— FOUR MODES OF NEIGHBOURHOOD GOVERNANCE: The View from Nanjing, China

In a context of global moves towards decentralization and neighbourhood governance, this article focuses on neighbourhood governance in Nanjing, China. We draw on interviews and observations in 32 neighbourhoods to examine how neighbourhood governance plays out in different neighbourhoods. We identify and describe four modes of governance––collective consumption, service privatization, civic provision and statesponsored––to argue that neighbourhood governance develops on the ground in diverse and complex ways, necessitating scholars to be cautious when seeking to generalize about it at the scale of the city, let alone that of the nation-state or the globe. We argue that relationships between actors are important units of analysis when considering how effective governance is achieved in different neighbourhoods; diversity and complexity in neighbourhood governance partly reflect the role of the state in these relationships. In turn the role of the state partly reflects processes of policy evolution in particular neighbourhoods. Introduction Across the world, over the past three decades, there have been moves towards decentralization and neighbourhood governance. The reasons for this ‘global drive’ towards decentralization have been varied (Ezcurra and Rodríguez-Pose,  2013). A ‘revisionist liberalism’ sees a need to supplement market deregulation with ‘good governance’ (Mohan and Stokke,  2000), including ‘sound institutions’ at the national, regional and local scales (Hafteck,  2003). In democracies, local institutions have been viewed as more likely to engage citizens (Blakeley,  2010) and to deliver services responsive to citizen-consumers (Mohan and Stokke, 2000). International governmental organizations have seen decentralization to local governments, private-sector firms and third-sector organizations as one means of promoting democratic participation and accountability and ultimately governmental effectiveness (Batterbury and Fernando, 2006). These moves have been accompanied by debates about whether decentralization and the rise of neighbourhood governance reflect developments in ‘global policy’ (Cochrane, 2007)––especially neoliberalization––or more national, regional and local processes. Brenner and Theodore (2002) assert that ‘the new localism’ should be viewed as the ‘spaces of neoliberalism’. Decentralization should be seen as a response to the absence of a sustainable regulatory fix at the national scale in the context of globalization. Within the literature on neoliberalism, there have been debates between ‘Marxists’ and ‘Foucauldians’ (Peck and Tickell, 2012). The former imagine an uppercase Neoliberalism: a class-based ideology, characterized by fixed attributes, hierarchical power and global reach (Ong, 2007). The latter imagine a lower-case neoliberalism: a logic of governing and technique of administration, characterized by contingency and hybridity (ibid.). These debates have generated a range of useful concepts for thinking about policy developments around the world. Brenner and Theodore (2016) distinguish between ‘neoliberal ideology’ and ‘actually existing neoliberalism’. Peck et al. (2009)


Introduction
Across the world, over the past three decades, there have been moves towards decentralization and neighbourhood governance. The reasons for this 'global drive' towards decentralization have been varied (Ezcurra and Rodríguez-Pose, 2013). A 'revisionist liberalism' sees a need to supplement market deregulation with 'good governance' (Mohan and Stokke, 2000), including 'sound institutions' at the national, regional and local scales (Hafteck, 2003). In democracies, local institutions have been viewed as more likely to engage citizens (Blakeley, 2010) and to deliver services responsive to citizen-consumers (Mohan and Stokke, 2000). International governmental organizations have seen decentralization to local governments, private-sector firms and third-sector organizations as one means of promoting democratic participation and accountability and ultimately governmental effectiveness (Batterbury and Fernando, 2006).
These moves have been accompanied by debates about whether decentralization and the rise of neighbourhood governance reflect developments in 'global policy' (Cochrane, 2007)--especially neoliberalization--or more national, regional and local processes. Brenner and Theodore (2002) assert that 'the new localism' should be viewed as the 'spaces of neoliberalism'. Decentralization should be seen as a response to the absence of a sustainable regulatory fix at the national scale in the context of globalization. Within the literature on neoliberalism, there have been debates between 'Marxists' and 'Foucauldians' (Peck and Tickell, 2012). The former imagine an uppercase Neoliberalism: a class-based ideology, characterized by fixed attributes, hierarchical power and global reach (Ong, 2007). The latter imagine a lower-case neoliberalism: a logic of governing and technique of administration, characterized by contingency and hybridity (ibid.). These debates have generated a range of useful concepts for thinking about policy developments around the world. Brenner and Theodore (2016) distinguish between 'neoliberal ideology' and 'actually existing neoliberalism'. Peck et al. (2009) indicate that we should focus on neoliberal ideology, but also on the ongoing process of neoliberalization, which happens across uneven institutional landscapes and produces 'varieties of neoliberalism' or 'localized neoliberalizations' (see also Brenner et al., 2010).
Such concepts have not been enough for some scholars, however, for whom the frame of neoliberalism, however well-developed, obscures too much of contemporary urban policy development. These scholars believe that cities may find or place themselves in global networks of (neoliberal) policy circulation, but are also territories with histories that make them distinctive places (Robinson, 2005). Furthermore, where cities do find or place themselves in policy networks, these networks are multiple and not only global but regional too (Parnell and Robinson, 2006).
Widespread moves towards decentralization and ongoing debates about neoliberalization provide the broad context for this article. It is based on a study of neighbourhood governance in urban China and makes three contributions. First, it reports on recent developments in the Chinese case, which is interesting because phenomena such as the rise of private neighbourhoods and the emergence of networked governance at the neighbourhood scale remind some commentators of developments in global or Western urban policy commonly interpreted as expressions of neoliberalism (Lee and Zhu, 2006;He and Wu, 2009;Wu, 2010;. However, the continued presence of a strong and interventionist national state in China has led some to recommend caution in making such interpretations (Cartier, 2011;Wu, 2018;Zhou et al., 2019).
Secondly, in this article we report on recent developments in neighbourhood governance in the case of Nanjing, which is interesting because Nanjing is both an 'ordinary city' of the kind often neglected in the construction of urban theory (Robinson, 2006) and a 'prototypical' city in the terms used by Brenner (2003) and in the context of China. For Brenner, prototypes are the first cases of something likely to become more generalized. When we focus on neighbourhood governance in China, Nanjing appears to be such a case. In 1999, it was selected as one of 12 Experimental Cities for Neighbourhood Governance and Community Reform. Between 2012 and 2019 it hosted eight National Experimental Zones for Community Governance and Service Innovation.
