Convicting Politicians for Corruption: The Politics of Criminal Accountability

Abstract

Why are politicians more likely to be prosecuted and convicted for corruption in some contexts rather than in others? Pulling together disparate threads of the literature on what we call the politics of criminal accountability, this review organizes current explanations along three levels of inquiry: (1) micro, encompassing characteristics of individual criminal-accountability agents and defendants, such as their partisanship and ideology, professional ethos, enforcement costs and judicial corruption; (2) meso, emphasizing the independence, capacities and coordination degrees of criminal-accountability institutions; and (3) macro, including the impact of political regimes, political competition, support from civil society, corruption levels and international norms. In doing so, we draw attention to methodological shortcomings and opportunities for research on the topic, providing a roadmap for this field of inquiry that also includes unexplored questions and tentative answers. Furthermore, we present new systematic data set that reveals a substantial increase in the conviction of former heads of government for corruption since 2000, underscoring the importance of the phenomenon and highlighting the need for further research into the politics of criminal accountability.

In recent decades, there has been a worldwide increase in the number of former heads of government who have been convicted for corruption by the judiciary of their own countries. The trend is widespread, covering presidents, prime ministers and dictators across continents, political regimes and countries with differing levels of economic development. Former leaders who faced this previously unprecedented fate come from affluent liberal democracies such as France and Italy, middle-income electoral democracies including Brazil and Ukraine, electoral autocracies like Egypt and the Philippines, and closed autocracies such as Sudan, to cite a few.¹

In some instances, this trend of increased domestic criminal accountability has been dramatic. In South Korea, two former military dictators and two former democratically elected presidents have been convicted since the mid-1990s; another former president committed suicide after investigations closed in on his allies and family; and the current president of South Korea was himself a leading prosecutor in two of these cases. Similarly, in Peru nearly all elected presidents since 1990 have been accused of corruption over the following decades, leading to one conviction, two temporary incarcerations, one presidential resignation and also, sadly, one suicide.²

Even if not all these convictions have been definitive – some were later overturned by other judicial decisions or pardoned by future heads of government – the fact that more domestic courts have been willing to declare their former heads of government guilty of criminal offences contrasts with previous historical record as well as with the fate of other leaders who, despite numerous claims of corruption, have escaped conviction – former presidents such as Suharto (Indonesia, 1968–1998) and Jean-Claude Duvalier (Haiti, 1971–1986), who were even prosecuted for allegedly having embezzled millions of dollars but nonetheless not convicted, illustrate the point (Hodess 2004: 13). At the same time, prosecutions and convictions for corruption are not always seen as just. While they are frequently perceived as demonstrations that no one is above the law, they are just as often recognized as products of overly politicized criminal justice systems. The topic, as a result, has important implications for the rule of law, corruption control and political stability in both authoritarian and democratic regimes.

In this article, we conduct an extensive review of the literature on what we call the politics of criminal accountability, which addresses why politicians are more likely to be prosecuted and convicted for corruption in some contexts rather than in others. We review and organize current explanations along three levels of inquiry: (1) micro, encompassing characteristics of individual criminal-accountability agents and attributes of defendants; (2) meso, emphasizing differences across criminal-accountability institutions; and (3) macro, including factors such as the adoption of international norms, a country's political regime and the level of political competition.

Additionally, to underscore the significance of understanding the factors driving prosecutions and convictions of political authorities, we introduce an original data set, Heads of Government Convicted of Crimes (HGCC), encompassing all convictions involving former heads of government (presidents, prime ministers and dictators) between 1946 and 2020. These data reveal a substantial increase in corruption-related convictions across all continents over the past few decades, highlighting the importance of the phenomenon for global politics in the 21st century.

This review article is structured as follows: first, leveraging the HGCC data set, we describe the temporal variation in judicial convictions against former heads of government. We then present a conceptual framework delineating our definition of criminal accountability and characteristics of the main actors and institutions. Next, to explore the factors contributing to variation in criminal accountability for corruption, we review the sparse but growing literature on the topic along three levels of analysis: micro (individual), meso (institutional) and macro (political). We conclude by suggesting directions for future research.

Convicting heads of governments in courts of law

To highlight the importance of the phenomenon for contemporary politics, we assembled the Heads of Government Convicted of Crimes (HGCC) data set. This compilation documents the universe of criminal convictions received by heads of government from the judiciary of the countries they once governed between 1946 and 2020. Figure 1 portrays the number of former heads of government convicted by decade, illustrating a notable upsurge in convictions for corruption, beginning in the 1990s. We distinguish between convictions that include corruption charges (e.g. influence in return for money, embezzlement, kickbacks, bribery, electoral fraud and money laundering) and those that do not. The number of heads of government convicted for corruption by domestic courts increased from 0 between the 1940s and 1960s to just 1 over each of the next two decades, to 8 in the 1990s, 14 in the 2000s, and 31 in the 2010s. This increase in the number of criminal convictions is an important and underappreciated outcome of the global anti-corruption movement of the last few decades. There has also been an increase in the number of convictions on grounds other than corruption in the 2010s.

Figure 1. Heads of Government Convicted for Corruption and Other Reasons, by Decade

Source: Da Ros and Gehrke (2024).

