The anti-corruption campaign (ACC), launched by President Xi Jinping after the 18th Congress of the Chinese Communist Party (CCP) in 2012, is undoubtedly one of the greatest bureaucratic overhauls in modern Chinese history. With over 4 million officials prosecuted since 2012, the ACC is unprecedented in both its scale and intensity.Footnote 1 Several high-ranking politicians in the politburo, the Party's highest decision-making body whose members traditionally enjoyed immunity from prosecution, have been expelled and indicted for corruption; it has also led to the removal of millions of street- and village-level bureaucrats for more petty wrongdoings. As many scholars have argued, in the past eight years the ACC has profoundly reshaped the political, economic and social landscape in China.
In this paper, we explore the effects of Xi's campaign on the pattern of prefecture-level personnel selection in China. As David Bulman and Kyle Jaros point out, the Chinese regime is constantly faced with a “loyalist–localist” dilemma in subnational leadership appointments.Footnote 2 On the one hand, officials with close ties to the centre tend to implement the regime's policy agenda more loyally but often lack the local knowledge and experience to govern effectively. On the other hand, locally embedded bureaucrats are able to accomplish policy goals more easily because of their familiarity with regional conditions, but their strong personal ties to their assigned locality can erode their loyalty to the centre and make them susceptible to “capture” by local elites. The upper-level authority needs to strike a balance between strengthening control over a locality and promoting effective, locally adaptable governance.
Our paper shows that under the ACC, provincial leaders tend to appoint locally disconnected, unfamiliar officials to govern prefecture-level cities whose ex-leaders have recently been prosecuted for corruption. From the provincial authority's perspective, corruption is a “team sport” which requires the complicity of colleagues and subordinates. Thus, even if local bureaucrats who served under the corrupt official did not explicitly collaborate with their superiors, they often bear the responsibility of turning a blind eye to their leader's misdeeds. Moreover, following the expulsion of a corrupt bureaucrat, the priority of the upper-level government is often to strengthen hierarchical control over the locality, eliminate the official's lingering personal influence and social networks, and reduce the negative socioeconomic impact of chaos and uncertainty that accompanies the downfall of a corrupt local leader. In this context, an outsider appointee without local vested interests is more suited to reasserting top-down authority and stabilizing local governance. Indeed, strong local connections could disqualify a potential candidate from filling that vacancy, because it raises suspicions about whether an official can remain independent from local influence and impartially implement the directions from above. Simply put, provincial Party chiefs can strategically leverage an appointee's lack of local embeddedness to tackle governance challenges in cities “tainted” by recent corruption scandals.
We tested our theory using an original dataset of all Party secretaries in China's 287 prefecture-level cities who served between 2013 and 2020. We collected information about those leaders’ backgrounds, previous work experience, local connections and reasons for their departure (whether they retired, were transferred to another post or were removed because of corruption). During the seven-year period, there were 551 cases of Party secretary turnovers across all prefecture-level cities. We then categorized those leadership turnovers into three types, based on whether those vacancies were caused by the abrupt expulsion or regular departure of previous leaders. If our theory is correct, we should observe that successors to local leaders who were expelled owing to corrupt activities are likely to be more locally detached and have less experience in their assigned locality compared to those who fill the vacancies left by retired or transferred officials.
To test this hypothesis, we performed a series of regression analyses to explore whether newly appointed officials' ties to the assigned locality are correlated with their immediate predecessor's reason for departure. We found that if a former local leader was prosecuted for corruption, his or her successor would be much less likely to have prior connections to the locality. This finding remained robust when we used various ways to measure successors' strength of local connection, including whether their immediate previous position was in the assigned city, whether they had any previous work experience in the locality, and the total duration of their related work experience in the city. These results offer strong evidence that locally connected officials are clearly at a disadvantage in the selection of successors to corrupt officials.
We further dissected the motivation behind the provincial Party chiefs’ anti-local bias in the leadership appointment of corruption-affected cities. We tested three possible explanations for this phenomenon: (1) provincial chiefs’ incentive to visibly signal their loyalty and commitment to the centre's anti-corruption agenda; (2) their need for a responsive and impartial substitute to reassert control and handle the after-effects of the purge; and (3) the lack of available candidates from the local bureaucracy following the crackdown. Based on a series of regression analyses, we argue that the provincial authority's need to strengthen hierarchical control and stability may be the most plausible reason that drives the assignment of non-local cadres to corruption-affected cities. In other words, the provincial leaders’ tendency to avoid appointing locally connected leaders following corruption scandals is not merely a symbolic gesture to signal loyalty or a result of a shortage of suitable local personnel; rather it is a strategic measure to improve the compliance and responsiveness of the local bureaucracy.
Our study makes two contributions to the existing literature on China's bureaucratic politics. First, it provides nuanced insights into how the Chinese regime weighs an official's local ties during political selection. In normal times, when the upper-level authorities’ priorities are economic growth and development, a moderate amount of local experience is often considered as a strength because it implies familiarity with local conditions and shows a candidate's suitability to serve leadership roles in the locality. However, during top-down bureaucratic purges such as the ACC, the centre's key priority is to strengthen vertical control over targeted localities and to enforce local government's compliance with the central political agenda. In this context, an official's local ties are viewed not as an advantage but as a burden that obstructs the effective implementation of the centre's directions. Hence, officials who lack local connections are more likely to be selected to govern those localities where centralized control is prioritized during the campaign.
