Bridging State and Nonprofit: Differentiated Embeddedness of Chinese Political Elites in Charitable Foundations

How are political elites embedded in both civil and political domains? I studied 246 Chinese political elites who also served on nonprofit foundations’ boards from 2011 to 2015 and conceptualized a framework to understand the state-nonprofit relationship before 2015. The political elites can build horizontal connections with foundations, helping the party-state reach nonprofit actors. They can also build vertical connections with political groups, helping foundations access political resources. The dual-role elites’ connectedness with foundations and in the polity varies by their primary positions in the political system. Effective communication between the state and nonprofit actors requires elites to have strong connections in both the civil and political domains, but the situation in China is far from ideal—those embedded in foundations are disconnected from the polity, and those connected in the polity are marginalized in foundations.


Introduction
The relationship between the party-state and nonprofits in China has been of interest to scholars, practitioners, and politicians for a few decades because of the expectation and fear that a burgeoning nonprofit sector may bring multiparty democracy to the authoritarian regime. A rich body of scholarship has been crafted from different macro perspectives; however, we are still in dire need of a framework to understand how the party-state and nonprofit actors interplay at the meso and micro levels.
Scholars have introduced numerous theoretical lenses to better understand the state-nonprofit relationship. The initial attempt started in the early 1990s, when scholars and policymakers primarily used a neo-Tocquevillian perspective and theorized the relationship as conflicting (e.g., [9,44]). The second wave started in the mid-2000s, when scholars and policymakers tried to theorize the relationship as being contingent and applied a lens of corporatism. Studies of this stream, for example, those on "graduated control" [31], "contingent symbiosis" [65], and "consultative authoritarianism" [67], framed nonprofit actors as the service arms of the state, leaving room for these organizations to grow. However, their survival is still contingent upon their focuses on nonpolitically sensitive areas. The third wave started in the late 2010s, when the relationship was theorized as being networked. Scholarship of this wave began to emphasize the active roles of nonprofit actors and the mutual embeddedness between the civil and political domains (e.g., [41,68]). A growing number of scholars have started to emphasize the blurring boundary between the party-state and nonprofit sector (e.g., [16,21,62,64,72,79]).
This study follows this research trend and advances our understanding of the mutual embeddedness between political and civil actors in China. By examining 246 political elites who also served on nonprofit foundations' boards from 2011 to 2015, I analyzed their profiles and networks and found that the dual-role elites who are popular in the nonprofit sector are marginalized in the party-state, and vice versa. In order to make the state-in-society strategy work [46], political elites need to have strong connections in both the civil and political domains. But the situation in China is far from ideal-those embedded in foundations are disconnected from the polity, and those connected in the polity are marginalized in foundations. We need to be cautious about generalizing the findings because this study only covers foundations and there are other channels between the party-state and nonprofit sector. However, as one of the first studies theorizing a framework at the meso and micro levels, this paper provides an essential baseline and can serve as a stimulus for future research.

Research Framework: Differentiated Embeddedness in a Multilayer-Network
The dual-role elites' double-embeddedness can be conceptualized through a multilayer-network perspective. From this perspective, an individual's purpose is a function of that person's embeddedness in a specific domain, and one individual can have multiple purposes that are defined by the different domains in which they are embedded [51,1469]. As Fig. 1 illustrates, the same political elites can be embedded in both the civil and political spaces.
The multilayer-network figure also presents two important notions in the literature on the state-nonprofit relationship in China: horizontal and vertical connections (e.g., [39,58]). A horizontal connection primarily considers how nonprofits are mutually connected (e.g., [22]), and a vertical connection measures these actors' access to political resources. The political elites who are embedded in both civil and political arenas are crucial-they can build horizontal connections with nonprofit groups, and they can also build vertical connections with the party-state. In a nutshell, they are unique players bridging civil and political spaces.
Combining the multilayer-network perspective and the two notions of connection, Table 1 illustrates the theoretical framework. The dual-role elites can be grouped into four categories according to their embeddedness. 1) Those who have strong connections in nonprofits are excellent state agents when it comes to reaching nonprofit actors. 2) Those who have strong connections in the polity can serve as access points to the party-state. 3) Those who are less connected in nonprofits are weak state agents. And 4) those who are less connected in the polity are ineffective access points to political resources. The sections below briefly review these elites' roles in the civil and political domains.