To be coded as convicted, leaders must be sentenced by a civilian court located in the country they once governed. Leaders who were only convicted by courts in other countries, by the International Criminal Court, or by ad hoc military trials are not classified as convicted in our data set. Nicolae Ceausescu (1965–1989), former dictator of Romania, was sentenced by an extraordinary military court in 1989 for genocide, subversion of state power and for trying to escape the country using one billion US dollars deposited in foreign banks. The trial lasted for about one hour only and Ceausescu was executed, together with his wife, just after the trial ended. Manuel Noriega (1983–1989), former dictator of Panama, was sentenced by a court in the United States for drug trafficking, racketeering and money laundering. Between 1993 and 1996, Noriega was sentenced by Panamanian courts for his involvement in the assassinations of political rivals. A few years later, Noriega was also convicted by a French court for money laundering. As such, while Ceausescu's conviction and Noriega's convictions outside Panama are not included in our data set, Noriega's conviction by Panamanian courts is included. This coding decision is built on our objective to capture the instances of criminal accountability in the countries leaders once governed. The data set includes all heads of government (presidents, prime ministers and dictators) who were in office for at least one full year between 1946 and 2020 and contains details of all convictions, the year of the judicial sentence and whether they were later pardoned or had their convictions overturned by other instances of the judiciary. Even if convictions were later pardoned or overturned, the initial coding as ‘convicted’ remains.³

Prior to the 1970s, it was extremely rare for former heads of government to be sentenced. Between the 1970s and 1990s, they were more likely to be convicted for human rights abuses, plotting the overthrow of governments and political violence than for corruption. This pattern has changed since the mid-1990s. Between then and 2020, the percentage of heads of government who had been convicted increased from 2% to 9%. Our data show that the proportion of former leaders who were convicted for corruption has risen significantly across all continents, suggesting the global reach of the phenomenon (Figure 2). The recent increases in the Americas and in Asia-Oceania are particularly striking.

Figure 2. Heads of Government Convicted of Crimes over Time by Continent, 1946–2020

Source: Da Ros and Gehrke (2024).

We focus on former heads of government because most constitutions guarantee immunity from prosecution while a leader is in power or require an impeachment or the authorization from the relevant legislature to initiate prosecution (Reddy et al. 2020).⁴ In addition, heads of government are often directly responsible for or have influence over the appointment of the attorney general.⁵ This might allow sitting heads of government to protect themselves from investigations and even to exert pressure on cases against their political adversaries, including former leaders who remain electorally powerful after leaving office and may try to come back to the highest executive position in the future (Baturo 2017; Epperly 2013).

Convicting a former head of government is an important stress test of a country's political stability and judicial independence. Most trials of former leaders are political events with enormous visibility. Attachment to these leaders, for personalistic, partisan or ethnic reasons, might condition citizens' perceptions of the fairness of the trial and raise concerns about biased investigations and questions about the level of politicization of justice. Likewise, criminal accountability may be important to hold political leaders accountable for what they did in office given the limitations of electoral accountability (De Vries and Solaz 2017; Dunning et al. 2019).

Interestingly, criminal convictions became more common as more violent and arbitrary fates of former leaders became less frequent (Figure 3). For the universe of heads of government exiting from office, the probability that a former leader would be killed, imprisoned or go into exile in the first year after leaving office decreased from more than 30% between 1960 and 1980 to 12% between 2000 and 2015.⁶ Unlike judicial convictions for corruption in our data set, which take on average 6.5 years (with a median of 5.5 years) after a leader has left office, imprisonment right after leaving office is more likely to be the result of an arbitrary decision.

Figure 3. Political Leaders' Post-Tenure Fate, 1946–2015

Source: Authors' analysis using non-parametric local regressions (LOESS algorithm) based on Archigos data set.

In addition to these more violent post-tenure fates, the possibility of being convicted in the future might also influence whether presidents and prime ministers are willing to accept electoral defeat and transfer power to a successor. An important literature on authoritarian politics and regime transitions has addressed the conditions under which dictators are likely to concede, including their future vulnerability to convictions for human rights abuses (Escribà-Folch and Wright 2015; Geddes et al. 2014). If their wrongdoing in office is likely to be discovered and punished, leaders, including democratically elected ones, may be less likely to accept defeat if that helps them to avoid criminal sanctions. Transfers of power thus have often included some type of amnesty, although there is always the possibility that future elected officials and prosecutors might renege on these promises (Posner and Young 2018).

Defining criminal accountability

This section defines criminal accountability, characterizing the institutions and agents involved in prosecuting, trying and convicting politicians for corruption. The literature on criminal accountability covers a much broader swathe of political authorities than just heads of government. That is, even though the data presented in the previous section are only about the convictions of heads of government, our review encompasses all instances in which elected and high-level appointed officials have been subject to prosecution or trial under corruption charges.⁷

One of the reasons we reviewed this broader literature is that the specific literature about criminal cases involving former heads of governments is small and predominantly legal or descriptive (e.g. Conaghan 2012; Lutz and Reiger 2009). By contrast, the emerging literature on the prosecution and trial of other types of public officials (including at the subnational level) is not only more abundant but has also been more focused on generalizability and testing causal claims. Similarly, our review is not limited to criminal convictions, but also includes stages that precede them (i.e. trials, prosecutions, investigations). So, reviewing this literature on the broader process of criminal accountability allows us to place the phenomenon of convicting heads of government within a relatively coherent, if also recent, scholarship, increasing the scope of potential explanations for the rise in convictions documented previously.

Accordingly, criminal accountability is the process of enforcing criminal sanctions against agents who failed to abide by the rules of criminal law. In corruption cases, criminal accountability is the enforcement of such sanctions against public officials who engaged in abuse of office for private or partisan gain. The latter typically involves practices such as bribery, embezzlement, procurement fraud, influence peddling and illicit campaign financing, as well as other ancillary activities, such as racketeering and money laundering, as defined in the criminal statutes of each country.⁸

What distinguishes criminal accountability from other forms of legal accountability, such as civil and administrative accountability, is the fact that it can result in imprisonment.⁹ Criminal accountability is therefore the type of accountability that enforces the harshest forms of punishment available in most societies (Bovens 2007; Da Ros 2019; Lindberg 2013). Because of that, criminal accountability is typically embedded in the separation of powers of modern democratic governments, so that the functions of making, executing and interpreting the law are split across different institutions and agents.

