Four lenses on people management in the public sector: an evidence review and synthesis

We review the literature on people management and performance in organizations across a range of disciplines, identifying aspects of management where there is clear evidence about what works as well as aspects where the evidence is mixed or does not yet exist. We organize our discussion by four lenses, or levels of analysis, through which people management can be viewed: (i) individual extrinsic, intrinsic, and psychological factors; (ii) organizational people management, operational management, and culture; (iii) team mechanisms, composition and structural features; and (iv) relationships , including networks, leadership, and individuals’ relationships to their job and tasks. Each of these four lenses corresponds not only to a body of literature but also to a set of management tools and approaches to improving public employees’ performance; articulating the connections across these perspectives is an essential frontier for research. We find that existing people management evidence and practice have overemphasized formal management tools and financial motivations at the expense of understanding how to leverage a broader range of motivations, build organizational culture, and use informal and relational management practices. We suggest that foregrounding the role of relationships in linking people and performance— relational public management —may prove a fertile and interdisciplinary frontier for research and practices.


I. Introduction
That people are an organization's greatest asset has become almost axiomatic among managers, but how to get the best out of employees is a practical and intellectual goal of the first order. This challenge is particularly salient in public-sector organizations, where many of the managerial levers available to private-sector firms are constrained by law, politics, or the differing nature of public-sector tasks. Yet taking a systematic and evidence-informed approach to people management is difficult because the relevant academic literature is fragmented across disciplines and research areas. This makes it difficult not only to know what evidence exists to guide decision-making, but also how to fit those pieces of evidence into a broader understanding of the questions, approaches, and interventions that constitute the field. This paper reviews and synthesizes key questions, research areas, and empirical evidence relevant to people management in the public sector. We organize our discussion around four lenses, or units of analysis, through which researchers and policy-makers have viewed issues of people management: (i) individuals; (ii) organizations; (iii) teams; and (iv) relationships. Table 1 summarizes the structure of the paper as well as key insights arising from our review and synthesis.
While the breadth of the literature resists easy summary, our review highlights two themes. First, across all four lenses we observe a relative over-representation of research on formal structures and practices, and on financial incentives and extrinsic motivation in both research and practice. Although these are essential aspects of people management, they represent only a narrow slice of the factors that determine employee performance and the tools available to improve it, and arguably do not deserve their prominence as the first, almost default approach to improving performance. Our review therefore seeks to give a broader and more balanced picture of the problems and the evidence on people and performance in the public sector, which also encompasses factors such as non-financial motivations, informal management practices, organizational culture, and the networks in which individuals and organizations are embedded.
The second theme of our paper is the potential value of thinking about people and performance through the lens of relationships among actors, rather than focusing on the actors themselves (whether individuals, organizations, or teams) in isolation. These relationships exist both across levels of analysis (e.g. the interconnection between the individual-level mechanisms of extrinsic and intrinsic motivation and organizationlevel performance management processes) and within levels of analysis (e.g. through the complementary and self-reinforcing processes among individuals through which individual behaviour and norms constitute collective culture). Throughout our discussion of the individual, organization, and team lenses in sections II-IV, we highlight instances where such relationships are important. In section V we go a step further and discuss three types of relationship that can themselves be understood as units of analysis (networks, leadership, and individuals' relationships to their jobs), and review the relatively scarce literature that foregrounds relationships rather than actors. We conclude by suggesting that a relationship-focused approach to studying management in public organizations-relational public management-has the potential to open up new research angles as well as shed light on current debates. 1 Any paper on a topic as broad and complex as this one needs to be amply caveated. Our aim is not to provide a comprehensive review of the thousands of academic books and articles on management or human resources (HR). Nor do we seek to recommend specific 'best practices' or any type of formula for people management, as the complexity and contextual specificity of people management would render any such effort futile. Rather, we aim to present a focused and concise picture of the state of the literature that can serve as a summary of where evidence does and does not exist, as an entry point for more in-depth study, and-perhaps most importantly-as a framework to help readers organize their own questions, reflections, and experience. Our review pieces together elements from the disciplines of public administration, management, human resources, sociology, economics, psychology, and political science, but does not claim to be fully representative of any of them. And while our primary focus is on management in public-sector organizations, we supplement our review with theory and evidence generated in private-sector or non-profit organizations where these lessons are applicable to public-sector contexts.

II. People as individuals
This section explores the evidence on the individual-level factors that influence employees' performance in government organizations. We specifically focus on three areas: (i) extrinsic factors, such as monetary incentives (salary and bonuses) and institutional incentives (promotion, leisure benefits, training); (ii) intrinsic factors, such as meaningful work, job satisfaction, and prosocial behaviour; and (iii) psychological factors such as cognitive biases, which can systematically affect public employees' decision-making process and prevent them from performing at the most optimal level. 2 Three main insights emerge from the literature. First, financial incentives-such as pay for performance-can sometimes work to improve public employees' performance but are more applicable when employees' tasks correspond to simple, clear, measurable targets. However, misaligned financial incentives can distort effort and may undermine intrinsic motivation. Second, there is strong evidence that intrinsic factors play a major role in motivating public employees and improving their performance, so better leveraging intrinsic motivations is a major opportunity for managers. However, there is more evidence about these motivations themselves than about how to effectively leverage them in management practice. Third, external non-financial rewards that leverage intrinsic motivation, such as social recognition, can be effective and avoid some of the drawbacks of financial incentives, and thus could be used more widely. Table 2 summarizes both the structure of the section and some of the key insights arising from the review and synthesis of the individual-level factors that influence public employees' performance.

(i) Extrinsic factors
Traditionally, organizations, in public and private sectors, assume that an increase in the financial incentives will improve individual performance (Gneezy and Rustichini, 2000). Therefore, a person with a better salary doing the same activity will decide to work more and better than a person with a lower wage (Lazear, 2000;Ariely et al., 2009b). This approach argues that external rewards, such as bonuses, an increase of salary, a promotion or any type of financial incentive, will trigger extrinsic motivation and increase an individual's performance (Bonner et al., 2000;Bonner and Sprinkle, 2002;Dal Bó et al., 2013). For our review, we define extrinsic motivation as whenever an activity is done in order to attain a tangible external benefit. Therefore, extrinsic motivation leads individuals to perform a task or activity for the instrumental value that it has for achieving a goal.
To increase public employees' productivity, governments have adopted pay-for-performance schemes (Varone and Giauque, 2001;Weibel et al., 2010;Bellé and Cantarelli, 2015). There are several studies that have shown that rewarding performance based on outcomes can have a positive impact on productivity (Lazear, 2000;Ariely et al., 2009b;Bandiera et al., 2017). For example, Gertler and Vermeersch (2012) found that offering performance-based pay to health care staff in Rwanda improved the provision of preand post-natal care (Basinga et al., 2011;Bandiera et al., 2017). Likewise, Muralidharan and Sundararaman (2011) found that an increase in teachers' salaries based on students' test scores has a positive effect on teachers' performance and students' learning outcomes (see also Levitt et al., 2016). However, some findings have shown that not every type of financial incentive works, well-designed financial rewards-simple and with measurable targets-linked to job outcomes are the ones that can have a higher impact on improving public employees' performance (Bandiera et al., 2017). In public organizations, identifying specific measurables targets can be a challenge, and promotion, leisure benefits.