Political Elites in the Nonprofit Sector: Access Points to Political Resources
Chinese political elites have been active players in the nonprofit sector since the founding of the socialist state. In the 1950s, numerous mass organizations were The image is only for illustration and does not reflect real observations. Non-political actors in the nonprofit network are hidden for visual clarity. Nodes in the same color indicate political elites from the same political organizations. For example, red nodes are all political elites from the Chinese Communist Party. Nodes in the same positions but in different networks represent the same individuals. Node size represents the number of direct connections of a node in that given network created by the state, and they are still quasi-government units today. The party-state also created the first nonprofit foundations in the early 1980s, and most of their staff in the following two decades were government employees. Even in the late 1990s, the majority of Chinese nonprofits were staffed by government officials serving as employees and political elites serving as board members [24,[232][233]. The regulations on how to register a nonprofit in China all employed an important strategy for the state to monitor the nonprofit sector-the "dual registration" system. 1 According to these regulations, a nonprofit was required to 1) have a government department serving as its "supervisory sponsor" and 2) register with the departments of civil affairs at different levels in the party-state. As a result, nonprofits in China differ from their counterparts in Western democracies: they have limited autonomy and are closely monitored or even directly established by the state [43]. These nonprofits, particularly those engaging in politically sensitive activities, must prudently manage their relationships with the state to achieve a "contingent symbiosis" [65].
Despite these operational constraints, the number of nonprofits in China has increased substantially since the 1989 Tiananmen Square protests, and the interaction between these organizations and the party-state is bidirectional. Nonprofits in China can have considerable policy influence through their connection to the state [56,210]; [45], and the state also relies on them as an indirect tool of social control [67]. In short, the party-state actors and nonprofits are mutually embedded.
For example, [68] found that nonprofits exploited their governmental supervisors as access points to the policymaking process, which they would not have been able to access otherwise, and formed policy networks with these policymakers that resemble those in Western democracies. Another study on Chinese nonprofit foundations found that, as a dominant power in the Chinese nonprofit sector, these organizations were able to form a decentralized organizational network through board interlocking relationships. Despite the widespread appearance of government officials on foundations' boards, Chinese nonprofits were able to cluster into several autonomous elite groups and form a multipolar structure that resembles what is found in liberal societies [41]. Moreover, [39] discovered that grassroots nonprofit actors formed "horizontal connections" with each other and that they also forged "vertical connections" with the state through retired government officials.
The literature has singled out the fact that the party-state relies on political elites to monitor the activities of nonprofits. But more importantly, nonprofits can also exploit these elites to reach the party-state. However, scholars have rarely investigated how exactly these dual-role politicians function.

Unifying politics and administration
Democratic regimes are characterized by a separation of power [37,38,55], but China's post-Mao era is experiencing the opposite. In the 1980s, political elites in China had already realized that the party-government relationship was the major challenge to reforming the political structure, but later legislation made no progress in separating the cadre and civil servants because such separation would fundamentally challenge the Chinese Communist Party's (CCP's) hegemony [66]. The enactment of the Law on Public Servants in 2005 formally institutionalized the combination of politics and administration by recategorizing public servant positions as "leading" or "non-leading" [10,389].
As Table S4 in the Appendix presents, Chinese political leaders in "leading positions" are ranked in ten categories, from the lowest rank of "Section/ Township Deputy" (xiangkeji fuzhi) to the highest rank of "State Leader" (guojiaji zhengzhi). These ranks also correspond to geographic levels, with the township level being the lowest and the state or central level being the highest [53, Article 16]; [10,393]. Leadership changes follow the "one-level-down" and "party control" principles, in that the party committees at a lower level prescreen the proposals to be submitted to the party organizations at a higher level for decision-making purposes [10,392]. For example, a party committee at the bureau/department level will review a list of candidates before submitting their proposal to the party committee at the province/ministry level for decision-making.
The number of cadres at different levels comprise a pyramid structure. In 1998, there were approximately 545,000 leading cadres who were ranked at the county level and above (i.e., xianchuji and above). Among them, over 500,000 were at the county level or below, 34,000 were at the bureau or municipal level, 2,500 were at the ministerial or provincial level, and less than 1,000 were at the national level [6,392]. Although leaders at the provincial level are not large in number, they are important brokers between local and national political elites [56,144].

