“Back to the Kitchen”: Corruption Charges Against Women Heads of Government

Abstract

While much of the scholarship on gender and corruption suggests that women in political office are, or are perceived to be, less corrupt than men, in just the past few years corruption accusations against Brazil's Dilma Rousseff and South Korea's Park Geun-hye have made headlines and led to their impeachment. In this article, we argue women heads of government are actually more likely to be charged with corruption due to pervasive beliefs that women, by their very presence, corrupt public office. Using cross-national data, we first demonstrate that women executives are significantly more likely to be formally accused of corruption than their male counterparts. We then present case studies of Brazilian President Rousseff and Turkish Prime Minister Çiller to demonstrate the powerful role of gendered discourse in motivating suspicion and inflaming elite and public sentiment and thereby driving corruption charges. These findings make a substantial contribution to the literature on gender, leadership and the politics of corruption.

Résumé

Alors que la plupart des études sur le genre et la corruption suggèrent que les femmes occupant des fonctions politiques sont, ou sont perçues, comme étant moins corrompues que les hommes, les accusations de corruption portées ces dernières années contre Dilma Rousseff (Brésil) et Park Geun-hye (Corée du Sud) ont fait la une des journaux et ont conduit à leur destitution. Dans cet article, nous soutenons que, contrairement à ces épisodes, les femmes chefs de gouvernement sont en fait plus susceptibles d'être accusées de corruption en raison des croyances omniprésentes selon lesquelles les femmes, de par leur présence même, corrompent la fonction publique. En utilisant des données transnationales, nous démontrons tout d'abord que les femmes dirigeantes sont beaucoup plus susceptibles d'être formellement accusées de corruption que leurs homologues masculins. Nous présentons ensuite des études de cas concernant la présidente brésilienne Rousseff et le premier ministre turc Çiller pour démontrer le rôle puissant du discours sexiste dans la motivation de la suspicion et l'exacerbation des sentiments des élites et du public, ce qui conduit à des accusations de corruption. Ces constatations apportent une contribution importante à la littérature sur le genre, le leadership et la politique de corruption.

Introduction

The literature on gender and corruption has largely focused on how likely women who hold political office are to engage in corruption (Barnes and Beaulieu, 2014; Treisman, 2007). Studies have found that women politicians tend to be less corrupt than their male counterparts and to be perceived by voters as less corrupt (Breen et al., 2017). At the same time, in just the past decade, women heads of government, from Brazil's Dilma Rousseff to South Korea's Park Geun-hye, have made headlines by being charged with corruption and removed from office. Other leaders, including Argentina's Isabel Perón, Bangladesh's “Battling Begums” Khaleda Zia and Sheikh Hasina, and the Philippines’ Gloria Macapagal Arroyo were all also charged with corruption. Are women heads of government disproportionately targeted by corruption charges (Swamy et al., 2001; Breen et al., 2017)?

On the one hand, pro-women, benevolent sexist stereotypes could benefit women caught up in corruption scandals, as people may be inclined to question (the extent of) allegations (Murray, 2010; Thomas and Adams, 2010). However, beliefs in women's exceptional honesty and purity could also trigger greater support for corruption charges against women (Jones et al., 2021).¹ We argue that women in leadership positions are more likely to be accused of corruption due to pervasive beliefs about the proper roles for women in public and political life (Tomashevskiy, 2021). While corruption is often described as fraudulent or illegal conduct in academic research, in political processes and in popular use, corruption signifies the erosion or debasing of an entity or body politic. Since premodern times, corruption has been associated with sexual immorality, impiety, ideas considered corrosive to society (Knights, 2018) and impure or deviant behaviour (Kroeze et al., 2018; Wickberg, 2019). In these conceptualizations, corruption implies not only specific acts of misconduct but the violation or subversion of spaces and norms. Rather than investigate the relative tendency of women to engage in abuses of office, or the double standards or inconsistent expectations and associated evaluations women face in political life, we focus on the link between gender, perceptions of corruption and long-standing narratives that make women heads of government more vulnerable to accusations.

In this article, we propose a novel extension of role incongruity theory. Eagly and Karau (2002) explain that role incongruity occurs when “social perceivers hold a stereotype about a social group that is incongruent with the attributes that are thought to be required for success in certain classes of social roles” (Eagly and Karau, 2002: 574). In a wide range of social contexts, politics has been conceptualized as a “male-dominated space” (Bos et al., 2021) where women are perceived as trespassers. This frame lends legitimacy to charges of malfeasance by forming “cognitive and affective glue between accusations” (Sosa, 2019: 724). Allegations of corruption levied against women in office, therefore, will often appear epistemically credible, resonating with and building on embedded narratives of women as dangerous presences in public life. While women may benefit from benevolent sexism when running for office, including gendered associations with honesty and purity, political setbacks—like low approval ratings, economic decline or other crises, or simply the act of attaining high office—can trigger hostile sexism against a woman leader.² In other words, women's perceived incongruity with the highest office will not be a constant, but rather conditional, and articulated through and reified by allegations of corruption.

This research makes two important contributions to the literature on gender and corruption. First, using updated cross-national data on heads of government, we demonstrate that women are significantly more likely to be formally charged with corruption than men and that this relationship is modulated by women's political empowerment domestically. Corruption is complex, can come from a variety of places and changes over time, and corrupt acts themselves are notoriously hard to identify and measure as they are typically covert and illicit. We focus on official corruption charges, pursued by legal-formal processes. This is an appropriate measure for this study as we are interested in how leaders’ gender affects the propensity to be charged with corruption, whether that corruption itself is real or imagined.³ This data allows us to add an observational and global component to the current literature, which primarily relies on survey experiments conducted in high-income Western countries. While experiments have advantages in causal identification, they, nonetheless, often fail to capture the likely role of priming and sociotropic factors that are hard to manipulate. Even when priming questions are present, these are arguably not as strong as real-life political discourse where people more easily attach stereotypes to candidates. Furthermore, findings from high-income Western countries might not generalize across societies globally due to institutional and cultural differences. Focusing on actual political processes, across a variety of contexts, we can more directly examine the role that gender plays in corruption allegations against women. In particular, this allows us to examine how corruption allegations are impacted by variation in women's political empowerment cross-nationally.

Second, we introduce a theory to explain this empirical pattern, focusing on the process by which charges of corruption are levied. We explore the cases of Brazilian President Dilma Rousseff and Turkish Prime Minister Tansu Çiller to investigate how their gender affected the corruption accusations against them. We find that accusations against women heads of government are empowered by narratives of women as agents of corruption. In particular, we look to how gendered discourses animate the corruption charges, and if and how they emphasize women as out of place and potentially dangerous to the legitimacy of the office. Both cases allow us to assess how the proposed mechanism can explain real-world corruption charges. While the statistical analysis provides evidence for the frequency of these charges against women, and how they are modulated by domestic indicators of women's empowerment, the case studies provide illustrations of the proposed causal mechanism: how gendered discourse motivates and validates accusations, resonates with embedded narratives of women as subversive presences in public life and inflames elite and public sentiment.

