How long will the US economy live with a banking system in which some institutions are too big to fail ? Not long, we should all hope, because large banks today, under federal protection, can raise short‐​term funds more cheaply than their smaller competitors, which are allowed to fail.

“Too big to fail” is an unstable system. Politically inspired constraints on large banks leave them not knowing what will come next out of Washington, while there is no way of knowing whether any given bank is just small enough to be let go or will be bailed out if it gets into trouble.

But here we are, more than a year after the rescue of Bear Stearns, without a plan for the future except for vague — and, as far as I can tell, totally empty — statements coming out of Washington about tighter regulation. What exactly does Washington have in mind?

Here is a proposal, not at all original but deserving of serious public discussion. As a condition of enjoying the benefits of a bank charter, every bank must issue 10‐​year subordinated notes equal to 10 per cent of its total liabilities. The specification can be adjusted, but this one serves to illustrate the proposal. The subordinated debt would be unsecured; holders would stand last in line among all creditors in the event that a bank had to be shut down. The sub debt requirement would be in addition to existing requirements for equity capital.

Genuine reform requires that four minimal requirements be met, and the sub debt proposal qualifies. First, banks need more capital to protect the federal deposit insurance fund. Second, there must be more market discipline: each bank would be forced to roll over maturing sub debt equal to 1 per cent of its liabilities each year. Third, financial stability requires that a bank not be subject to runs. Sub debt cannot run, because of the 10‐​year maturity.

Fourth, and critically important, some creditors and not just equity owners must be at risk, which is clearly the case with sub debt. Sub debt provides much more market discipline than equity, because a bank in trouble with a weak share price is not forced to do anything. Maturing sub debt, however, does discipline the bank and if the bank cannot roll over the debt, it must shrink by 10 per cent to live within its remaining outstanding sub debt. This system is stable because any bank can contract by 10 per cent within a year by letting loans run off and/​or by selling other assets. It is highly desirable that contraction be managed by the bank itself and not by regulators.

We cannot depend on regulatory agencies to prevent a recurrence of financial crisis. Ahead of the crisis, regulators, myself included, did not understand the risk of subprime mortgage paper in bank portfolios. Nor did the senior management and directors of the most sophisticated financial firms in the world. This is not the first time regulators and firms have failed to assess risk adequately. Large international banks accumulated Latin American loans in the 1970s; when Mexico defaulted in 1982 and other defaults followed, a financial crisis was narrowly averted.

Long‐​maturity bonds create much more market discipline than do short‐​term obligations. Holders of three‐​month certificates of deposit, for example, assume that they can always exit quickly by letting the CDs mature. It is for this reason that bond spreads over Treasuries of comparable maturity are systematically higher for longer maturities than for shorter maturities.

Banks hate the idea of a substantial sub debt requirement, because sub debt will be expensive. But bankers should think carefully about their opposition. Would they rather face market discipline from sub debt or much heavier Washington regulation, including opaque and changing rules? Given the scale of our financial crisis and taxpayer losses, intrusive regulation will be the norm for years to come. Do bankers really want to face unpredictable constraints such as they have seen on executive compensation?

I challenge leaders of our large banks to support a market‐​based reform such as the one I have outlined. A return to the status quo ante, with banks enjoying the benefits of “too big to fail”, does not seem likely. Regulators will not dare risk a repeat performance. Bankers who think that their political influence will control the regulatory process are in for a rude surprise.