Mapping the governing entities and their interactions in designing policy mixes for sustainability transitions: the case of electric vehicles in China

Recognising the limited consideration of administrative aspects in sustainability transitions research, this article investigates the governing entities behind China ’ s policy mix for electric vehicles (EVs), and their interactions from 2001 to 2020. Based on a social network analysis of policy documents we find that as the e-mobility transition unfolds, a complex and evolving network of governing entities has appeared in designing China ’ s EV policy mix. Specifically, a small group of highly interactive governing entities has played a critical role in coordinating and mobilising system resources, while some new entrants have also come to the fore in response to recent socio-technical changes. Moreover, our community detection analysis distinguishes various groups of governing entities, each performing different policy functions. Based on our empirical case, we discuss factors that influence changes to administrative arrangements for policy mixes. We conclude that the deliberate acceleration of sustainability transitions calls for further research on the administration of the associated policy mixes.


Introduction
The role of the state and policy interventions in accelerating low-carbon transitions has been widely acknowledged in innovation and transition studies (Markard et al., 2012;Schot and Steinmueller, 2018). In this literature, more interventionist approaches are being suggested to achieve rapid and fundamental transformation of whole socio-technical systems (Kanger et al., 2020;Markard et al., 2020). Considering the multiple and interconnected failures entrenched in the creation, adoption and diffusion of sustainable innovations (Weber and Rohracher, 2012), multiple policy instruments need to be designed and employed for the rapid and deep decarbonisation of currently unsustainable socio-technical systems (Geels et al., 2017), raising significant challenges for researchers and policymakers analysing and coordinating such policy endeavours (Flanagan et al., 2011). How to operationalise or "translate" these ambitions into feasible and well-coordinated policy solutions (Kern and Rogge, 2016;Schot and Steinmueller, 2018), however, has until recently been largely neglected.
The policy mix perspective has, in this context, been introduced over the past decade Rogge et al., 2017). Understanding policy mixes as the combination of multiple goals and policy instruments to achieve these (Kern and Howlett, 2009), the role of intentional policy interventions has been reconceptualised by adopting systemic transition frameworks (Kanger et al., 2020;Kivimaa and Kern, 2016). As such, scholars have not only investigated the features of individual policy instruments, but also their collective characteristics, such as the consistency of different policy goals and instruments (Huttunen et al., 2014) or the credibility of such policy mixes as perceived by target groups (Rogge and Schleich, 2018). Collectively, this literature provides lessons that can inform real-world transition policy mixes.
Yet, closer attention to the specific governing entities behind policy mixes for sustainability transitions is still relatively limited, other than to identify focal policy mixes (Ossenbrink et al., 2019). This presents an important research gap because these administrative actors may be crucial to the effectiveness of complex policy mixes (Rogge and Reichardt, 2016). We argue that the systematic identification of these governing entities is critical for understanding the underlying policy processes influencing the co-evolution of policy mix and socio-technical changes (Edmondson et al., 2019). Moreover, given the tensions between the need to mobilise massive resources, and the fragmentation of jurisdictional authorities in current governance systems (Marquardt, 2017;Meadowcroft, 2007), we argue that the interactions between different governing entities should be critically investigated in studying transition policy mixes.
Motivated by this research gap, we aim to initiate a closer discussion around the role of governing entities in designing policy mixes for sustainability transitions. For this, we have chosen the case of electric vehicles (EVs) for two main reasons: first, transitioning to electric mobility (e-mobility) has been widely recognised as a crucial pathway in decarbonising the current fossil-dominated transport sector, with electric vehicles representing a key technology for such a transition (Geels, 2012;Marletto, 2014); second, recent discussions concerning multi-system interactions 1 (Rosenbloom, 2020), deep transitions (Schot and Kanger, 2018) and the acceleration of unfolding transitions (Markard et al., 2020) have noted particularly complex governance challenges in cases such as the electrification of transport, which require the coordination of multiple governing entities across several socio-technical systems (Rosenbloom, 2020). Furthermore, we have selected China not only because of its significance in global efforts towards climate mitigation in general and in promoting EVs in particular, but also as a key country following a technocratic transition management approach (Cai and Aoyama, 2018;Ely et al., 2019). By going beyond the "European bias" in our field, we also contribute to greater geographical diversity (Markard et al., 2012). Specifically, we are interested in the following research questions: Which governing entities play a role in designing the national EV policy mix in China and how do they interact with each other? How can we capture the resulting cross-organisational structure and its temporal dynamics? By addressing these questions, we take a first step towards a better understanding of the "everyday experience" of policymakers and contribute to greater methodological diversity in policy mix research .
The remainder of this article is structured as follows. In Section 2 we provide a brief review of the literature on policy mixes and governing entities in sustainability transitions. In Section 3 we discuss case selection rationales, outline the boundaries of our empirical investigation and offer a brief review of EV policy mix changes in China. In Section 4 we introduce our research method, followed by the presentation of our main findings for governing entities involved in the design of China's national EV policy mix from 2001 to 2020 in Section 5. Finally, we discuss the implications of our findings in Section 6 and conclude by outlining future research avenues in Section 7.

Policy mixes for sustainability transitions
Over the past two decades, our understanding of the design and impact of policy mixesunderstood in its simplest form as the combination of multiple policy instruments to achieve policy objectiveshas advanced significantly as a result of interdisciplinary efforts by researchers from different fields, such as environmental economics, policy sciences and innovation studies (Bouma et al., 2019;Howlett and Rayner, 2007;Rogge et al., 2017). By going beyond a focus on individual policy instruments targeting specific socio-technical blockages, a growing number of innovation and transition scholars have attempted to integrate such policy mix thinking into the system innovation systems perspective, especially in the area of sustainability transitions (Borrás and Edquist, 2013;Flanagan et al., 2011;Kern et al., 2019).
In order to address the conceptual ambiguity in existing bodies of literature (Flanagan et al., 2011), a broader conceptualisation of policy mixes has been described at the level of socio-technical systems (Rogge and Reichardt, 2016). Three main analytical building blockspolicy mix elements, associated policy processes and policy mix characteristicsare incorporated in this approach . Integrating these into transition frameworks, such as the Transition Management approach (Kern and Howlett, 2009), the Technological Innovation Systems framework (Kivimaa and Kern, 2016) or the Multi-level Perspective (Kanger et al., 2020;Kern, 2012;Rogge et al., 2020), has further enriched our understanding of the policies and politics associated with transitions. Actionable methodological guidelines have been developed to support such analyses, including those on how to scope and delineate the relevantor focalpolicy mix under study (Ossenbrink et al., 2019). Collectively, these advances offer a more systemic lens to investigate, design and evaluate policy mixes in real-world transition processes .
While frequently acknowledged, less attention has been given to other dimensions of policy mixes, in particular the complex governance structures underpinning focal policy mixes and their temporal dynamics (Flanagan et al., 2011;Rogge and Reichardt, 2016). Although the role of existing governance structures in shaping the design of policy mixes and influencing their effectiveness has been stressed by policy scholars (Howlett and del Rio, 2015;Howlett and Rayner, 2013), the transition literature concerning this topic is still relatively limited. The complexity of governance structures involved in deepening, widening and accelerating unfolding 1 Accelerating and mainstreaming unfolding sustainability transitions necessitates broader efforts incorporating potential pervasive and cascading dynamics induced by their adjacent socio-technical systemssomething referred to as multi-system interactions (Rosenbloom, 2020). socio-technical transitions (Markard et al., 2020;Schot and Kanger, 2018) raises significant challenges for real-world transition policy practices that need to be designed by multiple governing entities across various policy domains and governance levels (Rogge and Reichardt, 2016). In this sense, the transformation of whole socio-technical systems calls for more research to investigate focal policy mixes and overlapping policy subsystems, and how they are embedded in governance structures. Moreover, although recent research has paid attention to the temporal dynamics of layered policy mix elements and their characteristics Schmidt and Sewerin, 2019), the role of key actors, their interactions and change over time remain understudied. Some promising advances in this regard have borrowed ideas from the policy feedback framework (Béland, 2010;Jacobs and Weaver, 2015;Pierson, 1993), to offer a systemic view capturing the interactions and interdependencies between policy mixes, technological development and socio-political backgrounds (Edmondson et al., 2019;Schmid et al., 2020). However, while these approaches explicitly acknowledge the role of actors' agency in hampering or accelerating transitions, they have largely neglected an in-depth investigation of the governing entities designing policy mixes and their interactions over time.

