The Research Problem of Identifying Factions

In the Chinese context, policymaking typically follows a top-down pattern, making elite politics a common concern for researchers.Footnote ¹ When attempting to understand elites – for example, their policy positions and promotional chances – researchers quickly realize that it is factional ties that mobilize and align them.Footnote ² Consequently, clarifying the factional factor has become an urgent research priority in the field of China studies. Earlier researchers relied on private information and personal judgements to parse the factional affiliations of Chinese elites. Later scholars developed more systematic approaches, including the background-based criteria employed by Sonja Opper, Victor Nee and Stefan Brehm, and Jérôme Doyon and Franziska Keller, as well as the patronage-based criteria of Victor Shih and Junyan Jiang.Footnote ³ Among these efforts, Jiang’s approach, which uses within-tenure promotion as an identification criterion, stands out as the most accessible and justifiable approach. However, there remains the possibility that some promotions are made without patronage consideration. We thus propose replacing Jiang’s measures with more stringent ones: “double promotion” and “promotional grooming.” We then test the available and verifiable faction detection approaches against the odds of China’s prefectural-level leaders crossing career thresholds between 2000 and 2020 to assess their validity in supporting classical factional hypotheses.

This article aims to improve the identification of factions in Chinese politics. After addressing the significance of factions in the next section, we categorize the existing methods of faction identification in the third section into three generations: the first relying on “rumours-have-it,” the second on “backgrounds-in-common” and the third on “practices-of-patronage.” Among these approaches, we find Jiang’s third generation within-tenure promotion method to be the best available option, although it still has limitations. Consequently, we propose two revisions to enhance Jiang’s measure. In the fourth section, we verify these approaches and present our test results. The fifth section discusses these findings and explores the applicability of these methods in the study of China to comparative elite studies as a whole.Footnote ⁴

Factions, Guanxi and the Dynamics of Chinese Politics

The significance of factions in Chinese politics primarily stems from the nature of Chinese society and its political landscape.Footnote ⁵ Historically, in the absence of alternative social organizations, such as the Christian church in Western culture, Chinese society has been dominated by families and extended families (jiazu 家族).Footnote ⁶ In this context, Chinese behaviour tends to be more particularistic and relationship-oriented, with trust typically extended within small circles. Consequently, Chinese society is often described as a “relationship society” (guanxi shehui 关系社会).Footnote ⁷ Particular social networks thus form the micro-level foundation for the social interactions and political alliances among Chinese people.Footnote ⁸

The structure of the Chinese polity also contributes to the rise of factionalism in several ways.Footnote ⁹ First, in a one-party dominated political system, political selection often follows a pattern of “sponsored mobility,” where subordinates’ appointments depend on their superiors’ support.Footnote ¹⁰ Second, as resources are closely tied to these appointments, most subordinates desperately seek their superiors’ patronage, resulting in highly imbalanced relationships. Third, superiors are also motivated to select those whom they trust, as seniors’ power depends on their subordinates’ loyalty.Footnote ¹¹ Finally, despite the significant impact of these patron–client relationships, the Chinese Communist Party’s (CCP) comradeship principle prohibits the public acknowledgement of such political alliances.Footnote ¹² Therefore, generations of China scholars have had to contend with the unavoidable challenge of deciphering factional affiliations.

Furthermore, recent progress in the study of both comparative authoritarianism and Chinese politics further underscores the importance of factionalism. Some research indicates that in political systems where institutionalized power transitions are absent, loyalty becomes the primary concern for politicians.Footnote ¹³ Other research suggests that political loyalty has been maintained through factional ties in the Chinese context.Footnote ¹⁴ Additionally, some scholars find that faction-mediated interactions inside the Chinese government can enable superiors to better understand their subordinates, thereby resolving the common principal–agent problem.Footnote ¹⁵ Furthermore, given the limited prospects for promotion, subordinates can be more effectively motivated by faction-based commitments, which in turn can generate the performance necessary for regime legitimacy.Footnote ¹⁶ In sum, we have many salient reasons to recognize the key role of political factions in both Chinese politics and elite politics more generally. As such, reliable tools to parse such factions would benefit both fields of study.

