Does shared leadership always work? A state-of-the-art review and future prospects

Purpose – The purpose of this study is to sort out the potential dark sides of shared leadership so as to promote a more comprehensive and balanced views of the impact of shared leadership and provide directions for future research. Design/methodology/approach – Throughextensivedatabaseandmanualsearches,766literaturerecords were obtained. After three roundsof literaturescreening, 17studieswere retained.On thisbasis,the 17studies were coded and analyzed. Findings – From the perspectives of individual motivation, hierarchical functionalism and leadership role configuration, the existing studies have explored the negative impacts of shared leadership on team members, formal team leaders and the overall work teams. Specifically, for team members, shared leadership may cause negative consequences like power struggle, role stress and knowledge hiding. For formal team leaders, shared leadership may cause negative consequences like psychological territorial loss, leadership motivation declines and the dualistic paradox of self and group. For the overall work teams, shared leadership may cause negative consequences like team performance inhibition, low decision-making efficiency, team responsibility dispersion and teamcreativitydecline.Meanwhile,contextualfactorsplayakeyroleindeterminingtheeffectsofsharedleadership. Originality/value – Through a systematic review of the negative impact of shared leadership, this study responds to the research calls for exploring the dark sides of shared leadership, provides the academic community with a more comprehensive and balanced view of the impact of shared leadership and identifies several directions for future research.


Introduction
In the past two decades, shared leadership, a dynamic leadership model that advocates the sharing of leadership roles and influence among team members (Carson et al., 2007) has received extensive attention from a range of disciplines, including organizational behavior, industrial and organizational psychology and strategic management. Studies on shared leadership have shown that this type of leadership has positive impacts on individuals, teams Dark side of shared leadership 51 and organizations, such as improving job satisfaction (e.g. Drescher and Garbers, 2016;Serban and Roberts, 2016), enhancing team creativity (e.g. Ali et al., 2020;Gu et al., 2022) and improving team or organizational performance (e.g. Chen et al., 2022;D'Innocenzo et al., 2021). In addition, some recent meta-analyses have identified the unique effect of shared leadership in explaining the variance in team performance when compared with the traditional vertical leadership (e.g. Nicolaides et al., 2014;Wang et al., 2014). As such, shared leadership gets its name from its positive impact on a variety of desired outcomes.
Despite the overwhelmingly positive view of shared leadership in the literature, some scholars have raised concerns about its potential dark sides (e.g. Pearce et al., 2007;Zhu et al., 2018). In particular, Zhu et al. (2018) pointed out that the inherent attributes of shared leadership, such as time-consuming, equal and decentralized power, will cause undesirable effects on team members, team's formal leaders and the overall work groups. In addition, empirical evidence also cautions against an overly optimistic treatment of shared leadership. For instance, empirical evidence is emerging that indicates shared leadership can lead to role stress (Wang and Peng, 2022), power struggle (Ji, 2018) and knowledge hiding (Zhao, 2013), which damage team effectiveness. The above phenomenon shows that while shared leadership brings positive effects, it can also cause destructive effects within the team. In this regard, scholars have called for the academic community to view shared leadership from a more comprehensive and balanced point of view (Zhu et al., 2018, p. 849).
However, compared with its positive impact, the current research on the negative impact of shared leadership is still scattered, with no literature review systematically summarizing its potential dark sides, the academic community lacks a comprehensive and clear cognition of its negative impacts. To fill this gap, the current study aims to provide a detailed literature review to summarize the negative effects of shared leadership and form a comprehensive analytical framework accordingly. Specifically, this paper attempts to explain what negative effects of shared leadership have been empirically or theoretically established, when and why shared leadership leads to these negative effects, what directions are promising for future research. By answering the above questions, the current study responds to the research calls for exploring the potential dark sides of shared leadership, provides the academic community with a more comprehensive and balanced view of the impact of shared leadership, and identifies several directions for future research.