Criminal accountability is hence an intrinsically interinstitutional process. At a minimum, it involves two different sets of institutions: the prosecutors' offices and the courts. Likewise, it comprises at least three different sets of actors: prosecutors, judges and the accused. Correspondingly, criminal accountability in corruption cases minimally refers to the sequential processes of prosecuting, judging and eventually convicting public officials who have been criminally accused of corruption.

Since criminal accountability involves interrelated yet distinct decisions (to prosecute, to judge, to convict, etc.), not all of the literature reviewed here examines all such decisions jointly. That is, research on criminal accountability occasionally analyses only corruption prosecutions or only corruption trials, for instance. Still, because all such decisions are part of the broader process of holding public officials accountable before the criminal justice system, throughout this review, we refer to the ‘outcomes’, ‘results’ or ‘processes’ of criminal accountability as categories that encompass corruption prosecutions, trials and convictions, either separately or jointly. Likewise, we refer to prosecutors' offices and courts as ‘criminal-accountability institutions’, and to prosecutors and judges as ‘criminal-accountability agents’.

Similarly, the terms ‘prosecutors’ office’ and ‘court’ encompass a variety of species because these institutions may assume distinct names and forms depending on the context. In fact, many explanations of the variety of outcomes of criminal accountability revolve around such variations. For instance, prosecutors' offices may be located within the executive branch and therefore subject to the politics of appointment within the administration, such as the Department of Justice in the United States (Gordon 2009). But they may also be institutionally located within the judicial branch as in Italy (Della Porta 2001) or be independent from any of the typical three branches of government as in Brazil and Chile (Ríos-Figueroa 2012). Specifically in corruption cases, moreover, prosecution offices may also take the form of semi-autonomous anti-corruption agencies (ACAs) that explicitly possess prosecution powers. Most ACAs do not have prosecutorial attributions, limiting their roles to the investigation of wrongdoing and the enforcement of administrative sanctions. A few ACAs, however, are legally entitled to file criminal charges in corruption cases, as in Indonesia and Romania, in which role they function as prosecutors' offices (Mungiu-Pippidi 2018; Schütte 2016).

The courts, likewise, exhibit various shapes and names. They may be trial courts presided over by a single judge or panels that comprise multiple magistrates. Similarly, courts may have their powers split into distinct investigative and adjudication bodies – the first working on the collection of evidence and the second properly judging the cases – as in France (Adut 2004). At times, because politicians in various countries enjoy provisions of immunity, they have special standing and can only be tried by the high courts, in some cases even after they leave office (Reddy et al. 2020). Lastly, there may be courts specialized in corruption and corruption-related practices such as money laundering, as in Albania, Brazil, Indonesia, the Philippines and Slovakia (Madeira and Geliski 2019; Stephenson and Schütte 2022).

But criminal accountability need not involve only prosecutors' offices and courts. Criminal accountability often encompasses other agencies in addition to these two. One institution that often participates in criminal accountability is the police. Again, it may have distinct names and shapes, including specialized divisions or offices, but its role is often pivotal to the investigation of corruption, a stage that typically precedes the prosecution (Arantes 2011). Other institutions that often take part in the investigation are the aforementioned ACAs (Sousa 2010), but it does not need to stop there. In fact, institutional multiplicity has been relatively frequent in recent large-scale investigations, at times involving auditing bodies, financial intelligence units and tax authorities, among others (Carson and Prado 2016). Lastly, because prosecutions often attempt to recover the money resulting from corruption, which was usually laundered abroad, they typically also rely on international cooperation (Acorn 2018; Sims 2011).

Reviewing the causes of different outcomes of criminal accountability, therefore, means that we are interested in explanations of both why they happen (if at all) and to what extent. Consequently, we take into account the agents, the institutions and the diverse ways in which they investigate, prosecute and judge corruption allegations.

Explaining criminal accountability

What are the causes of corruption prosecutions and convictions of political officials or lack thereof? Why are some politicians prosecuted and convicted for corruption and not others? What explains the varied performance of criminal-accountability institutions in punishing allegedly corrupt officials?

This section reviews three types of explanation corresponding to three distinct yet interrelated levels of inquiry. First, we address micro-level explanations, based on the characteristics of the individuals involved in criminal accountability – prosecutors and judges, on the one hand, and the accused, usually an elected official, on the other. Second, we review meso-level explanations, based on the institutional aspects of criminal accountability. These explanations consider factors such as the varying degrees of independence, capacities and coordination among criminal justice institutions. Third, we review macro-level explanations that address decisions made by political elites either to empower or disempower criminal-accountability institutions and agents. Within each of these three levels, we identify four types of explanation, summarized in Table 1.

Table 1. Summary of Explanations for Criminal Accountability

Source: Authors' compilation.

In this review, we cover a total of 118 studies in which the primary or secondary outcome centres on the occurrence or absence of prosecutions or convictions of elected officials due to corruption. Our inclusion criteria encompass studies that investigate the impact of individual or multiple factors on the presence, frequency and intensity of such events. We employ contemporary search tools (Research Rabbit and Google Scholar) and snowballing techniques to identify representative studies that provide insightful explanations grounded in theory. Our approach also involves excluding studies that are general in nature, such as descriptions of the functioning of the judicial branch in a country.

Additionally, we do not prioritize any method, discipline or region over others. We review recent empirical studies that addressed the substantive questions of interest using qualitative, quantitative or mixed methods. Likewise, even if most literature reviewed below comes from political science, it also reflects the efforts of neighbouring disciplines, such as economics, sociology, criminology and law. Lastly, although a few countries received more attention than others from the literature, our review covers a significant number of nations: considering only the case studies and small-N comparative studies reviewed in this article, 35 countries across all continents are covered. Our search strategy hence intentionally aims to achieve a well-balanced selection of studies from a wide range of geographical regions, disciplines and methodologies.

