We explore whether voters' willingness to approve government spending in bond elections is affected by how costs are presented. Using an original survey experiment, we examine willingness to approve bonds, randomizing both the total cost of the bond and the framing of the cost as either a personal cost or an aggregate amount. We find that respondents are less supportive of bonds when the bond is framed as a personal expense and that respondents are more cost-responsive when they see personal costs. There is also substantial heterogeneity based on the respondent's partisanship and the policy domain of the bond.