Definition and characteristics of shared leadership
Nowadays, the widespread application of cross-functional teams and the organization's growing need for leadership sharing between superiors and subordinates have promoted the rapid development of shared leadership theory and practice (Pearce and Conger, 2003). By definition, shared leadership refers to "an emergent team phenomenon whereby the leadership functions and responsibilities are broadly shared among internal team members to better achieve team goals" (Carson et al., 2007;Pearce and Conger, 2003). Different from the traditional vertical leadership theory that focuses on the styles and behaviors of formally appointed leaders, the shared leadership theory emphasizes the agency role of team members in the process of team leadership (Carson et al., 2007). Notably, the study by Zhu et al. (2018) distinguished the theoretical overlaps and differences between shared leadership and six other similar leadership constructs (i.e. collective leadership, empowering leadership, team leadership, emergent leadership, self-leadership and participative leadership), and refined the main characteristics of shared leadership into three aspects, including source of leadership influence, units of analysis and distributions of leadership influence.
First, for the source of leadership influence, shared leadership involves not only the top-down hierarchical influence that from team leaders to team members, but also the horizontal/lateral influence among team members and even the bottom-up influence that from team members to team leaders (Pearce and Conger, 2003). As a result, shared leadership is able to establish a complex influence network within the group. Second, for the units of analysis, shared leadership views leadership as a team-level phenomenon, which distinguishes it from leadership models that focusing on the individual-level or dyadic-level phenomenon, such as empowering leadership, self-leadership, participative leadership and etc. (Carson et al., 2007). Third, for the distributions of leadership influence, the leadership roles and influences are widely shared among team members under shared leadership conditions, resulting in numerous leaderfollower dualistic relationships within the group (Drescher et al., 2014). Meanwhile, the boundary between leader and follower becomes more blurred, as team members may perform both of these roles at a given time (Nicolaides et al., 2014). Finally, shared leadership is proposed to be dynamic in nature, that is, leadership will emerge dynamically among team members according to task requirements and team members' potential (Pearce and Conger, 2003).
The above characteristics make shared leadership a more effective leadership model as compared with the traditional vertical leadership in predicting individual and team effectiveness (e.g. D'Innocenzo et al., 2016;Lorinkova and Bartol, 2021;Wang et al., 2017). But at the same time, they also bring about a series of negative impacts, such as power struggle (Ji, 2018), role stress (Wang and Peng, 2022) and low decision-making efficiency (Zhu et al., 2018), which will be discussed in detail below.

Research design
3.1 Literature searching We conducted extensive database and manual searches to identify studies on the negative effects of shared leadership. First, computerized searches through the Web of Science, Scopus, EBSCO (BSP, ASP, ERIC and PsycINFO), ProQuest and CNKI databases were conducted in March 2022 using collocation search terms such as "shared leadership and dark side", "shared leadership and negative impact", "shared leadership and detrimental effect", etc. The time span was from January 2000 to March 2022, Table 1 lists the search strings used during the database search stage and the corresponding retrieval results. Second, a manual search was conducted for publications in major journals (e.g. "The Leadership Quarterly", "Journal of Applied Psychology", "Personnel Psychology", "Journal of Organizational Behavior") over the past 22 years. Third, we manually reviewed the reference list of qualitative and quantitative review articles (e.g. Nicolaides et al., 2014;Zhu et al., 2018) to identify additional studies. We ultimately obtained 754 records from the databases and 12 papers from the manual search processes, after removing 133 duplicate records, a total of 633 papers were included for title and abstract screening.

