While ethnic local autonomy has been considered as an institution of minority accommodation and conflict resolution, both why political leaders decide to introduce it in the first place and how their presence leads to political stability remain unclear. Drawing from the case of post-1949 China, I consider the granting of ethnic local autonomy in the context of authoritarian delegation. I argue that ethnic local autonomy, as an endeavor of local decentralization, allows the central leader to establish his supremacy over sub-national political elites while countering his inner-circle rivals. Using a novel index of elite connectedness and unique historical data on local political divisions, I find that Beijing is more likely to name EATs when the central leader faces a divided Politburo. EAT designations also tend to take place when the provincial Party secretary shares plenty of common knowledge more with the governor than with the central leader. Using an original dataset of district and county officials, I show that ethnic autonomous prefectures and counties are more likely to have ruling cadres appointed from the local communities than their non-EATs counterparts. Those in charge of sub-provincial EATs are also less likely to have worked with the provincial Party secretary. To extend the scope conditions, I assemble a cross-national dataset to study the presence of ethnic local autonomy in post-WWII authoritarian regimes. I conclude that one can consider ethnic local autonomy as an institution of agent control that can contribute to political centralization in multi-ethnic nation-states.