Witch Hunts? Electoral Cycles and Corruption Lawsuits in Argentina

Abstract

Courts prosecuting corruption serve a critical horizontal accountability function, but they can also play a role in moments of vertical accountability when voters can sanction corrupt candidates. This article documents the strategic use of corruption lawsuits, demonstrating the presence of an electoral cycle in filing new corruption accusations against politicians. Using an original dataset of daily corruption complaints filed in federal courts against members of Argentina's main political coalitions between 2013 and 2021, we document increased corruption accusations against and by politicians in the periods immediately preceding an election. A second dataset of daily media coverage of corruption accusations in two leading newspapers suggests that corruption is more salient before elections, offering politicians a temporal focal point to prepare and launch especially impactful lawsuits. Our findings shed new light on using courts for accountability and debates about the so-called ‘lawfare’ in Latin America.

A primary concern among those who study corruption prosecutions is the risk of seeing the rule of law morph into a ‘political weapon’ (Maravall 2003) or ‘lawfare’. The latter includes ‘resort[ing] to legal instruments, to the violence inherent in the law, to commit acts of political coercion, even erasure’ (Comaroff and Comaroff 2006, 30). Indeed, many studies document how incumbents politicize investigations to sideline opponents.¹ However, not all corruption politicization instances respond to lawfare logic or carry such extreme implications for regime openness. In this paper, we examine one example that takes place before the involvement of law enforcement, which is not necessarily aimed at political erasure and can be deployed against both incumbents and oppositions. We look at the decision to file corruption accusations against politicians in court and document that these accusations were timed strategically: it is normal for politicians to bring allegations to the attention of a prosecutor or judge during electoral campaigns.

Politicians can benefit from filing corruption lawsuits against their rivals because it can harm the reputation of their rivals, lower their approval rating, and possibly lead to a conviction. Electoral campaigns are a good time to launch accusations against rivals as voters have limited attention spans and prioritize new information (Healy and Lenz 2014; Huber, Hill, and Lenz 2012). However, making accusations come with risks. The authoritativeness and credibility of the judicial process depend on maintaining the ‘myth of legality’ (Bybee 2010; Gibson and Caldeira 2009; Nicholson and Hansford 2014; Nicholson and Howard 2003; Scheb and Lyons 2000; Zink, Spriggs, and Scott 2009). Opportunistic lawsuits, or investigations that are perceived as politically motivated, can quickly accumulate controversies, lose effectiveness, and embolden allegedly corrupt actors. Therefore, politicians must weigh the potential costs and benefits of involving the courts in investigating other politicians.

We investigate whether the ebbs and flows of the electoral cycle in Argentina track changes in the frequency of corruption accusations that make it to the federal docket. We use an original dataset of all cases filed in federal courts based in Buenos Aires (the Comodoro Py Courthouse) between 2013 and 2021, a period that covers eleven national elections and three administrations. Our data includes information on the timing of accusations, the identities of the defendants and accusers, and the crimes involved. Thus, we can identify complaints made against and by members of the main political coalitions that competed for office during this period.

Using this granular data, we examine daily variations in the filing of lawsuits against politicians before and after election day. Our local-linear estimates suggest that, on average, accusations against politicians are 26 per cent more likely in the days just before an election. We observe a similar difference using alternative estimators and for varying temporal windows. For instance, the number of days having a complaint against a politician is 24 per cent greater in the week before the election compared to the week after. When we analyse accusations against opposition and incumbent politicians separately, we observe that the impact is more pronounced for accusations against the incumbent (20 per cent compared to 15 per cent; p-values of 0.064 compared to 0.195). We also find more days with accusations filed by politicians just before election day (18 per cent). The estimates for accusations filed by incumbent and opposition coalition members are similar (0.08 compared to 0.09). Still, the precision of the estimates is higher for opposition politicians (p-value of 0.07 versus 0.112). We find only minor differences in the frequency and timing of accusations filed against or by politicians when controlling for the party in power – denunciation is not an activity dominated by right-wing incumbents. Finally, an analysis of a second original dataset of daily media coverage of corruption in two leading newspapers confirms that corruption is more salient prior to elections, offering politicians a suitable environment for their judicialization tactics.

This paper adds a new angle to the literature on corruption, which focuses on the macro-level determinants of corruption (Fisman and Golden 2017; Rose-Ackerman and Palifka 2016; Rothstein 2011; Treisman 2000); when and how politicians steal (Cooper 2019; Figueroa 2021; Gingerich 2013); and whether voters punish corrupt politicians (Anduiza, Gallego, and Muñoz 2013; Boas, Hidalgo, and Melo 2019; Chang, Golden, and Hill 2010; Chong et al. 2015; Dunning et al. 2019; Ferraz and Finan 2008; Klašnja, Lupu, and Tucker 2021; Pavao 2018; Weitz-Shapiro and Winters 2017; Winters and Weitz-Shapiro 2013).

First, this paper sheds light on the production of information about corruption by actors close to the action, a key step in the accountability chain (De Vries and Solaz 2017). Previous work by Balan (2011) studies when politicians leak information about corruption to the press. We take this further by clarifying when politicians find that acting on these leaks by involving the courts is wise. Our analysis calls attention to the strategic timing of corruption lawsuits and the merits and perils of activating horizontal accountability mechanisms over the electoral cycle.