A Fractured Political System
Although CCP members hold more than 96% of the leading positions [6,391], they serve in different political and bureaucratic institutions. Thus, these political elites may represent the interests of different departments that do not necessarily hold views consistent with those of the central leaders.
Such misalignment is real and emergent-the political elites are far from being a unified group. For example, the leaders of state-owned enterprises form a powerful group of corporate elites who are highly independent from government agencies [5]. Meanwhile, the People's Consultative Council, which comprises non-CCP elites from diverse social and economic professions, plays an important role in political participation and consultation [23]; [78,689]. Furthermore, as the CCP's primary youth organization, the Chinese Youth League (CYL) prepares youths to be the "assistants and reserve army" for the party. However, having this separate group may also be supporting the rise of the "CYL Clique" (tuanpai) within the party [32,[159][160][161][162]. Finally, intellectuals have opened up a robust public sphere for political debate [18,13]. Most, if not all, of these individuals are CCP members, but they serve in various institutions and are driven by different interests.
To handle the potential threat of a fractured polity, the CCP's central leaders began to bypass political and bureaucratic institutions and strengthen the party's direct governance at all levels during Xi Jinping's presidency [20]. In November 2018, the Central Committee of the CCP released the Regulations on the Operation of the Party Branches, which required all social, economic, and military "units" to establish CCP branches, and a unit was defined as having three or more CCP members, regardless of its registration status (e.g., as a formally incorporated cooperative company, private company, publicly listed company, or even a temporary working team; [76]). Although such a requirement has long existed since the 1990s, it became unprecedentedly compelling during Xi's presidency [48,2]. Besides establishing the CCP branches, the Xi administration has invested more effort in advancing the party's direct control. In 2013, only two regulations of this sort were made. The number was tripled in 2015. Then in 2018 alone, thirteen regulations were announced, followed by twenty-two in 2019. 2

Research Questions
Based on the theoretical framing and the analysis of the elites' roles, I am motivated to explore how these dual-role elites are embedded in the civil and political domains. Specifically, I develop this study from three perspectives.

1.
What are the characteristics of the dual-role elites? Answering this question involves examining the elites' demographics, their cadre ranks and positions, and how these characteristics change over time. 2. What are the profiles of the social and institutional networks? This switches our analysis from studying the individuals to describing the networks in which these individuals are embedded. Answering this question discloses the structures of these networks. 3. What are the patterns of embeddedness? Finally, I analyze the patterns of interactions between these elites. The analysis helps us group the dual-role elites as Table 1 theorizes, creating a framework and stimulus for understanding the differentiated embeddedness of political elites in the nonprofit sector.

Data Sources
I used two datasets: the Research Infrastructure of Chinese Foundations (RICF; [42]) for charitable foundations and the Chinese Political Elite Database (CPED; [29]) for political elites. The online appendixes have more details on data preprocessing and validation.

Research Infrastructure of Chinese Foundations
The RICF provides comprehensive data on Chinese foundations, including their institutional characteristics, financial status, and personnel from 2013 to 2016. The data descriptor of the RICF contains more technical details on dataset construction and validation [42]. Following this descriptor, I constructed a dataset on foundations from 2010 to 2016. Although each year is different, roughly 30% of all the foundations are eligible to raise funds publicly, 3 the board size is 11.72 (SD = 6.44) people on average, and there are about 140 political elites embedded in the foundation network each year. Tables S1 and S2 in the Appendix have more details about the foundations.

Chinese Political Elite Database
The CPED is a biographical database that contains the demographic and career information of over 4,000 Chinese political leaders who belong to one of the following groups: 1) "all city secretaries, mayors, and members of provincial standing committee The limitations of the data sources may bias the analysis, and selection bias is the most concerning issue. First, the foundations included are only a part of China's nonprofit sector. Therefore, using foundations as the primary units of analysis may ignore the activities of other nonprofits. However, empirical analysis shows that foundations are the most developed form of and the dominant power in the Chinese nonprofit sector [33]; [41,293], and they also contribute to our understanding of the state-civil society relationship in authoritarian countries [34,5]. Second, a substantial number of foundations are created by the government [70,308]. As a result, the analysis may merely reflect the assignments of the party-state. However, these government-created foundations are not isolated agents; the are instead structurally connected to other types of foundations [41,294] and lead the development of the foundation landscape [70,297]. Third, the CPED may miss cadres at the grassroots level, but the technical details of its construction and validation suggest that the dataset is better than acceptable [29]. Moreover, these grassroots cadres may have limited influence on the foundation landscape in comparison to high-level political elites. In general, we can expect that the biases of the analysis due to data source limitations are unavoidable but better than acceptable.