Governing entities in sustainability transitions
According to Ossenbrink et al. (2019), governing entities refer to the public actors "in charge of the design, implementation, and governance of the focal policy mix" (p. 3). Linking complex policy mix elements with their designers offers a novel analytical lens for analysing different building blocks of the focal policy mix (Rogge and Reichardt, 2016). We therefore follow this practice-orientated definition focusing on public actors, while acknowledging that non-public actors also play a role in designing policy mixes.
Referring to governing entities generally as a type of public actor or a specific organisational form of the state, transitions scholars have investigated their role and their interactions with other types of actors (Avelino and Wittmayer, 2016;Borrás and Edler, 2020;Braams et al., 2021). Theoretically, conventional transition approaches stress the dominance of structural notions, such as "paradigm" and "regime", and the resulting selection environment in shaping the trajectory of sustainable innovations (Smith et al., 2010). In this regard, transition research has considered governing entities as formal institutions often acting as supporters of the incumbent socio-technical regime(s). As such, exogenous pressures from outside the state may need to be exerted in order to bring about transition-friendly policy actions (Normann, 2015(Normann, , 2017. A set of relevant research topics, such as power relations (Avelino and Wittmayer, 2016;Stirling, 2014), vested interests and resistance (Geels, 2014;Penna and Geels, 2012), as well as expectations and narratives (Budde and Konrad, 2019;Yang et al., 2020) have been applied in analyses of the complex interplay between state actors and other types of actors involved in real-world systemic transitions. Concentrating on the role of governmental actors, existing studies have not only identified different functional roles that the state can play in the governance of socio-technical systems (Borrás and Edler, 2020), but also pointed out the differences in ideal governmental roles over different transition phases (Rotmans et al., 2001). Despite such largely conceptual research efforts on the "state" or "government", researchers tend to be "inconsistent in their reference to actors at different levels of aggregation" (Avelino and Wittmayer, 2016), thus a clarification of these terms at specific levels (sectors, organisational actors or individual actors) is critical.
Increasing research interest in the politics of sustainability transitions Meadowcroft, 2011) has offered more fine-grained definitions and insights about the agency of public actors in policymaking and implementation. Transition scholars have also suggested applying policy process theories in studying politically controversial transition processes (Kern and Rogge, 2018). Such studies have investigated the beliefs, interests and resources of different policy actors in order to better explain various policy preferences, the successes or failures of significant policy changes and the design of policy mix elements in the context of socio-technical transitions (Gomel and Rogge, 2020;Markard et al., 2016;Normann, 2017). For instance, by considering the systemic influence of fiscal and extended administrative policy mix feedbacks, Edmondson et al. (2020) show how the dominance of a ministry -Her Majesty's Treasuryin the UK's policy subsystem reduced system resources supporting zero carbon homes. Although research concerning the specific role of administrative actors is still rare, this strand of literature has started to advance our understanding of the activities of governing entities in such governance and policymaking processes.

Towards greater attention to governing entities in policy mix research
In this article we aim to overcome the limited attention given to governing entities involved in designing policy mixes for sustainability transitions, to provide more nuanced insights into their roles and interactions. We focus on the administrative aspects of transition policy mixes, rather than addressing the politics-administration dichotomy (Georgiou, 2014;Svara, 2001Svara, , 2008. With this dichotomy being controversial in the area of public sciences and public administration studies, 2 we are not suggesting a clear distinction between "politics" and "administration" in the context of sustainability transitions either. However, by flagging this strand of literature, we want to encourage transition scholars to pay closer attention to administrative aspects, which have been mostly 2 Apart from the theoretical criticisms of the politics-administration dichotomy itself, the empirical applications of this Weberian perspective in different institutional contexts are also questionable. The one-party system in China, for instance, calls for a rethinking of the clear boundary between the political actors and administrative agencies (See Jing, 2017). treated as a self-evident facet of politics in existing transition studies. In doing so, we contribute to a newly emerging literature that highlights administrative elements in sustainability transitions (Braams et al., 2021). Due to the urgency of the climate crisis, we assume that this topic will become increasingly important.
In this vein, one of the most pressing challenges for accelerating unfolding transitions is that the effective employment of transition policy mixes requires different branches of governmental bodies to co-produce well-coordinated strategic actions in established governance systems (Braams et al., 2021;Markard et al., 2020). Conceptually, this recognition resonates with nascent discussions concerning the administrative complexity in socio-technical transitions. For example, Gjoshevski (2016) argues that the complex structure of administrative organisations might influence the coherence of cross-sectoral sustainability policies. Similarly, Haley et al. (2020) point out that "added administrative complexity" (p. 6) needs to be carefully considered in developing robust and resilient institutional frameworks for governing sustainable energy programs. Regarding the effective design of integrated policy mixes, Axsen et al. (2020) highlight that some policy instruments, even though they have considerable potential in achieving certain policy objectives, might be difficult to employ because of the complexity of existing administrative structures.
In order to further understand the administrative aspects of transition policy mixes, we focus on the role of multiple governing entities and their cross-organisational interactions over time. Capturing the complexity of policy spaces is a significant challenge in analysing policy mixes (Ossenbrink et al., 2019), as many activities of governing entities are embedded in overlapping policy subsystems. Current studies fall short of comprehensively identifying governing entities and systematically tracing their dynamic interactions. Such operationalisation challenges have hampered collective knowledge advances , therefore a systematic and replicable mapping approach of governing entities involved in the design of complex policy mixes is required for further empirical inquiries. In this regard, we attempt to introduce a document-based network approach, described in detail in the methodology section, and apply it to the national policy mix for EVs in China.

