The idea that educational externalities affect local governments' incentive to provide such service was proposed over a century ago. Several studies have tested the relationship in the US, with mixed results. However, the theory has rarely been tested systematically outside of the US context. Paradoxically the unique household registration system in China makes the study of educational benefit spillover simpler to grasp theoretically and easier to handle empirically. A dynamic panel-data analysis that spans 17 years across 31 provinces of China suggests that a one-percentage-point increase in intra-provincial migration rate would decrease non-personnel spending in the following year by 23, 30, and 45 yuan per student for elementary schools, junior middle schools, and senior middle schools respectively. Narrowing the focus only on rural education funding, this study also shows that intra-provincial migration negatively affects rural spending per student by a similar magnitude. While migration and urbanization often bring about prosperity and efficiency gains, the substantial negative consequences of benefit spillovers under a decentralized education system should not be overlooked, especially for non-personnel expense and for rural schools.