Street-level discretion, personal motives, and social embeddedness within public service ecosystems

Drawing on the sense of community responsibility concept, we explore the enterprise policy ecosystem in an extensive qualitative study of Scotland. We present a processual model which explains how policies are shaped in an on-going dynamic through street-level managers ’ individual agency. Our findings reveal that driving the process is an interplay between personal motives (compassion, relational strength, esteem, coherence) with a social frame of reference (policy group, locality, public organization) which is based on embeddedness within specific policy contexts. This interplay guides how managers translate policy as either an opportunity or a threat which then directs how they enact their discretion to adapt, advocate change, or resist implementation. This process offers an explanation as to how situated value is created for specific policy areas within public service eco-systems. The implications are discussed in relation to the existing literature on policy implementation. Evidence for


INTRODUCTION
There is a growing body of literature detailing the influence that street-level workers (SLWs) have in public policy processes (Cohen & Frisch Aviram, 2021;Frisch-Aviram et al., 2020;Lipsky, 2010;Tummers & Bekkers, 2014).This literature mainly explores the discretion front-line workers (FLWs) have when implementing public services and the various behaviors they adopt as they explore the autonomy afforded to them to influence the way policy is delivered to clients (Cohen & Klenk, 2019;Lavee & Strier, 2019;Tummers & Bekkers, 2014).It also highlights that street-level managers (SLMs) have a great deal of discretion in their roles and can shape the way policy is implemented (Cloutier et al., 2016;Evans, 2016;Gassner & Gofen, 2018;Klemsdal et al., 2022).This departs from the notion that SLWs are impartial and prefer rule-following and routinization (e.g., Assadi & Lundin, 2018;Blau, 1963;Rothstein & Teorell, 2008) and poses the question of why those implementing policy choose to attempt to shape it in various ways (Cohen & Hertz, 2020;Cohen & Klenk, 2019;Frisch-Aviram et al., 2020).
Researchers have historically acknowledged that those working in public organizations are extrinsically motivated to perform work that is deemed worthwhile for society (Kim & Vandenabeele, 2010;Nowell et al., 2016;Perry & Vandenabeele, 2015).Increasingly, however, ecosystems of private, third-sector, and nongovernment organizations are collaborating to implement various policies to enhance public value creation in multiactor constellations (Lindsay et al., 2014;Osborne et al., 2022;Vermeiren et al., 2021).This emerging research has begun to understand more about complex policy contexts and the different street-level organizations (SLOs) that are embedded within them (Cui & Osborne, 2022;Frisch-Aviram et al., 2020;Lapworth et al., 2018;Lindsay et al., 2014).
However, for the managers of SLOs within policy ecosystems less is understood of their motivations for shaping policy.Indeed, Lapworth et al. (2018) find a wider array of motives in voluntary sector employees than typically found in public sector employees.Exploring the motives for and how and why SLMs of SLOs shape policy in situ, therefore, can provide a deeper understanding of policy implementation within public service ecosystems (PSE) (Cohen & Hertz, 2020;Klemsdal et al., 2022).This is particularly pertinent for enterprise policy, the context for this paper, which is a key pillar in many governments' efforts to grow and develop economies (Audretsch et al., 2020;Terjesen et al., 2016).Indeed, enterprise policy is a popular way for governments to meet social challenges and recover from economic recession (Wright et al., 2015), making it particularly relevant as governments look to recover from the COVID-19 pandemic.More frequently, governments are repositioning themselves, moving away from being direct providers of enterprise policy to facilitators of ecosystems with mixes of public, private, and thirdsectors organizations that implement policy at the street-level (Autio & Levie, 2017;Knox & Arshed, 2022).This places a greater demand on SLMs who are located at the "intersection of formal policy-making, local target populations, and ever-changing and highly contextual work" (Gassner & Gofen, 2018, p. 552).
Little attention has been focused on the role and motives of managers within SLOs and how their involvement impacts policy implementation (Arshed et al., 2016;Gassner & Gofen, 2018;Osborne et al., 2022).The relationships between discretion and various institutional, organizational, and individual-level factors have already started to be identified for FLWs (Cohen & Gershgoren, 2016;Tummers & Bekkers, 2014).However, this literature does not consider the situational and temporal nature of policy implementation and the process through which policy changes as it moves through multi-actor constellations (Klemsdal et al., 2022).Considering this gap, we seek to address two research questions: 1. How do SLMs involved in public service ecosystems influence the diffusion of policies as they are implemented?2. Why do SLMs in public service ecosystems look to shape policy implementation?
To frame our study, we draw on the PSE (e.g., Osborne et al., 2022) and sense of community responsibility (SOC-R) concepts (e.g., Nowell et al., 2016).We use an inductive case study research design focusing on Scotland, UK, to address our research questions (Gioia et al., 2013).We present a theoretical model which details how and why SLMs translate and enact on various policy directives based on their embeddedness within specific policy contexts.In doing so, we contribute to the existing literature detailing how street-level discretion and motives influence policy implementation.

