1. Introduction
No leader can govern a country alone; even the most powerful dictator needs a group of political elites.Footnote 1 To appoint those bureaucratic elites in the absence of effective electoral constraints, authoritarian leaders tend to rely upon a hierarchy of power, wherein the top leaders must be consistently concerned with the loyalty of their subordinates. Recent literature claims that concerns over loyalty lead authoritarian leaders to hire mediocre but loyal subordinates instead of competent but potentially threatening bureaucrats (Egorov and Sonin, Reference Egorov and Sonin2011; Reuter and Robertson, Reference Reuter and Robertson2012; Zakharov, Reference Zakharov2016). This logic also explains the emergence of political factions in authoritarian politics,Footnote 2 where political elites (clients) connected to a prominent national leader (the patron) tend to be promoted when the patron rises (Shih et al., Reference Shih, Adolph and Liu2012). However, scholars have not yet examined the extent to which loyalty to the patron might affect the cohesion of her network group. This raises an important question: do loyal subordinates prioritize the faction's interest (i.e., triumph as a group vis-à-vis other factions) over their individual interests (e.g., promotion and material resources)?Footnote 3
In this paper, we examine the existence and intensity of in-group competition within political factions using data from Chinese political elites. Specifically, we focus on factions that derive from interpersonal connections, through birthplace, education and work experiences, among the Politburo Standing Committee (PSC) members and provincial leaders (provincial party secretaries and governors). Scholars of Chinese politics have long debated the role of informal political factions in shaping key appointments and policy making (Nathan, Reference Nathan1973; Tsou, Reference Tsou1976; Dittmer and Wu, Reference Dittmer and Wu1995; Nathan and Tsai, Reference Nathan and Tsai1995; Tsou, Reference Tsou1995). Recent studies have revived the debate with rigorous empirical evidence. Shih (Reference Shih2008a) and Shih et al. (Reference Shih, Adolph and Liu2012), for example, claim that factional ties play a critical role in central-level elite promotion and resource distribution.
While these studies on political factions highlight a rather obscure aspect of the Chinese political system, they do not offer clear expectations regarding the behaviors of faction members, except to note that they tend to be loyal to their political patrons. Other political relationships within and across factions remain broadly understudied. This study addresses a remaining puzzle in the literature by approaching the issues of loyalty and elite competition in authoritarian politics from a different angle. In doing so, we demonstrate that loyalty to a patron is not identical to loyalty to the patron's faction. While remaining entirely loyal to their patrons, authoritarian elites often put the interests of other members at risk to maximize their own self-interest under fierce competition. Therefore, on the one hand, faction members compete against other factions to remain loyal to their patron, but on the other hand, they also compete internally to outrival in-group members to obtain more benefits from the finite pool of resources provided by their patron.
To analyze our claims, we collected data on Chinese PSC leaders and provincial leaders (both party secretaries and governors), as well as all inter-provincial media reports in local newspapers from 2000 to 2014 on corruption investigations and major industrial accidents. Our study attempts to understand the competitive behavior of faction members (here, provincial leaders) by analyzing inter-provincial news reports in those local newspapers. We focus on news reports across provinces regarding negative incidents under the jurisdiction of provincial leaders' faction peers, as compared to reports that concern members of different factions and elites without factional ties.
The analyses provide evidence supporting the notion of intense competition within factions: Chinese political elites, particularly those of closer rank, are more likely to promote news regarding their faction peers' political defects. The frequency and the share of negative reports in total inter-provincial reports reveal that the negative reporting toward other members within a patron's faction widely exists and is at least as intense as the negative reporting on other factions' members. We also conduct a series of robustness checks to examine whether the findings are driven by our choice of provincial leader (client) pool, statistical specification, competition measures, or data collection strategy. Finally, the analyses show that negative news reports reduce the promotion chances of reported cadres, while increasing the reporting elites' probability of promotion.
This study contributes to the literature on elite competition by clarifying the incentive structure of political subordinates under authoritarianism. Our empirical evidence on intra-faction competition among Chinese elites shows that clients under the same patron do not by default cooperate with one another; rather, they compete intensely for political resources and support from their patron. This does not mean that clients are disloyal. We find that they remain loyal to their patron, but that they do not prioritize the group's interest over their own, counter to an implicit assumption in many studies. Furthermore, by drawing empirical evidence from media reports, our study broadens the scope of data in authoritarian politics research. Media reports under authoritarian regimes are rarely considered to be accurate reflections of public opinion due to constraints on media freedoms (Egorov et al., Reference Egorov, Guriev and Sonin2009; King et al., Reference King, Pan and Roberts2013; Stockmann, Reference Stockmann2013; Lorentzen, Reference Lorentzen2014; Qin et al., Reference Qin, Strömberg and Wu2018). Instead of pointing out the limits of biased authoritarian media, however, this study exploits those potentially biased reports in order to measure the political intentions of authoritarian elites who supervise the media.
