Abstract
This paper examines how restrictions on religious expression affect women’s educational attainment. We study the 2010 removal of the headscarf ban in Turkish universities, which had long limited access to higher education for visibly religious women. Our empirical strategy combines cohort-level variation in exposure to the reform with individual-level variation in the propensity to veil within a difference-in- differences framework. We estimate veiling propensities using an early wave of the Turkish Demographic and Health Survey and predict them for a later sample using both machine learning and parametric methods. We show that lifting the ban significantly increased educational attainment among women with a higher propensity to veil. These gains appear to be concentrated around the transition into and progression through secondary school. The results remain similar when, instead of individual-level propensities, we use pre-reform veiling prevalence at the province level as an alternative exposure measure.