Dependent Variables: Nonprofit and Political Embeddedness
I used board interlocking relationships between nonprofit foundations to operationalize the dependent variables. In theory, the board interlocking relationship plays a crucial role in organizational governance and strategy. Although controversial in practice because organizational behaviors largely exist outside of boardrooms, the board interlocking relationship has been empirically tested as a valid proxy to study  the organizational relationships of companies and nonprofits (e.g., [13,15,17,40,47,73]).
I constructed two dependent variables to measure a political elite's horizontal and vertical connections: nonprofit embeddedness is operationalized by counting an elite's connections to nongovernmental actors, and political embeddedness is defined as an elite's connections to the party-state actors. Figure 2 illustrates how the two variables are constructed. I first create a bipartite affiliational network using board membership in which there are two types of nodes (i.e., foundation and board member; Subgraph 1 in Fig. 2). Then the bipartite network is converted to a unipartite board interlocking network in which nodes represent board members (Subgraph 2 in Fig. 2). Take node G in the figure for example: its nonprofit embeddedness is 2 (connected to E and J), and its political embeddedness is 3 (connected to H, F, and I). Meanwhile, individual D's nonprofit embeddedness is 4 and political embeddedness is 0.

Independent and Control Variables
The primary independent variable is political position. It is a categorical variable and coded according to an individual's primary position in a given year using the itemized curriculum vitae from the CPED. The variable has ten categories: 1) Chinese Communist Party (CCP; only when individuals directly serve as party officials, e.g., as the secretary of a party committee), 2) Chinese Youth League (CYL), 3) Judiciary, 4) Mass Organization (MO; e.g., the All-China Women's Federation), 5) People's Congress (PC), 6) People's Consultative Council (PCC), 7) State Council, 8) Central state-owned enterprise (CSOE), 9) Local state-owned enterprise (LSOE), and 10) School (e.g., university principals).
There are two caveats in coding these categories. First, the CYL is also a mass organization, but it is coded separately because of its direct affiliational relationship with the CCP (i.e., preparing youths to be the CCP's "assistants and reserve army") and because of the evidence of the CYL Clique [32,162]. Second, if individuals hold multiple positions simultaneously, I used the position that 1) has a longer track record or 2) has more important roles in policymaking [4, 28,57,80].
Control variables in the regression analysis include demographics (i.e., age, gender, education, and cadre rank) to control the influence of individual characteristics. A foundation's average board size is included to control organizational influence because individuals on larger boards can have more connections.