Thirdly, in this article we take a mid-level view of neighbourhood governance. Our focus is neither on national or citywide policies (a view from above that overlooks diversity on the ground), nor on only one or two neighbourhoods (a narrow focus that makes generalization and theory building difficult). Rather, in our study we compared neighbourhood governance on the ground in 32 different neighbourhoods of Nanjing. What was made visible by this view? Neighbourhood governance plays out in Nanjing in diverse and complex ways, and we should be cautious when generalizing about decentralization--at the scale of the city, let alone of the nation-state or the globe. Still, beyond a general claim about complexity and diversity, four modes of neighbourhood governance can be identified in Nanjing. We derive these four modes from critiques of existing frameworks--generated from both Chinese and Western contexts--and analyses of empirical material collected in Nanjing. The alternative framework we present distinguishes modes of neighbourhood governance by their dominant relationship (as opposed to their dominant actor or rationale) and captures better how neighbourhood governance is achieved and what governance arrangements are currently found in Nanjing's neighbourhoods. These arrangements are relevant to global debates about local governance and the role of the state, and at the same time firmly embedded in the particularities of Chinese urban society.

Neighbourhood governance in urban China
'Neighbourhood', in this article, is taken to mean shequ in the Chinese context--'the collective social body formed by those living within a defined geographic boundary' (Ministry of Civil Affairs of the People's Republic of China, 2000). Defined in this way, neighbourhoods often cover one or more residential estates (xiaoqus) and coincide with the administrative territory of the Residents' Committee (RC)--an important institution of Chinese local government (Yip, 2014).
These associations with xiaoqus and the RC bestow a triple identity on the neighbourhood in urban China. As a spatial entity with clear boundaries, the neighbourhood is a platform for material exchange based on contractual relationships and clearly defined property rights. The neighbourhood is a social entity where social ties develop and collective actions are organized on the basis of shared values and common goals. These first and second identities remind some observers of neoliberal policies emphasizing private property and market exchange alongside social capital and community self-governance. However, the Chinese neighbourhood is also a unit of administration. RCs are vehicles for party leadership at the neighbourhood scale. Through them, policy interventions are made, access to resources is provided, and opportunities for participation are selectively offered.
The development of neighbourhood governance can happen by quick and violent imposition from outside, or through more gradual and peaceful internal processes (Thurston, 1998). In the section that follows, we review existing studies of neighbourhood governance in China along these two lines. First, we consider topdown promotion of neighbourhood governance by way of strengthened state agencies. Thereafter, we consider the bottom-up shaping of neighbourhood governance by civilsociety organizations.
-Promoting neighbourhood governance from the top down The retreat of the Chinese state in the 1990s, witnessed in cities particularly in the demise of state-owned enterprises and the privatization of housing, left a vacuum in urban governance at the neighbourhood level (Wu, 2002). The top-down perspective on neighbourhood governance in China focuses on the actions of the state in filling this vacuum (see, for example, Tomba, 2014;Wu, 2018). Since the 1990s, the local state has been 'reorganized' in China (Sigley, 2006) through a national community building programme and various neighbourhood governance innovation projects. The principles of community building, as summarized by Shieh (2011), include state retreat from welfare responsibilities, maximizing the contribution of societal actors to service provision, and strengthening neighbourhood-based self-governance.
This process of decentralization has been 'fragmented' and 'ambiguous' (Zhou, 2014). In some communities, RCs, although legally defined as 'autonomous mass organizations' (National People's Congress, 1989), have been revitalized as 'nerve tips' of the state with new powers and responsibilities (Read, 2000). New neighbourhood service systems have been established, made up of party secretaries, outposts of government departments, professional community working stations and RC-led civic groups. RCs have been vertically integrated into the governance networks of local and super-local authorities (Ohmer, 2007;Heberer, 2009) focused primarily on maintaining social stability and enhancing state legitimacy (Yip, 2014). However, in other communities-especially gated communities--RCs have been relatively marginal actors (Min, 2009). Furthermore, where decentralization failed to find new social and private actors who were capable or willing to participate in neighbourhood governance, there has been a 'return of the state' in the form of direct intervention by state bureaucracies (Wu, 2018) and micro-governing by local authorities (Tomba, 2014). -

Constructing neighbourhood governance from the bottom up
This period was characterized by local state reorganization and, in dialectical relationship to this, also by changes in homeownership, the rise of homeowners as a social force and new institutions of neighbourhood governance reflecting these developments. The most important of these new institutions are the Property Management Company (PMC), a professional provider of 'territorial collective goods' (Foldvary, 1994), and the Homeowners' Association (HOA), a coordination system for collective consumption (Chen and Webster, 2005).
One way to understand PMCs is by using Buchanan's club theory (Buchanan, 1965). From a club theory perspective, privatized neighbourhoods become 'consumer clubs', where welfare services are not provided by the state but are allocated by the market as 'club goods' to homeowners who can afford them (Wu, 2005). Such clubs are efficient in theory because membership by homeownership limits freeriding while membership fees (or property management fees) structure collective consumption (Chen and Webster, 2005). In practice, however, there are numerous reports of homeowner disputes (weiquan--literally right-defending activities), often related to poor performance in the area of property maintenance by PMCs (Tomba, 2005;Yip, 2014).
This brings us to the HOAs. These often emerge from homeowner disputes as a social mechanism for protecting property rights. They have two main functions. First, they act as representatives of homeowners in the negotiation and implementation of property management contracts. In doing so, they counterbalance the power of the PMCs in the governance vacuum left by the retreat of the state (Tomba, 2005;He, 2015). Secondly, for individual homeowners, HOAs are platforms for collective decision making around collective consumption. Freeriding problems are managed by formal covenants or norms circulated through social networks (Shi, 2010;Fu and Lin, 2014). However, in neighbourhoods where social networks are weak and associated levels of trust and sense of community are low, collective action problems can raise transaction costs in the area of property management and also lead to disputes among neighbours (Shi, 2008).
