Abstract
Governments at times combat corruption intensively in an attempt to (re)gain political trust. While corruption crackdowns may demonstrate government resolve to fight corruption, the high-profile corruption uncovered may also shock the public. Therefore, how effective can anticorruption policies help boost political trust? We argue that anticorruption policies influence political trust through two channels: direct experience, that is, interactions with governmental bodies, and the media, that is, second-hand information culled from reporting on anticorruption. Differentiating between these two channels illustrates that anticorruption policies may have distinct effects on political trust for different social groups. We contextualize our theoretical framework with the latest anticorruption drive in China, combining longitudinal data from a national survey and field interviews and using difference-in-differences (DID) models. Our findings support our predictions. For state-system insiders (e.g., civil servants), increase of political trust is less pronounced than for outsiders because the former directly experience radical implementation processes and ineffective anticorruption outcomes. Similarly, political trust increases at a lower rate for groups with higher levels of education and greater access to information outside governmental propaganda than for their less-informed counterparts. Intensive anticorruption efforts are therefore more likely to increase political trust for the grassroots than for elites in China.
Introduction
Corruption erodes the quality of government (Rothstein and Teorell 2008). It reduces the efficiency and effectiveness of public services (Rose-Ackerman 1999), inflates government spending (C. Liu, Moldogaziev, and Mikesell 2017), undermines the rule of law (Tanzi 1998), undercuts political equality (Warren 2004), and reveals officials’ “ethical deficit” (Villoria et al. 2013, 86). Thus, corruption has the broader sociopolitical consequence of decreasing trust in government, an increasingly common concern. Declining trust ultimately undermines the legitimacy of the ruling regimes (Anderson and Tverdova 2003; Seligson 2002). Accordingly, governments go to great lengths to regain trust and increase public support through intensive anticorruption endeavors. In contrast to routine institutional measures to prevent and contain corruption (Klitgaard 1988; Perry and Hondeghem 2008), these kinds of anticorruption initiatives generally feature temporary escalations of enforcement, driven by a strong political will and resulting in a high number of officials arrested or otherwise held accountable for a brief period of time (Wedeman 2005). This pattern is common in countries where corruption is rampant. For example, in Brazil, incumbent presidents often initiate anticorruption policies to boost their public support in forthcoming elections (Avis, Ferraz, and Finan 2018). The Institutional Revolutionary Party (PRI) in Mexico regularly vows to fight corruption to maintain its ruling status (Morris and Klesner 2010). In China, the Chinese Communist Party (CCP) periodically investigates corruption intensively to showcase its commitment to rule of law and win public support (Manion 2004).
However, the effect of these intensive anticorruption efforts on political trust is not well understood. Enforcement initiatives may demonstrate a government’s resolve to combat corruption, but the high levels of corruption uncovered by a crackdown may also shock the public. Therefore, how effective is temporary intensive anticorruption enforcement at restoring political trust? To answer this question, we propose an analytical framework connecting macro-level policies with political trust at the individual level. We argue that anticorruption policies mainly influence political trust through two channels: direct experience, which refers to first-hand encounters with policy implementation or interactions with government agencies or officials; and the media, in which people rely on second-hand reports from different sources on anticorruption enforcement, such as its policies and outcomes. Accounting for these two channels highlights how governmental anticorruption efforts affect distinct social groups differently, depending on the prominence of each channel in mediating each social groups’ relationship to government agencies, officials, or policies.
We tested our theory in the context of the latest anticorruption drive in China. We adopted a mixed-methods approach, combining field interviews and quantitative strategies, using longitudinal public opinion data from the Chinese Family Panel Survey, conducted in 2012 and 2014 across Chinese cities, together with a self-compiled data set of local anticorruption endeavors measured by the number of mid-level officials arrested in each city annually. To test the impact of intensive anticorruption efforts on political trust in a municipal government, we used a difference-in-differences (DID) model. Overall, our results showed that anticorruption efforts have a positive effect on political trust. However, more importantly, we found that the positive effects vary by social group and that this variation depends on the main channel by which each group perceives anticorruption policies, through direct experience or the media. Trust was less likely to increase for state-system insiders, such as civil servants, or for individuals who interact often with government insiders (e.g., businessmen), than for state-system outsiders. Insiders’ direct experiences with radical implementation processes and sometimes ineffective anticorruption outcomes generate mixed opinions about the policies. Furthermore, those who learn about anticorruption efforts from official reports are more likely to increase their political trust because their information is filtered through government propaganda (Shirk 2011). However, this positive effect was limited for more-informed citizens, such as those with higher education levels or who used the Internet frequently. By 2014, anticorruption enforcement in China had a less significant effect on political trust for elites (i.e., state-system insiders and the better-informed) than for the grassroots (i.e., state-system outsiders and the less informed).