2. Elite competition in authoritarian regimes
The recruitment and promotion of bureaucrats stand among the principal tasks, and concerns, of authoritarian rulers. Recent studies on authoritarian governance have focused on the loyalty-competence tradeoff in authoritarian appointments. In most cases, those studies claim that dictators hire and/or promote loyal subordinates despite the fact that competent subordinates could govern more effectively, because more competent ones may betray and oust the dictator. Egorov and Sonin (Reference Egorov and Sonin2011) argue that a leader with a longer time horizon has stronger incentives to hire competent agents, while those with short shadows of the future choose loyal ones. Landry (Reference Landry2008) claims that the Chinese Communist Party (CCP) has managed to overcome this dilemma by promoting meritocratic and politically loyal cadres using a strictly hierarchical personnel system. In contrast, Reuter and Robertson (Reference Reuter and Robertson2012) show that authoritarian elections in Russia have led the leader to appoint less competent but loyal bureaucrats in order to bolster his political survival.
Our study is germane to this literature in that we analyze the loyalty-related political behaviors of authoritarian elites. At the same time, we raise several questions regarding the conventional assumptions in authoritarian recruitment and promotion studies. In previous studies, loyal-type and competent-type bureaucratic elites are exogenously distributed and discernible from one another. Therefore, the main question has typically been whom the leader would choose given the loyalty-competence tradeoff. Scholars also assume that as long as subordinates decide to remain loyal, i.e., to stay in the leader's faction or coalition, they will prioritize the political interests of the leader at any cost. We argue that these assumptions reflect the dictator's perspective, but not the subordinates'. From the subordinates' point of view, they can strategically decide the level and type of loyalty they confer in order to maximize their own political interests (Zakharov, Reference Zakharov2016). Whether, how, to whom and for what purpose they will be loyal may all be decisions that political elites weigh in the course of service to the political patron. If necessary, subordinates may hinder the faction's group interests as a whole, without betraying the patron's core interests, for their own benefit.
While addressing the broader issue of loyalty in authoritarian politics, this study more specifically contributes to the literature on political factions. Factions exist in any polity. In democracies, scholars have presented theories and evidence to analyze the role of political factions in the political process (Cox et al., Reference Cox, Rosenbluth and Thies1999; Persico et al., Reference Persico, Pueblita and Silverman2011; Boucek, Reference Boucek2012; Dewan and Squintani, Reference Dewan and Squintani2016). In authoritarian regimes, researchers show that factions launch coups, struggle, and share political power (Geddes, Reference Geddes1999, Reference Geddes2003; Bueno de Mesquita et al., Reference Bueno de Mesquita, Morrow, Siverson and Smith2003; Pepinsky, Reference Pepinsky2014). Moreover, in single-party authoritarian regimes like China, where no inter-party competition exists, factions represent the entities through which the political competition for power takes place.
In the context of Chinese politics, faction politics has long been at the center of academic debate regarding power struggles inside the CCP. Andrew Nathan's article (Nathan, Reference Nathan1973) and the rebuttals and surrebuttals that followed (Tsou, Reference Tsou1976; Dittmer and Wu, Reference Dittmer and Wu1995; Nathan and Tsai, Reference Nathan and Tsai1995; Pye, Reference Pye1995) built a theoretical foundation for the study of factionalism in China. Subsequent articles by Li (Reference Li2009, Reference Li2013, Reference Li2016) and other popular writings provide more concrete pictures of factional politics. Recently, scholars have shown that factional connections drive key political decisions, including personnel appointments and resource allocations (Huang, Reference Huang2000; Nathan, Reference Nathan2003; Shih, Reference Shih2008a; Shih et al., Reference Shih, Adolph and Liu2012; Jia et al., Reference Jia, Kudamatsu and Seim2015).