Second, this paper complicates narratives about lawfare in Latin America by looking at who targets whom with corruption lawsuits and when. Leftist leaders invoke the term, including those investigated as part of the different national chapters of the Lava Jato inquiry, arguably the biggest recent shock to Latin American politics (Gonzalez-Ocantos et al. 2023). Members of the Left use the notion of lawfare to accuse incumbents, and the ‘Right’ more generally, of resorting to authoritarian legality to thwart progressive agendas (de Sa e Silva 2020; Mota-Prado and Rodriguez-Machado 2021).² While our findings about the timing of corruption lawsuits suggest that politicians rely on the courts to score political points, we uncover patterns of politicization that differ from lawfare. In particular, our analysis shows that filing corruption lawsuits is not an activity dominated by the Right. More importantly, corruption lawsuits, including strategically timed ones, are not more likely to be initiated by incumbents than challengers. Furthermore, the incumbent coalition receives more accusations, both overall and closer to the election. These patterns are in line with what one might expect from healthy checks and balances, although we discuss possible downsides in the conclusion.

Finally, this paper innovates empirically. We use an original dataset covering thousands of corruption allegations and exploit short-term variation to examine the role of electoral motivations. While our analysis only identifies a descriptive difference in the frequency of accusations before and after elections, we exploit high-frequency data, pre-determined election dates, and short temporal windows, making it unlikely that important correlates of allegations (such as actual levels of corruption or detection capacity) drive the results (Figueroa 2021). This approach stands in contrast to those that draw conclusions either from qualitative assessments and case studies of the politicization of corruption or from the outcomes of investigations. Our study has the advantage of using official records covering every judicial complaint filed against Argentina's elites rather than relying on subjective expert assessments or information from the media. Thus, we can separate the judicial tactics used by political actors from their efficacy or salience.

The Electoral Timing of Corruption Accusations

The way politicians activate horizontal accountability has implications for the quality of representative democracy. Decisions about whether and when to approach the courts with corruption accusations against public officials and political rivals are no exception. For politicians, these decisions raise important normative and strategic questions.

Many instances of corruption are justiciable because they match codified criminal behaviour. In those cases, politicians have an obligation to alert law enforcement. When politicians scrutinize the exercise of power in this way, they act in accordance with what their role prescribes and fulfil a critical accountability function. Behaving in a principled manner might also yield pragmatic dividends. After all, corruption accusations are a straightforward way of scoring political points against opponents. While the effects may be short-lived, the damage potential can be high. Indeed, scholars have shown that corruption accusations in Latin America dent presidential approval (Carlin and Hellwig 2019) and play a role in executive downfalls (Hochstetler 2006; Pérez-Liñán 2007). Similarly, lawsuits in Spain helped the Right reach power in the 1990s after spending a long period in opposition (Maravall 2003).

Yet, accusing a rival of corruption is not without risks. Crying wolf too many times can backfire, even when there are good reasons to go to court repeatedly. From a strategic standpoint, doing so may dampen the ammunition or upend a politician's reputation, turning them into a serial finger pointer of diminished credibility, especially because proving corruption is very hard (Bertossa 2003; Della Porta 2001; Martini 2015). Raising a complaint of corruption can also be dangerous for politicians who aspire to be in power because the tide may turn against them relatively quickly. This is particularly likely in contexts where corruption is systemic. Argentina exemplifies these perils. Fernando de la Rúa ran for President in 1999 on an anti-corruption platform at a time when the incumbent administration had been the target of multiple corruption accusations. But less than a year into his administration, a scandal that other politicians might have survived destroyed de la Rúa's political capital (Charoski 2002). Similarly, a few years later, a leading candidate in the 2005 race for a high-profile legislative seat in the City of Buenos Aires represented a party that again made anti-corruption a flagship issue. When rivals accused him of omitting to declare over $2 million (US) from his financial records, the campaign took a massive electoral hit.³ Finally, in contexts where corruption is systemic, it is hard to predict how an accusation that initially focuses on particular figures might evolve. Italian politicians subject to the wrath of Mani Pulite and their Brazilian counterparts during Lava Jato can attest to this. It is unsurprising that Presidents Berlusconi and Bolsonaro, who rode the coattails of both anti-corruption probes, sought to suppress the affairs shortly after taking office.

Therefore, the decision to insert corruption into the public debate through litigation is complex. The balance of risks suggests that politicians filing lawsuits against political rivals need to administer them carefully to maximize their normative merits, horizontal accountability value, and potential for electoral gain. But deciding when to go to court is far from straightforward.

Timing lawsuits just before elections has several advantages. Given that voters have, on average, limited attention spans and give weight to new information more heavily than older information (Healy and Lenz 2014; Huber, Hill, and Lenz 2012), it is during a campaign that an accusation can make the biggest impact. Voters' propensity to give added weight to new information may result from a combination of psychological biases (for example, recency bias) and well-known news consumption cycles. Work on Mexico suggests that voters increase their attention to the news in the period preceding elections (Marshall 2019), and evidence from the United States shows that congressional candidates are negatively affected by corruption accusations, mainly in the short term (Pereira and Waterbury 2019). Ferraz and Finan (2008) further demonstrate that the public release of corruption audits close to election day reduces the re-election chances of Brazilian mayors.⁴ Allegations may have a greater impact if they are not just declared by a candidate during an interview or debate but brought to the court's attention. Going to court signals the seriousness of purpose and enhances credibility.

For parties that cultivate transparency as a flagship issue, accusing a rival can be beneficial because it increases the electoral salience of corruption. It may also harden the anti-corruption stances of core constituents, boosting turnout. For a party perennially accused of corruption, counter-accusations may help retain sympathizers or undermine a rival's competitive advantage in this area, effectively muting corruption considerations. Crucially, electoral campaigns are leader-centric affairs (McAllister 2007) that pit one political figurehead against another. By contrast, day-to-day politics tends to be more diffuse, focusing more on party politics (for example, infighting), the interplay between institutions, and abstract policy issues. As a result, it is easier to launch effective corruption allegations, which are usually personalized, during a campaign.

