Robust assessment of the institutionalist account of comparative development is hampered by problems of omitted variable bias and reverse causation, since institutional quality is not randomly assigned with respect to geographic and human capital endowments. A recent series of papers has applied spatial regression discontinuity designs to estimate the impact of institutions on incomes at international borders, drawing inference from the abrupt discontinuity in governance at borders, whereas other determinants of income vary smoothly across borders. I extend this literature by assessing the importance of sub-national variation in institutional quality at provincial borders in China. Employing nighttime lights emissions as a proxy for income, across multiple specifications I find no evidence in favour of an institutionalist account of the comparative development of Chinese provinces.