Policy entrepreneurship and local policy innovation

Embedded in the Multiple Streams Framework, the policy entrepreneurship literature has traditionally focused on the stage of agenda setting. Policy entrepreneurs are considered pivotal in conjoining three streams – problem, policy, and politics – and ultimately bringing them to the window of opportunity (Kingdon Reference Kingdon2013). This conception has been broadened in recent years. Scholars argue that these parallel streams persist and continue to flow beyond agenda setting (Zahariadis Reference Zahariadis, Sabatier and Weible2018; Goyal et al. Reference Goyal, Howlett and Chindarkar2020; Fowler Reference Fowler2022). Policy entrepreneurs are found to be active throughout the policy process, formulating novel policy designs, facilitating their implementation, and even promoting their evaluation (Mintrom and Norman Reference Mintrom and Norman2009; Cohen Reference Cohen2021; Mintrom and Luetjens Reference Mintrom and Luetjens2017). This expanded conception of policy entrepreneurs underscores the enduring significance of human agency throughout various stages of the policy process.

The potential of local policy entrepreneurs to engage in policy innovation arises from several significant factors. First, local governments have their own policy agendas and venues for policy formulation, providing policy entrepreneurs with legitimate jurisdictions in which to enact initial policy change, particularly in federal systems. The policy autonomy and administrative discretion bestowed on local governments grant them unique opportunities to start local pilot tests (Meijerink and Huitema Reference Meijerink and Huitema2010; Edri-Peer et al. Reference Edri-Peer, Silveira, Davidovitz, AviramFrisch, Shehade, Diab, Golan-Nadir and Cohen2023; Frisch Aviram et al. Reference Frisch Aviram, Cohen and Beeri2018). In an attempt to solve local governance problems, many of these small-scale pilots embrace unconventional ideas, often in the absence of explicit national policy guidelines (He Reference He2018; Aukes et al. Reference Aukes, Lulofs and Bressers2018). Furthermore, despite the localized nature of these pilots, ambitious local policy actors may act entrepreneurially to scale up such local government-led initiatives to a wider realm, provided there are favorable political conditions (Mintrom and Luetjens Reference Mintrom and Luetjens2017; Gassner and Gofen Reference Gassner and Gofen2018). Needless to say, promoting nationwide scaling up and inter-regional policy diffusion beyond one’s legitimate jurisdiction requires even strong entrepreneurial spirit and skills (Arnold Reference Arnold2020b; He et al. Reference He, Fu and Tang2025; Hong et al. Reference Hong, Bae and Choi2024).

Second, local governments’ and lower-level officials’ intimate familiarity with the field enables them to identify pressing social needs with a high level of acuity. Their close connections with the community facilitate the building of trust with citizens, as well as empathy with grassroots grievances. Working on the frontline, they typically understand what their citizens need most, while the professional training or specialized knowledge possessed by local technocrats grant them considerable authority in regard to identifying policy gaps and proposing new solutions (Arnold Reference Arnold2015; Cohen Reference Cohen2021; Erdi-Peer et al. Reference Edri-Peer, Silveira, Davidovitz, AviramFrisch, Shehade, Diab, Golan-Nadir and Cohen2023).

Third, vitally, as subnational agents of the national government, local policy entrepreneurs are tasked with implementing many national policies. Given the complexities of local circumstances, they are often granted substantial discretion to make adaptations during policy implementation, which creates an “opportunity structure” conducive to policy entrepreneurship. In other words, policy entrepreneurs remain essential for the successful implementation of national policies on the ground and they “continue to employ strategies to manipulate how policy implementers interpret policies” (Fowler Reference Fowler2022, p. 4). They convert ambiguous directives into concrete work plans and steer the implementation on the frontline. These responsibilities enable local policy entrepreneurs to play a critical role in catalyzing policy change (Arnold Reference Arnold2015; Petchey et al. Reference Petchey, Williams and Carter2008; Petridou and Mintrom Reference Petridou and Mintrom2021; Frisch Aviram et al. Reference Frisch Aviram, Cohen and Beeri2018).

The distinctive attributes of local policy entrepreneurs have attracted increasing scholarly attention in recent decades. Previous studies have examined the entrepreneurial activities of local officials across various levels and contexts (e.g., Petchey et al. Reference Petchey, Williams and Carter2008; He Reference He2018; Arnold Reference Arnold2015, Reference Arnold2020b; Lavee and Cohen Reference Lavee and Cohen2019; Frisch Aviram et al. Reference Frisch Aviram, Cohen and Beeri2018; Aviv et al. Reference Aviv, Gal and Weiss-Gal2021). Limited to one or two cases within a specific region, however, most of these studies rely on qualitative methods, so we know little about generalizable patterns of local policy entrepreneurship. This study employs the methodology of a systematic review to fill the research gap identified above.