Literature screening and coding
To ensure the rationality of the included studies, the following inclusion criteria were formulated to screen the retrieved records. First, the research object of the included studies should be shared leadership, and studies focusing on other similar leadership models (e.g. collective leadership and empowering leadership) are not within the scope of this literature review. Second, the studies retained for analysis should provide either empirical test or theoretical deduction that reveal the negative effects of shared leadership. Third, the studies had to be published in English or Chinese. Based on the above criteria, we first conducted a double screening of the titles and abstracts of the retrieved papers, for which 95.1% agreement was achieved. Next, we discussed inconsistencies that occurred during the screening process until all inconsistencies were eliminated. The title and abstract screening stage excluded 548 records, producing 85 potentially relevant studies for further full-text examination. We then conducted a full-text review of the remaining 85 papers independently; Dark side of shared leadership for which, the interrater-observed agreement reached 96.5%, indicating good agreement between the two screeners. Similarly, we discussed inconsistencies until 100% agreement was achieved. As the result, 41 papers were excluded for being irrelevant to the current research topic, 15 papers were excluded for investigating the positive effects of shared leadership only, 8 papers were excluded for investigating the antecedents of shared leadership and 4 papers were excluded for research unavailable. Finally, a total of 17 studies met all the requirements and provided useful information for the current review (for a flow chart of the literature searching and screening processes; see Figure 1). On this basis, the second author coded the information of analytic perspective, action object as well as main findings and implications for the included studies, and the first author checked what the second author had coded. Finally, we formed the coding table as shown in Table 2.

Dark side effects of shared leadership
As can be seen from Table 2, existing studies mainly explore the negative effects of shared leadership from the perspectives of team members, formal team leaders and the overall work teams. In this regard, the current study also focuses its research results from the above three perspectives. In addition, in each section, we further divide the negative effects obtained into Database Search strings a Results(N) Web of Science (TS 5 "shared leadership" OR TI 5 "shared leadership" OR AB 5 "shared leadership") AND (AB 5 "no association" OR AB 5 "negative effect" OR AB 5 "negative impact" OR AB 5 "inversely associated" OR AB 5 "unexpectedly associated" OR AB 5 "dark side" OR AB 5 "detrimental effect" OR AB 5 "not associated" OR AB 5 "no effect" OR AB 5 "harmful" OR AB 5 "adverse impact" OR AB 5 "negative influence" OR AB 5 "adverse effect") 22 Scopus ALL ("shared leadership") AND ABS ("no association" OR "negative effect" OR "negative impact" OR "inversely associated" OR "unexpectedly associated" OR "dark side" OR "detrimental effect" OR "not associated" OR "no effect" OR "harmful" OR "adverse impact" OR "negative influence" OR "adverse effect") 223 ProQuest TI, AB, SU, FT("shared leadership") AND AB("no association" OR "negative effect" OR "negative impact" OR "inversely associated" OR "unexpectedly associated" OR "dark side" OR "detrimental effect" OR "not associated" OR "no effect" OR "harmful" OR "adverse impact" OR "negative influence" OR "adverse effect") 20 EBSCO (ASP, BSP, ERIC, APA PsycINFO) (TI (shared leadership) OR AB (shared leadership) OR SU (shared leadership) OR TX (shared leadership)) AND AB ((no association) OR (negative effect) OR (negative impact) OR (inversely associated) OR (unexpectedly associated) OR (dark side) OR (detrimental effect) OR (not associated) OR (no effect) OR (harmful) OR (adverse impact) OR (negative influence) OR (adverse effect)) 396 CNKI b (SU 5 "shared leadership" or AB 5 "shared leadership" or TI 5 "shared leadership" or KY 5 "shared leadership" or FT 5 "shared leadership") and AB5 ("dark side" þ "negative effect" þ "not associated" þ "harmful" þ "inversely associated" þ "detrimental effect" þ "no effect") 93 Note(s): a TS 5 Topic; TI 5 Title; AB 5 Abstract; ALL 5 All Fields; ABS 5 Abstract; SU 5 Subject; FT 5 Full Text; TX 5 All Text; KY 5 Keywords. b CNKI is the largest and most comprehensive literature database in China   Table 1.
Search strategy and results different topics, then describe the findings and provide specific details about when and why these negative effects will occur. On this basis, we develop an analytic framework of the dark side of shared leadership as shown in Figure 2.