Our findings advance the literature on policies’ impact on political trust. By introducing an analytical framework, we show how policies at the macro-level can influence political trust at the individual level. We identify distinct channels by which public policy influences political trust and disaggregate society into social groups based on their exposure to different channels. By focusing our analysis on local government, we also contribute to the emerging literature on anticorruption efforts’ effects on political trust, which tends to primarily lodge trust at the national government (e.g., Ji and Meng 2017; Y. Wang and Dickson 2017; Zhu, Huang, and Zhang 2019). Political trust at the local level is also important to study because citizens are more likely to interact with local officials than at the national level. Furthermore, research has found that Chinese people who distrust local government tend to have lower overall levels of support for the regime, despite generally high levels of trust in the central government (Chen 2017). Finally, we alert decision-makers to the importance of proper policy implementation for maintaining political trust.
Anticorruption Enforcement and Political Trust
Political trust is a basic, evaluative orientation toward the government based on “how well the government is operating according to people’s normative expectations” (K. Hetherington 1998, 791). It is crucial for legal compliance and political stability (Chanley, Rudolph, and Rahn 2000). Low political trust reflects dissatisfaction with the government and may lead to social unrest (Paige 1971). Even in authoritarian states, higher political trust is important to regimes because they help governments achieve political tasks (Lianjiang Li 2013). Scholars also find that higher political trust often coexists with more government support (Dickson 2016; Tang 2016). People’s specific support to government is especially a cumulative outcome of a government’s performance, delivered by politicians and more easily influenced by government policies (Tang 2016). To gain specific support and trust, politicians have an incentive to use beneficial social policies that show good performance (De Mesquita et al. 2005).
Therefore, governments, especially those in countries with high levels of corruption, sometimes use intensive anticorruption enforcement to improve their performance and boost political trust. Rather than simply relying on popular institutional measures, such as reforming civil servants’ salary levels, improving government transparency (Cordis and Warren 2014), or screening and meritocratic recruiting (Gans-Morse et al. 2018; Maor 2004), which generally require good institutional environments and longer term efforts, political leaders may exploit their political mandate to launch new policy initiatives to clean up corruption in intensive, short-term bursts. Such harsh crackdowns are often meant to signal the government’s anticorruption commitment (Steves and Rousso 2003). For example, after assuming office, it is common for the Mexican President to spur intensive anticorruption campaigns to bolster political support (Morris and Klesner 2010). Leaders in a number of African countries, such as Kenya and Nigeria, similarly issue anticorruption policies after being elected, with an explicit aim to drum up political support (Adelopo and Rufai 2018). In India, Prime Minister Modi’s sudden and disruptive demonetization program was part of a broader anticorruption reform to expose corrupt officials (Vyas and Wu 2018).
However, societal responses to government anticorruption policies can be mixed. For example, Adelopo and Rufai (2018) found that Nigerian President Muhammadu Buhari’s anticorruption initiatives were both lauded and criticized. In India, while nationwide demonetization earned Modi political points, critics remained cautious about the broader anticorruption efforts (Vyas and Wu 2018). Zhu, Huang, and Zhang (2019) also found that Chinese netizens’ online comments about the recent anticorruption enforcement in China reflected conflicting opinions. In all of these cases, while some people generally consider government policies as a positive action to control corruption, others may question the government’s true intent and the specific measures to combat corruption, such as the policy design, implementation procedure, and effectiveness of reducing corruption. In addition, intensive anticorruption enforcement exposes corruption at all levels of government, which could have the unintentional effect of increasing public awareness of corruption and decreasing trust. For instance, Ni and Sun (2015) using data of one province in China found that more anticorruption efforts did not reduce the preexisting high corruption perception of the local residents. Steves and Rousso (2003) even found a significantly positive correlation between anticorruption activities and perceptions of corruption in transitional post-Soviet countries. Research has also shown that higher levels of perceived corruption reduce political trust, especially fiduciary trust in the public sector (Anderson and Tverdova 2003; Thomas 1998): corruption is seen as a betrayal of people’s normative expectations of public administration (K. Hetherington 1998). Given people’s mixed responses, the specific effects of temporary, intensive anticorruption enforcement on political trust remain an open question that requires theoretical and empirical analysis.
An Analytic Framework: Direct Experience versus Media
We argue that to systematically examine anticorruption enforcement’s effects on political trust, it is necessary to identify the channels through which large-scale anticorruption policies affect political trust at the individual level. Based on extant research on the influence of government policies, such as economy and health insurance program, on public opinions (M. J. Hetherington 1996; Lerman and McCabe 2017; Mutz 1992), we infer two channels by which individuals are exposed to anticorruption policies: direct experience and the media. For policies associated with immediate and tangible effects, citizens are more likely to evaluate governmental policies based on personal experiences. For instance, policies conducive to economic growth may improve public assessment of government performance and boost political trust through tangible indicators of improved economic well-being (M. J. Hetherington and Rudolph 2008; Z. Wang 2006). In contrast, policies beyond ordinary people’s direct experiences and whose outcomes are not directly apparent to the general public, such as space programs (Steinberg 1987), second-hand information, such as media reports, are critical in shaping public evaluations of government performance (M. J. Hetherington 1996). We therefore propose that anticorruption policies influence political trust through both direct experience and the media, but that each channel will have distinct impacts on different social groups and in different contexts.
