Management Innovation and Middle Managers: The Role of Empowering Leadership, Voice, and Collectivist Orientation

ABSTRACT Exposing under which conditions management innovation diffuses within firms, this study investigates at the individual level the mediating influence of middle managers’ voice behavior on the relationship between CEOs’ empowering leadership behavior and perceived management innovation. We also propose that the magnitude of this relationship depends on middle managers’ collectivist orientation. This study exploits a unique Moroccan sample of small and medium-sized enterprises (SMEs) and utilizes structural equation modeling to analyze the studied relations. We found that the positive relationship between CEOs’ empowering leadership behavior and management innovation is mediated by middle managers’ voice. This effect is conditioned by middle managers’ collectivist orientations, which positively influence their attention to CEOs’ signals and the value and frequency of their contributions to management innovation. While research has identified the external and organizational factors that shape management innovation, our study concentrates on the individual level and accentuates that middle managers’ closeness to management processes, combined with their access to technical knowledge, renders them essential to management innovation. We contradict arguments that middle managers may be less inclined to help management innovation to emerge. SMEs can systematically invest in management innovation by advancing their managerial capabilities and considering individual value orientations. 摘要 为了揭示管理创新在企业内部扩散的条件，本文考察了中层管理者建言行为对CEO授权与其感知到的企业管理创新之间的中介作用，并提出中层管理者的集体主义导向会调节这三者之间的关系。作者在摩洛哥收集了独特的中小企业样本，并采用结构方程模型进行分析，发现CEO授权型领导行为使得中层管理者更敢于建言，公司的管理创新因此越多。同时，中层管理者的集体主义导向影响他们积极关注CEO发出的信号，及他们在管理创新上所作贡献的价值和频率。虽然过往研究已经发现了一些塑造管理创新的外部因素和组织因素，但本研究聚焦于个体层面，强调中层管理者与管理过程的密切关系及其所掌握的技术知识对管理创新的重要作用。本文反驳了中层管理者可能不太愿意推进管理创新的观点，并启示中小企业可以通过投资中层管理者的能力并考虑个体的价值取向来系统地开展管理创新。

INTRODUCTION Management innovation appears in the workplace as new management-related practices, processes, or structures that intend to further organizational functioning (Birkinshaw, Hamel, & Mol, 2008;Vaccaro, Jansen, Van den Bosch, & Volberda, 2012). This is crucial for small and medium-sized enterprises (SMEs) because they compensate their lack of technological capabilities and engagement in R&D activities by not only relying on the market to drive change, but also by engaging in new marketing, design, or organizational practices (Hervas-Oliver, Ripoll-Sempere, & Moll, 2016). Moreover, management innovation helps craft a sustainable competitive advantage, as it is more difficult to replicate. It creates a web of individual innovations spanning various management processes and practices (Hamel, 2006). Although the definitions of management, administrative, and organizational innovation overlap (Damanpour & Aravind, 2012), research on specifically management innovation is scarce, albeit growing (e.g., Frynas, Mol, & Mellahi, 2018;Guzman & Espejo, 2019;Vaccaro et al., 2012).
The current study extends our understanding of the conditions under which management innovation diffuses within SMEs in two ways. First, it adds to innovation and leadership literature by analyzing the role of direct followers as internal agents of change who leverage management innovation. Research has demonstrated the important role of context in shaping management innovation, and 'leadership may eventually be the single most important contextual factor making management innovation work' (Frynas et al., 2018: 87). By creating and sustaining an organizational climate and culture that nurtures creative efforts and facilitates the diffusion of learning, leaders can significantly boost organizational creativity and innovation (Jung, Chow, & Wu, 2003). Similarly, leaders can communicate a shared vision and mobilize the trust and respect that followers have toward leaders. In so doing, this gives, in turn, weight to management innovation (Birkinshaw et al., 2008). However, little is known about the role of direct followers as change agents and their behaviors regarding the link between leadership behavior and management innovation (Chang, 2016). In this study, we focus on the role of middle managers, who are crucial to understanding why and how management innovation comes about (Birkinshaw et al., 2008). Middle managers are commonly considered essential to innovation because of their ability to make strategic innovation accessible to top executives and create novel knowledge in the innovation process. They have ample room to shape management innovation, as CEOs' intentions can only be a broad guideline rather than a detailed blueprint (Heyden, Sidhu, & Volberda, 2018). Specifically, management innovation centers on the changes in what managers do on a daily basis and how they do it as part of their job (Hamel, 2006;Vaccaro et al., 2012). Because middle managers are the intermediaries between organization leaders and employees, they can provide valuable information and suggestions that aid leaders in making better decisions.
Researchers have also previously emphasized that middle managers' strategic agency in driving organizational objectives (Woolridge, Schmid, & Floyd, 2008) and their access to multiple technical knowledge sources, along with their managerial expertise, can stimulate innovation (Delmestri & Walgenbach, 2005). Hence, middle managers' voice, through which ideas, concerns, and suggestions about work-related issues are shared, constitutes an essential source for the creation and implementation of management innovation (Su & Baird, 2018). However, questions persist on whether middle managers' involvement has a significant influence on firm innovation. Flat management theory proposes that middle managers bloat the management structure and lengthen the principal/agent chain, thus reducing overall organizational efficiency. Management innovation also interrupts the traditional managerial capabilities around which middle managers have developed their expertise and built their professional identity (Heyden et al., 2018). In addition, middle managers can delay, reduce the quality of, or even sabotage the changes advocated by top managers (Woolridge et al., 2008).
Given this context, the purpose of this study is to investigate, at the individual level, the mediating influence of middle managers' voice behavior on the relationship between CEOs' empowering leadership behavior and CEOs' perceived management innovation level. We turn to CEOs' empowering leadership behavior because it has been shown to be pertinent to innovation endeavors (see the meta-analytic reviews, Lee et al., 2020;Lee, Willis, & Tian, 2018), including management innovation (Hassi, 2019). Empowering leadership behavior entails emphasizing a sense of meaning in employees' work, favoring participation in decision making, eliminating bureaucratic constraints, and boosting confidence in individual performance (Ahearne, Mathieu, & Rapp, 2005).
Our second objective is to add to contingency-based innovation literature. While extant studies have highlighted how organizational, competitive, institutional, and technological contexts can act as levers for management innovation (Birkinshaw et al., 2008;Frynas et al., 2018), they have placed less emphasis on the characteristics of internal change agents. We propose that the magnitude of the relationship between empowering leadership behavior and management innovation via voice depends on middle managers' collectivist orientation. We base this proposition on signaling theory which sees leadership as an active signaling process (Grabo, Spisak, & van Vugt, 2017). We infer that empowering leadership behavior is positively related to middle managers' voice behavior, thereby promoting management innovation. Leaders signal that middle managers' active involvement in work-related issues is expected. In our study, such involvement is shown via voice, which provides opportunities for middle managers to have a say about work-related aspects.
Signaling theory also acknowledges that recipients differ in their attention to signals, and researchers have stated that more consideration could be devoted to the effect of followers' characteristics instead of leaders' traits (Sharma & Kirkman, 2015). Thus, we propose that middle managers' collectivist orientations may act as a boundary condition on the signaling effects of the relationship between empowering leadership behavior and voice behavior, in turn influencing management innovation. Collectivism is relevant in the leadership process, which involves managing relationships inside the organization and enhancing team solidarity (Yang, Zhang, & Tsui, 2010) because it captures the relative weight people give to personal interests and shared pursuits (Thomas, Au, & Ravlin, 2003;Wagner, 1995). Figure 1 shows the hypothesized relationships based on a moderated mediation framework tested on a unique Moroccan sample of SMEs.