Policy mobility and public service ecosystems
There are two current overlapping aspects that highlight the increasing mobility of public policies.First, the extant literature has emphasized the considerable discretion FLWs have to enhance public services for clients (Hupe & Hill, 2007;Tummers & Bekkers, 2014).Through their discretion, they effectively become the public policies they carry out, shaping them in the process of implementation (Lipsky, 2010).This view highlights the ability of FLWs to influence the way policy is delivered to clients through their practice.It challenges the traditional perspective which considers FLWs as passive actors who implement policy through standardized practices as it is passed down from senior government officials and elected policymakers (Assadi & Lundin, 2018;Blau, 1963;Brodkin, 2011;Fuertes & Lindsay, 2016;Stensöta, 2011).Despite SLMs being originally mitigated from Lipsky's work, they too have been highlighted as having significant discretion when it comes to implementing policies (Cloutier et al., 2016;Evans, 2016;Gassner & Gofen, 2018;Klemsdal et al., 2022).
Second, public services are being increasingly delivered through "loosely connected" and complex networks, which involve groups of public, private, and third-sector organizations (Lindsay et al., 2021;Osborne et al., 2022;Petrescu, 2019;Vermeiren et al., 2021).As such, the power of non-governmental organizations that are delivering front-line services to influence policy formulation is becoming more frequently documented (e.g., Fyall, 2016;Lindsay et al., 2014).Their involvement in delivering public services also provides different opportunities to enact influence upon the policy process (Lindqvist, 2019) and drive value creation (Lindsay et al., 2021).As such, policies can be regarded as mobile, constantly evolving, and influenced by many different SLOs (Fairbanks, 2019;Lovell, 2016;Xheneti, 2020).
In this article, we draw on the PSE concept which highlights the "inherently complex" nature of public service delivery, where interdependent actors and organizations "integrate resources and exchange services" (Leite & Hodgkinson, 2021, p. 4).The PSE approach moves away from the "transactional and linear approach" to policymaking and implementation associated with New Public Management (NPM) and "towards a relational model" where value is co-created amongst multi-actor configurations (Strokosch & Osborne, 2020, p. 436).The concept draws upon existing post-NMP models, such as New Public Governance (NPG), which advocates for the use of inter-organizational networks and public service systems as opposed to top-down bureaucratic control (Lindsay et al., 2014;Osborne, 2010).However, while NPG emphasizes efficiency through collaboration and wellfunctioning networks (Dickinson, 2016), PSE advocates for public value co-creation as the focal outcome (Osborne et al., 2022;Petrescu, 2019).Furthermore, critics highlight that governance networks are centrally controlled, whereas ecosystems are loosely connected, mutually dependent, and without central control (Kinder et al., 2021).This places significantly more agency in the hands of SLMs and advocates for more processual analysis of public service delivery (Sandfort & Moulton, 2020).
The PSE concept has emerged as an integrative framework that exists across four different levels and seeks to understand post-NPM public service delivery (Osborne et al., 2022;Trischler & Westman Trischler, 2022).The macro-level highlights that societal values and institutional norms impact upon value creation through public services (Osborne et al., 2022).This includes the impact of institutional values, rules, and norms upon the context and processes of value creation (Huijbregts et al., 2022;Osborne et al., 2022).It also includes the extent to which public services create societal or public value (Bozeman, 2019;Osborne et al., 2022).The meso-level concerns the organizational processes and norms of policymaking communities, or "multi-actor" constellations, and their impact on value creation (Trischler & Westman Trischler, 2022).The micro-level acknowledges the important role of individuals and individual action, while the sub-micro level expresses the impact of these individuals' values, motives, and beliefs (Osborne et al., 2022).In this study, we focus our attention on the micro-and submicro level as we are concerned with why SLMs are motivated to influence policy implementation and how their actions can generate value within specific contexts.
The PSE approach to public administration and management places a great deal of focus on agency and the role SLMs play in pulling together, diffusing, and enacting policies (Arshed et al., 2019).Ultimately, the agency of SLM interplays with context which creates the conditions in which policy can be implemented in different ways (Vega et al., 2013).At different points in time, and in different contexts, the opportunity to enact different practices is presented, thus the agency of different SLMs is situational and context-dependent (Klemsdal et al., 2022).This means various SLMs are afforded autonomy and discretion when implementing policy (Brodkin, 2011;Karlsson, 2019;Tummers & Bekkers, 2014).They can enact discretion due to the way policy is written and performed (Lipsky, 2010).That is, SLMs are able to interpret, translate, and synthesize policies as they move horizontally and vertically through the PSE.