This article proceeds as follows: first, we briefly review existing scholarship on gender and corruption as well as findings about perceptions of women in politics more broadly. Building on these findings, we derive our theoretical explanation that women heads of government are perceived to be corrupting the office itself, leading to an increased likelihood that charges of corruption will be levelled against them. We then present results from the statistical analysis demonstrating that women heads of government are significantly more likely to be formally accused of corruption than men. In the following section, we explore the proposed mechanism through brief case studies of corruption charges against Çiller and Rousseff. We conclude with a brief discussion and directions for future research.

Exploring Narratives: Gender and Corruption

In recent years, the question of whether women in public office behave differently than men, especially with regard to corruption, has received considerable attention from scholars. There is substantial evidence that women leaders are associated with both positive and negative gendered stereotypes that distinguish them from their male counterparts (Streb et al., 2008; Burden et al., 2017; Tremmel and Wahl, 2023). In particular, women in political office are widely perceived as less corrupt than men, and/or may be empirically less corrupt.⁴

Survey studies in a range of contexts have identified some level of stereotype endorsement about women as political leaders. Particularly relevant to this study, women are often associated with communal leadership traits, like co-operativeness, compassion, warmth, nurturing and kindness (Bauer, 2020a; Brescoll, 2016; Eagly and Karau, 2002; Huddy and Terkildsen, 1993; Koenig et al., 2011; Valdini, 2019). In some cases, these stereotypes can advantage women candidates, as they may be perceived as less corrupt than their male counterparts (Barnes and Beaulieu, 2014; Treisman, 2007).⁵

Other survey evidence suggests that voters may perceive women as marginalized within political institutions, or more risk averse, and thus less likely to engage in corruption (Barnes and Beaulieu, 2019). For example, a survey experiment found that the mere presence of a female candidate systematically reduced the probability that voters would express suspicions of election fraud (Barnes and Beaulieu, 2014). Armstrong et al. found that women finance ministers are more likely to be appointed when levels of corruption are high to signal cleanliness and that the office is above reproach (2022).⁶ On the other hand, Schneider and Bos show that stereotypes about women do not always apply to politicians in voters’ minds 2014).

There is also evidence that this stereotype reflects an empirical reality: women may be measurably less corrupt than men (Breen et al., 2017). Cross-national studies of representation in parliaments, for example, provide evidence that having more women legislators is correlated with lower levels of corruption (Dollar et al., 2001; Esarey and Schwindt-Bayer, 2019; Jha and Sarangi, 2018; Swamy et al., 2001). Experiments have also found that women are less likely to engage in corrupt behaviour while in office (Serra and Wantchekon, 2012; Rivas, 2013).

Other work has added complexity to these findings, suggesting that the effects of gender on corruption may be modulated by factors including social, cultural and institutional context (Alatas et al., 2009; Goetz, 2007; Le Foulon and Reyes-Housholder, 2021; Schwindt-Bayer, 2016; Stensöta et al., 2015; Sung, 2003). Esarey and Chirillo find there is a stronger relationship between gender and corruption in democracies, indicating that, in some contexts, women experience “greater pressure to conform to existing political norms about corruption” (2013: 362).⁷

There is evidence across social and political contexts that women candidates are often at least perceived to be less corrupt and more honest (Barnes and Beaulieu, 2014; Esarey and Schwindt-Bayer, 2019; McDermott, 1998; Schwindt-Bayer, 2010). Valdini finds that there is such a strong association between women and honesty that “when an institution is seen as untrustworthy,” for example, following a major corruption scandal, “elites are more likely to increase women's inclusion in an effort to change the perception of” the institution (2019: 15). Taken together, the existing literature on gender and corruption, which finds that women political leaders are perceived as less corrupt than their male counterparts, suggests that women leaders ought to be less likely to be charged with corruption.

However, the positive association between women and honesty can backfire if an audience comes to believe that a woman has engaged in corruption. Indeed, the belief in women's higher honesty and purity could lead to enhanced punishment for perceived transgressions. Expectancy violation theory suggests that individuals are punished for contravening stereotypes associated with their group (Eagly and Karau, 2002; Rudman, 1998). Indeed, due to stereotypes about women's lower association with corruption, women who are perceived to have engaged in corrupt or dishonest practices may be punished for seeming to defy this stereotype (Barnes et al., 2020; Courtemanche and Connor Green, 2020; Eggers et al., 2018; Pereira, 2020; Reyes-Housholder, 2020). Reyes-Housholder finds that in more corrupt contexts in Latin American countries, women presidents receive lower approval ratings. She argues that due to “pro-women” stereotypes that women are less corrupt, “female presidents often face higher standards for their moral leadership. When corruption accusations emerge implicating female-led administrations, ‘pro-women’ discourse may backfire, and the opposition can resurrect latent doubts about women's ability to govern” (2020: 540).⁸ There is evidence from other fields as well: women tend to receive more severe punishments following ethical violations at work (for example, Kennedy et al., 2016). However, punishment for violating expectations would suggest that these women leaders did in fact engage in illicit or fraudulent conduct. In other words, for expectancy violation theory or double standards to be a sufficient explanation, charges would need to be motivated by the belief that these women leaders violated the expectation by engaging in alleged financial or legal misconduct. However, in many cases, as we show below, charges are built on trumped-up allegations, and the investigations are more focused on defaming and launching sexist attacks against the leaders than uncovering financial or legal wrongdoings. The frequency of these charges, therefore, may be orthogonal to empirical patterns of actual corrupt behaviour.

Given these mixed findings, we ask, are women politicians disproportionately charged with corruption? We find that, in contrast to the logic of much of the existing literature on gender and corruption in politics, women heads of government are more likely to be charged with corruption.

Our research builds on work by scholars like Reyes-Housholder (2020), who finds that women presidents receive lower approval ratings in more corrupt contexts, specifically in Latin America, because they face a backlash when they fail to live up to higher moral standards. Our theory points in the same direction as Reyes-Housholder's work; however, we identify a different, albeit complementary mechanism. We argue that corruption in these contexts is not only (or even primarily) about women being held to a higher standard regarding malfeasance, but a manifestation of women's perceived incongruity with higher office. Even when men are engaging in corruption in these contexts, their gender is not conceptualized as antithetical to the space they are operating in.