Case selection
We consider China's national policy mix for electric vehicles a suitable research case for three main reasons. First, regarding the case choice of electric vehicles, we contend that it represents a promising technological contribution to the vital decarbonisation of the transport sector. Globally, transportation accounted for approximately 24% of direct carbon dioxide emissions by 2019, although this metric declined temporarily during the COVID-19 pandemic (IEA, 2020a(IEA, , 2020b. Decarbonising road transport has attracted particular attention (Geels, 2012;Marletto, 2014), as this subsector contributes much more to total transport emissions than aviation and shipping activities (IEA, 2021). Moreover, considering the multi-system nature of unfolding e-mobility transitions driven by the diffusions of EVs, the role of well-coordinated policy efforts across different policy domains has been highlighted by recent studies Kotilainen et al., 2019).
Second, regarding the specific case of China, the rapid development of technological capabilities, manufacturing systems and market sales of EVs in the country is increasingly recognised as "globally central to low-carbon mobility transition" (Tyfield and Zuev, 2018, p. 259). As Fig. 1 demonstrates, there has been a dramatic rise in gross EV sales in China over the past ten years, thereby making China one of the largest contributors to the global EV rollout. In this vein, closer attention to Chinathe largest emerging economyis needed within the context of the global low-carbon transitions (Ely et al., 2019;Tyfield et al., 2015). Significant developments of EVs in China have been shown to be largely policy-induced over the past decades (Trencher et al., 2021;Wu et al., 2021). The empirical focus on China's EVs enables us to explore the state-led approach stimulating directed socio-technical changes in the context of latecomer economies towards a green economy, as highlighted by recent research calls (Lema et al., 2020;Pegels and Altenburg, 2020).
Third, the selection of the Chinese EV policy mix case enables us to engage with rich data about various governing entities in the design of a complex policy mix over time. This policy mix has been observed at different governance levels, containing ambitious strategies (Wu et al., 2021;Xu and Su, 2016) and various instrumentssuch as strategic road-mapping (Prud'homme, 2016;Wu et al., 2021), R&D subsidies (Hao et al., 2014;Ye et al., 2021), demonstration programmes (Gong et al., 2013;Zheng et al., 2012), and many other elements administrated and coordinated by various interactive governing entities (Liu and Kokko, 2013;Wang et al., 2021). A systematic delineation of these associated governing entities is still missing; however our methodology (see Section 4) allows us to address this gap.

Empirical specification
Inspired by existing guidance in this area (Rogge and Reichardt, 2016;Ossenbrink et al., 2019), four dimensions, including strategic intent, governance level, geography, and time are utilised in setting the boundaries of our empirical work, as summarised in Table 1.
As pointed out by Rogge and Reichardt (2016), it is essential for researchers to consider multiple policy domains when analysing specific policy mixes in the context of sustainability transitions. In seeking a comprehensive actor mapping of governing entities in our study, the use of pre-defined policy domains is questionable. More reflectively, Ossenbrink et al. (2019) criticise the inherent conceptual ambiguity embedded in these socially constructed "domain descriptors" (e.g. "industrial policy" and "climate policy"). They argue that it hampers the efforts of policy mix scholars with respect to developing a shared and accumulated knowledge base, even if individual cases have yielded important empirical data. As an alternative, Ossenbrink et al. (2019) suggest researchers use "strategic intent", defined as "the presumed strategic rationale of a given policy mix" (p. 2), in operationalising their empirical inquiries of policy mixes. In this paper, we follow this suggestion by focusing on governing entities involved in the design of the policy mix for "stimulating the research, development and diffusion of EVs".
Beyond delineating strategic intent, we focus on the policy mix designed by governing entities at the national level (governance level) for stimulating research, development and diffusion of EVs in China (geography). This presents a unique governance context that deserves a brief introduction. At the top of China's national government hierarchy lies the State Council, where the central administrative power is exercised under the political guidance expressed by the Central Committee of the Communist Party of China. 4 The State Council is constituted of departmental componentscollectively, ministries and commissionsaround different functional areas, supported by its General Office (see Xu and Weller, 2016, for detailed discussions). In a system described as "fragmented authoritarianism" (Lieberthal and Oksenberg, 1988), structural tensions embedded in the Chinese bureaucracy are observed between the central leadership and across ministerial bodies and territorial governments (e.g. city and province), which have been increasingly considered in investigating the institutional context of China's energy transitions (Cai and Aoyama, 2018;Lema and Ruby, 2007;Zhang and Andrews-Speed, 2020). These central-peripheral and horizontal tensions, moreover, also bring additional operational challenges to the employment of complex policy mixes navigating green transformations in the country (Huang, 2019). In this work, we focus on the national level, while acknowledging the importance of multi-level policy interactions in understanding China's e-mobility transitions (Helveston et al., 2019;Huang and Li, 2020).
Lastly, we focus on the role of governing entities in designing China's national EV policy mix between the years 2001 and 2020 (time). This chosen time frame covers the 10th (2001-2005), 11th (2006-2010), 12th (2011-2015) and 13th (2016-2020) Five-year Plan (FYP) periods. It thereby includes the launch of the first EV-specific project affiliated with the National High-Tech R&D Programme (863 Programme) and covers the Energy-saving and New Energy Vehicles Industry Development Plan (2012-2020). The relevance of these time-periods and key developments in the evolution of China's national EV policy mix are discussed in the next subsection.  (2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014)(2015)(2016)(2017)(2018)(2019)(2020) The national FYPs formulated every five years in China offer overarching guidance to steer the country's economic growth and industrial development (Naughton, 2017), and have been identified as a critical component of China's national policy framework around EVs (Yuan et al., 2015). As top-level programmatic documents, these national FYPs offer key vision statements and policy targets for China's overarching direction over each five-year period, and are subsequently supplemented by a panoply of field-specific FYPs drafted by responsible ministries as well as local plans drafted by governments at different governance levels (Naughton, 2017). The official FYP-based policy cycle has been used as an analytical structure for understanding strategic priorities, broader contexts and temporal dynamics of governmental activities in China (Hu, 2016;Li and Taeihagh, 2020;Yuan and Zuo, 2011), including recent national policy efforts concerning low-carbon innovations and energy transitions (Hepburn et al., 2021;Tyfield et al., 2015). In the remainder of this subsection, milestone policy events and key policy mix elements are briefly summarised around the four FYP periods covered in this study, thereby providing the backdrop to our empirical analysis. 5

10th five-year plan period from 2001 to 2005
In 2001, the importance of fuel-economy and low-emission vehicles with energy-efficient internal combustion engines (ICEs) and hybrid powertrains was highlighted by the 10th national FYP, 6 providing the overarching direction of technological activities across China's automotive sector (Winebrake et al., 2008). As a specific policy action under the 10th FYP, a far-reaching policy instrument, the first EV-specific project under the 10th FYP 863 Programme, was launched by the Ministry of Science and Technology (MOST) to stimulate research and development in EVs (Gong et al., 2013;Feng and Li, 2019). This funding scheme continuously provided considerable support through the following 11th and 12th FYP periods, with one respective EV-specific project under the 863 Programme in each.
More importantly, a "three-by-three" technology roadmap, was first introduced by this funding project (Xu and Su, 2016). This roadmap highlighted three powertrain trajectoriesbattery electric vehicles (BEVs), hybrid electric vehicles (HEVs), and fuel cell electric vehicles (FCEVs) -and three technological componentsmulti-fuel powertrain control systems, driving motors and associated control systems, and batteries and associated control systemsas national technological priorities; by doing do, it has "guided China's new energy vehicle development for the next 15 years" (ICCT, 2021, p. 2).