Estimation Strategy
To understand how political actors in different positions are embedded, the two dependent variables (i.e., nonprofit embeddedness and political embeddedness) are regressed on political position and controls. An ideal analysis should take advantage of the dataset's panel structure, but the analysis of data quality ( Fig. S1 and Table S2) suggests the within-category variations of political position are small. Therefore, a fixed-effect regression is not appropriate if we use political position as the primary independent variable. Thus, the strategy is to use a pooled ordinary least squares regression to generate initial findings and then to conduct qualitative member checking to explain and confirm the findings-a quasi-"explanatory sequential design" in mixed methods [12,66].
Two limitations may bias the quantitative estimations. First is the presidential preferences. Because President Hu Jintao was the head of the CYL and his presidency has been considered to have ushered in the rise of the CYL clique [32], it is possible that cadres with previous CYL experience were favored during Hu's presidency. Therefore, I considered cadres' previous CYL experience in the robustness check.
Second, using only foundations' board interlocking relationships cannot capture "social ties beyond the boardroom" [2,37]. The ties formed through other channels, for example, political elites' work relationships and friendships made through private social clubs, can reduce the social distance between elites. However, the dependent variables and regression models cannot consider these interactions. Nevertheless, it is difficult to obtain affiliational data of this sort even in democratic countries where political and social operations are much more transparent. Because individuals who have connections with different social groups are structural holes who bridge heterogeneous information and are more likely to form weak social ties outside of boardrooms [8,19], I used the Herfindahl index of the types of political connections (i.e., embeddedness diversity) to approximate these social ties. 4 For the qualitative member checking, I conducted a series of interviews and focus groups following "explanatory sequential design" in mixed methods [12,66]. Using the dataset in this study, I identified a list of highly connected political elites in both the civil and political networks. Then I contacted a scholarly network in China and consulted a few scholars associated with several of the most prestigious Chinese universities. Because interviewing political elites in this study would be extremely difficult, if not impossible, these scholars helped me by 1) interviewing several key informants who were colleagues of these political figures and 2) arranging presentations at universities, foundations, and think tanks that were closely associated with these elites. The interviews and focus groups used the theoretical framework (Table 1) as an instrument and primarily focused on three questions: 1) the interviewees' opinions and understanding of the framework, 2) how they would group the political elites, and 3) the reasons for their grouping. Because the informants were selected through snowball sampling, 4 The Herfindahl index is calculated as ∑ n i=1 s 2 i , where s i is the share of all institutional connection i, and n is the total types of institutional connections. For example, individual a has four total connections, in which n = 3 and i 1 = 2, i 2 = 1, and i 3 = 1, the embeddedness diversity of individual a is (2.0)  Bridging State and Nonprofit: Differentiated Embeddedness… the qualitative fieldwork and analysis focused on process tracing to minimize the influence of selection bias [3,61].  Table 3 presents education and cadre ranking profiles. With respect to educational background, the political elites are a highly educated group: 49% of all leaders hold a master's degree or above, and 37% of the School leaders and 22% of the CCP leaders hold a doctoral degree. Although high-ranking officials may have ways of obtaining advanced degrees without substantial work, these political elites are still a well-educated group.

Profile of the Dual-Role Elites
With respect to the positions' ranks, 33% are political leaders at the local government level (i.e., Bureau/Department leaders who govern provincial or municipal areas). 5 Over half of the political elites are at the Province/Ministry levels, where they play important roles in bridging local and national political leaders. This spindle-shaped structure differs significantly from the pyramid structure of the cadre system as a whole [1, 693; 6, 391]. Table 4 presents the positional categories and ranks of the political elites by time. From 2011 to 2015, the CCP, PC, and State Council were the top three political groups that integrated nonprofits. Even in 2013, when Xi ascended to power, the percentages of positional categories remained largely stable, 6 and no substantial fluctuation in ranks occurred in the time period observed.
In general, the demographics suggest that the political elites embedded in Chinese foundations are a male-dominated and well-educated group who are focused on bridging local and national interests. Despite the high turnover rate of individuals (Fig. S1), the positional structure of these elites is well institutionalized and stable.  Figure 3 shows the political elites in nonprofit and political networks from 2011 to 2015. Contrary to many beliefs about authoritarianism that assume a dominant ruling party, the actors in the networks are diverse. This finding is consistent with prior empirical studies, which have found pluralistic and multipolar structures in policy and nonprofit networks [41,68]. The connections between different formal institutions (the bold edges in Fig. 3) are widespread (> 70% for all years; Fig. 4). Put together, the political elites embedded in the nonprofit sector form a well-educated and pluralistic group that connects local and national interests. The networks presented in Fig. 3 also suggest a crucial structural change: the formation and dissolution of political cliques along with presidential changes. I used transitivity, an important structural feature in social networks, to measure such structural change [25,26]. Transitivity presumes that if a chooses b as an ally, and b chooses c as an ally, then a will also choose c as an ally. As Fig. 3 shows, there were dense cliques (or "triangles" in social network analysis terms) both in the nonprofit and political networks in 2011 (during Hu's presidency, transitivity = 0.23). These cliques dissolved in 2013 (the year of presidential change, transitivity = 0.13) and then began to form again in 2015 (during Xi's presidency, transitivity = 0.14). 7 By analyzing the types of connections between individuals (e.g., between the CCP and State Council, the CCP and the PC, and the CCP and the PCC; Fig. 4), we can better understand how formal political institutions orchestrate policy in the nonprofit sector. As Fig. 4 presents, the links between different political organizations consistently make up more than 70% of all connections in all years observed. In contrast to many beliefs about authoritarian regimes, where the ruling party dominates the polity, the Chinese political system is quite pluralistic and collaborative.

