Historically, large changes in U.S. government spending induced fiscal efforts that were not all alike, with some using more progressive taxes than others. We develop a heterogeneous-agent New Keynesian model to analyze how the distribution of taxes across households shapes spending multipliers. The model yields empirically realistic distributions in marginal propensities to consume and labor elasticities, which result in lower responsiveness to tax changes for higher-income earners. In turn, multipliers are larger when spending is financed with higher tax progressivity—that is, when the tax burden falls more heavily on higher-income earners. This result is historically material. We estimate that, on average, tax rates increased more for top-income than for bottom-income earners after a spending shock. Thus, the typical U.S. spending shock was financed with higher tax progressivity. We further exploit the historical variation in the financing of spending to estimate progressivity-dependent multipliers, which we find consistent with the model.