What the factionalism theory claims regarding elite behavior is straightforward. First, within the vertical ties between a patron and a client, the relationship is reciprocal. Second, in the context of horizontal ties between patrons, relationships are competitive or even hostile. Intense competition among multiple patrons protects the system by preventing any particular patron from obtaining overwhelming power (Nathan, Reference Nathan1973). In the course of competition with others, a patron will promote her clients along the political career ladder, while demoting members of other factions (Nathan, Reference Nathan1973; Shih et al., Reference Shih, Adolph and Liu2012). Economic resources may also be distributed through factional ties, thereby enhancing the clients' career profiles (Shih, Reference Shih2008a; Jiang, Reference Jiang2018).
Yet, the implications of factional theory are less clear regarding the horizontal relationships among clients, both within factions and across them. The seminal work by Nathan assumed that the horizontal relationships between clients do not matter because the vertical, personal relationship between the patron and the client constitutes the fundamental element of a faction (Nathan, Reference Nathan1973, p. 42). To the extent that horizontal relationships among clients have been addressed in the emerging literature on factional politics, the implicit assumption is typically that factions function as a political entity in which the members share common political interests (Geddes, Reference Geddes1999, Reference Geddes2003; Bueno de Mesquita et al., Reference Bueno de Mesquita, Morrow, Siverson and Smith2003; Shih et al., Reference Shih, Adolph and Liu2012; Pepinsky, Reference Pepinsky2014). Our study challenges this implicit assumption and aims to examine the intra-factional behaviors of members. We argue that clients must always have other co-faction clients in mind. A patron's ability to distribute benefits, either personnel favors or generous resources, is always constrained by other factions. Those limited resources will further be distributed among multiple clients attached to a patron, and ultimately, only one client can succeed to the patron's position. It is clear, then, that not all clients, no matter how loyal they are, can achieve the same pinnacle of power. What, then, is the nature of relationships among clients? Are they comrades, friends, or rivals?
We hypothesize two competing answers, the distinction in which depends largely on how one conceptualizes the faction as a group. One possible conceptualization is that a faction works as a solidary group. A solidary group refers to a group sharing a common destiny, in which unity is crucial. Members are loyal to the leader and to the group as a whole. If the group becomes vulnerable, the natural reaction from members would be to protect the group together. The existing literature on authoritarian politics, which generally treats factions as political unities, implicitly supports this solidary group theory. If a faction functions like a solidary group, one can assume that faction members will work in cooperation for the benefit of the patron and the group. Thus, competition should occur across factional lines.
Hypothesis 1: If a faction represents a solidary group, co-faction members are less likely to compete with one another than with the members of other factions or those without a factional tie, if they compete internally at all.
The other conceptualization of a faction is that it functions like a tournament.Footnote 4 In a tournament, participants in the same group compete with one another to be selected as the winner of each round. Given the pyramidal structure of a faction, clients always outnumber patrons. In this structure, the patron can neither promote nor favor all clients at the same time or to the same degree. Patrons can only favor their clients to the extent that limited resources allow, given that the patrons themselves are in competition with other patrons (Francois et al., Reference Francois, Trebbi and Xiao2016). More importantly, a patron would not promote all clients equally in order to strategically maintain the highest level of loyalty. If promotions were predictable and guaranteed, the incentive of clients to behave loyally for the patron would likely diminish, as failing to demonstrate loyalty would not change their career trajectories. For this reason, having connections to a patron may be a necessary condition but not a sufficient one to achieve a promotion. This theory thus predicts in-group competition, rather than cooperation or protection, within a political faction.Footnote 5 Nevertheless, whether the co-faction members compete more intensely within the group than they do with the members of other factions is not straightforward. This will depend on the power distribution among patrons and who the clients perceive as their competitors for the positions they covet next.
Hypothesis 2: If a faction follows a tournament-style promotion, co-faction members are equally or more likely to compete with one another than they are with the members of other factions or those without a faction.
We draw our empirical evidence from inter-provincial media reporting in Chinese local newspapers. Under the decentralized administrative system, local media are tightly regulated and supervised by the local governments. This structure allows us to study how the distinct political incentives of local leaders affect the content of local news reports.
All local media in China are either operated or supervised by the local propaganda department. The local propaganda department, in turn, is supervised by the party committee at the corresponding level, which is under the authority of the party secretary. At the same time, the local propaganda department is also guided by the Publicity Department of the Communist Party of China (CCPPD). Sometimes, the CCPPD bans local media from reporting on or even conducting interviews on provocative topics, allowing only copy reports from the national Xinhua News or People's Daily. Otherwise, local media are generally able to publish news reports without central intervention (Zhou, Reference Zhou2011). Not all articles go through the party's scrutiny, although the party holds the authority to review all articles before publication. Instead, public officials typically exploit their personal networks to access articles related to them before publication and block potential negative news (Young, Reference Young2012).