Direct experience with anticorruption policies refers to the interpersonal channels connecting individuals’ political trust with government performance, such as policy implementation processes and outcomes. Trust generated through direct experience involves personal interactions with government agencies and officials (Thomas 1998). Positive interactions can serve as the basis for high levels of political trust, even for those skeptical of the government (Christensen and Lægreid 2005). Like e-government and education reform (Lü 2014; Tolbert and Mossberger 2006), good anticorruption policies can increase political trust when people experience higher government efficiency or responsiveness after a policy is launched. Moreover, because anticorruption policies mainly target government officials, those who work within the government or who have the opportunity to interact often with government officials are more likely to be influenced by direct experiences than those with few personal interactions with government.
Sentiments toward the government are also shaped by reports of anticorruption enforcement in the mass media, including both mainstream (e.g., newspapers, television) and social media. Although most people are influenced to some degree by media reports of anticorruption, the media are likely to be especially influential for citizens who lack opportunities to directly interact with government officials. Through agenda-setting and issue-framing, the media affect how people react to government policies, such as how much attention they pay to anticorruption efforts, how the success or failure of those efforts is evaluated, and, ultimately, their level of political trust (M. J. Hetherington 1996). In competitive media environments with high levels of freedom of speech, different media sources may provide competing framing of the policy, leading to more diversified opinions (Chong and Druckman 2007); influential criticism may even drag down a government’s overall levels of popular support. However, in government-controlled media environments, such as authoritarian countries, governments leverage state propaganda to induce people to “imagine” the achievements of anticorruption policies to mobilize and retain popular support, as is the case with other types of policy (Stockmann 2013). In the following, we contextualize our theory by examining the most recent anticorruption drive in China.
Recent Intensive Anticorruption Enforcement in China
Corruption in China has intensified in the reform era. To combat corruption, the Chinese government has embraced an anticorruption mandate that includes regular watchdog actions, such as audits, and also periodic enforcement crackdowns that generally follow from policy initiatives. These crackdowns typically involve “several short bursts of intensive enforcement” initiated by leaders at the highest levels, who call for “a major escalation” of publicity and use “harsh rhetoric” to condemn corruption and “strong words” to emphasize the importance of government integrity (Manion 2004, 86; Wedeman 2005, 93). Scholars sometimes describe the crackdowns as anticorruption campaigns or campaign-style anticorruption enforcement. With extraordinary political mobilization, political sponsorship, specific policy target, and public pressure (N. N. Liu et al. 2015), intensive anticorruption enforcement efforts avoid complex bureaucratic obstacles and can be implemented more rapidly than standard approaches (Zhu, Huang, and Zhang 2019). To claim success, the government points to statistics showing “a dramatic increase in the number of cadres arrested and convicted” as an indicator of anticorruption strength and effectiveness (Wedeman 2005, 93).
The most recent anticorruption policies were initiated by President Xi Jinping, who assumed the office of CCP Secretary General in mid-November 2012 and the Presidency of the state in March 2013. He began by calling for a fight against minor misconduct, such as extravagance or waste through lavish banquets in the end of December 2012 (Gong and Xiao 2017), followed by more intensive crackdowns targeting serious offenses, such as bribery and embezzlement of public funds from early 2013. The central government described its policy as bringing down both “flies” (i.e., low-level corrupt officials) and “tigers” (i.e., corrupt officials at or above the vice-provincial/ministerial levels). Starting from March 2013, many high-ranking officials, including several national-level leaders, were arrested and prosecuted for corruption, breaking the implicit rule of elite protection. Whereas some expected the policies to cease quickly, as previous campaigns had, the intensity has been maintained for years, with a high level of priority, rather than a short burst of intensive enforcements. These resulted in a crackdown much larger than all previous campaigns in the post-Mao era. Figure 1 shows a significant rise in the number of officials arrested for corruption after 2012. The CCP explicitly claims to boost political trust through these anticorruption endeavors. As Xi Jinping himself stressed, “All Party members must keep high mental vigilance and gain more public support for the Party by new progress in the anticorruption struggle” (Xi 2016).

Number of officials investigated and disciplined by anticorruption agencies (in thousands).
Moreover, in an authoritarian regime, where a single ruling party dominates the state hierarchy from the top down, the tight political environment can influence policy implementation at the local level. Although the central government initiated the recent anticorruption crackdown and the CCP’s Central Discipline Inspection Committee (CDIC) led the efforts, provincial, prefecture, and lower levels of government were all mobilized to conduct their own “local wars” against corruption. Local governments were encouraged to experiment with various anticorruption initiatives (Gong 2015). Central inspection teams went to the provinces to uncover problems and supervise local implementations. The CDIC also strengthened its control over lower level Disciplinary Inspection Committees (DICs) through personnel reforms, thus making DICs from the provincial to the county level more obedient to central command (Li et al. 2017). Provinces and prefectures actively investigated and arrested more local officials than usual as a response to the intense atmosphere (Zeng 2017).