In sum, neighbourhood governance in urban China has in recent decades been characterized by an increasing number of actors and the complex power relations between these diverse actors (He, 2015). Multiple actors now compete, disagree, co-operate and compromise in the everyday governance of neighbourhood life. What we need to know more about is how these power relations play out in each neighbourhood, which differ because each has its own relationship with a spatially uneven state (Gui et al., 2009) and each has its own social character (Shi, 2010) and civic capacity (Chen, 2016). The next section describes our research design, through which we sought to make visible the diversity of neighbourhood governance arrangements and experiences in Nanjing, and to make sense of that diversity.

Research design
Nanjing is one of the largest cities in the East China region, with an administrative area of 6,512 square kilometres and a permanent population of 8.34 million (Nanjing Statistical Bureau, 2018). A significant portion of this population--6.81 million urban residents--are organized in over 3,500 xiaoqus and 937 RCs (ibid.). Nanjing is particularly interesting from the perspective of neighbourhood governance: alongside RCs there are around 600 HOAs (He and Wang, 2015). The city as a whole was an Experimental City for Neighbourhood Governance and Community Reform in 1999 and hosted eight Experimental Zones for Community Governance and Service Innovation between 2012 and 2019.
Within Nanjing, we selected a sample of 32 neighbourhoods, focusing on the six inner-city urban districts of Xuanwu, Qinhuai, Jianye, Gulou, Qixia and Yuhuatai, and on the newly urbanized areas of Jiangning District--which together make up 18.5% of the land area and 80% of the population of Nanjing (see Figure 1). Sampling involved two stages. First, neighbourhoods were stratified into four groups based on the Chinese General Society Survey and other studies (see, for example, Yu and Tang, 2018;Zhang, 2018). In this typology, 'traditional neighbourhoods' refer to lane-or courtyardbased housing, usually built in the inner city by private or public-sector actors before the housing reform of the 1990s. 'Work units' refer to state-owned, self-contained 'microregions' with juxtaposed spaces of workplaces, residential areas and social service areas, usually built during the socialist era before the 1980s and privatized during the 1990s. 'Commodity housing estates' refer to gated and guarded housing with private amenities, built after the 1998 housing reform. 'Affordable housing' refers to welfare housing for relocated residents, migrants and low-income residents, provided by either the public sector or private developers at a subsidized price controlled by the government. The second stage of our research involved random sampling of six to 12 neighbourhoods from each of these four groups.
Our research involved eight months of fieldwork in Nanjing during 2017 and 2018. Four methods of data collection were used: interviews, site visits and observations, participant observation in neighbourhood meetings, and examining relevant policy and other documents. Interviews were our primary method of data collection. With the help of the Nanjing Civil Affairs Bureau, Ying Wang was introduced as an independent researcher to local RCs and community-based organizations. A total of 60 interviews were completed with government officers (four interviews), community directors and party secretaries (22 interviews), social workers (seven interviews), property managers (two interviews), members of HOAs (two interviews) and residents (23 interviews). 1 All interviewees and their neighbourhoods have been anonymized in this article. In the interviews we focused on the rationales for neighbourhood governance (effective service delivery, participation/self-government, social control, and so on), the key actors involved (RCs, HOAs, PMCs), their roles (as decision makers, service providers, consumers) and the relationships between them (for example, collaboration, contracting, integration). These focal points provided a framework for coding the transcripts Nanjing CHINA produced from interview recordings. During the data collection and analysis stage, four modes of neighbourhood governance gradually became apparent, distinguishable by their central relationship.

Four modes of neighbourhood governance
Existing frameworks for viewing neighbourhood governance in China have tended to focus on which actor is dominant--what might be termed 'the "who" question'--and to classify neighbourhoods according to whether they are led by an RC (representing the local state), an HOA (representing society) or a PMC (representing the market). These three organizations constitute a tripartite actor-based classification of neighbourhood governance arrangements, which has been widely adopted in Chinese literature on the subject.
An alternative framework for viewing neighbourhood governance was proposed by Lowndes and Sullivan (2008). While their framework is based on the English case, the authors also draw on political economy theory, which is more generally relevant, to identify four rationales for neighbourhood governance: the civic rationale (emphasizing participation), the political rationale (emphasizing accountability and responsiveness), the economic rationale (emphasizing efficiency and effectiveness) and the social rationale (emphasizing joint local action). These rationales form the guiding principles for organizing neighbourhood practices. Compared with the actor-based framework that dominates the existing Chinese literature, Lowndes and Sullivan's rationale-based framework is focused less on the 'who' question and more on the 'why' question, i.e. on the justification for neighbourhood governance. While their framework draws on the English case and may not be applicable to the Chinese case in all respects, it does suggest that alternative ways of viewing neighbourhood governance in China might be possible.
To develop our own framework--a framework that best captures the diversity of neighbourhood governance on the ground in Nanjing--we drew from the existing actor-based framework, Lowndes and Sullivan's rationale-based framework and the empirical material we collected in Nanjing. Diverse governing practices and hybrid forms of governance were observed that cannot be fully explained through the lens of either the dominant actor or the dominant rationale. On the one hand, if we were to classify the observed governance arrangements according to dominant actors, we would find that most sampled neighbourhoods would fall into the market-led category, leaving only a few led by the state and almost none led by society. This is because, as Wei (2008) has noted, the state-society-market framework fails to distinguish adequately between different neighbourhoods, since it only takes into account variation in the dominant actor and overlooks interactions between dominant actors and 'ordinary' organizations in the ongoing process of neighbourhood governance. On the other hand, if we were to apply the rationale-based classification, a number of the sampled neighbourhoods (particularly affordable and traditional neighbourhoods maintained by local state agencies) would not fit into any of the categories, since they are not dominated by concerns for, say, participation or accountability or efficiency. This is because Lowndes and Sullivan's framework does not accommodate China's strong state power and weak civil society at the grassroots level (Ohmer, 2007;Heberer, 2009) and is thus not 'directly applicable' to urban China (Yip, 2014: 4).