In this hierarchical and networked environment, editors and reporters of media companies have few incentives to criticize their own local government on a crucial topic, as their career paths are largely determined by evaluations from the local leader. Just as the central authorities use the national media to promote their political propaganda, local leaders also use the local media to propagate their achievements or to criticize their political rivals (Bandurski and Hala, Reference Bandurski and Hala2010; Shirk, Reference Shirk2011; Stockmann and Gallagher, Reference Stockmann and Gallagher2011). In this context, editors and journalists have more latitude to criticize other localities' mis-governance, since criticizing other local governments does not harm the political gains of their own leaders. It may instead contribute to their supervisors' political success, as it disparages those local leaders who may be in competitive relationships with their own leader.Footnote 6 This structure suggests that the empirical patterns reflect the following two potential processes: the provincial leadership may directly or indirectly nudge the local newspaper to publish more stories about negative events in a potential competitor's province. Or, the news editors or local propaganda department may choose to report those stories to signal loyalty to the local leaders. Although in our empirical setting we cannot distinguish between these two potential paths, the mechanisms are highly likely to coexist.
This, however, does not mean that local media will report negatively on other provinces as frequently as possible. Instead, they must strategically allocate negative news on others as doing so comes with constraints and costs. Excessive reporting on negative cases is not optimal for the communist party as a whole (Lorentzen, Reference Lorentzen2014). Another critical cost is the possibility of political retaliation. Thus, within the context of these constraints and potential costs, local media and leaders have an incentive to strategically report negative news from other provinces in a way that maximizes their benefits.
3. Empirical strategy
3.1. Data
We collected comprehensive data on the career backgrounds of all the central and provincial leaders in China, along with news content from the major local newspapers in China (all provincial newspapers and major prefecture newspapers).Footnote 7 Using the largest digitalized news content database of Chinese media, a Hong-Kong-based data vendor known as WiseNews (Qin et al., Reference Qin, Strömberg and Wu2018), we aggregate the content of 143 local mainstream newspapers in Mainland China from 2000 to 2014. Among these 143 newspapers, 49 are party-line newspapers directly controlled by the corresponding level party committees,Footnote 8 while the remaining 94 are commercial newspapers operated by local news corporations. Geographically, these newspapers cover 30 provinces in Mainland China, though not Beijing. Beijing newspapers are excluded from our local newspaper pool, as many Beijing newspapers are considered to be the national newspapers or to have special connections to the national ones.
The circulation of these local newspapers indicates that they are not trivial media. According to the China Newspaper Yearbook (2011), total circulation of all 143 local newspapers is 73.71 million readers, or 0.52 million per newspaper. A comparison with the circulation of major national newspapers, such as People's Daily (approximately 2.8 million), Global Times (1.04 million), and Guangmin Daily (0.5 million), suggests that the scope of circulation for local papers is notable.
Through keyword searches and article counts, we construct media coverage data over two types of local-level negative events: corruption investigations involving local governments and large-scale industrial accidents. For corruption investigations, we use “shuanggui” as the keyword. Shunggui is an internal disciplinary process conducted by the Central or Provincial Commission for Discipline Inspection of the Chinese Communist Party on party members suspected of graft. To mitigate the concern that our findings are sensitive to the specific keyword, we examine an alternative set of keywords, “luoma (dishonorable dismissal due to corruption)” with “fubai (corruption),” as a robustness check.Footnote 9 For industrial accidents, we use the list of industrial accidents compiled by the State Administration of Work Safety under the State Council,Footnote 10 along with a keyword search using the term “kuangnan (coalmine disasters),” and “yiwai” or “shigu” (accident). In addition, to measure the share of negative reports among all inter-provincial news reports, we establish a base of all news articles mentioning other province names. Finally, we match the number of news reports per year to over 26,000 dyads of provinces. More specifically, we identify and pair the provincial party secretary and governor where the newspaper article was published (news province, hereafter) and the same leaders from the provinces in which the event reportedly occurred (event province, hereafter).