Maximizing attention to instances of wrongdoing by filing lawsuits during campaigns may be more than just an electorally sound strategy; it could also enhance the accountability value of this tool. Timing accusations so that they surface when voters decide whom to vote for implies using the judiciary to activate vertical accountability. In weak rule of law contexts where electoral verdicts are often more powerful than legal ones, this might be a more effective strategy against corrupt political opponents.

Timing corruption lawsuits during campaigns can be advantageous but may also carry costs. Corruption prosecutions tend to get caught up in a debate between what Della Porta and Vanucci (2007, 846) call ‘judicialists,’ that is, those who emphasise the salutary effects of investigating the powerful, and ‘guarantors of the political system’ who caution against potential disruptions to democracy and the rule of law brought about by judicial involvement. This debate can resonate with voters. Voters value respect for procedures and fairness when evaluating judicial phenomena (Tyler 2001; Tyler 2004a; Tyler 2004b) and are responsive to whether the actors involved in the legal process behave in accordance with their role or turn the courts into a mere extension of politics (Ferejohn 2002; Gonzalez-Ocantos 2016). Performing and promoting the ‘myth of legality’ (Scheb and Lyons 2000) is crucial for preserving judicial legitimacy and the effectiveness of legal institutions (Bybee 2010; Gibson and Caldeira 2009; Nicholson and Hansford 2014; Nicholson and Howard 2003; Zink, Spriggs, and Scott 2009). But the electoral timing of lawsuits complicates this.

A recent study of the Lava Jato Operation by Gonzalez-Ocantos et al. (2023, 198–203; 215–25; 249–56) exemplifies some of these trade-offs. Investigations inevitably accumulate controversies and lose support when perceived as politically motivated or deviating from due process guidelines. Advocates for anti-corruption probes tend to find out that the intervention of judicial authorities can easily turn a valance issue into something highly contested (ibid., 172–4). Corruption probes create tension between citizens' demands for accountability and their partisan preferences (especially when prosecutors are perceived to behave in a partisan way). They also generate rule of law dilemmas; for instance, when decisions restrict the rights of key political figures or directly harm their electoral chances, including pre-trial detention orders or strategically timed searches. As corruption probes lose legitimacy or become divisive, they help embolden rather than weaken allegedly corrupt actors.

Launching a corruption accusation against a rival while simultaneously seeking to attract votes can deepen these lawsuits' political precarity and damage their potential for political use. These credibility and legitimacy concerns are, perhaps, more relevant for incumbents than oppositions, suggesting that the kind of strategic timing we explore in this paper is more likely to be practised by the latter. Incumbents who direct the law against their rivals during a campaign may have to deal with an opportunism charge and the more serious charge of intending to manipulate or pressure the courts at a critical time in the electoral cycle. This perception can undermine the credibility of the accusation and the incumbent's democratic credentials. By contrast, policing incumbent behaviour is one of the roles of oppositions in democratic systems, so these actors are less likely to encounter the same type of questioning. Relatedly, an incumbent fixated on wrongdoing by past administrations may seem less principled than an opposition politician who shines a light on recent incumbent behaviour, perhaps echoing media coverage of corruption.

While politicians may want to profit directly from an accusation by attaching their names to it, they can also diffuse authorship rights by enlisting political agents such as little-known associates, public interest lawyers, journalists, or NGOs to file complaints on their behalf. In Argentina, for instance, it is common for ‘independent’ lawyers and low-profile activists to file complaints. Politicians can also tip journalists with information about alleged crimes. The ensuing reporting often triggers new accusations.⁵

The implication of our argument contrasts sharply with the lawfare narrative that has emerged around the growing intensity of corruption prosecutions, especially in Latin America. The lawfare story makes one central claim: incumbents abuse their influence over the courts for political gain, especially by targeting the opposition to remove them from the electoral arena (Smulovitz 2022). Proponents of the lawfare narrative would expect incumbents to initiate corruption lawsuits more frequently and for opposition politicians to be more often the target of these cases. Instead, we hypothesize that both the opposition and the incumbent politicians have incentives to approach the courts to seek legal action during election season, with the opposition having slightly stronger incentives. This practice may serve frivolous purposes, such as discrediting rivals regardless of facts, or it may be used to bring attention to possible criminal behaviour by members of the opposite party.

Case Selection

Corruption is a relevant axis of political competition in Argentina. It is usually activated by non-Peronist candidates against Peronist rivals. Peronists have been in office for much longer and have been the subject of more investigations (Figueroa 2021; Manzetti 2014; Volosin 2019). Moreover, the historical sociological roots of non-Peronist parties make the defence of republican values key to their identity. Indeed, campaigning on anti-corruption platforms has helped these parties on at least two occasions (the 1999 and 2015 presidential races). Rebecca Weitz-Shapiro (2014) shows that middle-class voters, who tend to be anti-Peronist, punish candidates who use public office for private or electoral gain more severely than Peronists' poorer core constituents. While her focus is on clientelism, the logic also applies to grand corruption, incentivizing non-Peronist candidates to politicize it. But precisely because of this, Peronists sometimes try to neutralize the issue or even the score with their own corruption allegations.

