Combating corruption on the frontlines: analyzing penalties for street-level bureaucrats in China

Abstract

Penalties for corrupt acts are a vital component of integrity management systems within bureaucratic organizations, yet systematic inquiry into specific interventions is lacking. Drawing from the principal–supervisor–agent model, this article explores the criteria for imposing penalties on corrupt bureaucrats at the street level. We examine the impact of various factors on the severity of penalties for corruption, highlighting the influence of top-down anticorruption reform and local enforcement capacity. Utilizing text mining techniques, we build a comprehensive dataset containing 4025 cases of punishing corruption among grassroots officials in China from 2015 to 2023 and use multilevel analysis to explore these dynamics. Findings suggest that anticorruption reform directly led to an increase in the severity of penalties, but with substantial variation across practices. In particular, the analysis reveals a positive correlation between enforcement capacity and the severity of penalties. Moreover, while the relationship between harm and penalty severity represents a generalizable observation, significant disparities exist in the penalties imposed across various types of corruption. This study contributes to understanding the dynamics of multiple punitive measures and the factors leading to penalties imposed on corrupt street-level bureaucrats.

Full text

Vol.:(0123456789)
Global Public Policy and Governance
https://doi.org/10.1007/s43508-024-00096-3
RESEARCH ARTICLE
Combating corruption onthefrontlines: analyzing
penalties forstreet‑level bureaucrats inChina
RuiWang1 · XingNi2
Received: 9 April 2024 / Accepted: 10 September 2024
© The Author(s) 2024
Abstract
Penalties for corrupt acts are a vital component of integrity management systems
within bureaucratic organizations, yet systematic inquiry into specific interventions
is lacking. Drawing from the principal–supervisor–agent model, this article explores
the criteria for imposing penalties on corrupt bureaucrats at the street level. We
examine the impact of various factors on the severity of penalties for corruption,
highlighting the influence of top-down anticorruption reform and local enforcement
capacity. Utilizing text mining techniques, we build a comprehensive dataset
containing 4025 cases of punishing corruption among grassroots officials in China
from 2015 to 2023 and use multilevel analysis to explore these dynamics. Findings
suggest that anticorruption reform directly led to an increase in the severity of
penalties, but with substantial variation across practices. In particular, the analysis
reveals a positive correlation between enforcement capacity and the severity of
penalties. Moreover, while the relationship between harm and penalty severity
represents a generalizable observation, significant disparities exist in the penalties
imposed across various types of corruption. This study contributes to understanding
the dynamics of multiple punitive measures and the factors leading to penalties
imposed on corrupt street-level bureaucrats.
Keywords Corruption penalty· Street-level bureaucracy· Anticorruption reform·
Enforcement capacity
* Rui Wang
wrlyn2022@gmail.com
1 School ofManagement, Lanzhou University, Lanzhou, China
2 School ofPublic Administration andPolicy, Renmin University ofChina, Beijing, China

R.Wang, X.Ni
1 Introduction
A prerequisite for the durability of the state as well as its integrity management
is the capacity to hold lower-level agents accountable (Edin, 2003). Pervasive
street-level corruption not only weakens the legitimacy of public organizations
but also undermines public trust (Nieuwbeerta et al., 2003). Previous research
has examined the effectiveness of a wide array of anticorruption interventions,
such as building strong anticorruption agencies (Klitgaard, 1988; Quah,
2011), strengthening anticorruption laws (Mungiu-Pippidi & Dadašov, 2017),
implementing digital measures (Mutungi etal., 2021) and empowering citizens
to monitor public service (Molina et al., 2017). Among these measures,
punishment for corruption, which is often severe, has become salient and
popular in nondemocratic and hybrid regimes (Tsai etal., 2022). Instrumentally,
a considerable amount of research has focused on the benefits of anticorruption
punishment strategies by emphasizing that top-down bureaucratic control helps
address agents’ abuse of power (Grant & Keohane, 2005; Klitgaard, 1988).
Deontologically, the implementation of anticorruption punishment at the street-
level has also been shown to garner increased public support and signal political
will (Manzetti & Wilson, 2007; Tsai etal., 2022).
While scholars and policymakers are in consensus about the importance of
combating corruption at the front lines, the criteria for penalizing corruption can
be highly controversial. Previous literature typically discusses corruption penalties
within the framework of criminal law, based on the assumption that organizations
engage in a rational decision-making process when considering the punishment
of corrupt bureaucrats (Hylton, 2017; Saaty etal., 2015). Philosophers and legal
theorists strongly advocate for the principle of proportionality in punishment.
They noted that the severity of the penalties imposed should correspond to the
severity of the crime (Hart, 2008; Moore, 2010; Von Hirsch, 1992). In contrast,
another body of literature suggests that punishment is structured through explicit
standards and implicit norms, rather than an automated process (Pedersen
et al., 2018). Bureaucrats, especially those at the street-level, are sandwiched
between policymakers at the strategic level and local citizens, so they have to
manage the diverse expectations of different stakeholders (Page, 2006; Romzek
& Ingraham, 2000). Consequently, controlling corruption is a “black box” issue
that is largely context-specific and subject to variations in political reforms, local
characteristics, and individual behaviors.
This article is situated within the context of China, where street-level
corruption prevails and significantly undermines socio-economic development.
Despite repeated crackdowns, this form of corruption remains alarmingly high
and systematic, affecting critical sectors such as healthcare, education, and
land allocation. Street-level bureaucrats serve as critical agents for the state yet
holding them accountable has been difficult. Prior to 2018, the impact of the
anticorruption punishment system on grassroots officials was limited primarily for
two main reasons: First, the functions for controlling corruption were dispersed
across various agencies and departments, often resulting in a lack of coordination

Combating corruption onthefrontlines: analyzing penalties…
and coherence. Second, despite the availability of a variety of disciplinary tools,
the legislation remains ambiguous regarding the identification and punishment
of misconduct among nonparty cadres. This situation has changed since 2018,
following the establishment of the National Supervision Commission (NSC).
China has consolidated supervisory functions into a unified national supervisory
system, extending anticorruption measures to encompass not only party members
but also all public officials as well.
In this article, we draw on the principal–supervisor–agent model and empirically
explore how various factors influence the severity of corruption penalties imposed
on street-level bureaucrats. We propose that top-down anticorruption reform, local
enforcement capacity and the severity and types of corruption all have effects on
corruption penalties at the street-level. To test the above hypotheses, we obtained
rare access to collect data and built a comprehensive dataset containing 4025
cases of grassroots corruption reported by the Central Commission for Discipline
Inspection (CCDI) of China during the period from 2015 to 2023. Although those
cases are publicly available, they have not been systematically analyzed due to
the complexity, variability, and sheer volume of unstructured text data. Using text
mining techniques and natural language processing (NLP) tools, we identified the
details of corruption cases involving grassroots officials and the specific penalties
imposed on those individuals. Most importantly, we integrate punitive measures
with localized anti-corruption performance indicators to examine the influence of
various factors on the severity of corruption penalties.
We produced empirical findings about the effects of institutional, situational, and
individual factors on the application of corruption penalties at the street level. The
multilevel analyses reveal a wide variation in penalties for street-level corruption,
revealing complex dynamics that include party discipline, formal administrative
sanctions, and judicial prosecution. The findings confirm the signaling effect
whereby top-down anticorruption reform enhances the severity of penalties imposed
on corrupt bureaucrats at the street level. Moreover, by comparing the enforcement
capacity of provincial governments, we find that local enforcement practices can
directly influence the severity of penalties imposed on street-level bureaucrats.
The analysis results also highlight how the extent and types of the corrupt acts can
significantly impact the application and severity of punishment.
2 Literature andhypotheses
2.1 Corruption andanticorruption atthestreet level
Corruption incidence indicates a social pathology that distorts resource allocation,
undermines stability and reduces economic growth (Shleifer & Vishny, 1993). It
can take the form of grand corruption (Bauhr & Charron, 2018; Rose-Ackerman,
1975), such as senior officials within the political system engaging in the exchange
of power and profit (Gray, 2015). Although the extant literature suggests that
grand corruption primarily acts as a barrier to development, often described
metaphorically as “sanding the gears” of progress (Méon & Sekkat, 2005), it has