Research design

Apart from grasping the research landscape, a systematic review can also be used to synthesize either quantitative or qualitative data from previous studies (Grant and Booth Reference Grant and Booth2009). In order to explore more details about local policy entrepreneurship, this systematic review adopts a qualitative approach combining systematic and transparent data collection and an open inductive process to analyze literature data (Petticrew and Roberts Reference Petticrew and Roberts2006; Gough et al. Reference Gough, Oliver and Thomas2012). Such a strategy allows us to bring together a large number of case studies from different disciplines into a holistic theoretical overview (Faling et al. Reference Faling, Biesbroek, Karlsson-Vinkhuyzen and Termeer2019). The well-accepted Preferred Reporting Items for Systematic Reviews and Meta-Analyses protocol (Moher et al. Reference Moher, Liberati, Tetzlaff and Altman2009) is followed to ensure the reliability, transparency, and replicability of this study. In order to construct a meaningful scholarly dialog with important literature on policy entrepreneurship, we closely reference five recent similar reviews (Frisch Aviram et al. Reference Frisch Aviram, Cohen and Beeri2020a, Reference Frisch Aviram, Cohen and Beeri2020b; Faling et al. Reference Faling, Biesbroek, Karlsson-Vinkhuyzen and Termeer2019; Edri-Peer et al. Reference Edri-Peer, Silveira, Davidovitz, AviramFrisch, Shehade, Diab, Golan-Nadir and Cohen2023; Meijerink and Huitema Reference Meijerink and Huitema2010).

Following Faling et al. (Reference Faling, Biesbroek, Karlsson-Vinkhuyzen and Termeer2019), we use Web of Science and Scopus as the two databases to collect relevant studies published before November 2023. Title, keywords, and abstract were used as search domains. The extensive literature on policy entrepreneurship spans across multiple disciplines and approaches (e.g., political science, public administration, policy sciences, area studies, etc.). Therefore, we took an inclusive approach to sampling. Furthermore, the concept of policy entrepreneurs encompasses a broad range of local actors, including street-level bureaucrats, public managers, mid-level officials, legislators, and governmental experts. Thus, we employed a broad search string consisting of all these possible terms to discern the level and role of individual agency. We also included several interchangeable concepts associated with policy entrepreneurship, such as institutional entrepreneurship and public entrepreneurship, in order to not miss any relevant study. Studies analyzing policy entrepreneurship at national and international levels were excluded. Diagrammed in Figure 1 below, the search protocols described above yielded a sample of 8,835 results.

Figure 1.

The Preferred Reporting Items for Systematic Reviews and Meta-Analyses flow diagram.

Source: authors.

We performed two rounds of eligibility assessment to filter out irrelevant studies. In the first round, titles and abstracts of the entire collection were reviewed, with a focus on their relevance to this study concerning case studies on local policy entrepreneurs. According to Robert Yin (Reference Yin2014), a “case” in social sciences can be defined as the basic unit for analysis, encompassing individual, events, or programs. He also stressed that the tentative definition of case is determined by the research questions (Yin Reference Yin2014). Because one of the major objectives of this review is to integrate qualitative evidence about the process of policy change driven by local policy entrepreneurs, we define that a case to be included must meet at least two criteria: (1) it should be an event with a relatively intact storyline; and (2) the role of individual or collective policy entrepreneurs in policy change should be identifiable. Quantitative studies were excluded. We define local policy entrepreneurs as comprising four main groups of actors: local legislators, executive officials, career civil servants, and other personnel in public agencies (e.g. government-affiliated experts, auditors, etc.). This inclusive conception is intended to capture the variety of entrepreneurialism in the subnational sphere. In the second round, we thoroughly read the full texts of the studies to identify any inclusion biasFootnote ¹ . Ultimately, 64 articles were included in this systematic review.

We used a directed approach to classify the large quantity of qualitative data into an efficient number of quantitative categories that represent similar meanings (Hsieh and Shannon Reference Hsieh and Shannon2005). Content coding strictly followed a codebook that synthesizes key attributes and dimensions characterizing policy entrepreneurship in previous reviews (e.g., Frisch Aviram et al. Reference Frisch Aviram, Cohen and Beeri2020a; Faling et al. Reference Faling, Biesbroek, Karlsson-Vinkhuyzen and Termeer2019; Mintrom and Norman Reference Mintrom and Norman2009; Meijerink and Huitema Reference Meijerink and Huitema2010), as well as important literature on policy entrepreneurship (e.g., Arnold Reference Arnold2015, Reference Arnold2020b; Zahariadis and Exdaklos Reference Zahariadis and Exadaktylos2016; Cohen and Golan-Nadir Reference Cohen and Golan-Nadir2020). The initial codebook was tested through pilot coding of 10 randomly selected studies. This pilot run allowed us to refine the codebook (see Appendix 1). The final codebook comprises five sections. Section 1 captures the bibliographic items of the study, such as the title, author(s), journal, and year of publication. Section 2 seeks to gauge the contextual factors surrounding policy entrepreneurship, including the country and level of government involved. The nature, duration, and outcome of each innovation case are also captured to the extent that the textual information in the study is sufficient for us to do so. Section 3 records the profile of the policy entrepreneurs in terms of gender, status, administrative position, and personal experiences. Entrepreneurial motivations and the actual strategies used are identified in Sections 4 and 5, respectively. While multiple policy actors were reported to actively participate in a reform, we only identified the key policy entrepreneurs involved.Footnote ²

Using the codebook as a guide, we coded 64 articles in the sample. As the primary coder, the first author meticulously read the texts line by line. Regular meetings were held with the corresponding author to discuss any issues arising from the coding exercise. The corresponding author, as the secondary coder, verified the results, and both authors discussed the appropriateness of the codes where discrepancies emerged. The codes supported both descriptive purpose and multivariate analysis. We used hierarchical cluster analysis (HCA) to map out the specific patterns of strategy combinations among the policy entrepreneurs. This technique enables researchers to generate a hierarchy of clusters by identifying unique and similar values from observations, which is not sensitive to initial conditions and less sensitive to outliers and cluster shapes (Sharif and Chandra Reference Sharif and Chandra2022; Androniceanu and Georgescu Reference Androniceanu and Georgescu2023). Notably, HCA as a clustering method has been employed by Arnold (Reference Arnold2021) to explore the patterns of policy entrepreneurship. In order to understand the relationships between motivation and strategy combinations and other components of policy entrepreneurship, we performed a chi-square test to calculate the phi-coefficient, which allowed us to assess the strength and direction of associations. Python and SPSS 29 were used to perform the HCA and correlation analysis, respectively.

