What banks do -- sell short-term debt (like deposits) to fund long-term loans -- is inherently risky. In theory, shareholders bear this risk. In practice, much of it is dumped on citizens, even though they've no claim on the returns associated with the risk-taking.

The problem is that banks have too little capital to absorb losses without going bust. Even the new, stronger Basel III rules mandate a bare minimum ratio of bank equity to bank assets of just 3 percent. This means that a bank that loses more than 3 percent of its portfolio becomes insolvent.

Aside from this sliver of equity, banks fund themselves by issuing what some scholars call "money-like" debts. In addition to government-guaranteed deposits, these include commercial paper sold to money-market mutual funds, cash-management accounts offered to pension funds and corporations, and "repo" (the sale of a security with the promise to repurchase it later).

Rightly or wrongly, owners of these bank obligations generally treat them as if they were riskless. Deposits are insured up to a limit but the other liabilities aren’t. Savers who come to doubt that they’ll get back 100 cents on the dollar generally try to convert their money-like debts into something that has absolutely no risk of default, like a T-bill. (Inflation is a risk for T-bills, but the bank obligations are no better protected in that respect.)

Runs, in other words, aren’t confined to (uninsured) deposits. They’re so destructive that governments, though tolerating individual failures by small banks, never willingly let the system as a whole collapse. One way or another, the citizenry ends up covering the losses with a combination of direct fiscal transfers, monetary accommodation and regulatory forbearance. In effect, the government gives banks an implicit subsidy to take excessive risks.

Bloomberg View's editors have been writing a lot about this lately. (See Why Should Taxpayers Give Big Banks $83 Billion a Year? and follow-up articles on the subsequent debate and the banks' response.) Those who want to dig even further into the academic research on the subject may also want to read this, this and this.

Recently, reformers have focused their attention on resolving this equity shortage. "The Banker's New Clothes," by Anat Admati and Martin Hellwig, argues that the implicit subsidies would be removed if banks funded about 20 percent to 30 percent of their assets with equity. (We posted excerpts here, here and here.)

Equity-capital ratios in the range of 20 percent to 30 percent would make banks safer, so you'd expect the return on bank equity to fall. That's a feature not a bug. Bankers who have been feasting on profits from excessive risk-taking will see their pay fall too. Count that as a further benefit. It might do a little something to slow the 30-year trend toward greater income inequality.

Bankers might call these proposals radical, but in fact they're moderate. The structure of the banking system wouldn't change. Banks would still operate two essentially different businesses: selling short-term debt and making loans. The potential for a mismatch would remain. More capital would certainly help, and taxpayers would be less on the hook, but the risk of bank failure wouldn't disappear. The same goes for making banks smaller, so that more of them could be left to fail on their own. It would help, but it doesn't address the underlying problem.

There's a way to do that. Divide the banking business in two. Deposit-takers don't have to be credit-creators -- they can be told to hold entirely safe assets. Credit-creators don't need to take deposits -- they can fund their operations by borrowing in financial markets. As renowned Yale economist James Tobin once said, "The linking of deposit money and commercial banking is an accident of history." He and other thoughtful scholars have been discussing how to correct this "accident" for many years.

I'll discuss some of their ideas next time.

(Matthew C. Klein is a contributor to the Ticker. Follow him on Twitter.)