The negative effects of shared leadership on team members
Although growing evidence indicates positive impacts of shared leadership in improving team member's creativity (Gu et al., 2016(Gu et al., , 2022, performance (Fu et al., 2020;Scott-Young et al., 2019) and satisfaction (Robert and You, 2017;Wood and Fields, 2007). Some recent empirical studies have shown that shared leadership also causes negative effects among team members, such as power struggles (Ji, 2018), role stress (Wang and Peng, 2022) and knowledge hiding (Zhao, 2013). 4.1.1 Power struggle. From the perspective of hierarchical functionalism, the empirical study by Ji (2018) pointed out that the dynamic nature of shared leadership and its characteristic of mutual leader-follower relationship will lead team members to engage in power struggle behaviors, such as cliquing, impression management and slandering, in order to compete for important resources within the team.  Abfalter (2013) R T For creative expert teams, shared leadership (or selfleadership) is not a panacea since it shows a clear desire for leadership. By contrast, transformational leadership, which provides direction and autonomy, appears to be a better fit for these types of teams Boies et al. (2010) R T Shared passive avoidant leadership is negatively related to team trust, team potency and team performance Evans et al. (2021) R I When peer engagement is low and employees need to take interpersonal responsibilities, shared leadership will damage individual enjoyment and performance due to the need for frequent interpersonal interaction Fletcher and K€ aufer (2003) I L Shared leadership makes the formal team leaders face paradoxes or dilemmas in which they need to find a balance between self and the group Han et al. (2021) R T While relationship-oriented shared leadership has a positive effect on team performance through the mediating role of team positive psychological capital, task-oriented shared leadership is negatively related to team performance Hao (2016) I, R T Shared leadership has a negative impact on team performance when (1) leadership is unevenly distributed among team members; (2) the level of team monitoring is low; and (3) the level of team members' cooperative orientation is low Ji (2018) I M, T Shared leadership will stimulate the power struggle behaviors of team members, thus inhibiting team performance; the negative impact of shared leadership on power struggle will be mitigated under the condition of high-level of team functional background diversity and intensified when the team's formal leader exhibited high levels of transactional leadership style Kaur (2013) H T Due to the lack of designated hierarchical leaders, it is difficult for teams to reach consensus during the decisionmaking process. Therefore, shared leadership is not beneficial to the convergence stage of team innovation. In addition, the high team cohesion under shared leadership conditions will lead to the tendency of group thinking Li (2019) R T High levels of leadership role-enactment dispersion will inhibit team performance, and the negative effect will be intensified under high task complexity conditions Liu (2017) R M, T There is an inverted U-shaped relationship between shared leadership and employee's challenge-oriented organizational citizenship behavior. When leadership is over-shared, it will reduce employees' perception of psychological empowerment, thereby inhibiting their motivation to implement challenge-oriented organizational citizenship behaviors, which will damage the overall team performance Nordb€ ack and Espinosa (2019) R T Shared leadership will have a negative influence on team performance when leadership is highly shared within the group and uncoordinated Pasarakonda et al.
H T Shared leadership is not beneficial when task complexity is manifested as surgical difficulty, but exerts positive influences when it is manifested as unexpected events (continued ) First, dynamic is embodied in the process in which leadership roles and influences emerge dynamically among team members, which is contrary to the hierarchical functionalism that emphasizes "stability is a necessary prerequisite for hierarchical functions". As a result, this lack of stability "hierarchical structure" easily triggers power struggle behaviors among team members (Greer et al., 2018). Specifically, for the emerging leaders, in order to consolidate their power and status, they regard the decentralization behavior of team members as a potential threat, and try to deal with it by means of punishment, status discrimination or refusal of team members' constructive advice. While for team members, they tend to choose more competitive behaviors, such as adopting a hostile attitude toward the emerging leaders and disobeying the emerging leaders' instructions in order to improve their status (Ji, 2018). As a result, confrontations between the emerging leaders and team members emerge. In addition, the emergence and coexistence of multiple informal leaders under shared leadership conditions will also lead to power struggle behaviors. Since leadership roles and influences are shared within the team, team members may perform both the roles of team leaders and followers in a given time. According to the social identity theory, individuals are more inclined to define their own identity as team leaders for the purpose of maintaining or improving their self-esteem or self-concept (Tajfel and Turner, 1986). However, when all of these individuals define their own identities as team leaders, they will inevitably fall into disputes over leadership power and status (Kilduff et al., 2016;Zhu et al., 2018). The existence of the above problems will intensify the contradictions among team members, thereby inhibiting the operation and development of the team.