Distinct Support across Different Social Groups in China
We argue that the intensive anticorruption policies in China affect political trust through both the direct experience and the media, and at distinct levels for different social groups. Direct experience with the large-scale anticorruption drive is likely to significantly affect people’s interactions with government agencies or officials. The pragmatics of policy implementation and real reductions in corruption are likely to influence how these individuals assess government performance. In fact, many of the interviewees in our study mentioned judging the effectiveness of local anticorruption endeavors by the pragmatics of governmental operations, such as whether government is more efficient, whether officials will only serve people after receiving bribes, and whether leaders continue to organize lavish banquets and drink expensive wines. Therefore, direct experience is likely to be an important channel linking anticorruption enforcement to political trust.
Moreover, we propose that “state-system insiders,” that is, those who work inside the government or who interact often and closely with government officials, such as party members, cadres, businessmen, and state-owned enterprise (SOE) employees, are more likely to be influenced by direct experience and therefore likely to respond to anticorruption policies differently than state-system outsiders. Anticorruption policies in China are made to solve internal governmental problems (Biddulph, Cooney, and Zhu 2012). Those who work within the government and ruling party, such as party cadres and government officials, and those who engage in public services under CCP leadership, such as state organization and enterprise employees, are the primary targets of the policies. These state-system insiders experience and observe the implementation process of the anticorruption policies more closely than outsiders (Tsai and Xu 2018). Political pressure during intensive enforcement generally outweighs formal, legal procedures; repressive actions may be taken to demonstrate government efficacy (Tanner 2000).
The civil servants who we interviewed raised concerns that local governments may have overreacted to the central government’s mandates and implemented the policies too radically. 1 For example, fearing being accused of wasting state resources or violating party discipline, many state organizations canceled overseas visits or training for their staff, and cracked down on modest nonvital spending, such as small presents during public holidays. Government departments also strictly restructured their offices to adhere to new office-size regulations, despite the fact that this restructuring resulted in wasted expenditure and space. Officials also became conservative when embarking on new projects to avoid making mistakes or being charged with corruption. 2
State-system insiders have direct access to the outcomes of anticorruption policies and therefore know more about actual reductions to corruption than outsiders. Some interviewees reported observing obvious improvements, whereas others complained that corruption was getting worse. For example, a businessman suggested that the situation worsened after the government began its corruption crackdown, saying, Yes, high-ranking and junior officials alike have become wary about soliciting bribes, but the anticorruption effort did not stop them from doing so. They would instead ask for bigger benefits and favors factoring in the higher risks that are involved.
3
This observation is also supported by Vyas and Wu’s (2018) interviews with Chinese civil servants, who were skeptical about the effect of anticorruption policies, especially on low-level petty corruption.
In general, state-system insiders’ nuanced personal experiences may lead to reservations about the implementation and outcomes of anticorruption enforcement, which, in turn, influence their support for the policy and trust in the government. One civil servant told us, “I think the general public should have gained more trust in the government through the anticorruption enforcement. However, please note, I am only talking about the general public, not people like us working inside the government.” 4 This comment vividly illustrates the discrepancies between state-system insiders and outsiders. We thus suggest the following:
A government policy as strong as the latest anticorruption wave in China is highly likely to be accompanied by extensive media propaganda. The CCP routinely uses propaganda to promote its policies and garner political support (Kennedy 2009). In addition to suppressing critical voices, propaganda sets the public discourse agenda around negative issues through issue-framing and manipulating political narratives to give the impression that the government is actively responding to popular demands and seriously addressing a problem (Landry and Stockmann 2009). During corruption crackdowns, mainstream media reports are framed primarily in terms of the government’s success in its efforts and its commitment to clean government (Zhu, Lu, and Shi 2013). In our field interviews, many interviewees reported that the official media were their primary channel for learning about the anticorruption work, for example, mentioning that they “learned that anticorruption enforcement has helped control corruption effectively” “from the official news,” “television,” or “newspapers.” 5 Impressions that anticorruption efforts were successful thus appeared to be largely influenced by news reports from the mainstream news media. Skillful agenda-setting may therefore improve public perceptions of governmental responsiveness and efficacy (Tang 2016), ultimately increasing political trust.
However, the communist regime in China may not always succeed in using their propaganda machine to boost trust through the official media, especially for “better-informed” citizens, such as those who are better educated, younger, and who access alternative information via the Internet. Studies have shown that even in authoritarian regimes, education can still increase people’s critical capacities, knowledge, and understanding of politics, and therefore affect their reactions to government policies (e.g., Croke et al. 2016). Although government propaganda is embedded in education in China, especially at primary and secondary schools, those with a higher education level (i.e., above secondary school) tend to have a high level of literacy and greater skill in using the Internet to access a wider source of information. Those with a college education can be more critical of government propaganda, because college education includes more elements of critical thinking and exposure to ideas outside of China (e.g., Tang and Darr 2012). Younger people also tend to receive less orthodox education than the elders. In addition, the commercialization of mass media has greatly diversified the available sources of information in China (Stockmann 2013). The penetration of the Internet further accelerated the dissemination of information that competes with official sources (Huhe, Tang, and Chen 2018; Shirk 2011). During intensive anticorruption enforcements, online rumors often portray the policy as an internal struggle between political elites; stories about arrested officials are also often among the most eye-catching on the Internet and social media channels. Thus, although many Chinese netizens may not be particularly attuned to politics, they, in comparison with non-Internet users, are still exposed to broader political information (Wallis 2011), including online rumors, which may pose a direct challenge to official news reports and possibly eliminate propaganda’s effects among “netizens.” This better-informed group also tends to consider governmental propaganda crude, heavy-handed, and preposterous (Huang 2017). Therefore, we conjecture that through the media, government propaganda about anticorruption policies is more likely to improve sentiments toward the government among less-informed groups, such as the less-educated and the elderly, who are more likely to rely on official media for information. We thus develop the following:
In the following sections, we empirically test the two hypotheses among the different groups.