Based on these considerations, we propose a relationship-based classification of neighbourhood governance. This typology not only addresses the 'who' question by specifying the key actors involved in each governance network, and the 'why' question by considering the interests of these actors, but also emphasizes the 'how' question--how neighbourhood governance is achieved--by specifying the key actions in neighbourhood governance and identifying the interrelationships between key actors on which these actions depend. Four key inter-organizational relationships were identified from the sampled neighbourhoods in Nanjing: the relationship between the PMC and the HOA, between the PMC and homeowners, between the HOA and homeowners, and between local government and residents. These relationships and their place in the governance network allowed us to distinguish four ideal types of neighbourhood governance (see Table 1).
Neighbourhoods characterized by a dominant PMC-HOA relationship are managed according to the collective consumption mode. Here, collective decision making and collective consumption are organized by a fully functioning HOA that is able to contract out neighbourhood service provision to a professional PMC. All the sampled neighbourhoods that fitted into this mode are commodity neighbourhoods (see Table 1, column 2). When there is no effective self-governing mechanism for collective decision making, individual homeowners have to negotiate individually with the market institution about neighbourhood service delivery. In such cases, the PMC-homeowner relationship becomes the dominant relationship, leading to the emergence of the second mode: service privatization. This mode of governance arrangement can be found in some commodity neighbourhoods and other neighbourhoods where goods and services have been fully commodified (see Table 1, column 3). The third mode--civic provision--arises when homeowners actively participate in their HOAs (or other forms of self-governing organizations) and take full control of neighbourhood service provision. HOAs then act both as primary decision maker and service provider. In these neighbourhoods (usually privatized work units and traditional neighbourhoods--see Table 1, column 4), neighbourhood governance effectiveness is shaped by the relationship between the homeowners and the HOA, and public services are provided directly by a civic organization governed by the residents themselves. When either the civic organization (for example, the HOA) or the market actor (for example, the PMC) fails to govern effectively, the local government may intervene directly in neighbourhood issues. This is often the case in dilapidated neighbourhoods that suffer from varying degrees of social crisis (see Table 1, column 5). In these neighbourhoods, collective goods are not fully commodified--instead, they are provided, at least partly, as state welfare. The relationship between local state agencies and their constituents is therefore the key relationship that dominates this state-sponsored mode of governance.
In the remainder of this section, we view neighbourhood governance in Nanjing within this framework. We demonstrate that it contributes to our understanding of the diversity and complexity of neighbourhood governance in the city. All sampled neighbourhoods can be classified using this framework, and each classification is found in at least five neighbourhoods and can be distinguished adequately from the others. However, we add two caveats here. First, this typology presents ideal types of neighbourhood governance arrangement that accentuate one or more common points in the synthesis of 'a great many diffuse, discrete, more or less present and occasionally absent concrete individual phenomena' (Weber, [1949(Weber, [ ] 1997. These ideal types differ from 'working models' that exhibit all characteristics of some particular cases. Secondly, this classification is not static, since neighbourhoods are multifaceted entities situated in open networks shaped by internal dynamics and external forces. Governance arrangements in a particular neighbourhood may fit one mode at a particular time, and evolve to fit a different mode later. Therefore, we excluded some sampled neighbourhoods from Table 1 because they were less typical of particular modes of governance at the time of our fieldwork, and included only those that fitted most closely the ideal types in the discussion that follows.

-Collective consumption
The collective consumption mode is the ideal mode of neighbourhood governance according to club theory (see Buchanan, 1965) and theories of networked governance (Rhodes, 1996). It provides a possible solution, at least theoretically, to the optimal provision of public goods on the provider side and the enabling of democratic decision making on the consumer side.
In this mode of governance, a central relationship is founded and maintained between the key service provider (for example, the PMC) and the key collective decision maker (for example, the HOA), as indicated by the grey box in Figure 2. The HOA is usually at the centre of this relationship. According to the Nanjing Regulation of Residential Property Management (Nanjing People's Congress, 2016), a fully functioning HOA should deal with homeowners and the PMC at the same time--resulting in a double-edged governance structure. On the one hand, based on the 'associationmembership' model (Foldvary, 1994), a responsible HOA formulates collective choices over neighbourhood goods provision through norms and conventions circulated through the neighbourhood's formal and informal networks. On the other hand, the HOA is authorized by homeowners to negotiate and establish contractual relationships with the PMC--a professional service provider with capacity to respond to the demands of homeowners.
Many regard the enforcement of the property management contract as the most important part of neighbourhood governance (Fu, 2015). The extent to which such contracts are enforced determines the relationship between PMCs and HOAs, which varies considerably across Nanjing's neighbourhoods. In some neighbourhoods, such as neighbourhoods J and Y, PMCs tend to respond best when external pressure is applied to them by HOAs and state agencies. In the former case, the HOA has the capacity to monitor the performance of the corresponding PMC and exercise its legal right to dismiss the PMC if it does not meet the expectations of most homeowners (for example, for service quality, ownership of public facilities, and management of public spaces). In the latter case, Street Offices or SOs (the lowest level of government in urban China) also supervise the property management enterprises. Rectification notices and blacklists are standard measures the SOs adopt to hold PMCs accountable. These measures, however, are often regarded by local community workers as 'too soft' and 'too loose', since both the HOA and the SO lack enforcement measures to hold the PMC accountable on a daily basis. As one HOA member in neighbourhood T complained, 'the PMC did not listen to us; sometimes, they even cheated on us'.
Where property management contracts are not enforced effectively, conflicts and contentious actions tend to arise. In neighbourhoods T and Y, for instance, the relationships between the HOA and PMC were found to be antagonistic, with the HOAs attempting to dismiss their PMCs owing to their poor performance and their refusal to withdraw from the neighbourhood. HOAs took a variety of measures, including petitions and appeals in the media. In the most extreme case, in neighbourhood T, conflict with the PMC spilled onto the streets, as resident T recalled: 'To cope with security guards from the PMC, we [homeowners] called together a team of "guardians of homeowners" (yezhu huwei dui) equipped with shields, helmets, and vests … We fought against the PMC's security guards with water bottles and fire extinguishers'.