Intrinsic factors
Three main intrinsic motivators can influence public employees' performance: • Meaningful work • Prosocial preferences • Job satisfaction • Intrinsic motivation (Perry, 1996;Ryan and Deci, 2000;Esteve and Schuster, 2019) • Prosocial behaviour (Bénabou and Tirole, 2006). • Social incentives (Ashraf and Bandiera, 2018) Psychological factors Cognitive biases influence public employees' performance by affecting their • Prospect theory (Tversky and Kahneman, 1981) • Cognitive biases and public service (Bellé et al., 2017 pay-for-performance schemes can even have negative effects on work effort as they can crowd out the intrinsic motivation related to being a public servant (Gneezy and Rustichini, 2000;Gneezy et al., 2011). For example, Georgellis et al. (2011) found that higher predicted earnings and satisfaction with the extrinsic job characteristics reduce individuals' propensity to accept employment in the public sector. Likewise, Bellé and Cantarelli (2015) found that monetary rewards had no significant effect on the intended effort of executives working for the Italian central government. In section III below, we explore further how these individual-level responses to incentives translate into the practical use of financial incentive schemes by public organizations.
Another way to incentivize individual performance with external rewards is through promotion in public organizations. Promotion opportunities matter as they behave as an extrinsic motivator that allows employees who perform to an exceedingly high level to see that their effort can be recognized, and that the advancement is a realistic expectation. For example, Karachiwalla and Park (2017) found, using a tournament model of promotion and retrospective panel data, that teachers increase effort in years leading up to promotion eligibility, and they reduce effort if they are repeatedly passed over for promotion. Likewise, Whitford (2018), in his research on tournaments, argues that promotion tournaments in public organization hierarchies might be more efficient than pay-for-performance schemes. This type of external incentive can have a smaller crowding out effect on intrinsic motivation (Coccia, 2019). In this sense, promotion can be an effective instrument to incentivize public employees' performance. However, promotion incentives need to be based on performance and not on seniority, education, or connections. In some countries, the lack of data for monitoring and evaluating each individual's performance can be a challenge to implementing successful promotion schemes based on merit. We discuss this further in the next section.

(ii) Intrinsic factors
Individuals are not just motivated by self-interest and material concerns, but also by experiences, emotions, values, and identities (Frey and Oberholzer-Gee, 1997;Ryan and Deci, 2000;Paarlberg and Lavigna, 2010). Due to the nature of public service, intrinsic motivators play a vital role in improving public employees' performance (Heyman and Ariely, 2004;Kamenica, 2012;Esteve and Schuster, 2019). Here, we define intrinsic motivation as those internalized factors that spur effort without the inducement of a tangible external benefit.
Even though there is a vast literature that supports the importance of intrinsic motivators in public service, more knowledge is needed to evaluate and operationalize them into managerial practices in public organizations (Perry and Wise, 1990;Perry et al., 2009). Three main intrinsic motivators can influence public employees' performance: first, when public employees feel that their effort is meaningful (Ariely et al., 2008;Grant and Gino, 2010); second, when they are committed with prosocial activities and desire to serve the public (Gneezy and Rustichini, 2000;Mellström and Johannesson, 2008;Ariely et al., 2009a); third, when public servants feel a pleasurable or positive emotional state-job satisfaction-by doing their work. Additionally, social recognition of their effort and symbolic awards can also influence public employees' performance by leveraging their intrinsic motivation (Kosfeld and Neckermann, 2011;Bradler et al., 2016).
According to Ariely et al. (2008), work that is perceived to be meaningful is an essential ingredient to individuals' putting in effort and performing well. A job is considered meaningful to the extent that it is recognized and/or has a purpose. Recognition means that some other person acknowledges the completion of the work. Such recognition does not have to be linked to any financial incentives or to any non-tangible rewards such as praise or appreciation. Purpose means that the employees understand how their work might be connected, even tangentially, to important objectives. The literature on the impact of meaning on behaviour and performance is quite sparse, especially within economics and public administration. In economics, Loewenstein's (1999) study of mountaineering literature examines the role of meaning as an incentive. Preston (1989) and Leete (2001) look at whether individuals accept lower wages to work in the non-profit sector, while Stern (1999) examines whether scientists are willing to take a wage cut in order to be able to publish their work. In public administration, the work of Perry (1996) on public service motivation has studied the effect that meaningful public service has on performance (Perry and Wise, 1990;Durant et al., 2006;Perry et al., 2009). Considering this evidence, there are two central insights from goal-setting theory for increasing public employees' performance: (i) the importance of establishing clear, measurable, and meaningful goals (not outcomes); and (ii) the importance of clarifying organizational goals (Locke andLatham, 1990, 2002;Latham, 2004;Verbeeten, 2008;Wright et al., 2012).
Prosocial motivation can also be used to improve public employees' performance (Ashraf, 2013). Prosocial preferences consider the individual's internal desire for approval and external willingness to help others, which is especially relevant for public employees. Weber [1978Weber [ (1922] and Durkheim (1956) see the commitment to public service as the crucial factor in effectively delivering services to citizens. The idea of mission motivation in bureaucracies is also emphasized by Wilson (1989) and Tirole (1994). Evidence shows that making salient the social impact of public service can affect the behaviour and performance of public employees. For example, a study in a North Carolina hospital showed that prompts telling practitioners that hand hygiene protects patients from disease was much more effective at inducing them to wash, compared to merely reminding them that hygiene protects them (Ashraf, 2013). Likewise, the tools from value-based management can be useful to harness the positive aspects of public service-their impact on social and public good-and to connect with officials willing to help others (Grant, 2008;Grant and Sumanth, 2009;Paarlberg and Lavigna, 2010).
Research has shown that job satisfaction is positively related to motivation, job involvement, organizational citizenship behaviour, organizational commitment, life satisfaction, mental health, and job performance. Job satisfaction is defined as 'a pleasurable or positive emotional state, resulting from the appraisal of one's job or job experiences' (Locke, 1976;Kim, 2004). This emotional response is also negatively linked to absenteeism, turnover, and perceived stress (Judge et al., 2001;Kreitner and Kinicki, 2001). Using the data collected from 298 schools and 13,808 teachers, Ostroff (1992) supported the positive relationships between employees' job satisfaction and organizational performance. Public managers can use the tasks and relationships embedded within jobs to try to improve officials' job satisfaction-an idea we explore further in section V below.
Finally, social recognition and peer appreciation, which are external non-financial rewards, can have internal effects on motivation and a positive impact on intrinsic motivation (Besley and Ghatak, 2018;Kosfeld and Neckermann, 2011;Ashraf and Bandiera, 2018). Research has shown that appreciation and recognition can be more important than good wages, job tenure, and promotion opportunities. For example, Ashraf et al. (2014) experimentally confirmed that Zambian agents assigned to a nonmonetary reward treatment-namely, stars for performance plus a public ceremony for top performers-sold twice as many condoms as agents who were offered a modest financial margin on each pack traded. Such a study could help in public-service contexts where management and performance could benefit from new techniques of motivation. Moreover, recognition may enhance intrinsic motivation, for instance, by making the positive attributes of the effort more salient.