Bibliographic profile

The 64 articles in the sample came from 40 academic journals, with 60% published in public administration and policy journals. Journals in the fields of political science, area studies, interdisciplinary studies, and sociology hosted the rest of the sample. The pool of eligible articles consists of 98 case studies on local policy entrepreneurship. More than half of these cases (56%) were observed in developed countries. Approximately one-third (34%) of the cases were based in China. The cases in the sample encompassed eight policy sectors. Social welfare and services (30%), environment policy (28%), economic policy and development (14%), and administrative reforms (10%) were the popular sectors in which local policy entrepreneurship is often observed. The distribution of cases across these substantive policy sectors largely resembles that reported in Frisch Aviram et al.’s (Reference Frisch Aviram, Cohen and Beeri2020b) systematic review in the developing world.

Key characteristics of the cases are summarized in Table 1. We considered: (1) the form of entrepreneurship; (2) the nature of policy change; (3) the level of government involved; (4) the duration of the policy change; and (5) the policy change outcome, as reported in the article. The conventional conception of policy entrepreneurs as heroic individuals has been expanded in recent literature to include collective policy entrepreneurs who rely on “solidarity, cooperation, and effective interaction” to achieve the shared goals (Tang et al. Reference Tang, Cheng and Cai2020). Compared to individual entrepreneurship, collective efforts in promoting policy change offer distinct advantages. Actors from different arenas can leverage strategies, skills, and resources from their respective fields, enabling collective entrepreneurship to influence reform through diverse approaches (Meijerink and Huitema Reference Meijerink and Huitema2010; Arnold Reference Arnold2015, Reference Arnold2020b; He et al. Reference He, Fu and Tang2025). Furthermore, solidarity and cooperation within collective entrepreneurship allow individuals to effectively handle the overwhelming pressure associated with policy innovation, potentially smoothing out any resistance (Tang et al. Reference Tang, Cheng and Cai2020; Aviv et al. Reference Aviv, Gal and Weiss-Gal2021). Our results note the prominence of collective efforts in driving local innovations, as close to 60% of the cases in the sample manifest collective policy entrepreneurship, with multiple individuals involved.

Table 1.

Snapshot of local policy entrepreneurship cases

Source: authors.

In two-thirds of the cases, policy entrepreneurs work to champion policy change within their respective jurisdictions. We observe 31 cases (32%) in which policy entrepreneurs creatively implement higher-level policies and eventually introduce innovative solutions to their respective jurisdictions.Footnote ³ Exworthy and Powell (Reference Exworthy and Powell2004) argue that significant policy change on the ground is likely to occur only when “big” national windows match “little” local windows. This process of coupling the two types of policy windows provides local policy entrepreneurs in multilevel governance structures with ample opportunities to undertake innovations through top-down implementation. In other words, policy implementation also presents fertile soil for many types of entrepreneurial individuals at the middle or even lower echelons of the hierarchy to maneuver.

Our review notes that local policy entrepreneurship occurs in various layers of subnational authorities, with the municipal level being the principal avenue (44%). Approximately 34% of the policy change cases were pursued at the district (sub-municipal) or community levels. Around 22% of the cases took place at the level of province or state.Footnote ⁴ As many empirical studies document the chronological outline of policy entrepreneurship cases, we were able to gauge their duration. The results indicate that 31% of the cases took three to five years to yield change outcomes, while 33% of them lasted for six years or longer, mirroring the sustained efforts required on the part of local policy entrepreneurs.

The existing policy entrepreneurship literature suffers from significant “survivor bias,” in that most empirical studies document successful cases of policy change, whereas little is known about the role of policy entrepreneurs in failed cases or in cases where the status quo is defended (Zahariadis Reference Zahariadis, Sabatier and Weible2018; Petridou and Mintrom Reference Petridou and Mintrom2021). As a result, limited variance in the “dependent variable” further hinders empirical generalization. Yet, our review managed to identify both successful and unsuccessful cases of local policy change. We find that, while most cases (78%) present positive outcomes and achieve the desired change, one-fifth of them (20%) are eventually unsuccessful, even in the presence of policy entrepreneurs. Our further analysis suggests that continued policy entrepreneurship is a key factor leading to successful policy change outcomes. For example, Arnold (Reference Arnold2015, Reference Arnold2020b) compares two cases of policy innovations in state-level wetland governance in the United States. In State H, where the innovation yielded success, policy entrepreneurs proactively selected competent successors to steer the subsequent reform process before leaving their positions. In contrast, the successor was incapable of sustaining the policy coalition forged by this entrepreneurial predecessor, resulting in the ultimate termination of the innovation. In other circumstances, the radical design of policy innovation constituted a key reason for discontinued policy entrepreneurship. For example, Zhu (Reference Zhu2018) examines the entrepreneurial role of a controversial Chinese reformer, who serves executive roles in multiple localities. Inspired by the (then) flourishing new public management movement, the reformer championed the privatization of essential public services in his city, which triggered massive controversies regarding the appropriateness of such “load-shedding” reforms. These reforms were subsequently abandoned after the policy entrepreneur left his post. Detailed information regarding the outcome of each case and the identification of policy entrepreneurs’ personal background and motivation is exhibited in Appendix 2.