R.Wang, X.Ni
largely neglected another form of corruption in which street-level corruption has a
corrosive long-term effect on organizations (Jancsics, 2019; Jávor & Jancsics, 2016).
Street-level corruption, often referred to as “petty corruption,” is a form of corrupt
practice that occurs at a relatively small scale in the daily interactions between
citizens and lower-level public officials. For example, evidence has shown presented
that informal payments for public service delivery are prevalent (Davis, 2004), and
such corrupt practices distort the allocation of resources.
Lipsky’s classic study suggested that many street-level bureaucrats spend
their work lives in environments characterized by systemic inefficiencies and
corruption, which he describes as “corrupted worlds of service”(Lipsky, 1980).
Given the nature of street-level bureaucracy, research on the public administration
has highlighted that street-level corruption operates independently of corruption
occurring at higher levels of organizational hierarchies (Jávor & Jancsics, 2016).
Following the views of Lipsky and numerous other scholars, recent research
regarding street-level corruption invariably intertwines with themes such as
discretion, bureaucratic encounters, and accountability. A classic and enduring issue
in the public administration literature is how to achieve a proper balance between
administrative discretion and integrity. On the one hand, discretion is not only an
inevitable characteristic of street-level governance but also essential for the effective
implementation of policies (Thomann etal., 2018; Tummers & Bekkers, 2014). On
the other hand, the excessive discretionary power of street-level bureaucrats can
create convenient opportunities for corruption, especially in their direct interactions
with the public (Maynard-Moody & Musheno, 2000; Sandfort, 2000).
With an increased awareness of the pervasive impact of street-level corruption
on public services and political instability, strategies to combat it have become a top
priority within policy-making circles. However, there is a longstanding debate as
to whether anticorruption interventions show immediate results at the street level.
Typically, scholars generally assume that political leaders may lack the incentive
to implement anticorruption reforms because they often rely on the allocation
of spoils and patronage to maintain authority (Dal Bó & Powell, 2009), derive
personal gains from corrupt practices, and exploit anticorruption campaigns as
tools to target political opponents (Zhu & Zhang, 2017). Even putting aside political
considerations, controlling street-level corruption is very costly and difficult to
sustain in rigid bureaucratic systems, as it requires significant resources to detect
various corruption practices.
Among many anticorruption measures, sanction-based accountability is widely
considered to be a way to prevent corruption from becoming substantially worse. It
refers to the imposition of a punishment for transgressing a rule or poor performance
is detected (Tu & Gong, 2022). Theoretically, the central premise of sanction-based
accountability is that the certainty and severity of punishment can deter corrupt acts
by increasing the costs of engaging in such activities beyond the potential benefits
(Ortner & Chassang, 2018). In practice, governments in nondemocratic and hybrid
regimes have broadened their use of anticorruption punishment strategy in ways that
bolster public support and move in the direction of bureaucratic accountability. For
instance, Singapore has established rigorous laws and maintained severe penalties
for both the bribe giver and receiver (Ankamah & Manzoor E Khoda, 2018). The

Combating corruption onthefrontlines: analyzing penalties…
Nigerian government also declared zero tolerance for corruption and attempted
to convict influential elites on charges of corruption through the establishment of
anticorruption agencies (Justine & Okoye, 2014).
The importance of sanction-based accountability is rarely underestimated, yet
the literature on street-level bureaucracy is remarkably silent on the track record
of such a strategy (Gofen etal., 2022; Jancsics, 2019). On the one hand, existing
empirical studies are limited by their datasets, which fail to clearly distinguish
petty and grand corruption. Theorists have noted the emergence of the “street‐level
divergence” (Gofen, 2014), “collective ventures” (Gong, 2002) and uncontrolled
exercise of discretion (Kurniawan, 2022), yet quantitative evidence at the
street level has rarely been presented. On the other hand, the literature on street-
level bureaucracy is remarkably silent on how to understand the dynamics of the
enforcement of corruption penalties. Most existing anticorruption studies mainly
focus on the design and effectiveness of anticorruption strategies, however, there is
a wide knowledge and evidence gap on the “missing middle”, which is the practical
operation and implementation of anticorruption interventions. Therefore, there is a
need for both theoretical insights and empirical evidence to fill some of those gaps
in the literature.
2.2 Corruption penalties intheprincipal–supervisor–agent framework
Corruption often occurs when the principal(s) often cannot perfectly monitor the
agents. As such, enhancing a whole repertoire of sanctions and rewards is necessary
for minimizing the divergence between the principal’s objectives and the agent’s
actions (Han, 2020). Existing studies indicate that in settings where corruption
is systemic, traditional anticorruption reforms often fail because corruption is a
collective action problem (Persson etal., 2013). Research also shows that a collusion
problem arises when the supervisor (who is supposed to monitor and guide the
agent on behalf of the principal) forms an alliance with the agent to deceive or
work against the interests of the principal (Laffont & Meleu, 1997; Strausz, 1997).
These relationships constitute a complex network of multi-layered hierarchical
monitoring and control, providing a fertile ground for theoretical analysis under the
principal–supervisor–agent model. In corruption and anticorruption research, the
principal–supervisor–agent model has been widely applied to explain the dynamics
of corrupt behaviors and the implementation of contemporary anticorruption
strategies throughout the entire command chain (Brandt & Svendsen, 2013; Saaty
etal., 2015).
The principal–supervisor–agent model is a framework used in economics and
organizational studies to describe the hierarchical relationships between different
levels of authority and responsibility within an organization (Tirole, 1986). The
original purpose of anticorruption punishment was to hold bureaucrats accountable
at multiple levels and prevent corrupt practices. Therefore, the implementation
of punishment regimes, particularly in the context of anticorruption strategies
and governance, should be highly consistent with what the institutional design
requires. However, the credibility of these anticorruption efforts is undermined

R.Wang, X.Ni
by the selective implementation of policies among lower-level governments. As
noted by Persson and other scholars, anticorruption reform is likely to evolve into
a collective action problem of the “second order”(Persson etal., 2013). In other
words, as the central government continues to intensify top-down monitoring and
control over street-level bureaucrats, we argue that the enforcement capacity of local
governments is integral to a comprehensive anticorruption strategy.
Drawing on the principal–supervisor–agent model, this article analyzes how
anticorruption reform, initiated at a higher level of government, trickles down and
impacts the punishment of street-level bureaucrats. In this framework, there are
three key parties: the central government, which serves as the principal, is tasked
with formulating comprehensive anticorruption policies; local governments, which
act in a supervisory capacity, are responsible for the on-the-ground implementation
of these policies; and street-level bureaucrats, which function as agents, are held
accountable for their corrupt behavior. The control logic of formal hierarchical
structure is that authority flows from the top (higher levels) to the bottom (lower
levels). Incongruent goals and information asymmetry might incentivize local
jurisdictions to prioritize their own interests, thereby potentially undermining the
objectives of top-down anticorruption initiatives. This phenomenon especially
manifests in countries characterized by extensive geographical expansion and
significant interregional differences, where the centralized authority weakens
progressively as the chain of command extends and becomes more dispersed (Li,
2015).
It should be noted that local governments are not merely supervisors of street-
level bureaucrats, they also serve as agents of the central government. This dual role
requires local governments to interpret and adapt central anticorruption reforms to
their specific local contexts. However, our focus is on the interactions between dif-
ferent levels of government and the mechanisms of accountability. In this context,
we propose that the structure of the three-tier principal–supervisor–agent relation-
ship is central to understanding the penalties imposed on street-level bureaucrats
(Fig.1).
Fig. 1 Theoretical model