At the other end of the scale, the relationship between the HOA and PMC in some neighbourhoods (for example, neighbourhood SD) can be so close as to appear corrupt (to some of our interviewees in those neighbourhoods), with certain HOA members seeming to speak for the PMC in exchange for beneficial property management or parking fees. Resident SD described their HOA in the following disappointed terms: 'The HOA is nothing as imagined. The activists have their own concerns and interests. They would rather be thought of as "inside men" of the PMC than representatives of us [homeowners]'. The social basis of the HOA-PMC relationship can be further undermined by internal conflicts among homeowners. The relationships between homeowners are governed by a social contract detailing rules to prevent freeriding. However, such contracts tend to be voluntary agreements and focus on common visions and shared values rather than on sanctioning procedures to deal with freeriding. Interviewees reported a lack of incentives for good conduct and a lack of enforcement in cases of wrongdoing. They also reported a lack of familiarity with such contracts and the responsibilities described in them. In this context, much rests on social networks and their potential for generating trust, loyalty and reciprocity (Putnam et al., 1993), characteristics that are often lacking, especially in newly established commodity housing estates. Furthermore, increasing diversity among homeowners in the sampled neighbourhoods made negotiation and enforcement of social contracts even more difficult. For instance, in neighbourhood YY, some homeowners preferred better property management services and were willing to pay more, but others cared more about holding down costs. In the case of each group, activists would seek institutional space for articulating their demands, which escalated differences among homeowners to the organizational level and led to contentious actions and faction politics.
When such conflicts arise, the local state often attempts to intervene through the RC. Designed as a 'meta-governor', the RC can occupy a role as broker in the relationship between the HOA and the PMC, or the HOA and its members, through joint boards or joint conferences. However, our observations in Nanjing indicate that the RC has become a marginal figure in many neighbourhoods, having largely withdrawn from direct service provision and now often lacking in administrative resources. Therefore, the ability of RCs to monitor contracts and arbitrate between other actors tends to be limited. The RC director of neighbourhood Z told us: 'We don't have any enforcement power, nor do we have the legitimacy to intervene in social tensions among the people. We can only console residents, most of whom will become less angry after some time'. Some residents interpreted this limited ability of RCs as prevarication, some terming it 'sloth administration'. A resident in neighbourhood Y commented: 'The RC or the SO? I would not turn to those jacks-in-office for help anymore. They just sit in their office every day and do not care whether the PMC encroached on our rights'.
In this section we described how urban neighbourhoods are governed in the collective consumption mode. In its ideal form, this mode of governance involves active HOAs, responsible PMCs, cooperative homeowners and facilitative RCs all acting in partnership to achieve good neighbourhood governance. In reality, however, collective consumption in many neighbourhoods in Nanjing deviates from this ideal form because one or more actors or relationships in the governance networks are absent or fail to work effectively. When this happens, neighbourhood governance may take on one of three alternative forms: service privatization, civic provision and state-sponsored governance.

-Service privatization
The HOA is a central actor in the collective consumption mode, but a recent survey in Nanjing found that more than half of HOAs are in 'hibernation' (Liang and Xu, 2018). The situation appears to be the same or worse in other cities. Less than 10% of HOAs were found to be active in Shanghai (Wang, 2014). On top of this, many urban neighbourhoods do not have an HOA at all--whether active or inactive. In Nanjing, it is believed that only 47% of residential communities are covered by HOAs (He and Wang, 2015). Nationally, the corresponding figure is thought to be only 26% (China Consumers' Association, 2019). In neighbourhoods without active HOAs, many commodified neighbourhood services and goods are consumed individually rather than collectively, through a second mode of neighbourhood governance: service privatization.
In the Chinese context, this mode privileges effective service delivery and professional management over community engagement and collective decision making. In service privatization, the PMC becomes the key actor and its relationship with homeowners determines the effectiveness of neighbourhood governance. The PMC has been seen as a location for privatization of local government functions concerning public goods provision (Foldvary, 1994;Wu, 2012). It provides services--mainly property maintenance, but also security in poorer neighbourhoods, housekeeping in richer neighbourhoods, and much else in between--and takes responsibility for the effectiveness and efficiency of service provision. These services can be viewed as 'clubbed goods' available exclusively to the homeowners who buy into the neighbourhood. In the absence of an HOA, the service provider and the consumers are linked directly (see Figure 3). The link is not one-to-one, as when the PMC works in partnership with the HOA. Rather, it is one-to-many, with homeowners needing to perform numerous roles from consumer (of services) to negotiator (of contracts) to monitor (of PMC performance).
Such one-to-many relationships have been criticized for their low efficiency and lack of accountability (Chen and Webster, 2005). While a transaction must happen only once between a PMC and an HOA, it must happen many times between a PMC and multiple homeowners. With every additional transaction come additional costs and increasing chances of encountering a hold-up problem--an important category in contract theory (see, for example, Grossman and Hart, 1986;Ellingsen and Johannesson, 2004). The hold-up problem, also known as the commitment problem, describes a contractual relationship in which one party makes a prior commitment that gives the other party bargaining power, thus positioning the former party as vulnerable to subsequent exploitation, which ultimately is associated with generalized inefficiency and underinvestment. Deng (2002) introduced the hold-up problem to the study of urban neighbourhoods in China. In his analysis, the consumption of real estate and the consumption of 'territorial collective goods' (Foldvary, 1994) provided by the PMC are bundled together. In such a situation, both the homeowners and the service provider can find themselves 'held up' by the other party.

Residents' Committee
Market contract One-to-many relations In our Nanjing study, there were only a small number of cases where the PMC worked acceptably well without an HOA. In most sampled neighbourhoods, homeowners reported finding it difficult to govern the performance of PMCs as individuals. Consequently, their needs and desire for services are not being met by the PMC, but the PMC can ignore their complaints, knowing that the homeowners will probably not move away from the neighbourhood for this reason alone. Many residents in the sampled neighbourhoods reported such concerns, describing their PMCs as 'powerful, rude and aggressive' organizations that 'own rather than serve the neighbourhood' (resident H).
Conversely, the PMCs we interviewed complained about being 'held up' by irresponsible homeowners who were described as 'self-serving and lacking in public spirit', especially those 'refusing to pay the PMC fees every month' (interview with PMC manager in neighbourhood Q). A vicious circle therefore developed in some neighbourhoods, observed by us as researchers but also identified by some interviewees. The party secretary of neighbourhood H told us: 'Homeowners are not satisfied with the service the PMC provides and refuse to pay the PMC fees. Consequently, the PMC cannot function effectively owing to financial problems. Some PMC members just washed their hands of the property management matters, which, in turn, aggravated homeowners' dissatisfaction'.