Profile of local policy entrepreneurs

We identified 122 sets of subnational policy entrepreneurs in the cases. Table 2 summarizes the key characteristics of these individuals. Out of the 84 cases in which the gender of key individuals was identifiable, 62 entrepreneurs were men and 22 were women. This imbalance did not come as a surprise, considering the widespread existence of gender inequity in administrative systems worldwide. Particularly in Asian countries, men tend to hold dominant positions in government agencies, granting them more opportunities to become policy entrepreneurs.

Table 2.

Personal background of local policy entrepreneurs

Source: authors.

Note: *In most cases of collective policy entrepreneurship, gender is unidentifiable. **We code professional qualifications of policy entrepreneurs based on whether they have a postgraduate degree or professional certificate in an area.

When categorizing the administrative status of different individuals, we exercised utmost caution, given the vast differences in administrative norms and personnel rules across countries. In this review, the term “local legislators” refers to elected representatives who serve in subnational legislative bodies. Executive officials are typically chiefs and deputy chiefs of the government (e.g., the mayor, governor, or party secretary in communist systems, etc.) or functional departments (e.g., the Education Bureau, the Department of Finance, etc.), or statutory agencies (e.g., the local housing authority, tourism board, etc.). In our operational definition, career civil servants do not take departmental or agency-level leadership positions, but rather typically work in the middle or lower strata of their respective organizations. As expected, they constitute the largest group of local policy entrepreneurs (47%) in the sample. This is very close to the proportion of local executive entrepreneurs, which stands at 44%. We noted a smaller percentage of entrepreneurial roles played by “other personnel” (7%) in the cases. These individuals include researchers in government-affiliated institutes, government auditors, and so forth.

Work experience and professional qualifications constitute endogenous and exogenous personal attributes of public sector professionals (Toh et al. Reference Toh, Diong, Boo and Chia1996). We classify work experience into frontline experience and cross-sectoral experience. The former category typically refers to experience of working at the grassroots level of the bureaucracy. For example, despite their high positions in the local bureaucracy, some entrepreneurial executive leaders had worked as trainees (Mackenzie Reference Mackenzie2004), medical practitioners (Oborn et al. Reference Oborn, Barrett and Exworthy2011; He et al. Reference He, Fu and Tang2025), or street-level bureaucrats (Zhang Reference Zhang2016; Leo and Andres Reference Leo and Andres2008; He Reference He2018; Mintrom et al. Reference Mintrom, Salisbury and Luetjens2014). These frontline experiences gave them valuable grassroots know-how and nurtured their close emotional attachment to the community. Cross-sectoral experience is another valuable asset for subnational policy entrepreneurs, because they often entail boundary-spanning efforts that in turn require personal networks and multi-sectoral knowledge (e.g., Oborn et al. Reference Oborn, Barrett and Exworthy2011). Prior experience in other policy domains and even in the private sectors can help policy entrepreneurs better appreciate the positions and interests of other stakeholders and is often crucial for inter-agency collaboration (Petridou and Mintrom Reference Petridou and Mintrom2021).

We also considered the professional qualifications held by policy entrepreneurs, which not only give them powerful skillsets with which to diagnose the social problem at hand and devise appropriate solutions but also allow them to claim authority and expertise in a specific domain. Here, qualifications are assessed based on whether the entrepreneur holds a professional degree (e.g., Master of Public Administration, Master of Business Administration, etc.) or acquired professional certificates (e.g., as a lawyer, certified accountant, medical doctor, registered engineer, etc.). The results indicate that nearly two-thirds of policy entrepreneurs (45%) at the subnational avenue have frontline experience, and half (50%) hold professional qualifications. More than one-third of them (38%) have work experience in other sectors. Notably, 87% of the policy entrepreneurs in the sample have more than one type of experience. Such experiences certainly help increase the social acuity and operational capacity of local policy entrepreneurs.

Motivations

In Agendas, Alternatives, and Public Policies, Kingdon (Reference Kingdon2013) identifies several types of entrepreneurial motivations, including career advancement, the desire to promote their values, and an interest in catalyzing policymaking. In recent years, there is a growing literature seeking to provide a more nuanced characterization of motivations for middle- and street-level bureaucrat policy entrepreneurs. Their motivations for engaging in entrepreneurial actions include sympathy for the disadvantaged situations of their clients (Lavee and Cohen Reference Lavee and Cohen2019) and seeking solutions for local problems (Teets et al. Reference Teets, Hasmath and Lewis2017; Feiock and Carr Reference Feiock and Carr2001). These studies also indicate that local policy entrepreneurs are often driven by a mixture of motivations. Our result reveals that 94% of the policy entrepreneurs hold more than one entrepreneurial motivation, highlighting the significance of mixtures of motivations.