We also construct a panel data set that consists of the political careers of Chinese politicians in provincial governments and in the PSC to identify factional connections and political turnover.Footnote 11 We first identify provincial party secretaries and governors over our period of analysis by collecting the names from the China Communist Yearbook. To obtain factional connections and promotion data, we extract information from their personal biographies, including a number of individual characteristics such as age, gender, place of birth, education, and work history, using the Chinese internet search engine Baidu Encyclopedia. By matching the personal information and work histories of provincial officials with information on the PSC members, we construct a faction network for each provincial leader in our sample. Our definition of factional connection follows Shih et al. (Reference Shih, Adolph and Liu2012): national leaders and local officials are connected if they share birthplaces, educational institutions or workplaces over one year within two administrative steps.Footnote 12 Two officials belong to the same faction if they are connected to the same PSC member. All variables are summarized in Table A.2.
Some conspicuous cases from our data illustrate that intra-factional competition can be notably fierce. When Wang Yang was the party secretary of Guangdong (2007–2012), numerous news reports on Guangdong's corruption cases appeared in newspapers published in Shandong province. The party secretary of Shandong at that time was Jiang Yikang (2008–2017), who was tied to the same patron, Wen Jiabao, as Wang Yang. The frequency of news reporting from Guangdong on corruption cases in Shandong also increased at a remarkable rate, exceeding even the number of reports on corruption in Chongqing, a province governed by Bo Xilai, a famous political rival of Wang Yang from a different faction. Another example reveals the competition among the members of the Communist Youth League faction (also known as “tuanpai”). When Hu Chunhua became the party secretary of Guangdong in 2012, the news coverage of corruption cases in Guangdong spiked in Hunan's newspaper, where another tuanpai member, Zhou Qiang, was serving as the party secretary.
3.2. Specification
Our baseline model employs a province-pair panel from 2000 to 2014:
Our key dependent variable NewsCount d(i,j) t represents the number of bilateral news reports published in news province i on events that occurred in event province j in year t. To capture the dyadic reporting dynamics, we create three different news reports variables: the number of negative reports by i on corruption investigations in j [A], total news reports by i on j [B], and the share of negative news among all news reports [A/B].
Our key explanatory variable measures the faction affiliation of the provincial leader dyads. We construct a dummy variable that identifies whether the two provincial leaders (party secretaries or governors) belong to the same faction. To estimate the likelihood of political competition aside from factional ties, we use three measures as control variables: the age gap, the tenure year gap between the two leaders, and the gap of GDP per capita between the provinces.Footnote 13 We also add an indicator variable for cases where the party secretary and the governor have the same factional tie, on the logic that belonging to the same faction may encourage the leadership team to attack the other province more frequently or may insulate the province from negative media reports.
In addition, we include province dyad fixed effects (γij) and year fixed effects (λt). Province dyad fixed effects address not only province-specific effects of news provinces and event provinces, but also dyad-specific effects, controlling for the effects of any provinces with a special relationship, such as rivals.Footnote 14 We primarily rely on the linear model with multiple levels of fixed effects.Footnote 15 Finally, the robust standard errors are clustered at the dyad level to address the complex structure of dyadic data where dyad units are networked (Cameron and Miller, Reference Cameron and Miller2014; Aronow et al., Reference Aronow, Samii and Assenova2015).
We further examine whether the media coverage on negative events affects the political careers of provincial leaders using the following specification:
Following Li and Zhou (Reference Li and Zhou2005), Turnover pt is coded as an ordinal variable, with promotion taking the value of 4, lateral transfer to positions of the same rank and/or staying in office 3, retirement to the advisory position in the People's Congress or Political Consultative Conference 2 and demotion 1. To explain the impact of negative events, independent from the media coverage, on the promotion prospects of provincial leaders, we include the number of major corruption cases (NegEvents pt). We collect the major corruption case data from the prefecture-or-above-level corruption cases reported in Procuratorial Daily. The key explanatory variable is the total number of news reports on corruption investigations having occurred in province p (NewsReported pt) and the total number of news reports by province p on corruption investigations in the other provinces (NewsReporting pt) during year t. We also control for a number of covariates that previous studies have found to be related to cadre promotions such as local GDP per capita, local GDP growth, cadre's factional ties, age (in quadratic form), and education level (Li and Zhou, Reference Li and Zhou2005; Shih et al., Reference Shih, Adolph and Liu2012; Yao and Zhang, Reference Yao and Zhang2015), along with the province (γprov) and year (λt) fixed effects.
4. Results
We first show the loyalty-oriented behaviors of provincial leaders. We search and analyze local newspaper articles containing the name of PSC members to reflect the personal connection between the current local leaders and the PCS members. The results in Table A.3 confirm the findings of Shih (Reference Shih2008b), demonstrating that provincial party secretaries with factional ties to a PSC member publish substantially more news on their patron(s) in their local newspapers (137% in Model (1)).