Given all these difficulties, it is perhaps surprising that approximately one third of the sampled neighbourhoods fitted the service privatization mode. Why would this be so? A plausible answer lies in the hostility towards HOAs found among PMCs (some of whose members regarded HOAs as the stereotypical 'mother-in-law'--controlling, judgmental and overbearing) and local state agencies (some of whose members regard HOAs as 'trouble makers', in the words of the RC director of neighbourhood YY). More generally, collective action theory (Olson, 1965) proposes that actors in some neighbourhoods may perceive the costs of the collective action required in the collective consumption mode to outweigh the benefits returned to them as individuals. We found some evidence for this in sampled neighbourhoods exhibiting relatively low levels of neighbourhood social capital, which failed to counterbalance the relatively high levels of individual transaction costs in the establishment and operation of the collective decision making body (for example, the HOA). One such transaction cost is the cost of bargaining in the establishment of an HOA. In neighbourhoods D and Z, for instance, hardly any residents were willing to contribute to community self-governing activities. Instead, inactive residents would rather 'sweep the snow from their own doorsteps' (a community worker in neighbourhood D). This is because, as resident Z told us, they had 'no trust in HOAs' and 'no spare time'. Even in areas where some homeowners did volunteer to lead HOAs, whether they achieved accountable community representation (Chaskin, 2003) or effective entrepreneurial leadership (Purdue, 2001) is another matter. Studies in neighbourhoods H, S and Z all found a level of dissatisfaction regarding selfelected neighbourhood activists. In neighbourhood S, such dissatisfaction centred on conflicts over different plans for community development. A lighting project proposed by some neighbourhood activists was strongly opposed by some residents who viewed the project as 'a trivial issue' and 'not worth the investment' (resident S). Such dissent was even greater among residents in neighbourhood H, where some homeowner activists 'worked for the PMC and became its hardcore supporters' (resident H), while others sought to dismiss the PMC.
High transaction costs are evidence of the difficulty of organizing neighbourhood collective action and provide a plausible explanation as to why the service privatization mode is widely found across the sampled neighbourhoods in Nanjing, given its low level of effectiveness in some respects. In some other neighbourhoods, however, actors appeared to recognize the alternative costs associated with the governance problems of the service privatization mode (including the hold-up problem). They sought institutional solutions to those problems--providing our third and fourth modes of neighbourhood governance.

-Civic provision
One response to the hold-up problem and associated governance problems has been the (re)introduction of HOAs and/or strong local government to neighbourhoods (via RCs and SOs), and their subsequent institutional integration within neighbourhoods (Deng, 2003). This can take the form of horizontal integration between PMCs and HOAs (the civic provision mode), or vertical integration between PMCs and SOs (the statesponsored mode--discussed in the next section).
In the civic provision mode, residents get to participate in HOAs or other forms of self-governing organizations, such as Self-management Associations (SMAs) and Deliberative Councils (DCs), to influence service provision and other aspects of neighbourhood governance. In some cases, where integration between the HOA and PMC is complete, residents may even be involved in providing their own collective goods--a complete integration of decision makers and service providers (see Figure 4).
There are a variety of institutional pathways by which residents might participate in neighbourhood governance, from HOAs (more institutionalized and democratic) to SMAs (less institutionalized) to DCs (under the guidance of local RCs). The powers and responsibilities devolved to these participatory bodies can vary significantly between neighbourhoods. This influences their ability to promote neighbourhood participation and to enable horizontal integration.
One of the most common approaches to horizontal integration is the empowerment of neighbourhood self-governing organizations where professional PMCs are absent or incapable of providing the necessary collective goods. These bottom-up initiatives often take the form of an HOA, such as in neighbourhood G, or an SMA or another neighbourhood civic group, such as in neighbourhoods C, D and X, where the legal requirements for establishing an HOA or recruiting HOA members could not be satisfied. Empowerment of these civic organizations is achieved through specially designed participatory mechanisms, which pass more political power to individuals and enable them to exercise greater 'choice' and 'voice' over local service delivery (Lowndes and Sullivan, 2008). The effectiveness of these participatory mechanisms varied significantly across the sampled neighbourhoods. The involvement of local residents was found to be important and relatively higher in privatized work units compared with other sampled neighbourhoods. One resident in neighbourhood X, where an SMA was established to replace the poorly-performing PMC, provided a convincing explanation as to why this might be the case: 'We had a fundraising campaign for the SMA. Each household was asked to pay 15 CNY a month … most residents here used to work for the same work unit. They could hardly resist doing such a small favour for their former colleagues when approached for the fees'.

Horizontal integration
Besides the effectiveness of horizontal integration depending on selforganization and participation, it also to a significant extent hinges on the power and responsibilities devolved to neighbourhood civic groups. According to our interviewees, in some neighbourhoods (for example, neighbourhoods N and BS), SMAs were no more than 'window-dressing' organizations that 'cannot fully satisfy our daily needs' (Resident WT). In other neighbourhoods, such as neighbourhoods WT and X, civic groups were granted decision-making powers for some neighbourhood issues. In yet other neighbourhoods, such as neighbourhoods C and D, there was a further step towards self-governance: not only decision-making powers but also the rights for enforcing those decisions were transferred to the HOAs/SMAs. The functions of HOAs and PMCs were thus completely integrated into these empowered civic groups, which can have wide-ranging responsibilities, from collecting fees and hiring staff (security guards, cleaners, and so on) to delivering services (for example, property maintenance). Such horizontal integration was often well-received by local residents. A community worker in neighbourhood D commented, 'Not just our residents, but those living nearby spoke highly of our approach, saying that they would pay less [in property management fees] but have more say in neighbourhood issues. Inspired by our success, two adjacent neighbourhoods recently dismissed their PMCs and set up SMAs'.