Combating corruption onthefrontlines: analyzing penalties…
2.3 Factors associated withcorruption penalties
Anticorruption is a complex issue that involves both legal enforcement and
policy implementation. The “anticorruption consensus” within the prevailing
development paradigm suggests that corruption originates in institutional failures
and maintains that institutional tinkering can alleviate corruption (Meyer &
Rowan, 1977; Shleifer & Vishny, 1993; Treisman, 2000). There are two major
ways in which anticorruption reform affects the severity of penalties for corrupt
officials. First, legal and institutional frameworks can be classified as positive
reinforcements that encompass a wider range of corrupt actives and establish
clear penalties for corrupt practices. Based on the deterrence theory, prior
research has suggested that legal and institutional frameworks are indispensable
for reducing corrupt acts (Quah, 2011), particularly in those countries with weak
institutional structures. Typically, strategies for combating corruption seek to
establish stringent controls by enhancing the likelihood of detecting and enacting
severe penalties for officials who are guilty of corruption. This finding is also
confirmed in laboratory settings (Boly et al., 2019). In fact, the introduction
of stricter laws and regulations is often accompanied by mandatory sentencing
guidelines for corruption-related offenses, which are intended to enhance harsher
penalties, such as longer prison sentences, heavier fines, or increased scrutiny and
monitoring postconviction.
Simultaneously, driven by strong political will, the initiation of anticorruption
reform is commonly accompanied by strategies such as power centralization
(Carothers, 2022), institutional integration (Li & Wang, 2019), and bureaucratic
control and penetration (Gong & Yang, 2022; Wang & Mou, 2021). Despite
the established correlation between democratic governance and the efficacy
of corruption control, combating corruption has for decades been a major
agenda issue for many nondemocratic and hybrid regimes. Recent research has
highlighted substantial anticorruption efforts and successful reforms undertaken
by authoritarian regimes and argued that these regimes navigate challenges
by leveraging their unique institutional strengths rather than by mimicking
democracies (Carothers, 2022). Moreover, strong links between political will
and anticorruption reform send signals about government efficiency and credible
commitment, fostering sustained support for enforcing credible sanctions
(Brinkerhoff, 2000). The purpose of anticorruption punishment, which is
often severe, is not only to act as a tool for political control but also to boost
public support (Tsai etal., 2022). Therefore, it is reasonable to hypothesize the
following:
Hypothesis 1 The initiation of anticorruption reform will significantly increase the
severity of penalties imposed on street-level bureaucrats.
While curbing corruption has assumed the form of a global crusade, marked
by the introduction of a wide array of anticorruption reform packages, researchers
contend that there is a notable discrepancy between the aspirational goals of these

R.Wang, X.Ni
anticorruption measures and their tangible effects (Brinkerhoff, 2000; Persson
etal., 2013). Previous literature typically suggests that the success or failure of
anticorruption reforms can primarily be attributed to the “missing middle” the
specific processes, resources, and actions required to effectively implement
them (Heeks & Mathisen, 2012). More specifically, anticorruption measures are
largely context-specific and subject to local integrity management systems. In the
context of anticorruption reform, the enforcement capacity of local government
institutions represents their ability to detect, investigate, prosecute, and penalize
corrupt activities. Evidence abounds that by complementing national-level
initiatives, local enforcement and anticorruption efforts ensure that top-down
policies are effectively implemented on the ground (Gemperle, 2018; Lacatus &
Sedelmeier, 2020; Prado & Carson, 2016). For example, scholars employ panel
data on corruption convictions and demonstrate that greater prosecutor resources
result in more convictions for corruption, which supports the “system capacity”
arguments (Alt & Lassen, 2012).
Scholars also agree that local protectionism and factionalism pose a genuine
threat to corruption investigations (Guo, 2014), especially when local politicians
and officials derive benefits from corrupt practices or patronage networks. A
laboratory experiment provides strong evidence that in-group loyalty undermines
the punishment of corruption, while out-group loyalty increases it (Solaz etal.,
2019). This is important because corruption is embedded in patron–client forms
of social organization (Uberti, 2016). Nevertheless, local governments may not
consistently cooperate with top-down anticorruption reforms. This argument has
received much empirical support. In a field study of monitoring corruption, while
traditional top-down monitoring has the potential to curb grassroots corruption,
special attention must be given to minimizing free-rider problems and restricting
elite capture (Olken, 2007). This is especially acute in countries undergoing
rapid political system transitions. In contexts where corruption has become the
norm rather than the exception, implementing stringent penalties can reinforce
accountability and cultivate a culture of integrity among local officials. In the
context of China, provinces possess significant autonomy and responsibility
for implementing and enforcing national policies at the local level. Local
governments’ enforcement capacity not only serves as a deterrent for corrupt
behavior but also increases accountability at the street-level. Therefore, the unit
of our analysis for assessing local governments’ enforcement capacity is at the
provincial level. Given the foregoing discussion, the following hypothesis is
proposed:
Hypothesis 2 Local governments with stronger enforcement capacities will impose
more severe penalties on street-level bureaucrats.
Finally, we posit that the nature of agents’ misconduct, particularly the
severity and types of corruption at the street-level, significantly influences
the determination of specific penalties. Along with legal positivism theory,
proportionality is used as a criterion of fairness and justice in law, aiming to

Combating corruption onthefrontlines: analyzing penalties…
balance the sanctions imposed by the authorities and the severity of the nature
of the prohibited act (Tonry, 2012). Scholars have proposed that offenders
convicted of violating laws should be subjected to penalties proportional to
the value of the damage inflicted upon others (Becker, 1968; Becker & Stigler,
1974), incorporating adjustments for the likelihood of evading detection. Here,
proportionality theory typically focuses on two factors essential to accountability:
the seriousness of improper behavior and the severity of punishment.
One approach involves identifying the outcomes of corruption and implementing
stringent penalties accordingly. There is abundant evidence that the monetary value
associated with corruption cases is a critical determinant in imposing penalties,
offering a tangible and quantifiable basis for punishment (Gong etal., 2019; Rose-
Ackerman, 1975). Another critical practice involves distinguishing between various
types of corruption and employing an asymmetric punishment strategy to effectively
address these issues. Examining the patterns of government corruption in China,
scholars find that bribery cases are punished more harshly compared to other
occupational crimes (Liu & Liu, 2017). Lab experiments differentiate the impact
of symmetric versus asymmetric punishment regimes on the behavior of potential
bribers and recipients. The findings indicate that more collusive corruption occurs
when punishment for bribers is more lenient (Engel etal., 2016). Reflecting on the
resulting discussion, we contend that both the financial magnitude and the distinct
types of corruption are crucial determinants in establishing suitable penalties.
Tentatively, the following is hypothesized:
Hypothesis 3 The greater the monetary value involved in a corruption case, the
more severe the penalties imposed on street-level bureaucrats.
Hypothesis 4 Varying types of corruption precipitate different levels of punishment
severity.
3 Research design
3.1 Penalties forstreet‑level bureaucrats inChina
Corruption is predominantly observed in developing and transitional nations
(Olken & Pande, 2012). China presents a good case for studying the effectiveness
of anticorruption reforms and the challenges in implementing them. Corruption
has remained a significant challenge in China since the founding of the Chinese
Communist Party (CCP). As a critical feature of a Leninist vanguard party, the
internal control system was imported to China, but it also followed a distinctive
historical trajectory. In the initial decades following the establishment of the
People’s Republic of China, the central authorities launched numerous periodic
yet intense campaigns of rectification aiming to use mass campaigns to bypass
bureaucracy and manage government officials’ illicit behavior. As Perry noted in
her study of the Four Cleans campaign, the targets of mass mobilization were local
cadres accused of misappropriating the fruits of land reform, embezzling collective