Our argument is not just about perceptions of corruption, but whether women leaders are more likely to be punished for perceived corruption via investigation and possible removal from office. We therefore extend the literature on leadership, gender and corruption by accounting for not just who may be perceived as more corrupt, but which leaders are more likely to face tangible consequences for such perceptions. This is especially important given divergent findings in the literature about whether gendered standards and perceptions ultimately hurt women politicians or not, with several recent studies finding that gendered perceptions are ultimately not harmful for women politicians. Brooks (2013), for example, finds that gender stereotypes do not harm women candidates, while Teele et al. (2018) find that, ceteris paribus, most respondents prefer women candidates.⁹ Indeed, some even find that gendered stereotypes about the honesty and trustworthiness of women advantage women political leaders in contexts with higher corruption (Funk et al., 2021). We demonstrate that gender biases inform corruption charges and cause tangible harm to women political leaders.¹⁰

Gendered Charges

Existing scholarship does not provide clear indications of whether women heads of government are more or less likely to be charged with corruption, but the focus of much of this work is on legal malfeasance rather than the political process by which charges are levied. Levying charges suggests not just that a figure has engaged in malfeasance but also that an audience is motivated to punish them publicly for the behaviour. We contend that perceptions of women occupying these offices as inherently subversive can easily be triggered and thus substantially increase the likelihood that they will be charged with corruption.¹¹ In this way, corruption charges can serve as a gendered mechanism for redressing perceived gender-role incongruity (Eagly and Karau, 2002; Rudman, 1998).¹²

As previously noted, there are challenges to conceptual clarity in the study of corruption. Although “corruption” is often used in scholarship to narrowly describe particular illegal acts in public office (often of financial nature), this is not how the term is primarily used in popular and political discourse (Kroeze et al., 2018). As Sofia Wickberg notes, “‘corruption’ comes from the Latin corruption/corrumpere–destroy or ruin,” and the Oxford English Dictionary includes definitions such as “the destruction or spoiling of anything, especially by disintegration or by decomposition with its attendant unwholesomeness,” as well as “the perversion of anything from an original state of purity” (Wickberg, 2019; Oxford English Dictionary, 1989: 141). This definition is common in popular use—for example, when people describe the “corruption of youth” or the “corruption of innocence.” And while we do not intend to engage in etymological debates, this origin and connotation can be found across many languages with Latin influences. We argue that, as opposed to technocratic and objective, corruption is conceptually socio-political, both in terms of what constitutes it and how it is perceived, investigated and charged. While scholars tend to focus on legal-rational definitions, the definition relating to ruin, destruction and a subversion of norms is inextricably linked to women heads of government's vulnerability to corruption charges. These leaders are more vulnerable to this rhetoric because they are easily cast as corrupting the political space and are subject to salient socio-cultural narratives of women as temptresses and subversives.

The political sphere has often been defined in feminine terms, in relation to the men who control it (Sjoberg, 2009: 44).¹³ As Wilcox explains, “the ‘imagined community’ of the nation depends upon the homosocial relations of men to protect the nation-as-women's-body against foreign incursion” (Sjoberg, 2009: 37). Feminist theorists (Kittay, 2019; Okin, 2013), for example, have long interrogated the division between the public (the political) and the private (the familial), noting that the public and political spheres are contrasted with the feminine. As Wright explains, “The public sphere was marked originally as an exclusively male preserve and the private sphere of the home as that to which women were relegated” (Sjoberg, 2009: 193).

Even as more women are shattering glass ceilings, there is a lingering unease with women occupying these leadership roles. Scholars have found that political elites may be more likely to nominate women for political positions to signal trustworthiness (Funk et al., 2021; Valdini, 2019), but we argue that by virtue of attaining positions of heads of government, the image of these women's inherent honesty can easily be transformed. In her book Women, Power, Politics, Sylvia Bashevkin coins the phrase “women plus power equals discomfort” to describe “social unease” about women inhabiting leadership roles in the political arena (Bashevkin, 2009: 27).¹⁴ Media and popular discussion about women leaders often focus on their clothing, romantic relationships, age and other attributes associated with gendered stereotypes as a means of conveying this unease with women inhabiting leadership roles. “The combination of plus-perfect expectations, followed by damning indictments that declare every [woman leader] has fallen short,” Bashevkin argues, “work to reinforce an underlying disconnect between women and political power” (2009: 11). While Bashevkin explores this phenomenon in the Canadian context, she notes that the women plus power equals discomfort phenomenon is one that “we share with many other political systems” (2009: 10). Indeed, if such discomfort surfaces in a developed democracy that is highly ranked on women's empowerment indices, it may be present in similar contexts too, as well as in societies where women's rights are less assured.

This unease creates a substantial barrier to women's entry and taxes women in political office.¹⁵ In the United States, for example, Sheeler and Anderson explain that “popular frames applied to women running for the presidency since the 19th century [erode women's] credibility by accentuating the oddity of a woman as president” (2013: 242). In this context, women in political leadership are not seen simply as out of place, but as debasing political life. This is articulated in Sarah Lewis's popular 1839 book, Woman’s Mission: “Women would risk succumbing to selfishness if they were to enter the public sphere of the state. With their nature ruined, they would thus cease to be the ‘potent agent for the amelioration of mankind’, leading to the degeneration of civilization” (1839: 48–9).¹⁶ Williams describes a similar phenomenon in the ascension of Australian PM Julia Gillard: “Women politicians are almost always seen as deviant due to being women in highly traditional masculine spaces” (2017: 552). Similarly, in an analysis of editorial cartoons depicting Malawi's first female president Joyce Banda, Chikaipa found that the commentary indicated “a president should be ideally viewed in terms of a father who will fend for and stand for the country,” and that President Banda was often thought of as an unsuccessful leader because of her “woman characteristics” (2019: 21). In analyzing the UK's second 2014 general election leadership, Harmer et al. found “Female success is presented in terms of the capacity to emasculate male rivals; male success is presented as a consequence of displays of strength, aggression, and masculinity” (Harmer et al., 2017: 937). Women in office are perceived as inherently guilty of subverting the integrity of the office, and charges of corruption connote this. As Sosa explains, “…Corruption is not only a judgement about ethics, but a judgement about belonging” (2019: 735).

From Eve and Jezebel to Lady Macbeth, Cersei Lannister, and the Marquise de Merteuil, the trope of the powerful, subversive femme fatale, seducing and corrupting men pervades movies, literature and religious discourse across a variety of socio-cultural contexts. While the salience of these narratives vary, these lenses have been applied to coverage comparing Hillary Clinton and Argentina's Cristina Fernandez to Lady Macbeth. In China, Fang illustrates through digital ethnography, discourse analysis, and interviews “that a cultural script derived from the Chinese storytelling tradition of ‘women are a source of trouble’ serves as a contemporary narrative of corruption”(Chen, 2017: 67). We can see this in the literature on the Freikorps in Germany: as Hooper describes, “in their novels and memoirs, the male body was depicted as dry, clean, hard, erect, and intact, but always threatened by contamination from feminine dirt, slime and mire. Women's bodies were seen as messy, open, wild, and promiscuous—as engulfing swamps in which men could be annihilated. Women, communists, and the rebellious working class represented a ‘flood’ or ‘tide’ threatening to break down both masculine integrity and established social barriers” (2001: 83). These cultural referents articulate and reinforce an image of women's leadership as eroding legitimacy and subverting the “natural” order–fundamentally a corrupting force.