Social and Institutional Networks Created Through Foundations
The connections between and within the CCP and State Council are consistently the highest, indicating the close relationship between the two political groups and their important roles in influencing the nonprofit sector. The CCP-State Council coalition separates two institutional spaces: a legislative space with the PC and a political participation space with the PCC. The two institutional spaces would not be able to reach each other without routing through the CCP-State Council coalition in all years observed. Such a coalition also resembles the politics-administration tension studied in democracies [37,38,55].
A shift from the PC to the PCC as the main entity tied to the CCP-State Council coalition occurred after the leadership change from Hu to Xi. In 2011 and 2012 (during Hu's presidency), it was the PC that connected to the CCP-State Council coalition; between 2013 and 2015 (during Xi's presidency), the PCC assumed the PC's role. The shift from the PC to the PCC indicates both an institutional improvement and a retreat. In theory, the PCC is an important institution for non-CCP elites' political participation. Policy bargaining and participation within the PCC will not threaten the CCP's hegemony because the PCC has no legislative authority [23,78]. The close relationship between the PCC and the ruling party is propitious for building an inclusive regime, but the detachment of the PC from the CCP signals a retreat because the PCC cannot reach the final legislative body, even though it can reach the ruling party.
There was also a shift from the State Council to the CCP as the primary institutional broker during the two presidencies. The PC and PCC were both connected to the State Council initially, but they all ultimately switched their connections to the CCP. In all the years observed, the PCC had no or a low (< 2.6%) number of direct or indirect connections with the PC. Thus, the CCP has built itself as a necessary broker in the institutional network.
In general, the structures of the individual and institutional networks show the pluralism under authoritarianism. The connections between political elites Fig. 4 Links between political institutions: Separating political participation and legislation. Notes: The CCP-State Council coalition separates two institutional spaces. The space on the left has PCC for political participation, and that on the right has PC for legislation. Only showing links greater than 10% of the total connections. Percentages in parentheses represent the proportion of links that connect different political institutions. Detailed statistics and raw counts are in Table S6 are widespread and diverse, and they systematically differ across presidencies. The institutional political networks embedded in the nonprofit sector separate legislation and political participation by blocking non-CCP institutions from reaching legislative authority-a key strategy for balancing civic participation and the ruling party's hegemony. Table 5 presents more inferential statistics by regressing political affiliations and controls on nonprofit and political network degrees in both presidencies. 8 The table reveals some presidential differences: compared to Hu's presidency, CSOE's nonprofit embeddedness and CYL's political embeddedness dropped substantially in Xi's presidency, whereas the political embeddedness of MO, PCC, and School significantly increased. The results of CYL are worth highlighting as well: the CYL cadres had larger degrees of nonprofit embeddedness, but were marginalized in the political arena. Such a pattern survived the presidential change.

Main Regression Results: A Strategy of Differentiated Embeddedness
The estimations of the current presidency can be grouped into four categories as Table S5 in the appendix summarizes. 1) Strong nonprofit embeddedness includes CYL. 2) Weak nonprofit embeddedness includes Judiciary, LSOE, State Council, and CSOE. 3) Strong political embeddedness includes MO, State Council, and School. And 4) weak political embeddedness includes CYL and CSOE.

Checking Robustness: Considering Presidential Preferences and Diversity of Embeddedness
As Table 6 presents, the addition of embeddedness diversity and previous CYL experience substantially improves the adjusted R 2 of the models that predict embeddedness in the political network (from just over 0.1 to almost 0.6). Statistically, the test is effective because the stability of coefficients and the increase of R 2 are beneficial to evaluate robustness [50]. Table 6 confirms the differentiated embeddedness presented in Tables 5 and S5. It also supports the anecdotal observations of China's fractured political system [59,60]. Hu was the head of the CYL, and his presidency has been considered to have ushered in the rise of the CYL Clique [32]. As Table 6 shows, previous CYL work experience could help a cadre gain a significant political advantage under Hu's administration. But during Xi's presidency, such a privilege disappears, and current CYL leaders were also marginalized in the polity. 8 In general, more observations can generate smaller standard errors in regression analysis, resulting in more significant estimates. Accepting these estimates as significant can increase the risk of false positive conclusions (i.e., false claims accepted as true). Given the relatively small number of observations we have here, we are susceptible to the risk of false negative conclusions (i.e., true claims not supported by the analysis) but not false positive conclusions. This actually adds more confidence to the findings because the significant estimates are less likely to be false.

