The uneven reach of the state: A novel approach to mapping local state presence
提出一种利用机器学习算法，结合基础设施数据和居民调查数据，预测撒哈拉以南非洲地区地方国家存在感的方法，并验证其对发展结果的影响。
The ability of states to exercise authority often varies considerably within their borders, yet we lack reliable empirical measures of the uneven reach of states. In this paper, we develop a methodology to predict state presence at granular spatial resolutions and demonstrate the approach using data from Sub-Saharan Africa. We link a range of indicators of state presence, e.g., infrastructural data, with geolocated survey data of residents’ experiences with subnational governance. Then, we employ a machine learning algorithm that learns how the input variables relate to experienced state presence and extrapolates the predictions to all of Sub-Saharan Africa. We validate the predicted measure through a range of tests and document how local state presence influences development outcomes. • We present a novel approach to mapping local state presence. • A machine learning model trains on geo-located survey data to predict state presence. • The features comprise variables that signal state presence, e.g. road infrastructure. • The index is validated against alternative data on state presence. • State presence moderates the relationship between oil wealth shocks and conflict.