(iii) Psychological factors
Psychological factors such as cognitive biases can have an impact on public employees' performance by systematically affecting their decision-making process and preventing them from performing at the most optimal level. In this section, we explore how these cognitive biases can influence public employees' behaviours in two main issues: (i) judging performance information, and (ii) innovation and risk behaviours.
Scholarship from behavioural science and behavioural public administration has shown that individuals' perception and judgement of performance metrics can be affected by comparisons, points of reference, and framing of the information and emotions (Tversky and Kahneman, 1981;James et al., 2020). Fuenzalida et al. (2020) confirmed that officials are susceptible to framing effects when judging performance information. More specifically, the authors experimentally showed that public-service managers and professionals tend to evaluate metrics about target achievement and job satisfaction more negatively when these performance rates are negatively presented (as opposed to their logically equivalent percentages presented under a positive frame). Bellé et al. (2017) found that anchoring and halo effects systematically biased performance appraisal. On the one hand, the anchoring effect affects public employees' decisions by establishing a starting point that will shape their subsequent estimations. For example, in an artefactual field experiment, Bellé et al. (2017) identified that average scores were higher when public managers were exposed to a high rather than a low anchor. On the other hand, the halo effect influences the performance evaluation that public managers do by creating a general assessment of the different performance dimensions based on the like (or dislike) of a person and regardless of, or even contrary to, available information (Battaglio, 2015;Bellé et al., 2017).
Framing effects and status quo biases can affect the decision process of individuals regarding implementing new policies, taking more risks, and ethical behaviours (Tversky and Kahneman, 1981;Banuri et al., 2019). Bellé et al. (2017), through a series of experiments, demonstrated that the way that policies' outcomes are presented (framing) could impact policy-makers' evaluations and behaviours. Specifically, they highlighted the effect of gaining or losing framing. For example, they asked individuals to select one of two public policies; the expected value of the outcome of the two public policies was the same. Nonetheless, the outcome of one policy was expressed as a sure thing, whereas the outcome of the other policy was expressed in probabilistic terms (Bellé et al., 2017;James et al., 2020). The result of the experiments showed that individuals prefer the policy with the certain outcome when the outcomes are framed positively and prefer the policy with the probabilistic outcome when they are framed negatively. Gómez et al. (2018) conducted a lab experiment with police officers in Mexico and found that loss aversion framing motivates a more dishonest behaviour. Thus, under the framing effect, decision-makers tend to be risk-averse in the domain of gains and risk-takers in the domain of losses (Tversky and Kahneman, 1981).
While there is thus ample evidence that psychological factors influence decision-making in public servants (as in other groups), this literature has so far succeeded more in establishing that these factors exist than in providing actionable insights for managers on how to counteract or leverage them. This poses both a challenge and opportunity for the nascent field of behavioural public administration (Grimmelikhuijsen et al., 2017) as well as for social scientists and policy-makers more broadly.

III. People in organizations
This section reviews the evidence on how organization-level processes and practices affect individual-level performance. While the vast literature on this subject makes a comprehensive survey impossible, we highlight three areas of particular relevance for people management: (i) human resources management (including hiring, career progression, and performance management and performance-linked pay); (ii) operational management (e.g. managing and monitoring work processes); and (iii) organizational culture. We exclude external, environmental, and contextual factors that affect organizational performance but are typically outside the control of public-sector managers.
Three broad themes emerge from the literature. First, the nature and quality of these organization-level processes and practices have large effects on the performance of individuals within the organization, even after controlling for individual characteristics and workforce composition. This implies that significant performance improvements are possible for public-sector organizations even when resources and the ability to hire or fire employees are constrained-as they usually are in the public sector. Second, many processes that influence individual performance are not classic human resources functions, although appropriate human resource management practices may be complements to better operational management processes. Third, while much of research and practice focuses on using formal, financial incentives to improve performance, these are difficult to implement effectively, and their impacts are highly variable across contexts and functions-especially in core civil service settings. Non-incentive-based management practices may thus have a broader potential. The scope and modalities for this is a frontier of research.