R.Wang, X.Ni
funds and engaging in other forms of corruption (Perry, 2021). Mass mobilization,
however, proved controversial, often resulting in violent struggles.
Since economic reforms began in 1978, rapid growth has led to a high demand
for capital and created numerous new opportunities for officials to pursue illicit
profit-sharing. At this stage, weak institutional capacity and inefficient legal systems
were blamed for worsening corruption. To prevent corruption from becoming
substantially worse, the CCP did not simply restore the CCDI; it also issued
policies constraining the behaviors of government officials. However, the spread of
corruption has outpaced institutional efforts against it (Gong & Tu, 2022), and ad
hoc anticorruption campaigns have failed to impose constraint on corrupt village
cadres (Wang & Mou, 2021). China embarked on an extensive anticorruption
campaign following the conclusion of the 18th National Congress of the CCP in
2012.
The importance of grassroots cadre management is rarely underestimated, yet
disciplining such management has been difficult because of ambiguous legislation
(Wang & Mou, 2021), a multi-track anticorruption system (Li & Wang, 2019) and
patronage networks (Wang & Yan, 2020). This situation has changed since 2018.
The formation of the National Supervisory Commission (NSC) consolidated
anticorruption resources to the central authorities and was aimed at expanding
its battle against corruption to cover all types of public servants. First, grassroots
cadres are now on an equal legal footing with other public agents and subject to state
laws. Second, following the establishment of the “comprehensive leadership” over
village affairs, the practice of “one shoulder-pole”—combining the elected village
committees and the party committees—has proven to be an effective way for the
Party to exert control over local agencies and grassroots cadres. Third, a campaign
waged by the CCP took decisive action to address “misconduct and corruption
that occur at people’s doorsteps” to promote social fairness as well as policy
implementation.
The constituent weight of three tools—party discipline actions, administrative
sanctions, and judicial prosecution—has been commonly used as an internal
mechanism for combating and punishing corrupt officials of all types. Anticorruption
investigations in China are carried out by the CCDI and the NSC. The CCDI (and
its local affiliates) are dedicated to enforcing the inner-party supervision system
and imposing party discipline sanctions. The NSC is co-located with the CCDI
to supervise all types of public officials and impose administrative sanctions. The
procuratorate at all levels takes responsibility for prosecuting and punishing all
types of corruption. These three types of penalties serve different functions but are
often interconnected in practice. For example, when a party member’s misconduct
involves criminal behavior, the CCDI may refer the case to judicial authorities for
prosecution, and administrative sanctions often coincide with party disciplinary
measures.

Combating corruption onthefrontlines: analyzing penalties…
3.2 Data andvariables
Our primary data source comprises two parts: one from street-level corruption cases
reported by the CCDI on its website and the second from institutional anticorruption
performance reports provided by the provincial CDIs for each year. Given that the
CCDI website has regularly published accounts of typical nationwide misconduct
and corruption cases occurring at the ‘people’s doorsteps’ since 2015, this new
dataset offers a unique opportunity to examine the determinants of punishment for
corrupt acts. We constructed a sample of 4025 corruption cases at the grassroots
level from 2015 to 2023. This period encompasses significant developments in anti-
corruption policies and the reform of the national supervision system, which are
crucial to understanding the trends and patterns of street-level corruption.
To reduce the costs of analyzing extensive collections of corruption cases, we
used natural language processing (NLP) tools to identify and capture specific
pieces of information from unstructured texts. First, we applied rigorous data
preprocessing, which included removing irrelevant content, correcting errors, and
standardizing formats. Second, we used named entity recognition (NER) models to
identify and classify named entities in the text, such as the names of street-level
bureaucrats, organizations, locations, dates, the type of corruption, and the severity
of punishment. Third, based on contextual patterns and keywords, we also leverage
domain-specific knowledge to extract variables. This resulted in a final sample of
4025 street-level bureaucrats who were sanctioned for their corrupt acts.
3.2.1 Dependent variable: theseverity ofpenalties
The dependent variables for this analysis are indicators of the severity of penalties
for bureaucrats at the street level. We investigate three dependent variables that
capture the consequences faced by individuals for their actions or misconduct:
judicial prosecution, party disciplinary actions, and administrative sanctions. (1)
Judicial prosecution is a dummy variable coded as 1 if an official is brought to
trial for allegedly committing a crime and 0 if an official is not subject to judicial
convictions. Judicial prosecution is a good dependent variable because it is
considered one of the harshest penalty mechanisms and such judicial punishment
sends out the most vital signals to deter corrupt officials. (2) Party disciplinary action
contains different severity levels of specific measures, code 1 for warnings, code
2 for probation within the party, code 3 for demotion, code 4 for suspension, and
code 5 for expulsion from the party. (3) Administrative sanction refers to penalties
imposed by government agencies, code 1 for warnings, code 2 for demerits, code
3 for severe demerits, code 4 for demotion, code 5 for removal from the office, and
code 6 for expulsion.
This variable is significant due to the expansion and reform of administrative
sanctions related to post-2018 corruption behavior. We employed binary coding to
represent whether judicial action was taken in each case. It is important to note that
this coding system was not uniformly applied across all types of penalties. For party
disciplines and administrative penalties, we focused on cases where a penalty was

R.Wang, X.Ni
clearly documented. The different levels of punishment reflect the severity of the
actions taken.
3.2.2 Key independent variables
To determine top-down anticorruption reform, the most straightforward method is
through official documents and government records that specify its implementation
timeline. The specific proposal for the National Supervision Commission
was formally announced in October 2016. After the pilot phase, the national
implementation began with the formal establishment of the National Supervision
Commission in 2018. While there is a set of variables that capture key milestones
in the reform process, scholars suggest that “reform implementation” has a more
significant impact than “reform announcement” (Persson et al., 2013; Thomas
& Grindle, 1990). Therefore, to mitigate measurement errors and enhance the
empirical robustness of the study, we construct reform implementation according to
the official date and set a binary variable to 1 for post-reform, and 0 for per-reform.
To assess the extent of the enforcement capacity of local governments, we used the
total number of public officials annually punished for corrupt acts at the provincial
level. This variable is found in local reports on anticorruption performance and is
logged to mitigate the presence of outliers. Given that anticorruption punishment is a
concrete action rather than mere slogans or declarations, actual punishment for corrupt
activities refers to local governments’ credible commitment to curb corruption.
Individual corruption factors are measured with two indicators. The amount of
money involved is a numerical variable that quantifies the extent of the abuse of
power. It serves as an indicator of the degree to which resources are diverted from
public services. Different types of corruption often result in varying severities of
punishment. For instance, systemic embezzlement might be seen as more severe
than a single instance of bribery due to its larger scale and impact. Corruption
can manifest in various forms, including bribery, embezzlement, fraud, and other
illicit activities. The severity of punishment for each form of corruption varies,
influenced by its specific impact and the corresponding legal frameworks. In this
study, corruption type is delineated as a categorical variable. The coding scheme
is as follows: 1 for embezzlement, 2 for bribery, 3 for misappropriation, and 4 for
abuse of power. This classification is based on a careful review of each case, with
the primary type of corrupt act being identified and coded accordingly.
3.2.3 Control variables
The severity of punishment is context-specific and is subject to both individual
factors and institutional parameters. The first set of control variables is the influence
of individual characteristics. While previous research underlines the importance
of considering the individuals’ positions, bureaucrats at the street-level often have
less pronounced rank differences than do compared to higher-level officials. Hence,
we filter three individual parameters, namely, village jurisdiction, dual position
status, and leadership position. We set village jurisdiction as a dummy equal to 1