The evolution of the party system in the aftermath of the 2001 economic collapse reinforced the centrality of the corruption/anti-corruption cleavage (Gervasoni 2018). This was a function of two factors. First, as left-of-centre Presidents Kirchner (2003–2007) and Fernández de Kirchner (2007–2015) tightened their grip on Peronism and sidelined other internal factions, a few Peronists abandoned the government – they mounted external challenges, citing corruption as a key differentiating characteristic. Second, the 2001 crisis severely weakened the Unión Cívica Radical (UCR), the oldest non-Peronist party, and led to the balkanization of that side of the political spectrum (Lupu 2016; Torre 2006). Former radicals formed their own parties (for example, Elisa Carrió's Coalición Cívica – CC). Political newcomers, including future president Mauricio Macri, did so too (Propuesta Republicana – PRO). The fight against corruption became a key strategy deployed by new non-Peronist parties to position themselves in an overcrowded field and by the UCR to mount a comeback. The politicization of corruption gained momentum as the Kirchners became embroiled in serious allegations (Manzetti 2014) and as non-Peronist parties overcame divisions to form competitive coalitions that challenged Peronist incumbents. One of those coalitions brought together the UCR, CC, and PRO. With Macri as the nominee, it dislodged Kirchnerism from power in 2015 (Lupu, Oliveros, and Schiumerini 2019; Murillo, Rubio, and Mangonnet 2016), only to lose it again in 2019, when Fernández de Kirchner returned as vice-president in a Peronist government led by Alberto Fernández.

Two characteristics of Argentina's criminal justice system enhance the appeal of using corruption accusations to score political points or alert voters about wrongdoing by opponents. First, prosecutors are allowed to investigate ex officio, but they usually react to accusations brought to their attention by others. Importantly, standing rules are maximally lax, meaning that individuals do not need to show a legitimate interest in a case to be able to file an accusation. Therefore, accusers face very low entry barriers. Second, courts invariably fail to make progress with corruption investigations. A report by the National Judicial Council shows that impunity for corruption was rampant. In the cases initiated between 1996 and 2016, only 2 per cent of defendants reached the trial stage, and less than 1 per cent were convicted. The performance is especially dismal among the twelve pre-trial federal courts based in the City of Buenos Aires (commonly known as Comodoro Py), which investigate corruption crimes involving federal officials, programmes, or money. The average duration of investigations was two years and nine months, with some cases taking more than ten years.⁶ This means that no one who files a lawsuit reasonably expects it to produce results on the criminal legal front, at least not in the short term. There is, therefore, every chance that actors go to court mainly for the expressive value of formalizing accusations. This is especially true for members of the opposition because incumbents are known for being able to negotiate case outcomes with judges.

Anecdotal evidence suggests that some corruption accusations are electorally timed. In 2013, non-Peronist parties formed a coalition to compete for legislative seats in the City of Buenos Aires. The mid-terms were scheduled for October, but the parties agreed to determine the composition of their lists during the mandatory primary to be held on the second Sunday of August. A few days before a televised debate, one of the candidates filed two corruption cases against the incumbent Kirchner administration, the first for illicit enrichment against the Army Chief and the second for embezzlement against the head of the Social Security Agency. Elisa Carrió, an anti-corruption crusader, timed these accusations to burnish her credentials and distinguish herself from internal competitors. When referring to her coalition partners during the debate, she said: ‘They are all stupid. They will not give everything they have to fight corruption.’ The debate quickly descended into a bidding war in which the others emphasized that they could also be trusted on this issue. Some diligently went through the cases they had filed against Kirchnerism, implying that their accusations were more credible than Carrió's: ‘We didn't base our complaints on newspaper clippings but on more than 300 pages of evidence.’⁷ An editorial published in a pro-government newspaper days later complained about the politicization of corruption by the opposition, longing for a bygone era when few corruption accusations were made against officeholders. There were ‘accounts [of government corruption], some more compelling than others […] but there were practically no cases filed in court […]. Today's media show […] is more political than ethical and […] generates more confusion than clarity.’⁸

More recently, Cristina Fernández de Kirchner expressed her belief that corruption accusations against her were strategically timed to maximize electoral damage. She cited a case from 2015, just before her term was about to expire, in which the opposition took allegations about currency manipulation to a federal court. As the trial was concluding, the former president told the judges: ‘You contributed to Macri's victory […] Five days after it was clear that the [runoff of the 2015] election would be between Daniel Scioli [representing Kirchnerism] and Mauricio Macri, only then did [two opposition congressmen] file the complaint […] On 17 November 2015, [five days before the runoff, judicial officers] searched the premises of the Central Bank. What was the goal? To generate a run on the currency, a devaluation, a disaster.’⁹ Conspiracy theories aside, and regardless of whether the case impacted the election, the opposition did file the complaint during the campaign.

Despite complaining about the timing of corruption lawsuits, Kirchnerists also resorted to this tactic. For example, two weeks before the 2017 legislative elections, during which the incumbent Macri administration sought to improve its minority status in Congress, a Kirchnerist legislator filed a case against the president and his minister of energy for influence peddling, arguing that Macri had asked the tax authority to shelve a complaint against the cabinet member. Interestingly, the congressman had gone public with the accusation three weeks earlier but only decided to take it to court on the eve of the election.¹⁰

Two years later, Macri was fighting for re-election amidst a major economic crisis that included an unpopular rescue agreement by the IMF. Kirchnerism again went to court. Days before the August primaries, an activist with close ties to the Kirchners filed charges against the president, the finance minister, and the former IMF chief, Christine Lagarde. The brief accused them of being part of a criminal conspiracy to use the international loan for electoral purposes and asked the judge to put the agreement on hold. The unmistakable goal was to shine additional light on one of the incumbent's key electoral liabilities. The accuser made sure to publicize the move in an online video.¹¹

Data and Main Outcome Variables

To investigate the electoral timing of corruption accusations, we assembled an original dataset with all the federal criminal accusations that entered the Comodoro Py docket between July 2013 and December 2021. Neither judges nor prosecutors intermediate the timeline of when cases appear in the dataset. A case appears in the docket when the accuser files the accusation, and information becomes publicly available immediately thereafter. Data comes from the federal judiciary's official information system (Lex 100) and is available on the website of the Judicial Information Centre (Centro de Información Judicial). We retrieved all available information for a total of 38,828 cases. For each one, the dataset includes the filing date, the crimes under investigation, the court selected to instruct the case, and the names of the defendants and plaintiffs.