11th five-year plan period from 2006 to 2010
Given the increasing strategic importance of EVs across different policy domains during the period from 2006 to 2010 (Howell et al., 2014), the Chinese government started to establish an EV-specific regulatory framework and to nurture the niche market for the early rollout of EVsparticularly those produced by domestic manufacturers (Bohnsack, 2018). Notably, there was a round of administrative reformthe so-called "super-ministry reform" -at the national level in 2008, which led to an amalgamation of various existing entities to establish five new "super-ministries" (Yeo, 2009). For instance, the Ministry of Industry and Information Technology (MIIT) was established in 2008 for facilitating industrial development with coordinated strategic policy actions at the national level.
Within this context, EVs shaped the strategic direction in the structural adjustment and technological upgrading of China's automotive sector in 2009, 7 and were listed as one of seven strategic emerging industries in 2010. 8 To cope with such emerging sectoral dynamics, an entry regulation for EV manufacturers and their products was issued by the National Development and Reform Commission (NDRC) in 2007, 9 and amended by the newly established MIIT in 2009. 10 Inspired by the successful EV demonstration at Beijing Olympics in 2008, the first batch of national public demonstration pilot projects (2009)(2010)(2011)(2012), known as the "Ten Cities Thousand Vehicles" (TCTV) programme, was initially launched by the Ministry of Finance (MOF) and MOST in early 2009 11 and scaled up with MIIT and NDRC in 2010. 12 This programme mainly targeted the adoption and diffusion of EVs in the field of public transport, such as via buses and taxis, with subsidies offered by the central government and matched by local subsidies provided by the pilot cities (Bohnsack, 2018;Xu and Su, 2016).

12th five-year plan period from 2011 to 2015
The publication of the Energy-saving and New Energy Vehicles Industry Development Plan (2012-2020) 13 -thereafter, 2012 NEV Development Planissued by the State Council in 2012 was the most important milestone in the 12th FYP period. This 2012 NEV Development Plan provides a set of strategic policy elements relevant for the design of China's national EV policy mix, including a set of key targets (e.g. the total sales of BEVs and Plug-in HEVs to reach 500,000 by 2015 and more than 5000,000 by 2020) and associated 5 More detailed discussions about the historical development of China's EV policies can be found in a recent report published by ICCT (2021 principal plans (e.g. the Industrial Technology Innovation Programme for New Energy Vehicles). In the same year, the first fieldspecific 12th FYP on the science and technology (S&T) development of EVs 14 was issued by MOST, confirming the BEV-orientated strategy for achieving technological catch-up and providing a detailed list of prioritised technological directions. In addition, in 2015, a top-level industrial strategy -Made in China 2025 15 -was published by the State Council, in which the development of EVs was outlined as one of ten key industrial areas.
Guided by these strategic orientations, the second round of the national subsidy scheme funding public demonstration and deployment projects 16 (2013-2015) was inaugurated in 2013, in which the scale of pilot areas was extended from individual cities to cities and regions (Konda, 2022;Xu and Su, 2016). In order to incentive private buyers of EVs, a set of tax-based policy instruments was introduced by MOF, the State Taxation Administration (STA) and MIIT during the 12th FYP period, such as exemptions on the vehicle and vessel tax in 2012 17 and the vehicle acquisition tax in 2014. 18 Moreover, the role of EV charging infrastructures was raised at the national level toward the end of the FYP period (Xu and Su, 2016;Zhou et al., 2020), with two strategic documents issued by the general office of the State Council.

13th five-year plan period from 2016 to 2020
A number of policy mix elements were introduced from 2016 to 2020, the last FYP period covered by the 2012 NEV Development Plan. First, integrating the 863 Programme and many other public R&D projects, the newly established National Key R&D Programme also included an EV-specific funding call 19 for stimulating EV-related research and innovation activities within the 13th FYP period.
In this period, the strategic importance of EVs for combating air pollution problems 20 and achieving carbon emission reduction goals 21 was promoted by the State Council, and more policy pressures were exerted on conventional ICE vehicles. For this, a new policy instrument, the Dual Credit System was adopted in 2018, 22 requiring automotive manufacturers to adjust their product portfolio strategies by introducing the Corporate Average Fuel Consumption Credit and the New Energy Vehicle Credit (See Ou et al., 2018). With respect to addressing batteries as a key technological bottleneck for the industrial development of EVs, a battery-specific industrial development plan 23 was published by MIIT, NDRC, MOST and MOF in 2017; and another development plan concerning smart and connected vehicles 24 was issued by MIIT in 2018, in order to capture cross-sectoral cutting-edge technological opportunities.
Remedying the decrease in subsidies on EV purchases, financial support for the deployment of EV charging infrastructures 25 was encouraged by the central government, where the National Energy Administration (NEA) started to play a critical role in coordinating national subsidy schemes, alongside MOF, MOST, MIIT and NDRC. Meanwhile, to address a set of regulatory challenges around the construction of EV charging points and stations in certain regions such as residential areas 26 and national highways, 27 the increasing policy involvement of the Ministry of Housing and Urban-Rural Development (MOHURD) and the Ministry of Transport (MOT) was witnessed from 2016 to 2020.

Document-based network approach
To investigate the governing entities and their interactions in designing China's national EV policy mix over time, a documentbased Social Network Analysis (SNA) approach is applied. The SNA approach has been widely recognised in analysing complex dynamics of social systems in the field of social sciences (Borgatti et al., 2009). Recently, this method has seen increasing employment in studies of policy networks (McGee and Jones, 2019; Varone et al., 2017) and transition actors (Giurca and Metz, 2018;Scherrer et al., 2020). The SNA approach not only considers the role of individual actors in the big picture, but also enables researchers to investigate the interactions between multiple actors and to showcase the underlying agency structure. This approach, therefore, is very suitable for addressing our research questions.
We use policy documents, i.e. the physical manifestation and administrative form of designing policy mix elements, as a footprint of governmental activities within the focal policy subsystem(s). Such documents are drafted and issued by governing entities in the focal policy subsystem(s), which thereby exercise their authorised power and express their agency within the country's given governance structure (Duygan et al., 2019). In China, various (and even competing) ministerial bodies work together to develop substantial policies which translate the programmatic guidelines issued by the top authority, such as the State Council (Chen and Naughton, 2016). The "authorship" of these policy documents, in this sense, may be regarded as a proxy for the agency that governing entities exercise in designing policy mixes. Similarly, the "co-authorship" of policy documents is utilised as a way of measuring the formal interactions between different governing entities in such processes (Huang et al., 2015;Sun and Cao, 2018). Whilst "co-authorship" has limitations (especially in terms of capturing informal interactions between governing entities), we see it as a reliable proxy for the kinds of formal interactions which are the focus of this paper. Fig. 2 below illustrates how governing entities involved in the focal policy mix and their institutional network can be extracted and analysed in an intuitive and systematic way, even without the need for a prior delineation of all relevant policy mix elements.