Management Innovation, Empowering Leadership Behavior, and Middle Managers' Voice
The role of internal change agents is crucial to understanding how management innovation emerges (Birkinshaw et al., 2008;Hamel, 2006), particularly in leadership style (Frynas et al., 2018;Vaccaro et al., 2012). In the current study, we focus on empowering leadership. This leadership form relates to the concept of empowerment, which can be defined as the creation of a conducive environment that enhances individual's self-efficacy and control while eliminating the factors that uphold feelings of powerlessness (Arnold, Arad, Rhoades, & Drasgow, 2000). Empowered followers take a more proactive approach to shaping and influencing their work environments (see Sharma and Kirkman (2015) for a review), as they benefit from empowering leadership behavior through the development of self-determination and intrinsic motivation (Zhang & Bartol, 2010). Furthermore, in complex organizations, leaders who focus on follower empowerment and self-management by coordinating and coaching followers rather than instructing them achieve far greater fitness and creativity than those in top-down, leader-controlled organization hierarchies (Marion & Uhl-Bien, 2001).
Turning to empirical evidence, Lee et al. (2020) found in their meta-analytic review that empowering leadership was among the leadership styles with the strongest correlations with individual innovation. They tentatively suggested that 'employees are better able to innovate […] when their leaders become less "leader-like" in the traditional sense. That is, when leaders act as facilitators and support and empower employees' (Lee et al., 2020: 15; see also Lee et al., 2018). In the hotel sector, Hassi (2019) revealed that empowering leadership exerts a positive effect on management innovation. However, literature that explains innovative adoption decisions from the contingency perspective is more limited. Research has focused on the influence of organization contextual factors (Hassi, 2019;Mol & Birkinshaw, 2009), leadership behavior, and control mechanisms (Su & Baird, 2018;Vaccaro et al., 2012) on management innovation. To augment our understanding of how management innovation emerges and considering the potentially important role of middle managers in this process, we focus on middle managers' voice behavior as a mediating variable between empowering leadership behavior and management innovation.
Scholars define voice behavior as the voluntary communication of ideas, concerns, and suggestions about work-related issues, stressing the expression of a constructive challenge intended to improve rather than simply criticize (Morrison, 2011;Van Dyne & LePine, 1998). Voice is uniquely focused on verbal expressions (directed up, down, or horizontally) explicitly intended to bring about positive change and make improvements in the organization rather than benefit just the individual (Morrison, 2011). It tends to improve management systems through increased understanding and information and employee ideas for development (Dundon, Wilkinson, Marchington, & Ackers, 2004), which helps organizations innovate (Guzman & Espejo, 2019;Morrison, 2011). However, voice's positive influence on innovation may differ, that is, be greater for nontechnological innovations, including organizational and process innovations (Andries & Czarnitzki, 2014;Uhlaner, van Stel, Duplat, & Zhou, 2013).
Middle managers' voice cannot be coerced or readily integrated into job's in-role requirements (Detert & Burris, 2007;Morrison, 2011). It constitutes a proactive form of extra-role behavior not considered part of the job's requirements. It arises from individuals' desires to make a difference and challenge the status quo by improving organizational issues because they want to, not because of pressure or fear (Elsetouhi, Hamad, Nagm, & Elbaz, 2018). Middle managers' voice is pertinent to management innovation because it changes the way in which management sets goals, makes decisions, aligns activities, and mobilizes employees (Hamel, 2006). Additionally, middle managers provide information and suggestions that top management may not otherwise have been aware of (Morrison, 2011). However, extra-role activities can be draining, and those who go the extra mile carry the risk of being scapegoated for failures (Glaser, Stam, & Takeychi, 2016;Heyden et al., 2018). Hence, middle managers must be motivated and are likely to pay particular attention to CEOs' cues when deciding whether to engage in voice behavior (Morrison, 2011;Xu, Qin, Dust, & DiRenzo, 2019). Such cues from CEO behavioral preferences can be derived from CEO leadership as an active signaling process that overcomes the information asymmetries between middle managers and CEOs (Spence, 1973). Indeed, middle managers observe CEOs, as they have legitimate power that originates from the authority lines within organizational hierarchies when deciding on the implementation of middle managers' suggestions (Bandura, 1986). CEOs also have reward power because they oversee middle managers' job assignments, remuneration, and career opportunities (Pearce & Robinson, 1987), as well as control whether they will be compensated or penalized for voicing suggestions.
When subordinates perceive that their leaders listen to them, value their ideas, and occasionally act on their ideas (Detert & Burris, 2007), they gain a sense of direct access to the leaders, which entices their voice behavior (e.g., Ashford, Sutcliffe, & Christianson, 2009). This resembles the participation dimension of empowering leadership behavior, according to which leaders share power and actively stimulate follower input and involvement (Ahearne et al., 2005). Empowering leadership requires leaders to develop trust in their followers by delegating high levels of discretion and decision-making authority (Lorinkova, Pearsall, & Sims, 2013). Empowering leaders convey confidence in their followers' performance and fairly consider the ideas presented, helping endorse a strong feeling of follower competence. Their followers are thereby more likely to focus on tasks rather than worry and hesitate. Thus, they become willing to take risks to express their ideas (Gao & Jiang, 2019). Middle managers, who are given a chance to voice their opinions and concerns, also have a heightened sense of control over their work, which facilitates higher levels of satisfaction and motivation (see Morrison, 2011).
Taken together, empowering leadership behavior can affect organizations' overall innovation, and depending on the signaling behavior of their leaders, middle managers may decide to make suggestions for organizational improvement (Dedahanov, Lee, Rhee, & Yoon, 2016). Meanwhile, if a leader's signal to act on middle managers' voice is absent, middle managers may stay silent (Donovan, O'Sullivan, Doyle, & Garvey, 2016). Hence, it is plausible to assume that the effect between empowering leadership behavior and management innovation is transmitted through middle managers' voice behavior. We therefore argue that voice behavior plays a mediating role in the relationship between empowering leadership behavior and management innovation and propose the following: Hypothesis 1: The positive relationship between CEOs' empowering leadership behavior and management innovation is mediated by middle managers' voice behavior.

