Spillovers, migration and public school expenditures: The repetition of an experiment