Table 3 further reports key findings regarding motivations and their combinations. Several observations are evident. First, the most salient motivations appear to be interest in or commitment to a specific policy area and the desire to seek solutions for local problems, with 86% and 80% of entrepreneurs manifesting these gustoes, respectively. This result echoes with Edri-Peer et al. (Reference Edri-Peer, Silveira, Davidovitz, AviramFrisch, Shehade, Diab, Golan-Nadir and Cohen2023) that low-level policy entrepreneurs are generally motivated by the desire to promote and improve local interests. In some cases, even when advocating for initiatives to enhance international collaboration, these reformers may still seize such opportunities to promote the interests of their own localities (e.g. Mintrom and Luetjens Reference Mintrom and Luetjens2017).

Table 3.

Motivation frequency

Source: authors.

Second, career incentives and completing tasks assigned by superiors also represent salient motivations of local policy entrepreneurs. Our results suggest that approximately 30% and 34% of the policy entrepreneurs in the sample have shown such motivations respectively. We are certainly aware of possible bias due to challenges in identifying the latent motivations of policy entrepreneurs. Most qualitative studies heavily rely on in-depth interviews for data collection, and it is common for policy entrepreneurs to rarely cite career incentives as the principal motivation behind them. In another systematic review focused on street-level bureaucrat policy entrepreneurs, a similarly low percentage (26%) is noted with regard to self-interest as the key motivation (Edri-Peer et al. Reference Edri-Peer, Silveira, Davidovitz, AviramFrisch, Shehade, Diab, Golan-Nadir and Cohen2023). Another possible explanation for the relatively low emphasis on career incentives is the presence of a glass ceiling for career advancement in local bureaucracy. As most street-level and middle-level officials are typically aware of the terminal position they could possibly attain, many choose not to actively participate in the race for promotion (Teets et al. Reference Teets, Hasmath and Lewis2017; Folke and Rickne Reference Folke and Rickne2016). As a result, career incentives may play a less significant role for local policy entrepreneurs, but this result should be interpreted with caution concerning executive leaders, as Schneider and Teske (Reference Schneider and Teske1993) have demonstrated that career incentives remain an intrinsic factor leading these senior officials to undertake entrepreneurial innovations.

Third, the results highlight the intention to earn resources and empathy as crucial motivations of local policy entrepreneurs, with approximately 20% and 19% of them manifesting such motivations, respectively. The intention to earn resources has a relatively low presence than expected. Previous studies suggest that resource constraints are a significant characteristic facing local policy entrepreneurs, particularly those at the street level (Lipsky Reference Lipsky2010; Cohen Reference Cohen2021). The discrepancy between resource constraints and the relatively low presence of the motivation to earn resources may be explained by the intrinsic nature of resources – particularly financial ones—as a means to achieve policy goals. Despite the critical importance of resources, local policy entrepreneurs may view earning resources as a means to achieve their goals, rather than as an end in itself. The presence of empathy as a motivation appears to be associated with certain policy areas. In poverty alleviation and social welfare in particular, local policy entrepreneurs frequently interact with marginalized groups, thus evoking sympathy and compassion among them (Lavee and Cohen Reference Lavee and Cohen2019; Aviv et al. Reference Aviv, Gal and Weiss-Gal2021; Cai et al. Reference Cai, Shen and Tang2022).

The chi-square test and phi-coefficient analyses yield additional insights into the factors associated with entrepreneurial motivations, as reported in Table 4. We find significant correlations between administrative status and specific entrepreneurial motivations. Specifically, career incentives appear to be a dominant motivation for senior-level entrepreneurs (phi = 0.345, p < 0.001). Yet, these senior entrepreneurs exhibit negative associations with other motivations, such as earning resources (phi = −0.289, p < 0.001), completing tasks (phi = −0.410, p < 0.001), and empathy (phi = −0.454, p < 0.001). This could be attributed to their positions as keen participants in the leadership tournament, compared to other groups of entrepreneurial individuals. This distinction in motivations is further supported by the negative association between career incentives and middle- and low-level entrepreneurs (phi = −0.173, p < 0.1 and phi = −0.212, p < 0.05, respectively). As discussed above, rank-and-file bureaucrats often face career advancement barriers, while senior-level bureaucrats tend to have greater chances of overcoming these barriers by mobilizing resources and demonstrating political performance. In contrast, middle-level entrepreneurs tend to exhibit a stronger preference for seeking resources, such as finances and reputation, through catalyzing policy innovation (phi = 0.341, p < 0.001). It is not surprising that they also display a strong intention to fulfill tasks assigned by administrative supervisors (phi = 0.223, p < 0.05). Furthermore, they are less likely to be motivated by interest or commitment in a particular policy area (phi= −0.189, p < 0.01). Similarly, low-level entrepreneurs demonstrate a strong motivation to connect policy innovation with implementation (phi = 0.217, p < 0.05). For entrepreneurial officials at the street level, aligning their innovative ideas with policy implementation proves to be an effective strategy for overcoming obstacles such as resource constraints (Lavee and Cohen Reference Lavee and Cohen2019). Moreover, low-level policy entrepreneurs tend to exhibit a particularly strong association with empathy (phi = 0.500, p < 0.001), underscoring the unique motivation mix rooted in their proximity to citizens.

Table 4.

Phi-coefficient for entrepreneurial motivations and other variables

Source: authors.

Note: *, **, and *** denote significance at 0.05, 0.01, and 0.001, respectively. As a result of the exploratory nature of this review, we also mark statistical significance at 0.10 as +.