The Moderating Role of Middle Managers' Collectivist Orientation
Management research has demonstrated that signaling effectiveness is partly determined by receiver characteristics, in our case by middle managers (Connelly, Certo, Ireland, & Reutzel, 2011;Xu et al., 2019). Among these characteristics, the receiver's attention, that is, whether the middle manager is looking for signals from leadership behavior, has been identified as a key attribute according to which signaling effectiveness differs (Connelly et al., 2011). Following this logic, we focus on middle managers' characteristics. Researchers have referred to receivers' value orientations in the signaling process as well (Li & Sun, 2015;Xu et al., 2019), including the collectivist orientation (Yang et al., 2010). We thus propose that middle managers' collectivist orientation as a specific receiver characteristic influences the indirect positive effects of CEOs' empowering leadership behavior on management innovation via middle managers' voice behavior.
Collectivism refers to the tendency to view oneself as interdependent with select others, be concerned about the consequences of behavior for the in-group's goals, and be more willing to sacrifice personal interests for the group's welfare (Thomas et al., 2003;Wagner, 1995). In general, collectivism is a characteristic of interdependence, group embeddedness, and personalized relationships (Triandis, Bontempo, Villareal, Asai, & Lucca, 1988). Interdependent individuals see themselves as part of an encompassing social relationship and recognize that their behavior is determined, contingent on, and to a large extent organized by relationships (Markus & Kitayama, 1991). Therefore, collectivism is relevant to the leadership process, which involves managing relationships inside the firm and enhancing follower solidarity (Schaubroeck, Lam, & Cha, 2007;Yang et al., 2010).
According to the above arguments, empowering leadership behavior's effect on voice is likely to be moderated by a collectivist orientation. Middle managers with high collectivist orientation might be more attentive to signals for proactive voice behavior, as such behavior would allow them to bring about positive change and make organizational improvements (Morrison, 2011), thereby fostering shared pursuits rather than personal interests. In contrast, individuals with low collectivist orientation are less enticed to seek achievements in shared pursuits and hence lack the attention to signals that enables this interest. In view of this, we propose that middle managers with high collectivist values have a greater tendency to display behavior for collaborative work, such as voice, than individuals with a low collectivist orientation.
We also suggest that collectivism moderates the relationship between middle managers' voice behavior and management innovation. Successfully implemented management innovation can facilitate the achievement of common organizational goals (Birkinshaw et al., 2008). Middle managers with high collectivist orientations are more interdependent with their community and concerned about the consequences of their behavior regarding the group's goals (Thomas et al., 2003), even if it means sacrificing personal interests (Wagner, 1995). They follow a cognitive and calculative process when engaging in voice to decide whether it is worthwhile to speak up vis-à-vis remaining silent. This choice depends on individual characteristics (Guzman & Espejo, 2019;Morrison, 2011); therefore, middle managers do not necessarily display voice at the same frequency (Van Puyenbroeck, Stouten, Hofmans, Van Meervelt, & Vande Broek, 2020). We propose that individuals concerned with the effects of their conduct on common organizational goals more frequently engage in voice behavior, thereby increasing the number of ideas for management innovation to assess and implement successfully (Axtell, Holman, Unsworth, Wall, & Waterson, 2000). Altogether, when contributing to management innovation as a shared pursuit, the voice behavior of middle managers with high collectivist orientations is expected to be of a higher value and frequency than that of middle managers with low collectivist orientations. In other words, we expect the positive effect between voice and management innovation to be stronger for middle managers with high collectivist orientations compared to those with low collectivist orientations. Thus, we propose the following hypothesis: Hypothesis 2: The indirect relationship between CEOs' empowering leadership behavior and management innovation through middle managers' voice is moderated by middle managers' collectivist orientation, such that the relationship is stronger under a high rather than low collectivist orientation.