Micro-level explanations

Given the various agents that take part in prosecuting, trying and eventually convicting politicians for corruption, we start with the micro-foundations of such processes. So, the questions here are: why are some prosecutors and judges more active in punishing allegedly corrupt politicians than others? And, relatedly, why are some individuals (politicians, businesspeople, etc.) targeted by prosecutors and judges more frequently and severely than others?

The first question asks about characteristics of criminal-accountability agents that may affect their decision-making, while the second asks about characteristics of those accused of corruption that may affect the same decisions. These two sets of characteristics interact, of course. Four types of explanations for variation in the results of criminal accountability derive from such characteristics: partisanship and ideology; professional identities; enforcement costs; and judicial corruption.

The partisanship and ideology of prosecutors and judges may affect how they perform their functions largely as an extension of the logic driving political competition (Balán 2011). Accordingly, prosecutors and judges may go after, disregard or even collude with politicians allegedly involved in corruption depending on their own partisan or ideological leanings, as well as that of the potentially targeted politicians. That is, prosecutors and judges appointed by one party may treat politicians from other parties more severely than those from their own party or that of their appointing principals. The net result is a criminal-accountability process that is weaponized to attack opponents or to protect allies, or both (Maravall 2003). Overall, the evidence produced by Sanford Gordon (2009), Milena Ang (2017) and Maria Popova and Vincent Post (2018) in distinct contexts of corruption prosecutions – the United States, Mexican states and Eastern European countries, respectively – points in this direction. Adding to this perspective, corruption convictions in the United States seem more frequent in electoral years when a state is electorally salient and when governors and presidents are politically aligned (Davis and White 2021; Pavlik 2017). In many works in this line of inquiry, the individual behaviour of criminal-accountability agents seems to derive from the institutional setting – and, specifically, from the rules that govern the appointment of prosecutors, which affect their degree of independence from the administration. Even in a context of highly independent prosecutorial and court systems, the ideological leanings of prosecutors and judges may also impact their behaviour, as works by Andrea Ceron and Marco Mainenti (2015) and Lucia Manzi (2022) about Italian magistrates suggest.

Professional identities refer to the expressed ethos of prosecutors and judges as they perform their jobs. They encompass professional self-conceptions, values and beliefs about how they are expected to behave, comprising predominantly immaterial incentives for action or inaction in face of allegations of impropriety by political officials. This applies especially to contexts where prosecutors' offices and courts enjoy high levels of independence from the political branches. The literature that examined periods of increased charges and convictions in countries like Italy, France and Brazil suggested that such occurrences were, to some extent, influenced by changes in the ideas and perspectives of prosecutors and judges. In all such cases, predominantly young, low-ranking, overzealous prosecutors and judges behaved as ‘true believers’ in the fight against corruption from the bench as they attempted to purge what they perceived to be overly corrupt political systems (Adut 2004; Da Ros and Taylor 2022; Della Porta 2001). The absence of this ideational component, in turn, failed to lead to an active stance against corruption from the bench in Germany, according to comparative research by Kimberly Sims (2011). Relatedly, variations in professional identities are frequently aligned with the development of and adherence to different legal doctrines, including law and order, penal populism, defence of due process, or deference to political elites resulting from the political question doctrine, among others. Exposure to such legal doctrines may arise from the different training and academic experiences of criminal-accountability agents, as well as with their interactions with peers in other countries. Similarly, anti-corruption may fuel internal struggles within different legal careers, being deployed to empower one group of legal actors within prosecutors' offices or the courts against others (Adut 2004; Engelmann 2020).

The concept of enforcement costs encapsulates another perspective, centred on the characteristics of the accused. From the viewpoint of legal officials, there are different costs associated with prosecuting, judging and eventually convicting politicians who hold different kinds of offices.¹⁰ Prosecuting a single former independent city councillor from a small town is much less costly for criminal-accountability agents than charging a large group of powerful sitting senators. The stakes differ because not all politicians can wield the same power against criminal-accountability agents and institutions. Accordingly, the literature has abundant evidence of political backlashes against criminal-accountability processes, which seem to be more pronounced as more political figures of national stature are accused of corruption (Conaghan 2012; Da Ros and Taylor 2022; Hein 2015; Vannucci 2009). Overall, this suggests that characteristics of the accused may influence the decisions to prosecute or to convict because they entail different risks to prosecutors and judges, including public scrutiny and potential retaliation. Much here has to do with the differences between frying ‘small or big fish’ and the respective capacities of the accused to fight back against the accusations or, more to the point, their accusers. More formally, such differences refer to the public positions held by the accused, including their stature (e.g. federal or local authorities, mayors and governors who administer larger or smaller cities and states), participation in the sitting administration (e.g. being in the government or the opposition, being a current or former public official), political leverage (e.g. the levels of congressional and popular support enjoyed by the accused) and the sheer number of accused officials (i.e. how widespread corruption allegations may be among political elites and within the state).

Evidence from Brazil, for instance, suggests that federal officials exhibit lower conviction rates than local officials (Levcovitz 2020), that current mayors had lower chances of conviction than former mayors (Bento et al. 2020), and that mayoral candidates charged with misconduct are less likely to be convicted if they win, even by a small margin, rather than lose an election (Lambais and Sigstad 2022). Similar evidence holds in India, where politicians from the ruling party who win legislative elections are more likely to be acquitted than politicians who lose elections or who are not from the ruling party (Poblete-Cazenave 2023). Criminal-accountability agents in authoritarian systems may also be constrained in their capacity to favour the prosecution of popular subnational elites because of the potential backlash from the public (Buckley et al. 2022). Because of such varying enforcement costs, prosecutors and judges may act strategically, picking the battles they are more likely to win, courting support (from political opponents, the media, civil society or the international community) and pushing the limits of criminal law to overcome these costs and target specific defendants who may be considered too powerful (Da Ros and Taylor 2022; González-Ocantos and Baraybar 2019).