Street-level discretion
There is an emerging body of work detailing the ability of various SLWs (both FLWs and SLMs) to influence policy at the street level (Cohen & Frisch Aviram, 2021).FLWs have been shown to use the discretion afforded them, and their position within the policy process, to influence how policy is formulated, as well as implemented.SLMs who were originally "bracketed-off" from Lipsky's work but have also been found to enact discretion in their roles (Evans, 2016;Klemsdal et al., 2022).Existing research has identified different types of work that SLWs do to lobby, form coalitions, and innovate service delivery (Arnold, 2015;Lavee & Cohen, 2019).This line of literature has also begun to identify the factors that can shape SLWs' involvement in policy entrepreneurship across three levels-institutional, organizational, and individual (e.g., Cohen & Golan-Nadir, 2020).
At the institutional level, the existing literature suggests that political, regulatory, and culture can shape discretionary behavior (Brodkin, 2011;Cohen, 2018).In this sense, a SLW will look to influence the implementation of a policy if they believe it is socially acceptable to do so.At the organizational level, the existing research has identified numerous factors that influence SLWs' behavior, including the influence of peers, managers, and the availability of organizational resources and incentives (Keiser, 2010;Tummers et al., 2012).At the individual level, the extant literature suggests personal preferences and interests can explain SLW's motivations for enacting discretion.This is despite Lipsky's work indicating that SLWs enact discretion primarily due to the contingencies they face at the organizational and institutional levels.On one hand, a body of literature highlights that SLWs can be motivated by self-interest which can include professional benefits, financial incentives, and a sense of selfefficacy (Bovaird et al., 2023;Brodkin, 2011;Cohen & Gershgoren, 2016).On the other hand, a greater body of literature emphasizes the pro-social motivations for SLWs' behaviors (Hupe & Hill, 2007;Tummers & Bekkers, 2014).
This literature generally acknowledges that SLWs are extrinsically motivated to perform work that is deemed worthwhile for society (Kim & Vandenabeele, 2010;Nowell et al., 2016;Perry & Vandenabeele, 2015).Much of this literature focuses on Perry's (1996) seminal public service motivation (PSM) construct and four dimensions-attraction to public policymaking, commitment to the public interest and civic duty, compassion, and self-sacrifice.This construct, however, does not capture the situational nature of policymaking and the intrinsic motivation of policy actors (Esteve & Schuster, 2019;Nowell et al., 2016).It does not consider the contexts in which SLMs operate and the extent to which an individual identifies with their policy context (Perry & Vandenabeele, 2008, 2015).To address this, we turn to the SOC-R perspective which stresses the importance of community to public service provision and how the motivation to take responsibility is fostered within situational policy contexts.

Sense of community responsibility
SOC-R is defined as "feelings of duty and obligation to take action to advance the well-being of a specific group and its members that is not directly rooted in an expectation of personal gain" (Nowell et al., 2016, p. 665).It draws on social identity theory (e.g., Ashforth & Mael, 1989), referring to one's feeling of belonging, of mattering to and sharing faith with a group (Boyd & Nowell, 2020).Here, individuals act to protect or enhance the benefits of a collective (Nowell & Boyd, 2010).In turn, the social collective is a resource for meeting an individual's psychological and physical needs (Boyd & Nowell, 2020).
SOC-R states that individuals develop values and beliefs through engagement and embeddedness within various communities and relational networks (e.g., organizations, geographic territories, or interest groups) which act to shape their identities (Nowell et al., 2016).Individuals are, therefore, motivated to act in ways that affirm their identities within a community to retain psychological coherence.The driver of SOC-R, therefore, is not an expectation of a personal gain, but a need for psychological coherence between identity and action (Boyd & Nowell, 2020).
There is debate on whether SOC-R and PSM are integrated or distinct constructs (Brincker & Pedersen, 2020;Toubøl et al., 2022).On one hand, both are predominately based on the premise that through social interaction individuals develop personal values, which acknowledge social identity as the core driver of behavior (Boyd & Nowell, 2020;Nowell et al., 2016).On the other hand, distinctions have been offered.PSM is acknowledged as a broad construct where individuals have predispositions related to self-sacrifice, compassion, commitment, and duty to public service (Perry, 1996).The SOC-R concept, however, does not emphasize a predisposition toward certain motives (Nowell et al., 2016).SOC-R can account for a larger array of individual-level motives beyond selfsacrifice and compassion, considering both personal and social-specific motives.
Early empirical evidence suggests that SOC-R can be useful to explain what drives various behaviors, such as employee engagement (Boyd et al., 2018), organizational citizenship (Boyd & Nowell, 2017), visionary leadership, and political influence (Pedersen et al., 2020).However, what is missing from this literature is a processual understanding of how these constructs interact and play out in situ (Kinder et al., 2021;Sandfort & Moulton, 2020).This construct, therefore, offers a promising lens as to which to investigate how and why SLWs look to shape policy implementation within PSEs.This is particularly pertinent as the wide array of SLWs within PSEs likely differ in their motives, values, and beliefs (Lapworth et al., 2018;Osborne et al., 2022).

METHODS
We adopted a grounded, interpretative methodology designed to understand the motives and influence of SLMs implementing policy.Our study is built upon a single in-depth case detailing the inner workings of enterprise policy in Scotland, UK.Our research design was inductive, and the purpose was to build a conceptual explanation for how and why individual actors within the PSE influenced the implementation of enterprise policies (Gioia et al., 2013).