Empirical Strategy
Data and Methods
We tested our hypotheses by examining variations in subnational levels of political trust in relation to different local anticorruption efforts. Data were collected from two surveys of the China Family Panel Studies (CFPS) in 2012 and 2014. 6 The CFPS, first launched in 2010, is a nationwide representative biennial longitudinal social survey conducted by the Institute of Social Science. The CFPS collects information on individual’s demographic backgrounds, including education levels and work experience, in addition to opinions on social or political issues. The samples include sixteen thousand households and more than thirty thousand respondents from cities in major provinces (Xie and Hu 2014).
Because the anticorruption initiatives were launched at the end of 2012, we used the 2012 and 2014 CFPSs for comparison. We measured the dependent variable, political trust for municipal government, by respondents’ answers to the question, “To what extent do you trust local officials, on a range of 0–10?” 7 The panel data allow us to use a quasi-experimental design, comparing individual responses before and after the anticorruption policies were initiated. We use a DID approach with individual fixed effects to identify the effects of anticorruption enforcement, controlling for any city-level confounding factors that might affect trust.
The treatment, or the independent variable, in the study is city governments’ anticorruption efforts. This is measured by the number of government officials at the bureau, department, or county levels that were arrested between December 2012 and December 2014. This measure follows previous research (e.g., Zhu and Zhang 2017) and is justified on several grounds. First, as mentioned previously, the government itself uses arrests as an indicator for the success of anticorruption endeavors; thus, more arrests can represent more intensive anticorruption efforts (Manion 2004). Second, given the rampant corruption in China, most officials charged are for past activities rather than recent behaviors; hence, the number of arrested officials is not identical to the level of corruption (Wedeman 2012) but reflects more anticorruption intensity. Third, the officials collected here are mid-level officials who have some power locally and especially subject to city governments to monitor. Thus, the arrest data can particularly reflect a city government’s anticorruption efforts. Larger arrests data also presumably manifest stronger propaganda on anticorruption endeavors because governments usually want to publicize their efforts and achievement. We also have a direct measurement of propaganda strength in the robustness check.
We collected the number of officials arrested from Tencent (www.tencent.com), which provides a searchable database of arrests, including officials’ names, former positions, and localities.
In addition, the 2012 CFPS survey was conducted from mid-2012 to the end of the year, except a few individuals interviewed in January 2013 and after. We excluded the respondents surveyed in 2013 from the statistical analysis to avoid any potential contamination of the pretreat sample. As discussed in an earlier section, although President Xi came into power in mid-November 2012, intensive crackdowns began from early 2013. Therefore, December 2012 is a safe cutoff to ensure that the vast majority of respondents not yet received the “treatment.” 8 Equation (1) is the baseline regression model:
where Yijt is the dependent variable, respondent i’s trust in the municipal government j in year t. The independent variable, LnAnti, is the logged number of arrested mid-level government officials (i.e., Num. CrrptOffs) in city j between 2012 and 2014 and addresses the skewed distribution of arrested officials in each city. This variable interacts with a dummy variable for the year to identify the treatment effect.
Hypothesis 1 holds that anticorruption efforts will have different impacts on state-system insiders and state-system outsiders due to their respective levels of direct experience with government officials or agencies. To test this, we identify four main groups that were more likely to be directly affected by anticorruption policies or to interact with officials personally: party members, government/SOE employees, businessmen, and cadres. We generate four dummy variables accordingly. 9 Hypothesis 2 predicts that anticorruption efforts will have different effects on better-informed groups and less-informed groups. To test this, we identify three classic groups of critical citizens: young people, the highly educated, and Internet users. We create three sets of dummy variables accordingly. 10 We include an interaction term between the respective dummy variables and the independent variable in the baseline model to compare changes in political trust among each group and its counterparts. In the following analyses, clustered standard errors at the city level are used to correct for potential heteroscedasticity. The descriptive statistics are reported in Table 1.
Descriptive Statistics.