To identify corruption allegations against or by prominent politicians and their networks within this universe of cases, we followed an approach that leverages the fact that most corruption accusations include a set of politicians as defendants. We first identified all accusations targeting the leaders of the two political coalitions that dominated Argentine politics between 2013 and December 2021: Fernández de Kirchner and Macri. Those appearing as co-defendants in these cases subsequently formed the basis of our lists of individuals who are members of each political coalition.¹²

We are confident that our list of defendants for each network contains the most important politicians active during the period of interest, including prominent cabinet members and other high-profile executive branch officials; presidents' closest relatives/aides; leaders of Peronist and non-Peronist parties; and ranking members of Congress. The final networks contain 374 co-defendants for Macri and 435 for Kirchner.¹³ Only a small number of co-defendants appear in our dataset as accusers: 11.7 per cent for Macri's network and 12.8 per cent for Kirchner's.

A potential concern with our method is that it might miss cases involving politicians who, while belonging to a coalition, were never accused alongside their leaders. In the appendix, we present our main results using an alternative list of politicians and public officials. To this end, we obtained the list of members of Congress that served during the period and the lists of public officials in the national and City of Buenos Aires governments. We obtained the latter from public payrolls and the sworn affidavits that officials are required to submit each year detailing their income and assets.

Our co-defendant classification, however, has the advantage that it identifies network connections regardless of their role, including politicians, businessmen, relatives, and civil society activists, among others.¹⁴ With this information, we reviewed all cases filed in federal courts to identify accusations in which at least one co-defendant of either president appeared as a defendant or an accuser. We use approximate string distance matching (Jaro-Winkler distance <0.1) to ensure that our classification identifies name matches even in the presence of middle and second last names and minor typos or spelling mistakes. The final sample includes 1,488 cases with political defendants. Table 1 presents the distribution of cases broken down by year and the defendants and accusers' political affiliations.

Table 1. Distribution of cases against Kirchnerist and Macrist networks

If the reader is not familiar with Argentine politics, these figures may come as a surprise because Kirchnerists are most often associated with corruption. By contrast, we reveal that both coalitions have been the targets of roughly the same number of corruption accusations. Temporal variation also suggests that complaints against a group increase when it is in power. For this reason, we further collapse our data into claims filed against incumbent and opposition politicians. Cases involving Cristina Fernández de Kirchner and her co-defendants during the period when her party was in power (between July 2013 and 10 December 2015, and after 10 December 2019 until 31 December 2022) and those involving Macri and his co-defendants during his presidency (10 December 2015 to 10 December 2019) are incumbent cases. Cases involving the parties outside of these periods are classified as opposition.

In addition to being confident that the list of defendants includes relevant political figures, we are quite certain that the universe of complaints is corruption-specific.¹⁵ Figure 1 presents the distribution of the crimes under investigation in cases against politicians from both networks.¹⁶ For the incumbent (opposition), approximately 86 per cent (77 per cent) of the cases include standard corruption offences, known in Argentina as ‘crimes against the public administration’. These include abuse of power, bribery, conflict of interests, money laundering, illicit enrichment, and fraud. Only in a few cases (6.7 per cent for the incumbent and 10.8 per cent for the opposition) we find no crimes specified in the complaint, mainly because the accuser simply asked the judge to investigate whether the defendant had committed any corruption crime.

Figure 1. Crimes under investigation.

Note: cases may include none (Not specified) or multiple crimes under investigation. See Appendix B for the full table.

Table 2 presents descriptive statistics for our main dependent variables. Corruption complaints against politicians in our sample are pervasive. We document a complaint against a politician 43 per cent of the time – roughly every other day. According to the lawfare narrative, we should observe more accusations against the leftist Kirchnerist coalition; more of these should materialize when it is in opposition, but that is not what we find. Accusations are evenly distributed between coalitions, and they are almost twice as likely when parties are in power. We documented 731 lawsuits against the incumbent and only 370 against the opposition. There are fewer days when politicians appear as accusers (15 per cent of all days in our sample). Indeed, the number of cases with a political accuser is much lower: only 328 cases, with a notable difference between Macrist (225) and Kirchnerist (119) accusers. As a result of the credibility concerns mentioned above, we suspect many accusations are linked to politicians, even if they are not filing them directly. The fact is that the Macrists and Kirchnerists received a similar number of accusations. Still, the Macrists seem more adamant than Kirchnerists about putting their names to lawsuits, which is highly suggestive of this strategy. After all, Macri's coalition emphasizes corruption; therefore, it makes sense that they are more likely to claim credit for activating judicial probes.

Table 2. Summary statistics by network, incumbency, and status

Note: Summary statistics correspond to the number of days with at least one case, with the exception of Total, which corresponds to the sum in the number of cases.

