How do political leaders rise to positions of power in China? Students of Chinese politics have presented two important models to explain elite mobility: the factional model and the performance model. The factional model proposes that personal ties to dominant leaders are the keys to success in the careers of political leaders. In the factional model, dominant leaders at the top lead various factions through personal exchange relations with their followers. The survival of factions depends on the capacity of leaders to provide rewards to their followers, and factional struggles centre on replacing rival factions' members with sympathetic members.Footnote 1 As Pye stated, “the real motivation [of joining a faction] is that of career security and enhancement.”Footnote 2 On the other hand, the performance model emphasizes the achievement of elite members as the primary determinant of their career advancement.
Students of Chinese politics disagree about which model is dominant in elite politics of the post-Deng era. Some point to the prevalence of factionalism. For instance, elite politics in Hu Jintao 胡锦涛's era has been seen as factional rivalry between Hu's faction and that of Jiang Zemin 江泽民. Hu's faction has been known to be based on the personal network he developed while serving on the secretariat of the Chinese Communist Youth League (CCYL). Jiang's faction has been known to be based on his associates in the Shanghai region. Numerous studies have explained the rise and fall of the political elite in Hu's era as outcomes of competition between the two factions.Footnote 3 While prior research has suggested the existence of factionalism in Hu's era, the question as to what extent factionalism remains prevalent is still unanswered. Teiwes argued for the limited utility of the factionalism concept in explaining Chinese politics during the reform period, because objective criteria such as age, qualifications and tenure were crucial for determining career patterns of political leaders.Footnote 4 Previous studies on factionalism, as well as Teiwes's study, have not provided a systematic test to evaluate the degree to which factional ties affect political mobility.
Others emphasize the prevalence of the performance model during the reform period. The Chinese Communist Party (CCP) appears to depend on economic performance to provide legitimacy during the reform period. According to this position, the CCP encourages economic development by linking the political mobility of its elite members to their performance in economic development. Li and Zhou found that between 1979 and 1995, provincial leaders who achieved higher GDP growth rates were more likely to be promoted and less likely to be terminated.Footnote 5 Bo found that between 1978 and 1998, revenue contributions to the central coffer were important in determining the promotion and demotion of provincial leaders. Based on this finding, Bo further asserted that the performance approach “challenges conventional models of factionalism.”Footnote 6 The finding that performance affects political mobility, however, is not sufficient to challenge the factional model. Performance and patronage are not necessarily mutually exclusive criteria for determining political mobility; it is possible that both are important but that they carry different weight at different levels of leadership.
While existing studies on the Chinese elite have provided some evidence for both the factional and performance models during the reform period, several questions remain unanswered. First, is one model dominant over the other in shaping elite politics in the post-Deng era? Alternatively, are they both relevant? Second, if performance matters for elite mobility, does this apply to both factional members and non-members to the same extent? Or, does performance affect career trajectories only for non-factional members but not for factional members? Third, if members of Jiang's faction enjoyed advantages in promotion in Jiang's era, does the influence of his faction still persist in Hu's era?
This study attempts to answer these questions by analysing a person-year dataset of career mobility of provincial Party secretaries and governors for inclusive years 1989 to 2009, which covers both Jiang's and Hu's eras. It should be noted that the dynamics of elite politics can be different at various echelons of power. Chinese political elites can be categorized into five levels of power: top, upper-middle, lower-middle, upper-lower and lowest levels. At the top, around 25 leaders command the highest authority, taking the positions of Politburo and high-ranking positions in the Party, government, parliament, and military.Footnote 7 The upper-middle level leaders consist of ministry-level leaders including provincial Party secretaries and governors. Within this echelon of power, the position of provincial Party secretaries is higher than that of governors. Party secretaries and governors at the prefectural-level city are at the lower-middle level. Those at the county and township levels belong to the upper-lower and lowest levels, respectively. Borrowing from the organizational theory of Mintzberg, Lee suggests that elite politics of the CCP at the top echelon remains informal, even after elite politics at the lower level has become formalized.Footnote 8 The career trajectories of officials at the middle and low levels are more likely to be determined by objective criteria such as performance than those of leaders at the top.
This study analyses career trajectories of provincial leaders. Because the provincial leadership position is one level below the top leadership position and because some succeed in taking a top leadership position, provincial leaders are likely to be involved in factional politics. At the same time, the performance model is applicable to provincial leaders. With the economic decentralization during the reform period, provincial leaders have had considerable autonomy in economic decision making in their jurisdiction.