Research context: Enterprise policy ecosystem in Scotland
The UK was one of the first countries to adopt NPM practices to reform government in the 1980's and 90's (Cooper et al., 2022).However, more recently, intense post-NPM-style reforms based on e-government, transparency, citizens' engagement, collaborative governance, and coordination are prevalent (Christensen & Laegreid, 2022).This extends to Scotland, a devolved government since 1999, where enterprise policies have typically been delivered under a national umbrella through local governments and national agencies since about 2007 (Brown & Mason, 2012).The Scottish Government's overarching aim is in "delivering a more supportive business environment, focused on delivering a competitive tax system, reducing costs for business and allowing more businesses to get the help they need to flourish" (Scottish Government, 2019).
However, the public sector is more frequently turning to a munificent ecosystem of private and third-sector partners present within Scotland to formulate and implement public policies aimed at developing entrepreneurship (Scottish Enterprise, 2019).This rich ecosystem of providers had a particularly important part to play in supporting, implementing funds, and complementing the Scottish Government's response to COVID-19.This ecosystem approach has been shaped by two important policy developments.In 2011, The Christie Commission on the future delivery of public services expressed a recommendation to move toward a whole system approach, where public, third, and private sectors work together to drive efficiency (Cairney et al., 2016).In 2013, the Scottish Government launched its Scotland CAN DO strategic plan to become a world-leading entrepreneurial and innovative nation.This strategic plan outlined a collective approach that "that brings companies, universities, public agencies and customers together to exploit more of the opportunities that drive growth" (Scottish Government, 2013, p. 17).
Additionally, in 2014, the Scottish Government focused on a more concentrated regional approach to economic development (Clelland, 2020).Stemming from a 2011 UK Government policy introducing City Region Deals, agreements have been made between the UK Government, Scottish Government, Local Authorities, private partners, universities, and the voluntary sectors to encourage local economic growth, which to date covers each Local Authority in Scotland (Copus et al., 2022).Given this collaborative and decentralized turn over the last decade, the enterprise policy ecosystem in Scotland is a particularly fruitful context to explore.

Data collection
Our extensive qualitative study took place between 2018 and 2019, pre-COVID-19 pandemic.In line with best practice for inductive theorizing, multiple data sources were utilized (Gioia et al., 2013).Our primary source of data was semi-structured interviews, which were triangulated with participant observation from policy meetings, event participation, and document analysis.Table 1 presents the data collected, the main contextual settings for each of our informant groups, and the identifying code we allocated to each group.
Our unit of analysis was the organization of streetlevel public service delivery with a specific focus on SLMs • 8 semi-structured interviews with local authority representatives in GCR, 3 in TCR, and 2 in FVR (n = 13).• 9 semi-structured interviews with contracted organizations in GCR, 1 in TCR, and 2 in FVR (n = 12).• 11 semi-structed interviews with sponsored organizations in GCR, 4 in TCR, and 2 in FVR (n = 17) • Reviewed multiple reports and strategic documents outlining policy approach to economic development and enterprise support.• Field notes from multiple ecosystem coordination events in each region.• Many ESOs provided the researchers with reports, strategic documents and media articles outlining their approach and impact.• The researchers collected information from websites and archives regarding ESOs support offerings, objectives, activities, and impact.
Front line workers (FLW) • Advisors from a national public sector enterprise programme that provides various support to businesses.Programme is delivered by local authorities (BA).
• Review of multiple internal documents outlining job role, performance, and monitoring.

Policy managers (PM)
• Enterprise policy workers in the Scottish Government and senior officials in economic development agencies that have a strategic remit, focusing on business growth and innovation and upskilling the workforce (NG).
• Field notes from several policy meetings within each case and workshops between Scottish Government, National Economic Agencies, and regional development teams.• Review multiple reports, ministerial briefings and strategic documents outlining policy.• Review of The Scottish Parliament archival database, covering Parliamentary and Committee hearings.

Service users (SU)
• Perceptions of how business owners use and navigate the enterprise support ecosystem to get the resources they need to start and grow (E).
• Field notes from multiple "ecosystem coordination" events in each region.

External stakeholders (ES)
• Organizations that support business owners to start and grow.Including social enterprise, education, growth, network organizations, incubation and acceleration programs and investment and finance groups (ESO).
• 8 semi-structed interviews in GCR, 5 in TCR, and 2 in FVR (n = 15) • Field notes from multiple ecosystem coordination events in each region.• Many ESOs provided the researchers with reports, strategic documents and media articles outlining their approach and impact.• The researchers collected information from websites and archives regarding ESOs support offerings, objectives, activities, and impact.
who operated the various organizations that implemented enterprise policy.Our approach looked to gain insights from multiple informant groups, however, to ensure triangulation.We conducted a total of 110 semistructured interviews with different groups that make up the enterprise policy ecosystem in Scotland.In line with our research focus, our main informants were SLMs from local government delivery services (n = 13), enterprise support organizations with contracts to implement public services (n = 13), and organizations with sponsorship to deliver their own enterprise support services (n = 16).We triangulated this data with other informant groups, including FLBs (n = 10), policy managers (PM, n = 9), service users (SU, n = 34), and external stakeholders who provided non-public services within the ecosystem (n = 16).
Our sampling procedure involved initially purposefully targeting key SLOs, identified in key policy documents.From these initial semi-structured interviews our sampling snowballed as informants identified other important informant groups in the ecosystem.Each semi-structured interview lasted between 30 and 90 min and followed a broad thematic protocol aimed at capturing their role, influence, and motives for engaging in the policy process.