Sample and Procedures
To test our hypotheses, we studied a database of Moroccan SMEs. SMEs constitute about 95% of Moroccan firms and play an important role in the country's employment and wealth creation (Hassi, 2016). According to the Global Innovation Index 2017, Morocco ranks in the bottom half of countries in terms of its innovative orientation and output (Dutta, Lanvin, & Wunsch-Vincent, 2017). In particular, how conducive Moroccan firms are to innovation activity shows room for improvement, which could occur through better knowledge absorption and more knowledge workers (Dutta et al., 2017). Building a firm's capabilities by promoting managerial capacity and positive behaviors among organization members is considered a challenge that management research can address so companies can seize opportunities to grow (George, Corbishley, Khayesi, Haas, & Tihanyi, 2016).
To create a sufficiently large database, we called the Moroccan firms listed in a governmental dataset. The companies were screened to ensure they had the required level of technological innovation, as management innovation is needed to foster or reinforce this type of innovation (Hervas-Oliver et al., 2016;Mol & Birkinshaw, 2009). We asked whether they had created a new or substantially improved production process with new equipment or engineering over the last three years (OECD, 2005). With these management innovation parameters in place, 800 SMEs agreed to participate in our study. The data were collected between October 2017 and November 2018 using a drop-and-collect procedure to enhance the participation rate and efficiently match the completed questionnaires (described below) to each participating company.
We distributed two types of questionnaires: one for the top management tier and holding positions, such as owners, CEOs, director generals, and general managers (CEOs for short), and the other for middle managers (Appendix I). The latter were those who ranked lower than the top management team but higher than firstline workers, including professional and technical workers (Huy, 2001). In our study, middle managers were the responsible heads for the production department because, in this functional area, voice appeared relevant for innovation activity (Andries & Czarnitzki, 2014). The CEO questionnaire gathered information about the organizations' management innovation, while the middle manager questionnaire collected data about their CEOs' empowering leadership behavior, their voice behavior, and their collectivist orientation. We separated the evaluation of the current model's predictors and criterion variables in the questionnaire to reduce the common method variance (Podsakoff, MacKenzie, Lee, & Podsakoff, 2003). Following Kock (2015), we have conducted a PLS-SEM with the variables of the mediation model to calculate the variance inflation factor (VIF) resulting in VIF of 1.395 and, thus, below the threshold of 3.3. Therefore, there is no collinearity problem, and the model can be considered free of common method bias (Kock, 2015).
We obtained 474 completed sets of questionnaires from the 800 SMEs we contacted. From there, 19 sets were discarded because of missing data or incomplete information. A total of 455 sets of questionnaires were thus considered valid, representing a response rate of 56.9%. The participating firms operated in various industries, mostly supplying automotive companies (17%), IT and electronics (16%), food processing (17%), textiles and clothing (15%), building materials (17%), and furniture manufacturing (17%). Most firms had been operating for an average of 17.3 years (SD = 8.84) and were pursuing an innovation strategy (89%). Regarding size, the firms had an average of 42.8 full-time employees. The middle managers averaged at 37.6 years old (SD = 6.04), had an average tenure in their present position of 7 years (SD = 3.86), and were mainly male (68.7%).