This review also finds correlations between entrepreneurial motivations and individual experiences. Local policy entrepreneurs with frontline experiences are more likely to exhibit empathy, even if they currently hold senior positions (phi = 0.474, p < 0.001). Holding professional qualifications appears to produce a negative impact on policy entrepreneurs’ career incentives (phi = −0.339, p < 0.001), possibly because many of them are technocrats working in middle- and low-level positions within the government and tend to hold moderate political ambitions. Finally, we find that two high-frequency motivations – seeking solutions for local problems and having an interest in a specific policy area – are not significantly associated with the individual-level factors mentioned above. This result implies that these twin motivations are common among local policy entrepreneurs, regardless of their hierarchical positions or individual experiences.

Entrepreneurial strategies

The actual behaviors of policy entrepreneurs bridge the gap between their intentions and contextual factors, ultimately leading to policy change outcomes (Zahariadis and Exadaktylos Reference Zahariadis and Exadaktylos2016; Mintrom and Norman Reference Mintrom and Norman2009). Building on the systematic review by Frisch Aviram et al. (Reference Frisch Aviram, Cohen and Beeri2020a) and other works (Faling et al. Reference Faling, Biesbroek, Karlsson-Vinkhuyzen and Termeer2019; Lavee and Cohen Reference Lavee and Cohen2019; Zahariadis and Exadaktylos Reference Zahariadis and Exadaktylos2016), this study synthesizes seven types of entrepreneurial strategies: problem framing, venue shopping, policy learning, teamwork strategies, leading by example, anchor work, and the use of procedural policy tools. Promoting policy innovations requires entrepreneurs to navigate complex contexts, necessitating not only multiple skills but also effective strategies to address the challenges at hand. The astute combination of different strategies is crucial for successful policy entrepreneurship. Our analysis of individual entrepreneurial actions regarding strategy mixes yields several crucial findings (see Table 5 below).

Table 5.

Strategy frequency

Source: authors.

Note: *, **, and ***denote significance at 0.05, 0.01, and 0.001, respectively.

First, teamwork strategies stay at the center of the strategy mix, reflected in their high frequency of adoption by entrepreneurs (84%). Previous studies have highlighted the lack of resources as a defining characteristic of policy entrepreneurs, particularly those in middle and lower positions, who often need to build teams to mitigate resource constraints (Lavee and Cohen Reference Lavee and Cohen2019; Tang et al. Reference Tang, Cheng and Cai2020; Arnold Reference Arnold2020b). Even chief executives in local governments need to rely on the expertise and entrepreneurship of others to develop policy solutions and steer implementation (Goyal et al. Reference Goyal, Howlett and Chindarkar2020; Gong Reference Gong2022; Oborn et al. Reference Oborn, Barrett and Exworthy2011). As Mintrom (Reference Mintrom2019) underscores, the gathering of political intelligence and development of strategies often happen in team settings. Teamwork strategies thus provide a solid foundation for other entrepreneurial actions, such as policy learning and anchor work, thereby explaining their central role in the strategy mix.

Problem framing represents a highly useful strategy for policy entrepreneurs looking to strengthen legitimacy and earn resources (Mintrom and Norman Reference Mintrom and Norman2009; Frisch Aviram et al. Reference Frisch Aviram, Cohen and Beeri2018). While most previous studies treat problem framing as a broad concept that can describe all activities related to coupling problem streams, there is a multitude of framing skills, such as highlighting the failures of current policies (Baumgartner and Jones Reference Baumgartner and Jones1993), presenting scientific evidence (Arnold Reference Arnold2015), and linking the benefits of proposed policies to the interests of other stakeholders (He Reference He2018; Capano and Galanti Reference Capano and Galanti2021). Drawing on applied behavioral science, Aukes et al. (Reference Aukes, Lulofs and Bressers2018) developed a five-fold typology of problem-framing activities based on Dewulf and Bouwen (Reference Dewulf and Bouwen2012), including accommodating, disconnecting, incorporating, spotlighting, and reconnecting.Footnote ⁵ However, this typology appears to be too complex to apply in public policy research. In particular, the boundary between accommodating and incorporating, as well as that between spotlighting and reconnecting, are often too blurry to identify in real cases. Therefore, we consider three essential framing skills: incorporating, disconnecting, and spotlighting. Compared to disconnecting and incorporating, spotlighting has a much higher frequency of use in the sample (52% compared to 23% and 7%, respectively). This result suggests that highlighting the merits of a policy proposal and connecting it with the interests of other policy actors appear to be common strategies in local policy entrepreneurship.

Defined as the transfer of decision-making authority to a new policy arena where there are fewer opponents, venue shopping is frequently noted by previous studies as a crucial entrepreneurial strategy used to gain support from key policy actors (Faling et al. Reference Faling, Biesbroek, Karlsson-Vinkhuyzen and Termeer2019; He and Ma Reference He and Ma2020; He et al. Reference He, Fu and Tang2025). This study adapts the categorization of venue shopping strategies put forth by Faling et al. (Reference Faling, Biesbroek, Karlsson-Vinkhuyzen and Termeer2019) to distinguish vertical venue shopping from horizontal venue shopping. This distinction allows us to identify the nuances in which local policy entrepreneurs maneuver within complex networks to amass political capital and resources. Vertical venue shopping involves moving policy issues to higher-level authorities, while horizontal venue shopping entails transferring proposals to other policy domains for support. Such strategies policy entrepreneurs with important tools for building coalitions and earning resources. As shown in Table 5, vertical venue shopping is frequently employed by about 48% of entrepreneurs when promoting policy change. Clearly, maneuvering within the hierarchy is pivotal for local policy entrepreneurs, given their relatively low position in the administrative system.