Second, our study reveals the ad hoc, task-responsive nature of bureaucratic appointments in China. Previous research often views political selection in China as a static process that can be consistently explained by a number of fixed factors, such as education, economic performance, factional affiliation and political connections. Our study, in contrast, shows that bureaucratic selection can be an ad hoc response to specific priorities and challenges at a given moment in time. Under the ACC, provincial Party chiefs adapt their political selection criteria based on a locality's observed history of leadership corruption. If a city's ex-leader is prosecuted for corruption, the provincial Party chief will intentionally install a locally disconnected successor to strengthen authority and control. However, once it becomes less imperative for provincial leaders to address the centre's anti-corruption agenda, they no longer discount local connections in the subsequent appointment of sub-provincial cadres. We hope to show that during a top-down purge like the ACC, personnel management, rather than following a one-size-fits-all procedure, is situational and responsive to particular circumstances.
Bureaucratic Management and Corruption in China
Loyalty and expertise in China's bureaucratic selection
China, like most modern regimes, relies on a large, complex bureaucracy to implement its policy objectives. From the perspective of political leadership, the two most desirable qualities in a bureaucrat are loyalty and competence: a bureaucrat must be faithful and responsive to a leader's directions but must also possess the capability and skills to effectively implement assigned tasks. However, a body of literature on bureaucratic selection shows that the loyalty and competence of bureaucrats sometimes constitute a trade-off. Highly skilled bureaucrats who possess superior knowledge and expertise tend to be less obedient to the principal's rules and expectations. This is because they are more likely to leverage their skills to pursue their private interests rather than follow the regime's agenda.Footnote 3 Meanwhile, mediocre bureaucrats tend to display a high degree of loyalty to the political leadership, because their lack of skills makes their career more dependent on the mercy of the principal.Footnote 4 The challenge facing any political leadership in bureaucratic selection, therefore, is to strike a proper balance between the two qualities.
Scholars have explored the selection and promotion of bureaucrats in modern-day China. Most agree that during the reform era, with economic growth being the key objective, a main criterion for the promotion of officials is competence. In order to move up the bureaucratic ladder, officials must out-perform their peers in their ability to promote economic growth and meet development targets. The positive correlation between economic performance and the likelihood of promotion is well supported by various analyses of political turnovers at all levels of government.Footnote 5 Political loyalty and connections, however, also play an important role in China's bureaucratic promotion. Studies show that officials who are connected to high-ranking political elites usually have a better chance of being promoted to higher positions.Footnote 6 Some also find that as an official moves further up the ranks, chances of promotion are less correlated with performance and depend more on factional allegiance.Footnote 7
Local leadership management in China: the loyalist–localist balance
In comparison with the rich literature on vertical bureaucratic promotion in China, few studies pay specific attention to the mechanism of leadership appointment at the sub-provincial level or address the question of what type of official is considered suitable to serve as a Party or government chief in townships, counties and prefecture-level cities. One unique factor often taken into consideration in sub-provincial leadership selection is a candidate's local embeddedness: the strength of his or her socioeconomic ties and connections to the assigned locality.
A major challenge facing the Chinese regime in the management of local bureaucrats, as Bulman and Jaros point out, is the so-called “loyalist–localist dilemma.”Footnote 8 Simply put, an official's local embeddedness can bring both harm and benefit to the regime. On the one hand, local connections can erode an official's political loyalty to the principal, because they can make an official susceptible to nepotism and local influence and lead to non-compliance with higher-level government directives.Footnote 9 Historically, a millennium-old political tradition from the Ming Dynasty prohibited mandarins from serving in their native province in order to avoid collusion between bureaucrats and the local gentry.Footnote 10 In line with this tradition, the post-1978 Chinese regime developed two formal rules to prevent bureaucrats from being captured by local interests, “avoidance” (huibi 回避) and “rotation” (jiaoliu 交流). Typically, local officials at the township, county and prefecture level are prohibited from assuming leadership roles in their place of birth or upbringing, and they must be reassigned to a different post in another locality after a certain period of service.Footnote 11 Recent studies show that owing to these institutional constraints, only a small fraction of prefecture city chiefs are native to their assigned jurisdiction and the average length of their tenure is usually less than five years.Footnote 12
Native knowledge and experience can also increase a bureaucrat's competence, allowing him or her to govern the locality with greater access to information. By installing officials who are already familiar with local conditions, the regime can better penetrate and control local communities, obtain accurate information about public opinions and attitudes, and achieve policy goals more easily through informal means and practices. Studies have also shown that locally connected officials tend to provide better public goods and services, perform better in maintaining social stability and make more investments in health and education.Footnote 13