Data collection, tidying and analysis
Operationalising the document-based SNA approach discussed above, we developed a stepwise process of collecting, tidying and analysing network data, as illustrated in Fig. 3.
In the first step of data collection, we chose the PKULAW policy database 28 of policy documents, as its comprehensiveness and reproducibility have been tested by recent studies (Li and Taeihagh, 2020;Yang and Huang, 2022). This choice avoids possible selection bias when using archival policy records provided by pre-identified core governing entities (Ossenbrink et al., 2019). As suggested by past research (Xu and Su, 2016;Yang et al., 2021), we utilise a set of four Chinese terms -New Energy Vehicle (in Chinese "新 能源汽车"), Electric Vehicle ("电动汽车"), Hybrid Electric Vehicle ("混合动力汽车") and Fuel Cell Electric Vehicle ("燃料电池汽车")in running full-text searches of national policy documents archived in the PKULAW policy database. With this keyword-based search strategy, our network dataset can comprehensively capture policy documents representing national policy efforts around all key EV technological trajectories highlighted by the Chinese government (see Section 3.3.). By implementing full-text search queries within the PKULAW database, both EV-focused policy documents and other documents considering EVs are collected. This addresses our research intention of offering a comprehensive mapping of governing entities in designing the policy mix for EVs across different policy domains (see Section 3.2.). As a result, we obtained a total of 1785 policy documents in our raw dataset. 29 For the second step of tidying the collected raw dataset, we proceeded through two sub-steps. First, to tackle the analytical challenge that the names of certain governing entities may have changed with administrative restructuring and integration over the past decades, 30 we compared and integrated historical names of these governing entities, and used their current names in our analysis in order to trace possible temporal differences over different FYP periods, as recommended by existing research practices Sun and Cao, 2018;Xu and Su, 2016). Second, in order to focus on the administrative aspects of transition policy mixes, we only kept documents issued by the State Council or at least one ministry-level (or vice ministry-level) governing entity listed in the current structure of the State Council. 31 By doing so, we ensured that all policy documents are related to the exercise of formal administrative authority while also including key non-administrative actors who collaborate with these listed governing entities, such as some organisational bodies of the Communist Party of China. Accordingly, our tidied dataset includes 1699 policy documents issued by 75 governing entities. 32 Finally, regarding data analysis we follow Varone et al. (2017) who stress that three elementsstructures, actors and relationsshould be considered in operationalising the SNA approach in investigating policy-relevant cases. Therefore, our data analysis proceeds in three sub-steps. In the first sub-step, we focus on the overarching network structure of governing entities in designing China's national EV policy mix, to gain an understanding of the big picture of how all governing entities work together. In this regard, a set of network-level metrics (Sun and Cao, 2018;Varone et al., 2017) are calculated and discussed across different FYP periods. In the second sub-step, at the node leveleach node representing a governing entitywe capture the practices of different governing entities and potential changing dynamics in designing the national policy mix stimulating research, development and diffusion of EVs. For this, we utilise three widely recognised but different centrality measuresdegree centrality, betweenness centrality and eigenvector centrality (Bonacich, 2007;Freeman, 1978) in investigating the specific role of governing entities. In the third sub-step, we extend our analytical focus from the node level to analyse interactions amongst different governing entities. By calculating the most frequent bilateral linkages between governing entities for each FYP period and applying the Louvain method for detecting intense communities 28 See www.pkulaw.cn. The PKULAW database contains policy documents issued by the Chinese government at different governance levels, and enables users to develop personalised search strings for identifying targeted policy documents within its retrieval system. 29 The raw dataset was accessed and downloaded from the PKULAW database on 5 January 2022, and all included policy documents mentioned at least one of the used four Chinese terms in their full texts. Documents concerning only administrative and legislative procedures, which are mostly removed in other research focusing on the delineation of policy mix elements (e.g. Xu & Su, 2016), have been retained in our raw dataset, as we contend that these documents also indicate agency-performing activities of their issuing governing entities within the government. 30 For instance, the State Environmental Protection Administration was promoted and renamed to be the Ministry of Environmental Protection in 2008, and then integrated into the newly established Ministry of Ecology and Environment in 2018. 31 See http://www.gov.cn/guowuyuan/zuzhi.htm, accessed 29/9/2022. Reference to this structure ensures a focus on governing entities with formal administrative power delegated by the State Council. 32 The full list of these identified 75 governing entities is provided in the appendix. (Blondel et al., 2008), we identify dense connections amongst governing entities in the whole networkshowcasing who is co-producing core policy efforts for systemic socio-technical changes.
Details of the main network metrics and analytical techniques employed for our empirical inquiries are summarised in Table 2. The igraph package (Csardi and Nepusz, 2006) of the R program is utilised for processing our network data, calculating main metrics and implementing the community detection algorithm. We utilised Gephi (Bastian et al., 2009) for visualising the network structures presented in our results section.
Moreover, to enhance the validity of our research and to better interpret its main quantitative insights (Wald, 2014) we supplemented our quantitative network analysis with qualitative insights drawing on milestone policy documents. These key documents, as introduced in Section 3.3., were identified by combining: 1) a list of milestone policy records (covering the period from 1991 to 2015)

Results
Over the past two decades, there has been a significant expansion of national government activity around EVs in China, as illustrated by an increase from 8 documents in 2001 to 144 documents in 2020. On the basis of the numbers of EV-related policy documents issued per year (Fig. 4), it is obvious that the Chinese government has made significant efforts to develop the EV policy mix in recent years. Meanwhile, from the increasing number of jointly-issued policy documents, it is also clear that an increasing number of policymaking activities involve collaborations across different governing entitiesfrom 2 in 2001 to 37 in 2020. Notably, there is an increasing percentage of policy documents with multiple issuing entities over the past ten years. It is these jointly-issued documents (23.66% of all policy documents issued between 2001 and 2020) which provide the empirical foundation for investigating crossorganisational interactions between governing entities in the design of China's national policy mix.

Network evolution
With the expanding governmental activities addressing the research, development and diffusion of EVs, the number of governing entitiesindicated by the number of nodes in Table 3 has significantly increased over time.
Since the 10th FYP, two emerging intertwined policy tendenciesthe rising industrial priority of the automotive sector, and the increasing ambition to save energy use and reduce fuel emissionshelp explain the employment of systemic EV policy efforts carried out by the Chinese government (Gan, 2003;Winebrake et al., 2008). From 2001 to 2005, only 7 governmental ministries and commissions were observed to carry out policy activities concerning EVs, which contrasts with 49 in 2011-2015 and 67 in 2016-2020. During the 11th FYP and the 12th FYP, a series of national-level strategic documents, such as the 2012 NEV Development Plan, were issued by the State Council and its general office, providing strategic and comprehensive guidelines for steering the development of EVs from the top level (Xu and Su, 2016). During the 13th FYP period, there is a significant rise in policy activities around EVs in China, as the government aimed to accelerate the market deployment of EVs across the whole country (as planned by the 2012 NEV Development Plan). This highlights the expanding domains of relevance for China's national EV policy mix, which led to the involvement of more governing entities (Liu and Kokko, 2013;.

Table 2
Main network metrics and analytical techniques in this work. Source: Authors' own summary.