(i) Human resources management
A range of evidence has shown that organizational hiring processes can be used to draw in more skilled and motivated recruits-albeit in sometimes surprising ways. The classic theoretical dilemma in public-sector hiring is whether extrinsic motivations crowd out intrinsically motivated recruits (e.g. Prendergast, 2007). On the one hand, the evidence supports the idea that intrinsic factors such as public-service motivation and person-organization fit play a major role in attracting high-quality applicants, and that mission alignment can substitute for extrinsic incentives (Perry and Wise, 1990;Besley and Ghatak, 2005). On the other hand, a growing body of studies-including several recent field experiments-find that extrinsic motivations either do not deter or may even crowd in recruits with higher motivation and/or subsequent job performance. This is true of resource-intensive types of extrinsic rewards such as pay levels (Dal Bó et al., 2013) and performance-linked pay schemes (Leaver et al., 2019b), but also of non-financial extrinsic factors such as career progression opportunities, personal benefits, and person-organization fit (Chapman et al., 2005;Ashraf et al., 2015;Linos, 2018). While intrinsic sources of motivation are doubtless important, this emerging evidence suggests that organizations must develop recruitment strategies that are extrinsically appealing to the type of people with the intrinsic motivation that the organization wishes to attract. These extrinsic appeals include not just initial wage levels or performance-linked financial incentives, but also longer-term personal and professional development opportunities.
Despite the importance of career considerations, however, there is relatively little theory-and still less evidence-about how governments should design promotion and career progression structures. Within economics, a relatively small literature on internal bureaucratic labour markets and personnel economics in private firms (Lazear and Oyer, 2012;DeVaro and Waldman, 2012;Ke et al., 2018) sets out some of the key theoretical considerations, such as the tension between productive efficiency and the use of promotions as a reward for good performance. However, these models focus on management levers (e.g. high-powered financial incentives, discretionary promotion and firing) that are not usually available in the public sector, which instead tends to be characterized by low-powered (if any) formal incentives, rigid career progression systems, and almost non-existent firing powers (except for discretionary political appointees). And while there are detailed descriptions of the different ways governments can organize career structures, such as with career vs position-oriented systems or the use of generalist vs specialist classes, there is little rigorous evidence about the effects of such design choices in all but the broadest terms. Nevertheless, careful studies such as Teodoro (2009Teodoro ( , 2011 and Bautista-Chavez (2020) do demonstrate that the career opportunities facing government managers-in particular the opportunity to advance by making diagonal moves across agencies and roles-can encourage managers to take risks by introducing innovations. In contrast, less flexible structures that allow only vertical advancement within a given career or agency might discourage such innovative behaviour.
Aside from career progression, performance management and performance-linked pay are the main channels through which human resource practices affect employee performance. These are typically implemented through formal periodic appraisal cycles (usually annual) in which employees' performance is evaluated against a pre-defined set of targets (intended to be as objectively measured as possible), with some form of rewards and/or punishments linked to this evaluation. In addition to the empirical literature documenting individuals' responses to incentives discussed in section II above, there exist several studies which focus more on the organizational and managerial aspects of implementing financial incentive schemes in the public sector (Perry et al., 2009;Gerrish, 2016;Weibel et al., 2010;Hasnain et al., 2014;Arnaboldi et al., 2015). As with the individual-level literature, these studies tend to converge around the view that: (i) well-designed performance-linked pay can sometimes increase the performance of employees in lower-level, more frontline roles with more narrowly defined and measurable tasks; (ii) in practice, many public-sector performance-linked pay schemes are not well-designed or are misapplied and often fail to be implemented, lead to gaming or effort distortion, or harm employee morale; (iii) there is less evidence on potential effectiveness for core civil servants in mid-level or senior managerial roles, as their tasks are even more difficult to pre-specify and measure objectively and their roles are more politicized. But while this literature has focused heavily on the role of financial incentives, performance management routines typically bundle these incentives with roledefinition and performance review functions that might have their own independent effects (Williams and Yecalo-Tecle, 2019). This suggests that focusing on improving these aspects of performance management (without linking them to high-powered incentives) might be a more widely effective approach for non-frontline public-service roles.

(ii) Operational management
Organizational management practices that do not bear directly on human resource issues, such as the management and monitoring of work processes, can nonetheless have a significant impact on the productivity of individual staff. Extensive literatures in economics, public administration, and even sociology document the existence of what Gibbons and Henderson (2012) call 'persistent performance differences' among similar organizations, and trace these differences back to differing levels of management quality and the organization of work (e.g. Meier and O'Toole, 2002;Boyne, 2004;Lazear and Oyer, 2012;Bloom et al., 2014;McDonnell, 2017;Leaver et al., 2019a). While empirical studies differ in the extent to which they can control for the quality of human resource inputs, these performance differences are still substantial even in the most tightly controlled studies (e.g. Syverson, 2004). This literature also finds strong evidence of complementarity between operational management and human resource management: some types of operational practices, such as more flexible production practices, benefit from supporting HR practices, and vice versa (Ichniowski and Shaw, 2003;Brynjolfsson and Milgrom, 2013). While this has been investigated for private firms, there is little rigorous evidence in the public sector about how the introduction to the public sector of operational practices such as e-government, agile management, and remote working might interact with different approaches to people management.
Another theme that emerges from this literature is the tension between two distinct approaches to improving operational management: (i) process standardization, monitoring, and control; and (ii) improving employees' exercise of discretion and autonomy through flexibility, empowerment, and communication. As Miller and Whitford (2016) note, this tension aligns to an extent with an old debate within public administration about the nature of bureaucrats and their roles: are they lazy, self-interested agents, or diligent and public-spirited professionals? This dichotomy also points to the impossibility of separating out people management from operational management. While much of the empirical evidence within economics has found positive results from better incentives and monitoring practices (among others) in private firms, schools, and hospitals (see various examples in Bloom et al., 2014), studies in public-sector organizations and professional organization contexts have identified positive results from enhanced discretion and autonomy and even potential negative associations from over-reliance on top-down control approaches (Rasul and Rogger, 2018;Rasul et al., 2020;Bandiera et al., 2020). As with the discussion on performance-linked pay above, this is perhaps best explained as a matter of different approaches being effective for different types of tasks or roles, but the potential interactions between management styles, task types, and individuals' intrinsic motivations is a fertile ground for further research.
An even more nuanced question (and potential research frontier) relates to the interactions and blending of these different practices with each other: the optimal management mix surely includes elements of both standardization and monitoring as well as discretion and flexibility, but which elements, when, and why? This question is the subject of a great many management tracts, and countless qualitative studies shed light on specific aspects of it. But there is much less to draw on in terms of big-picture theory that could help elucidate general principles that could be used to navigate through the infinite range of practical dilemmas managers face, or quantitative studies that could be used to test hypotheses derived from these principles.

(iii) Organizational culture
A third organization-level factor that influences individual performance is culture, in the sense of shared expectations, norms, and cognitive frames among members of the organization (Schein, 1985). 3 The same individual operating in different organizational cultures might react to other management practices (both HR-related and operational) quite differently, and engage differently in important behaviours such as voice and innovation (Edmondson, 1999;Ashford et al., 2009). Long a core concept in management studies (Schein, 1985) and governance (Grindle, 1997), more recent quantitative studies have shown various dimensions of culture to directly affect measures of performance in settings such as hospitals (Martinez et al., 2015;Curry et al., 2018) and to interact with the effectiveness of reforms to operational management in private firms (Blader et al., 2020). Culture-the shared set of expectations, norms, and cognitive frames-is not just shaped by the members of an organization and by organizational-level processes and practices, but also shapes them.
But while few dispute that organizational culture matters for individual and organizational performance, much less is known about how these cultures transform over time and can be shaped-despite a nascent theoretical and lab-experimental effort to do so within economics (Chassang, 2010;Gibbons, 2020;Gibbons et al., 2020). While most studies of organizational culture formation are long-term case studies of particular sets of high-performing organizations (e.g. Grindle, 1997;Tendler, 1997;McDonnell, 2017), Azulai et al. (2020) conduct a randomized controlled trial with Ghana's civil service and show that a one-day training emphasizing bottom-up culture change delivered to mid-level bureaucrats from across the service can lead to persistent and widespread improvements in organizational culture and performance. However, delivering the same training to existing work teams led to no changes in culture or performance. This illustrates both the possibility and the complexity of efforts to improve organizational cultures. In particular, the difference between the individual-and team-training branches of the intervention demonstrate that while culture is held by individuals and tends to cohere within organizations, it is produced by repeated interactions and social relationships within teams. To begin to unpack these complexities, we now shift our attention to two different levels of analysis: first to the dynamics of teams within an organization, and then to the existence of networks within and across them.