In the first decade of reform (1978–1988), localism was on a rising trend in China's sub-national politics, driven by Beijing's need for local knowledge and experience to promote economic recovery and growth.Footnote 14 A 1988 survey shows that 43.8 per cent of provincial Party secretaries and governors were native to their assigned province.Footnote 15 Between 1990 and 2013, although the centre made increasing efforts to promote the cross-regional rotation of cadres, the fraction of provincial leaders who were native to the region of their posting was still as high as 41.8 per cent.Footnote 16 Below the provincial level, the rule prohibiting the appointment of native cadres was often circumvented. For example, studies of local politics in Shanxi and Yunnan in the early 2000s find that although no county-level leaders were native to the specific county in which they served, all of them were native to the prefecture above their assigned county.Footnote 17 The presence of a large fraction of locally embedded officials between 1978 and the 2000s, scholars argue, reflected the centre's encouragement of reform initiatives and policy experimentation at the local level, which was considered a key factor in China's economic success.Footnote 18
Corruption in China: from tacit permissiveness to Xi's great crackdown
Corruption has been an enduring problem in China since the beginning of the reform. It can take a variety of forms, ranging from explicit embezzlement and bribery to rent-seeking in land auctions and favouritism in public project bids.Footnote 19 It is estimated that at least half of China's cadre force in the early 2000s engaged in some form of corruption.Footnote 20 Before Xi's rise to power, the Chinese regime largely adopted a passive attitude towards corruption, seeing it as an inevitable by-product of China's growth or an ordinary part of political life.Footnote 21 As a result, the anti-graft efforts made by previous Party leaderships were symbolic and largely insufficient, and failed to contain the rampant corruption at all levels of government.Footnote 22 The regime's tacit tolerance of corruption before the Xi era, as argued by some scholars, was actually a rational choice under the loyalty–competence trade-off. Corruption could drive officials to seek more business deals with private firms that would produce personal benefits as well as promote growth. Thus, although rent-seeking behaviour could undermine the Party's reputation and political authority, it also offered the space for competent bureaucrats to deliver better economic outcomes.Footnote 23
Anti-corruption efforts before the Xi era, some scholars argue, were frequently politicized and weaponized by politicians to attack rival factions or strengthen their own. For example, Jiangnan Zhu and Dong Zhang show that between 1996 and 2012, inter-factional competition among provincial-level leaders could drive up anti-corruption investigations in a province by up to 20 per cent.Footnote 24 On the other hand, an official's connection to powerful leaders could offer protection from corruption charges. A recent study of all anti-corruption cases from 2000 to 2014 shows that a provincial leader's personal ties with the incumbent Party general secretary could reduce his or her probability of being convicted by roughly 70 per cent.Footnote 25
The ongoing anti-corruption crackdown, launched by Xi Jinping in December 2012 shortly after his inauguration as the paramount leader at the 18th Party Congress, marks a remarkable paradigm shift from his predecessors. Early analyses often dismiss Xi's campaign simply as a cynical manoeuvre to crush rival factions and strengthen his personal authority.Footnote 26 The subsequent unfolding of the campaign, however, has largely contradicted those views. Officials who have been prosecuted for corruption come from diverse political backgrounds, and factional ties have not affected the likelihood of being investigated.Footnote 27 Additionally, political connections to high-ranking politburo members have not offered officials any protection from prosecution, which implies that the crackdown is not merely a politically motivated attack on Xi's rivals.Footnote 28 Overall, most scholars agree that Xi's campaign seems to be a systematic attempt to discipline and control the Party's cadre force and preserve the regime's legitimacy in the eyes of the public, rather than a Machiavellian strategy to achieve short-term, personal political goals.Footnote 29
Observers also predicted that Xi's crackdown on corruption would be limited in scope as the abrupt expulsion of too many officials could hurt the economy and undermine the morale of bureaucrats.Footnote 30 However, the campaign has actually evolved into one of the most overarching purges within the Chinese bureaucracy since the reform began. At the top of the political hierarchy, prosecutors have indicted former politburo standing member Zhou Yongkang 周永康 along with three other ex-politburo members, breaking a long-held unwritten rule that retired top Party leaders are usually immune from prosecution.Footnote 31 Large-scale prosecutions have occurred across all levels of the bureaucracy. Official statistics show that 4.09 million cadres were prosecuted for corruption between 2013 and 2021, including 392 officials at or above the provincial level, 17,000 at the county level, and over 616,000 at the township level.Footnote 32 Xi's anti-corruption purge signifies a new norm of bureaucratic management in his “new era” – there is now less focus on officials’ economic performance and more focus on their political loyalty and subordination.
One question that attracts increasing scholarly attention is how Xi's anti-corruption drive has impacted the logic of bureaucratic appointment in China. Zeren Li and Melanie Manion, for example, find that Xi's purge has led to a remarkable decline in factionalism among local leadership appointments.Footnote 33 Their analysis shows that provincial Party bosses tend to avoid promoting prefecture Party chiefs who are members of their own faction. The bias of provincial leaders against their co-faction members, the authors argue, is a “risk reduction” strategy to signal their aversion to factionalism and their compliance with Party discipline.Footnote 34
Subnational Leadership Selection during the Anti-Corruption Crackdown
Our study builds upon this strand of literature to further explore the impact of the ACC on sub-provincial political selection. Xi's purge has led to a mass expulsion of corrupt local leaders; however, we have limited knowledge about the upper-level authority's approach to replacing them. What type of bureaucrat is likely to be appointed to fill the vacancies left by expelled officials? Is there a divergence in the backgrounds of local officials who govern cities with and without corruption?
