Chapter 1: I develop a theoretical model for investigating the cost of failure to commit in the provision of bailouts to financial institutions. When a financial institution fails, the fiscal authority often deviates from its ex-ante no-bailout commitment: the ex-post best response is to bailout. The fiscal authority’s time inconsistency creates moral hazard. I calculate the welfare loss from the failure to commit. In the model, as long as the fiscal authority is able to commit to a pre-determined bailout policy, the outcome is typically constrained efficient. Furthermore, a higher probability of bank run is not always welfare reducing. Increased run probability can be beneficial by making financial institutions more cautious, thus decreasing the moral hazard loss. Regulations on short-term interest rates offered by financial institutions can deter moral hazard, particularly when the run probability is small. Chapter 2: This chapter analyzes the optimal bailout policy in an interconnected banking system. Banks are allowed to deposit in each other to hedge against idiosyn- cratic liquidity shocks. When a fraction of banks in the economy are hit by a liquidity shock and become insolvent, there are potential spillovers to solvent banks. In this case, the optimal bailout policy is not always either a full bailout or zero bailout. It is sometimes optimal for the fiscal authority to provide partial bailouts that are just sufficient to prevent spillovers. The decision of the fiscal authority depends on how much pressure from taxpayers and banks. If the ur- gency to save the banking sector outweighs the utility from public goods, a full bailout is optimal, and vice versa. When the two effects are comparable, the optimal decision of the fiscal authority is partial bailout.