Data analysis
Data was analyzed by following guidelines to bring rigor to inductive theorizing, where the theoretical perspective emerges from the data (Gioia et al., 2013).Initially, the first author categorized observation notes, policy documents, and interview transcripts into initial open codes (Strauss & Corbin, 1998).These codes were then discussed iteratively between the authors and refined into a more manageable number (Charmaz, 2006).These first-order codes were then organized into more abstract secondorder constructs which explained how individuals influenced the diffusion of policy and provided an indication of the reasons as to why they may do this.We then distilled our second-order categories into aggregate dimensions that conceptually presented a process of how and why individuals influenced policy diffusion in the PSE.The analytical process is presented in a data structure in Figure 1.Further evidence for the second-order constructs of each aggregate dimension can be found in Supplementary Appendix Tables A1-A4.

FINDINGS
The four main dimensions that emerged from our analysis are the following: personal motives, social frame of reference, policy translation, and enacting discretion.The "personal motives" referred to individual unique professional drivers.The "social frame of reference" referred to the specific policy contexts that individuals identify with."Policy translation" referred to how various policies were interpreted based on the (mis)alignment of personal motives and social frame of reference.Finally, "enacting discretion" referred to the actions that individuals could take to influence policy.
Figure 2 shows how these different aggregate dimensions interlink conceptually.Driving SLMs' discretion in implementing policy is an interplay between personal motives with a social frame of reference which is based on their embeddedness within specific policy contexts.The outcome of this process was an adapted policy agenda, which created value for local policy contexts.In the following section, we develop our analysis to explain how and why individual actors in the PSE influence the diffusion of policy.

Personal motives
We found four different personal motives-client-centered compassion, relational strength, esteem, and coherence.The prevalence of these motives amongst informant groups is presented in Supplementary Appendix Table A5.All motives were equally prevalent amongst SLMs, except for esteem which was less frequently identified in the analysis.These motives related to each manager's own unique drivers, which motivated them when delivering public services and interacting with other policy actors.Example evidence for these motives is presented in Table A1 in the Online Appendix.The first personal motive, client-centered compassion, was demonstrated by street-level workers whose primary driver was to help clients.These participants frequently emphasized the importance of helping clients and focused on creating personal bonds, experiences, and interactions with clients: "We are here to help businesses, students and universities, our paying customers.They are our first and foremost priority.We're not here to serve egos, we're not here to pat ourselves on the back, we're not here to justify our jobs, we're here to serve our customers and that's why we exist.To help our customers solve their problems.If we [rub people] the wrong way or something, that's just too bad.We're here to make a difference" (TCR-ESO-04).
The second personal motive, relational strength, referred to informants that created narratives placing them in the center of a social group.The motivation was to create attachments with people they interacted with and to build their own social capital.In interviews, informants who expressed this motive frequently looked to emphasize the importance of forming strong relationships: "That's a different approach.We're not going to send them the survey, we're just going to sit down and talk to them, because we've got relationships with them already.They're a bit more established.They've got completely different sets of issues and challenges in the business.And that's what we're going to do with those guys, and we're going to inform them of what we can do to support them" (FVR-ESO_PC-02).
The third personal motive we found was a need for esteem.Informants would emphasize their own individual achievements and successes.They would frequently cite their own credentials, experience, and knowledge, claiming client successes as their own.The focus would be on their own career-path, means of advancement, and reputation within their job role: "It comes from experience.It comes from knowledge.I'm an experienced adviser; I've been doing this now for 13, 14 years" (GCR-BA-04).
F I G U R E 1 Data structure.

PUBLIC ADMINISTRATION REVIEW
The final personal motive we identified was a need for coherence.These informants seemed to be driven by a need to reduce uncertainty, and having clear concise remits and structures.They would focus on maintaining order and describe in-depth how they organized their jobs.They looked for consistency, mainly by stressing how rules can be followed to create homogenous processes and practices.Clients were not referred to individually, but collectively, in which they could be categorized and processed.These actors would struggle when they found aspects of their work disorganized: "Although, personally, coming from outside, that seems a bit woolly.I'm trying to find somewhere that says, these are the criteria for putting people into the growth.So, I think that would be a bit woolly.It seems to be, I've seen it written down somewhere but it's difficult to find a hard and fast rule" (GCR-BA-02).

