Banks finance illiquid assets with demandable deposits, which discipline bankers but expose them to damaging runs. Authorities may not want to stand by and watch banks collapse. However, unconstrained direct bailouts undermine the disciplinary role of deposits. Moreover, competition forces banks to promise depositors more, increasing intervention and making the system worse off. By contrast, constrained central bank intervention to lower rates maintains private discipline, while offsetting contractual rigidity. It may still lead banks to make excessive liquidity promises. Anticipating this, central banks should raise rates in normal times to offset distortions from reducing rates in adverse times.

Loans are illiquid when a lender needs relationship-specific skills to collect them. Consequently, if the relationship lender needs funds before the loan matures, she may demand to liquidate early, or require a return premium, when she lends directly. Borrowers also risk losing funding. The costs of illiquidity are avoided if the relationship lender is a bank with a fragile capital structure, subject to runs. Fragility commits banks to creating liquidity, enabling depositors to withdraw when needed, while buffering borrowers from depositors' liquidity needs. Stabilization policies, such as capital requirements, narrow banking, and suspension of convertibility, may reduce liquidity creation.

Views on the role played by banks in the economy have evolved greatly over the last 125 years, as have arguments on the need, as well as the best way, to regulate them. Some of the key insights in the debate have been published in the Journal of Political Economy. In what follows, we will outline the main contributions to the debate in recent years, with an emphasis on work done at the University of Chicago or published in the JPE. We want to emphasize work that has relevance today, but despite this caveat, we will probably end up doing injustice to work published long ago. We begin with a framework for organizing the theories of intermediation. We then draw out the implications for what the theories say about regulation and note that in many respects the motivation for regulation has been only loosely tied to the theory of intermediation. We close with some open questions for regulators and economists interested in banking. We do not survey the research that has followed up on work published in the JPE, nor will we attempt to provide a detailed overview of the entire academic literature on banking. For that, we refer the reader to the excellent work by Gorton and Winton (2003) and Freixas and Rochet (2008).