Combating corruption onthefrontlines: analyzing penalties…
if an official operates within village jurisdictions, dual position status as a dummy
equal to 1 if an official bears dual or multiple responsibilities in the context of
Chinese bureaucracy, leadership position as a dummy equal to 1 if a grassroots
official is in primary leadership roles. The second set of control variables is the local
determinant. We utilize provincial GDP per capita as a measure of local economic
development, as economic development is related to anticorruption punishment.
3.3 Model specifications
Although street-level bureaucrats are subject to the same institutions and policies,
the punitive measures they face are contingent upon specific local contexts. We
used a multilevel model to test our hypotheses, which allows for modeling and
interpreting nested data structures (Hox etal., 2017). This technique has been widely
used for analyzing how context affects individual-level outcomes and has proven
to be effective in revealing corruption dynamics (Ni & Su, 2019). The dependent
variables are both binary and ordinal, necessitating the use of multilevel mixed-
effects logistic regression and multilevel mixed-effects ordered logistic regression,
respectively. We also address potential issues such as multicollinearity and missing
data through correlation analysis and sensitivity analysis.
For a two-level model, where level-1 units are nested within level-2 units, the
multilevel models are specified as:
Level 1 focuses on individual factors, reflecting both the nature and severity of
corruption, as well as the personal characteristics of corrupt officials. On the other
Corruption penalties
ij
=𝛽
0j
+𝛽
1
log (Corrupt money)
ij
+𝛽
2
Corruption types
ij
+
⋯
+ϵ
ij
𝛽0j
=𝛾
00
+𝛾
0j1
Anticorruption reform
j
+𝛾
02
log (Enforcement capacity)
j
+⋯+u
0j
Table 1 Descriptive statistics of
key variables Variables Obs Mean SD Min Max
Dependent variables
Judicial prosecution 4025 0.415 0.493 0 1
Party disciplinary action 3757 3.662 1.615 1 5
Administrative sanction 949 5.095 1.643 1 6
Independent variables
Anticorruption reform 4025 0.291 0.454 0 1
Enforcement capacity (log) 3402 9.482 0.973 6.258 11.019
Corrupt money (log) 3304 2.667 1.800 0.020 11.429
Corruption types 4025 2.765 1.149 1 4
Control variables
Village jurisdiction 4025 0.532 0.499 0 1
Dual position status 4025 0.139 0.346 0 1
Leadership position 4025 0.496 0.500 0 1
GDP per capita (log) 4016 10.793 0.336 10.131 12.100

R.Wang, X.Ni
hand, level 2 concentrates on institutional variables, which include information
about institutional reform and local anticorruption performance.
4 Research ndings
Table 1 presents the descriptive statistical analysis of the variables. The total
number of cases is 4025. Of these, 3757 cases involve party disciplinary action, and
949 cases involve administrative sanctions. Overall, 41.5% of the individuals in our
dataset were subject to judicial prosecution. The mean of party disciplinary action is
3.662 (5-point scale), with a standard deviation of 1.615, indicating that the average
disciplinary action taken is closer to the higher severity end of the spectrum. The
mean number of administrative sanctions is 5.095 (6-point scale), showing that the
sanctions imposed are, on average, of greater severity. It is worth mentioning that
we added a time variable (pre- and post-reform) to account for the period effect. In
addition, the local enforcement capacity reflects the scope and intensity of measures
implemented to combat corruption in various localities, showing a considerable
range of effectiveness. The resulting log-transformed variable, with a mean of 9.482
and a standard deviation of 0.973, is designed to minimize the impact of outliers.
Approximately 53.2% of the corruption cases in our analysis involved officials at the
village level, while a mere 13.9% of the cases were associated with officials holding
dual position status. The mean of leadership position is 0.496, closely approaching
the midpoint of 0.5, which indicates a nearly even distribution between regular and
deputy positions among the officers in our data.
This section analyses the determinants of the severity of punishment for corrupt
bureaucrats at the street-level by combining top-down anticorruption reform, local
governments’ enforcement capacity, and the severity of corruption in one analytical
framework. Table2 reports our main findings, and the empirical regression results
confirm our theoretical hypotheses. Table 3 also reports the coefficients, robust
standard errors, percent change in odds ratio, Akaike information criterion (Akaike,
1974), and Bayesian information criterion (Schwarz, 1978).
Model 1 focuses on judicial convictions, suggesting that after controlling for
other factors, top-down anticorruption reform has a significant positive effect on the
dependent variable. The implementation of institutional reform is associated with
approximately doubling the odds of an official being judicially convicted compared
to scenarios where such reform is absent. Moreover, local enforcement capacity mat-
ters greatly. For each one-unit increase in the logged measure of local enforcement
capacity, the odds of an official being judicially convicted increase by approximately
48.6%. Our research substantiates a proportional relationship between the severity
of punishment and the amount of corrupt money involved. This demonstrates that an
increase in the logged value of corrupt funds significantly elevates the likelihood of
a street-level official’s judicial conviction. Specifically, our findings indicate that for
each increment in the logged amount, the odds of conviction almost double, reflect-
ing a 99.9% increase in odds ratios. The odds of being convicted of misappropriation
and abuse of power are reduced by 60.5% and 85.5%, respectively, when compared

Combating corruption onthefrontlines: analyzing penalties…
Table 2 Multilevel modeling on determinants of corruption penalties
***p < 0.01, **p < 0.05, *p < 0.1
Variables/models Judicial prosecution Party disciplinary action Administrative sanction
Model 1 Model 2 Model 3
Coefficient (standard
error)
Change odds ratio Coefficient (standard
error)
Change odds ratio Coefficient (standard
error)
Change odds ratio
Anticorruption reform 0.739*** (0.146) 1.095 0.731*** (0.118) 1.077 0.526* (0.279) 0.692
Enforcement capacity
(log)
0.396*** (0.090) 0.486 0.221*** (0.079) 0.247 0.388*** (0.121) 0.473
Corrupt money (log) 0.693*** (0.036) 0.999 0.771*** (0.036) 1.162 0.760*** (0.078) 1.138
Corruption type_ bribery 0.238 (0.157) 0.269 0.252 (0.163) 0.287 0.660* (0.363) 0.936
Corruption type_
misappropriation
− 0.905*** (0.124) − 0.595 − 0.929*** (0.122) − 0.605 − 0.992*** (0.320) − 0.629
Corruption type_ abuse
of power
− 2.016*** (0.142) − 0.867 − 1.932*** (0.121) − 0.855 − 2.577*** (0.322) − 0.924
Village jurisdiction − 0.0322 (0.124) − 0.032 0.469*** (0.108) 0.599 − 0.261 (0.372) − 0.23
Dual position status 0.547*** (0.147) 0.728 0.248* (0.132) 0.281 0.442 (0.526) 0.555
Leadership position − 0.145 (0.118) − 0.135 − 0.145 (0.100) − 0.135 − 0.145 (0.285) − 0.135
GDP per capita (log) 0.526* (0.282) 0.693 0.317 (0.271) 0.373 1.045** (0.438) 1.842
Observations 2838 2720 568
Number of groups 30 30 30
AIC 2789.331 5664.475 950.2708
BIC 2860.741 5753.101 1019.745
Wald chi2 (df) 570.67 (10) 807.64 (10) 186.81(10)