We propose an incentive-based theory to explain how a candidate's local embeddedness is weighed in sub-provincial leadership appointments during the campaign. Our key theoretical prediction is that the provincial authorities have a selective anti-local bias towards cities tainted by corruption scandals. We suggest that if a city's ex-leader is prosecuted for corruption, the provincial Party chief will strategically appoint a successor who has few local ties to govern the city. In contrast, for cities whose leaders have not recently been indicted for corruption, the provincial authority will not necessarily discriminate against locally connected candidates in subsequent leadership appointments.
We argue that provincial Party chiefs’ selective anti-local bias for appointments in cities with corrupt predecessors could be owing to three reasons: virtual signalling, hierarchical control and a shortage in local suitable candidates. First, the ACC is a highly politicized crackdown that is being used by Beijing to test an official's loyalty. Thus, provincial leaders may need to credibly signal their alignment with the centre's anti-corruption agenda by visibly distancing themselves from their corrupt subordinates. Second, the removal of corrupt leaders may lead to chaos and uncertainty within the local bureaucracy and affect the socioeconomic stability of the city. Thus, a key priority of the provincial chief following a corruption scandal is to reshuffle the local bureaucracy, eliminate the corrupt officials’ personal influence, and restore order and stability in the locality. An appointee with little connection to the local political establishment, therefore, could impartially carry out those tasks and help the provincial leadership reassert hierarchical control. Lastly, the purge of a corrupt city leader may simultaneously bring down a number of high-ranking officials in the local authority, which could lead to a shortage of potential local replacements. As such, the provincial authority may resort to non-local officials to fulfil the leadership vacancy.
Politicization of corruption as an indicator of disloyalty
A defining feature of Xi's anti-corruption campaign, which differs from similar efforts made by his predecessors, is the politicization and securitization of corrupt behaviour. Instead of seeing the corruption of Party officials as an economic crime or a violation of state law, Xi views it as a sign of political disloyalty and defiance of the Party line. In his speeches, Xi variously frames corruption as the “biggest threat” or “life-and-death issue” facing the Party.Footnote 35 An editorial published by the Central Commission for Discipline Inspection (CCDI), the CCP's official disciplinary organ, explicitly characterizes corruption as a political crime: “anti-corruption is a severe political struggle … corrupt officials are politically degenerate, have shown disloyalty and dishonesty to the Party, and have undermined the Party's leadership and unity.”Footnote 36
Official announcements on the prosecution of corrupt local officials also characterize their alleged wrongdoings in a highly politicized manner. The majority of official press releases by the CCDI include language that points to the political disloyalty of corrupt officials, such as “losing one's political ideals and convictions” or “failing to stay true to the Party's founding mission.”Footnote 37 Such language implies that officials who engage in corruption not only violate the law but also demonstrate their political disobedience by challenging the Party line.
Centre holds provincial leaders accountable for subordinate corruption
When confronted with the highly politicized climate during the ACC, how do provincial leaders respond to corruption scandals involving prefecture city chiefs within their jurisdiction? According to China's nomenklatura system, provincial Party chiefs are responsible for appointing prefecture-level officials and overseeing their behaviour and performance. From the perspective of provincial leaders, corruption scandals involving those under their supervision, if not properly handled, could point to their own failure to enforce the Party line and thus damage their own political career. Therefore, once a corruption scandal is exposed, provincial Party chiefs have to be seen to be making a strong effort to reshuffle the local bureaucracy and must clearly signal their resolve to support the centre's anti-corruption priorities.
Provincial Party secretaries who fail to control the corruption of subordinate local officials risk damaging their own careers. Yuan Chunqing 袁纯清, the provincial Party secretary of Shanxi, was removed from office in 2014 owing to his negligence in controlling the graft of local officials. While Yuan was in office, more than 11,879 officials in Shanxi were expelled for corruption, including 26 prefecture-level officials and 336 county-level leaders.Footnote 38 Although Yuan did not intentionally shield his corrupt subordinates, he was accused of “lacking discretion and oversight in selecting appropriate local cadres” and was stripped of power. An online People's Daily editorial piece referring to Yuan's removal commented that “President Xi emphasized that improper personnel appointments can bring severe damage to the Party and the state, and Shanxi's leadership set such an example.”Footnote 39
Selective delocalization: virtue signalling, hierarchical control, and cadre availability
How does the provincial leadership appoint local leaders during the ACC? Based on existing qualitative evidence, our main theoretical prediction is that the local embeddedness of a newly appointed local leader is correlated with a locality's recent history of corruption. If a city's ex-leader is removed because of corruption, the leadership vacancy is unlikely to be filled by a successor with local ties; in contrast, if a city's former leader leaves office through regular bureaucratic rotation, the provincial authority will not intentionally disfavour locally connected officials in the appointment of a successor. We therefore propose the following hypothesis:
Hypothesis 1 (Selective Delocalization). During the anti-corruption campaign, successors appointed to replace expelled leaders are more likely to have fewer political connections to their assigned locality, compared to officials whose predecessors departed in a regular manner.