Measures
All variables were assessed using established scales and measured on a 5-point Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree). We applied the construct measures with established multi-item scales from prior research with most of them having been applied in the Middle East and/or Islamic majority countries. The English versions of the scales were translated into French (the language used in the questionnaires) and then back-translated into English by two independent bilingual individuals to ensure their meaning was maintained (Brislin, 1980). Empowering leadership. We conceptualized and operationalized empowering leadership behavior as an individual perception. We considered leadership to have emerged when followers constructed and represented their views of their leaders in their thought systems (Meindl, 1995). We combined the four subscales of the empowering leadership behavior measure from Ahearne et al. (2005) into one overall empowering leadership behavior measure. These subscales are (1) enhancing the meaningfulness of work, (2) fostering participation in decision making, (3) expressing confidence in high performance, and (4) providing autonomy from bureaucratic constraints. Prior research using the same four subscales indicated not only support for a higher-order global construct (Cheong, Spain, Yammarino, & Yun, 2016) but also its applicability in the Moroccan context (Hassi, 2019).
Voice behavior. To measure middle managers' voice behavior, we used a 6-item scale developed and validated by Van Dyne and LePine (1998) with very minor modification. Since middle managers self-report, all items were changed to begin with 'I' as suggested by Van Dyne and LePine (1998). This construct has been widely used across African countries including Islamic majority countries and was found to show greater validity and reliability compared to other voice constructs (Fisher et al., 2019).

Management innovation.
We used the 6-item management innovation scale developed by Vaccaro et al. (2012) which had been applied previously in the Moroccan context (Hassi, 2019). Because management practices, processes, structures, and techniques are applied at the organization level and refer to alterations in the way middle managers' work is carried out (Chang, 2016;Volberda, Van den Bosch, & Heij, 2013), we operationalized management innovation at the individual level as CEO's perception of firm management innovation.
Collectivist orientation. To measure the middle managers' collectivist orientation, we used Dorfman and Howell's (1988) 6-item scale, an adaptation of Hofstede's (1980) original cultural measures to the individual level. This measure has been used in several past studies with acceptable internal consistency values, including in non-Western contexts (Clugston, Howell, & Dorfman, 2000;Dorfman & Howell, 1988;Yang et al., 2010).

RESULTS
We conducted structural equation modeling using the sem command in the Stata 16 software package (StataCorp, 2019). Following Kline (2010), we employed a maximum likelihood estimation method with robust standard errors together with the Satorra-Bentler rescaled chi-square statistic (Satorra & Bentler, 1994). This statistic compensates for non-normality and corrects for heteroscedasticity (i.e., the vce(sbentler) option in Stata). Furthermore, we used item parceling to better approximate the linear factor model assumption of multivariate normality (Bandalos & Finney, 2001;Kline, 2015). The items of the latent variable CEOs' empowering leadership behavior were parceled theoretically according to its measure's four subscales. The items of the three remaining latent variables (management innovation, middle managers' voice, and middle managers' collectivist orientation) were assigned based on similarities, with factor loadings, mean values, and standard deviations parceling two items to form three parcels for each latent variable. Additionally, we conducted the Durbin-Wu-Hausman chi-square test for endogeneity between middle managers' voice, management innovation, and individual-level variables (i.e., ivregress 2sls in Stata). The Durbin-Wu-Hausman test was conducted after an instrumental variables regression in which we instrumented middle managers' voice behavior by the middle managers' age (logarithm of age) and tenure (logarithm of years in the organization). A significant p-value indicated an endogenous mediator. Therefore, following Antonakis, Bendahan, Jacquart, and Lalive (2010), we correlated the disturbances of the dependent variable (management innovation) and the mediator (middle manager's voice behavior) in our subsequent analyses, omitting the instruments and control variables. Lastly, we found the Hansen J statistic to be nonsignificant, indicating that the model was not overidentified.

Descriptive Statistics
The model variables' descriptive statistics, correlations, and reliabilities (Cronbach's α) are provided in Table 1. The reported reliabilities of the four model variables showed satisfactory internal consistency (Cronbach's α: 0.65-0.87). The correlation results were also highly significant (p < 0.001). As expected, CEOs' empowering leadership behavior was positively related to middle managers' voice (r = 0.53) and management innovation (r = 0.53). Middle managers' voice was also positively related to management innovation (r = 0.33). Middle managers' collectivist orientation was positively related to CEOs' empowering leadership behavior (r = 0.31) and middle managers' voice (r = 0.44), but not management innovation (r = −0.01).

