imbalances accumulate over time, followers may increasingly seek to understand the causes behind the perceived leader behaviors, thus facilitating rumination and, eventually, stronger negative affect. However, the methods we have used herein were not suitable to test such intricacies, as this would require repeated measures over a longer period of time. Alternatively, experimental work could use neuroscience methods to examine neural activity associated with emotions and cognition (Waldman et al., 2019). Therefore, shedding light on the exact interplay and sequence of affect and cognition in the response to exploitative leadership constitutes a promising future research direction.

Another limitation is that we did not consider individual characteristics of followers that could influence how they perceive and react to destructive leadership. In fact, as both abusive supervision and exploitative leadership reflect a subjective evaluation (Schmid et al., 2019; Tepper et al., 2017), it will be fruitful in future work to control for individual traits that may influence how individuals appraise, process, and deal with abusive supervisory and exploitative leadership behaviors. For instance, while prior research indicates that the perception of interpersonally hostile leader behavior is related to followers’ trait negative affectivity or hostile attribution style (Tepper et al., 2017), the overestimation of one’s own contribution or holding the exaggerated belief that one is better than others (Moore & Schatz, 2017), are more likely to cause employees to interpret their leader's behavior as exploitative. A valuable addition to such studies would be the inclusion of followers’ attributions regarding the motives behind their leader’s destructive leadership. Previous research has shown that abusive supervision results in more negative outcomes, including negative affect, when targets attribute the mistreatment to the motive of causing harm, as opposed to the motive of promoting performance (Liu et al., 2012; Yu & Duffy, 2021). It would be intriguing to explore whether and to what degree reactions to exploitative leadership might be influenced by similar attributions.

Finally, the specific ways in which leaders and followers interact in the process of abusive supervision or exploitative leadership also raise the possibility that both forms of leadership operate differently at the group level. Exploitative leadership may often be performed more covertly, while abusive behaviors “tend to be more spontaneous, harder for the leader to control […], and more likely to occur in the presence of witnesses” (Krasikova et al., 2013, p. 19). Moreover, in organizations with a strong hierarchical or even exploitative culture (Livne-Ofer et al., 2019), exploitative leadership may not be an exception but a normality, whereas abusive leader behaviors may still represent a norm violation and thus increase their attentional pull. This would mean climate perceptions that emerge from abusive supervision versus exploitative leadership vary considerably in their strength and in turn, differentially affect group-level outcomes. Future research should therefore investigate these interesting and relevant group phenomena accordingly.

Conclusion

Over time, researchers have introduced numerous new constructs and labels to describe various aspects of unethical and destructive leadership, often resulting in overlapping or redundant concepts. This proliferation can lead to confusion, inconsistencies, and a lack of clarity in the literature. In this study, we set out to investigate the difference between abusive supervision and exploitative leadership as two types of destructive leadership. Describing them as different forms of aggression, we find that both negative leader behaviors undermine satisfaction with the leader, albeit through different mechanisms. Abusive supervision, being fundamentally rooted in interpersonal provocation, primarily impacts followers emotional experiences within the social exchange relationship with the leader. Exploitative leadership, with its inherent focus on self-interest, primarily influences followers through their cognitive understanding of the social exchange. By clarifying the conceptual distinctions between exploitative leadership and abusive supervision and providing empirical support for their distinct mechanisms, this study helps to refine the existing body of literature on destructive leadership, ensuring more precise and nuanced insights into the processes associated with destructive leadership.