Research Design

We define the campaign season as the window consisting of the days before primary and general elections (Figueroa 2021). In Argentina, primary and general elections always occur on the second Sunday of August and are mandatory for voters and parties. For presidential races, a run-off is possible if no candidate secures more than 45 per cent of the vote or if the frontrunner secures more than 40 per cent and the difference between the frontrunner and the runner-up is less than 10 per cent. Run-offs are scheduled for November.¹⁷

Given the nature of electoral competition in Argentina, the incentive to use corruption accusations against members of rival coalitions is similar across primary and general elections, which is why the analysis pools all elections that took place during this period. In a context where primaries are competitive, such as in the US, one might expect more ‘friendly fire’ during primaries and more enemy crossfire in general elections.¹⁸ In Argentina, however, primaries are notoriously uncompetitive. A dataset of all district-level nominations for Chamber of Deputies elections between 2011 and 2017 shows that only 18.48 per cent of primaries included multiple lists, whereas the rest were uncontested (Skigin 2022). This is because the law mandates that all parties, regardless of whether they have multiple contenders for specific elected offices, must go through the primary process to qualify for the general election. Since parties rarely face internal competition during primaries, primary elections amount to an at-scale opinion poll prior to the general election. As a result, most politicians aim to garner more votes than their rivals in other parties to signal that victory in October is a fait accompli. The predominance of uncontested primaries suggests that it is highly unlikely that those accusing others of corruption will belong to the same party or coalition as the accused. In other words, the incentive to undermine the electoral prospects of a rival is as strong in the run-up to primaries as it is during the campaign ahead of the general election.

To explore whether corruption accusations increase in the run-up to the election, we compare the number of days with and without accusations shortly before to shortly after the elections (our unit of analysis is days). We take two approaches. First, we run a simple regression model (akin to computing a difference-in-means) for varying temporal windows around the election:

(1)$$Y_t = \beta _0 + \beta _1\;{\rm Campaign}\;{\rm Season}\;t + \varepsilon _t$$

where Y_t is a dummy variable that takes the value 1 if at least one corruption complaint was filed on day t against or by politicians from the incumbent or opposition parties, CampaignSeasont is a dummy variable that indicates whether the complaints were filed before or after the election, and ε_t is the idiosyncratic error term. We expect $\hat{\beta }_1$, the coefficient of interest, to be positive, indicating that the number of days with corruption accusations increases before the election. We estimate Equation 1 using ordinary least squares (OLS) regression and compute Huber-White robust standard errors (of the HC2 type).

We focus our temporal windows on one-day intervals within five and thirty-seven days before and after the election. Because we analyse eleven elections in total and only include business days in our sample, the number of observations in our regressions varies from 77 to 539. For instance, for a five-day bandwidth, we have three business days before Sunday (election day) and four days after. We selected this thirty-seven-day range because (1) elections are held on Sundays, so the last day to file a complaint before the election that can make it to newspaper headlines is Thursday (three days before the election); and (2) the modal distance between elections in our sample is seventy-seven days, thus a window of thirty-seven days splits the pre- and post-election working days in half, leaving one day in between.

Our second approach is to estimate a local-linear regression using the number of days away from the nearest election to compute triangular weights. Within a fixed bandwidth to/from election day, we fit two separate regressions for the days before and after the nearest election, respectively, and estimate the difference between the two intercepts. We report the bias-corrected confidence intervals and p-values developed by Calonico, Cattaneo, and Titiunik (2014). In the regression, this estimator assigns more weight to the days closest to the election, both before and after. Our analysis used a thirty-seven-day bandwidth, meaning that our local-linear regressions allow for a maximum of thirty-seven days before and after the election on each side of the discontinuity.

We present the results by grouping all elections for all politicians and incumbent and opposition defendants/claimants separately. In the appendix, we examine our results using the alternative classification of political cases, estimators (for example, n-degree polynomials), and bandwidths (for example, the ‘optimal’ bandwidth selection method developed by Calonico, Cattaneo, and Titiunik 2014).¹⁹

More Corruption Cases Against and by Politicians are Filed in the Run-up to Elections

Our argument is about accusers' incentives to time their accusations strategically. We argue that those who accuse politicians in court – especially politicians who accuse other politicians – may have a strategic incentive to file accusations shortly before an election to maximize potential electoral pay-offs. Yet testing this argument has challenges, mainly because accusers often delegate legal work to ‘independent’ lawyers and/or lesser-known activists, making it difficult to attribute direct connections to a political coalition. We thus test an observable implication: if those accusing politicians are more likely to do so shortly before an election, accusations against politicians should be higher at that time. As a result, we rely on the identity of the defendants for the main empirical examination of our theory.

Figure 2(a) shows OLS estimates for whether any case was filed in Comodoro Py against a politician for a large number of temporal windows, collapsing political coalitions (left panel), incumbents (centre panel), and opposition politicians (right panel) separately. Each point estimate in Fig. 2 comes from a separate regression following Equation 1, where we compare the number of days with lawsuits involving a politician before and after election day using one-day intervals for each regression. For instance, the first figure in panel (a) looks at accusations against incumbent and opposition politicians. For the five-day window, we compare the number of days with accusations filed between Wednesday and Friday before election day with the number of days with accusations filed between Monday and Thursday after the election. The number of observations in that regression is seventy-seven (seven days times eleven elections).

Figure 2. Difference in the number of days with a corruption lawsuit before and after the nearest election Panel (a) Panel (b).

Note: Outcome variables are dummy variables for whether at least one case was filed on that day for each category. Point estimates come from separate regressions for varying windows, as described in Equation 1. The top x-axis indicates the number of observations for each regression. Bars represent robust 95 per cent confidence intervals. Panel (a) examines cases in which politicians are the defendants; panel (b) examines cases in which politicians are the accusers.

In all cases, the point estimates are positive and become smaller as we move away from election day. We find similar results when considering all politicians and when focusing on accusations against the incumbent.²⁰ For these groups, the difference in the number of days with a complaint between the days before and after the election is positive (that is, more accusations are filed before the election) and statistically significant for days closer to the election. Our local-linear estimates confirm these results, shown in Table 3, panel (a). On average, a complaint against a politician is 26 per cent more likely just before the election (p-value <0.05) and 20 per cent more likely for incumbent party politicians (p-value <0.1). The difference is smaller (15 per cent) for opposition politicians and not significant at conventional levels. However, the difference between the two estimates is not in itself statistically significant, a point we discuss below (see Appendix A for full results).²¹

Table 3. Local-linear estimates

Note: Local-linear estimates with triangular weights for days with at least one case just before the nearest election. Confidence intervals and p-values are calculated using the “robust” procedure developed by Calonico, Cattaneo, and Titiunik (2014).