R.Wang, X.Ni
to embezzlement. However, the odds of being judicially convicted of bribery do not
significantly differ from those of being judiciously convicted of embezzlement.
Party disciplinary actions vary widely in the context of China, ranging from
warnings to expulsion from the party. Model 2 shows the extent to which multiple
factors affect the severity of party disciplinary actions. The empirical regression
analysis reveals a statistically significant positive relationship between top-down
institutional reform and the incidence of party disciplinary actions. This outcome
suggests that the reform of the supervision system may correspondingly lead to
strengthened disciplinary practices within the Chinese Communist Party (CCP).
Model 2 indicates that the implementation of reform is associated with more than
double the odds of a higher level of disciplinary action (moving from a lower to a
higher category of severity) compared to when there is no reform implementation.
Increased local enforcement capacity appears to correlate with a greater likelihood of
street-level officials facing more severe disciplinary actions. Specifically, our analysis
indicates that the odds ratios of such outcomes increase by approximately 24.7%.
Additionally, a one-unit increase in the logarithm of the amount of corrupt money
amount is associated with a 116.2% increase in the odds of imposing a higher level
of disciplinary action, controlling for other variables. These coefficients also show
how the severity of punishment changes for each type of corruption, relative to the
base category (in our analysis, embezzlement). In cases involving misappropriation
and abuse of power, the likelihood of encountering more severe disciplinary action
is comparatively lower than that in cases involving embezzlement. This is evidenced
by decreases in the odds ratios of 60.5% and 85.5%, respectively.
Model 3 reveals that not all factors exert a substantial influence, which is inconsistent
with the findings of Model 1 and Model 2. While the odds of street-level bureaucrats
receiving a higher level of administrative sanction increase by approximately 69.2% after
institutional reform, this result is significant only at the 10% level. A potential explanation
for this could be the ambiguity inherent in the rules governing administrative sanctions,
which previously did not encompass a wide range of street-level bureaucrats, such as
village cadres, prior to the supervision system reform. Consequently, this ambiguity
complicates efforts to compare the extent of administrative sanctions before and after
the implementation of the reform. The model also shows that local enforcement capacity
intensifies the severity of administrative sanctions imposed on street-level bureaucrats,
with the odds ratios increasing by 47.3%. As the logarithm of the amount of corrupt
money amount increases, the likelihood of imposing higher levels of disciplinary action
on street-level bureaucrats also increases, and the odds ratios are 1.138 times greater. In
addition, different forms of corruption are met with varying degrees of administrative
penalties, as evidenced by the consistent results across Model 1 and Model 2.
In terms of the control variables, the regression results also exhibit many
interesting estimated effects. Our analysis reveals that street-level bureaucrats
operating under village jurisdiction, as opposed to those who are not, experience
a 60% increase in the odds of encountering more severe disciplinary actions. This
outcome indicates that party disciplinary actions on village cadres play a crucial
role in the governance of rural areas because they also serve to maintain the party’s
authority and legitimacy at the grassroots level. Moreover, after controlling for other
factors, street-level bureaucrats with a dual position status are approximately 72.8%

Combating corruption onthefrontlines: analyzing penalties…
more likely to be convicted than those without dual position status. In addition,
the variables of local GDP per capita demonstrate interesting findings. Unlike to
the findings in previous anticorruption research, economic determinants only have
significantly positive effects on the severity of administrative sanctions.
This study offers a detailed analysis of robustness considerations related to the
main findings. First, we implement our models as ordinary least squares regressions
with standard errors clustered by local government. These findings confirm that the
results align closely with the significant and consistent main findings presented in
Table2, supporting our theoretical hypotheses. Second, as we examine the effect
of reform implementation, our results are potentially threatened by temporal
confounding. Although the precise measurement of the implementation timeline for
institutional anticorruption reform is essential for evaluating its impact, we cannot
rule out the potential influence of signal effects associated with the announcement of
the reform. To address this concern, we utilize the timing of the announcement of a
pilot reform of the supervisory system as an independent variable to test the expected
effects of the reform announcement. As shown in “Appendix” Table3, except for
this variable, the direction of the estimates of all other variables is consistent with
our main results. This finding confirms our theoretical expectation that in the context
of reforming China’s inspection system, the actual on-the-ground implementation of
these reforms significantly enhanced the effectiveness of corruption penalization, far
exceeding the impact of merely announcing the intention to enact such reforms.
5 Discussion andconclusion
Despite widespread consensus on the necessity of combating corruption at the
street-level, few studies have linked individual, situational, and institutional factors
to the extent to which penalties are imposed on corrupt street-level bureaucrats.
Drawing from the principal–supervisor–agent model, we propose that corruption
penalties rely on the complex considerations of top-down anticorruption reform,
local governments’ enforcement capacity, and the severity and types of corruption.
By utilizing data from large-scale, street-level anticorruption cases and integrating
local anticorruption performance indicators, this article seeks to investigate the
relationships between high-level policy strategies and their tangible impacts on the
ground.
We examine the above arguments in the context of the supervisory system
reform in China, where street-level anticorruption initiatives have undergone
a significant transformation. The establishment of the NSC in 2018 marks a
significant restructuring of China’s anticorruption framework, signaling an increase
in the party-state’s capacity to monitor and control its bureaucracy. The empirical
study further demonstrated that top-down anticorruption reform, initiated by the
central government, trickles down, influencing disciplinary actions against street-
level bureaucrats. This reform proved effective as it served as a deterrent against
future corrupt acts, demonstrated a credible commitment to controlling corruption,
and fostered a more accountable governance structure. Higher-level anticorruption
reforms have promoted sanction-based accountability, indicating an upward trend in

R.Wang, X.Ni
the use of various penalty practices. These findings align with the broader literature
on the effectiveness of various anti-corruption strategies and a paradigm shift in
disciplining bureaucrats (Gans-Morse etal., 2018; Ni & Su, 2019; Wang & Mou,
2021).
Further analysis indicates that local governments’ enforcement capacity positively
affects the severity of corruption penalties at the street level. Driven by a strong political
will, political leaders often circumvent unreliable monitoring agencies, adopting a
top-down approach reinforced by severe sanctions to disrupt the highly integrated
equilibrium of group interests (Rothstein, 2011). Although temporary anticorruption
campaigns can immediately impact corrupt practices, they seldom sustain a willingness
to curb corruption and lack long-term effects (Di Tella & Schargrodsky, 2003). A
principal reason for this is the discrepancy between the institutional framework and
its actual implementation (Persson et al., 2013). Indeed, the broader sociopolitical
repercussions of local enforcement capacity are frequently overlooked. More
accountable agents may signify not only more widespread corruption but also a
heightened level of enforcement by anticorruption agencies within local governments.
The capacity of local enforcement and accountability systems is critical not only for
ensuring that national directives and policies are effectively enacted but also for
dismantling patronage networks that compromise fair governance.
The third noteworthy finding is that different types of corruption incur a variety of
penalties due to the multifaceted nature of corrupt acts and their disparate impacts on
government. The penalties for embezzlement are markedly more severe than those for
the misappropriation of public funds and the abuse of power. Embezzlement, typically
characterized as the intentional misappropriation of state or public assets for personal
enrichment, results in direct and quantifiable financial loss. This form of corruption is
perceived as particularly grievous, hence warranting more severe penalties. However,
the available evidence is insufficient to definitively assert that the punitive measures for
embezzlement significantly differ from those for bribery.
This study makes several important contributions. First, this article aims to clarify
the theoretical limitations of anticorruption interventions by drawing on insights
from the principal–supervisor–agent model. Specifically, our analysis reveals that
contemporary successful anticorruption reforms extend beyond mere outliers. In
China, the reform of the national supervision system grounded in a central-local-
grassroots relational framework, disrupted the state’s entrenched corrupt equilibrium
by imposing corruption control from a top-down perspective. Second, our study
contributes to the sanction-based accountability literature through a comparative
analysis of various accountability mechanisms implemented to sanction corruption
among street-level bureaucrats. Moreover, this study advances the current discourse
on the application of textual analysis within the realms of public administration and
public policy research. Through the analysis of large-scale semi-structured text data,
this study gains unique access to information pertinent to corruption cases, thereby
filling existing data gaps. This approach presents a promising avenue for addressing
the issue of data scarcity in corruption research.
Overall, this article seeks to integrate the literature on anticorruption with the study
of street-level bureaucracy, aiming to enhance the understanding of the interactions
between macro-level mechanisms and their on-the-ground enforcement. Although the