Identifying Factions in Chinese Politics: The Efforts of Three Generations

Given the centrality of factions in Chinese politics, researchers must confront the challenge of elucidating factional affiliations among political elites.Footnote ¹⁷ Unfortunately, this task is complicated by several characteristics of Chinese political factions. First, factional ties are fundamentally informal social relationships, which makes them highly varied and flexible. Thus, defining these ties using simple, fixed criteria proves to be very difficult. Second, factional activities are usually concealed, rendering them difficult to trace through publicly available data.Footnote ¹⁸ Moreover, earlier Western studies on factions were predominantly ethnographic and also lacked well-developed, testable methods to identify these ties.Footnote ¹⁹ Consequently, China scholars have continuously struggled to generate effective approaches for identifying factions. These efforts can be roughly categorized into three generations: the first generation follows a “rumours-have-it” approach, the second looks towards the “backgrounds-in-common,” and the third focuses on “practices-of-patronage.” In the remainder of this section, we review the approaches, origins, strengths and weaknesses of each generation.

Faction identification 3.0 and beyond: the practices-of-patronage approach

The “backgrounds-in-common” approach offers a systematic way to track Chinese factional affiliations, but many problems remain. Third-generation China scholars therefore seek to enhance its validity without sacrificing clarity. Drawing inspiration from theories of clientelism, the improvement strategies can be split into two approaches, determined by which half of the patron–client dyad receives more focus. One highlights the loyalty aspect; hence the practices of subordinates are traced to demonstrate their allegiance to their superiors. This approach is best demonstrated by Shih’s exemplary study.Footnote ³³ Conversely, the second approach emphasizes the patronage aspect of factional relations, so researchers investigate the practices of superiors in ensuring the promotion of their subordinates. Jiang’s work offers a representative case of this second perspective.Footnote ³⁴

These new approaches that target “practices-of-patronage” offer a direct and relatively valid way to identify political factions. However, it remains to be seen whether these new efforts have the potential to supersede the earlier second-generation approach. An examination of Shih’s methodology, which tracks subordinates’ “displays of loyalty,” provides insight. First, it is not fully clear whether those displaying loyalty are actually the clients researchers most want to find. As Shih’s logic goes, the loyalty displays of subordinates are designed to win the trust of their superiors. If superiors already trust their juniors, however, there is no need for these clients to publicly and excessively demonstrate their loyalty. Rather, it is those subordinates whose loyalty is most in question who have to express their faithfulness. Therefore, using public displays, such as those found in local newspapers, as a primary indicator of factional affiliations may be misleading.

Second, the valence of “lip service” in loyalty displays remains ambiguous. In the Chinese political system, even when subordinates disagree, they do not express their disagreements publicly. In turn, superiors know better than to blindly trust the public expressions their juniors make; after all, there are many people who “signal left but turn right” or “wave the red flag while opposing the red flag” (dazhe hong qi fan hong qi 打着红旗反红旗).Footnote ³⁵ Furthermore, expressions of loyalty can also be conveyed in an implicit manner, varying according to the needs of particular occasions. In sum, there is an absence of well-defined and consistent criteria to evaluate these “displays of loyalty.” For scholars, it remains unclear through which channels, with what signals and to which extent displays of loyalty should be considered indicative of factional allegiance. Therefore, while Shih’s proposal is inspiring, its criteria are difficult to apply, and its data require additional efforts to access and curate into a format suitable for analysis. Consequently, few researchers are able to build on Shih’s exemplary and innovative research.

Tracking factions from the opposite direction, Jiang’s proposed measure of “provision of support” identifies factions based on the promotions of subordinates during their superiors’ tenures. This approach has several advantages over other methods. First, the focus of identification is well-defined, emphasizing only the most important practice of patronage – appointments or promotions – between direct dyads of superiors and subordinates. Second, the information required for this approach is readily accessible and, in most cases, already available to most researchers of Chinese elite politics. Additionally, Jiang’s method sheds light on the future dynamics of factions; its identification criteria – the “provision of support” and its metric (appointments or promotions) – encompass not only the fulfilment of faction-based commitments but also the recruitment of new members. For these reasons, Jiang’s approach is probably the most effective method currently available for identifying political factions in the Chinese context.