We further propose three explanations for the provincial authority's propensity to appoint non-localized officials to govern cities “tainted” by corrupt predecessors. The first possible reason is virtual signalling – that is, such appointments aim to display a provincial leader's resolve to distance him or herself from corrupt former subordinates and align with the centre's anti-corruption objectives. The second possible motivation is political control. The abrupt dismissal of a local leader creates an urgent need for the provincial authority to strengthen vertical control over the affected locality, eliminate the disgraced official's personal networks and influence, and restore socioeconomic order and stability. A new appointee without local ties would be less constrained by the personal pressure exerted by local political elites and would therefore be more capable of stabilizing the local political and social order. The third explanation concerns the availability of suitable local candidates to replace the disgraced leader. If the downfall of a corrupt local leader leads to the collateral expulsion of multiple officials within the same local authority, the shortage of potential local candidates can leave the provincial Party chief with little choice but to appoint an outside candidate to fill the leadership vacancy. We therefore propose the following hypotheses:
Hypothesis 2a (Virtue Signalling). During the campaign, provincial leaders tend to appoint successors without local ties to replace expelled corrupt officials in order to credibly signal their loyalty and commitment to the centre's political agenda.
Hypothesis 2b (Hierarchical Control). During the campaign, provincial leaders tend to appoint successors without local ties to expelled corrupt officials in order to restore top-down control and stability in the affected locality.
Hypothesis 2c (Cadre Availability). During the campaign, provincial leaders tend to appoint successors without local connections to replace expelled corrupt officials because the crackdown has created a shortage of available local cadres.
Part C of the appendix presents an illustrative case study of Huang Lixin 黄莉新, an official from Jiangsu province known as “the professional firefighter” of corruption. Huang was appointed as the Party secretary of Wuxi in late 2011 and was later transferred to Nanjing as the Party chief in 2015, both to fill the power vacuum left by corrupt predecessors. As shown in the case study, Huang's appointment decisions reflect the political logic of both loyalty signalling and political control. Facing upward, provincial Party chiefs could signal their determination to correct past mistakes made in personnel management and reaffirm their loyalty to the Party line. Facing downward, a non-local appointee could be more capable of reshuffling the malfunctioning local bureaucracy and enforcing the upper-level authority's policy agenda.
Research Design
Data collection
To understand how Xi's anti-corruption crackdown on local leaders impacts the pattern of local leadership appointments, we manually constructed a panel dataset of local leaders of all 287 prefecture-level cities between 2013 and 2020. Our dataset includes three categories of information: (1) biographical background, including age, gender, place of origin and length of CCP membership; (2) previous work experience, including length of previous service in the assigned locality and in the provincial government, most recent position before assuming the current leadership role, and the time and reason for departure from the current leadership role; (3) biographical information about the provincial Party secretary overseeing the city official, including name, age, gender, length of CCP membership, hometown, local work experience and personal connection with the local official. Additionally, we extracted prefecture-level socioeconomic data from China City Statistical Yearbooks from 2013 to 2020, including GDP, population, land area, revenue income and unemployment numbers in order to control for the confounding effect of economic performance on personnel appointment.
Main variables of interest
The main objective of our empirical strategy is to explore whether officials who are installed to replace a disgraced predecessor are less connected to the assigned city compared to those who fill the roles of retired or transferred officials. Thus, the main unit of our analysis is leadership successions occurring between 2013 and 2020. Based on our dataset, we count an event of leadership succession in year t if the Party secretaries of a given city i in year t and t – 1 are two different individuals.Footnote 40 Using this formula, we identified 551 instances of Party secretary replacement between 2013 and 2020. Table 1 displays a breakdown of the backgrounds of outgoing and incoming local leaders involved in those leadership successions.
Table 1: Summary Statistics of 551 Prefecture City Leadership Turnovers
Dependent variable: local embeddedness of incoming officials
The dependent variable of our analysis is an incoming official's strength of local connection during the leadership replacement. Our main analysis uses two dummy variables indicating the incoming official's immediate previous position before the turnover: whether the official is promoted from an incumbent leadership position in the same city;Footnote 41 and whether an official is “sent down” from a position at an upper-level government authority (including both provincial and central authorities).Footnote 42 The first measure aims to capture successors’ personal ties with the political establishment in the locality. The second measure, on the contrary, reflects officials’ detachment from the assigned locality: officials who are directly sent down from upper-level authorities are less tied to local political elites, and their appointment would raise fewer concerns about collusion and disobedience.
In subsequent robustness checks, we used two additional measures to capture an official's local ties on a longer time horizon: a dummy variable to capture whether the official has any work experience in the assigned locality, and a continuous variable indicating the total number of years the official has served in the locality. The two variables, which account for an official's overall local experience, provide alternative measures for an official's connection with the assigned locality.
Key independent variable: departure pattern of outgoing officials
The key independent variable in our analysis is the predecessor's reason for departure during each leadership replacement. In the main analysis, we code an outgoing official's departure reason as a dummy variable based on whether he or she was expelled from office on corruption changes. Alternatively, in robustness checks, we further break down an outgoing official's departure pattern into three categories: corruption charges, transfer to a different post or retirement from public service.Footnote 43 If our theoretical assumption is correct, then we should observe that incoming officials who replace corrupt officials are more likely to be disconnected from the locality compared to those incoming officials who follow regular bureaucratic rotations.