Another means by which horizontal integration can be achieved is through a neighbourhood council system established by local government. In such a system, the DC provides reliable institutional spaces for conflicting parties to negotiate a solution for neighbourhood issues. In some neighbourhoods, a further step was taken, called the 'union of deliberation and execution' (yizhi heyi). In neighbourhood A, for example, we observed how those who proposed matters during DC meetings were made directly responsible for implementing the decisions made by the DC. In this way, some responsibilities that once belonged to the PMC (for example, property management) were transferred to the DC (and further to the citizens), and empowered DCs could thus be regarded as a form of horizontal integration. The local RC members we interviewed all spoke highly of the multiple roles empowered DCs played in civic provision, since they 'significantly relieve the fiscal and administrative pressure on the RC' (RC member in neighbourhood W) and ultimately serve as an alternative to public spending.
A final point here, demonstrated by the Nanjing case, is that civic provision of neighbourhood goods and services can hardly succeed without the assistance of the local state. This is true not only for RC-led DCs: even HOA-and SMA-led civic groups seek self-governance in ways that are in accordance with, and sustained by, the local state. Let us consider the recruitment of volunteers as an example. A community worker in neighbourhood D, preparing for the SMA election, complained that 'nobody wants to serve the neighbourhood … We have approached many residents, but no one wants to do some real work … If no volunteers can be found by the end of next month, the last thing we can do is to turn to the RC. They are good at "ideological works" (sixiang gongzuo) and may persuade existing members to serve for another term of office'. Where the RC provides support to HOAs and other self-governing organizations in this way, but HOAs still exist to make decisions and provide services, we have the civic provision mode of neighbourhood governance. However, in some neighbourhoods state intervention extends beyond support for HOAs, and HOAs--if they exist at all--are bypassed, as state agencies deal directly with PMCs and absorb responsibilities for decision making and service provision. In these neighbourhoods, we have state-sponsored neighbourhood governance. -

State-sponsored governance
Whereas the civic provision mode seeks to solve the hold-up problem by horizontal integration of PMCs and HOAs (or other participatory bodies), the statesponsored mode aims to solve this problem through the vertical integration of PMCs and SOs--which function as local centres of administration (see Figure 5).
In this mode of neighbourhood governance, service delivery is no longer contracted out to commercial organizations. This is because privatization does not always lead to effective provision of collective goods, especially in poorer neighbourhoods where PMCs are held up by residents who are unable to afford service fees. In this regard, our study echoes Wu's (2018) study in Nanjing. According to our findings, such neighbourhoods included traditional neighbourhoods in dilapidated inner-city areas (for example, neighbourhoods DS and GT), degraded work units (for example, neighbourhood SY) and affordable and resettlement neighbourhoods (for example, neighbourhoods JM and N). When privatization fails, the state often intervenes, resulting in state-sponsored neighbourhood governance. By incorporating service delivery into local administration, local state agencies attempt to ensure that essential services (for example, unblocking sewage pipes or fixing broken windows) are provided at affordable rates to residents so that, as one interviewed officer from Street Office M put it, 'none would be left behind'.
Local SOs and RCs act as leading organizations in state-sponsored neighbourhood governance. By establishing PMCs or subsidizing commercial PMCs, SOs shape a welfare-oriented property management system to distribute essential services and reinforce social security in disadvantaged areas. Such a governance arrangement has its advantages, which include reduced transaction costs for PMCs and delivery of basic

Vertical integration
Market contract with subsidies FIGURE 5 State-sponsored governance (source: produced by the authors based on their research) services where they are needed (Tomba, 2014;Wu, 2018). However, it can also produce new hold-up problems for actors in the network. Interviewees reported concerns among residents and community leaders that service provision had become too dependent on local state agencies. They spoke of 'waiting, depending, [and] wanting ' (deng, kao, yao). Such dependency can significantly increase the administrative burden on RCs, as the RC director of neighbourhood GT explains: 'They [the residents] are used to government rescues and lack the common sense of "paying for service". Most of them refuse to pay PMC fees, as they feel all services should be provided by the party … If they lack something, they just turn to the RC for help'.
The efficacy of neighbourhood governance is commonly evaluated not only by its service provision but also against the RC's capacity to guide community participation (Tomba, 2014). Our research found this capacity to be constrained in multiple ways in many of the sampled neighbourhoods. Participatory platforms provided by RCs cannot really be interpreted as initiatives of self-governance or reflections of democracy, since they are guided, monitored and audited by SOs and higher levels of government. What RC members do, stated a community worker in neighbourhood GT, 'needs to satisfy the leaders [from the SO] first'. The RC route is thus not a realistic route by which residents can challenge SOs and express their own needs regarding service delivery. Furthermore, institutional spaces created by state agencies, as the Nanjing case shows, do not always transform into organizational sources for governance, unlike in the civic provision mode. This is because participation opportunities are limited to 'more capable and more qualified people' (CPC, 2010), who are able to convey 'organizational intentions' (zuzhi yitu). After careful screening, only political and social elites sharing intimate relationships with the state tend to be included in the RC governance system. Interviewees in the sampled neighbourhoods regarded most RC-led participation as tokenism, since it involved limited decision making, and such 'democratic decoration' offered limited opportunities (resident JM).
Attention to the state-sponsored mode of neighbourhood governance, then, helps to expand our understanding of 're-statization' in urban China (Sigley, 2006;Heberer, 2009). The vertical integration of neighbourhood services and grassroots administration transforms neighbourhood institutions into combinations of 'authoritarian government'--a government that provides controlled and constrained opportunities for collective decision making--and a 'local welfare state', where basic levels of neighbourhood services are guaranteed to vulnerable social groups.

Conclusion
In a context of global moves towards decentralization and neighbourhood governance, in this article we have focused on the rise of neighbourhood governance in China and how it plays out differently in different neighbourhoods. We presented the case of Nanjing because it is both an ordinary city (Robinson, 2006) deserving more attention in urban studies, and a prototypical city (Brenner, 2003) deserving attention for its role as an experimental zone for neighbourhood governance within the context of urban China. We designed our study as a comparison of neighbourhoods across Nanjing that included interviews and observations in 32 neighbourhoods.
