Under what conditions can repression, generally assumed to induce social obedience through violence, consolidate authoritarian power in a non-violent way? I propose a novel theory of the effects of repression, accounting for regional heterogeneity of state capacity in authoritarian countries. I start from the premise that authoritarian leaders need to compete with regional governance structures of opposition to establish authority and therefore face dogged resistance to their policies of co-optation. Repression in such “restive” regions has downstream effects that complement autocrats' co-optation policies by rendering local communities legible. To test my theoretical predictions, I exploit the attempted coup in July 2016 in Turkey, a unique event that led to the replacement of pro-Kurdish mayors with government trustees in Turkey’s restive regions. Adopting regression discontinuity and difference-in-difference designs and relying on a unique dataset that combines administrative documents, protest events and procurement contracts, I demonstrate evidence supporting my theory.  This paper provides an alternative view as to when repression can generate favorable outcomes for authoritarian survival.

Do the electoral calculations of autocrats in competitive authoritarian regimes shape the distribution of potentially contentious public goods, such as education? If so, how are such public goods targeted and do they lead to attitudinal changes in the autocrat's favor? We offer a set of hypotheses linking autocrats' political strategies to the distribution of educational resources and also theorize their impact. We test our theoretical predictions in the case of Turkey, analyzing how the Justice and Development Party (AKP) regime has distributed a particularly contentious form of religious public schooling--imam hatip middle schools--across competitive municipalities. Using a regression discontinuity design to account for endogeneity issues, we show that, among municipalities with a high degree of electoral competition, the AKP disproportionately distributes religious middle schools to opposition-controlled districts after the 2014 municipal elections. We also find suggestive evidence of changed norms and beliefs in these areas. Specifically, Turkish municipalities targeted with an increase in religious middle schools exhibit increased demand for subsequent religious education, greater youth identification with religious values and dress-codes, and diminished youth support for secular parties. The paper contributes to the literature on distributive and authoritarian politics by demonstrating that the provision of contentious public goods under competitive authoritarian regimes can follow a distinct "indoctrination" logic and that such a strategy can successfully shift societal norms in the autocrat's favor.

How can we explain precipitous declines in religiosity in the MENA region? Relying on religious cleavages in Turkey, we study the relationship between political polarization along the Islamist-Secularist cleavage and religiosity. Our results demonstrate that political polarization considerably decreases religiosity, and this association is robust to various measures of polarization and estimation models. While the effects of polarization are significant throughout Recep Tayyip Erdogan's rule, they have accelerated since the 2016 coup. We also maintain that these effects are driven by partisanship: while Secularists have become more distanced from religion, their AKP counterparts have become moderately more religious. Surprisingly, these polarizing consequences hold among both the youth and older individuals. We argue that our results reflect social identity dynamics and specifically the increased salience and polarization of ‘Secularist’ and ‘Islamist’ identities. This study shows how political polarization can drive social change in developing societies, initiating a short term secularization process. 

What are the underlying political motivations driving the construction of mosques in authoritarian countries? To address this inquiry, I rely on a unique dataset that meticulously tracks the annual issuance of construction permits for mosques in Turkey, along with their corresponding funding sources. Employing a regression discontinuity design, I provide evidence presenting the central government's strategic focus on constructing mosques in electorally competitive opposition districts, when compared with electorally competitive incumbent districts. However, such targeting patterns are not observed for mosques funded by municipalities or private entities. Further analysis reveals that the content of Friday sermons delivered in these mosques bears a resemblance to President Recep Tayyip Erdogan's speeches, particularly during electoral periods, when compared to the speeches of the opposition's leader, Kemal Kilicdaroglu. These findings shed light on how sacred places, such as mosques, can be manipulated to serve autocrats' political agendas.