Number of Nodes Indicating the extensiveness of governing entities by counting the number of network nodes Number of Edges
Indicating the size of the network of governing entities by counting the number of network edges Edge Density Indicating the denseness of the network of governing entities by calculating the value that the actual number of edges is divided by the possible number of edges amongst all nodes in that network (e.g. Sun and Cao, 2018) Components Ratio Indicating the cohesion of the network of governing entities by calculating the value that the number of network components minus one is divided by the number of network nodes minus one (e.g. Park et al.,

2019) Clustering Coefficient
Indicating the degree to which governing entities in each network tend to interact with each other as clusters by calculating the ratio of triangles and connected triples in a certain network (e.g. Sun and Cao, 2018) Governing Entity Degree Centrality Identifying governing entities with most connections that could play a leading role in mobilising other actors (e.g. Sun and Cao, 2018) Betweenness Centrality Identifying key governing entities playing a brokerage role in bridging various and less connected actor groups (e.g. Howlett et al., 2017), measured by the number of shortest paths each node lies Eigenvector Centrality Identifying key governing entities surrounded by the most influential neighbours and involved in the core actor group of the network (e.g. Huang et al., 2018), measured by the values of the first eigenvector of the adjacency matrix

Number of Bilateral Linkages
Indicating most frequent bilateral interactions for each FYP period by counting the number of shared edges between governing entities Community Detection Identifying most intense groups of governing entities by applying the Louvain method with the multilevel optimisation of modularity ( With respect to the size of networks, as indicated by the number of network edges in Table 3, an increasing number of crossorganisational interactions can be found in the expanding network of governing entities behind China's EV policy mix: up from 4 interactions in the 10th FYP period to 1047 in the 13th FYP period. It is important to stress the large increase seen between 2016 and 2020, as this trend corresponds with a shift in the strategic focus of the Chinese government after the 12th FYP period towards massmarket adoption and diffusion of EVs and the nationwide deployment of charging infrastructure (Trencher et al., 2021), for which two programmatic documents were issued by the State Council in 2014 35 and 2015. 36 For instance, the National Energy Administration (NEA) started to play a role in planning public subsidy schemes (highlighting EV charging infrastructures) for the 13th FYP period, 37 collaborating with existing responsible agenciesthe Ministry of Finance (MOF), the Ministry of Science and Technology (MOST), the Ministry of Industry and Information Technology (MIIT) and the National Development and Reform Commission (NDRC) -that had   38,39,40, . 41 Although the numbers of governing entities and interactions have increased continuously over the past two decades, the metrics concerning overarching network structures show that this evolutionary process does not follow a linear pattern (see Table 3). From 2006 to 2015, although more and more actors were involved in the design of the EV policy mix, the network became relatively less dense. Indeed, several isolated groups of governing entities (i.e. those not actively connected to the rest of the network for certain FYP periods) appeared and the network became in general less clustered. For instance, the Ministry of Transport (MOT) was isolated from the rest of the network during the 11th FYP period, as at that time most policy documents solely issued by this entity only considered the potential of EVs in achieving transport-specific energy-saving targets. 42 This implies many governing entities still worked within their own jurisdictions, and the collaborations between different policy domains were relatively low, despite the increasing number of governing entities in this period. However, this trend has been reversed since 2015 as the network became very dense and cohesive, which suggests considerable numbers of governing entities from different policy domains worked together through a multitude of policymaking practices performed in the 13th FYP period. To display these evolutionary dynamics in the network structures of emobility governing entities, visualisations of subgraphs for these four FYP periods are offered in Fig. 5.

Governing entities
In order to track the positions and the changes of governing entities during such network evolution dynamics, we distinguish between three types of centrality measuresdegree centrality, betweenness centrality and eigenvector centralityin our analysis for better understanding the divergent roles of governing entities in designing China's EV policy mix at the national level (see Table 4).
In terms of degree centrality, NDRC has occupied the most prominent position throughout the period of study, from 2001 to 2020, which means that it has the highest number of linkages with other actors in the observed network. As a key element in the institutional legacy of China's planned economy, NDRC has a higher authority than other ministries and commissions under the State Council (Yeo, 2009), and the scope of its jurisdictional authorities covers not only industrial development but also energy and climate strategies (Tsang and Kolk, 2010). In this sense, it is not surprising that it has taken on a key role in coordinating and mobilising cross-sectoral policy efforts around EVs in China (Zheng et al., 2012). While NDRC has retained a key role over different periods, however, it is worth also noting the role of MOST and the Ministry of Commerce (MOFCOM) in the earlier two periods (2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010). The development of national R&D capacity on EVs was an important policy goal in this earlier phase, as highlighted in the 10th FYP for S&T development. 43 This therefore required a set of policy instruments, such as public research programmes and subsidies for EV-related technology imports and foreign direct investments (FDI), to mobilise cross-sectoral policy actions (Liu and Kokko, 2013), for which MOST and MOFCOM were the main responsible entities during these periods. For instance, the Catalogue for FDI in High-tech Products, 44 which EVs were highlighted, was jointly-published by MOST and MOFCOM in 2003. Similarly, MOF and MIIT also played critical roles in terms of coordinating and mobilising massive resources for public demonstration and market diffusion of EVs through purchase subsidies and industry support programmes in the 12th FYP and 13th FYP (Wu et al., 2021).
Regarding the brokerage role indicated by betweenness centrality, MOF displays the highest measure in the established network in the overall period from 2001 to 2020, which means that MOF has occupied the most significant position in terms of connecting different pairs of governing entities as an institutional broker. On the one hand, it is important to recognise the large public financial resources used in supporting the design of China's EV policy mix during this period. Since the deployment of the first round of national public demonstration projects in 2009 (Xu and Su, 2016), MOF has played a principal role in coordinating and monitoring these subsidy-related activities for pilot cities and regions, public service sectors and individual customers (Xu and Su, 2016;Zhang and Bai, 2017)   are not normally considered EV-related governing entities such as the Ministry of Justice (MOJ) and the Supreme People's Procuratorate (SPP), which meant that it was critical in terms of connecting the "core" and the "periphery" layers of the established network.
The eigenvector centrality reveals the core governing entities surrounded by the most influential neighbouring entities over time. MIIT has taken the top position since the 12th FYP period, which means that MIIT has been at the core of the leading group of most influential governing entities in designing the EV policy mix over the past decade. Established in 2008, MIIT took over from NDRC and

Table 4
Top 3 leading governing entities ranked by centrality metrics. Source: Authors' own calculation. became the main governing entity responsible for technology standardisation, emissions monitoring and project evaluation (Zheng et al., 2012). Before this, MOST can be regarded as the core governing entity managing national policies in research, development and demonstration of EVs. It is widely recognised that the EV-specific 863 Programme, launched by MOST in 2001, guided systemic policy actions and specific technological directions of EVs at the national level in China during this earlier period (ICCT, 2021). As China's e-mobility transition proceeded, the role of MOST became less prominent, compared to its dominance in the pre-development and exploration phase over the first ten years. However, this shift of administrative leadership, as Wachtmeister (2014) points out, was not very smooth as the struggles between MIIT and MOST influenced the promulgation of the 2012 NEV Development Plan. The relatively stable core position occupied by MIIT since that time can be interpreted as a strategic requirement for deepening and widening the ongoing e-mobility transition at China's national level, through processes such as the acceleration of market diffusion, the development of battery recycling systems and the rollout of charging infrastructures (Trencher et al., 2021;Wu et al., 2021). During the 13th FYP period, the role of the State Administration for Market Regulation (SAMR) was also strengthened, as it interacted more with leading entities, such as MIIT and MOF, throughout the design and revision of the Dual Credit System 50, . 51