Empirical Results
We first report results from the baseline model, which tests the general effect of anticorruption efforts on political trust. Column 1 of Table 2 reports estimates with individual and time fixed effects and city-level controls only. Column 2 includes controls that capture the potential importance of individual characteristics that can change over time, such as health and urban residence status. Results show a marginal effect of .110, which means a 100 percent increase in the number of arrested officials (e.g., from two to four) will increase political trust by 0.11 percentage points. Considering that eight is the average number of arrests, 10.77 is the standard deviation, and ninety-two is the maximum number of arrested officials, cities have considerable room to boost political trust by doubling or tripling the number of arrests. We further checked the robustness of our baseline findings. In column 3, we control for the dummy “city leaders,” which equals 1 if a city’s party secretary or mayor was arrested in the previous two years and 0 otherwise, because the party secretary and the mayor are more important and recognizable to the general public; their arrests would usually expose further corruption at lower ranks. In column 4, we drop three cities where a significantly greater number of officials were arrested than the average to mitigate possible outlier effects. The significance of LnAnti was consistent in both models. We also controlled for the level of corruption in a city, as this was likely to be related to the number of officials arrested. Because no direct measure of local corruption is available, we use a dummy variable for whether a city was inspected by the provincial DIC team as a proxy. These inspections were initiated by the provincial government and were therefore exogenous to municipal anticorruption efforts. Furthermore, the inspections were usually prompted by specific reports of significant local corruption; as such, these inspections were a strong signal of relatively high levels of corruption. The regression results controlling for these inspections are reported in column 5. LnAnti remains statistically significant and the magnitude of the estimated coefficient increases. We conducted an additional test to reject the pretrend concern and present the findings in Online Appendix B. In general, the analyses support our measures of local anticorruption efforts and show that stronger anticorruption efforts increase overall levels of political trust in local government in China. 11
Anticorruption Efforts and Political Trust.
Robust standard errors in parentheses.
p < .1. **p < .05. ***p < .01.
Tables 3 and 4 report the results of our hypotheses tests. In column 1 of Table 3, we examine the effect of anticorruption on changes to party members’ levels of trust in the government. The interaction term shows that party members’ trust increased less than nonparty members’ when the number of arrested local officials increased, although the overall anticorruption effect remains significantly positive for this group. Column 2 examines the government and SOE employees. Consistent with Hypothesis 1, the negative and statistically significant coefficient suggests that trust increased less among these groups than among state-system outsiders. The overall effect of anticorruption on political trust for the government and SOE groups is still positive, although only marginally significant. In column 3, we check the effect of anticorruption on cadres, who hold middle-to-top administration positions within the government and SOEs. We found no statistical significance in the interaction term as the number of arrested officials increased. This result raised the question as to whether there was some level of preference falsification, as cadres might be more sensitive to political questions and thus intentionally respond more positively than ordinary employees. We address this concern in the following robustness check section. Column 4 presents findings for businessmen who interact with government officials closely and who were therefore also aware of the policy implementation process and outcomes. Again, the statistically significant negative result confirms our hypothesis that their levels of trust in the government increased less than state-system outsiders as anticorruption efforts became more intensive, although their overall trust nonetheless increased. 12 We visualize the interaction terms between insider groups and local government anticorruption efforts in Figure 2.
State-System Insiders’ Response to Anticorruption Enforcement.
Robust standard errors in parentheses. The “overall LnAnti effect on identified group” means the effect of anticorruption on social groups identified in each column. From models 1 to 4, it is the effect of anticorruption on party members, government and SOE employees, cadres, and businessman, respectively. Some lower order terms, such as “businessman” and “cadre” are dropped under individual fixed effect because of no value change between 2012 and 2014. DV = dependent variable; SOE = state-owned enterprise.
p < .1. **p < .05. ***p < .01.
Media Effect of the Anticorruption Enforcement.
Baseline group for education is illiterate/primary school. The reference group for age is elderly. Robust standard errors in parentheses. In model 1, the “overall LnAnti effect of identified groups” is anticorruption effect on middle school graduates, and “overall LnAnti effect of identified groups_2” is anticorruption effect on college graduates. Similarly, in model 2, the two overall LnAnti effects are for middle-aged people and young people, respectively. In model 3, that effect is for netizens. No participant changes Internet usage status, so the lower order term is automatically dropped in model 3 with individual fixed effect. In model 2, because two-year does not change cohort effect significantly, the lower order term of age group is automatically dropped with individual fixed effect. DV = dependent variable.
p < .1. **p < .05. ***p < .01.

Interaction term between identified groups and LnAnti.
For the media channel, we first examine the effects of anticorruption enforcement on people with different levels of education (column 1 of Table 4). Compared with respondents who never attended school or had only primary school education, political trust increased less for both middle-school graduates and college degree holders. The larger absolute marginal effect of college graduates (−0.073) relative to those with only a middle-school education (−0.056) also supports our hypothesis that access to more information further reduces the efficacy of propaganda at increasing political trust. However, the overall effect of anticorruption on trust for both groups still increased by 0.084 and 0.066, respectively. In column 2, we test responses across age groups: the elderly (sixty years and above), middle-aged (between thirty and sixty), and the young (below thirty). As expected, the middle-aged and the young were less affected by anticorruption propaganda than the elderly, and the marginal effects were more strongly negative for the younger group (−0.074) than the middle-aged (−0.042), suggesting that youth are most rebellious against propaganda. Again, the overall effects of anticorruption on trust for the two groups remain statistically significant, at 0.106 and 0.074, respectively. In column 3, we examine “netizens” and detect a negative interaction between Internet usage and anticorruption, suggesting that the propaganda was less effective in increasing political trust for netizens. This group also showed no significant increase in political trust due to anticorruption, overall. In general, our results support Hypothesis 2 that alternative information alleviates the effects of propaganda. 13 The interaction terms for the informed groups are shown in Figure 2.
