This paper studies how governments manage public employee pensions and how this affects insolvency risk. I propose a quantitative model of governments that choose their savings and risk exposure by borrowing/saving in defaultable bonds, borrowing in non-defaultable pension benefits, and saving in a pension fund that earns a risk premium. In insolvency, the government can receive transfers from households who may differ from the government in their preferences for public services and private consumption. I match the model to a panel of CA cities and a hand-collected record of fiscal emergencies. The model predicts that governments are highly vulnerable to another stock market bust. A hypothetical shock to pension funds in 2015 produces twice as many fiscal emergencies as the original 2008-10 shock. In the quantified model, the government undersaves and take excess risk relative to what households would choose. Savings requirements that limit spending to essential services plus 0.3% of cash-on-hand produce large welfare gains for households. Requiring the pension fund to invest more in safe assets decreases household welfare because the lower average return discourages the government from saving.