This study found that the dynamics of elite mobility were somewhat different between provincial Party secretaries and governors. For provincial Party secretaries whose promotion meant taking a top leadership position, both performance and patronage mattered for their promotion. On the other hand, for provincial governors for whom promotion meant taking a ministry-level position, only performance determined their chances of promotion. Among provincial Party secretaries, the extent to which performance affected the likelihood of promotion did not vary between factional members and non-members. This suggests that even factional members needed to show good performance to enhance the likelihood of their promotion. The extent to which members of Jiang's faction enjoyed advantages in being promoted during Hu's era was not different from that in Jiang's era. This suggests that the power of Jiang's faction in promoting their members to a top leadership position did not decline even after Jiang's retirement from the position of general secretary of the CCP in late 2002.
Conceptual Framework
We construct ideal types of political systems depending on whether performance and/or factionalism affect elite mobility. The upper right cell in Table 1 is the performance-dominant type system in which political mobility is determined by performance but not by factionalism. In this type, top leaders have little influence over the career advancement of their favoured followers. The lower left cell is the factionalism-dominant type system in which career mobility is determined by factionalism but not by performance. The upper left cell is a mixed type in which both factionalism and performance matter for political mobility. The lower right cell belongs to other types in which neither factionalism nor performance affect political mobility. For instance, a political system in which seniority is the main criterion for promotion of the political elite belongs to this type.
The Types of Political Systems for Elite Mobility
In the mixed type where both factionalism and performance are relevant for elite mobility, we can distinguish two sub-types, depending on how factionalism and performance interact. In the dual-track type, the paths to power are separated between the patronage path and the performance path; some rise to power solely by their loyalty to a dominant leader, whereas others climb the ladder entirely through good performance. In other words, performance does not affect political mobility for factional members, whereas it does for non-members; that is, those who take the path of patronage can disregard their performance for political mobility. On the other hand, in the cross-track type, although factional members have advantages in career mobility, their promotion also depends on performance. In other words, in the cross-track type, the effect of performance on political mobility is the same for factional members and non-members. Distinguishing the dual-track type from the cross-track type is useful because these two sub-types have different implications for elite cohesion. Dual paths to power are likely to undermine cohesion among the elite. On the other hand, in the cross-track type, as performance criteria are also applied to factional members, differences in qualification among the elite are less distinct compared with the dual-track type. The cross-track type comes into being if powerful factional leaders can agree upon broad criteria of performance for political mobility.
To which type of system does China belong? It may depend on the echelon of power being examined. Whereas studies of Chinese elite applying the factional model tend to focus on political leaders at the top, those applying the performance model are inclined to focus on political leaders at the middle and low levels.Footnote 9 Studies report that the cadre evaluation system, which specifies performance criteria for evaluation, is applied to township leaders.Footnote 10 County-level leaders are also subject to such types of cadre evaluation. In 1995, the Organization Department of CCP released a “Circular on strengthening and improving the cadre evaluation of the Party and government leaders at the county (city) level” (Guanyu jiaqiang he wanshan xian (shi) dangwei, zhengfu lingdao banzi gongzuo shiji kaohe de tongzhi 关于加强和完善县(市)党委, 政府领导班子工作实绩考核的通知).Footnote 11 It specified criteria for cadre evaluation at the county level, among which economic development was an important part. To this author's knowledge, no equivalent documents for evaluating the performance of provincial leaders have been identified. Nonetheless, it is plausible to hypothesize that economic performance affects the career trajectories of provincial leaders. China has had the one-level down personnel management system since 1984, meaning that the centre's direct authority over personnel management extends only to provincial leaders. Under this system, the centre might need to employ economic development as a criterion for evaluating provincial leaders so that they have incentives to lead sub-provincial leaders to work for economic development.
Others argue that factionalism is relevant in Chinese elite politics, particularly at the top level. The Leninist one-party system officially condemns factionalism in favour of unity. While this does not eliminate factionalism, it gives rise to a specific type of factionalism: what Beller and Belloni call “elite factionalism.” Unlike open factionalism in which factions are organized extensively covering the whole organization, elite factionalism is confined to client group followings among elite at the high echelon of power.Footnote 12 The one-party system provides fertile ground for factionalism because the banning of competition among political parties may lead politicians to form factions to enhance their power.Footnote 13 Moreover, the nomenklatura system in which top leaders have authority over appointment provides room for exerting influence over the career trajectories of elites. These characteristics of the Leninist party system remain intact in the reform era. Furthermore, the absence of supreme leaders in the post-Deng era is likely to intensify factionalism. As Tsou rightly points out, supreme leaders such as Mao Zedong and Deng Xiaoping were able to exert influence on decision-making on their own, relatively free from factional politics.Footnote 14 Lacking charismatic authority, current top leaders are likely to have stronger incentives to cultivate clientelist ties to secure their power base.
