Should banking regulations ever have forced the use of bond credit ratings from certain ratings agencies for investment decisions and risk-based capital requirements? No. Mandatory use of credit ratings was a big regulatory mistake. It fostered a government-sponsored credit ratings cartel. It created a concentrated point of possible failure, generating systemic risk, which did indeed fail in the bubble (along with other things).

But should banks be able to use credit ratings in the normal course of considering investments as a logical external source of analysis, which is, especially for smaller firms, very difficult and expensive (if even possible) to replicate? Of course they should. This makes perfect sense, the problems of the bubble with mortgage-backed securities notwithstanding. Should banking regulations refer to and discuss how credit ratings can sensibly be used by banks in credit risk management? Of course they should. Whoops—Dodd-Frank prohibits that, in a truly silly provision.

When you are doing 2,000-page bills, you are bound to make a lot of mistakes, as the creators of Dodd-Frank did. In this case, they took an excellent idea — that the use of credit ratings should not be made mandatory by regulation — and confused everyone by adding the provision that regulations must not even mention credit ratings.

Is it too late to fix now? No. It is never too late to fix a serious mistake, especially when the fix is extraordinarily easy.  Is there already too much post-Dodd-Frank regulatory water under the bridge? I don't think so. There are always costly future regulatory waves coming down the banking river.

Dodd-Frank's Section 939A provides that all federal regulatory agencies must "remove any reference to or requirement of reliance on credit ratings." Here in the heat of the battle, they overachieved. The obvious fix is simply to delete the three words, "reference to or." The provision would then read that all agencies must remove "any requirement of reliance on credit ratings."

Yes, that's the real goal: no mandatory use. With this simple fix, regulatory agencies could "refer to" credit ratings, which certainly makes sense, while never "requiring reliance" on them, which doesn't.

Section 939A goes on to say the agencies must "substitute in such regulations such standard of credit worthiness as each respective agency shall determine as appropriate." This has already been done by regulators, using hardly intellectually earthshaking, but commonsensical definitions. For instance, "investment grade" means securities "where the issuer has an adequate capacity to meet the financial commitments" and "the risk of default by the obligor is low," according to the Office of the Comptroller of the Currency.

It is highly interesting that the OCC "does not intend for the elimination of references to credit ratings, in accordance with the Dodd-Frank Act, to change substantively the standards national banks must follow." The Federal Deposit Insurance Corp.'s Supervisory Insights Journal likewise opines that "removal of references to credit ratings from regulations has not substantively changed the standards institutions should consider." The FDIC also emphasizes that "the OCC and FDIC regulations are not envisioned to significantly change the scope of permissible investments."

If the standards do not change and the investments do not change, one might ask: what is the point? Is the point that they do not change in substance, only in form and burden—a regulatory attempt to fix the mistake by Congress? Or do regulators believe that smaller financial institutions can do a better job of analyzing nationally traded securities than rating agencies? If so, that is a very dubious proposition.

The FDIC and the OCC advance an even more dubious proposition. Some bonds can be purchased "without having to make an investment-grade determination," the FDIC explains. Now which bonds do you suppose those would be? How about some politically favored ones? Yes, "federal agency bonds will not require credit analysis." Banks can thus, for example, buy Fannie Mae, Freddie Mac and Farm Credit bonds without any analysis at all and, of course, without loan-to-one-borrower limits.

That would be the Fannie Mae which collapsed in the most recent crisis, after having greatly helped inflate the housing bubble. That would be the Freddie Mac which likewise collapsed, after likewise inflating the housing bubble. That would be the Farm Credit Banks, which collapsed in the previous financial crisis of the 1980s, after having greatly contributed to inflating the farmland bubble. One can hardly view this aspect of the rule as serious systemic risk management.

"The transition away from reliance on credit ratings," the FDIC concludes, will "entail a learning curve for both bankers and examiners." No doubt, and no doubt, it's an expensive one. But this learning curve could be much improved, with a better cost-benefit ratio for smaller financial institutions, by the deletion of three words from the Dodd-Frank Act.

Alex J. Pollock is a resident fellow at the American Enterprise Institute in Washington. He was president and CEO of the Federal Home Loan Bank of Chicago 1991-2004.