Table 1 present the main results for all party secretary dyads. Model (1) employs a simple specification with relevant fixed effects to see whether dyads of provincial party secretaries sharing the same faction leader report more or less on the other's corruption investigations. A pair of party secretaries sharing the same faction leader is 36.7% more likely to report on corruption cases involving the other than average inter-provincial reports.Footnote 16 In addition, other variables that highlight career incentives demonstrate expected results. Party secretaries at a similar place in the political cycle, measured by age or years in office, are more likely to report on one another's political defects.
Table 1. Factional ties and interprovincial news reports on corruption investigation (party secretaries)
Notes: Robust standard errors clustered at the provincial dyad level are shown in parentheses. Variables not shown include provincial dyad fixed effects, year fixed effects, and constants. +p < 0.10, *p < 0.05, **p < 0.01, ***p < 0.001.
Note that we contrast dyads sharing a faction to all other cases, including different faction members and those without factional ties, in Model (1). However, grouping together dyads of party secretaries with no factional ties with dyads who belong to different factions can be problematic. To address this problem, we separate the reference group into four categories: [1] only the event province leader has a factional tie (News = 0 & Event = 1); [2] only the news reporting province leader has a factional tie (News = 1 & Event = 0); [3] neither leader has a factional tie (News = 0 & Event = 0); and [4] the two leaders in a dyad belong to different factions. By including groups [1] to [3] as control variables in the analysis, we leave the dyads with different factional ties as the reference category in Model (2). This specification confirms the intra-faction competition argument contrasted to cross-faction competition. Model (3) further tests whether multiple factional ties, in contrast to a single factional tie, matter in terms of reporting on other party secretaries' negative news; here, the results indicate that having single or multiple patrons does not change the results.
A possible alternative explanation to the results in Models (1) to (3) is that members of the same faction may report more about their peer's province in general. To test this possibility, we collect data on all news articles in local newspapers where the names of other provinces appear. Models (4) to (6) replicate Models (1) to (3) using total news reports on the event province as the dependent variable. We find that co-faction members are not simply more likely to report about their co-faction member's province.Footnote 17
In the subsequent models, we examine whether factional ties affect the share of negative news by dividing the number of corruption investigation news articles by the total number of news reports on the event province.Footnote 18 The results reveal that co-faction members report negatively on their own peers. Being in the same faction increases the corruption report share by 1.4% (Model (9)). We acknowledge that compared to the absolute count measure, the ratio measure provides weaker evidence for intra-faction competition. Nonetheless, these findings, along with the positive findings from earlier models using the count variable, militate against the solidary group hypothesis.
We also examine the outcomes using other pools of provincial leaders. In theory, a party secretary leads the communist party committee, while a governor administers the local government. By structure, the political authority of party secretaries thus takes precedence over the governors'. Estimations reported in Table 2 replicate the analyses of Table 1 using the pool of all party secretaries and governors, where the effects are smaller but consistent with those in Table 1. Table A.6 presents statistical analyses using governors only. Given the more administrative and less political nature of the position, we find the effects to be somewhat weaker among dyads of governors.
Table 2. Factional ties and interprovincial news reports on corruption investigation (party secretaries and governors)
Notes: Robust standard errors clustered at the provincial dyad level are shown in parentheses. The full table is available in Table A.5. +p < 0.10, *p < 0.05, **p < 0.01, ***p < 0.001.
One may be concerned that the pervasiveness of corruption shapes the media reports. Measuring the actual level of corruption is an empirical challenge beyond the scope of this paper. We attempt to address this issue by adding the event province-year fixed effects, which captures the province-year specific corruption trends, such as a big scandal in a province. The results in Tables A.9 and A.10 are largely consistent with the previous findings, suggesting that our results are not driven by any bias from the omitted corruption magnitude.
If intra-faction competition explains elite behavior in authoritarian regimes in a systematic manner, we should expect to find similar patterns regarding other politically salient issues. We choose massive industrial accidents as another critical issue in China (Jia and Nie, Reference Jia and Nie2017). The majority of industrial accidents in our dataset are coalmine related, followed by accidents in the chemical industry. Coal mining has been a notorious cause of death in China due to unsafe work conditions (Hong, Reference Hong2018). Table 3 shows similar patterns in inter-provincial news reporting of industrial safety accidents compared to those regarding corruption investigations. Provinces in which the respective provincial party secretaries are members of the same political faction more frequently report on each other's industrial safety accidents in their local newspapers. However, we find no significant pattern regarding the share of accident news reports among total news reports. Table A.8 replicates the models from Table 3 using the pool of party secretaries and governors. The results remain qualitatively the same.