However, there are still a few unresolved concerns with Jiang’s approach, specifically related to his “within-tenure promotion” criterion. In the Chinese government, routine promotions can also occur when there are position vacancies or when incumbents have served well beyond their expected tenure. Promotions in these circumstances have little to do with faction-based patronage. Therefore, based on our field experiences, we propose two minor improvements to Jiang’s approach to better distinguish between “intentional promotions” and “routine promotions.”Footnote ³⁶

First, we propose replacing Jiang’s “within-tenure promotion” criterion with “within-tenure double promotion.” This new, more selective measure can help to exclude non-patronage based promotions, as the likelihood of subordinates receiving two consecutive “routine” promotions within the same tenure from a single superior is very low. Our second proposal extends Jiang’s criteria by considering how superiors use indirect channels to facilitate their juniors’ career advancement. Specifically, when promotion opportunities are unavailable owing to excessive requirements or a scarcity of vacancies, superiors may arrange training at a Party school for their client subordinates. Such training essentially grooms the recipient client for future success. This training is extremely valuable for future promotion prospects not only because attendees acquire new competencies but also because attendance signals that they are “ready for appointment or promotion.” As discovered during fieldwork, these options provide superiors with adaptive flexibility in determining how best to support their favoured subordinates: one option is more direct, open and fearless of any suspicion; the other is indirect, almost covert and paves the way for subordinates’ advancement.

Table 1 summarizes the features of the different faction-identification approaches developed by the past three generations of scholars focused on elite Chinese politics.

Table 1. Contrasting the Major Approaches to Identifying China’s Political Factions

Empirical Tests of Common Approaches to Factional Identification

To date, several methods have been proposed to identify factions, yet almost none of them has been systematically and quantitatively verified. In this section, we test the following three approaches against the general odds of China’s prefectural-level leaders crossing career thresholds between 2000 and 2020: (1) backgrounds-in-common, (2) practices-of-patronage, and (3) this study’s revised practices-of-patronage. With these tests, we can assess the effectiveness of these approaches in fulfilling the classical hypothesis of Chinese factionalism.

In our test design, the dependent variable is the career breakthroughs of China’s prefectural-level Party secretaries. We focus on their promotions because, in Chinese politics, promotion is a key mechanism for distributing power and resources.Footnote ³⁷ Hence, following the logic of factional politics, i.e. the norm of reciprocity between patrons and clients, promotion could serve two primary functions: first, to reward followers and ensure their loyalty, and second, to empower followers and consolidate the patron’s power base. We focus on prefectural-level leaders due to data availability and define breakthroughs as promotions to vice-provincial positions with real power, such as vice-governors and members of the Standing Committee of the provincial Party committee. We exclude vice-provincial positions without executive power, such as the vice-chairs of the province’s people’s congress and the CPPCC. Also, we only focus on Party secretaries, as it is highly unlikely that other prefectural-level leaders will be directly promoted to the vice-provincial level. In Chinese politics, such promotions are critical for local leaders as many would otherwise be trapped indefinitely at the prefectural level.Footnote ³⁸ Moreover, aspiring national-level officials largely follow a career pattern of “small sprints,” meaning that breakthroughs must be achieved successfully and smoothly; otherwise, these cadres will not have sufficient time to advance beyond the provincial arena to join national-level leaders at the top.Footnote ³⁹ Hence, the duration of each appointment is also a consideration.

When examining these breakthroughs, we analyse several key dimensions: (1) whether these officials successfully cross the prefectural career threshold to become provincial leaders; (2) whether they cross the threshold quickly, which we define as within 3 years, given that the average tenure is 4.4 years; and (3) whether they cross the threshold through “manufactured performance,” meaning that their appointment positions are known to ease the delivery of superior performance.Footnote ⁴⁰ This dimension is measured by the average economic growth rates in the five years preceding their appointment, as higher rates facilitate the demonstration of performance; and, finally, (4) whether the officials secure their provincial-level positions through “manufactured vacancy,” indicating that their prospective positions were deliberately vacated to accommodate them. This item is measured by the term lengths of their predecessors in their next positions, with shorter terms suggesting a concerted effort to empty the position for the new appointee. Taking all these dimensions into account, we can better detect the functioning of factionalism behind the scenes.

The dependent variable of career breakthrough is used to test the validity of the scholarly approaches. Thus, the independent variable in our study is the factional factor identified with each respective generation’s scholarly approach, i.e. the factional affiliations between the prefectural Party secretaries and their immediate superiors, the provincial Party secretaries. We exclude the first generation “rumours-have-it” approach from our testing because it is not amenable to systematic application. We test the second generation’s “backgrounds-in-common” approach by following the exemplary model provided by Opper, Nee and Brehm. Subsequently, we examine the third generation’s “practices-of-patronage” approach according to Jiang’s design. Finally, we assess the validity of our suggested improvements using the two new criteria of “within-tenure double promotion” and “within-tenure promotional grooming.” The measurement of “within-tenure double promotion” is straightforward; “within-tenure promotional grooming” involves a more detailed assessment, focusing on the participation of leaders in formal Party school training programmes, specifically, the “Training programme for young and middle-aged leaders” (zhongqingnian lingdao ganbu peixunban 中青年领导干部培训班) for prefectural-level leaders.