The results are similar when we examine trends focusing on accusations filed by politicians. Figure 2(b) reports the difference in means for one-day intervals for these cases. We also find more days with corruption lawsuits as we move closer to election day. Table 3, panel (b), reports local-linear estimates. A complaint filed by a politician is 18 per cent more likely just before the election (p-value <0.05) and 8 per cent (p-value >0.1) to 9 per cent (p-value <0.1) more likely for incumbent and opposition party politicians.

These results are consistent with the idea that politicians privilege the possibility of maximizing the benefits of horizontal accountability by filing lawsuits at a time when voters are paying more attention than usual to politics. The electoral impact of corruption allegations should be higher as we approach election day, so politicians disproportionately file cases in a relatively narrow temporal window.

While some of our results suggest differences in the timing of accusations against and by incumbent and opposition networks, the magnitude of the estimates is very similar. However, the differences are slightly larger and more precisely estimated for lawsuits against the incumbent rather than the opposition. This runs against the lawfare narrative (which suggests the opposite relationship) and, instead, is better explained by our argument emphasizing the credibility and legitimacy risks of strategically timed lawsuits. The nature of corruption means that accusations are more credible and can potentially accrue greater political dividends when raised against those in control of the state apparatus and are more likely to have misused public office for private or electoral gain in the recent past. By contrast, opposition members can usually be accused only of things they did under past administrations. Such accusations may fail to resonate with voters, appear extemporaneous, or stink of opportunism, making lawsuits against opposition politicians less valuable from a purely electoral standpoint.

Several placebo tests increase confidence that the main results reflect politicians' electoral considerations. Figure 3 reports local-linear point estimates with 95 per cent robust confidence intervals for several tests.²² One threat to inference comes from the possibility that judicial activity increases before the election for reasons unrelated to politicians' judicial strategies. To examine this possibility, we first look at whether the electoral cycles we document are specific to corruption cases or whether they apply more broadly. To this end, we replicate our main estimation by reviewing the differences before and after the election for non-corruption crimes; for example, drug and human trafficking and falsifying public documents or money. There were no significant differences in the number of cases that entered the Comodoro Py docket before and after the nearest election for these offences.

Figure 3. Placebo tests.

Note: estimates come from local-linear regressions. Bars represent ‘robust’ 95 per cent confidence intervals as developed by Calonico, Cattaneo, and Titiunik (2014).

Second, the number of corruption accusations may drop after the election if, for instance, politicians take time off once the campaign is over. Indeed, we document that politicians file more complaints just before the election, and accusations decrease in frequency after that. But politicians also file complaints for other offences that do not involve corruption crimes or other politicians. If they take time off, we should observe a drop in these accusations in the election's immediate aftermath. However, we find that the overall number of cases filed by politicians in non-corruption cases does not change discontinuously after an election.

Third, we consider the possibility that the results pick up a seasonal effect in filing corruption accusations rather than the strategic considerations of politicians. We test for this using a placebo to look at the number of cases filed before versus after the dates when elections would have taken place had non-election years been election years.²³ We find that there are no statistically significant differences in the number of cases before and after the dates of these placebo elections.

Media Coverage of Corruption Also Follows the Electoral Cycle

To complement the analysis of temporal patterns in case filings, we assess whether media reporting on corruption inquiries also follows the electoral cycle. The goal is to paint a picture of the environment in which the politicization of corruption lawsuits takes place, determining whether the campaign season is a moment during which the issue salience of corruption, as well as journalists' demand for this type of news, is higher. If this is the case, politicians can exploit and exacerbate the comparatively higher salience of corruption by going to court.

To this end, we built a second dataset with information from two of Argentina's leading newspapers, Clarín and Página 12. We selected these dailies because they have been among the most critical of the Fernández de Kirchner/Fernández and Macri administrations, respectively, and are, therefore, likely to pick up news about corruption by either group. We web-scraped the content of the front pages published up to thirty-seven days before and after each election between 2013 and 2021. To gauge the extent to which the dailies reported news about corruption inquiries, we identified whether the front pages included the keywords most frequently used in news covering corruption investigations and coded two variables: a dummy indicating whether any of the keywords appeared on the front pages and a count of the number of keywords.²⁴ Using the same estimation techniques featured in the previous sections, we then analysed whether Clarín and Página 12 were more likely to cover corruption news just before an election than the days and weeks after an election.

The results, reported in Fig. 4 and Table 4, provide additional rationale for politicians' tendency to go to court during the campaign season: news about the judicialization of corruption is more frequent in the days before an election. Before the election, newspapers are 14 percentage points more likely to include information related to corruption on their front pages than after the contest is over, and they include, on average, roughly 0.9 more corruption-related keywords.

Figure 4. Corruption news coverage before and after elections (Clarín and Página 12).

Note: Point estimates come from separate regressions for varying windows, as described in Equation 1. Bars represent robust 95 per cent confidence intervals.

Table 4. Local-linear estimates: corruption keywords

Note: Local-linear estimates with triangular weights for days with at least one case just before the nearest election. Confidence intervals and p-values are calculated using the ‘robust’ procedure developed by Calonico, Cattaneo, and Titiunik (2014).

While this does not prove that politicians' judicial tactics find a direct echo in the press, it does suggest that the electoral cycle in corruption lawsuits is part of, and likely contributes to, an informational ecosystem in which corruption is comparatively salient. As with corruption claims, these differences become smaller and/or lose statistical significance as we move away from election day.