Cross-organisational interactions
Changes in the positions of governing entities have resulted in a very complex and evolving network in the past 20 years, with notable cross-organisational interactions. As Table 5 shows, the most frequent bilateral collaboration involved MIIT and NDRC, both of which played a critical role in guiding the research, development and diffusion of EVs in China, especially for the most recent 13th FYP period. It is worth noting that 5 of the top 10 collaborative relationships involved MIIT, providing further evidence for its dominant role in designing China's EV policy mix.
Overall, it seems that MIIT, NDRC, MOST and MOF represent a highly interactive group in employing policy actions for the development of EVs in China, as 8 of the top 10 bilateral collaborative relationships include at least one of them. However, other actors, such as MOFCOM, SAMR and STA, should also be noted, for their interactions with other governing entities in supporting the trade, financial subsidies and market regulations.
Going beyond the above focus on cross-organisational interactions at the two-actor scale, we use the Louvain community detection technique to identify multi-actor groups or communities constituted by intensive ties in the institutional network, extracted from the whole dataset from 2001 to 2020. By doing so, we identify 11 actor groups (for an overview of the membership of governing entities see Table 6).
First, combining the above discussions on centrality metrics and two-actor mode interactions, Group I, including MIIT, NDRC, MOST, MOF and other actors, could be regarded as the core community. Interestingly, NEA was found in this core group, indicating has been playing an increasingly important role in terms of accelerating the deployment of EV charging infrastructure. This implies the policy importance of whole-system transformation has been recognised by the Chinese government, which also seems to be reflected by the intensive interaction amongst governing entities in Group V, including MOT, Ministry of Housing and Urban-Rural Development (MOHURD), Ministry of Ecology and Environment (MEE), Ministry of Public Security (MPS), and others.
Second, governing entities in Group II, such as MOFCOM, SAMR, National Natural Science Foundation of China (NSFC, managed by MOST), can be regarded as the background support of China's national innovation system for general research and innovation activities, and thus also of relevance for EV.
Third, the existence of Group III and Group IV is consistent with findings from previous research (Sun and Cao, 2018), which implies that the Central Committee of the Communist Party of China and the State Council (Group III) and their general offices (Group IV) exist on a higher administrative level than other governing entities. They do not jointly issue policy documents with governmental ministries in China's political system, while the strategic policy documents issued by them, such as the 2012 NEV Development Plan, have supreme authority in China's policy system.
Fourth, although the number of actors involved in Group VI is the largest, we contend these actors provide complementary support to general policy items that do not target technology-specific activities. For instance, the Ministry of Education (MOE) and the Ministry of Human Resources and Social Security (MOHRSS) have issued a series of policy documents concerning the EV-related professional education and labour training projects, while research and innovation activities are not included much in these policy initiatives.
Fifth, the remaining actor groups only include one actor, which suggests these actors hardly interacted with others. Although most of them do not particularly relate to China's e-mobility transition, the Central Leading Group for the Reform of Official Vehicle Systems (CLGROVS, launched by the State Council) has been coordinating and encouraging the public purchase of government vehicles across all policy domains.
Although these different actor groups have been discussed individually, it is important to recognise that they also interact with each other at the network level. A visualisation of the whole institutional network over the period 2001-2020, in which the nodes have been coloured by their group membership, is provided in Fig. 6. Clearly, most actor groups interacted with each other substantially, and some of the aforementioned brokers, such as MOF, MIIT and MOFCOM, have bridged different portions of the whole network. 50 See http://www.gov.cn/xinwen/2017-09/28/content_5228217.htm, accessed 29/9/2022. 51 See http://www.gov.cn/zhengce/zhengceku/2020-06/22/content_5521144.htm, accessed 29/9/2022.

Discussion
Based on the case of China's national policy mix for electric vehicles (EVs) -one of the central technologies for the transition to lowcarbon e-mobilityour research reveals the temporal dynamics of multiple governing entities and their cross-organisational interactions in the design of complex policy mixes for multi-system sustainability transitions. Such transition policy mixes should be comprehensive, including setting long-term objectives, designing principal plans and adopting a combination of different policy instruments guiding and accelerating the processes of "creative destruction" within whole socio-technical systems (Kivimaa and Kern, 2016;Rogge and Reichardt, 2016). As our analysis suggests, no governing entity on its own possesses all the necessary jurisdictional authority or adequate resources to design such transition policy mixes. Instead, the complexity of the task raises significant administrative challenges for the state-led approach to eliciting systemic changes across different governance contexts (Borrás and Edler, 2020;Trencher et al., 2021). Here, we discuss three main insights into such administrative aspects of transition policy mixesan area that has so far been largely neglected in transition studies.  Table 6 The list of group membership of governing entities. Source: Authors' own calculation.

Group
Actor Group Actor Group Actor

Temporal dynamics
Our first insight concerns the temporal dynamics of the administrative aspects of transition policy mixes. More precisely, as transitions advance, we find that administrative arrangements tend to become more dynamic and complex, thereby responding to changing policy priorities induced by phase-specific socio-technical changes. This significant temporal dynamic, in our view, can be understood as an institutional response to changing policy challenges at different transition phases (Rotmans et al., 2001), which appear through a co-evolutionary dynamic of policy mix changes and socio-technical changes (Edmondson et al., 2019). In China, the increasingly extensive participation of interacting governing entities across different policy domains could be seen as a result of the transition to e-mobility entering its acceleration phase (Wu et al., 2021). In this context, multi-system interactions and their coordination have thus become more relevant for adding and revising various EV policy mix elements over time.
Regarding the changes in China's strategic priorities for EVs and the wider e-mobility transition, recent studies have noted that the scope of the national EV policy mix has been extended from stimulating and supporting alternative automobile technologies to embedding them into current mobility (and other relevant) systems (Trencher et al., 2021;Wu et al., 2021). This is mirrored by the rise of certain governing entities in the administrative structure underpinning the design of China's national EV policy mix. We find that newly added policy goals were addressed by employing various policy instruments designed and "owned" by different governing entities (Ossenbrink et al., 2019). For instance, the increasingly significant role of NEA and MOHURD in the observed network can be attributed to specific policy goals around the deployment of EV charging points, 52 which has become a strategic priority due to accelerating socio-technical changes in recent years. Interestingly, we also found the appearance of relatively new governing entities, as existing policy instruments have been revised. For instance, in 2009, the first document concerning the EV public demonstration  54,55, . 56 Another example is the vehicle and vessel tax exemption for EVs that was initially proposed by MOF, STA and MIIT in 2012 57 ; later in 2018, MOT started to play a role in issuing new policy documents for the administration of this policy instrument. 58 As such, this increasing complexity of administrative arrangements resonates with the "patching" pattern identified by scholars in analysing policy mix changes . More precisely, we find that those newly emerging governing entities have not replaced the leading incumbent agencies such as MIIT in designing China's national EV policy mix, but instead have supplemented them.