How can we reconcile the trope of the subversive woman with normative expectations that women will be more honest and the increasing representation of women cross-nationally? More specifically, how can women come to office under these conditions? We argue that these are, in fact, not competing narratives but rather two sides of the same coin. Like many other binary or trichotomous narratives regarding women (Sjoberg and Gentry, 2007), women are perceived as either self-sacrificing and beyond reproach or subversive and sexually deviant (Hametner et al., 2021). This discussion can also be situated in distinct manifestations of sexism that are contextually contingent. Benevolent sexism—attitudes or beliefs that women are naturally pure and generous—primarily benefits women in traditional or feminized roles. Alternatively, hostile sexism—which would view women in highly masculinized spaces as dangerous—would affect women in political leadership more. One can, and many often do, simultaneously hold benevolent sexist beliefs toward women in traditionally feminized roles and hostile sexist attitudes toward women that are seen to perform traditionally masculine roles.¹⁷ This can also help explain how women may be able to attain office and be subject to different types and levels of sexism once attaining it, especially once crises emerge (Pérez, 2022). These narratives are deeply embedded and can easily be activated or catalyzed (by a poor economy, for example). When these charges are levied against women, their veracity is reinforced by pre-existing narratives and, therefore, the accusations are perceived as intrinsically true. The narratives themselves, and how they appear and are understood, vary across cultures, as gender is ultimately a social construct. While striving to not essentialize women or cultures, we endeavour to demonstrate how pervasive these gendered narratives are across the globe.

In the next sections, we test three hypotheses associated with our theory:

• • H₁: Women heads of government are more likely to be formally charged with corruption.

• • H₂: Women heads of government are more likely to be formally charged with corruption in countries with lower levels of women's inclusion.

• • H₃: Rumours and accusations of corruption resonate with embedded narratives of women as subversive presences in public life, inflaming elite and public sentiment.

The statistical analysis in the following section tests H1 and H2. In the third section, we demonstrate the proposed mechanism described in H3 through studies of corruption charges against Turkish Prime Minister Çiller and Brazilian President Rousseff.

Empirics

Statistical analyses

We conduct an initial analysis using the Archigos dataset on leaders (Goemans et al., 2009). The dataset includes information on individual heads of state, such as the manner in which rulers enter and leave political power, the post-tenure fate of rulers and other personal information. The dataset also includes historical case descriptions that focus on challenges leaders experienced during their tenure, regardless of whether these challenges resulted in exit from political power. From these case descriptions, we created a new variable coding cases where formal charges for corruption have been brought against the leader (regardless of whether the leader was subsequently imprisoned), either during their tenure or up to five years after tenure ended (as the process of bringing charges often extends to after a leader leaves office). For this, we searched the Archigos case descriptions for leaders who were in power between 1945 and 2015, using the term ‘corrupt*’¹⁸ Leaders who were accused of corruption by foreign governments, by the public or by political rivals but not formally charged were not coded as charged. As discussed above, we were interested in allegations brought by domestic judicial or legislative institutions in order to capture formal charges of corruption. This is a good, albeit imperfect, measure, as we aim to assess the frequency of charges, and thus corruption's socio-political underpinnings, rather than the absolute extent of corruption or whether charged individuals were also found guilty, which would be the more technocratic definition of corruption.

Out of 2119 leader observations, we have 14 cases of women leaders charged with corruption (out of a total of 68 women leader observations) and 37 cases of men charged with the same crime (out of a total of 2051 male leader observations). Given that we base our coding of corruption on an existing dataset with limited information about each leader, it is possible that we are undercounting cases of corruption charges. Measurement bias in the dataset is, however, likely not systematic; in other words, there should not be any specific gender-related bias in how cases of corruption are counted in the dataset. Even if there is an underlying gender bias in the corruption coding, it would have to be quite large to explain the findings described below.

Descriptive statistics show a clear difference between the rate of women leaders and the rate of men leaders who were charged with corruption. Eighteen per cent of women were charged with corruption compared to only two per cent of men, and there is also a small difference between the rate at which women leaders were imprisoned, conditional on corruption charges (see Figure 1). To further probe these differences, we use our new measure of corruption charges as our dependent variable, and leader gender as our main independent variable, and conduct logit analyses on data spanning 1945-2015. The unit of analysis is the leader, and we include various controls at the leader level as well as the country level, along with standard errors clustered at the country level and specifications for country and year-fixed effects for the year the leader left office.

Figure 1. Corruption Charges and Post-tenure Fate.

As mentioned, our main models contain a battery of standard control variables, including regime-type measures from Polity (Marshall and Gurr, 2020), GDP growth (Bolt and van Zanden, 2014), and V-Dem indicators (Coppedge et al., 2015) of judicial constraints on the executive, executive corruption and media bias. All of these could impact readiness to charge and imprison corrupt leaders. We control for regime type using the country's Polity score during the leader's last year in office, as this could impact the readiness to charge leaders with corruption as well as the degree and manner in which leaders are prosecuted. According to a large literature, GDP growth can serve as a rough proxy for a leader's approval, which could also impact corruption charges (Fauvelle-Aymar and Stegmaier, 2013). We also control for judicial constraints on the executive using V-Dem's index on the extent to which the executive respects the constitution and complies with court rulings (and to what extent the judiciary is able to act in an independent fashion), and executive corruption as measured by V-Dem's index on executive bribery and corrupt exchanges. More corrupt, but judicially independent, countries could exhibit more cases of corruption charges and imprisonment as the systemic nature of corruption incentivizes misbehaviour on the one hand, and the judiciary has more leeway in bringing charges of corruption on the other hand. At very high levels of public corruption, we could, nevertheless, see a reverse effect, as leaders would have the means to stop such charges from becoming official. Media bias against opposition parties or candidates, also from V-Dem, could also positively impact corruption charges and imprisonment, as it is oftentimes through independent journalism that such cases come to light in the first place.

Beyond these, we add controls that proxy women's power in politics, as well as the networks and resources leaders have at hand to influence potential corruption charges and sentencing. We include a control for women's political empowerment in the country, an index incorporating fundamental civil liberties, women's open discussion of political issues and participation in civil society organizations and the descriptive representation of women in formal political positions from V-Dem. We also include a control for having a previous woman head of government. These variables, we argue, roughly proxy different socio-cultural contexts with respect to the acceptance of women in politics and thus the degree to which they are seen as subversive and charged with corruption. From Archigos, we include several leader-level variables, such as months in office, family ties to other leaders and mode of entry (whether regular, irregular or by foreign imposition). These proxy individual leaders’ resources in politics, which could also impact corruption charges. Descriptive statistics on all variables can be found in the Appendix.

