Bank Runs, Deposit Insurance and Liquidity (Journal of Political Economy, Vol. 91, No. 3, June 1983)

"And providing liquidity is where banks come in. Banks, according to Diamond and Dybvig, are in the business of transforming illiquid assets – assets that pay off in the future – into liquid assets, so that people can undertake the spending and productive projects they want now. Banks accomplish this by pooling the risks of many individual people. It may be difficult to predict whether an individual person will choose action A rather than B, but it is relatively easy to predict what proportion of a large group of people will choose action A. This is the same principle that insurance companies rely on. So, let’s say it’s Thursday and there’s a large population of people. Each person has some wealth but doesn’t know when she will need to spend it – that is, whether her spending needs will occur on Friday or on Saturday. There is also a productive project. Each dollar invested in this project on Thursday yields $1.50 of output on Saturday. If a person doesn’t have to spend until Saturday, then she can benefit handsomely from the project. On Thursday she can invest her wealth in it, and on Saturday end up with 50 per cent more than she started with. But what if, instead, her spending needs turn out to occur on Friday? In that case, she has to take her money out of the project then, losing the 50 per cent return she would have received had she been able to wait until Saturday. So, here’s what a bank does. Lots of people deposit their wealth in the bank on Thursday and it then invests these deposits in the productive project. Even though the bank doesn’t know for sure whether any given person will want to withdraw her money on Friday, it knows accurately what proportion of people will do that. So it can, in effect, give those people insurance. Rather than just giving them their deposits back – which is what these people would get if they were on their own without a bank – it can pay them interest, in exchange for reducing the interest rate for the Saturday withdrawers below 50 per cent. In other words, the Saturday people subsidise the Friday people (in the same way that, with health insurance, people who remain healthy are subsidising those who get sick). And the bank makes this possible. But there’s a problem with this arrangement. Suppose that on Friday people get worried that the bank may not have enough money to pay off all the depositors. Then everybody – not just the people who need to spend on Friday, but also those who would otherwise have waited until Saturday – will try to withdraw their deposits on Friday. That’s a bank run. And if it occurs, the bank actually won’t have enough money to pay everyone off. Yes. This is an important feature of banking: at any given time, a bank doesn’t have enough money – shouldn’t have enough money, if it’s doing its job well – to repay all its depositors then and there. The money has been invested in some productive project; it’s illiquid. Having all its assets in liquid form would be unproductive, inefficient for the bank. But, nevertheless, if everyone decides to withdraw at the same time, the bank is stuck. And furthermore, if a bank run takes place at one bank, then the same thing may well happen at other banks, because their depositors get worried too – and we soon have a financial crisis. So there’s an important role for government here. Government can promise depositors that it will repay them if the bank cannot – it can offer deposit insurance. The government can make this pledge credibly because it has the power to tax: it can get the money to repay depositors by taxing other people. And once the government has made the pledge, the threat of bank runs vanishes: depositors will no longer all try to withdraw their money on Friday. Now what is the relevance of all this for the current financial crisis? Well, deposit insurance has pretty much eliminated traditional bank runs. But many financial institutions have experienced the equivalent of a bank run. Bear Stearns was brought to the brink of collapse because investors had lost confidence in it and had therefore withdrawn their financial support. The government bail-out of Bear Stearns was something akin to paying off on deposit insurance. That’s one way in which the Diamond and Dybvig framework is incomplete: it ignores the fact that if banks know their deposits are insured – or that they are going to be bailed out – they may change their behaviour. Diamond and Dybvig is certainly a good starting point, and a great deal of the subsequent work on banks and on liquidity follows in their footsteps. But they don’t take account of moral hazard. There had been previous models of bank runs – after all, bank runs are a very old phenomenon – but this particular explanation is new to Diamond and Dybvig. As Diamond-Dybvig show, it would be very damaging for the government to refrain from insuring deposits or bailing out banks. Bank runs are not only scary for depositors, but interfere with production. If all depositors try to withdraw on Friday, the bank has to take its entire investment out of the productive project. So there’s no output on Saturday. Deposit insurance and bail-outs are essential to prevent an economy’s production from grinding to a halt, but they create a moral hazard problem. And moral hazard is something that Holmstrom and Tirole explicitly take into account in their paper, which can be thought of as an extension of Diamond and Dybvig."