If at First You Don't Succeed: Budgeting by a Sequence of Referenda

Abstract

An econometric model, based on the Romer-Rosenthal model of agenda control by budget-maximizing bureaucracies is used to analyze the budgetary and voting outcomes of referenda in a cross-section of Oregon school districts. In addition to estimates of the effects of agenda control, the model permits estimation of the spending effects of voter failure to perceive the .availability of lump-sum intergovernmental grants. Budgets are set via referenda. In the event of a failed referendum, a limited number of additional votes may be taken. The model permits estimation of the degree to which the agenda setter (e.g., the school superintendent) learns about voter preferences from the outcomes of failed referenda. The endogenous variables in the model are the budget proposals and voting outcomes of each referendum in the sequence of referenda held in each school district. The effects of proposals on voting behavior and the effects of learning appear via structural parameters in the error structure. The model is estimated by non-linear maximum likelihood. The results (1) support the theoretical model of agenda control and the effect of the setter's proposals on voting behavior; (2) indicate that voter failure to perceive state grants leads to important increases in spending; (3) fail to indicate any learning by the setter.