Robustness Check
We conducted a series of robustness checks for the main findings presented in Tables 3 and 4. First, some may suggest that people only have an approximate perception of the level of their local government’s anticorruption efforts (e.g., high, medium, low), instead of the exact number of the arrested officials. We therefore convert the independent variable into an ordinal variable to run the major regressions. Results are largely consistent with the main findings (Online Appendix C).
Second, we are aware of the potential threat of selection bias affecting our findings of the effect of the direct experience of anticorruption policies. The CCP tends to recruit competitive people into the party and government. State-system insiders may therefore be better educated and more informed. To reduce the confounding effect of alternative information, we restrict our sample to those less-informed according to our previously outlined criteria (i.e., illiterate/primary school education, nonnetizens, and those above thirty years), and compare state-system insiders with the outsiders. 14 The regression results in column 1 of Table 5 identify a significant and negative relationship, which confirms Hypothesis 1. We also restrict the sample to a highly informed group and find that insiders still increase political trust less than outsiders with higher anticorruption efforts (see Online Appendix D). Second, to further identify respondents who had personal interactions with government officials, we construct a dummy variable, “experience,” for those who responded negatively when asked about interactions with government officials. 15 The interaction term, shown in column 2 of Table 5, is significantly negative. This further supports our hypothesis that for individuals with direct experience with government officials or agencies, intensive anticorruption efforts are less likely to yield a positive effect on political trust.
Robustness Check of State-System Insiders’ Political Trust.
Robust standard errors in parentheses. In model 2, because people with corruption experience have no status change over time, the lower order term is automatically dropped due to individual fix effect.
p < .1. **p < .05. ***p < .01.
Third, to reduce the confounding effects of one’s status as a state-system insider on the media channel, we ran a robustness check using only the state-system outsider subsample. The results in columns 1 to 3 of Table 6 show a similar pattern to Table 4. Political trust increases for state-system outsiders who access information from alternative sources. This lends further support to Hypothesis 2.
Robustness Check of Better-Informed Group’s Political Trust.
Baseline group for education is illiterate/primary school. The reference group for age is elderly. Robust standard errors in parentheses.
p < .1. **p < .05. ***p < .01.
Fourth, we have thus far assumed that locations with greater anticorruption efforts also had more anticorruption achievement propaganda. For a robustness check, we built an alternative measurement for anticorruption propaganda. We collected the number of news articles from city government–owned newspapers that covered corruption (fubai), embezzlement (tanwu), anticorruption (fan fubai), party working style (dangfeng), clean government (lianzheng), disciplinary committee (jiwei), and four poor work styles (sifeng), terms commonly used in official reports regarding anticorruption policies, based on the assumption that higher frequencies of such reports from local party mouthpieces reflect stronger anticorruption propaganda. We constructed a continuous variable, All Report, to represent the number of reports using all seven themes for the years 2012 and 2014. The results in Table 7 support our hypothesis that better-informed people are less affected by government propaganda.
News Reports as Measurement of Propaganda.
Baseline group for education is illiterate/primary school. The reference group for age is elderly. Robust standard errors in parentheses. DV = dependent variable.
p < .1. **p < .05. ***p < .01.
Fifth, we address preference falsification among respondents that may have shrouded their genuine views due to social or political pressures (Kuran 1997). We follow a practice developed by Jiang and Yang (2016) that uses questions about one’s political views, but that are less sensitive (i.e., “How do you anticipate your future turning out?” Score 1 for “very badly,” to 5 for “very well”?). Generally, we found that insiders’ genuine political support may have been lower than reported, which actually lends stronger support to Hypothesis 1, that state-system insiders respond to anticorruption policies less positively than state-system outsiders. Thus, preference falsification was not a major problem in our data. Online Appendix F provides more discussions.
Finally, we rule out the possibility that the less increase of political trust of insiders is due to a possible “ceiling effect.” The distributions of political trust of insiders and outsiders were comparable, and neither group had many respondents with very high political trust scores, such as 9 or 10, in 2012. We also ran regressions with a restricted sample of political trust scoring below 9. Results are consistent with the main findings and provided in Online Appendix G.
Conclusion
This study attempts to unpack intensive anticorruption enforcement’s effects on political trust. We proposed that anticorruption policies influence political trust through two channels, direct experience and the media, and may therefore have distinct impacts on different social groups as the two effects work unequally. We applied our theoretical model to recent anticorruption enforcement efforts in China, using data from a large, longitudinal national survey. We found that anticorruption policies generally had a positive effect on political trust.