Combating corruption onthefrontlines: analyzing penalties…
main empirical results are based on the Chinese anticorruption practices, our find-
ings about corruption penalties and street-level governance have broader theoretical
implications for nondemocratic and hybrid regimes. This study indicates a significant
escalation in the penalties imposed on corrupt officials at the grassroots level in China
following the implementation of anticorruption reform. However, such high-pressure
anticorruption strategies sometimes lead to several unintended consequences that may
undermine the effectiveness of these measures. Further research is warranted to inves-
tigate the above question across diverse situational and institutional settings. We also
propose that future research should focus on the overlaps and interconnections between
different types of penalties, with particular attention to the thresholds at which adminis-
trative or party penalties might trigger judicial intervention.
Appendix
See Table3.
Table 3 Multilevel Modeling on corruption penalties (the announcement of a pilot reform)
***p < 0.01, **p < 0.05, *p < 0.1
Variables Judicial prosecution
coefficient (standard
error)
Party disciplinary action
coefficient (standard
error)
Administrative sanction
coefficient (standard
error)
The announcement of a
pilot reform − 0.0861 (0.115) 0.162 (0.106) 0.380 (0.254)
Enforcement capacity
(log)
0.275*** (0.0884) 0.154* (0.0882) 0.404*** (0.120)
Corrupt money (log) 0.684*** (0.0362) 0.764*** (0.0355) 0.754*** (0.0776)
Corruption type_ bribery 0.243 (0.156) 0.239 (0.162) 0.654* (0.361)
Corruption type_
misappropriation
− 0.878*** (0.123) − 0.892*** (0.122) − 0.892*** (0.328)
Corruption type_ abuse
of power
− 2.048*** (0.142) − 1.944*** (0.121) − 2.573*** (0.323)
Village jurisdiction 0.0306 (0.122) 0.508*** (0.107) − 0.223 (0.370)
Dual position status 0.523*** (0.147) 0.244* (0.132) 0.442 (0.528)
Leadership position − 0.165 (0.117) − 0.149 (0.1000) − 0.155 (0.284)
GDP per capita (log) − 0.0421 (0.284) − 0.217 (0.305) 0.950** (0.418)
Observations 2838 2720 568
Number of groups 30 30 30

R.Wang, X.Ni
Funding The research was supported by the National Social Science Found of China, 21CZZ027.
Data availability The data that support the findings of this study are available on request from the
corresponding author. The data are publicly available also from the governmental website:https:// www.
ccdi. gov. cn/
Declarations
Conflict of interest On behalf of all authors, the corresponding author states that there is no conflict of
interest.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License,
which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long
as you give appropriate credit to the original author(s) and the source, provide a link to the Creative
Commons licence, and indicate if changes were made. The images or other third party material in this
article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line
to the material. If material is not included in the article’s Creative Commons licence and your intended
use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permis-
sion directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/
licenses/by/4.0/.
References
Akaike, H. (1974). A new look at the statistical model identification. IEEE Transactions on Automatic
Control, 19(6), 716–723.
Alt, J. E., & Lassen, D. D. (2012). Enforcement and public corruption: Evidence from the American
states. The Journal of Law, Economics, and Organization, 30(2), 306–338.
Ankamah, S. S., & Manzoor E Khoda, S. (2018). Political will and government anti-corruption efforts:
What does the evidence say? Public Administration and Development, 38(1), 3–14.
Bauhr, M., & Charron, N. (2018). Insider or outsider? Grand corruption and electoral accountability.
Comparative Political Studies, 51(4), 415–446.
Becker, G. S. (1968). Crime and punishment: An economic approach. Journal of Political Economy,
76(2), 169–217.
Becker, G. S., & Stigler, G. J. (1974). Law enforcement, malfeasance, and compensation of enforcers. The
Journal of Legal Studies, 3(1), 1–18.
Boly, A., Gillanders, R., & Miettinen, T. (2019). Deterrence, contagion, and legitimacy in anticorruption
policy making: An experimental analysis. The Journal of Legal Studies, 48(2), 275–305.
Brandt, U. S., & Svendsen, G. T. (2013). Why does bureaucratic corruption occur in the EU? A princi-
pal–supervisor–agent model. Public Choice, 157, 585–599.
Brinkerhoff, D. W. (2000). Assessing political will for anti-corruption efforts: An analytic framework.
Public Administration and Development, 20(3), 239–252.
Carothers, C. (2022). Taking authoritarian anti-corruption reform seriously. Perspectives on Politics,
20(1), 69–85.
Dal Bó, E., & Powell, R. (2009). A model of spoils politics. American Journal of Political Science, 53(1),
207–222.
Davis, J. (2004). Corruption in public service delivery: Experience from South Asia’s water and sanita-
tion sector. World Development, 32(1), 53–71.
Di Tella, R., & Schargrodsky, E. (2003). The role of wages and auditing during a crackdown on corrup-
tion in the city of Buenos Aires. The Journal of Law and Economics, 46(1), 269–292.
Edin, M. (2003). State capacity and local agent control in China: CCP cadre management from a town-
ship perspective. The China Quarterly, 173, 35–52.
Engel, C., Goerg, S. J., & Yu, G. (2016). Symmetric vs. asymmetric punishment regimes for collusive
bribery. American Law and Economics Review, 18(2), 506–556.

Combating corruption onthefrontlines: analyzing penalties…
Gans-Morse, J., Borges, M., Makarin, A., Mannah-Blankson, T., Nickow, A., & Zhang, D. (2018).
Reducing bureaucratic corruption: Interdisciplinary perspectives on what works. World Develop-
ment, 105, 171–188.
Gemperle, S. M. (2018). Comparing anti-corruption agencies: A new cross-national index. International
Review of Public Administration, 23(3), 156–175.
Gofen, A. (2014). Mind the gap: Dimensions and influence of street-level divergence. Journal of Public
Administration Research and Theory, 24(2), 473–493.
Gofen, A., Meza, O., & Perez Chiques, E. (2022). When street-level implementation meets systemic cor-
ruption. Public Administration and Development, 42(1), 72–84.
Gong, T. (2002). Dangerous collusion: Corruption as a collective venture in contemporary China. Com-
munist and Post-Communist Studies, 35(1), 85–103.
Gong, T., & Tu, W. (2022). Fighting corruption in China. China Review, 22(2), 1–19.
Gong, T., Wang, S., & Li, H. (2019). Sentencing disparities in corruption cases in China. Journal of Con-
temporary China, 28(116), 245–259.
Gong, T., & Yang, S. L. (2022). Controlling bureaucratic corruption. In B. G. Peters & I. Thynne (Eds.),
Oxford encyclopedia of public administration. Oxford University Press. https:// doi. org/ 10. 1093/
acref ore/ 97801 90228 637. 013. 1463
Grant, R. W., & Keohane, R. O. (2005). Accountability and abuses of power in world politics. American
Political Science Review, 99(1), 29–43.
Gray, H. S. (2015). The political economy of grand corruption in Tanzania. African Affairs, 114(456),
382–403.
Guo, X. (2014). Controlling corruption in the party: China’s central discipline inspection commission.
The China Quarterly, 219, 597–624. https:// doi. org/ 10. 1017/ s0305 74101 40007 70
Han, Y. (2020). The impact of accountability deficit on agency performance: Performance-accountability
regime. Public Management Review, 22(6), 927–948.
Hart, H. L. A. (2008). Punishment and responsibility: Essays in the philosophy of law. Oxford University
Press.
Heeks, R., & Mathisen, H. (2012). Understanding success and failure of anti-corruption initiatives.
Crime, Law and Social Change, 58, 533–549.
Hox, J., Moerbeek, M., & Van de Schoot, R. (2017). Multilevel analysis: Techniques and applications.
Routledge.
Hylton, K. N. (2017). Whom should we punish, and how: Rational incentives and criminal justice reform.
William & Mary Law Review, 59, 2513.
Jancsics, D. (2019). Corruption as resource transfer: An interdisciplinary synthesis. Public Administra-
tion Review, 79(4), 523–537.
Jávor, I., & Jancsics, D. (2016). The role of power in organizational corruption: An empirical study.
Administration and Society, 48(5), 527–558.
Justine, I.-I.C., & Okoye, J. C. (2014). Anti-corruption crusade in Nigeria: More words than deeds. Inter-
national Journal of Public Policy and Administration Research, 1(2), 47–63.
Klitgaard, R. (1988). Controlling corruption. Univ of California Press.
Kurniawan, T. (2022). Discretion as a factor in corruption: A case from Indonesia. Public Integrity, 24(7),
692–701.
Lacatus, C., & Sedelmeier, U. (2020). Does monitoring without enforcement make a difference? The
European Union and anti-corruption policies in Bulgaria and Romania after accession. Journal of
European Public Policy, 27(8), 1236–1255.
Laffont, J. J., & Meleu, M. (1997). Reciprocal supervision, collusion and organizational design. Scandi-
navian Journal of Economics, 99(4), 519–540.
Li, L. (2015). “Rule of Law” in a Party-State: A conceptual interpretive framework of the constitutional
reality of China. Asian Journal of Law and Society, 2(1), 93–113.
Li, L., & Wang, P. (2019). From institutional interaction to institutional integration: The national supervi-
sory commission and china’s new anti-corruption model. The China Quarterly, 240, 967–989.
Lipsky, M. (1980). Street level bureaucracy: Dilemmas of the individual in public services. Russell Sage
Foundation. https:// doi. org/ 10. 7758/ 97816 10447 713
Liu, G., & Liu, S. (2017). Corruption crime and punishment: Evidence from China’s state corruption
audits. Journal of Financial Crime, 24(4), 601–619.
Manzetti, L., & Wilson, C. J. (2007). Why do corrupt governments maintain public support? Compara-
tive Political Studies, 40(8), 949–970.

