current projects

The democratic attitudes of economic elites—rather than regular people—are a far stronger predictor of future regime changes. This article finds that as economic elites turn against key democratic principles then democratic decline becomes increasingly likely and public attitudes—whether more democratic or authoritarian—have little bearing on the process. On the other hand, future democratization is much more likely when elites and the public are both committed to democracy. The evidence shows that whereas democratic decline is most often an elite driven process, cross-class coalitions do indeed appear to guarantee a stronger democracy. Distributional conflict helps explain these two diverging paths away from and toward democracy. In states with worsening economic inequality, wealthy elites are more likely to hold authoritarian attitudes and their preferences have a stronger relationship with regime changes. This argument is borne out by time series analyses of 84 countries over 25 years, from 1995 to 2020. 

If democratic backsliding occurs due to the gradual undermining of democratic institutions by incumbent presidents and prime ministers (Bermeo 2016; Svolik 2015), then which institutions are most vulnerable to manipulation by these self-aggrandizing executives? Inferring from the literature on authoritarian stability, the usual suspects are political parties (Brownlee 2007; Magaloni 2008), legislatures (Gandhi 2008), elections (Blaydes 2010), cabinet appointments (Arriola 2009), and advisory councils (Boix and Svolik 2013). I argue that cabinet positions and top-level advisor appointments provide would-be autocrats the quickest and most effective means to coopt elites and build a stable authoritarian coalition. Such appointments represent a mechanism for making promises of patronage that are more credible and more directly under the executive’s control than patronage facilitated by their party. In other words, a leader with expressly authoritarian ambitions would prefer to use the executive to build their coalition, rather than their party, because parties are accountable to a wider assortment of actors and are more likely to be governed internally by democratic norms (Mainwaring 2018; Kaufman 2011). Elites who join the executive’s coalition via such appointments have incentives to maintain their privileged access and therefore are more likely to support institutional changes that entrench the position of the current executive. 

In almost any regime context, the choice to defect from the party in government risks political marginalization and economic repercussions. In competitive authoritarian regimes, defections should be diminishingly rare when a dominant party controls the distribution of economic benefits and enjoys what Levitsky and Way (2010) call “an uneven playing field,” due to unequal access to resources, media, and the law. Yet, in Turkey, where the ruling AKP is exactly this kind of dominant party, hundreds of party elites and rank-and-file party members—defined as party representatives who manage the party’s regional and municipal headquarters—have defected to the opposition since 2017. According to a cost-benefit logic based on material incentives, which predominates in the literature (Reuter and Szakonyi 2019; Esen and Gumuscu 2020), we would expect loyalty. This paper provides a novel approach for explaining defections by party members whose actions defy economic incentives.

Methodologically, this article uses longitudinal analysis of social media behavior to observe the long-term motivations of political elites, and to distinguish between the material and ideological reasons for defections. Therefore, this approach makes both empirical and methodological contributions. It tests whether individuals who publicly express commitments to key democratic principles are more likely to defect from the AKP. I use network and sentiment analysis to build a qualitative dataset that maps individuals’ democratic values and motivations for leaving a dominant party, thereby helping us better understand how authoritarian parties collapse or remain resilient against defections.

This paper presents a text-based method for measuring illiberalism. Our approach can distinguish between political actors within the same country and is sensitive to different subtypes of illiberalism—such as anti-pluralism, nationalism, religious fundamentalism, or villainization of political opponents. The model combines word embeddings, trained on actor- and time-specific text, with dictionaries of liberal and illiberal terms. By applying this model to the cases of Hungary (2010-2022) and Austria (2008-2019), we show how it is possible to map differences between the government and opposition in six policy areas: education and culture, the environment, foreign policy, gender, immigration and citizenship, and social policy. By identifying illiberal subtypes, we can measure not only whether parties in power are illiberal on policy areas such as immigration, but also how they are illiberal.