Data and Measurement
This study conducted statistical analyses by incorporating the following variables into the model: factional ties and performance. Including these two variables in one model enabled us to evaluate whether the effect of factional ties held once the performance variables were controlled for. The unit of analysis was a person-year, covering years from 1989 to 2009 (21 years). Statistical analyses were conducted separately for provincial Party secretaries and governors. The number of cases was 643 for provincial Party secretaries and governors, respectively.Footnote 15 The data on the career patterns of provincial leaders were drawn from several compilations of bibliographies published in Beijing, Hong Kong, and Japan.Footnote 16 Various statistical yearbooks were referenced to examine the performance of provincial leaders.Footnote 17
Since China does not officially acknowledge the existence of factions, it is not easy to identify factional affiliation of political elite. The conventional way is to identify whether a political leader has former associations with a factional leader. Since factions are based upon personal relationships, it is assumed that those who have been in acquaintance from the past are likely to feel comfortable with each other. Based upon this convention, Li provides a systematic guideline for identifying Jiang's and Hu's faction. This study relies on his extensive study of Chinese elite to determine the factional affiliation of provincial leaders.Footnote 18 The members of Jiang's faction were those who had connections with Jiang when he served in the First Ministry of Machine Building Industry and in Shanghai as mayor and Party secretary. The latter group is so-called the “Shanghai faction.” The basis of the Shanghai faction is not whether they were born in Shanghai but whether they advanced careers in Shanghai through their association with Jiang.Footnote 19 Those who experienced career advancement in Shanghai but had close relationships with Zhu Rongji 朱镕基 were not included as Jiang's faction.Footnote 20 The members of Hu's faction were connected with Hu while he served on the secretariat of the CCYL. Specifically, those who worked at the national or provincial levels of the CCYL for some time between 1982 and 1985 when Hu served on the CCYL secretariat were included as Hu's faction. Those who worked at the lower level of CCYL in the early 1980s were not regarded as Hu's faction.Footnote 21 Using these criteria for identifying factional affiliation, we found that only Li Yuanchao 李源潮 belonged to both factions.
Aside from these factions, this study examined whether family connections contributed to the career advancement of provincial leaders. The term “princelings” refers to the children or sons/daughters-in-law of the high-ranking elite.Footnote 22 The high-ranking elite included veteran revolutionaries, military officers, and those who served at the level of vice minister and above.Footnote 23 This study did not consider princelings as a faction since a princeling group did not fit into a narrow definition of factions that this study employed, that is, clientelist ties between a leader and followers. As Xi Jinping 习近平, son of Xi Zhongxun 习仲勋, former vice premier of the State Council, is rising as a strong candidate for the general secretary of the CCP, it would be interesting to examine whether princelings enjoy advantages in career advancement in a systematic way.
Figure 1 shows the proportions of Jiang's faction, Hu's faction, CCYL membership, and princelings out of total provincial Party secretaries by year. The proportion of Jiang's faction reached 10 per cent in 2002 when Jiang retired from the position of General Secretary of the CCP. The proportion of Jiang's faction remained undiminished until 2007, but then it declined to 3 per cent in 2008. The proportion of Hu's faction jumped from 10 per cent in 2002 to 26 per cent in 2006. One might argue that since the CCYL is an institution nurturing leaders of the next generation of CCP, it is not surprising to see a high proportion of Hu's faction in the Chinese elite. To examine whether this is the case, Figure 1 includes CCYL membership. All members of Hu's faction held CCYL membership, but not vice versa. The proportion of CCYL membership fluctuated sharply; it recorded as low as three per cent in 1995. This suggests that CCYL members are not necessarily accessed for the position of provincial Party secretaries.Footnote 24 The fluctuation in the proportions of the factional affiliation among provincial Party secretaries appears to confirm the factional model. Jiang placed his followers in provincial Party secretary positions before he retired. Once Hu became the general secretary of the CCP in the late 2002, he successfully advanced his followers to provincial Party secretary positions. Another interesting feature is that the proportion of princelings increased from zero in the 1990s to 16 per cent in 2007.Footnote 25
Proportions of Provincial Party Secretaries Belonging to Jiang Zemin's Faction, Hu Jintao's Faction, CCYL Membership, and Princelings
Figure 2 shows the proportions of Jiang's faction, Hu's faction, CCYL members and princelings out of total provincial governors by year. The proportion of Jiang's faction remained very small among provincial governors. On the other hand, the proportion of Hu's faction increased from 10 per cent in 2002 to 32 per cent in 2008. As in the case for the position of provincial Party secretaries, it is not natural for CCYL members to take many seats in provincial governor positions. For instance, the proportion of CCYL members out of total provincial governors in the mid-1990s was merely 7 per cent. It reached as high as 58 per cent in 2008. This proportion was even much higher than that of Hu's faction (32 per cent) in that year. Li noted that many CCYL members who had associates with Li Keqiang 李克强, a strong candidate for the position of Premier, took positions as provincial governors.Footnote 26
Proportions of Provincial Governors Belonging to Jiang Zemin's Faction, Hu Jintao's Faction, CCYL Membership, and Princelings
This study classified the political mobility of provincial leaders into four categories: promotion, status quo, retirement, and demotion.Footnote 27 Promotion referred to advancement in rank. For provincial Party secretaries, their promotion means taking a national leadership position, such as becoming a member of the Politburo. For provincial governors, their promotion includes moving to a provincial Party secretary position and to a ministry-level position in the central government. The status quo referred to the maintenance of the same rank. It includes both the maintenance of the same position as a provincial Party secretary or a governor within one province and the lateral transfer to a position at the same rank in another province or in the central government. Those who reached age above 64 and did not keep the position without promotion or lateral transfer were recorded as retirement. Demotion referred to taking a position of lower rank before age of 64.
