The SEC couldn't catch Bernie Madoff, and the combined resources of BP and the U.S. government needed three months to plug the Gulf oil leak. So it should come as no surprise that financial regulators frequently get it wrong when they try to figure out how much capital that complex, modern banks require. As "Risk Lies in the Risk-Based Capital Approach of Basel III" [American Banker, July 15] observes, several banks that appeared to be best capitalized under the Basel rules were the first to be bailed out.

The article is surely correct to question whether refinements to Basel II can be relied on to prevent another crisis. Any set of fixed rules for measuring capital adequacy will fail to capture a great deal of risk, and rules will never match bankers' ingenuity for engaging in regulatory arbitrage. Basel II, which has been adopted in most developed countries other than the U.S., improves on Basel I by plugging loopholes, but its very complexity makes the whole enterprise harder for managers, shareholders, creditors, rating agencies, regulators and politicians to follow.

Considering the limitations of mechanical rules, the article's advice that "effective capital rules would have to [make sure] … risk isn't pawned off on entities outside the system" seems backward. Transferring credit risk to insurance companies, hedge funds and anyone else who wants it will free the banking system as a whole from dependence on shaky formulas that no one fully trusts.

The article's main example considers Bank A that buys a corporate bond, then buys credit protection from Bank B, which in turn buys credit protection from an insurance company. Bank A's capital charge is reduced by 80% under Basel I while Bank B's capital charge is minuscule. The total capital the two banks together have to hold is a fraction of the amount required if Bank A simply held the bond. (The calculation for Bank B in the article mixes the Basel I and Basel II calculations and would not apply to any particular bank. However, in either regime, it is true that Bank B would not have to hold much capital against a matched pair of credit default swaps assuming it buys the protection from a highly rated counterparty.)

Should we be troubled by this outcome? The insurance company at the end of the chain bears the ultimate credit risk. Assuming Bank B prudently manages its counterparty relationship with the insurance company, the banking system no longer needs to worry about the bond issuer's credit. As a result, the two banks should have to hold less capital to support their exposure to the bond.

Let's make this example more realistic to show how "pawning off" can create value for all the participants in the chain. Bank B doesn't add anything to the analysis, so assume Bank A directly faces the nonbank credit protection seller. Make the credit protection seller a hedge fund. Since the bank would have no particular reason to buy a bond as in the original example, assume instead that the borrower is a long-term client of Bank A seeking a term loan in excess of its existing credit lines.

Then the bank can buy credit protection from the hedge fund on the amount of the loan above the existing credit line. The client gets the loan it wants, the bank makes a profit from intermediating and keeps its customer satisfied, and the hedge fund earns a credit spread if it bets correctly and loses otherwise.

Admittedly, the hedge fund is unregulated, but so what? What better place to put credit risk than in a large number of independent funds, backed by wealthy individuals and institutional investors whose profits and losses are of concern only to themselves and not to the public? The hedge fund would usually post cash collateral to the bank to guarantee its own performance. The amount of capital the bank has to hold against the loan is proportionately reduced — this is fair, as the bank has reduced credit exposure to the counterparty by purchasing credit protection.

Most important, this approach works whether or not regulators know the right model for estimating the loan's probability of default, loss given default or correlation of default to the business cycle. If the borrower defaults, a sovereign wealth fund that backs the hedge fund might grow slightly less wealthy. The bank soldiers on unharmed. Using nonbank investors to absorb credit exposure can start to pose systemic risk once the hedge funds grow "too big to fail." That is far from a concern now and one the SEC can easily control.

Finally, it is worth pointing out that the Basel II rules are not quite as foolish as the article makes them out to be. It is simply not true that, "As is currently standard, government debt would be considered riskless." Under Basel II's Standardized Approach, sovereign risk weights depend on the issuer's credit rating. Triple-B-rated sovereign debt, for example, attracts a 50% risk weight and double-B-plus is weighted 100%. Under Basel II's Internal Ratings Based Approach, banks must compute the sovereign default probability and hold appropriate capital as with any other asset.

As the article points out, Basel residential mortgages have a risk weight of 35% under Basel II (at least under the Standardized Approach); on the surface, mortgage lending may appear unworthy of favorable treatment given its role in the crisis. However, this weight applies only to mortgages originated in "accordance with strict prudential criteria," and nothing in recent experience outside the U.S. suggests that 35% is too low. A risk weight of 35% and capital ratio of 10% implies that 3.5% of capital is held to back a home mortgage loan. Assuming 100 basis points of spread over the cost of funding and a five-year average life, this amount of capital will cushion a 1.7% annual loss rate. Even in the U.K. in 2009, the increase in cumulative losses in Moody's U.K. Prime Index was only 6 basis points.

Richard Robb is the chief executive of Christofferson, Robb & Co., an investment management firm, and professor of professional practice at Columbia University's School of International and Public Affairs.