Our simple keyword search picks up news related to corruption inquiries in general. A close reading suggests that crimes committed by politicians (corruption or otherwise, including using intelligence officials to spy on others or participating in torture sessions during the last military dictatorship) are amply covered on these front pages. Often covered is the decision of the judge or the prosecutor to act in response to one politician's accusation against another. Appendix E provides several examples, but two from 2013 are particularly illustrative. Clarín dedicated half a page to an accusation by Elisa Carrio (from the opposition party Coalición Cívica) against Cesar Milani, Cristina Kirchner's army chief, for illicit enrichment. According to our dataset, Carrio filed the accusation on July 16, and the front page is from two weeks later (July 30), immediately after the prosecutor pressed charges. Interestingly, this happened twelve days away from the primary election. We also find headlines covering politicians going to court to bring ‘new dirt’ on their political rivals. For example, eight days before the 2013 primary, Clarín dedicated space to an accusation filed by prominent opposition politicians against Cristina Kirchner for abusing her presidential prerogatives during the campaign.

Conclusion

Granular data from Argentina suggests politicians are more likely to file and be the target of criminal investigations in the days prior to an election. This pattern is consistent with politicians using these lawsuits to activate horizontal accountability. Understanding how and when corruption lawsuits become politicized has implications for debates about the rule of law. On the one hand, our findings could suggest that democratic competition produces favourable conditions for voters to access quality information about corruption when they need it the most to exercise their accountability functions. In other words, the results could point to possible synergies between horizontal and vertical accountability and undermine the lawfare narrative promoted by parts of the Latin American left. Yet they could also lend themselves to a more pessimistic interpretation.

First, the war of attrition can easily escalate, with accusations becoming an accepted part of campaign strategy. Underlining this risk, our data shows that the two main political coalitions attack each other in similar measures. The abusive use of corruption accusations can undermine rather than strengthen the role of this information on accountability. On the one hand, the political source and timing of the allegations may ultimately do little to boost voters' propensity to exercise their vertical control function. For example, in the absence of high levels of sophistication that enable voters to separate the wheat from the chaff (Weitz-Shapiro and Winters 2017), accusations could merely reinforce pre-existing attitudes, turning corruption into a highly partisan (as opposed to a valence) issue. An opposite but similarly dangerous outcome is that voters may become numb to the issue after being bombarded with allegations from all sides. As Pavao (2018) persuasively demonstrates, voters cannot use corruption to decide whom to vote for if everyone is allegedly corrupt. This may be why recent research finds that voters are responsive to information about corruption in hypothetical scenarios but not in real elections (Boas, Hidalgo, and Melo 2019).

Second, following Maravall (2003), if, instead of serving a neutral and permanent watchdog function, rule of law institutions are selectively activated when political actors need to score points, democracy can distort accountability, and accountability can distort democracy. Compounding things further, if those who happen to be the target of lawsuits at any given point react by constructing a lawfare narrative and casting themselves as victims, the result may be a perilous form of cynicism about the role of courts as horizontal accountability bodies.

With these considerations in mind, future research could move forward in various directions. One possibility is to study the actors who accuse politicians in court in more detail. While our theoretical framework and empirics focus on the motivations of politicians, other actors – including citizens, corruption watchdogs, and journalists – also approach the courts with accusations. Indeed, while we document an electoral cycle for lawsuits filed by politicians, the number of cases with political defendants is substantially larger. One limitation of our study is that we cannot distinguish whether politicians orchestrate the accusations filed by other actors. Also, we cannot determine if some of these accusations should be considered friendly fire. Because politicians often try to detach themselves from these accusations, it is hard to tell which ones are politically motivated.

Future research should explore different aspects of political competition that could reinforce or ease electoral cycles in corruption accusations. For example, higher levels of opposition fragmentation may make these attacks against incumbents more frequent because opposition politicians may find it attractive to stand out as anti-corruption watchdogs in a crowded field. And, if dislodging incumbents through traditional programmatic politics proves hard, the attractiveness of activating corruption as an alternative axis of competition likely increases (Maravall 2003). By contrast, the dynamics of coalitional presidentialism, typical of some Latin American countries (Chaisty, Cheeseman, and Power 2018), may temper the electoral use of corruption allegations. This outcome is especially likely if incumbents rely on oversized coalitions and/or shifting coalitions to pass specific policies. In those instances, the line between the incumbent and the opposition becomes blurred, and side payments are common, thus reducing the incentives for opposition actors to politicize corruption.

Finally, the vast literature on how voters process corruption information tends to draw pessimistic conclusions about the promise of electoral accountability (Dunning et al. 2019). In this regard, exploring whether the strategic timing of corruption allegations underpins these accountability failures is interesting. The calculations that lead some actors to politicize corruption during campaigns may not correspond to how voters interpret those developments. Does politicizing corruption during campaigns increase the informational value of accusations or diminish their credibility? Does it matter whether rivals own these allegations or diffuse authorship rights? And what is the most likely outcome: persuading swing voters or hardening core support?

The patterns in our data reveal the growing centrality of federal courts as arenas of intense politicking, which is part and parcel of a broader trend in Latin America. In recent years, the region has seen intense battles around historic waves of corruption prosecutions. The rhetoric of unyielding virtue of some political actors (the accusers) and the victimization cries of others (the accused) have proven extraordinarily disruptive to democratic politics in countries like Brazil, Ecuador, and Peru. Argentina is no exception. Whether these corruption accusations constitute lawfare, as many on the Left claim, lies in the eye of the beholder. But our data does provide some perspective on the extent to which this type of lawsuit has redrawn the boundaries of political competition.