Judicial corruption refers to the fact that the prosecutors and judges may themselves be corrupt (Basabe-Serrano 2013; Gloppen 2014; Wang and Liu 2022). One possibility here is that legal and political officials participate in a single corrupt network where they collude to protect each other, as implied by Michael Johnston's ‘elite cartel’ corruption syndrome (Johnston 2005). Another possibility is that politicians, prosecutors and judges may each have their own corrupt schemes, which are enabled by mutual non-interference arrangements – for example, prosecutors may take bribes to shelve cases and judges may fill judicial positions with their relatives, while politicians break campaign finance rules. These may also interact, with prosecutors and judges taking bribes to overlook political corruption. In either way, as Maria Popova (2012a) explains in reference to Bulgaria, the typical net results are prosecutorial and judicial passivity towards political corruption, since going after someone else's corruption may eventually turn against one's own (corrupt) interests. That said, data on judicial corruption are known to be exceedingly hard to obtain, and so is causal research about it.¹¹ Correspondingly, it is even harder to provide evidence that judicial corruption affects judicial performance in corruption cases proper. One example of how such research might be conducted, however, is provided by John McMillan and Pablo Zoido (2004), who reported data of bribes paid to judges and prosecutors during the 1990s in Peru under Alberto Fujimori. Less overt tools can also be used: Supreme Court judges in India are more likely to rule in favour of the incumbent government when they are about to retire and are rewarded with more prestigious jobs when they do so (Aney et al. 2021).

Meso-level explanations

Why are some courts and prosecution offices more active or severe in punishing allegedly corrupt politicians than others? Answers to this question often revolve around the two basic dimensions of the state as applied to criminal-accountability institutions: autonomy and capacity. The former is often referred to as ‘independence’ and can be understood as the inverse of the level of control exerted by political principals over an institution and its agents – the lower such control, the higher the independence. Capacity relates to the tools and resources available to an institution to execute its functions (Fukuyama 2013; Geddes 1994). The capacity of criminal-accountability institutions, however, is an exceedingly broad category, so we have split it here into legal and organizational capacities. Likewise, because criminal accountability is an interinstitutional process, we incorporate explanations that take into account the coordination (or lack thereof) across different institutions. Consequently, there are four major institutional explanations for different results of criminal accountability: independence, legal capacity, organizational capacity and interinstitutional coordination.

The independence of courts and prosecutors' offices has consistently held a central role in the realm of corruption studies. Scholars such as Susan Rose-Ackerman and Bonnie Palifka (2016) and Alina Mungiu-Pippidi (2015) attribute lower levels of corruption to higher levels of judicial independence. That is so because judicial and prosecutorial independence have often been understood as preconditions for the expected impartiality of such institutions. At stake here is the fact that some of the political principals who may be targets of criminal accountability are the ones who may also want to control criminal-accountability institutions to avoid being punished. As a result, the degree of independence enjoyed by prosecutors and judges is seen as an important incentive for their action, given the enforcement costs inevitably associated with criminal accountability. Independent courts provide a credible third-party signal in the context of high information asymmetries, including about the validity of corruption charges, and may also safeguard against politicization (Stephenson 2003). Because judicial independence is a relatively vague concept, it is often translated into a few dimensions, such as different forms of selection (e.g. merit-based self-recruitment vs political appointments), tenure in office and removal of judges and prosecutors. Overall, it is assumed that the less control political elites have over the processes of hiring, retaining, promoting and removing criminal-accountability agents from office, the more credible criminal-accountability results will be (Butt and Schütte 2014; Della Porta 2001; Quah 2010; Ríos-Figueroa 2012; Van Aaken et al. 2010).

Another dimension of judicial independence, and one that is often overlooked, is the administrative autonomy enjoyed by courts and prosecutors' offices, which they may use to build capacity to fight corruption. In fact, highly autonomous judicial institutions that have produced significant criminal accountability over the last few decades have been capable of doing so largely based on such administrative latitude, as in the Brazilian Lava Jato investigation (Da Ros and Taylor 2022; Rodrigues 2020). But independence is not always conducive to increased criminal accountability. Interestingly, Maria Popova (2012a) notices that magistrates may fail to fight corruption precisely because they may be too independent: as they reap the benefits from a secure office and fighting corruption is inevitably costly, only a few magistrates may end up doing so. That is, because judges and prosecutors may be too insulated from society, they may express passivity in face of corruption allegations, failing to confront actors who may otherwise disturb their comfortable professional positions. Critically, Julio Ríos-Figueroa (2012) suggests that overly independent judges may themselves become corrupt, as they are subject to limited forms of accountability.

Legal capacity refers to the breadth and focus of laws that allow corruption to be fought from the prosecution office and the bench. Because criminal accountability generally demands strict adherence to legal rules, the law must minimally authorize prosecutors and judges to act in such cases so that their actions are not easily thwarted by the high courts. Ultimately, this means saying that the law matters. This does not mean that only the law matters, but that the legal framework contributes to shaping the diverse outcomes of criminal accountability that are observed empirically. Particularly, two types of legal capacity matter: criminalization and prosecution tools. First, the criminalization of corruption proper and of corruption-related practices means that explaining variation in corruption prosecutions and convictions has to take into consideration what types of corrupt behaviour the law of each country defines as a crime (Sousa 2002). In short, the more distinct types of abuse of power for private gain are defined in law as criminal behaviours, the more likely criminal accountability becomes. These include typical corruption practices, such as bribery and embezzlement, but also activities that are usually auxiliary to corruption, such as procurement fraud, illegal enrichment, irregular campaign finance provisions, racketeering and money laundering. Brazil's Lava Jato investigation is often considered a corruption investigation but has been largely based on money-laundering charges (Da Ros and Taylor 2022; Fontoura 2019).