On this basis, the study by Ji (2018) pointed out that the negative impact of shared leadership on power struggles will be mitigated under the condition of a high-level of team functional background diversity and intensified when the team's formal leader exhibited high levels of transactional leadership.
4.1.2 Role stress. A second voice supporting the potential dark side of shared leadership argued that the wide sharing of leadership functions and responsibilities will cause role stress Dark side of shared leadership on team members (Evans et al., 2021;Wang and Peng, 2022). Role stress is a general term for the psychological pressure caused by the inability of an individual to perform multiple roles, which is specifically manifested as role ambiguity, role conflict and role overload.
Role ambiguity refers to the phenomenon that employees lack a clear understanding of their job responsibilities or expectations. It is proposed that employees under shared leadership conditions are more likely to experience role ambiguity, as the informal leader and follower roles assumed by these employees place completely different role expectations on them, which may lead to a lack of clear understandings of their role expectations when facing a specific task.
Role conflict refers to the phenomenon of psychological contradictions and conflicts caused by incompatible role expectations when individuals assume multiple roles. As mentioned above, employees under shared leadership conditions may assume both the roles of leader and follower at a given time, and need to switch between these two roles dynamically according to task requirements. However, the great differences between the roles of leader and follower in terms of working ideas, task focus and responsibility scopes make employees face great difficulties in the process of switching between these two roles, thus resulting in a feeling of role conflict.
Role overload is a kind of role stress caused by the lack of necessary skills or time to realize the role expectation when an individual is faced with multiple roles. It is proposed that employees under shared leadership conditions need to assume the responsibilities of both leader and follower, and are placed under more role expectations, which will bring greater work pressure to them and lead to the perception of high role load (Wang and Peng, 2022). In addition, the empirical study by Evans et al. (2021) indicated that when peer engagement was low and employees needed to take interpersonal responsibilities, shared leadership will damage individual enjoyment and performance due to role stress (i.e. the need for frequent interpersonal interactions). To sum up, shared leadership may be an important factor that causes role stress of employees.

Knowledge hiding.
Although studies have shown that shared leadership helps to stimulate the knowledge sharing behaviors of team members (e.g. Han et al., 2018), there is no lack of doubts. For example, the empirical study by Zhao (2013) has pointed out that leadership power is highly unstable in the context of shared leadership, which makes it possible for team members to emerge as the informal leaders of the team. Out of the desire for leadership status and power, team members will choose to hide their unique information and knowledge in order to maintain an advantage in the competition. Therefore, contrary to the view that shared leadership inspires knowledge sharing behaviors of team members, empirical evidence suggests that shared leadership may also lead to employees' knowledge hiding behaviors.

The negative effects of shared leadership on formal team leaders
As a powerful supplement to the traditional vertical leadership, shared leadership can effectively alleviate the work pressure of formal team leaders by sharing leadership functions and responsibilities among team member (Wood and Fields, 2007). But at the same time, studies have pointed out that shared leadership will also bring a series of negative effects to the formal team leaders, such as the loss of psychological territory, the decline of leadership motivation and the dualistic paradox of self and group (Fletcher and K€ aufer, 2003;Zhu et al., 2018).
4.2.1 Psychological territorial loss and leadership motivation decline. Under shared leadership conditions, a large number of employees emerge as informal leaders of the team, and play a pivotal role in the process of planning, organizing, leading and controlling. In contrast, the power and influence of the team's formal leader are greatly reduced, and sometimes they even need to obey the leadership of their subordinates, which creates a sense of psychological territory being violated (Zhu et al., 2018).