We first test H1 that women heads of government are more likely to be formally charged with corruption. Table 1 shows that controlling for the above-mentioned country- and leader-specific characteristics, women are much more likely to be accused of corruption. Holding covariates at their means in Model 1, we find that the relative probability of being accused of corruption for women is 110 percentage points higher than for men. Models 2 and 3 show that the statistical significance of this effect holds when using country- and year-fixed effects at the point of a leader's exit. In the Appendix we explore a variety of other model specifications, including alternative controls such as leaders’ political orientation and approval ratings (Table A2), alternative measures for institutional context (Tables A3 and A4), covariates for year of entry and average over the years as opposed to covariates for year of exit (Tables A5 and A6), and use multinomial simultaneous equation models to ascertain the effect of gender on leader post-tenure fate, controlling for corruption allegations (Table A7). In all models, the size and statistical significance of the effect remain similar.

Table 1. Corruption Charges Models

Note: Standard errors in parentheses

^∗p < 0.05; ^∗∗ p < 0.01; ^∗∗∗ p < 0.001

To test H2 that perceptions of women as subversive forces increase the likelihood of women heads of government being charged with corruption, we use an interaction term with women's political empowerment. Table 2 presents two models, one with and one without country-fixed effects. To interpret the findings, we present predicted differences in the probability of corruption charges between men and women for the model with fixed effects in Figure 2.

Table 2. Corruption Charges Models with Interaction

Note: Standard errors in parentheses

^∗p < 0.05; ^∗∗ p < 0.01; ^∗∗∗ p < 0.001

Figure 2. Predicted Difference in Probability.

The confidence intervals around this difference make clear that gender has an effect only in contexts where women's empowerment is not high—confidence intervals do not overlap the 0 line in contexts of low women's empowerment, whereas they do in contexts of high empowerment. Thus, in contexts of high empowerment, these types of corruption charges are less likely to be catalyzing as women enjoy more acceptance in political life, and thus the gender gap in corruption charges is larger when men dominate the political system, as predicted by H2.

It is worth noting here that there may be a selection effect with respect to our findings: women who attain the highest office could be systematically more or less corrupt in practice than men. However, because corruption is necessarily covert behaviour that is not always discovered, it is very difficult (if not impossible) to measure and compare the objective “level” of corruption that leaders engage in (we do, however, try to proxy for this in our analysis by using the aforementioned V-Dem variable on executive bribery and corrupt exchanges). To the extent that it can be measured, there is substantial evidence that women in office do engage in less corruption than men. Thus if anything, a selection effect should bias in the opposite direction of our findings. Nevertheless, adjudicating whether women heads of government are actually more corrupt and are therefore appropriately more likely to be punished for it, or are more likely to be accused and punished for corruption because of their gender, is necessary if we are to believe the results above. To do so, and further explore the causal mechanisms explaining our findings, we turn to two case studies.

Case studies

In the following case studies, we explore the causal mechanism identified in H3, demonstrating how rumours and accusations of corruption resonate with embedded narratives of women as subversive presences in public life, inflaming elite and public sentiment.¹⁹ Drawing on media coverage of Turkish Prime Minister Çiller and Brazilian President Rousseff at the time they served in and left office, as well as secondary sources, allows us to illustrate the causal mechanism at work (George and Bennett, 2005).

These two cases were selected from the dataset because they provide variation on a number of key factors, but have the same value for the dependent variable (that is, both women were charged with corruption). We use process tracing to demonstrate that our theorized mechanism operates in a similar way across these two positive cases. The cases vary across both time (1993 to 1996 and 2011 to 2016) and geographical space and feature variation in institutional context. While Brazil is a presidential democracy with lower levels of executive constraints (6 Polity Score), Turkey is a parliamentary democracy characterized by more executive constraints for the duration of the case (7 Polity Score).²⁰ Most importantly, the cases show variation in women's representation and political participation writ large. When Çiller was in office, women accounted for only 2.4 per cent of the MPs in the legislature. Alternatively, at the beginning of Rousseff's second term, women accounted for 9 per cent of the lower house compared to 13.6 in the upper house (Inter-Parliamentary Union, 2020). More generally, levels of women's political empowerment were substantially higher (.83) in Brazil in 2015 than in Turkey in 1993 (.54) (Coppedge et al., 2015).²¹

Though the two short cases cannot definitively prove the proposed theoretical mechanism, they can provide preliminary support. To discern if allegations of corruption levied against women in office will appear inherently credible, resonating with and building on embedded narratives of women as subversive presences in public life, we explore the gendered discourses animating the corruption charges in these cases. Misogyny in this context would provide evidence of our mechanism if it emphasizes women as incongruous with the office or subversive to the body politic.²² We find that corruption charges in both cases were profoundly shaped by gender, and perceptions of women as out of place in, and potentially dangerous to, the office of head of government increased their vulnerability to the charges.

President Dilma Rousseff

Dilma Rousseff's gender augmented the plausibility of the corruption allegations and substantially increased the efficacy of the campaign against her. Building from the work of Sosa (2019) and Encarnación (2017), we investigate how corruption was used as a catchall by Rousseff's opposition and how she was seen as guilty by virtue of holding the office of the presidency.

Socio-politically, the office of the presidency in Brazil is deeply gendered. As Sosa explains, “Brazilian populist leaders such as Getulio Vargas in the 1930s and Lula da Silva in the 2000s often positioned their charismatic leadership in terms of father figures, not only in order to establish a relationship with the people, but also to position their political family” (2019: 18). Brazil's government is dominated by men at all levels and has the lowest levels of women in the legislature in Latin America (Inter-Parliamentary Union, 2020). Rousseff, however, still managed to earn a reputation as an effective, no-nonsense bureaucrat when serving in the Lula da Silva administration. In 2005, Rousseff became President Lula da Silva's chief of staff and was handpicked to be his successor at the end of his second term.²³

Despite her credentials, many viewed Rousseff as “Lula's puppet” (Encarnación, 2017: 85), “Lula in a skirt” (Sosa, 2019: 718), or “the woman Lula gave to Brazil” (Franceschet et al., 2016: 12). Both opponents and supporters cast Rousseff as a placeholder, expecting a continuation of Lula da Silva's policies. Before her presidential campaign, Rousseff worked with consultants on her image, but her candidacy challenged antiquated conceptions of women's roles. Rousseff was a former Marxist guerrilla, divorced, and as Omar Encarnación describes,

Brazilians thought her age (68 at the time of her impeachment), short hair, and professional attire were an affront to conventional standards of femininity, which emphasize youth, flowing locks, and voluptuous features. For the media, Rousseff's status as an unmarried divorcee supported the view that she was aggressive and lacked sex appeal. Like many other female politicians, Rousseff has also been criticized for being too serious, being too much of a wonk, and lacking charisma (Encarnación 2017: 89).