Table 3. Factional ties and interprovincial news reports on industrial accidents (party secretaries)
Notes: Robust standard errors clustered at the provincial dyad level are shown in parentheses. Total news analyses are not reported as they are identical to Models (4) to (6) in Table 1. The full table is available in Table A.7. +p < 0.10, *p < 0.05, **p < 0.01, ***p < 0.001.
A crucial assumption underlying our baseline analyses is that local newspapers publish news articles based on the explicit interest of the provincial leaders. If our assumption is correct, one should expect that party-line newspapers, which are more closely aligned with the politicians, show a clearer pattern of politicized news reporting. Commercial newspapers, by contrast, have sales as their primary concern and should thus be less likely to utilize news reports as a tool for political competition between party leaders (Stockmann, Reference Stockmann2013; Lu and Ma, Reference Lu and Ma2019). The analyses in Table A.11 confirm that the mechanism behind our main findings is intra-faction competition. Negative media reports on co-faction peers are more pervasive in party-line newspapers than in commercial papers.
One may be concerned that our keyword choice (shuanggui) drives the results. To address this issue, we adopt another set of keywords, luoma (dishonorable dismissal due to corruption) and fubai (corruption), collect a new set of data, and run new analyses using the same models. We use these two words together to minimize potential noise in the data generating process, as both words, if used separately, may also indicate affairs other than corruption investigations. The results in Table A.12 are not only consistent with the previous findings, but in fact provide even stronger support for our argument.
Additionally, we analyze the effects of a breakdown in the power-sharing system in China. In the theory section, we noted that within-faction competition is facilitated by a power-sharing among factions. Under balanced power, the spoils should be distributed across factions, and each patron should be able to control only a limited number of posts for clients. However, when such a power-sharing arrangement at the patron level breaks down, it likely incites inter-faction competition, to either take or defend spoils from the other factions, and reduces intra-faction competition. To investigate this hypothesis, we exploit the years after the 18th Party Congress, the period that many China observers describe as a period of political power concentration, unprecedented over the last few decades (Li, Reference Li2016). The results in Table A.13 support this observation. After the 18th Congress, within-faction competition declined. The local news reports on co-faction province's corruption investigations is not significant under the 18th leadership regardless of measures or sample pools.
Finally, we examine whether the negative reports by other provinces, frequently those published in the newspapers from a co-faction peer's province, reduce the promotion prospects of the reported-on provincial leaders. This examination is critical to understand the nature of inter-provincial reports on corruption and other potentially sensitive issues. Some may argue that the inter-provincial reports among co-faction members may be aimed at achieving a positive impact. For instance, the reporting province newspapers may intend to promote the co-faction province's anti-corruption efforts. If this is the case, an analysis of turnover and promotion will reveal positive effects of inter-provincial reporting on the promotion probability of reported leaders. In contrast, if the incentive is to outrival other peer cadres, the results will incline toward negative findings.
Table 4 addresses the political turnover of provincial party secretaries. Model (1) first tests whether the event itself, i.e., the presence of big corruption cases, negatively affects the probability of promotion. In Model (2), even after controlling for the impact of the events we find that other provinces' news reports significantly reduce reported-on leaders' promotion prospects.Footnote 19 Interestingly, the more frequently provincial newspapers report on corruption investigations in other provinces, the more likely their own party secretaries are to be promoted. Furthermore, Model (3) indicates that repeated negative reports from the same faction peers carry even more severe political costs for the reported-on leader, while bringing larger benefits to the reporting provincial leaders. Models (4) and (5) use the ratio of corruption news to total news reports in the event province; these analyses further confirm that negative reporting by other provinces significantly harms the promotion prospects of the event province leader, while other variables have no statistically significant impact. To confirm the robustness of our findings, we reanalyze the provincial leaders' promotion probability using the extended pool of party secretaries and governors (Table A.14). We also employ a binary variable to measure promotion (Table A.15).
Table 4. Corruption news reports and provincial party secretaries' promotion
Notes: Robust standard errors are shown in parentheses. Variables not shown include provincial GDP growth, local GDP per capita, cadre's factional ties (single tie or multiple ties), age, age2, years of education, province and year fixed effects and constants. +p < 0.10, *p < 0.05, **p < 0.01, ***p < 0.001.