3
Bridging State and Nonprofit: Differentiated Embeddedness… Another concern is the All-China Women's Federation (ACWF), which has occupied a unique role between the party-state and civil society since its creation [71]. 9 Melting it with other organizations in the MO category may introduce a heterogeneity problem to the measure. I tested this concern by removing ACWF cadres from the regression analysis, and Table S7 in the Appendix shows no substantial difference. According to my analysis, ACWF individuals are not heavily embedded in nonprofit foundations. This finding is congruent with their operating strategy-"the politics of concealment": they camouflage their agenda using the party's language and efface themselves by attributing accomplishments to the party's leaders [71,[17][18]. Although this robustness test does not alter the findings, it cautions us that the negotiation between political and civil actors can happen through other channels and that the findings should thus not be overgeneralized.

Qualitative Member Checking
Following the member checking procedures described earlier, I conducted a series of interviews and focus groups. The fieldwork generally confirmed the findings, and the appendix has more details.

Discussion
By studying over two hundred high-profile political elites embedded in Chinese nonprofit foundations from 2011 to 2015, I conceptualized a framework of differentiated embeddedness to understand the relationship between the state and the nonprofit sector in China at the meso and micro levels. Table S5 summarizes the framework and primary results. According to the elites' embeddedness in the civil and political domains, they can be grouped into four categories. 1) Those who have strong horizontal connections in nonprofits are excellent state agents when it comes to reaching civic resources (i.e., the CYL). 2) Those who have strong vertical connections in the polity can serve as access points to the party-state (i.e., Mass Organization, intellectuals, and government bureaucracy). 3) Those who are less connected in nonprofits are weak state agents (i.e., the legal system, state-owned business, and the government bureaucracy). And 4) those who are less connected in the polity are ineffective access points to political resources (i.e., the CYL and state-owned business). The four groups of political elites constitute a strategy of differentiated embeddedness, which reminds students studying the relationship between the state and the nonprofit sector to shift their theoretical lens from a conventional dyadic paradigm to a network paradigm.
There are still a few lingering questions, first of which is the role of CCP in the differentiated framework. As the ruling party, it is not a dominant actor in all networks. Second, the special role of CYL should be detailed. Third, directions for future studies are worthy of discussion. The remaining sections clarify these questions.

CCP as the Vanguard: Strategy for Supervising Foundations
Veering away from conventional thoughts about authoritarianism, the Chinese political network leaves room for pluralism. As Fig. 3 shows, CCP leaders are not dominant actors in all political networks. Moreover, political leaders with different affiliations are widely interconnected, which provides a structural basis for the negotiations between different political groups. Meanwhile, CCP leaders are not the most connected political group in the foundations' board interlocking network either. The degree of CYL leaders' connections is twice as great as that of the CCP leaders', indicating that many more social elites are connected with CYL cadres than CCP cadres. We can speculate that the CCP's influence in the nonprofit sector may be mostly "behind the scenes," so it is not as direct as we conventionally assume. However, if we conclude that the nonprofits and the polity are sufficiently pluralistic to threaten the CCP's control, we are missing the unique features of the multilayered network system-in the party's official narrative, the CCP should serve as the vanguard of the polity, and its leadership, even without having direct control over all works, is the most essential feature of "socialism with Chinese characteristics." One strategy is to build the CCP as an essential broker to reach the PC, the legislative authority in the political network. The institutional space in Fig. 4 is separated into two: the left PCC space for political participation and the right PC space for legislation. By routing all connections through CCP, the party creates room for political participation while securing access to legislation. Yan [78,75] suggested that the CCP needs to "have an overwhelming majority in the PC to control legislative power, but intentionally keeps itself as minority in the [PCC]." I advance this control strategy further-it is not only the number of individuals that matters but also how they are organized.
Another strategy is to embed party branches in nonprofits directly. As the Regulations on the Operation of the Party Branches states, any "unit" with three or more CCP members shall establish a party branch. The party branches are independent political organizations and only report to the CCP system. Although not examined in this paper, it would be intriguing to study how these party branches function.

