Abstract
Although the median-voter hypothesis has been around for nearly fifty years, direct tests of the hypothesis have not been forthcoming. This study first develops a model to consider the impact of alternative voting rules on the desired level of debt in a community. If the median-voter hypothesis holds, then voter preferences should dominate agent preferences and debt levels should be the same whether a simple-majority referendum is employed or whether the debt decision is left to local public officials. The model also predicts that when some fraction greater than a majority is required for referendum approval then debt levels in the community will be lower. Cross-section empirical results for the fifty states supports both sets of hypotheses.