Anchor work and using procedural policy tools both have a high frequency of adoption (86% and 77%, respectively). Anchor work, as defined by Frisch Aviram et al. (Reference Frisch Aviram, Cohen and Beeri2018), refers to activities aiming to secure policies through regulation, rule-making, and actual implementation. Policy entrepreneurs can consolidate the progress of policy change through formal institutionalization. They can also anchor the shift through informal practices, such as persuading peer bureaucrats or lower-level officials to use new policy instruments in their daily work without modifying existing regulations (Arnold Reference Arnold2015, Reference Arnold2020b), similar to the strategy of “scaling up change processes” in Mintrom (Reference Mintrom2019). Both approaches to anchor work have been observed in previous studies. For example, Lu et al. (Reference Lu, Mayer, Wellstead and Zhou2020) found that a middle-rank official championing an urban greening proposal in Shanghai made constant efforts to enhance the amendment of local regulations on urban planning. Arnold (Reference Arnold2015, Reference Arnold2020b) also illustrated how local officials in the United States innovated evaluation metrics for environmental protection and persuaded frontline implementers to use these tools to expand the influence of their innovations.

Previous studies have also noted that policy entrepreneurs strategically use bureaucratic procedures to influence decision-making and implementation (Meijerink and Huitema Reference Meijerink and Huitema2010; Zahariadis and Exadaktylos Reference Zahariadis and Exadaktylos2016). Here, we use the concept of “procedural policy tools,” adapted from Bali et al. (Reference Bali, Howlett, Lewis and Ramesh2021), to describe such strategies. Unsurprisingly, both anchor work and using procedural tools are crucial for consolidating the progress of policy change. Institutionalization and the full-scale implementation of policy change are undoubtedly the ultimate goal pursued by policy entrepreneurs. Procedural tools, such as authority, treasure, and organization, are often used to enhance the anchor work of policy entrepreneurs. For example, entrepreneurial officials often establish provisional organizations, such as steering groups and task forces, to facilitate bureaucratic coordination at both the strategic and operational levels (Oborn et al. Reference Oborn, Barrett and Exworthy2011; Gong Reference Gong2022). Policy entrepreneurs also take advantage of opportunities to present their proposals to senior leaders during sharing sessions or high-level conferences, thereby seeking political endorsement for their initiatives (He Reference He2018).

Policy learning and leading by example are the last two types of entrepreneurial behaviors observed in the strategy mix. To address problems in a specific professional domain, policy entrepreneurs often need to acquire knowledge and build narratives and policy solutions in an informed manner. Therefore, policy learning as “the updating of beliefs” concerning policy based on lived or witnessed experiences, analysis, or social interaction (Dunlop and Radaelli Reference Dunlop and Radaelli2013), is pivotal for entrepreneurs aiming to effectively engage in policy innovation. They may learn from different sources. External sources include those outside the government, such as advice from external experts, third-party research, and peer experience from pioneering regions (Lavee and Cohen Reference Lavee and Cohen2019; Ye and Wu Reference Ye and Wu2022; Capano and Galanti Reference Capano and Galanti2021). Internal sources include channels within the bureaucratic system, such as track records from previous policies, top-down guidelines, learning from peer bureaucrats, and bottom-up feedback from subordinates (Lu et al. Reference Lu, Mayer, Wellstead and Zhou2020). Our results indicate that external sources appear to play a bigger role in the sample, as a higher percentage of local policy entrepreneurs learn from the outside (49%), rather than the inside (33%).

Leading by example denotes taking actions to prove the workability of a policy, particularly through pilots or demonstration projects (Mintrom and Norman Reference Mintrom and Norman2009). Through such strategies, policy entrepreneurs increase the likelihood of convincing other policy actors to participate in the innovation, as the benefits of the policy become tangible and observable. Additionally, setting up pilots offers policy entrepreneurs a crucial tool with which to identify shortcomings in the current policy proposal, thus helping improve the policy design. In our sample, more than half of the cases (55%) that adopted this strategy were based in China, reflecting the country’s long tradition of experimentalist governance interwoven with local policy dynamism (Teets and Noesselt Reference Teets and Noesselt2020).

A significant finding from this study is the strong correlation between successful policy change and two prominent strategies, namely, anchor work (phi = 0.318, p < 0.01) and teamwork (phi = 252, p < 0.05). The statistical significance of anchor work highlights the pivotal role of institutionalization and scaling up in local policy entrepreneurship. Because the continuation of policy innovations is often threatened by job turnover of key individual officials, such a strategy presents the key to secure continued use of innovative practices. Furthermore, local policy entrepreneurs who pursue innovations collectively were found to have a greater chance of policy change. Individual entrepreneurship seems to have much less chance of success given their lack of authorities and resources at the subnational arena.