At the same time, the weakening of the power and influence of the team's formal leader under shared leadership conditions will make them realize that taking the role of formal team leaders cannot provide them with the expected benefits. As a result, the motivation and selfefficacy of formal team leaders to exert leadership are weakened, which slows down or inhibits the further improvement of their leadership ability (Zhu et al., 2018).
4.2.2 The dualistic paradox of self and group. In addition to the above two negative effects, research has also pointed out that formal team leaders may have to face a number of paradoxes and contradictions under shared leadership conditions (Fletcher and K€ aufer, 2003). For example, shared leadership requires the creation of a flatter and more adaptable team structure to facilitate team processes like learning or coordination, which often requires deep corporate cultural change and strong leadership. In this regard, organizations tend to turn to "hero CEOs" who, paradoxically, are asked to share their power and influence under shared leadership conditions. This seems to create a paradox in which the formal team leaders need to be different and above the group, but at the same time interact as an integral part of the group, even equal to the other members.
In addition, although team efforts such as teamwork or collective learning are essentially the key to team success, post-mortem reviews of team success tend to attribute it to the Dark side of shared leadership triumph of individual leadership. Meanwhile, influenced by the traditional notions of heroic individualism, those behaviors that seem mundane and routine but are critical to team success (e.g. share information and offer help) are often considered personality traits rather than leadership skills. As a result, this presents a new paradox for formal team leaders in which mundane but effective behaviors do not make them stand out. As Heifitz and Laurie (1999) pointed out in their study, despite the need for new leadership practices, "managers and leaders rarely receive promotions for providing the leadership required to do adaptive work and enhance organizational learning" (p. 65).
These are the paradoxes and contradictions that formal team leaders need to face under shared leadership conditions. If they cannot be properly solved, it will not only lead to the loss of individual leadership, but also make the team itself doubt its shared leadership principle (Fletcher and K€ aufer, 2003). Therefore, how to effectively deal with the above paradoxes is a major challenge that formal team leaders need to overcome.

The negative effects of shared leadership on work teams
In addition to discussing the negative impacts of shared leadership on team members and formal team leaders, existing studies have also explored the potential negative impacts of shared leadership on the overall work teams. In general, its negative effects are manifested as team performance inhibition, low decision-making efficiency, team responsibility dispersion and team creativity decline (e.g. Kaur, 2013;Nordb€ ack and Espinosa, 2019;Pearce et al., 2007;Zhu et al., 2018).
4.3.1 Team performance inhibition. Although the positive relationship between shared leadership and team performance has been confirmed in numerous empirical studies and meta-analytic reviews (e.g. Chen et al., 2022;Hoch and Kozlowski, 2014;Klasmeier and Rowold, 2020), research evidences suggest that shared leadership may also be harmful to team performance in certain situations.
First, the negative consequences such as power struggles and knowledge hiding caused by shared leadership are not conducive to achieving good task performance (Ji, 2018;Zhao, 2013). On this basis, some studies have further explored the boundary conditions of the impact of shared leadership on team performance. For example, through a survey of 30 surgical teams, the study by Pasarakonda et al. (2021) suggested that shared leadership exhibited diametrically opposite effects on team performance under different task complexity conditions. Specifically, shared leadership was not beneficial when task complexity was manifested as surgical difficulty, but exerted positive influences when it was manifested as unexpected events. In addition, the empirical study by Hao (2016) indicated that shared leadership had a negative impact on team performance when (1) leadership was unevenly distributed among team members, (2) the level of team monitoring was low and (3) the level of team members' cooperative orientation was low.
Second, some empirical studies have pointed out that the impact of shared leaders on team performance depends on the specific leadership styles adopted by employees in playing leadership roles. Boies et al. (2010), for example, suggested that shared leadership had a negative impact on team performance when employees shared passive avoidant leadership styles. In addition, the study by Han et al. (2021) indicated that although shared leadership had an overall positive impact on team performance, task-oriented shared leadership often showed a negative impact on team performance due to low decision-making efficiency.