Confirmatory Factor Analyses
Before testing the two hypotheses, we examined the construct validity of the measures by performing four confirmatory factor analyses. The first model included four first-order factors derived from their respective item parcels (four parcels for CEOs' empowering leadership behavior and three parcels each for management innovation, middle managers' voice, and middle managers' collectivist orientation). In the following models, we merged the first-order factors (Table 2), but in the last model, we loaded all parcels onto one factor. The models were compared based on both relative and absolute goodness-of-fit indices (Acock, 2013). The χ 2 goodness-of-fit statistic and root mean square error of approximation (RMSEA) were the two absolute indices. The comparative fit index (CFI) and standardized root mean square residual (SRMR) were the two relative/standardized indices. Following Hu and Bentler (1999), good-fit models should have a nonsignificant χ 2 and an RMSEA less than 0.05 (between 0.05 and 0.08 for an acceptable fit). The values for the other indices should be >0.90-0.95 (CFI) and <0.08 (SRMR) for an acceptable fit. Table 2 shows that the hypothesized four-factor model of CEOs' empowering leadership behavior, middle managers' voice, management innovation, and middle managers' collectivist orientation exhibited a better fit than the other models (χ 2 = 183.360, df = 59, p = 0.00, RMSEA = 0.068, CFI = 0.940, SRMR = 0.054). These results showed that the current study's variables are different constructs and can be used in subsequent analyses.

Hypotheses Testing
Hypothesis 1 proposed that the positive relationship between CEOs' empowering leadership behavior and management innovation is mediated by middle managers' voice behavior. Considering the confounding variables by correlating the disturbances of the dependent and mediator variables, we found that the unstandardized indirect effect of middle managers' voice behavior was statistically significant (b = 0.104, SE = 0.034, 95% CI = 0.037; 0.171), thus supporting Hypothesis 1 (Table 3). In other words, greater CEO empowering leadership behavior was associated with greater middle manager voice behavior, which increases management innovation. The model further showed an acceptable fit (χ 2 = 91.081, df = 32, p = 0.00, RMSEA = 0.054, CFI = 0.975, SRMR = 0.045).
Hypothesis 2 predicted that the mediation relationship between CEOs' empowering leadership behavior, middle managers' voice, and management innovation is strengthened by middle managers' collectivist orientation. Following Preacher, Rucker, and Hayes (2007) and Hayes (2013), our moderated mediation model (e.g., conditional indirect effect model) examined whether the indirect effect of CEOs' empowering leadership behavior on management innovation through middle managers' voice behavior depends on middle managers' collectivist orientation. To test this moderated mediation, we used the Johnson-Neyman technique for probing interactions, which 'points along the continuum of the moderator where the effect of the focal predictor transitions between statistically significant and nonsignificant' (Hayes & Matthes, 2009: 928). With this technique, we ensured more specific information regarding the range of moderator values for which the predictor significantly affects the criterion. In support of Hypothesis 2, our results showed that mediation was moderated by middle managers' collectivist orientation, with positive relationships increasing as middle managers' collectivist orientation increased. More precisely, the indirect path between CEOs' empowering leadership behavior, middle managers' voice, and management innovation varied significantly for low (1 SD below average), average, and high middle managers' collectivist orientations (1 SD above average; Table 4). Therefore, the conditional indirect effect increases as the value of the moderator increases, meaning that the positive relationship between empowering leadership behavior, middle managers' voice behavior, and management innovation is strengthened by any middle manager collectivist orientation. The model also showed an acceptable fit: χ 2 = 136.336, df = 53, p = 0.00, RMSEA = 0.059, CFI = 0.958, SRMR = 0.045.

DISCUSSION
The current study extends innovation and leadership literature by providing empirical insights into the role of middle managers in the relationship between CEOs' empowering leadership behavior and management innovation. It is the first study to demonstrate that leadership is a process whereby the relationship between Table 3. Unstandardized coefficients, standard errors, and 95% confidence intervals for the mediation model (Satorra- empowering leadership behavior and a more distal outcomemanagement innovationis transmitted through a more proximal mediating variablemiddle managers' voice. While Frynas et al. (2018) identified leadership as probably the most important contextual variable, our findings indicate that direct followers (in our case, middle managers) are important change agents to leverage management innovation. The mediating effect of middle manager's voice is significant but small. This could be because of other intermediary influences that are, for instance, related to the dyadic relationship of empowering leadership (Zhang & Bartol, 2010) and the fact that management innovation is affected by trusting relationships within a company (Vaccaro et al., 2012). Nevertheless, considering SMEs' understudied organizational context (Hervas-Oliver et al., 2016), both leaders and middle managers are important for a more encompassed understanding of how management innovation comes about. Middle managers are essential because they can bring novel knowledge to the innovation process by proactively engaging in voice behavior. Furthermore, middle managers' closeness to management structures, processes, and practicesthe potential targets of innovation activitycombined with their access to multiple technical knowledge sources renders them crucial for management innovation to emerge. However, as leaders, CEOs are not less important because they can encourage extra-role behavior, which in our study is middle managers' voice.
CEOs' observed behavior acts as a signal to middle managers whether proactive voice behavior is desired or not. Hence, it is instrumental to mobilizing middle managers and encouraging their voice to achieve management innovation. This study also adds to contingency innovation leadership literature by demonstrating that the magnitude of the relationship between empowering leadership behavior and management innovation via voice varies per middle managers' collectivist orientations.

