The US tax system allows taxpayers to deduct local taxes from their taxable incomes. Using school district referendum results, we employ a continuous treatment two-way fixed-effects framework to provide causal evidence of a positive relation between the demand for local public goods and the share of residents deducting local taxes. We find that a 1 percentage point decrease in the share of residents deducting property taxes reduces tax and bond referendum approval rates by approximately 0.97 percentage points. Because these federal tax deductions disproportionately benefit higher-income individuals, they potentially widen disparities in public service provision across jurisdictions.