IV. People in teams
Teams represent an intermediate lens and unit of analysis, between the individual and the organization, for studying people management and performance. Teams are collective and inherently social, which makes them a prominent building block for studying how shared cognition and culture are built within organizations. They are also smaller in scale than organizations and thus analysing them allows for within-organization heterogeneity, and because they can exist both within and across formal organizational sub-divisions they allow researchers and managers to think about performance more flexibly than does an official organizational chart. 4 In practical terms, the growing importance of teams per se as a managerial unit within the public sector has been driven by the need for more flexible and responsive organizational forms in government. Some authors even call for the adoption of agile government-emulating agile software development-as a new production method for public goods and services (OECD, 2015;Roseth et al., 2018;Mergel et al., 2020).
In this section, we discuss some essential mechanisms and characteristics that contribute to team attitudes, behaviour, and performance (Cohen and Bailey, 1997;Mathieu et al., 2019). We organize our discussion into two main sub-sections: (i) the mechanisms through which teams work, including processes and emergent states; and (ii) their composition and relationship to external structures.
Three key insights emerge from the literature. First, the relative novelty of teams as a recognized managerial unit in government is matched by the scarcity of public management studies focusing on this organizational level, which strikingly contrasts with the extensive literature on teams in other areas such as (private-sector) management, psychology, and human resources. 5 Second, there exist a variety of performance factors that are perhaps more important for teams than for other levels of analysis, such as developing shared group mindsets or peer-appraisals. Other important team-level levers interact with individual-and organizational-level practices, which speaks to the value of encouraging a networked and integrated approach for people management policies and practices in public organizations across levels of analysis. Third, empirical evidence confirms that 'soft' factors such as cognitive, affective, and behavioural states and processes are often more important for teams' effectiveness than formal rules, external conditions, or the characteristics of the teams' members. This implies that there is more room for managing teams than is often assumed. Thus, the mechanisms by which team members interact to produce outputs and results primarily explain the gap between what a team can achieve-considering the abilities of its members, external conditions, and established rules-and the ultimate team effectiveness (e.g. Barron, 2003). This stresses the importance of relationships within teams.

(i) Mechanisms of teamwork
The primary mechanisms underlying the operation of teams can be viewed in terms of emergent states and team processes (Marks et al., 2001). Emergent states are cognitive, motivational, and affective states that vary dynamically-depending on the structural context and networks where the team performs, its inputs and compositional features, processes, and outcomes-and also determine the way a group works. Team processes are cognitive, verbal, and behavioural members' interpersonal activities to organize taskwork to attain team goals, and therefore to convert team inputs to outcomes.
Two major emergent states for effectiveness in groups are cohesion and trust. The former leads members to ask and offer opinions and hence to increase knowledge sharing (van Woerkom and Sanders, 2010), and the latter enables subjects to disengage from defensive behaviour in response to the potential harm (they presume) by colleagues. This attitude leads individuals to focus their efforts on working collaboratively towards achieving group goals, rather than pursuing personal interests (Jones and George, 1998;Dirks, 1999;Mayer and Gavin, 2005;Joshi et al., 2009;De Jong and Elfring, 2010;De Jong et al., 2016). On the other hand, team conflicts jeopardize team performance and cause the deterioration of individual attitudes and behaviour in teams, such as their motivation, satisfaction, and identification (Hülsheger et al., 2009;De Wit et al., 2012). Disputes are especially compromising when they are relational and about interpersonal incompatibilities, tension, hostility, and irritation among group members (Huang, 2012). Team empowerment is another emergent state positively associated with employees' performance as well as with a broad range of attitudes and behaviour, including their job satisfaction and organizational commitment, while also preventing them from strain and turnover (Seibert et al., 2011).
Another primary emergent state when managing collective production in a group is achieving shared cognition on fundamental notions for its performance. Every individual has their own interpretation of the world through knowledge structures or mental models, whereby they make inferences, predictions, and decisions (Johnson-Laird, 1983). However, shared mental models are the enabling knowledge to work as a team, including tasks, equipment, roles, goals, and abilities (Cannon-Bowers and Salas, 2001;Lim and Klein, 2006). Studies confirm the positive link between shared mental models and team performance (Ensley and Pearce, 2001;DeChurch and Mesmer-Magnus, 2010), as well as the improvement they mean for several group processes that ultimately affect team effectiveness, such as the strategy formation and coordination, cooperation, and communication (Mathieu et al., 2000;Marks et al. 2002). Transaction memory systems are vital to attaining shared mental models as they emulate a 'group mind' by which teams collectively encode, store, and retrieve knowledge (Hollingshead, 2001). Thus, a tacit mechanism arises in the team's assigning responsibility for knowledge and expertise to its members and bringing awareness about who knows what (Mathieu et al., 2008). As a result, everyone has their own as well as others' information to access (Brandon and Hollingshead, 2004). Communication has an essential role in developing transaction memory systems overtime because discussions serve to continuously improve the mapping of team members about who is an expert in certain areas (Liang et al., 1995;Rulke and Rau, 2000;Hollingshead and Brandon, 2003;Lewis, 2004). 6 Several empirical studies, mostly targeting regular individuals in labs or educational environments, confirm that better performing teams have well-communicated functional networks (Brewer and Holmes, 2016;Amelkin et al., 2018;Marks et al., 2002). An alternative approach to attaining shared cognition is reinforcing knowledge and information sharing, because groups can make the most of the initial know-how and informational resources of their members, and therefore improve team effectiveness, creativity, and innovation (Hülsheger et al., 2009;Mesmer-Magnus and DeChurch, 2009;Kessel et al., 2012). These sharing processes should centre on unique knowledge and information to specific team members instead of resources commonly held by most individuals in a group (Mesmer-Magnus and DeChurch, 2009;Mesmer-Magnus et al., 2011).
Within the range of processes affecting team effectiveness, two types are the most relevant: (i) those related to the planning of work; and (ii) those linked to task accomplishment and performance monitoring. Indeed, a critical tool for improving team performance is the use of peer-review processes. There is evidence that developmental peer-appraisals improve several group attitudes and behaviours in teams, such as individual satisfaction, motivation, etc. (Druskat and Wolff, 1999). Studies also confirm the positive effect these practices have on the performance of team members. For instance, Behrens and Chemin (2019) show that non-binding peer reviews lead to higher effort and team productivity. These effects are more substantial for low-ability individuals in low-ability teams. Also, Ho (2017) experimentally confirmed that peer reviews improve the effectiveness and consistency of food safety inspections across public servants. Team performance is also structurally determined by the goals it must achieve. As these groups are usually subsets of larger organizational units in public agencies, there is a high risk of only relying on the super-ordinate institutional goals rather than developing particular group goals. Establishing clear goals for teams improves their performance (van der Hoek et al., 2018) as well as individual-level performance (Sonnentag and Volmer, 2010)