Social frames of reference
We found three different social frames of reference amongst our informants-policy groups, local place, and public sector organizations.The prevalence of these frames of reference is presented in supplementary appendix Table A6.The social group frame of reference was more prevalent than local or public sector organization for SLMs.However, it was not identified as being a reference for FLWs.The local frame of reference was the least likely for PMs and public sector identification was the least likely reference point for external stakeholders.These frames of reference presented a sense of social belonging with which informants identified with.They provided a lens through which individuals directed their attention.Example evidence for these themes is presented in Table A2 in the Online Appendix.
The first social frame of reference was with regard to a specific policy group.These are related to both specific sectors (e.g., technology, cultural, energy), or specific of business (e.g., high-growth, womenowned, social enterprises).These various policy groups were regarded by informants as exceptional, and had unique challenges and characteristics which were deserving of special attention and support.There was also frequent mention of other organizations working to support these policy groups, which formed part of a policy community: "Our aim is to facilitate building the tech start-up community and allow a positive environment for companies to come off at any size, really, and to start their businesses, build on their ideas in a place where there's a community that can support them" (FVR-ESO_PS-01).
The second social frame of reference was about local geography.Here, informants demonstrated attachment and identification with geographically bound places.As such there was a commitment to and promotion of the local area.Policy actors would focus on the benefits they could provide (both economically and socially) to the area, and have a deep understanding of local networks and stakeholders: "[Our aim] is to create employment in Clackmannanshire, wealth in Clackmannanshire, encourage businesses to move into Clackmannanshire by developing a wide range of premises.We have a large property portfolio of industrial and office units.Very flexible leasing term and a very good property department.But we're not all about making money, we are here to encourage businesses to move in and expand within the Clackmannanshire area" (FVR-ESO_PC-01).
The final social frame of reference prevalent in our data was the public sector organizations where informants were employed.Informants identified with either the public sector organization they worked for specifically, or the public sector in general (encompassing public agencies and local authorities).These informants made frequent references to job descriptions, organizational processes and protocols, and the internal structures of organizations.They would focus on the public sector as the "in-group", and highlight other policy stakeholders (third-sector, private) as outsiders.As such, they very much viewed solutions to policy challenges as being delivered by the public sector: "What we're trying to do as a local authority is encourage the key sectors in the city to identify what their action plan would be for the next five years.What are the big challenges that their industries face over the next five years, and how can we come together as a public sector network, Scottish Enterprise, Skills Development Scotland, the council, to help support and realise some of their ambitions and address some of the challenges that these organisations have?"(GCR-LG-04).
Policy translation: Linking personal motives and social frame reference to discretion Recall we refer to policy translation as how various policies were interpreted based on the (mis)alignment of personal motives and social frame of reference.Policy translation, therefore, is the perception of a policy action, mandate, or interaction as either posing a threat or an opportunity to a policy actor.For example, one participant with a relational personal motive, and identification with a specific policy group (life science cluster), perceived the new Glasgow City Region deal as an opportunity to create wider connections within the policy ecosystem: "If there's a recognisable life sciences cluster or a set of expertise within a particularly academic context, we will run programmes to allow that technology to be transferred into a potentially commercial opportunity, obviously working very closely with the tech transfer offices in a number of universities.Specifically, within the Glasgow City Region deal area, that would be Strathclyde and Glasgow, would be our two biggest sources of the opportunities" (GCR-ESO_PS-02).
Adversely, another participant with a personal motive for coherence, and a local place-based identification, perceived the City Region deal as a potential threat to the local agenda and was wary of committing to an agenda outside of their own Local Authority: "I suppose we've had our own focus locally until the City Deal stuff crystallised in 2015.
We are pretty much sitting alongside parallel to what's happening in City Deal, taking the learnings from that but also looking after our own Local Authority" (GCR-LG-01).
The means by which individual actors translated policies ultimately guided their behavior.Example evidence for policy translation is presented in Table A3 in the Online Appendix.

Enacting discretion
Our informants at a local level detailed the processes and practices in which they had an influence on how national policies were delivered to create value in their context.We distinguished three overall types of discretion that were enacted: (1) adapting policies to the local context; (2) advocating different policies to influence top-down change at a national level; and (3) resistance practices that were enacted to maintain policies at both local and national levels.Example evidence for these discretionary practices is presented in Table A4 in the Online Appendix.
The first discretionary practice looked to adapt and innovate policy as it was transferred through to local organizations in charge of delivery.Several street-level workers would look to adapt the policy to fit in with the needs of their social frame of reference-either their local context, the wider policy context, or existing organizational practices.
These informants were proactive and were more willing to take risks and bend rules.Indeed, two mid-level Local Authority workers claimed "we are entrepreneurs" when talking about different policy innovations they had introduced to their local area.These informants would also attempt to setup new policies based on the needs of their frame of reference.For example, one street-level business advisor with a strong local frame of reference setup different support groups: "I've set up two, soon to be three, networking groups in the area.Again, that was as a result of feedback from clients, people looking for opportunities to work together, learn from each other etc.So, we've set up ED-Net which is [Local Authority] Networking.We have a very strong women in business group as well" (GCR-BA-01).
The second means to influence policy was through advocacy practices to influence top-down change.We found evidence that these informants would actively seek additional responsibilities, which would strategically position themselves within policy discussions.Frequently, they would work across areas to be heard in multiple arenas and had extensive social capital.One street-level business advisor, for example, attempted to influence the formation of a procurement policy to create value for a policy group she identified with: "I spent quite a bit of time obviously trying to find out more about the tender and try and influence our procurement colleagues to do it in such a way that it was a bit more SME friendly" (GCR-BA-05).
Finally, many informants demonstrated a reluctance toward change and focused their policy work on following set procedures.On the one hand, these informants would be resistant to policy changes and reluctant to engage with opportunities to influence policy formulation.On the other hand, they would inadvertently influence policy implementation by resisting any changes that were passed down.There was a proclivity for these informants to have esteem or coherence personal motives and an organizational frame of reference.They would act to "protect" the status quo, acting as a buffer between top-down and bottom-up changes: "Because I've got one or two members of my team who keep on talking about creative industries, and I'm like yes, but how many jobs is that going to create?So, to me, it's all very well.It's pie-in-the-sky thinking" (GCR-LG-05).