R.Wang, X.Ni
Maynard-Moody, S., & Musheno, M. (2000). State agent or citizen agent: Two narratives of discretion.
Journal of Public Administration Research and Theory, 10(2), 329–358.
Méon, P.-G., & Sekkat, K. (2005). Does corruption grease or sand the wheels of growth? Public Choice,
122, 69–97.
Meyer, J. W., & Rowan, B. (1977). Institutionalized organizations: Formal structure as myth and cer-
emony. American Journal of Sociology, 83(2), 340–363.
Molina, E., Carella, L., Pacheco, A., Cruces, G., & Gasparini, L. (2017). Community monitoring inter-
ventions to curb corruption and increase access and quality in service delivery: A systematic review.
Journal of Development Effectiveness, 9(4), 462–499.
Moore, M. S. (2010). Placing blame: A theory of the criminal law. Oxford University Press.
Mungiu-Pippidi, A., & Dadašov, R. (2017). When do anticorruption laws matter? The evidence on public
integrity enabling contexts. Crime, Law and Social Change, 68, 387–402.
Mutungi, F., Baguma, R., Ejiri, A. H., & Janowski, T. (2021). Digital anti-corruption typology for public
service delivery. International Journal of Computer Applications, 975, 8887.
Ni, X., & Su, X. (2019). Institutional anticorruption in China: Effectiveness on bribery incidence. Public
Administration Review, 79(4), 538–551.
Nieuwbeerta, P., De Geest, G., & Siegers, J. (2003). Street-level corruption in industrialized and develop-
ing countries. European Societies, 5(2), 139–165.
Olken, B. A. (2007). Monitoring corruption: Evidence from a field experiment in Indonesia. Journal of
Political Economy, 115(2), 200–249.
Olken, B. A., & Pande, R. (2012). Corruption in developing countries. Annual Review of Economics,
4(1), 479–509.
Ortner, J., & Chassang, S. (2018). Making corruption harder: Asymmetric information, collusion, and
crime. Journal of Political Economy, 126(5), 2108–2133.
Page, S. (2006). The web of managerial accountability: The impact of reinventing government. Adminis-
tration and Society, 38(2), 166–197.
Pedersen, M. J., Stritch, J. M., & Thuesen, F. (2018). Punishment on the frontlines of public service
delivery: Client ethnicity and caseworker sanctioning decisions in a Scandinavian welfare state.
Journal of Public Administration Research and Theory, 28(3), 339–354.
Perry, E. J. (2021). Missionaries of the party: Work-team participation and intellectual incorporation. The
China Quarterly, 248(S1), 73–94.
Persson, A., Rothstein, B., & Teorell, J. (2013). Why anticorruption reforms fail—Systemic corruption as
a collective action problem. Governance, 26(3), 449–471.
Prado, M. M., & Carson, L. (2016). Brazilian anti-corruption legislation and its enforcement: Potential
lessons for institutional design. Journal of Self-Governance and Management Economics, 4(1), 34.
Quah, J. S. (2011). Curbing corruption in Asian countries: An impossible dream? Emerald Group Pub-
lishing Limited.
Romzek, B. S., & Ingraham, P. W. (2000). Cross pressures of accountability: Initiative, command, and
failure in the Ron Brown plane crash. Public Administration Review, 60(3), 240–253.
Rose-Ackerman, S. (1975). The economics of corruption. Journal of Public Economics, 4(2), 187–203.
Rothstein, B. (2011). The quality of government: Corruption, social trust, and inequality in international
perspective. University of Chicago Press.
Saaty, T. L., Özdemir, M. S., & Shang, J. S. (2015). The rationality of punishment—Measuring the sever-
ity of crimes: An AHP-based orders-of-magnitude approach. International Journal of Information
Technology and Decision Making, 14(01), 5–16.
Sandfort, J. R. (2000). Moving beyond discretion and outcomes: Examining public management from
the front lines of the welfare system. Journal of Public Administration Research and Theory, 10(4),
729–756.
Schwarz, G. (1978). Estimating the dimension of a model. The Annals of Statistics, 1978, 461–464.
Shleifer, A., & Vishny, R. W. (1993). Corruption. The Quarterly Journal of Economics, 108(3), 599–617.
Solaz, H., De Vries, C. E., & De Geus, R. A. (2019). In-group loyalty and the punishment of corruption.
Comparative Political Studies, 52(6), 896–926.
Strausz, R. (1997). Collusion and renegotiation in a principal–supervisor–agent relationship. Scandina-
vian Journal of Economics, 99(4), 497–518.
Thomann, E., van Engen, N., & Tummers, L. (2018). The necessity of discretion: A behavioral evaluation
of bottom-up implementation theory. Journal of Public Administration Research and Theory, 28(4),
583–601.

Combating corruption onthefrontlines: analyzing penalties…
Thomas, J. W., & Grindle, M. S. (1990). After the decision: Implementing policy reforms in developing
countries. World Development, 18(8), 1163–1181.
Tirole, J. (1986). Hierarchies and bureaucracies: On the role of collusion in organizations. The Journal of
Law, Economics, and Organization, 2(2), 181–214.
Tonry, M. (2012). Punishment and politics. Willan.
Treisman, D. (2000). The causes of corruption: A cross-national study. Journal of Public Economics,
76(3), 399–457.
Tsai, L. L., Trinh, M., & Liu, S. (2022). What makes anticorruption punishment popular? Individual-level
evidence from China. The Journal of Politics, 84(1), 602–606.
Tu, W., & Gong, T. (2022). Bureaucratic shirking in China: Is sanction-based accountability a cure? The
China Quarterly, 249, 259–274.
Tummers, L., & Bekkers, V. (2014). Policy implementation, street-level bureaucracy, and the importance
of discretion. Public Management Review, 16(4), 527–547.
Uberti, L. J. (2016). Can institutional reforms reduce corruption? Economic theory and patron–client
politics in developing countries. Development and Change, 47(2), 317–345.
Von Hirsch, A. (1992). Proportionality in the philosophy of punishment. Crime and Justice, 16, 55–98.
Wang, J., & Mou, Y. (2021). The paradigm shift in the disciplining of village cadres in China: From Mao
to Xi. The China Quarterly, 248(S1), 181–199.
Wang, P., & Yan, X. (2020). Bureaucratic slack in China: The anti-corruption campaign and the decline
of patronage networks in developing local economies. The China Quarterly, 243, 611–634.
Zhu, J., & Zhang, D. (2017). Weapons of the powerful: Authoritarian elite competition and politicized
anticorruption in China. Comparative Political Studies, 50(9), 1186–1220.
Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps
and institutional affiliations.