The mid-level view from these 32 neighbourhoods enabled us to identify four modes of neighbourhood governance in Nanjing: collective consumption, service privatization, civic provision and state-sponsored governance. We generated this framework on the basis of multiple sources. First, we offered a critique of existing frameworks of neighbourhood governance in China (especially the state-marketsociety framework), which recognize the specificity of the Chinese case but focus narrowly on the question of who dominates the action in neighbourhood governance (the answer to which does not address related issues, including how effective neighbourhood governance is achieved). Secondly, we offered a critique of existing frameworks of neighbourhood governance beyond China (especially the Lowndes and Sullivan framework), which have a broader focus that includes the multiple rationales for different forms of neighbourhood governance, but are themselves context-specific and do not translate easily to the Chinese case. Thirdly, our framework was derived from the empirical material we collected in Nanjing, and based on our finding that certain relationships between actors were crucial in shaping effective neighbourhood governance in the different neighbourhoods. Our resultant framework, we argue, clearly shows the most important features of neighbourhood governance in Nanjing--more so than other existing frameworks--while at the same time contributing to the general literature on neighbourhood governance.
One contribution that is relevant to the broader literature is that relationships between actors are important units of analysis when considering how effective governance is achieved in neighbourhoods. In the case of Nanjing, it was the relationship between the HOA and the PMC, the PMC and the homeowners, the HOA and the homeowners, or the SO and the PMC that was important. In other cities and countries, it might be the relationship between citizens--perhaps positioned as homeowners, but perhaps positioned in other ways--and other state, civil-society and private-sector organizations. Another contribution, which is of particular relevance to debates on 'global policy' and neoliberalism, is that we found neighbourhood governance to be diverse and complex even at the scale of one city (Nanjing), let alone that of the nationstate or the globe. This implies that scholars should be cautious when attempting to make generalizations about how recent moves towards neighbourhood governance around the world play out in practice.
In terms of current debates on neoliberalism, our third contribution is that diversity and complexity in neighbourhood governance reflects in part the varying role of the state. In Nanjing, the state--in the form of SOs and/or RCs--can largely retreat from service provision, support service provision by other organizations, or deliver services directly to homeowners. Moreover, the role the state plays in different neighbourhoods is largely shaped by developments over time. The starting position in many neighbourhoods is retreat (the collective consumption mode). The service privatization mode arises when HOAs are absent or become dormant over time, making collective consumption less viable. Civic provision and state sponsorship arise when PMCs and homeowners hold each other up--a problem that then has to be solved by horizontal or vertical integration. In some neighbourhoods, this process of evolution eventually leads to state agencies stepping back in as neighbourhood governance facilitators and even as direct service providers.
How does this discussion relate to debates on neoliberalism? On the one hand, we argue that neighbourhood governance plays out on the ground in complex and diverse ways, so that generalizations need to be made with caution. On the other hand, the process of evolution--from state retreat to state return--is reminiscent of what Peck and Tickell (1994: 317; term 'process-based analyses of neoliberalization'. Their analysis in this classic article focused on how neoliberalization proceeds through different moments: destructive and creative moments; moments at which old forms of regulation are 'rolled-back' and new forms are 'rolled-out'; the moment when 'jungle law breaks out' (ibid.), followed by the moment when new market rules are imposed. This is not quite what happens in neighbourhood governance in Nanjing, but there is a process whereby the state retreats and then has to respond to the consequences of that retreat by adopting new and more active roles. The literature on neoliberalism, therefore, provides some categories and storylines that help us conceptualize neighbourhood governance in China, even if the Chinese context differs in many ways from that of North America and Western Europe, where these categories and storylines were developed. Also, such categories are not limited to those of Peck and Tickell. Hay (2007) offers 'depoliticization' and 'repoliticization', the former referring to delegation or privatization of responsibilities by the state, the latter to reactions against those moves (whether in government or in the public sphere). Becker et al. (2015) offer the term 'remunicipalization', which refers to privatized companies being repurchased by municipalities. None of these concepts quite fit our particular case; all were generated within other contexts--the UK for Hay and Germany for Becker et al. However, the Nanjing case aligns with this broader literature in showing clearly that, while neighbourhood governance evolves in response to global and national policy, it also develops as a consequence of such policy in particular neighbourhoods. Where those consequences involve the unintended breakdown of effective service provision, the state is pushed towards a more central and active role.
We also raise at least two further questions for future research. First, there is the explanatory question of why certain modes of neighbourhood governance are found in certain neighbourhoods--or, put differently, why neighbourhood governance evolves in different ways in different neighbourhoods. What path dependencies are at play? One factor may be the type of homeownership in a neighbourhood. Our research showed that commodity neighbourhoods tend to have active and strong PMCs that lend themselves to the collective consumption or service privatization modes of governance. We also discovered that some modes of governance are found in multiple neighbourhood types. Notably, state-sponsored governance is found in some traditional neighbourhoods, some privatized work units and some affordable housing neighbourhoods. What unites these neighbourhoods is not a particular type of homeownership, but a lack of resources among residents--meaning that they struggle to form effective HOAs or negotiate successfully with PMCs as individuals.
It seems that other factors therefore need to be considered. One of these might be neighbourhood wealth or poverty. We stated earlier that residents are more likely to hold up PMCs if they are poor and struggling to afford service fees. State sponsorship is thus a more appropriate mode in poorer neighbourhoods (especially where SOs sponsor PMCs or subsidize commercial service providers). Another factor might be neighbourhood social networks and social capital. One suggestion from this study of Nanjing is that HOAs function best in tackling freeriding--and thus the collective consumption mode functions best where contracts are supplemented by norms circulated through well-developed social networks--especially when they include responsible neighbourhood activists.
Finally, there is the evaluative question as to the strengths and weaknesses of certain modes of neighbourhood governance. We suggest that the collective consumption mode is ideal from the perspective of club theory. It involves collaboration between actors to maximize overall interests and minimize transaction costs. Participation is achieved via the HOA, and service delivery via the PMC. The service privatization mode, by contrast, can exist in neighbourhoods lacking HOAs (a strength), but makes heavy demands on homeowners who must govern the PMCs as individuals, and requires a high number of transactions (between PMCs and individual homeowners), making it costly and inefficient. Transaction costs may be reduced by integration in cases involving civic provision and state sponsorship (a strength), but in the latter case, a corresponding weakness of vertical integration is constrained participation, which must happen via RCs. All these questions and suggestions deserve further empirical research.