Abstract This study reexamines the US federal revenue-spending nexus by applying continuous wavelet analysis to observations of 1792–2020. Specifically, we use the cross-wavelet phase difference and wavelet Granger causality test to make inferences concerning the lead-lag relationship between federal income and expenditures. The two methods agree on three empirical results. First, the spend-and-tax hypothesis is confirmed as the dominant pattern. Second, the negative tax-and-spend hypothesis also received moderate support. Third, an adverse effect of spending on revenue, which has not been suggested by any theory before, recently appeared in cycles shorter than four years. These findings’ theoretical, methodological, and policy implications are discussed at the end.