Third, from the perspective of leadership role configuration, studies indicated that both the leadership sharedness (the degree to which the leadership role is shared by team members) and the leadership role-enactment dispersion (the degree of overlap between different leadership roles) will have an impact on its final effect. For example, the empirical study by Liu (2017) suggested that there was an inverted U-shaped relationship between shared leadership and employee's challenge-oriented organizational citizenship behavior. That is, when leadership is over-shared, it will reduce employees' perception of psychological empowerment, thereby inhibiting their motivation to implement challenge-oriented organizational citizenship behaviors, which will damage the overall team performance. Similarly, Nordb€ ack and Espinosa (2019) found that shared leadership had a negative impact on team performance when leadership was highly shared within the group and uncoordinated. In addition, the leadership role-enactment dispersion was also proposed to be a critical factor that affects team performance. Specifically, Li (2019) found that the lower the degree of overlap between the operation modes of different leadership roles, the more significant the negative impact of shared leadership on team performance, and the negative impact would be intensified with the increase of task complexity.
4.3.2 Low decision-making efficiency. The second negative effect scholars think that shared leadership has on work teams is that it reduces team decision-making efficiency. Specifically, studies have pointed out that compared with the traditional vertical leadership structure, the shared leadership structure lacks clear hierarchy attribute. This makes it difficult for teams to reach a consensus during the decision-making process, and usually consumes more time, resulting in low decision-making efficiency, which makes enterprises with high environment dynamics face more challenges (Kaur, 2013;Zhu et al., 2018).
4.3.3 Team responsibility dispersion. Shared leadership may also lead to the dispersion of team responsibilities, which breeds negative phenomena such as free riding and social loafing, especially in larger teams (Zhu et al., 2018). In addition, in a debate with Pearce et al. (2007) about whether the horizontal leadership model is more advantageous than the hierarchical leadership model, Locke pointed out that the advantage of the hierarchical leadership model is that it has a clear hierarchical leader to make the final decision and be responsible for the corresponding results, which effectively eliminates the emergence of team chaos and anarchy; while the horizontal leadership model (i.e. shared leadership) shows great defects in these aspects.
4.3.4 Team creativity decline. Studies have also indicated that shared leadership causes decline in creativity in a work group. For example, Kaur (2013) suggested that the low decision-making efficiency caused by shared leadership will have a negative impact on the convergence stage of team innovation. In addition, the high team cohesion under shared leadership conditions will lead to the tendency of group thinking, especially in teams with low cognitive diversity (Zhu et al., 2018), which will damage team creativity (Kaur, 2013). Similarly, the empirical study by Wu and Kathryn (2016) suggested that the strength of shared leadership networks (the frequency of contacts among team members) had an inverted U-shaped effect on team creativity. Excessive communication within the network will lead to homogenize views, which is not conducive to promoting team creativity. Abfalter (2013), from the perspective of creative expert teams, pointed out that shared leadership was not a panacea since it showed a clear desire for leadership. By contrast, transformational leadership, which provided direction and autonomy, appeared to be a better fit for these types of teams.

Summary
Based on the above research findings, it can be seen that while shared leadership brings positive influences to the team members, formal team leaders and the overall work teams, and it also has negative effects on them. On the basis of summarizing and sorting out existing research findings, this paper forms an analytical framework of the dark side of shared leadership as shown in Figure 2. As can be seen from Figure 2, from different analytical perspectives (i.e. individual motivation perspective, hierarchical functionalism perspective and leadership role configuration perspective), shared leadership may have a series of Dark side of shared leadership negative impacts on team members, formal team leaders and the overall work teams. Meanwhile, contextual factors (e.g. task complexity, team monitoring level) play a key role in stimulating the negative effects of shared leadership. Therefore, we suggest that the academic circle should take a more comprehensive and balanced view to treat this leadership model.

Directions for future research
Although only a small amount of literature has explored the negative effects of shared leadership, these studies have raised many interesting and important research issues while revealing the potential dark sides of shared leadership, which will be discussed in detail below.