DISCUSSION AND CONCLUSION
Despite a growing body of work detailing the influence street-level workers have in shaping the policy process in public service ecosystems, there is a lack of research explaining why those implementing policy choose to attempt to shape it in various ways (Cohen & Hertz, 2020;Cohen & Klenk, 2019;Frisch-Aviram et al., 2020).To address this, the aim of this study was to explore individuals' motives for and how and why they shape policy in situ.
Our findings, presented in Figure 2, show policy actors have a personal motive, which drives their work within a specific social context.On a personal level, we found SLMs have a predisposition for compassion, relational engagement, esteem, or coherence which guides how they enact their role within the policy process.On a social level, we found policy actors identify with specific policy groups, local places, or public sector organizations, which act as a reference point to them.Based on this, they translate policy as either a threat or opportunity to advance both their personal and social agenda.This triggers how they enact discretion-either attempting to adapt, advocate change, or resist the implementation of policy to create value for their policy communities.Ultimately, they look to drive public service value creation by aligning policy with their personal motives ("how do I like to work") and with their social frame of reference (who do I like to work for?).
While compassion is a well-known motive within the public service literature (Perry, 1996), and has been found to drive bureaucrats to enact discretion (Masood & Nisar, 2022), relational engagement, esteem, and coherence are less frequently documented.Indeed, esteem can be regarded as a self-interested motivation where an individual is driven by being well-regarded in certain policy communities.It could be that managers of SLOs within PSEs are not motivated by financial or performance incentives, nor the compassion typically seen in public sector organizations, but by need for a sense of competence and importance within a specific social circle (Brodkin, 2011;Cohen & Gershgoren, 2016).It is also important to highlight that personal motives often interplay, and informants often express multiple personal motives to describe different behaviors.That is, in one situation an informant could act based on a desire to increase their esteem, while in another situation they could be driven by compassion.However, while it is important to recognize the heterogeneous nature of public sector workers in the PSEs, our findings indicated that a single dominant motive was present and, in combination with a social frame of reference, acted as a main driver for behavior.
The current literature also identifies organizational values, norms, and incentives as important for driving SLWs' behavior as they help to create a sense of what is socially acceptable (Cohen, 2018;Keiser, 2010;Tummers et al., 2012).Our findings suggest that the relationship between these organizational factors and individual behavior is indirect and interacts with personal motives and social identification.Organizational level factors can help to shape the level to which an individual identifies with a policy context.If supportive peers, managers, and wider culture are perceived favorably, this could influence the level of identification an SLW will have with an organization or policy group.This will, therefore, act to align their behavior with the specific interests of this social group.
Considering the emergence of the PSE approach to policy implementation (Osborne et al., 2022), and the emergence of wider varieties of organizations delivering public services (Vermeiren et al., 2021), these findings offer further insights into the individual-level drivers within wider PSE contexts and why attempts are made to adapt public policies.We have started to unpack the complex nature of PSE actor networks at a meso-level and how they are shaped by the sub-micro beliefs and microlevel actions of individuals (Osborne et al., 2022;Trischler & Westman Trischler, 2022).Considering the heterogeneous nature of PSEs, the findings in this study indicate that SOC-R can be used as a contextual viable framework to explore how the various actors within a PSE at a micro-level navigate complex policy implementation contexts and influence value creation.
Within the PSE, public value is co-created between multiple different actors (Strokosch & Osborne, 2020).Considering at a sub-micro level individuals have different motives, the co-creation of value can be highly situational (Osborne et al., 2022).Our findings indicate that the embeddedness within a policy context directs efforts to create value.This may have the potential to create a fallout where value is not generated in another interdependent policy context.For example, in our study, the case of the informant who lobbied colleagues to place SMEs at the heart of public procurement strategy; while this created value for the SME policy group, it may have come at the expense of another group (e.g., community-led, or multinational enterprise) and acted to destroy public value (Cui & Osborne, 2022).Policy actors, therefore, are likely to have both direct and indirect impacts when enacting discretion.Future research should investigate these interdependencies within PSEs to understand how public value can be co-created (and reduced) across policy contexts.