Later, in her second term, Rousseff's image would be juxtaposed with her vice president's wife, Marcela Temer, who was described as “beautiful, maiden-like, and a housewife” (Argolo, 2018). This comparison to Temer's wife articulates that the ideal or appropriate role for women is that of a housewife. Implicit in these and other statements was that Rousseff did not belong in public office by virtue of her gender.

In 2010, Rousseff was elected with 56 per cent of the vote. She maintained strong public support early in her first term, however, by 2013, the aftershocks of the Global Financial Crisis produced a recession in Brazil. Rousseff's popularity ebbed: while over 60 percent of respondents believed she was doing a “good/excellent job” in May 2013, less than 30 percent thought so by July (Jalalzai, 2015: 208). During her re-election campaign in 2014, she received a disproportionate amount of negative coverage compared to male candidates, and the Brazilian Democratic Party indicated that they would not accept the election results (Feres Junior and Sassara, 2018).

Ultimately, Rousseff was re-elected to her second term in 2014 by the narrowest margin since the country democratized. While most newly elected leaders experience a short honeymoon period, the disproportionately negative coverage of Rousseff continued after the election, only increasing until her ousting (Feres Junior and Sassara, 2018). Following a speech on International Women's Day on March 8, 2015, popular protests took hold against her. In addition to the poor economic climate, public anger was compounded by the construction of expensive sports facilities for the 2014 World Cup and 2016 Olympics.

Shortly following the devastating Zika outbreak in April, the Tribunal de Contas da União (TCU) announced its decision to delay approval of the 2014 budget. The TCU requested Rousseff explain why she had allegedly masked the size of the budget deficit. Rousseff was charged with delaying payments to the Central Bank during her re-election campaign in 2014. However, this was a common practice, used by both her predecessors and successors (Sosa, 2019: 717). That October, the TCU rejected the government's account. The opposition sent another impeachment request (which had previously failed), which was accepted by Speaker of the House Eduardo Cunha in December 2015. Some claimed Cunha was angry at the president for not protecting him against an investigation in the House Ethics Committee—later that August he would be charged with taking a five-million-dollar bribe related to Petrobras (Marcello, 2015). Members of the president's coalition soon turned against the president, while aides of Rousseff suggested “the impeachment charges were retribution for her refusing to pre-emptively pardon key members of Congress from charges linked to the investigation” (Sosa, 2019: 736).²⁴ In fact, on a leaked recording, Senator Romero Juca explained the impeachment was meant “to ‘staunch the bleeding’—i.e. shut down the corruption probe and save everyone else from jail” (Encarnación, 2017: 87).²⁵ Rousseff was scapegoated despite the fact that many of her male peers were undeniably guilty of at least equally serious (if not more serious) crimes. As Watts argued, “Rousseff's popularity nosedived as a result of corruption allegations…Despite being widely perceived as ‘one of the few politicians in Brazil not to accept bribes’” (Watts, 2016).²⁶ So why then was Rousseff charged with corruption, and why was the campaign to oust her so effective and ultimately successful?

The sexism that animated the charges was both instrumental and sincere, deployed by actors who saw Rousseff as unfit because of her gender and understood the mobilizing power of this rhetoric. Leaders within Rousseff's coalition, including Vice-President Temer, were never comfortable with a woman as president, and the opposition routinely implied—if not explicitly claimed—Rousseff was a “lesbian,” “communist,” “whore” and “subversive” (Encarnación, 2017; Sosa, 2019). This coincided with the rise of the Bancado da biblia, boi e bala or the bible, rural, and bullet benches in Brazilian politics, who as Macaulay describes, “combined moral panics about gender roles, sexuality, family and reproduction, sin, crime, and minority rights” (Macaulay, 2017: 136). This and other language indicate that Rousseff was not being penalized for failing to live up to heightened expectations of honesty or a double standard, but rather that, as a woman, Rousseff was corrupting the office. During the impeachment process, for example, Senator Magno Malta described Rousseff's presidency as a gangrene, claiming, “If we amputate the leg, we save the body” (Miroff, 2016). These and other statements also seem to indicate not only anger but disgust, an emotional response tied to feelings of aversion, particularly toward foreign objects that threaten to contaminate.

During the Chamber of Deputies’ vote, legislators held placards reading “Tchau Querida” (“goodbye darling”) (Encarnación, 2017: 88). Legislators justified their vote as a means to “‘save the county,’ ‘end communism,’ and ‘restore the foundations of Christianity’” (Encarnación, 2017: 88). In this way, misogyny served as the connective tissue between the charges, emphasizing Rousseff's gender as a threat to the office and nation. Rather than violating a higher standard of behaviour, the crime of delaying payments to the Central Bank operated as a vehicle through which Rousseff could be formally charged, and elites could articulate the more egregious socio-political violation—that Rousseff, by virtue of her gender, was corrupting the integrity of the office. In fact, in an analysis comparing the impeachment of Brazilian President Fernando Collor in 1992 to that of Dilma Rousseff, Pérez identifies a significant difference in the gendered articles in the proceedings, “suggesting a fundamentally sexist motivation behind the impeachment of Rousseff” (Pérez, 2022: 46).

Public opposition to Rousseff was similarly gendered and included sexualized fantasies of punishment. One of the more insidious examples was “sticker(s) showing Rousseff with her legs spread apart around their gas tank openings, sexually violating her image every time they filled up” (Krook and Sanín, 2020: 749).²⁷

When representatives voted in April, they claimed to impeach Rousseff “in the name of their families and God” (Feres Junior and Sassara, 2018: 228). By the end of August, the process was approved and Rousseff was removed from office, despite the fact that impeachable offences in Brazil must both violate the law and produce national instability, criteria that the allegations against Rousseff did not meet according to the findings in the Senate's proceedings. In their analysis, Dos Santos and Jalalzai found that “misogyny was an important element during the crisis that culminated in Rousseff's impeachment” (2021). When Temer ascended to the presidency, he appointed an exclusively white and male cabinet (Sosa, 2019).²⁸

Prime Minister Tansu Çiller

As in the case of Dilma Rousseff, Turkish Prime Minister Tansu Çiller's gender made corruption allegations more plausible and increased the efficacy of the campaign to charge her with corruption. Building on a number of secondary sources, as well as media accounts of Çiller during her tenure and her subsequent fall from grace published in Turkey and abroad, we demonstrate that Çiller's gender became intertwined with accusations of corruption.

