Some readers may have found these results of greater interest were we able to confirm the directional hypotheses of the certainty setter model in addition to finding that spending is “related” to the reversion – the much weaker prediction of the uncertainty model. Yet the failure of the certainty
When base elections were optional, their infrequent occurrence might be expected under either a median voter or a setter model. For those districts that did hold base elections, the observations appear consistent with a setter model, but are either inexplicable by or inconsistent with a median voter model. Finally, both (a) the relationship between the number of budget elections and the presence of base elections, and (b) the outcome of mandatory base elections in 1978 provide a set of observations that are challenging to either model.
A similar challenge was posed by our investigation of budget elections when we called attention to the inadequacy of static, full information models. Rather than resolving the question of the simple setter model against the simple median voter model, our results indicate that both may be inappropriate and that endeavors dealing with complexities omitted here are warranted.
Romer, T., &Rosenthal, H. (1982). Median voters or budget maximizers: Evidence from school expenditure referenda. Economic Inquiry, 20(4), 556-578.