Differentiated Embeddedness and the Chinese Youth League
The CYL cadres were far younger than the non-CYL cadres at the same rank, indicating that they gained a substantial age advantage in promotion that may have helped them institutionalize the formation of the CYL Clique [32,162]. As a result, although the CYL is meant to prepare youths to be the CCP's "assistants and reserve army," it is also a potential rival to the CCP in the polity. Many anecdotal papers have reported the CYL to be one of the major factions within the Chinese political system [11,35,36]. This paper supports these claims with empirical resultsalthough political elites could not enter the political network while they were active CYL cadres, those with previous CYL work experience were significantly favored in Hu's administration (Table 6).

3
Bridging State and Nonprofit: Differentiated Embeddedness… The CYL system plays important roles in developing volunteerism and grassroots nonprofits [69,77], but such roles are limited to the grassroots level and are marginalized, even isolated, in the political network. This study reveals that the CYL cadres are the most connected political leaders in nonprofit networks but that their connections are unquestionably restricted-CYL leaders have no connections in the political network. Their cadre ranks tend to be low as well: all CYL leaders in this study were Bureau/Department leaders at the local level (Table 3), indicating that they were unable to reach the center of the polity directly.
The CYL actors are marginalized in political networks because of numerous institutional and promotional arrangements. For example, CYL leaders have to "reenter" the regular party or governmental departments through other political channels [14,54]. As of 2019, all CYL leaders in my dataset had been reassigned to CCP or bureaucratic positions. Moreover, Xi made a series of reforms to "keep CYL leaders grassroots" [7] and required CYL leaders to be "youths' friends but not young cadres" [52]. The youth organizations are also decentralized, preventing CYL leaders from forming cohesive political groups [14,439]. In all scenarios, we can expect that 1) the CYL leaders' future political careers and decisions can be influenced by their social roles and experience obtained from nonprofits and that 2) these leaders are strategically placed by the CCP so that their autonomy is limited in the party-state.

Policy Implication and Future Directions
The connections between the party-state and nonprofits have already been institutionalized following a strategy of differentiated embeddedness. However, this embeddedness is inefficient. Effective communication between the state and the nonprofit sector requires elites to have strong connections in both the civil and political domains, but the Chinese political elites who are well embedded in the nonprofit sector are disconnected from the polity, and those connected in the polity are weakly tied to nonprofits. The CYL elites are the most promising group that can bridge the civil and political spaces because they are widely connected in nonprofits. However, the party-state cannot effectively absorb social elites and receive feedback from the nonprofit sector if it keeps marginalizing CYL leaders in the polity.
Mass Organizations, intellectuals, and the government bureaucracy are all limited in serving as brokers. For example, Mass Organizations have been criticized as being institutionally bureaucratic and "out of touch with the people" [74]. Intellectuals, especially political scientists, can be widely co-opted by the state [49]. Moreover, government officials have been strictly regulated with regard to their involvement in the nonprofit sector [75].
Governing the relationship between the state and the nonprofit sector in authoritarian China is not an easy task, and the framework of differentiated embeddedness may only serve as a stimulus for future projects. As discussed previously, this study has numerous limitations regarding data sources and concept operationalization, and the findings and conclusions should be interpreted within the context of nonprofit foundations. Scholars can advance this topic from two general perspectives to improve the research design and generalizability. First, future studies should include data that can better describe the elites' connections within political domains (e.g., their coappearances in political events; [27]). Second, the longitudinal observation of the embedding strategy is worth tracking. The Charity Law of the People's Republic of China was officially released in early 2016 and represents a landmark in the development of the Chinese nonprofit sector. Furthermore, many sociopolitical situations have changed since 2015. Has the embedding strategy stayed consistent after 2015, and will it stay so throughout Xi's presidency? How do the elites' roles evolve over time? This study provides a snapshot of the pre-2015 period, which future studies can compare their findings to.