Before 1914 the U.S. economy experienced frequent bank runs and financial panics. Runs occurred when a shock to one or a few banks—such as crop failure, a major loan default, or a corruption scandal—caused many depositors to attempt to withdraw their money simultaneously. Banks lend out most deposits and hold only a fraction as cash, so widespread demand for withdrawals makes it likely the affected banks will fail. Worse, a run on one bank can increase depositor concerns at nearby banks, leading to contagion and financial panic.
The frequency of bank runs and panics was a crucial reason for creation of the Federal Reserve System in 1914. The idea was that the Fed would provide an “elastic” currency, injecting cash into the economy during periods of high credit or currency demand and withdrawing it in periods of low demand. This policy appeared to work initially, since bank runs disappeared from 1914 through 1928. Runs and panics returned with a vengeance in 1929–1933, however, and the huge number of bank failures contributed significantly to the Great Depression.
In response to the runs and panics during the Depression, the U.S. Congress created federal deposit insurance in 1934. Under this policy the government reimburses the depositors of failed banks. Funding comes from insurance premiums paid by the banks, and from general government funds if necessary. In principle the insurance applies only to deposits below a specified ceiling, but in practice coverage is essentially complete. Depositors can split large accounts across banks, and the FDIC typically covers all deposits.
The public and many economists now take as given that government deposit insurance is good policy.
At first glance this conclusion seems warranted. Bank runs have not occurred since 1934, and insured depositors have not lost a dollar in that time. Further inspection, however, suggests that an alternative approach to limiting bank runs might be superior to deposit insurance.
The problem with deposit insurance is that it generates a moral hazard: When banks know their deposits are insured, they have an incentive to purchase riskier assets. If these assets generate high returns, banks make good profits, while if they fail, deposit insurance cushions the blow. Thus banks assume more risk than warranted by market fundamentals. That is why current regulation tries to limit bank holdings of risky assets while also requiring a minimum degree of capitalization.
In principle the combination of balance-sheet regulation and deposit insurance can both limit runs and prevent excessive risk taking. In practice banks can innovate around much regulation, so they still end up taking more risk than is appropriate. This is precisely what occurred in the run-up to the 2007–2008 financial crisis. By using derivatives, off-balance-sheet vehicles, and “structured finance,” banks were able to assume huge risks within the confines of existing regulation. For several years the excessive risk taking generated large profits, but eventually the underlying fundamentals crashed, pushing several large banks to the brink of failure. Widespread failure did not occur, thanks to the Treasury bailout, but the adverse implications for taxpayers were at least as bad as those that failure would have imposed.
In response to these events, many observers have argued that the United States needs more regulation of banks. It is not obvious, however, why additional regulation would be any less subject to manipulation than past regulation. Thus approaches that involve less regulation, not more, are worth considering.
The crucial regulation in this context is the longstanding regulatory ban on bank suspension of convertibility. This regulation means that when depositors request cash withdrawals, banks are legally obligated to comply. In the absence of legal constraints, banks might offer deposit contracts that allowed them to suspend partially or fully. These contracts would presumably offer different terms from a standard demand deposit. For example, they might require that interest be paid on any suspended balances, or they might specify the length of time a bank could suspend without incurring penalties. They would not, however, commit the bank to always meeting demand withdrawals both immediately and in full.
If banks can suspend convertibility, depositors know that runs merely precipitate suspension. This greatly reduces depositor incentive to panic and run. Allowing banks the right to suspend would probably not eliminate all runs, but it would plausibly limit them to banks that are insolvent rather than merely illiquid.
The question, then, is whether a banking system with less regulation—no prohibition on suspension and no deposit insurance—might work better than current regulation—prohibitions on suspension, combined with deposit insurance and balance-sheet regulation.
The evidence from the pre-1914 era suggests that the regime with less regulation has promise. Banks were not legally allowed to suspend convertibility during this era, but many did so anyway, sometimes with explicit approval of, or even encouragement from, regulators. This did not eliminate runs and panics, but the record suggests that suspension reduced contagion and failure in these episodes. A few panics were associated with substantial declines in output, but many others were short-term and confined to a few cities or parts of the country. Even in cases where recession and panic coincided, some of this correlation no doubt reflects the effect of recession on bank solvency, rather than panics causing recessions. It is plausible, moreover, that suspension would be even more effective in limiting runs and panics if banks were able to experiment with different types of contracts and use suspension without fear of legal jeopardy.
It is also plausible that the socially desirable number of runs is not zero; after all, runs discipline banks that take excessive risks. In an idealized setting the mere threat of runs might be sufficient to prevent them; actual runs need never occur. In the real world, however, the occasional run is most likely necessary to close down irresponsible or incompetent banks and to remind others to behave.
Both theory and evidence, therefore, suggest that regimes with less regulation deserve as much consideration as those with more regulation. Designing and enforcing regulation is difficult because it complicates incentives and generates unintended consequences, as recent events have shown. Perhaps, therefore, markets are better than regulation at disciplining the banking system.