In contrast with Rousseff, whose non-stereotypical attributes were highlighted by the opposition to suggest that she lacked sex appeal and was aggressive, Çiller was painted as stereotypically feminine, typically depicted in a Western-style skirt suit. Yet in a similar way that Rousseff's gender was used against her, Çiller was eventually seen as violating the bond she had with women voters by becoming part of a coalition with her Islamist party opponents. When she entered office, Çiller's gender represented a fresh start for the country and a more youthful, pro-business and Western-oriented approach to policy. Turkish headlines declared “Ours is the Prettiest” and “It's a Festival for Tansu,” as her ascension represented the setting aside of the old male guard. A prominent Turkish businesswoman commented that “there is an awakening, a feeling of independence and that we're catching up with the West. Çiller is a very good role model, and a big encouragement to younger people to aspire to new things” (Pope, 1993). Yet by the time she left office, Çiller was perceived as having violated her femininity and pro-West approach by aligning herself with the Refah Party. Once a feature of her appeal as a candidate, her femininity was instead used to depict her as mendacious, grasping and corrupt.

Tansu Çiller, Turkey's first (and so far, only) woman PM, served in the role from 1993 to 1996. Çiller, who served as both Deputy PM and Minister of Foreign Affairs between 1996 and 1997, faced several allegations of corruption during and after her time in office. In 1996, the Turkish parliament opted to set up a special commission to probe corruption charges against her, claiming that she had improperly tampered with the privatization of two companies. In February 1997, parliament voted against sending the investigation to the Supreme Court (The Irish Times, 1997; Couturier, 1997). In June 2000, she was again cleared by parliament, this time for charges that she had siphoned money from a covert intelligence fund used to finance her party (Deseret News, 2000).

Like Rousseff, Çiller's public self-presentation challenged traditional Turkish gender norms. Çiller embraced a more modern form of femininity to appeal to voters. Her image as a professional, highly-educated young woman who had earned a graduate degree in the United States was closely linked to her political views as a secularist and a modernizer (Bennett, 2010; Kesgin, 2012; Inal, 2017: 110, 38, 17–18). In keeping with this image, Çiller campaigned vigorously against Islamist opposition party Refah in the 1995 elections, contrasting their positions with her secular politics. Çiller pledged to defend Turkey's secularism against Refah. Her campaign argued that only Çiller was courageous enough to oppose Necmettin Erbakan, leader of Refah, and that a vote for a third, centre-right opponent was a vote for Islamists. A campaign newspaper ad declared, “Mothers, sisters, brides, girls, fiancees… I warn all of you. The Welfare Party cadres of Yilmaz are against the rights Turkish women have gained. Don't forget that the votes you give to Yilmaz will help the Welfare Party cadres…In turn, I have shielded my bosom from those who want to take back the rights Ataturk gave our women” (Arat, 1998: 16).²⁹

At the same time, from the beginning of her tenure, Çiller's gender was painted as a corrupting force rather than evidence of her purity. Analysts of Turkish politics have noted that gender played an important role in how she was perceived in stark contrast with many of her male peers. While Çiller's political mentor Süleyman Demirel was depicted as a deft political actor with a “moderating and conciliatory style,” Çiller was described as having a “lust for command” (Bennett, 2010: 131). Çiller was also subject to chants of “Tansu, back to the kitchen,” perhaps the clearest articulation that she was not seen to belong in the public sphere (Watson et al., 2005: 70).

When Çiller was selected by the True Path Party as candidate for PM, just 12 of the party's 1,169 delegates were women. One party member claimed that “a woman cannot be a prime minister; a woman cannot even lead a funeral prayer!” while Çiller was campaigning; according to Arat, “others said their party cannot ‘shoulder’ a woman president; they want a leader they can (literally) carry on their shoulders” (Arat, 1996). A political cartoon that ran in Milliyet, a Turkish daily newspaper, depicted Çiller seated with her legs crossed, a position viewed as improper for a Turkish woman, in front of her political mentor, President Süleyman Demirel, who was telling her “not that she could not be prime minister but that she could not be a woman.” Arat writes, “Though she appealed to women when she needed their votes, [Çiller] acted like a man in politics” (Arat, 1996).

Çiller's party won the second-largest share of seats in parliament in December 1995. In March 1996, Çiller and Mesut Yılmaz, head of the centre-right Motherland Party, formed a coalition government with Yilmaz as PM, but this government proved short-lived. Opposition leader Erbakan announced his intention to call for a parliamentary investigation into Çiller's role in the privatization of Tofas and Tedas. Although he was her coalition partner, Yilmaz—who had also called for an investigation into Çiller's role in the Tofas and Tedas privatizations during the campaign, voted in favour of the investigations (Meyer, 1997).

In a move that shocked many, Çiller then formed a coalition with Refah in June 1996, with Erbakan as PM and Çiller as Deputy PM and Minister of Foreign Affairs. Çiller's agreement appeared to be designed to avoid corruption charges in parliament, since Erbakan reportedly agreed to drop the charges in exchange for her support. Politicians in Turkey accused Çiller “of forging the alliance to shield herself from a series of inquiries into her personal finances” (Kinzer, 1997). The coalition government quickly drew opposition from secularist forces in Turkey, including the powerful military. In June 1997, the military forced the government out of power in a soft coup (Cinar et al., 2002; Bennett, 2010).

As in Rousseff's case, the sexism animating corruption charges was deployed by actors who both sincerely saw Çiller as unfit for higher office because of her gender and understood that her gender could be used instrumentally against her. Her opponents quickly portrayed her decision to form a coalition government with Refah as a venal political move, a charge that was amplified by her gender. The New York Times noted that “some Turks view Ms. Çiller, who has denied all wrongdoing, as cynical and mendacious” (Kinzer, 1997). Çiller was described in a 1997 New Republic article in harsher terms, as “the singularly ambitious, pathologically duplicitous former prime minister whose craven deal-making with the fundamentalists precipitated the whole crises in the first place”; the article was accompanied by a cartoon apparently depicting Çiller as a witch wearing a dunce cap (Bardach, 1997: 16–17). A prominent Turkish journalist, Mehmet Ali Birand, said of Çiller that “she will make a deal with anyone, anyone who will keep her in power” (Bardach, 1997: 18). A 1998 book by journalist Faruk Bildirici, described Çiller as a “scheming, stubborn woman driven by a lust for power” (News, 1998). Her decision was seen as a betrayal of her gender and a corruption of the office itself.

A political cartoon by Turkish cartoonist Musa Kart published in 1995 illustrates the ties between femininity and corruption. The image, which ran in leading Turkish weekly magazine Nokta, depicted Çiller dressed in feminine Western attire, in a white and purple skirt suit with matching hat, pumps, and manicure. Çiller is shown carrying a pile of elaborately wrapped gifts while spraying an aerosol can in the startled face of a man dressed in traditional Turkish attire. “I brought you bits of air from Paris” her speech bubble reads (Kart, 2019).