Second, variation in outcomes of criminal accountability may derive from the existence of different tools that can enable prosecutors to investigate and charge corruption. Well-known examples include plea-bargain agreements, which have been pivotal in exposing large-scale corruption such as in Mani Pulite in Italy (Della Porta and Vannucci 1999) and Lava Jato in Brazil (Rodrigues 2020). Other, more controversial powers available to prosecutors may include the ability to conduct wiretaps without judicial warrants, as performed by Indonesia's Komisi Pemberantasan Korupsi (KPK or Corruption Eradication Commission), an anti-corruption agency with prosecution powers (Butt and Schütte 2014). Inversely, the absence or timidity of criminal accountability may be explained by the existence of immunity enjoyed by political elites, so that they cannot be held legally accountable for certain types of behaviours while in office or prosecuted before regular trial courts (Eggers and Spirling 2014; Reddy et al. 2020). In some cases, politicians may also have formal powers to stop investigations, as in the cases of a few Latin American nations where congressmembers have to authorize indictments to proceed against members of the executive and legislative branches, potentially providing a ‘legislative shield’ against prosecutions (Conaghan 2012; Mann 2011). Overall, the likelihood of prosecuting and convicting political leaders for corruption increases as the legal definitions of corruption and corruption-related practices become more encompassing, the prosecution offices are equipped with broader legal tools, and the legal safeguards enjoyed by former and current political elites are minimized.

Organizational capacity is about the flesh and bones of criminal-accountability institutions. If the law matters, it can do little if it lacks the ‘material support’ to litigation (Epp 1998: 23). Organizational capacity encompasses the set of tools and resources made available to or produced by different criminal-accountability institutions as they perform their functions. This includes the varying degrees of professionalization possessed by prosecutors' offices and courts, as well as their budgets, expertise, internal complexity and the technology at their disposal. James Alt and David Lassen (2012) suggest, for instance, that the number of US attorneys per state is positively associated with the number of corruption convictions at the subnational level. The presence of prosecution offices and courts with specialization in corruption and financial crimes, likewise, appears to have contributed to similar results in countries such as Indonesia, Romania and Brazil (Bütt and Schütte 2014; Kerche 2021; Mungiu-Pippidi 2018). Similar findings hold for temporary specialization (i.e. task forces, special prosecutors) in Brazil, Ecuador, Peru and the United States (Davis et al. 2021; Ginsberg and Shefter 1999; González-Ocantos et al. 2023; Harriger 2000; Rodriguez-Olivari 2020).

Interinstitutional coordination is the ability of otherwise separate institutions to work together. Given the institutional multiplicity that characterizes criminal accountability, many explanations of the varying intensities of criminal responses to corruption have acknowledged the profound effect coordination may have. Often this involves how closely judges, prosecutors and investigative authorities work together on corruption cases, as illustrated once again by accounts of Mani Pulite in Italy and Lava Jato in Brazil (Della Porta 2001; Rodrigues 2020). At times, there is even integration with intelligence agencies, such as the cooperation between Romania's Direcția Națională Anticorupție (DNA, or National Anti-corruption Directorate) and the country's intelligence agency (Stoian 2020). Inversely, the absence of coordination has been a contributing factor to suboptimal results in criminal-accountability efforts, according to a variety of studies about the Brazilian case (Aranha 2017; Carson and Prado 2016; Da Ros 2014; Taylor and Buranelli 2007).

Macro-level explanations

Macro-level theories ask why criminal-accountability institutions differ across nations and over time in their independence, capacities and coordination. Specifically, this last level of explanation asks about the decisions made by political elites to empower or disempower such agencies. Five primary explanations have been advanced in the literature concerning the roles of political regimes, political competition, support from civil society, corruption levels and international norms.

Political regimes matter for criminal accountability because democracies and autocracies are often expected to perform very differently in the matter of punishing wrongdoing by political elites. Whereas democracies would lead to increased criminal-accountability results emerging from the system of checks and balances (including independent courts), the concentration of powers that is typical of autocracies would incentivize ruling elites to quash corruption cases against them and their allies. Still, prosecutions and convictions for corruption do exist in authoritarian regimes (Carothers 2022). Accordingly, most recent empirical evidence based on analyses of countries such as China and Russia suggest that they serve primarily to reinforce processes of concentration of power of ruling elites, so that prosecutions and convictions target opponents and potential rivals disproportionately (Popova 2017; Zhu and Li 2020; Zhu and Zhang 2017).¹² Nonetheless, as investigations come closer to the political nucleus of the regime, the accusers, not the accused, are likely to become the new targets of accusations of abuse of power. This is illustrated by the aftermath of the Uzbek corruption scandal of the late 1980s, in which prosecutorial investigators Tel'man Gdlyan and Nikolai Ivanov accused members of the Soviet Politburo and were later accused of wrongdoing themselves (Maor 2004: 12–13). An important nuance to the scepticism of combating corruption in autocracies is provided by Christopher Carothers (2022), who explains how a combination of discretionary power (i.e. independent of pressures by ‘democratic-like’ institutions or competition) and state capacity resulted in effective anti-corruption reforms.