Top-level guidance and coordination
Our second key insight addresses the relevance of top-level guidance and coordination in the design of China's EV policy mix. That is, top-level coordination of leading governing entities appears to be critical in employing transformative policy efforts. Given the complexity of the diverse policy goals emerging within different transition phases, the strategic vision, outlined by the State Council (SC) and its General Office (GO_SC), fulfilled a critical function by providing top-level guidance for the active involvement of key governing entities. As outlined in Table 7, a set of EV-specific strategic documents has been issued by the State Council or its general office, which supported the sequential development of the policy mix over time.
To address the coordination challenge of managing increasing interactions between the growing number of governing entities, an Inter-ministerial Joint Meeting System was established in 2013 -as mandated by the 2012 NEV Development Plan. Managed by MIIT, by 2020 this inter-ministerial coordination mechanism included 20 ministries, commissions and agencies under the State Council. 59 Its role can be illustrated with two examples. First, following the Guideline on Accelerating the Deployment of New Energy Vehicles 60 published in 2014, various governing entities proposed or revised a set of policy instruments targeting different aspects of the market deployment of EVs (Xu and Su, 2016), and the coordinating role of the Inter-ministerial Joint Meeting System was reiterated. Second, in 2015, the Guideline on Accelerating the Construction of Electric Vehicle Charging Infrastructures 61 was issued specifically to promote the development of EV charging infrastructures in China, highlighting the role of NDRC and NEA in mobilising and coordinating different ministries and commissions via the established Inter-ministerial Joint Meeting System. These findings resonate with the insights offered by Braams et al. (2021), who highlight that coordination mechanisms between public agencies are key for developing internal capabilities and governance structures in order to address transition goals. Moreover, recent policy mix research has also pointed to the importance of introducing new practices and routines to existing organisations. This contributes to destabilising current regimes, thereby improving policy mix coherence . In our view, the case of China provides an interesting example of how strategic policy initiatives (such as the above-mentioned NEV Development Plan, Inter-ministerial Meetings and Guideline) can play a role in supporting the coordination of new practices and routines amongst various governing entities within the existing administrative structure.

Broader political factors
A third key insight captures the influence of broader political factors. In this sense, our analysis supports Edmondson et al. (2019) who note that broader political factors can affect the policy subsystem directly as "exogenous conditions". A prime example to this is that both the current leading governing entity (MIIT) and the "rising star" (NEA) in designing China's national EV policy mix were launched in 2008, when the Chinese government underwent the political reconfiguration known as the "super ministry reform". This reform aimed to maintain legitimacy and improve efficiency and coordination in the face of insufficient market regulation, energy shortage, environmental pollution and social security (Yeo, 2009). Clearly, this profound reconfiguration of China's administrative structure, under significant political pressure from top-level leaders, was not specifically designed for governing EVs, however, it Table 7 Key EV-specific strategic documents issued by the State Council and its General Office. nonetheless laid the organisational foundation for the observed evolutionary development of China's EV policy mix in the following periods.
We therefore call for more attention to the impact of such broader political factors on the administrative aspects of policy mixes for sustainability transitions in future research. We also argue that the political feasibility of radical governance reforms, such as the establishment of new "super ministries", deserves further examinations in other country-specific transition cases (Markard et al., 2020).

Conclusion
Investigating the governing entities behind China's national EV policy mix, this work has offered empirical insights into the increasing administrative complexity involved in governing sustainability transitions beyond the emergence phase. By selecting electric vehicles (EVs) as one of the core low-carbon alternatives at the intersection of electricity and mobility systems and by tracing the development of policies promoting this cross-sectoral technology from 2001 to 2020, our focus on governing entities enabled us to identify the growing cross-organisational interactions involved in e-mobility transition policies. As our document-based network analysis shows, although an increasing number of governing entities have become involved in the design of China's national EV policy mix, a small group of core governing entities with intensive interactions has emerged. We find that together with these evolving administrative dynamics, increasingly complex policy efforts spanning different policy domains are mobilised and coordinated. Key aspects of these dynamic administrative arrangements concern the role of phase-specific policy priorities, top-level strategic guidance and broader political factors.
While our paper provides a first step into unpacking the governing entities and interactions behind policy mixes for sustainability transitions, we acknowledge four limitations and based on these derive future research avenues for investigating the administrative aspects of transition policy mixes. First, although the analytical approach developed in this work has the benefit of not requiring researchers to identify and delineate policy mix elements up-front, by itself it is not able to shed sufficient light on the qualitative linkages between policy mix elements and policy design processes (Rogge and Reichardt, 2016), and especially on the role of politics and underlying power dynamics. Building upon the mapping insights offered by our methodological approach, future research should investigate the role of these governing entities and their interactions in articulating specific policy mix elements and the resulting policy mix changes through in-depth qualitative methods. In this regard, the full politics-policies-administration nexus could be considered as a conceptual lens to investigate transitions, with special consideration of the "policy coordination failure" identified by Weber and Rohracher (2012).
Second, our empirical analysis relies on a single type of data (i.e. policy documents) that only reflects what happened with respect to formal cross-organisational interactions in policymaking processes. Future research could pay more attention to other, more informal interactions between governing entities, not only during policy formulation but also in other policy stages such as monitoring and evaluation of policies. This may present methodological challenges around access to meeting records and require new analytical techniques for novel data sources.
Third, we have not explicitly searched for governing entities associated with the regime (internal combustion engine vehicles), and thus acknowledge insufficient attention to the destabilisation of this regime and broader processes of exnovation. This shortcoming could be addressed by extending or changing the search keywords used to map the documents and governing entities associated with these processes as a precondition for better understanding the corresponding changes in their agency and interactions.
Last but not least, our analysis was focused on the national EV policy mix in China, a country with significant socio-political differences to many other countries. It would therefore be valuable if future research would investigate governing entities and their interactions in other countries' transitions. Further, we suggest that more attention to those countries from the Global South, especially these emerging economies in which the state plays a significant role in promoting green transformations, will be very important in overcoming "European bias" in transition studies (Markard et al., 2012). We also suggest applying our method to different governance levels (e.g. sub-national province or state levels). Together, these extensions would enable cross-country comparisons at multiple scales that could deepen our insights into governing transitions.
Overall, we conclude that there is a need for more research on the administration of policy mixes for sustainability transitions, so as to better understand the practical administrative challenges and better inform those involved in transition policy making.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability
Data will be made available on request.
the European Union's Horizon 2020 research and innovation programme (grant agreement No 852730). The authors thank two anonymous reviewers for their constructive comments on earlier drafts of the manuscript. The authors also thank Nicholas Goedeking and Asli Ates for fruitful team discussions throughout the development of the paper. The preliminary results of this work were presented at the 2021 International Conference on Sustainability Transitions (Karlsruhe), and the authors would like to thank all participants for their insightful feedback. QS is grateful to Professor Yuan Zhou (Liu et al., 2011).