Allegations of corruption quickly became a wedge that political elites, including the opposition as well as former supporters, used to push Çiller out of power after her political star fell. Media reports increasingly focused on Çiller's “‘wealth, family, character, leadership style,’ and political dealings” (Bennett, 2010: 130–1). A Turkish Daily News editorial claimed “There have been widespread allegations that Çiller and people associated with her have been using their offices to gain undue advantage” (Turkish Daily News, 1997). The corruption allegations against Çiller were undoubtedly both numerous and serious—indeed, many believed at the time that they could lead not just to her fall from grace but to imprisonment or exile (Kinzer, 1997). But her behaviour was not substantially different from the male politicians who surrounded her. As Arat notes, Çiller “acted like the male leaders before her in a number of ways. Neither her priorities nor accusations of corruption surrounding her were novel to Turkish politics” (Arat, 1998: 18). An expert on Turkish politics commented at the time that “Turkey has proved reluctant to deal with corruption. It is significant that the National Assembly is handling alleged corruption in such a way” (Yackley, 2000). And although Çiller had been accused of corruption during her time in office, she had faced little push-back within her party around earlier corruption allegations. Rather, her “Unforgivable sin was the coalition agreement with Refah,” a political manoeuvre that was seen to conflict with her modern feminist image (Meyer, 1997).

Conclusion

In this article, we intervene in the debate regarding the relationship between gender and corruption by focusing on the process by which women are charged and why these charges tend to stick. We demonstrated that women heads of government are significantly more likely to be charged with corruption than their male counterparts, holding other factors equal. Through case studies of Brazilian President Dilma Rousseff and Turkish PM Tansu Çiller, we argue that perceptions of women in the public sphere as inherently subversive increase the likelihood that corruption charges will be levied against them.

We find evidence that the logic of the charges against Rousseff were buttressed by narratives of her as a subversive (Encarnación, 2017; Sosa, 2019). In Çiller's case, similar narratives around her political motivations for aligning with a different party in a coalition depicted her as venal and self-serving. In both cases, the corruption charges were strengthened by narratives of women's presence in political office as inherently subversive. Though allegations against Rousseff were strengthened by suggestions that she wasn't feminine enough corruption charges against Çiller were augmented by framing her leadership in hyper-feminine language. While we contend that the double standard mechanism is complementary, the discourse, especially surrounding Rousseff's impeachment, did not primarily emphasize her failure to live up to a particular threshold of honesty, one higher than that expected from men. Rather, politicians emphasized her incongruity with the office. The case study analysis is not a complete content analysis of media coverage but rather an effort to illustrate the causal mechanism suggested by the theoretical and empirical analysis. Future research can further interrogate these findings.

Though our argument centres on the link between gender and perceptions of corruption that make women heads of government more vulnerable to accusations, rather than double standards or the relative frequency with which women engage in corruption, it is important to address these as potentially alternative explanations for the evidence we find.

First, if women heads of government were generally more likely to engage in actual corruption (such as fraudulent or illegal behaviour), we would expect this to be reflected in a higher rate of charges. Though we do not test this alternative explanation directly, most research has found that women's representation in political office is correlated with less, rather than more corruption (Dollar et al., 2001; Esarey and Schwindt-Bayer, 2019; Jha and Sarangi, 2018; Serra and Wantchekon, 2012; Swamy et al., 2001; Rivas, 2013).

A second competing argument is that women may be elected to office when levels of corruption are higher, and their increased probability of being charged is simply a reflection of this. To account for this possibility, we control for executive corruption (as measured by the V-Dem index on executive bribery and corrupt exchanges) in our analysis. Additionally, in the Appendix, we present a multinomial simultaneous equation model to assess if women are more likely to be imprisoned than men, controlling for corruption charges. If women are also more likely to be imprisoned, this could mean that they are indeed more corrupt. We find, however, that women are still more likely to be charged with corruption, but that they are not more likely to be imprisoned, exiled or assassinated.³⁰

Another challenge to our argument could be that the increased rate of charges is a byproduct of women being held to a higher standard of behaviour, and they are therefore penalized more significantly for engaging in corruption. However, we argue that this explanation is contingent on a broad perception that these leaders are engaging in actual corrupt practices, and as we attempted to demonstrate, the charges were not exclusively or even primarily about legal malfeasance, but a perception that these leaders were, by virtue of their gender, out of place and thus corrupting public office. Though these latter two explanations could be framed as competing, we believe that our argument can be seen as complementary. In other words, women may be more likely to come to office in more corrupt states and still more likely to be charged because of the perception that they are out of place. Similarly, while we contend that these charges are often not about actual corruption, per se, different actors will have different perceptions of the leaders’ culpability—it may be that because women are perceived to be corrupting the office, both publics and elites are more likely to believe they engaged in actual malfeasance.

In this article, we build on literature investigating women at various levels of analysis cross-nationally—including national legislatures, cabinet posts, heads of government as well as various sub-national and local positions. While we should expect some continuity in the gender dynamics in public office, these will also vary in different social, political, and institutional contexts. As such, future research could investigate and theorize how corruption dynamics differ at different levels of governance and across socio-cultural contexts.

There are significant implications for this research regarding political stability and the prospects for future women leaders. First, following evidence presented by Reyes-Housholder (2020) and Carlin et al. (2020) that women leaders’ approval ratings are more negatively impacted by corruption charges, our findings imply that not only are women more likely to be charged with corruption, but they are also more likely to face more substantial backlash due to corruption allegations. This could more readily undermine the stability of women's tenure in office. Surfacing discomfort with women political leaders in contexts like corruption allegations where the gendered aspect is hidden is also important in ensuring that women are represented in the highest office around the world. Many women presidents and prime ministers in recent decades represent a first for their country, and as a result, a great deal rides on how they are perceived and treated by the electorate, the media and fellow elites. If women in office are persecuted more harshly for corruption, other women may observe this and choose not to run for office themselves. Voters and party leaders alike may also associate women leaders with electoral loss, concluding that it is not worthwhile to nominate more women for leadership roles (Bashevkin, 2009: 8–10). Indeed, if voters see women executives “as less ‘successful’ or ‘more corrupt’ than male—particularly on corruption issues where citizens thought these women would be more effective—public support for women in politics may erode” (Reyes-Housholder, 2020: 541). Ultimately, as Krook (2017) notes, “sexist hostility and intimidation have driven female politicians out of politics…young women appear to internalize these lessons in ways that reduce their own political ambitions” (84). Women in executive office are important sources of symbolic power. If their tenures are disproportionately associated with corruption, this will affect the prospects of women running for political office in the future.

Ultimately, we hope these findings contribute to a small but growing literature on how women heads of government fare in the highest office. This research is critical to our understanding of politics in a world where more women are breaking the highest glass ceiling by becoming heads of government. As research in this area advances—and more women serving in leadership positions yields more cases to study—future research, such as content analysis of media coverage of these women leaders, can further test and refine these explanations.