Characterizing strategy mixes

The theoretical interest of this study goes beyond merely describing the strategies employed by local policy entrepreneurs. We maintain that their divergence in skills and positions may lead to divergent entrepreneurial behaviors (Arnold Reference Arnold2020a; Mintrom Reference Mintrom2019). However, within similar contexts, such as the same stage of policy process, their adoption of strategies may still present certain commonalities, yielding some “bigger patterns” of strategy mixes (Goyal et al. Reference Goyal, Howlett and Chindarkar2020; Arnold Reference Arnold2021). When seeking the “bigger patterns” of subnational policy entrepreneurship, HCA is employed as a useful technique to explore the similarities (Arnold Reference Arnold2021). Our systematic review utilizes the simple linkage algorithm that can generate a set of clusters at each level (Graham and Hell Reference Graham and Hell1985), or for each threshold value which produces a new partition (Murtagh and Contreras Reference Murtagh and Contreras2012). In this study, the HCA initially considers each policy entrepreneur as a single unit and successively merged pairs of clusters based on the cooccurrence of strategies in each unit as a continuous variable, which indicates the common “tool kit” of policy entrepreneurs. The distance between the strategies is measured by the reciprocal of their cooccurrence. Here, we are predominantly interested in the “tool kits” for executive officials and career civil servants for two reasons. First, they are the two largest workforces in subnational governments. Second, the comparison between these two groups of local policy entrepreneurs allows us to further explore the influence of bureaucratic positions on policy entrepreneurship within a relatively large sample size.

A dendrogram is a common way to present HCA results (Arnold Reference Arnold2021; Sharif and Chandra 2023). Figs. 2 and 3 below present the strategy combinations of career civil servants and executive officials at different hierarchies, respectively. This study intentionally sets a threshold of 0.1 in distance to distinguish the central cluster and peripheral strategies. Specifically, one strategy within the central cluster co-occurs with another one from the same cluster at least ten times, which is a reasonable threshold to define “frequent adoption” by policy entrepreneurs in our sample. Two central clusters of strategies emerge from career civil servants and executive officials respectively. For career civil servants, their central cluster encompasses eight specific strategies, including “anchor work,” “teamwork strategies,” “using procedural tools,” “spotlighting,” “learning from external sources,” “vertical venue shopping,” “learning from internal sources,” and “disconnecting.” In comparison, “using procedural tools,” “anchor work,” “teamwork strategies,” “vertical venue shopping,” “spotlighting,” “leading by example,” “learning from external sources,” and “learning from internal sources” agglomerate into the central cluster of executive officials.

Figure 2.

Hierarchical analysis of career servants’ strategy mix.

Source: authors.

Figure 3.

Hierarchical analysis of executive officials’ strategy mix.

Source: authors.

Two key findings emerge from this clustering of strategies of local policy entrepreneurs. First, “anchor work,” “teamwork strategies,” and “using procedural policy tools” share the closet distance with each other (from 0.023 to 0.025 for career civil servants and 0.022 to 0.026, respectively) within the clusters of both groups, forming the strategy foundation of local policy entrepreneurs. Anchor work is crucial for policy entrepreneurs because it helps secure the desired policy change. While anchor work can be conducted through both formal and informal approaches as mentioned above, both approaches practically rely on the proper use of procedural policy tools within the bureaucracy. As Meijerink and Huitema (Reference Meijerink and Huitema2010) commented: “[p]olicy entrepreneurs operate within a particular institutional setting but many also try to change that setting.” (p. 21). Although executive officials and career civil servants are bound by procedures and rules of policymaking, they also gain sufficient knowledge about the processes as insiders. Such knowledge enables them to utilize procedural tools to increase success of substantial change of anchor work (Zahariadis and Exadaktylos Reference Zahariadis and Exadaktylos2016). This explains why local policy entrepreneurs across different countries are often found to use various types of procedural policy tools, such as information, organization, and treasure, to institutionalize their innovations, scale them up, or secure their sustained implementation (Zapata and Campos Reference Zapata and Zapata Campos2019; Mintrom and Luetjens Reference Mintrom and Luetjens2017; Brown and Cohen Reference Brown and Cohen2019; Frisch Aviram et al. Reference Frisch Aviram, Cohen and Beeri2018; Durose Reference Durose2011; Aviv et al. Reference Aviv, Gal and Weiss-Gal2021; Pautz et al. Reference Pautz, Wright and Collins2021).

In addition, given the limited capacity of individuals, teamwork plays a pivotal role in the adoption of anchor work strategies. The HCA demonstrates that compared to executive officials, local career bureaucrats are more likely to align anchor work with teamwork strategies, as anchor work has shorter distance with teamwork strategies in the clustering of career civil servants (0.023 compared to 0.026 in the cluster of executive officials). This result may be explained by the fundamental differences between career civil servants and executive officials in local policy entrepreneurship. As Teske and Schneider (Reference Teske and Schneider1994) explained, compared to executive officials, career civil servants lack sufficient power, making it difficult for them to promote significant policy change independently. Therefore, establishing coalitions becomes an entrepreneurial necessity.

Second, as “anchor work,” “teamwork strategies,” and “using procedural policy tools” form similar nuts of strategy mixes for local policy entrepreneurs, the clusters at the higher hierarchies emerge as different shells for executive officials and career civil servants. Specifically, career civil servants are more likely to combine strategies in the core cluster with “spotlighting,” “disconnecting,” and “learning from external sources” (distance in 0.036, 0.071, and 0.038, respectively) compared to executive officials (distance in 0.040, 0.111 and 0.045, respectively). On the contrary, “vertical venue shopping,” “leading by example,” and “learning from internal sources” have shorter distances with the core strategy mixes within the clustering of executive officials (0.033, 0.043, and 0.056 vs. 0.045, 0.083, and 0.059). Such differences further present distinctive features between career civil servants and executive officials when acting as policy entrepreneurs.