Comment on Interbank Lending and System Risk

In much of the recent debate on payment system risk, it has been assumed that the ultimate goal is a system with zero risk. Hence the debate has largely focussed on how to achieve this goal at the lowest transaction cost; for instance, should the Fed step in as a counterparty to all interbank transactions, much in the nature of a futures exchange? Should interbank claims be senior to all other claims so as to make the payment system secure? The paper by Rochet and Tirole seeks to shift the terms of the debate. Rather than take as given the goal of immunizing the payment system from risk, it suggests a trade-off. Minimizing ex post risk to banks has the collateral effect of taking the most effective monitors of a bank other banks out of the monitoring function. As the paper argues, the banking system may be riskier as a result. Furthermore, there are other channels, which the paper does not explore, through which banking system risk could increase when payment system risk is reduced. For instance, if interbank risk exposure is eliminated by making claims of banks on a distressed bank senior, it increases the exposure of less informed lenders to the distressed bank, which could increase the possibility of panics. Having refocussed the question, the authors ask how an optimal system would minimize risk and disruption ex post while preserving the ex ante incentives to gather information and exercise judgment. The paper first shows why banks may want to insure against situations of illiquidity by holding liquid assets or contracting for central bank assistance. As in Diamond (1991) or Holmstrom and Tirole (1995), the necessity of providing bank management the incentive to run the bank well leads to a situation where a bank suiTering an adverse "liquidity shock" is solvent (there is a future stream of positive rents the bank generates) but illiquid (the bank cannot raise funds from outsiders against these rents because if the rents are promised away, the bank will be run into the ground). Rochet and Tirole then argue that when banks can monitor each other, allowing banks to be exposed to each other's liquidity shocks will enhance their incentive to monitor. While the intuition is straightforward, they derive a number of interesting implications. First, if a bank does not monitor other banks, it does not make sense to