In these tests, we control for two additional groups of variables. The first group involves demographic characteristics of the Party secretaries, including their age (plus whether they are overage), education, year of term, previous position and career prospects (as represented by the next position of their predecessors). These are the key factors related to the incentives and resources of these leaders.Footnote ⁴¹ The second group contains the key socio-economic features of the localities administered by the leaders, such as the GDP per capita and GDP growth rates, which significantly influence the performance evaluations of these leaders. To better align with Jiang’s explanatory framework, we also include GDP growth rates as an intermediary variable in our models; this serves as an indicator of leadership performance for each test group. This inclusion implies that factional operations affect the promotion of leaders by manipulating their performance profiles to ensure members cross career thresholds smoothly.

We verify these approaches using data taken from the career profiles of officials and the socio-economic statistics of their localities from 2000 to 2020. We focus on prefecture-level data because county-level data are incomplete, while the prospects of provincial leaders are too diverse to be systematically coded. In addition, the prefecture is neither too low nor too high within the administrative hierarchy, ensuring that the impacts of key factors affecting prefecture-level promotions are neither magnified nor diminished.Footnote ⁴² The selected time period corresponds with the establishment of China’s merit-based personnel management, which fully took shape in late 1990s.Footnote ⁴³ Notably, after the year 2000, both profile and statistical data became reliable and available. The years from 2020 to the present are excluded to avoid the dramatic impact caused by the COVID-19 pandemic. This design allows for a comprehensive verification of the different factional identification approaches used by China scholars. The results are reported in Table 2 for “backgrounds-in-common,” Table 3 for “practices-of-patronage,” and Table 4 for our revisions to “practices-of-patronage.”

Table 2. Testing the Backgrounds-in-Common Criterion

Notes: Clustered standard errors at the prefecture level are reported in the parentheses;

* p < 0.1, ** p < 0.05, *** p < 0.01.

Table 3. Testing the Within-Tenure-Promotion Criterion

Notes: Clustered standard errors at the prefecture level are reported in the parentheses;

* p < 0.1, ** p < 0.05, *** p < 0.01.

Table 4. Testing the Two Improvements to Jiang’s Approach

Notes: Clustered standard errors at the prefecture level are reported in the parentheses;

* p < 0.1, ** p < 0.05, *** p < 0.01

Table 2 presents the effects of the factional factor identified using the “backgrounds-in-common” approach. The results of Models 2–4 clearly indicate that the factional affiliations between prefectural Party secretaries and their respective provincial Party secretaries do not have significant effects on any aspect of their promotions. Moreover, Model 5 implies that the identified factional affiliations actually prolong the waiting time for further promotion. The validity of the second-generation’s approach is thus not supported by the data.

Next, the factional affiliations identified with Jiang’s “practices-of-patronage” approach are tested, with the results displayed in Table 3. As shown in Model 1 of Table 3, consistent with Jiang’s assertions, factional affiliations do motivate Chinese leaders and lead to better performance. However, the results from Models 2, 3, and 4 in Table 3 reveal a surprising trend. The factional affiliations of prefectural Party secretaries, as identified by Jiang’s approach, appear to exert a negative influence on their promotions: their promotions are less successful, take longer to achieve, and they are appointed to positions where superior performance is unlikely to occur. These perplexing results prompted further investigation, leading to one potential explanation: superiors may prefer to keep their followers in their current positions, either to maintain control over the locality or to ensure continued performance.Footnote ⁴⁴

To investigate this possibility further, we introduce two additional models that examine whether the career patterns of superiors influence the promotions of their followers. We use interaction terms to explore the effects of superiors’ career advancements, considering two scenarios: first, their later promotion to the Politburo, and second, their subsequent step-down from key leadership positions (tuiju erxian 退居二线). The results of the tests are presented in Models 6 and 7. These findings support the hypothesis that the future career patterns of the affiliated superiors significantly affect the promotions of their followers.

Nevertheless, the measure proposed by Jiang continues to exert negative influences on the promotion of the prefectural Party secretaries, contradicting the logic of political factionalism. It is also possible, however, that Jiang’s “within-tenure promotion” approach may require additional revisions to ensure its validity. To test whether this is indeed the case, we use a similar research design to interrogate our revisions of Jiang’s approach to examine whether more precisely calibrated criteria can offer a more accurate assessment of factional affiliations.