(ii) Composition and structural features of teams
Team composition can be discussed in terms of three general themes: the team size, what characteristics of its members should be considered for enabling certain performance levels, and what are the distributional features of team members. Two significant streams of evidence regarding team size set up a balanced perspective here. Studies have confirmed the positive effects of larger teams on the performance of top management groups (Haleblian and Finkelstein, 1993) as well as on group productivity to monitor, classify, and map real-time information during humanitarian crises such as earthquakes or hurricanes (Mao et al., 2016). Nonetheless, the advantages of a larger team size should be calibrated to the higher coordination, communication, and managerial demands it involves (Haleblian and Finkelstein 1993). Another critical issue refers to the features that members should have to achieve the functions and goals the group expects. Average cognitive abilities such as subjects' mental ability and expertise-i.e. experience and education-predict team performance (Stewart, 2006). High conscientiousness, agreeableness, and extraversion of members also contribute to better performance of groups (Bell, 2007). Finally, there is no consensus about the effects of demographic and psychological diversity on team effectiveness (Bell et al., 2011;Opstrup and Villadsen, 2015), but there is evidence in this line suggesting the importance of a balanced diversity of functions and roles in teams (Bell, 2007;Humphrey et al., 2009). Such balance is attained by having the right mix of backgrounds in groups and ensuring the coverage of the functional areas required for their production.
The most salient structural factor regarding team effectiveness is its interdependence, which refers to features of the team-usually task-driven inputs such as resources, workflows, goals, and reward mechanisms-determining the interconnectedness of members (Wageman, 1999) 7 . Higher levels of task cohesion and interdependence positively affect group performance by creating explicit expectations about the intensity of members' interactions to produce the expected outcomes (Castaño et al., 2013;Chiocchio and Essiembre, 2009). Moreover, outcome interdependence encourages and incentivizes members to build and maintain relationships to achieve the expected collective results that the team pursues (Courtright et al., 2015).
Teams thus form an important building block through which individuals are aggregated into organizations. Still, they nevertheless presume a sort of coherence, intentionality, and self-consciousness that does not always match the messy and piecemeal reality of life inside organizations. In the following section, we instead place relationships front and centre as the lens through which people management is viewed. 7 Courtright et al. (2015) synthesize many forms in which scholars have defined, classified, and integrated various representations of interdependence into two main concepts: task interdependence and outcome interdependence. The former corresponds to how taskwork is designed so that members depend upon one another for access to critical resources and create workflows that require coordinated action, while the latter refers to the way the outcomes of taskwork are measured, rewarded, and communicated at the group level, as a way to emphasize collective outputs rather than individual performance (Courtright et al., 2015(Courtright et al., , p. 1828).

V. People in relationships
A fourth potential unit of analysis for studying people management is the relationship. Relationships link individuals, organizations, and teams to each other both across and within levels of analysis. But the characteristics of a relationship are not completely determined by the features of the actors that it links, nor by fully formalizable contracts or practices. Thus far in the paper, we have sought to show how an understanding of these interconnections can both complicate and enrich our view of the link between people management and performance at the individual, organization, and team levels.
In this section, we discuss two well-established areas of research in public-sector people management that go a step further by taking the relationship itself as their key unit of analysis: networks and leadership. We also propose that a relational perspective might also be applied to understanding how individual bureaucrats' tasks and roles shape and are shaped by these bureaucrats-in other words, bureaucrats' relationships to their own jobs. The remainder of this section discusses existing theory and evidence in these three areas, including that which does not explicitly take a relational perspective. 8 Across each area, we find more evidence that relationships are substantively and analytically important than concrete guidance on what this means for good publicsector people management. This suggests to us that relational public management may be a productive frontier for further work-an idea we expand on more speculatively in the paper's concluding section-but is not yet a fully developed approach or method.

(i) Networks
Networks have been an important topic of study for public-and private-sector management since the 1990s, with foundational contributions in the public sector authored by Provan andMilward (1995), andO'Toole (1997). The literature on networks is vast, with several excellent existing literature reviews (e.g. Provan and Lemaire, 2012;Hu et al., 2016), so rather than attempt to survey this literature we instead highlight two aspects of it which are important for understanding public employees' performance: (i) the importance of inter-organizational networks for service delivery; and (ii) the impacts of employees' personal and professional networks on their behaviour.
The core insight of the literature on inter-organizational networks is that public service delivery outcomes are the result not just of actions taken by individuals or organizations in isolation, but of inter-related networks of government organizations, NGOs, communities, and beneficiaries. Provan and Lemaire (2012) refer to this as the 8 While thinking of a relationship (rather than an actor) as a key unit of analysis may be foreign to many scholars, precedents do exist within social science. Within organizational economics and theory, a small but influential stream of work conceives of employment as not just an economic relationship but also a social one (Baron and Kreps, 2013). Within sociology, the influential school of relational sociology is premised on the foregrounding of relations rather than actors (e.g. White, 1992;Mutch et al., 2006;Crossley, 2010;Padgett and Powell, 2012). Within psychology, Fiske (1992) has developed a framework for a unified theory of social relations. However, neither field has applied this relational perspective to the empirical study of public-sector management. Within public administration, the closest analogue that exists in the empirical literature on inter-organizational policy systems and coordination (e.g. Milward, 1982;O'Toole and Montjoy, 1984;Jennings and Ewalt, 1998;Bryson et al., 2006), while Stout (2012) offers a philosophical discussion of relational ontologies and their application to theoretical questions within public administration.
'whole network' approach to network analysis, arguing that the appropriate level of study is the network (rather than individuals or organizations, in more common 'egocentric' approaches to networks). Schneider et al.'s (2003) work on stakeholder networks in estuary management provides an example of a case where policy outcomes can only be understood as the joint outcome of many organizational and individual decisions. Similarly, Kapucu and Garayev (2016) show how differing network structures affected disaster response in comparative case studies. This area of literature thus focuses mainly on within-level relationships among a particular set of actors, usually organizations or individuals.
From the 'whole network' perspective, managers thus need to see their objectives not just as maximizing individual or organizational performance in a narrow sense, but of improving network effectiveness (Provan and Milward, 1995) and engaging in the governance and even design of networks (Rhodes, 1996;Provan and Kenis, 2008). While the complexity of network functioning precludes simple recommendations for managers, the main implication from this literature is that public managers need to see themselves both as managers in networks (in the sense that their key objectives depend on actors outside their own organization) as well as managers of networks (in the sense of actively participating in, drawing resources from, and building such networks) (Provan and Lemaire, 2012).
The second key insight of the networks literature for public-sector people management is that individual employees are embedded not just in formal organizational hierarchies, but also in intra-, inter-, and extra-organizational networks that might influence their behaviour within their official organizational roles. For example, Moynihan and Pandey (2008) show that the strength of employees' intra-organizational social networks is positively related to employee retention, while extra-organizational social networks are (contrary to theoretical expectations) only weakly associated with turnover intention. Understanding the structure and operation of these networks is of practical importance for managers as important exogenous contextual features that may interact with management strategies and tools, as well as potentially endogenous outcomes that they can seek to shape and manage over the medium and long term to improve public employees' performance. These networks comprise not just other individuals but also other organizations, professional fields, and social communities, and thus represent both within-and across-level relationships. However, there exists relatively little empirical literature examining the effects of networks on actual service delivery outcomes (as opposed to intermediate outcomes like organizational commitment), and still less that rigorously evaluates the impacts of management strategies on network structure or outcomes, so this is an important gap for future empirical work.