However, the positive effect was less significant among two important elite groups whose political trust can influence regime stability, particularly in authoritarian countries. The first elite group was state-system insiders, such as party members, cadres, government and SOE employees, and businessmen. Members of these groups are usually selected carefully and placed in important positions to help maintain state control (Levitsky and Way 2002). Relative to outsiders, these groups are more likely to have direct experience of anticorruption policies. However, a deeper knowledge of policy implementation and outcomes can result in more reservations; that is, insiders’ trust in government increases less than does state-system outsiders’. The second elite group was better-informed citizens, with higher levels of education and who access alternative information sources. This group tends to comprise the “critical citizens” leading political opinions and social movements (Norris 1999). While mainstream media remain an important channel through which people learn about anticorruption efforts, the better-informed group is less influenced by government propaganda. Their trust in government was increased less than the less-informed group. We verified these findings using a DID research design; this helps to reduce the potential threat of causal linkage building, which could not be sufficiently controlled for in previous empirical studies on anticorruption policies and trust using cross-sectional data.
This research has both theoretical and practical implications. Theoretically, while many discussions have focused on whether anticorruption measures, such as auditing and replacing individual actors, are effective at reducing corruption (Avis, Ferraz, and Finan 2018; Klitgaard 1988), we expand the study to evaluate the broader impact of anticorruption on political trust. In addition, most research on political trust focuses on central government, which is assumed to play a driving role in nationwide anticorruption campaigns (e.g., Ji and Meng 2017; Z. Wang and Dickson 2017). We find that if local government focuses on fighting corruption, it can also earn trust, even when the campaign is initiated by the central government, instead of being a “doomed loser” as commonly presumed. Finally, we identify two main channels, direct experience and the media, as bridging the government’s macro-level performance with individual opinions. Understanding these channels is particularly important for policies such as anticorruption, which may not generate immediate or tangible benefits for the public to easily perceive and measure.
Practically, while this study provides empirical evidence of the positive effects of anticorruption policies on political trust, our analytical framework actually indicates that policy effects vary. Our framework alerts decision-makers to the importance of tracing divergent outcomes across social groups. For policies aimed at changing government employees’ behaviors, such as corruption control, trust generated from interactions with the government is probably more important than government propaganda. High-profile anticorruption will certainly raise the public’s expectations of the government, and the gap between reality and rising expectations can eventually undermine governmental legitimacy. Hence, actual behavioral changes at lower levels of government should be enforced to ensure that political trust is sustainable.
In addition, understanding how policy affects political trust should also remind policymakers to examine how responses to a policy differs by group, rather than simply assuming the general public consists of similar individuals. For intensive anticorruption enforcement activities, radical policy implementation processes and ineffective policy outcomes may lead state-system insiders to question the government’s capacity to handle corruption and result in less significant increases to political trust. Maintaining buy-in among this group is especially crucial, as the government relies heavily on these insiders to manage the state. As Fu (2016) pointed out, anticorruption enforcement needs to be orchestrated to not substantially threaten core supporters; otherwise, they will undermine the regime’s stability. Therefore, policymakers should consider different social groups’ needs when designing policies. Ideally, anticorruption efforts should evolve from a political campaign to institutionalized approaches, as the experiences of both Singapore and South Korea show.
We are aware of the limitations of our research. First, as the survey questions were limited, we were unable to compare political trust in central government before and after the campaign was launched. It would be valuable to examine whether trust in the central government follows a similar pattern to trust in local government, or if it is even stronger. Second, our policy implications are based on the assumption that government sincerely attempts to boost political trust through anticorruption policies. Undeniably, governments may hide their real intention for other political purposes, such as purging political rivals (Zhu and Zhang 2017). Policy implications would then differ greatly. In addition, scholars may examine how anticorruption enforcements have further affected political trust with new waves of CFPS; this is particularly important, as concerns have arisen that intensive enforcement efforts may backfire and result in administrative paralysis. Finally, current empirical findings are based on the case of China, which has a particular media environment and structure of political elites. How would anticorruption policies affect political trust through the two proposed channels in other countries with different political environments? We leave these questions for future studies to explore.
Supplemental Material
Online_Appendix_1 – Supplemental material for Do People Trust the Government More? Unpacking the Distinct Impacts of Anticorruption Policies on Political Trust
Supplemental material, Online_Appendix_1 for Do People Trust the Government More? Unpacking the Distinct Impacts of Anticorruption Policies on Political Trust by Siqin Kang and Jiangnan Zhu in Political Research Quarterly
Footnotes
Acknowledgements
We thank Professor Baishun Yuan for the generous sharing of interview notes. We also thank Abbie Chan, Elizabeth Chu, Maggie Huang, Arthur Poon, and Iris Wen for the excellent research assistance. We appreciate the valuable comments provided by Ting Gong, Yijia Jing, Ciqi Mei, Xing Ni, Hanyu Xiao, Diyangnan Zhang, and all the participants of the First Workshop of Empirical Political Science Research in Hong Kong. We also thank the Institute of Social Science Survey at Peking University for sharing the China Family Panel Studies (CFPS) data.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
Supplemental Material
Supplemental materials for this article are available with the manuscript on the Political Research Quarterly (PRQ) website.
Notes
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