The test results of this study’s proposed refinement of Jiang’s “practices-of-patronage” approach are reported in Table 4. First, Models 2–5 show that factional affiliations, as identified with the “within-tenure double promotion” criterion, enable officials to cross career thresholds successfully and quickly. Furthermore, these affiliations enable strategic manipulation of performance metrics and vacancy opportunities. Similarly, Models 7–10 confirm that the factional affiliations associated with “within-tenure promotional grooming” also support cadres to effectively cross these thresholds, leading to comparably successful strategies regarding manufactured performance and vacancies.

In sum, both of the modifications to Jiang’s approach effectively identify the factional factor in Chinese politics. In comparing the two patronage tactics, direct promotion and indirect grooming, as suggested by Models 1 and 6, tactic choice may be contingent upon the expected performance of the client Party secretaries. If their patrons expect their client Party secretaries will deliver superior performance with additional support, superiors may prefer the more direct and visible option: “double promotion.” In contrast, for clients who cannot deliver superior performance, patrons are inclined to employ the more indirect and less visible option: promotional grooming. Hence, while factional tactics still require further exploration, the validity of our proposed revisions to Jiang’s “practices-of-patronage” approach has been preliminarily confirmed.

This result can be further elucidated using the concepts of Type I error (a false positive conclusion) and Type II error (a false negative conclusion) in statistical hypothesis testing.Footnote ⁴⁵ When comparing Jiang’s measures to ours, we find that Jiang’s is too permissive, identifying 60.07 per cent of the total transfers of prefectural Party secretaries as “factionally affiliated.” However, in the Chinese government, routine promotions can also occur due to position vacancies or lengthy service, making Jiang’s measure more susceptible to Type II errors. In contrast, our measures are considerably more restrictive, identifying 40.59 per cent and 10.89 per cent of the total transfers as factional affiliated. By tightening the criteria, our measures reduce the likelihood of Type II errors but increase the risk of Type I errors. But based on the results of Tables 3 and 4, our proposal performs better, with a slight increase in Type I errors but significant reduction in Type II errors. Thus, our measure is a better choice for identifying political factions in Chinese politics.

Conclusion: Parsing Chinese Factions and Researching Elite Politics

Scholars of China studies widely acknowledge that understanding Chinese politics necessitates a consideration of the factional factor. If researchers are unable to discern the factional affiliations among Chinese elites, they will struggle to systematically account for factionalism, thus will continue to either omit key variables or commit measurement errors. Searching for an effective way to identify factions has become a perennial and arduous task for China researchers. Yet, the field’s continuous attempts to identify the covert membership and activities of Chinese factions have also contributed towards China studies’ unique focus on understanding elite behaviour more generally.Footnote ⁴⁶

This paper reviews the efforts of three generations of China scholars to identify factional affiliations among political elites, namely: “rumours-have-it,” “backgrounds-in-common” and “practices-of-patronage.” Of these, we found Jiang’s third-generation “practices-of-patronage” approach, “within-tenure promotion,” the most rigorous and accessible. We then proposed two revisions to this method, “double promotion” and “promotion grooming,” and then tested all verifiable approaches with empirical data. The “backgrounds-in-common” approach was not supported by the data, while Jiang’s “within-tenure promotion” criterion was partially supported. In comparison, our proposed improvements to Jiang’s approach performed better and thus can be considered a new option for identifying covert political factions. However, further fieldwork, theory testing and methodological verification are needed. As Melanie Manion wisely suggests, the future of China studies will emerge from the mutual reinforcement of intensive fieldwork and theory-informed hypotheses, along with cross-examinations of creative ideas and rigorous falsifications.Footnote ⁴⁷ In this vein, this study makes the following contributions: first, it reviews and critically evaluates existing approaches for identifying factions in Chinese politics; second, it designs and empirically tests these prevalent approaches; and third, it proposes a more effective measure and demonstrates its validity for future research

Through such efforts, China studies could contribute new, verifiable approaches to the comparative study of political elites. The people factor is particularly salient in the overlapping fields of Chinese politics and elite politics. As this review demonstrates, researchers in China studies have dedicated substantial efforts to uncovering and understanding the factional relationships that align these elites. The field’s accumulated experiences have produced multiple tools and considerable insights into elite behaviour. Being able to share these resources with the field of elite politics presents an encouraging prospect for China studies, which has been criticized for functioning as a consumer rather than a producer of theories and methods in comparative politics.Footnote ⁴⁸ This study could be one small step in that worthwhile and rewarding direction.