Our study makes two contributions to the policy implementation literature.First, we address a prevalent gap in the literature by detailing the motives, goals, and strategies of SLMs who are non-traditional bureaucrats from non-governmental organizations (Cohen & Frisch Aviram, 2021).We add an individual-level perspective to the literature that details factors, which facilitate streetlevel discretion, as well as extending understanding across multiple levels and covering multiple different actors within the policy process (Frisch Aviram et al., 2021;Lavee & Cohen, 2019;Osborne et al., 2022).We show that the motivation to enact discretion in public service ecosystems is driven by a combination of both personal motives and embeddedness within a specific policy group they identify with.This places managers of SLOs as embedded within certain policy contexts, working to shape policy to align with their own personal and social agenda.This, therefore, indicates that policies which are "touched" by multiple actors are likely to be particularly malleable.
Second, we contribute to the emerging PSE literature by showing that SOC-R as a motivational construct drives various behaviors (Boyd & Nowell, 2020;Nowell et al., 2016).This offers an alternative perspective to the dominance of PSM in public management, which denotes that public sector workers have a broad predisposition related to self-sacrifice, compassion, commitment, and duty to public service (Perry, 1996).The SOC-R construct can account for a wider array of specific personal motives, and indeed our findings show multiple different motives present within the PSE.Extending this, we add a social frame of reference which acts to guide individuals toward a certain community of interest.The combination of motives can describe why individuals are motivated to enact their discretion in various situations, connecting the micro and meso levels of the PSE (Osborne et al., 2022).
Our findings also have important implications for practitioners.From a normative point of view, there are benefits and drawbacks to managers of SLOs enacting discretion.On one hand, bottom-up insight into the needs of clients can be provided in policy implementation (Cohen & Frisch Aviram, 2021;Lavee & Cohen, 2019).On the other hand, SLWs can lack the broader pictures of public policy, often are at an information disadvantage, and their decisions can cause fall-out by negatively affecting an interdependent part of the policy process (Cohen, 2021).In this debate, we highlight the importance of understanding the interdependencies that exist in PSEs and the potential value destruction that SLOs can cause to wider service delivery when enacting agency.
One potential way to mitigate any disadvantages of individuals enacting their discretion within PSEs is to aim for alignment between individual and job roles.Public organizations could assess different stakeholders' personal motives and social frame of reference when considering policy collaboration, to ensure the motivations of delivery agents are aligned with their policy agenda.Public service organizations should also investigate how to best align job roles and responsibilities with their personal motives and social frames of reference.Training programs and strategies to foster a sense of community in different social settings can be valuable means to ensure alignment in policy workers' interests.It could be that mismatched placement of street-level workers could contribute to value destruction (Cui & Osborne, 2022).This is considering the importance of collaborative arrangements to deliver public services post-COVID-19 pandemic (Christensen & Laegreid, 2020;Huang, 2020;Steen & Brandsen, 2020), and the propensity for governments to turn to enterprise policy to meet social and economic challenges (Wright et al., 2015).
Our study was carried out in the specific context of enterprise policy in Scotland, where collaborative implementation is commonplace.However, the model is likely to have resonance in other contexts and situations where public services are being delivered through complex networks of public, private, and third-sector organizations.This can include emergency situations where diverse administrative networks respond to frequent updates and changes to policy to mobilize and distribute resources (e.g., Grizzle et al., 2020); in efforts to incentive regional economic development through collaborative networks (e.g., Lee et al., 2012;Zou, 2023); in efforts to diffuse innovation and technology throughout public organizations across geographies (e.g., Yi & Chen, 2019); and the collaborative and decentralized implementation of programs with multiple components and partners (e.g., Lindsay et al., 2021).
Future research, however, is required to explore the various motives, frames of references, and links to the various behaviors enacted to influence policy across public service provision contexts.Although we did not identify any determinate patterns in our qualitative approach, it is likely that public sector workers are drawn to work in contexts that align with their sense of community.It is important through further analysis to understand the various institutional and organizational values that interact with individual level motivations across various contexts to illuminate the complexities involved in policy implementation.Our findings also captured a static snapshot of the PSE in Scotland.Considering there is evidence that indicates that values and beliefs can co-evolve within policy networks (e.g., Siciliano et al., 2017), it is important to explore the relationships between personal motives and social frames of references over time.This can provide an indication of how street-level discretion can shape PSEs and how this influences policy implementation.

•
T A B L E 1 Informant groups, contextual settings, and data collection.Local government representatives that formulate local policy and administer implementation (LG).• Organizations that are contracted by the public sector to support business owners to start and grow (ESO_PC).• Organizations that are sponsored by the public sector to deliver their own support services to business owners to start and grow (ESO_PS).

F
I G U R E 2 A model of policy diffusion in public service ecosystems.