Control variables
In accordance with the established literature on political selection, we also include three sets of control variables that are potentially correlated with both our outcome variable and key independent variable. The first group of controls concern the local socioeconomic conditions of the prefecture city that could impact the selection of local leadership, including its population size, land area, per capita GDP, GDP growth rate, amount of revenue income, unemployment rate and percentage of public servants in the population. The second group includes other personal characteristics of the outgoing official, including age, gender, length of Party membership, years of work experience and years of previous work in the locality. The third category includes the characteristics of the provincial Party secretary in charge of appointing the incoming local leader, including age, gender, length of Party membership, length of local work experience, length of central government service and whether he or she was convicted on corruption charges later on.Footnote 44 Furthermore, we also control for the provincial Party chief's political connection with the outgoing leader, using whether the outgoing leader was directly appointed by the provincial Party leader as an indicator. If, as argued by Li and Manion, provincial chiefs tend to have an “anti-faction bias” during the ACC, they would have greater incentives to install locally disconnected successors to signal their impartiality if they were personally connected to the corrupt former official.Footnote 45
Treatment–control group comparison
We begin our analysis by comparing the personal characteristics of successors to expelled and regularly departed officials. In Table A2 of the online appendix, we present a series of two-sample t tests to show whether any substantive difference exists between the two types of successors. Apparently, the two groups are very similar in their average age, gender composition, length of Party membership and length of total work experience (with p-values all greater than 0.05). The only substantive difference is the length of previous work in the assigned locality (p = 0.0003). Successors to expelled corrupt officials on average only have 0.465 years of prior local experience, implying that a majority of them may have never worked in their assigned locality before. In contrast, successors to retired or transferred officials have spent almost two years serving in the locality before their assignment. This comparison suggests that corrupt local leaders are more likely to be replaced by officials who have no or few previous ties to the locality.
Regression Estimates
Baseline analysis
In our regression analysis, we employ a series of linear probability models (LPM) to estimate the impact of predecessors’ departure type on the local embeddedness of successors. Our baseline analysis, presented in Table 2, evaluates the effect of a predecessor's departure pattern on two main outcome variables: whether the successor is promoted from the same city's leadership team; and whether the successor is sent down from an upper-level political authority. Models 1 and 3 use the whole sample (all 551 leadership turnovers) for estimation, while Models 2 and 4 only include leadership turnovers that occurred in cities with at least one anti-corruption purge between 2013 and 2020 (n = 84; hereafter “corruption-affected cities”). The purpose of this subgroup analysis is to mitigate the concern of endogeneity. It is possible that certain types of cities are likely to be specifically targeted in the anti-corruption crackdown, and those cities can also be subject to different leadership appointment patterns. Therefore, by testing the correlation between the predecessor's departure pattern and successor's local connection within the particular subgroup of corruption-affected cities, we can present stronger evidence that the removal of corrupt predecessors indeed affects the local embeddedness of incoming appointees to the same city. Furthermore, in line with established norms for panel data analysis, we include provincial and year fixed effects to mitigate between-group variations, and cluster all standard errors on the prefecture city level to account for heterogeneity.
Table 2: Baseline Analysis
Note: For a complete list of control variables and the full regression results, please see Table A4 in the appendix. Only completed observations are used in the regression. Significance levels: ∗p < 0.1; ∗∗p < 0.05; ∗∗∗p < 0.01.
The main regression results are consistent with the theoretical predictions of Hypothesis 1. As shown in Models 1 and 2, replacements for corrupt officials are much less likely to be chosen from among incumbent officials from the same locality, compared to appointees in regular leadership turnovers (p < 0.001). Furthermore, as shown in Models 3 and 4, successors to expelled officials are also more likely to be sent down from upper-level authorities, which implies an effort to assert hierarchical control. The main results provide support for our theory that the ACC does indeed lead to a “selective delocalization” effect: provincial Party chiefs tend to appoint outside officials to govern those cities whose former leaders were disciplined for corruption, while regular bureaucratic rotations do not lead to any significant displacement of localized officials. In Appendix Tables A5–A7, we perform a series of robustness tests to validate our main finding, by changing the model specifications by including different sets of controls. As shown in the results, the key findings remain robust and consistent: regardless of how we specify the model, the predecessor's pattern of departure has always been a strong predictor of the successor's local embeddedness.
In Appendix Tables A8 and A9, we provide further support for Hypothesis 1 by using two alternative outcome variables that capture an incoming official's local connection on a longer time horizon: whether he or she has any previous work experience in the assigned locality and the total number of years served in the locality.Footnote 46 The results are consistent with our main findings: replacements for corrupt officials are much more likely to have no or little prior experience serving in the assigned locality, which suggests that local ties are strongly disfavoured in the selection of leaders for corruption-affected cities.
Analysis of competing explanatory mechanisms
In Table 3, we present a series of mechanism analyses to explore which of the three potential reasons proposed above could better explain provincial leaders’ propensity to appoint non-local leaders to govern corruption-affected cities. Models 2 and 3 examine Hypothesis 2a – that the anti-local bias is primarily driven by virtue signalling. If provincial authorities install non-local replacements because they want to visibly display their commitment to fighting corruption, then provincial chiefs who are considered more likely to shield their guilty subordinates would have a stronger incentive to appoint outsiders in order to dispel any suspicions that may be cast upon themselves. To test this logic, we interact our key independent variable with two indicators that imply a provincial chief's potential culpability by association: (1) whether the predecessor is appointed by the provincial leader, and (2) the length of the provincial leader's local work experience. In other words, provincial chiefs who are more locally embedded or who have personally appointed the corrupt city leader are likely to make stronger efforts to credibly distance themselves from their corrupt subordinate.