Theoretical Implications
By shining a light on middle managers as the 'voicers' rather than merely the target or active managers of voice mechanisms, the present study identifies the importance of middle managers as stimuli of management innovation. Conceptualizing middle managers' voice in innovation research helps accurately echo organizational reality, in turn providing actionable knowledge to support business leaders in their decisions regarding how to create management innovation. Given the study's empirical context, how to create management innovation is particularly relevant not only for the increasing number of expatriates from China, India, and Europe in Morocco when building local managerial capacities, but also for local entrepreneurs' activities (Walumbwa, Avolio, & Aryee, 2011). Hence, the findings concentrating on the role of individuals leveraging management innovation complement prior studies at the firm or subunit level that focus on internal contextual factors (Guzman & Espejo, 2019;Heyden et al., 2018;Su & Baird, 2018). Our study also provides a compelling empirical argument for research on the dynamics within leader-manager dyads that enable a firm's innovation capabilities (e.g., Heyden et al., 2018) to maximize actors' performance in their shared pursuit of management innovation.
In addition, our empirical findings contradict the argument that middle managers may be less inclined for management innovation to emerge (Heyden et al., 2018). Management innovation tends to arise out of organizational necessity (Birkinshaw & Mol, 2006), and middle managers may see management innovation as a welcome reason to upgrade their skills through self-study, experimentation with management innovation, and special training, which boosts their employability. The latter is an important facet in times of organizational de-layering and restructuring (Huy, 2001). Rather than obstructing it, middle managers may have a self-interest in fostering management innovation to maintain an attractive knowledge and skill profile for later employers.
Furthermore, the current study increases our scholarly and practical understanding of under which circumstances empowering leadership behavior is the best 'fit' for particular work settings, here based on middle managers' differences. Research has found the magnitude of the relationship between leadership and innovation to be largely variable (Hughes, Lee, Tian, Newman, & Legood, 2018), and our findings increase our understanding of the within-context variables that influence this relationship. Although research has mainly focused on leaders' cultural orientations as an antecedent to empowering leadership (Sharma & Kirkman, 2015), our findings indicate that followers' value orientations are important to the magnitude of empowering leadership behavior effectiveness. Our findings corresponding to collectivist orientation in Moroccan SMEs could be extended to contexts with similar underlying commonalities in cultural value orientations, such as Asian cultures (Walumbwa et al., 2011). It is also likely that other moderators exist, meaning empowering leadership researchers can contextualize specific followers' value orientations using our results. The religious and sociocultural norms related to Islam could be of heightened interest because they are specific to Northern Africa and foster a strong convergence in this region's cultural orientations, thus affecting leadership practice (Dorfman, Javidan, Hanges, Dastmalchian, & House, 2012).

Managerial Implications
Fostering management innovation is key for SMEs to gain a sustainable competitive advantage (Hervas-Oliver et al., 2016). The current study suggests that firms can consciously and systematically invest in management innovation by advancing managerial capabilities and middle managers' behaviors. CEOs may need to focus on middle managers' value orientations and strategically manage their managerial employees depending on the nature of their value systems. As such, CEOs' leadership skills become important and further accentuate the important role of the upper echelon's capabilities in SME innovation performance (Uhlaner et al., 2013).
Furthermore, organizations should encourage middle managers to actively engage in voice behavior to tap into their managerial and technical knowledge. Even for SMEs operating in the emerging Moroccan market, for which the data generally confirmed a lack of human capital in terms of knowledge and expertise (World Economic Forum, 2017), middle managers' voice behavior can work indirectly, allowing for empowering leadership behavior to positively influence management innovation. Therefore, organizations are advised not only to implement formal, multiple voice mechanisms to facilitate empowering leadership's indirect effect on management innovation, but also to foster this particular leadership type in general, which seems to be especially useful in the SME context, where human capital is generally more limited and middle managers' knowledge needs harnessing. This is a relevant and practical finding because locally owned organizations and multinational subsidiaries are slowly starting to replace traditional human resources management approaches with newer culturally contingent approaches marked by individual involvement (Hassi, 2016;Walumbwa et al., 2011). Empowering leadership therefore fits well as a means of driving management innovation and thereby SME performance in an African context.