Political competition has been an important explanation for varying outcomes of criminal accountability, offering nuanced insights into the distinctions between democracies and autocracies. On the one hand, political competition helps to explain why political elites empower criminal-accountability institutions in the first place. According to an influential yet controversial explanation usually referred to as insurance theory, when a formerly dominant political elite perceives a threat to its grip on power due to escalating political competition, it may opt to strengthen judicial institutions. This proactive measure aims to shield the elite from potentially politicized prosecution after leaving office (Dixon and Ginsburg 2017; Epperly 2013). In this way, judicial empowerment and investigations, charges or convictions are not necessarily positively associated. That is the case because courts empowered under such circumstances would have been empowered precisely to enforce defendants' rights and to prevent politicized or arbitrary punishment promoted by new rulers. This is in line with evidence that in new democracies, judicial independence does not lead to more investigations of former leaders (Bahry and Kim 2021) nor to a higher level of indictment of ministers for corruption (Popova and Post 2018). However, insurance theory is challenged by studies suggesting that under certain conditions (e.g. high electoral volatility, lower levels of trust in the judiciary) more commonly found in developing democracies, short-term political competition or insecurity (e.g. risk of a presidential impeachment) might hamper the independence of courts (Aydin 2013; Helmke et al. 2022; Popova 2012b). So, in isolation, insurance theory does not provide an explanation for why prosecutions and convictions eventually occur. Therefore, what may account for the fact that some independent courts favour criminal accountability is the argument that underscores the unintended consequences of insurance provision over time. Once empowered, courts and prosecutors' offices may eventually respond less and less to their creators and instead become powerful creatures with their own interests – including a commitment to the enforcement of anti-corruption laws.

On the other hand, explanations based on the impact of political competition help to address why political elites struggle to weaken criminal-accountability institutions once these entities pose threats to them. Here, according to a well-known hypothesis of the judicial politics literature, the political fragmentation and gridlock that are typically associated with increased levels of political competition make it harder for governing elites to interfere politically with judicial institutions (Castagnola 2017; Ingram 2015; Ríos-Figueroa 2007). As a result, criminal-accountability agents and institutions find fertile terrain for their efforts in the relative absence of credible threats of interference coming from political principals. Political competition, in other words, helps to reduce enforcement costs, making it harder for those accused of corruption to target the ones accusing them.

Support from civil society may be useful both to induce reform and to help in enforcement efforts. Activists, social movements, non-governmental organizations and the media all have played significant roles in pushing for reforms that strengthen or shield existing criminal-accountability institutions. In some cases, reforms also resulted from episodic responses to particularly salient scandals, whereby hard-hit political elites met public pressure for reform with legal changes that empowered prosecutors and judges. Empowering prosecutors in the wake of scandals, for instance, was the story that led to the establishment of the special prosecutor in the United States (following Watergate); that enhanced the investigative powers of prosecutors in Brazil (following the 2013 public demonstrations); and recently sparked debate about prosecutorial independence in South Korea (Chisholm 2021; Da Ros and Taylor 2022; Ginsberg and Shefter 1999). At times, intense societal mobilization takes place in the absence of a specific scandal, as with the enactment of a 1990s law that criminalized vote-buying in Brazil and led to numerous prosecutions afterwards (Nichter 2021). Regarding enforcement, support or pressure from civil society mirrors the more general mechanisms of social accountability (Fox 2015), while the media is a powerful player that amplifies the visibility of criminal-accountability efforts in ways that inhibit political interference (Guarnieri et al. 2020; Mancini et al. 2017; Maor 2004). Likewise, the vibrant engagement of various social actors may fuel criminal-accountability agents even in hostile environments. The comparison between Peru's more proactive enforcement of Lava Jato's ramifications than in Mexico highlight precisely the role of civil society: despite the more robust legal capacity enjoyed by legal actors in Mexico and the higher enforcement costs in Peru (owing to the more extensive role of Odebrecht in the Peruvian economy), the latter – not the former – exhibited increased criminal-accountability outcomes precisely due to greater societal involvement in the case (Pimenta and Greene 2020).

Corruption levels matter because they help define the costs of enacting anti-corruption reforms that may empower criminal-accountability institutions in the first place. If corruption is widespread across a large number of actors, then the costs incurred by political elites to consent to such reforms are much higher than if corruption is rare. Providing prosecutors' offices and the courts with the tools to tackle corruption is hence easier if only a few politicians are ever expected to get caught as a result. Inversely, if empowering criminal-accountability institutions jeopardizes a plethora of powerful political figures, it is unlikely that any such reforms will ever pass unless they occur in response to exogenous shocks such as particularly salient scandals or pressure from the international community. This reasoning follows Matthew Stephenson's general hypothesis, according to which ‘there exists a “high-corruption equilibrium” (a “vicious cycle” or “corruption trap”) in which high corruption begets high corruption, as well as a “low-corruption equilibrium” (a “virtuous cycle”) in which corruption's rarity makes it easier to control’ (Stephenson 2020: 193). Just as corruption convictions cannot be considered adequate indicators of corruption, so the latter cannot be reliably inferred from the former (Treisman 2007). As Rasma Karklins summarizes,

the naïve observer may think that if just a few cases are prosecuted, this indicates a low level of corruption. In fact, just the opposite is likely to be true: the scarcity of prosecutions can indicate a very high level of corruption … if many corruption cases are brought to trial, this can indicate an active fight against corruption and a low level of it. (Karklins 2005: 35)

Lastly, international norms that criminalize corrupt practices have been thoroughly disseminated around the globe over the last three decades. There have been numerous conventions signed by different international organizations – for example the United Nations (UN), Organisation for Economic Co-operation and Development (OECD), European Union (EU), Organization of American States (OAS), Financial Action Task Force (FAFT), and so on – that have pushed countries in different regions to become more assertive in fighting bribery, money laundering, organized crime and a variety of practices associated with corruption, through criminal law. Even if these international norms are not aimed solely at criminalization, many of the capacities of criminal-accountability institutions over the last decades have benefited from this process. At times, political elites have incorporated international conventions into domestic law because of their perceived need to obtain membership in international organizations, such as the EU and OECD, or as a precondition for development aid (David-Berrett and Fazekas 2020). The implementation of these new laws may not be identical across countries but contributed to the spread of legal capacities in criminal accountability. At the same time, a vast anti-corruption industry has grown in influence, helping to promote the importance of the tools to fight corruption. This, in turn, serves multiple purposes that may help criminal-accountability institutions and agents to perform their tasks domestically. These include training judicial personnel, disseminating ‘best practices’ (e.g. the use of task forces in corruption prosecutions), facilitating international cooperation in corruption cases that go beyond the borders of a single nation and providing support against political interference (Acorn 2018; Johnston and Fritzen 2021; Katzarova 2019; Lacatus and Sedelmeier 2020; Sims 2011).