Table 3: Test of Competing Mechanisms for Selective Delocalization
Note: For a complete list of control variables and the full regression results, please see Table A10 in the appendix. Only completed observations are used in the regression. Significance levels: * p < 0.1; **p < 0.05; ***p < 0.01.
Models 4 and 5 focus on Hypothesis 2b, which states that provincial leaders’ anti-local bias when appointing a new leader for a corruption-affected city is driven by the need to reassert hierarchical control and order following the abrupt removal of the corrupt ex-leader. If this conjecture is true, then provincial chiefs would be less likely to appoint non-local successors to cities where the provincial authority has stronger pre-existing control or cities that report better socioeconomic outcomes. For the former, we interact our explanatory variable with a dummy that equals one if the city is either a provincial capital or a sub-provincial city (fu shengji shi 副省级市). Such “key” cities are major political and economic centres of a province and usually come under greater provincial oversight and control. For the latter, we interact our explanatory variable with a city's per capita GDP in the previous year in order to examine whether provincial chiefs would feel less need to install outsider appointees in cities that display more robust socioeconomic outcomes.
Lastly, Model 6 examines the possibility that provincial authorities appoint non-local officials to govern corruption-affected cities because there is a shortage of available local cadres. We further break down the key independent variable (predecessor expelled) into two sub-categories of whether the city Party secretary is the only official who has been expelled for corruption, or if one or more members of the city's Party Standing Committee (PSC) have been simultaneously removed alongside the corrupt Party chief.Footnote 47 In the local power hierarchy, the PSC is a group of high-ranking officials immediately below the Party secretary who are seen as potential candidates to fill the leadership role. Thus, if Hypothesis 2c is indeed correct, then provincial authorities will be more likely to appoint non-local successors to cities that suffer from the collateral loss of one or more PSC members, which makes the selection of a local replacement more difficult.
Table 3 illustrates the relative strengths of the three competing explanations. As the results for Models 2 and 3 have shown, the statistical evidence in favour of Hypothesis 2a is weak: neither the provincial chief's connection to the outgoing city leader nor his or her local work experience increases the likelihood of appointing a non-local official to govern a city if its former leader has been expelled on corruption charges (p > 0.1). This negative result suggests that virtue signalling may not be the main reason that drives the anti-local bias.
On the contrary, Models 4 and 5 show strong evidence for Hypothesis 2b. As the interaction coefficient of Model 4 shows, although leadership corruption in non-major cities leads to a significantly higher likelihood of a non-local successor being appointed, it has little effect on the successor type for provincial capitals and sub-provincial cities (with the joint effect close to zero). This result apparently shows that asserting control is a more plausible explanation than virtual signalling. Intuitively, if the provincial leader's objective is virtue signalling, then major cities should have displayed a stronger anti-local bias when appointing successors because they have greater salience and visibility; however, if the reason for appointing non-local leaders is to strengthen top-down control and order, the provincial chief would install non-local officials in less important localities where hierarchical control and oversight are weaker than they are in major urban centres. Model 5 lends further support for the latter mechanism. As the interaction coefficient has shown, cities with better socioeconomic outcomes are less likely to be assigned non-local officials to replace corrupt ex-leaders, which implies that the need to exert control and improve governance is the driving force behind non-local appointments.
With regard to Hypothesis 2c, we do not find sufficient evidence that the tendency to appoint non-local officials is driven by a shortage of local candidates. Among the 40 corruption-related expulsions, only a small fraction (eight cases) resulted in the simultaneous purge of one or more local PSC members, which suggests that the impact of local leadership expulsion on the pool of local candidates is limited. Additionally, as shown in Model 6, cities where only the Party secretary is removed are still more likely to be assigned a non-local successor (p < 0.001), which indicates that the lack of available local cadres is not a very plausible reason for non-local appointments.Footnote 48
Lastly, Table A13 in the online appendix presents additional evidence that the observed effect is driven by the strategic decisions of the provincial leadership, rather than the intervention of the central government. In Table A13, we exclude from our data all leadership appointments in which (1) the successor is transferred from a different province or (2) the successor is a former central official sent down to the locality. In both scenarios, the appointment decisions are more likely to be made by the centre than by the respective provincial authorities. As shown in the results, the observed effect still remains robust when the impact of the central intervention is removed.
The Durability of Post-crackdown De-localization
We have shown that the exposure of a corruption scandal in a locality can prompt the provincial authority to appoint non-local officials to strengthen oversight and reassert control, but how durable is this effect? Does the expulsion of corrupt local leaders only affect the selection of their immediate replacement, or does it have a long-lasting effect on the leadership appointment pattern in the affected locality? Despite the limited time scope of our dataset, we are able to offer preliminary insights into this question by examining the local ties of all subsequent leaders following the expulsion of corrupt officials.
We use two approaches to test the durability of the anti-local bias in leadership appointments in corruption-affected cities. First, we regress an incoming official's local embeddedness on a dummy variable indicating whether any former leader since 2013 had been prosecuted before being appointed to the assigned city. As shown in Table A12 in the appendix, the effect of predecessor corruption on an appointee's local ties is no longer significant if we take into account both immediate replacements of corrupt officials and all subsequent appointees afterwards. In other words, although the provincial authorities tend to appoint non-local officials to replace corrupt officials, we find no evidence that there is any bias against locally connected candidates for later leadership appointments in the same city.
