Does the U.S. have too much public debt? Is the U.S. overly reliant on debt financing crises?  Many commentators have raised debt sustainability concerns yet strong investor demand for safe assets has kept interest rates on Treasury bonds low. Policymakers face a trade-off between reducing the public debt and providing markets with safe assets.  To quantify the importance of this trade-off, I solve for the optimal Ramsey policy of a government that issues safe assets (i.e. risk-free and liquid), taxes labor, and insures against aggregate risk.  Introducing quantitatively realistic liquidity demand into an otherwise standard optimal debt model dampens policymakers' tax-smoothing and precautionary savings motives. Calibrating to standard parameters, the model closely matches historical U.S. debt policy.  Although debt expansion was the optimal response to recent crises, the fiscal responses to the Great Recession and Pandemic were overly debt-financed.  This over-reliance on debt financing pushed the debt-to-GDP ratio 10 p.p. above the model-implied optimum. Given the projected path of government expenditure policymakers should reduce the debt to 75\% of GDP over the next 10 years. In extensions of the baseline model, I consider the optimal debt policy during liquidity and default crises.

How does government spending affect the macroeconomy?    We provide a new monthly series of government spending shocks, spanning 1960-2023.  Using high-frequency data on U.S. defense contractors' stock returns and exploiting institutional features of the federal budget process, we identify important fiscal policy events. By assuming defense spending news is revealed in daily windows around these events, we identify defense news shocks.  Our shock series matches historical fluctuations in defense spending.  The shocks have statistically and economically significant effects on aggregate activity.  Positive defense news shocks lead to large increases in defense spending, consumption, income, and output.  We estimate a fiscal multiplier of 2.0.   We test for asymmetric fiscal multipliers and find that multipliers are largest when (i) there is slack in the economy (ii) monetary policy is accommodative (iii) recessions are demand-driven (iv) government spending is declining, and, (v) spending growth is small.