(ii) Leadership as relationship
There is a growing acknowledgement by public-sector leadership scholars that relationships are important as a critical mechanism for connections between individuals within organizations and systems. While much leadership scholarship has viewed leadership as a matter of leaders' individual characteristics or styles of leadership, such as the debate between transactional and transformational leadership styles (Trottier et al., 2008), recent literature has increasingly discussed 'relational leadership' approaches. In contrast to the focus on the leaders themselves, relational leadership emphasizes: the importance of understanding reciprocal relationships between leaders and followers; leadership as both 'context-creating and context-dependent' (Ospina, 2008;Wallace and Tomlinson, 2010); and the idea that leadership emerges from particular situations as well as the context within which it operates (Gittell and Douglass, 2012;Gittell, 2012;Ospina, 2008). This shift towards relational approaches to leadership is partly in response to the limitations of individual-focused leadership styles (which serve vertically within bureaucratic forms) in addressing the evolving horizontal needs of public-sector institutions to work laterally within the organization and collaboratively with other institutions (Drath et al., 2008).
To delve further into how relational leadership approaches are increasingly emerging as critical components in the integrative application of leadership styles, the remainder of this sub-section explores literature on the intersection between leadership and relationships. We organize our discussion according to three questions as depicted in Table 3 The question 'what should a leader do?' has been most prominently explored in the literature by examining the hypothesis that good leaders lead satisfied and motivated followers and manage organizations that successfully transition through change (Moynihan et al., 2012;Paarlberg and Lavigna, 2010;Trottier et al., 2008). At the operational level, evidence shows that good leadership entails ensuring that followers have the encouragement, support, necessary resources, and skills to perform well, and at the executive level, they work towards facilitating change by instilling values and culture that drive organizational change (Van Wart, 2013). This focus of recent literature on leaders enabling and guiding organization members (as opposed to the common bureaucratic view of a leader as a holder of decision authority atop a hierarchy) emphasizes Behaviour of a leader must balance: ▪ Management theory (Fernandez, 2008) ▪ Ethical leadership (Ciulla, 2016) ▪ LMX leadership theory (Hassan and Hatmaker, 2015) ▪ Achieving competing demands (task completion vs people management) ▪ Managing interdependent relationships Where should leadership come from?
Source of leadership emerges from: ▪ Horizontal leadership (Ospina, 2008) ▪ Collaborative leadership (Ansell and Gash, 2008;Crosby and Bryson, 2010 the idea that good leadership is defined primarily by strong relationships, rather than the attributes of the leader in isolation (although the two are, of course, related). With respect to how a leader should act, the literature suggests that leaders can benefit from creating high-quality relationships with followers, because the positive nature of the relationship can motivate followers both to undertake their prescribed roles and to take on extra roles which contribute to organizational effectiveness (Hassan and Hatmaker, 2015). At the same time, leaders must strike a balance between building these relationships and fulfilling the various demands emerging from the multiple roles they play (Pitts, 2005;Fernandez et al., 2010). This balancing act emphasizes the idea discussed above that managing networks (and managing in networks) is a key aspect of the complexity of modern leadership, especially in the public sector (Provan and Lemaire, 2012). In terms of these multiple demands, research has therefore explored the challenges that managers face to balance the pressures of task accomplishment and people management (Fernandez, 2008), to deliver results, communicate goals, and ensure follower diversity and satisfaction (Pitts, 2005), while contending with a changing environment, evaluating progress, and acting ethically with integrity (Ciulla, 2016). In terms of understanding the dynamics of relational leadership, studies on leader member exchange (LMX) leadership theory (which considers the quality of relationships between the leader and followers as the primary unit of analysis) have shown positive organizational outcomes on work meaningfulness (Hassan and Hatmaker, 2015) and organizational commitment (Tummers and Knies, 2013), which emphasizes the idea that building strong relationships is in many cases a way for leaders to achieve objectives rather than being seen as simply another competing demand.
The question of where within the organization leadership should come from has traditionally focused on leaders as individuals at the apex of hierarchies (Hennessey, 1998;Kaiser et al., 2008). However, due to challenges of contemporary governance, public organizations are increasingly seeking out collaborative arrangements that rely on the creation of horizontal relationships to meet demands for collaboration across agencies to conduct a myriad of functions. For instance, one of the dilemmas of leadership across networked systems is how to cultivate cooperation among public organizations to perform interdependent and interconnected tasks such as delivering services to neglected city communities (Kenis and Raab, 2020), sharing power among partner organizations to allow for devolved and decentralized decision making in cross-sector collaborations (Crosby and Bryson, 2010), and sharing information for effective coordination across teams and networks (Jackson and Stainsby, 2010). In all these instances, forms of horizontal, collaborative, and adaptive leadership styles, which are essentially relationship-based, have emerged as critical to the success of collaborative efforts of public organizations (Ansell and Gash, 2008;Emerson et al., 2012). In this view, leadership is collective and the role of individual leaders is to manage these relationships, so leadership cannot come from the individuals atop organizational hierarchies alone.
The complexity of leadership as a construct and the importance of context in determining optimal leadership strategies makes it difficult to draw simple recommendations or summaries for leaders. However, recent research on leadership has overall put forward a view of optimal leader behaviour as less hierarchical, more supportive, more flexible, and more relationship-oriented than usually perceived, especially in bureaucratic settings. Adding more theoretical specificity to this broad recommendation and exploring how relational leadership styles might differ in effectiveness across contexts is a key task for future research in this area.