Limitations and Future Research Directions
Our study is not without limitations. First, one of our primary interests was investigating the organization-level phenomenon of management innovation. However, we operationalized and measured it via a set of items that assessed each CEO's perceived management innovation in their firm, which is an individual-level perception (Baer & Frese, 2003). This imperfect level alignment between conceptualization and operationalization/measurement may potentially hinder the accuracy of our study's empirical findings and interpretations (Cheong, Yammarino, Dionne, Spain, & Tsai, 2019). Hence, future research can conceptualize management innovation more explicitly at the individual level (Yammarino, Dionne, Chun, & Dansereau, 2005) with an innovation process perspective on management innovation and by focusing on the generation, rather than the adoption, process. Indeed, the generation and adoption of management innovation are separate processes led by different actors and can be studied at different levels of analysis (Damanpour, 2014). Middle managers and their voices are important internal change agents who create management innovation that results in novel alterations in what managers do daily and how they do it (Volberda et al., 2013).
Second, we used CEOs' perceptions about management innovation, meaning that the main effects from our analysis could have been influenced by the method employed. Research has often argued that the results of the relationship between leadership behaviors and innovation may be larger when the outcomes are selfrated rather than objectively measured (see Hughes et al., 2018). However, a recent meta-analysis on the link between leadership and innovation showed that for the effects of many leadership variables, 'there were no significant differences based on these study design issues' (Lee et al., 2020: 16). Nevertheless, future studies can combine both subjective and objective innovation outcome measures to echo the calls to use stronger research designs, including instrumental variables (e.g., Hughes et al., 2018).
Third, the findings show that the relationship between empowering leadership behavior and management innovation is mediated by middle managers' voice. Voice was operationalized as an extra-role, participative behavior that allows organization members to have a say in work-related issues (Van Dyne & LePine, 1998). However, this ignored the types of messages and information voice delivers. Even though it is a mature scale and still widely used in the literature (Li & Sun, 2015), future research can apply refined voice scales to allow a more defined understanding of voice effects. For example, Liang, Fahr, and Fahr (2012) distinguished between promotive voice (i.e., suggestions to improve the organization) and prohibitive voice (i.e., concerns to prevent the organization from harmful events), differentiating between the content, aim, and function of each. This differentiation could help identify voice utility in different stages of the management innovation process (e.g., the creation and implementation stages). Moreover, given that the mediation effect is smallwhich is often the case (Walters, 2019)it would be insightful to include further mediators in future studies, particularly those mediators identified through the meta-analysis of empowering leadership research by Lee et al. (2018) and to disentangle their explanatory power in the empowering leadership/management innovation relationship.
While all constructs used in this study have been applied in the context of the Middle East or Islamic majority countries in prior research, leadership effectiveness has been argued to be culturally contingent (Cheong et al., 2019). Hence, future research could include additional dimensions of empowering leadership behavior unique to the Moroccan context, reflecting the desire for high integrity or team orientation (Dorfman et al., 2012) which might lead to a better fit with the Moroccan context and yield more explanatory power.
Despite these limitations, the current study provides unique contributions to contingency-based leadership and innovation literature by empirically accentuating the role of middle managers in the relationship between empowering leadership behavior, middle managers' voice, and management innovation. Indeed, middle managers' collectivist orientation is an important boundary condition for boosting the positive indirect effect of middle managers' voice.

NOTES
This project has received funding from the European Union's Horizon 2020 research and innovation programme under the Marie Sklodowsda-Curie grant agreement No. 734447. The authors are grateful to AOM and EURAM 2019 reviewers for their helpful comments on an earlier version of the manuscript and extent their gratitude to Hicham Amine for his assistance in the data collection process. The content of this article does not reflect the official opinion of the European Union. Responsibility for the information and views expressed lies entirely with the authors.

APPENDIX I Management Innovation
(1) Rules and procedures within our organization are regularly renewed.
(2) We regularly make changes to our employees' tasks and functions.
(3) Our organization regularly implements new management systems.
(4) The policy with regard to compensation has been changed in the last three years. (5) The intra-and inter-departmental communication structure within our organization is regularly restructured. (6) We continuously alter certain elements of the organizational structure.

CEO's Empowering Leadership Behavior
Enhancing the meaningfulness of work: (1) The CEO helps me understand how my objectives/goals relate to that of the company.
(2) The CEO helps me understand the importance of my work to the overall effectiveness of the company.
(3) The CEO helps me understand how my job fits into the bigger picture.
Fostering participation in decision making: (1) The CEO makes many decisions together with me.
(2) The CEO often consults me on strategic decisions.
(3) The CEO solicits my opinion on decisions that may affect me.
Expressing confidence in high performance: (1) The CEO believes that I can handle demanding tasks.
(2) The CEO believes in my ability to improve even when I make mistakes.
(3) The CEO expresses confidence in my ability to perform at a high level.
(1) I develop and make recommendations concerning issues that affect this work group.
(2) I speak up and encourage others in this group to get involved in issues that affect the group.
(3) I communicate my opinions about work issues to others in this group even if my opinion is different and others in the group disagree with me. (4) I keep well informed about issues where my opinion might be useful to this work group. (5) I get involved in issues that affect the quality of work life here in this group. (6) I speak up in this group with ideas for new projects or changes in procedures.

Middle Manager's Collectivism Orientation
(1) Group welfare is more important than individual rewards. (2) Group success is more important than individual success.
(3) Being accepted by members of your work group is very important.
(4) Employees should only pursue their goals after considering the welfare of the group. (5) Managers should encourage group loyalty even if individual goals suffer. (6) Individuals may be expected to give up their goals in order to benefit group success.