Furthermore, we focus on the 84 leadership turnovers in 39 corruption-affected cities and make three-way comparisons between (1) officials appointed before the anti-corruption expulsion, (2) those who immediately succeed the expelled official, and (3) those who are appointed in subsequent turnovers. We perform a series of pairwise t tests of difference and present the results in Table A15 in the online appendix. Clearly, city leaders appointed in subsequent leadership turnovers have stronger local ties than immediate successors to corrupt officials and show no statistical difference from those appointed prior to the anti-corruption purge. This result suggests that the tendency to install non-localized appointees is an ad-hoc strategy to respond to challenges arising from the abrupt removal of corrupt officials and does not necessarily bring a permanent or long-lasting effect on the further political selection patterns in corruption-affected localities. Figure A3 in the appendix visually captures the “rebound” of locally connected appointees after the purge-induced turnovers. Furthermore, in Table A16, we provide some preliminary insights into the short duration of the de-localization effect following anti-corruption purges.
Conclusion: “Task-responsive” Political Selection
A key objective of President Xi Jinping's administration is to consolidate central power and strengthen the Party's authority. Further, to ensure officials’ exclusive allegiance to the centre, it is necessary to separate them from their embedded personal networks. As previous studies have shown, bureaucrats are appointed to localities where they have few social ties and little work experience so that they can implement central policies more effectively without interference from the local elite.Footnote 49 In this article, we provide a more nuanced look into how de-localization facilitates political control and policy implementation in the context of Xi's anti-corruption campaign. We show that the de-localization of sub-provincial leadership is not indiscriminate but rather selectively targets specific local authorities that show signs of political disobedience or governance challenges. Provincial Party chiefs tend to appoint non-local, external candidates to “rein in” prefecture-level cities whose ex-leaders have been prosecuted for corruption. Free from personal networks and vested interests that tie their hands, outside officials are able to take tough measures to reshuffle the local bureaucracy, implement their superiors’ directives and restore control and stability in their assigned locality.
Another important implication of our study is the declining effect of factionalism and informal politics during the anti-corruption campaign. As argued by Jiangnan Zhu and Dong Zhang, anti-corruption crackdowns before the Xi era were often leveraged by local leaders to weaken their factional rivals, while a superior official with stronger ties in a locality would make less effort in enforcing anti-corruption measures.Footnote 50 In contrast, our analysis finds that a provincial Party chief's political connection with the outgoing local official has no effect on the Party chief's appointment decisions during anti-corruption purges in either direction. Specifically, provincial leaders who are connected to corrupt local officials are neither more likely to appoint a candidate from the incumbent local bureaucracy to shield their wrongdoings, nor will they explicitly install a non-local replacement to distance themselves from their corrupt subordinates. This pattern reflects the tendency to downplay the role of factions in bureaucratic management in the Xi era. As recent studies have suggested, the Chinese regime under Xi not only opposes the forming of factionsFootnote 51 but also discourages using factions as a factor in personnel management.Footnote 52 For example, Kainan Gao and Margaret Pearson find that a corrupt official's political network has no impact on either the length of investigation or the severity of punishment in the Xi era.Footnote 53 Even officials connected to Zhou Yongkang, Xi's chief rival, did not experience drastically different treatment during anti-corruption investigations. Thus, our study provides further evidence of the changing pattern of bureaucratic control in Xi's China.
Last but not least, we hope that our study can prompt a re-thinking of the logic of political selection in China. Previous scholars often treat China's bureaucratic appointment as a standardized, one-size-fits-all process where officials are uniformly evaluated against a number of fixed criteria such as education,Footnote 54 performanceFootnote 55 or political connection.Footnote 56 In line with a bourgeoning literature,Footnote 57 we challenge this premise and show that personnel appointment patterns are contingent upon the regime's needs to address certain short-term challenges and crises. We show that during a top-down purge like the ongoing anti-corruption campaign, local leadership appointments can be leveraged by upper-level authorities as an ad-hoc response to problems and crises in local governance, and selection criteria can be adapted to specific tasks and priorities varying in time and location. In this sense, we hope our work can renew the debate on the logic of political control in Xi's China.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S0305741023000644.
Acknowledgements
We would like to thank Yuhua Wang, Xiaobo Lü, Zeren Li, Mai Hassan, Xuedong Yang, Yu-sung Su, Kyle Jaros, Xiran Liu, Jiaqi (Jackie) Lu, Jing Qian, Erico Dong Yu and the participants at the Second Chinese Politics Research in Progress (CPRP) Workshop for their invaluable comments. Additionally, we also want to thank Zhenze Huang, Yike Gao, Jiaqi Li, Xiaochuan Fan, Junxiao Fei, Yifei Liu and several other undergraduates at Tsinghua University for their excellent research assistance and data collection work. All errors are the authors’ own. Funding for this research project is partially covered by the Scott A. Harris Graduate Writing Scholarship from the political science department at the University of Wisconsin–Madison.
Competing interests
None.
Jingyuan QIAN is a PhD candidate in the political science department at the University of Wisconsin–Madison.
Feng TANG is an assistant professor (jiangshi) in the political science department at the University of International Business and Economics, Beijing, China. The two authors contributed equally to the manuscript.
Open access