(iii) Individuals' relationships to their jobs
In both economics and public administration, roles and tasks have typically been viewed as given or fixed, and the responsibility of people management is to fill those jobs with bureaucrats with the appropriate personal characteristics or to establish processes that appropriately enable and incentivize them to undertake these pre-defined tasks. Foundational texts in both disciplines have theorized how the properties of tasks or roles (e.g. output or outcome observability, multi-tasking, coordination requirements, ease of monitoring) might interact with management practices and affect performance (e.g. Kaufman, 1960;Wilson, 1989;Dixit, 2002). The implications of the differing nature of tasks across different institutions and different roles have permeated throughout scholarship on individuals, organizations, and teams-as the previous sections of this paper have discussed. Since public-sector organizations typically cannot unilaterally change their core tasks in the same way a private firm might, the nature and organization of these tasks has been widely taken as given, and attention has thus focused on better staffing and managing these roles.
In sociology and management, however, the idea that the nature of roles is exogenous to the agents populating them has come under challenge. For instance, an influential literature on job crafting shows how individual workers actively compose both what their job is physically, by changing a job's task boundaries, what their job is cognitively, by changing the way they think about the relationships among job tasks, and what their job is relationally, by changing the interactions and relationships they have with others at work…Job crafters act upon the task and relational boundaries of the job, changing their identity and the meaning of the work in the process. In doing so, job crafters create different jobs for themselves, within the context of defined jobs. (Wrzesniewski and Dutton, 2001, p. 180) While empirical studies of job crafting have focused mainly on private firms or NGOs (e.g. Berg et al., 2008;Tims et al., 2013;Rudolph et al., 2017), Wrzesniewski and Dutton give examples of hospital cleaners who reconceive their jobs to integrate themselves into patient care and nurses who manage task boundaries in order to centre care on patients, and Leana et al. (2009) discuss job crafting among teachers.
Recognizing the empirical prevalence of job crafting expands the scope of bureaucratic discretion beyond merely choosing how to execute a pre-defined task to choosing what tasks to accomplish, and makes organizational design a collective and ongoing process rather than a one-off management decision. This could be viewed as increasing the number of dimensions along which unmotivated employees can shirk, but perhaps more importantly also increases the potential value of empowering motivated employees to effectively exercise their discretion-particularly in the complex, coordination-intensive, high-discretion context of many public-sector tasks. For instance, public managers can thus use job design principles to understand jobs as a collection of relationships as well as a collection of tasks, recognizing that employees are motivated to the extent to which they perceive that their jobs affect the well-being of others (Hackman and Oldham, 1980;Grant, 2007;Paarlberg and Lavigna, 2010). For example, job tasks can be defined in such a way that public employees can have an interaction with their direct beneficiaries, such as the general public or internal customers/colleagues, or other agencies (Paarlberg and Lavigna, 2010). Indeed, bureaucratic entrepreneurship and innovation (e.g. Teodoro, 2009Teodoro, , 2011 could be viewed as an extreme example of job crafting, and there are obvious implications for the literature on job satisfaction and engagement discussed in section II. Yet we have little theory or evidence about how job crafting manifests itself in core public-sector roles, or about how managers should structure their organizations to facilitate performance-enhancing (and minimize performancedetracting) job crafting.
A broader implication of this relational perspective is that individuals' performance in a particular role or task is not fully determined by either the characteristics of the individual or the task, or even by static relationship measures such as person-organization fit. Instead, individuals have active and dynamic relationships with their jobs-much as they operate within active and dynamic networks across and within organizations, active and dynamic relationships with their supervisors, and active and dynamic relational contracts and organizational cultures. The individual-job relationship does not fit simply into the distinction used elsewhere in this paper of relationships as being either across or within levels of analysis, as the 'job' is not an actor but an inanimate abstraction. While this terminology might thus push the boundary of what 'relationships' are, exploring the potential for studying and theorizing such relationships (and how to change them) may provide a fresh perspective for scholars and policy-makers alike.

VI. Conclusion
Our review demonstrates that there is a rich body of evidence about the effects of different people management practices that spans a wide range of academic disciplines and empirical contexts. Although the effectiveness of management practices is inherently variable across organizations, there is enough accumulation of evidence to be able to make some generalizations, such as the importance of intrinsic motivations and other insights highlighted in Table 1. Thus, effective people management requires far more than simply establishing a set of formal personnel rules and regulations. However, our review also shows that there still exist numerous evidence gaps, particularly once one moves away from formal management practices and financial incentives. In many cases, the best evidence available derives from private-sector firms and thus may have debatable applicability to public-sector contexts. By reviewing this evidence, we hope to have provided researchers and practitioners alike a clearer understanding of the evidence that does-and does not-exist on these topics, to serve as a guide for further research and practical experimentation.
While we have organized our review according to the four lenses of the individual, organization, team, and relationship, there are obvious interconnections across and within these different units of analysis in terms of how people management practices are related to performance. The effect of performance management practices adopted and implemented at the organization level, for example, is likely to depend on individuals' responses to incentives and feedback, on the effect of these responses on collaboration among work teams, and on how the leaders that carry out the policy relate to their employees. Moreover, substantial benefits might emerge when these practices are integrated and comprehensively implemented across different levels (organizations, teams, individuals, and relationships) simultaneously. While no one would deny the existence of these interconnections, and some empirical studies do take account of them either implicitly or in an ad hoc manner, we found few theoretical frameworks that clearly articulate the connections between these levels of analysis. Both at the conceptual and empirical level, then, there is immense scope for a new wave of research to enrich our understanding of how to piece together the disparate shards of evidence available to us.
Finally, we suggest that there is immense potential to develop further evidence on how management practices can affect relationships-as distinct from the actors that comprise them. Among our four lenses for examining people management, the lens of the relationship was the most unorthodox inclusion, as most theoretical and empirical treatments take actors (whether individuals, teams, or organizations) as their primary units of analysis. Yet we found ample evidence of how relationships across and within classes of actors mediated or moderated the dynamics of people management and performance, and identified three sets of relationships that could themselves be understood as units of analysis. In other words, relationships are not epiphenomenal to individual behaviour and organizational performance, they are often active elements. While foregrounding relationships through a relational approach to public management may be unfamiliar to many researchers, relationships and exchange are the building blocks of both social and economic interaction, so in this light emphasizing them seems only natural. Of course, this suggestion raises more questions than it resolves, and further specifying what such an approach (or approaches) might look like represents a long-term agenda, but we are confident that efforts to do so will lead to scholarship on public management that is both more nuanced and more useful.