Bank runs are a feature of how banks get their money, John Cochrane says, not where they invest their money:
So a better approach, in my view, would be to purge the system of run-prone financial contracts — that is, fixed-value promises that are payable on demand and cause bankruptcy if not honored, like bank deposits and overnight debt. Instead, we subsidize short-term debt via government guarantees, tax deductibility, and favorable regulation, and then we try to regulate financial institutions not to overuse that which we subsidize.
[...]
[I]f we purge the system of run-prone financial contracts, essentially requiring anything risky to be financed by equity, long-term debt, or contracts that allow suspension of payment without forcing the issuer to bankruptcy, then we won’t have runs, which means we won’t have crises. People will still lose money, as they did in the tech stock crash, but they won’t react by running and forcing needless bankruptcies.
This sounds radical to Arnold Kling, but I’ve long held that demand deposits fail spectacularly.
There is no market failure as bad as the coercive intervention intended to fix it.
And then more regulation to block the loopholes in this regulation, and so on ad infinitum until the only way for a bank to function is to know who to bribe and how much.
Just let banks fail and soon enough depositors will learn that demand deposits are untrustworthy.
The problem of maturity transformation is that it causes bank runs, and the problem of bank runs is that they cause a swift fall in the supply of credit, and so a swift increase in interest rates. An economy with maturity transformation will be used to low interest rates, and as they rise, businesses go bankrupt and it takes a while for new ones that can cope with higher rates to start up.
So the problem is: how do we get from here to there?
I don’t think it would be difficult at all to move consumer banking away from fractional reserves and toward a mix of full-reserve checking accounts, supported by fees, and money-market accounts of some kind.