Discussion

Criminal convictions of political leaders have become an issue of global importance over the past two decades. While previously only a handful of unlucky politicians were ever charged with corruption, former presidents and prime ministers have increasingly not only been prosecuted, but also convicted for bribery, embezzlement, money laundering and electoral fraud. This article has reviewed a vibrant (if still recent) literature on criminal accountability which draws on a variety of disciplines, methods and countries. The findings of this literature mean that we have a better sense as to why politicians became targets of prosecutions in the first place, and why prosecutors and judges have become more aggressive on this front. To conclude, we point to five possible directions for future research on this extremely salient but still understudied topic.

First, as criminal justice systems move their records online, the availability of data from digitalized cases is becoming more widespread and the sheer number of cases is increasing. This may open up more possibilities for quantitative tests with improved empirical designs. This is any quantitative researcher's dream, for it allows the testing of several hypothesized explanations reviewed above. The richness of case data introduces the possibility of using panel data methods to account for individual and time-specific characteristics, discontinuity designs based on close electoral races, and the use of random assignment of cases to judges and judges' mandatory retirement ages. Another aspect that deserves more attention and raises the possibility of identifying potential bottlenecks in criminal accountability is the sequential analysis of various criminal-accountability stages – that is, prosecution, trial, conviction – which are usually examined in isolation from each other.

Second, some issues are difficult to fix simply by increasing the number of cases. One such issue is the content of the law, where minute differences across nations and legal systems make comparative research complex (Sousa 2002), despite the diffusion of international norms. Recognizing this issue, some of the literature now focuses on subnational research, which allows researchers to hold legal capacity constant for comparative purposes (e.g. Ang 2017; Da Ros 2014) or to compare units of analysis that exhibit a similar ‘point of departure’ (i.e. a ‘treatment’), such as the signing of the same international convention or an unfolding international scandal that involves different nations (González-Ocantos and Baraybar 2019; Pimenta and Greene 2020; Sims 2011). Similarly, more in-depth case studies are also in order. One case cited in our introduction, South Korea, is illustrative. The remarkable fact that ‘all former Korean presidents have faced investigations of corruption or embezzlement against themselves or close family members’ (Kalinowski 2016: 637, emphasis added) tells us not only about the extent of corruption in the country, but especially about the relative strength of the South Korean criminal justice system. Still, to the best of our knowledge, it is much easier to find academic research that addresses the former rather than the latter issue. In-depth studies on the workings of South Korean criminal-accountability institutions could add to the literature on intense criminal-accountability episodes in other nations (e.g. Mani Pulite in Italy, Lava Jato in Brazil). Inversely, the existing case studies in the literature seem biased towards positive outcomes (i.e. the incidence of criminal accountability, rather than its absence). This means that some of the reviewed explanations may be overestimated, and that more ‘negative’ cases should be added to this body of literature, to help explain why criminal accountability fails to happen. This is in line with similar realizations in the anti-corruption literature (Johnston and Fritzen 2021).

Third, we want to stress that, ultimately, the quest for a single, unified explanation that addresses such a widespread, diversified phenomenon may be futile. Different causal pathways are possible to explain why prosecutions and convictions of politicians vary across polities and over time. To that end, future research could develop typologies that tie together the different types and levels of explanation addressed here. In some cases, criminal accountability can derive from the interaction between international norms that found their way into domestic law, thus helping to build legal and organizational capacities within courts and prosecutors' offices, which in turn motivated prosecutors and judges to confront more aggressively corrupt yet competitive political systems. In other instances, it may be that a rise in political competition leads to increased politicization of appointments in courts and prosecutors' offices, so that criminal-accountability agents gradually behave more like their appointing principals. Still in others, it may be that civil society pressed a collusive political system from the outside-in and supported otherwise fringe actors to punish incumbents, temporarily shifting the playing field. As these three brief possibilities suggest, there may be numerous ways for integrating different levels of explanation to address variation in corruption prosecutions and convictions.

Fourth, given the global scale and concurrent timing of this phenomenon, another promising line of inquiry concerns the different mechanisms that may account for the impact of international norms. It may be that the provision of legal capacity through the ratification of international conventions plays a critical role. Alternatively, international norms may raise awareness about corruption, putting it on the agenda and helping to mobilize domestic actors to tackle it. Or it may be that international influences on domestic dynamics induce the training of local actors via transnational networks that later collaborate in cases with extraterritorial jurisdiction. Of course, all these may matter jointly, but future research may better indicate how and under which circumstances they affect varying levels of corruption prosecutions and convictions. Ultimately, the impact of international norms may be felt beyond domestic courts. Future research, for instance, may also incorporate decisions from courts in other countries and decisions from proposed international anti-corruption courts that may come to exist at a global or regional level (e.g. at the European Union).

A fifth direction for future research is to focus on the consequences of corruption convictions and the question of whether they help reduce corruption. Convictions may help to induce greater deterrence among other political figures. By contrast, convictions may put political elites in ‘survival mode’, such that they fight back against criminal-accountability institutions and agents, with potentially deleterious effects for both the independence of the courts and corruption. Seen from this angle, corruption prosecutions and convictions may be a useful tool for leaders to concentrate power gradually (Levitsky and Ziblatt 2018) as well as a source of political instability, similar to presidential impeachments (Pérez-Liñán and Polga-Hecimovich 2017). These directions have the potential to offer more empirical tests of existing theories and nuanced insights into several aspects of a multifaceted literature, thereby enhancing our understanding of the landscape concerning corruption-related legal action against political authorities.