First, future research should further explore the boundary conditions under which shared leadership functions. Through this literature review, we can find that shared leadership has both positive and negative effects on team members, formal team leaders and the overall work teams. Meanwhile, shared leadership may exert diametrically opposite effects in different organizational contexts. In this regard, it is particularly important to identify the boundary conditions under which shared leadership works. Specifically, future research can further explore in which organizational contexts or personnel characteristics, shared leadership is more likely to have negative impacts. For example, whether shared leadership is more likely to lead to negative consequences such as power struggle, knowledge hiding and trust crises when the organization is operated in collectivistic cultures rather than individualistic cultures, or when team members are competitive oriented rather than cooperative oriented. In addition, future research can also explore the corresponding boundary conditions from the perspective of inhibiting the negative effects of shared leadership, such as exploring whether leadership skills training can effectively alleviate the negative impact of shared leadership on team members' role stress, or whether team supervision can effectively reduce the negative impact of shared leadership on power struggle within the team. A good understanding of the above boundary conditions helps practice managers effectively avoid or restrain the negative effects brought by the implementation of shared leadership model, so as to better improve team members' satisfaction and team effectiveness.
Second, future research should further subdivide shared leadership when exploring its impact. The findings of this study suggest that sharing negative leadership styles (i.e. share passive avoidant leadership or task-oriented shared leadership) is not only not conducive to the improvement of team performance, but can even have a destructive impact on it. Similarly, the meta-analysis of Wang et al. (2014) indicated that shared traditional forms leadership (e.g. transactional and participative leadership) showed a lower correlation with team effectiveness than shared new-genre leadership (e.g. transformational and authentic leadership). Therefore, in addition to focusing on the strengths and weaknesses of the overall shared leadership structure, future research should further subdivide it and explore the impact when specific leadership style was shared by team members. For example, when abusive supervision is shared among team members, will it exacerbate the relationship conflict between team members and thus have a destructive impact on team performance. In addition, future research can also explore whether shared leadership can better improve team performance when emerging team leaders exhibited multiple leadership styles, that is, sharing diverse leadership styles, rather than sharing a single leadership style. It is worth mentioning that the exploration of the above questions helps to improve the development of shared leadership literature, and provide theoretical guidance for practice managers to better enhance team performance.
Third, future research can further explore the impact of different shared leadership role configurations on individual and team-level outcomes. Notably, the findings of this study indicate that both excessive sharing of leadership and excessive leadership role-enactment dispersion will bring negative influences on employees and work teams. In this regard, future research can explore under which role configuration shared leadership will play the greatest role. For example, based on the social network measurement approach, the role configuration of shared leadership can be further divided into four types, including (1) high density and high decentralization, (2) high density and low decentralization, (3) low density and high decentralization and (4) low density and low decentralization. On this basis, future research can further explore which of the above-mentioned leadership configurations is most conducive to improving team performance, and whether this optimal role configuration varies with the change of team type (e.g. top management teams vs. worker teams), team size, task characteristics (e.g. task complexity, task interdependence) and other contextual factors. The answers to the above questions are also helpful to promote the development of shared leadership theory, and provide theoretical guidance for practice managers to better construct the leadership role configuration to improve team effectiveness.

Conclusion
With the deepening of the shared leadership research, more and more studies have pointed out that shared leadership can also be harmful while bringing positive effects to employees and work teams. Therefore, in order to help the academic community better understand the potential dark sides of shared leadership, so as to form a more comprehensive and balanced view of its effects, we conducted a review study that focused on the negative impacts of shared leadership. Through a systematic literature review, we found that the existing studies mainly explored the negative impacts of shared leadership on team members, formal team leaders and the overall work teams from the perspectives of individual motivation, hierarchical functionalism and leadership role configuration. Meanwhile, contextual factors play a key role in determining the effects of shared leadership. On this basis, this study constructed an analytic framework of the dark sides of shared leadership and identified several directions for future research. Zhu, J.L., Liao, Z.Y., Yam, K.C. and Johnson, R.E. (2018)