Table 2 shows the frequency of promotion, status quo, retirement, and demotion events among provincial Party secretaries and governors from 1989 to 2009. Within the category of status quo, we distinguished lateral transfer from keeping the position. Keeping the position was a predominant occurrence. If provincial Party secretaries reached the end of their term without serious fault and they were not yet at retirement age, the most likely occurrence was lateral transfer (5.9 per cent) and the second most likely was promotion (4 per cent). Since a provincial governor position is lower than a Party secretary position, governors had a higher chance for promotion (9.5 per cent). Lateral transfer was also a likely event for provincial governors (7 per cent). Who, then, got promoted? This is the question that this paper aims to investigate.
The Frequency of Promotion, the Status Quo, Retirement, and Demotion of Provincial Party Secretaries and Governors (1989–2009)
Previous studies showed that economic performance, such as GDP growth rates and fiscal revenue growth rates, affected the career trajectories of local officials.Footnote 28 Based upon these findings, this study used GDP growth rates and fiscal revenue growth rates to measure economic performance. These indicators were adjusted for inflation. This study used the average of economic performance during provincial leaders' tenure.Footnote 29 For example, the second-year performance was measured by the average of the first and second years, the third-year performance was measured by the average of the first, second and third years, and so forth. For the years with leadership changes, those who were in their position for more than half of the year were credited with the economic record.
Several variables were included in the model to control for their effects. First, the Age variable indicated the age of provincial leaders at the time of their service. Because of the retirement system, the likelihood of older provincial leaders being retired was expected to be greater. Second, the Tenure variable indicated the number of years that provincial leaders served in their position. Third, the Municipality variable indicated whether a provincial leader served in a provincial-level municipality such as Beijing.Footnote 30 Fourth, year dummy variables were included to control for time effects. Fifth, the variable of Hu's era, which was included only in model 3, is a dummy variable, recording years from 1989 to 2002 as zero and those from 2003 to 2009 as one. Table 3 shows descriptive statistics of these variables.
Descriptive Statistics of Variables
Statistical Analyses
Multinomial logit analyses were conducted because there were four dependent variable categories: promotion, status quo, retirement, and demotion. Logit models for multinomial categories pair each category with a baseline category. This study used the status quo as a baseline category. This paper reported statistical results that showed the chance for promotion compared with the status quo.Footnote 31 The variables indicating performance – GDP growth rates and fiscal revenue growth rates – were highly correlated. Thus, instead of including both variables in the same model, the models were estimated separately. Statistical outcomes, including the variable of GDP growth rates, turned out to be consistent with those including fiscal revenue growth rates. Since GDP growth rates showed stronger associations with the chance for promotion, this paper reports only models with GDP growth rates.Footnote 32
Table 4 reported the estimated effects of factional ties and performance on the chance for promotion of provincial Party secretaries. Model 1 included variables of factional affiliation and GDP growth rates. If GDP growth rates had been the same, the members of Jiang's faction and Hu's faction would have been more likely to be promoted (vs. maintaining their status quo) than non-members. If factional affiliation had been the same, higher GDP growth rates would have increased the likelihood of promotion of provincial Party secretaries. In order to examine whether CCYL membership per se, rather than Hu's faction, affected the chance for promotion, we estimated a model including CCYL membership. It turned out that CCYL membership had tenuous associations with promotion.Footnote 33
Multinomial Logit Analysis on the Chance for Promotion of Provincial Party Secretaries Compared with that for the Status Quo
Note:
Year indicators are included in Model 1 and Model 2 but not shown.
