Fiscal policy and economic activity: new causal evidence

Abstract Utilizing a quasi‐natural experiment design, we identify an exogenous cut in local taxes accompanied by an equivalent reduction in local government spending, and we estimate the impact of these exogenous changes on income. We exploit a unique regional dataset that combines local income data with local voting outcomes on current expense tax levies. Taxes and the associated spending change abruptly at the 50 percent vote share cutoff below which a tax levy fails to pass. This cutoff determines which observations serve as controls and which receive treatment: a reduction in local taxes and government spending. Voting percentages around the cutoff are a source of exogenous variation, with observations around this quasi‐randomly assigned and very similar across characteristics. We find that balanced budget reductions in taxes and spending cause a large drop in local incomes in the first two years after the vote, suggesting that government expenditure effects on income are larger than fiscal revenue effects. The cumulative government spending multiplier of a balanced‐budget change in spending for our baseline is a sizable 1.5. This effect of local tax‐financed government spending is prominent in low‐income and high‐poverty areas, suggestive of mechanisms related to the share of liquidity‐constrained agents.