Aggregation bias in estimates of demand for local services: Empirical evidence

Abstract

Studies of demand for local public services have long been of interest for the value they can have for state and local public finance policy. While both macro and micro studies of these demands are common, most of these evidences are derived from macro studies based on the median-voter model. Specification error arising from a median aggregation can be quite significant. This paper finds a substantial difference in the coefficients on property tax prices and education service resulting from aggregated data compared with those from their micro data. These findings are important for the allocation and design of local budgets and state programs for intergovernmental revenue sharing.