* denotes p < 0.05, **p < 0.01, and ***p < 0.001
Model 2 included the variables of Age and Tenure for controlling their effects. Once the Age variable was included, the variable of Hu's faction became no longer significant. This suggests that members of Hu's faction tended to be younger. After controlling for the effects of age and tenure, the variables of Jiang's faction and GDP growth rates remained significant. Princelings and non-princelings showed no difference in their chance for promotion.
The coefficients in the logit model did not allow for an intuitive understanding of their meanings. Thus, the coefficients were transformed into odds ratios through exponentiation. The odds of an event's occurrence are represented by the ratio of the number of times the event is expected to occur to the number of times it is expected not to occur. For example, an odds ratio of four indicates that the event would be four times more likely to occur than not. In model 2, the members of Jiang's faction were 5.3 times more likely to be promoted, compared with non-members. If GDP growth rates increase by one per cent, provincial Party secretaries were 1.9 times more likely to be promoted (vs. maintaining the status quo).
Model 3 in Table 5 examined whether the effects of performance on the chance for promotion depended on factional ties. The interaction variables of factional affiliation and GDP growth rates were not significant. This indicated that the effects of GDP performance on the chance for promotion were the same between factional members and non-members. Model 4 examined whether the impact of Jiang's faction on promotion declined in Hu's era. The interaction variable of Hu's era and Jiang's faction was not significant. This suggested that the impact of Jiang's faction on promotion did not change in Hu's era, compared with Jiang's era.
Multinomial Logit Analysis on the Chance for Promotion of Provincial Party Secretaries Compared with that for the Status Quo
Note: Year indicators are included in Model 3 but not shown.
* denotes p < 0.05, **p < 0.01, and ***p < 0.001
Whereas Jiang's faction was consistently associated with the enhancement in the likelihood of promotion in all models, the impact of Hu's faction was significant only in Model 1 and Model 4. The volatility of impact of Hu's faction is in part due to the fact that Hu's era was still underway. As discussed earlier, many members of Hu's faction assumed provincial leadership positions in the last few years before the study. As the average tenure of provincial leaders is approximately three years, it is still too early to fully evaluate the effect of Hu's faction on the likelihood of promotion.
Table 6 reports the effect of factionalism and performance on the chance for promotion of provincial governors (vs. the status quo). Factional affiliation did not affect the chance for promotion of provincial governors. Provincial governors who achieved higher GDP growth rates were more likely to be promoted. The younger governors were more likely to be promoted. The longer one served in one province, the more likely one was to be promoted.
Multinomial Logit Analysis on the Chance for Promotion of Provincial Governors Compared with that for the Status Quo
Note: Year indicators are included but not shown.
* denotes p < 0.05, **p < 0.01, and ***p < 0.001
Conclusion
We laid out a typology of political systems in the section of conceptual framework: the factionalism dominant type, the performance dominant type, and the mixed type in which both factionalism and performance matter for political mobility. To which type does China belong? It depends on the echelon of power. For provincial Party secretaries whose promotion means rising to a top national position, the mixed type appears to be relevant. On the other hand, the performance dominant type appears to be applied to provincial governors. This suggests that factional struggles are intense at the top elite level but not as intense at the level below.
Within the mixed model, to which sub-types (dual-track type and cross-track type) does the mobility pattern of provincial Party secretaries belong? Statistical analysis showed that the effects of performance on promotion were the same for both factional members and non-members. Since statistical analysis only allows us to gauge effects on average, we cannot exclude the possibility that a few people rose to the top position purely due to patronage. On average, however, performance mattered for factional members for promotion to the same extent that it did for non-members. This suggests that the mobility pattern of provincial Party secretaries has some elements of the cross-track type, even if it may not perfectly fit into this type.
Top leaders in China appear to maintain a delicate balance between the performance-based system and the faction-ridden system. The current system appears to be the optimal solution for major factional leaders in their efforts to build a solid power base through patronage and to garner legitimacy through performance. Considering these vested interests of major factional leaders, the current system is likely to be maintained. However, the system may change if there is a breakdown in the current consensus among major factional leaders over the broad criteria of performance for political mobility. In such a case, the factional model would be dominant. The likelihood that the performance system would be dominant at the top echelon of power appears to be slim. As long as China lacks a mechanism to assure smooth political succession, it is likely that top leaders will seek to consolidate their power base through factions, and thus, they will not be willing to surrender their power and influence over appointing their followers to powerful positions.