Why do negative news reports in other provinces' local newspapers, apart from the negative events themselves, harm the promotion probabilities of reported-on provincial leaders? It is unlikely that national leaders follow all local newspapers on a daily basis. Nor is it very likely that the central government collects local news reports and systematically employs the data to evaluate provincial leaders. One possibility is a resonance effect. Once a corruption case or a large-sized safety accident in a province goes viral in other regions' newspapers or among the public, it becomes considerably more likely that newspapers in Beijing will pick up the news, and that national leaders in Beijing notice the case or the public response to it. To examine this possibility, we test whether provincial level reports are associated with the national news reports (Table A.16). We find suggestive evidence that the number of local news articles on corruption investigations and industrial accidents increases coverage of the same events in Beijing's media.Footnote 20
Conclusion
Existing studies often assume that loyalty keeps authoritarian elites within a leader's subgroup cohesively committed to the leader. In this paper, we argue instead that authoritarian elites seek to maximize their own self-interest, not the group's interest, while remaining loyal to their leader. The empirical evidence that we presented strongly supports a logic of in-group competition, generated by loyal but self-interested elite behavior. Members of the same faction at a similar rank with similar turnover timing are more likely to tattle on one another's bad news or misgovernment than are members of other factions or provincial leaders with no factional ties to the current PSC members. These reports are politically effective for provincial leaders: those who report more frequently on other provincial leaders' poor performance are more likely to be promoted, whereas those more frequently reported on are less likely to be promoted.
Our analyses offer a fresh take on how authoritarian politicians behave under a factional structure with competitive pressures for promotion. An important question that remains concerns the substantive implications of this intra-factional competition for authoritarian regimes and for ordinary citizens. One potential effect of intra-faction competition, particularly in the form of media reports, is expanded access to information for the public. In democratic politics, Dewan and Squintani (Reference Dewan and Squintani2016) show that informal intra-party factions benefit the party by facilitating information sharing. Due to limited media freedom in authoritarian regimes, particularly in places like China, revelations of information disadvantageous to incumbent public officials are less frequent than in democracies. In particular, reporting on corruption cases is especially tricky, as it can directly affect the public's trust in the government (Wang and Dickson, Reference Wang and Dickson2017). Hence, reporting on the misgovernment of competing local leaders enables other public officials and citizens to have more information on their government. How broadly the information available through intra-factional competition benefits citizens, however, requires deeper research. Although the revelation of information is generally meaningful in otherwise opaque authoritarian regimes, the degree to which the revealed information is unbiased and truthful is another question.
Another possible effect of intra-faction competition is the potential for efficiency gains through competition. In-group competition may benefit the regime and citizens by leading otherwise simple loyalty-seekers to compete among factional peers to stand out. Although the authoritarian recruitment literature has long recognized the trade-off between loyalty and competence, this does not necessarily imply the absence of performance competition among loyal subordinates. If all subordinates are loyal, the patron has no reason not to consider competence in promotion. Nevertheless, it is also noteworthy that factional competition is informal, and hence not necessarily based on defined criteria. Therefore, the basis of competition and its effects on promotion largely depend on the patron's needs and willingness to acknowledge the clients' performance. Future research might investigate how patrons' political power affects the competition within and between factions.
A final caveat is worth noting. We do not intend to argue that loyalty does not work for elite cohesiveness. Neither do we claim that members of different factions do not compete. Rather, the aim of this paper is to serve as a counterpoint to the existing view on elite competition which suggests that authoritarian political factions act as a united group with an identical political goal. Internal politics within an elite group can be as dynamic as inter-factional conflicts in an authoritarian setting. Furthermore, faction politics do not necessarily imply reduced competition among lower level politicians compared to a model of meritocratic competition. While vertically loyal and thus competing in some manner with members of other factions, faction clients are also consistently competing with co-faction members not just on loyalty but also on other aspects of political performance.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/psrm.2019.61
Acknowledgments
We would like to thank anonymous reviewers, Jidong Chen, Haifeng Huang, Junyan Jiang, Franziska Keller, James Kung, Jonghyuk Lee, Xiao Ma, Adam Przeworski, Victor Shih, Erik H. Wang, Yuhua Wang, Tianyang Xi, and participants of the 2016 APSA Mini-Conference on Chinese Politics, seminar participants atthe National School of Development, Peking University and at Beijing Normal University for helpful suggestions and comments. We deeply thank Leo Y. Yang for helping the data collection. The project was supported by the School-based Initiative Grant from the Division of Social Science, Hong Kong University of Science and Technology (SBI16HS08). All errors are our own.
