02/13/2009 05:12 am ET Updated May 25, 2011

SEC Rules Don't Go Far Enough

The Securities and Exchange Commission is about to adopt rules for ratings firms that were previously published for comment, but they will not address the agencies' internal compensation policies, which rewarded revenue production over rating accuracy. Nor will they go far enough to eliminate rating shopping, where the issuer hires the agencies that give it the highest ratings.

These were main drivers of the inaccurate ratings of securities backed by subprime mortgages.

Credit ratings have an extraordinarily broad effect on our financial system -- witness their role in the subprime crisis. Though the agencies did not single-handedly cause the crisis, they could have singlehandedly prevented or at least minimized it.

I was involved with two of the three major rating agencies from 1968 into 1998. During that time agency culture emphasized the importance of ratings and the attendant responsibility to be accurate. But it appears from recent SEC findings that, somewhere during the last five years, the emphasis shifted from being accurate toward generating revenue.

It needs to shift back. One way to accomplish that is to change the internal compensation structures of the rating agencies.

When an agency registers with the SEC as a nationally recognized statistical rating organization, its ratings for securities carry many regulatory benefits, including lower capital requirements for banks, insurance companies, and brokers, along with investment eligibility for mutual funds, public retirement plans, and money market funds. These benefits create investor demand for ratings and are important reasons issuers pay for them.

The SEC's rules should not allow agencies to register as NRSROs and confer these benefits unless analysts' compensation is reasonably aligned with the performance of their ratings. Bonuses could be paid over several years, and if the rating proved wrong as a result of poor analysis, the unpaid balance could be forfeited. That bonus period could be three years, since most mortgage defaults occur in that time.

Analysts responsible for challenging rating methodologies should be paid as much as those who assign the ratings, even if a challenge threatens major revenue streams, such as the robust fees from rating mortgagebacked securities. The analysts who assigned an initial triple-A rating to a security that was subsequently downgraded to junk status were probably paid much more than the analysts who challenged those ratings.

The SEC rules should eliminate this gap, signaling that accuracy is valued at least as much as revenue. It should require agencies to disclose these compensation policies in their registration statements,so they become a basis of liability if the agencies do not comply with them.

If these rules had been in place, analysts might have questioned the underwriting of mortgage originators that securitized -- sold off -- substantially all their risk. Agencies might not have relied as much on FICO scores, given their poor performance as default predictors.

Rapidly rising home values might have signaled inaccurate appraisals. Increased homeownership, loans that started with low monthly payments but spiked later, and the failure to verify borrowers' income might have raised affordability issues. Maybe the agencies would have changed their models to better reflect the absence of any performance history for the new types of loans.

"Rating shopping" also contributed to inaccurate securities ratings. The Credit Rating Agency Reform Act of 2006 opened up the market to more NRSROs, on the flawed theory that more competition would yield more accurate ratings. But the NRSROs are more likely to compete by giving high ratings that misrepresent the issuer's creditworthiness and abuse the regulatory benefits that come with them. For example, an insurance company can "shop" for the highest rating for a bond it purchased, so it can put up less capital against that investment -- a recipe for insufficient capital to pay claims.

The proposed SEC rules address rating shopping by requiring that information given to the NRSROs hired to rate structured financings also be given to those that were not hired, and that the latter commit to rating "a minimum number" of those deals. This would help, but it is not enough.

A better solution is to recognize only a few NRSROs and require issuers to hire all of them to rate all structured financings. Then the regulation of rating fees -- unthinkable before the crisis -- should be considered.

Mr. Baron, a former counsel to Standard & Poor's Corp. and a former vice chairman and general counsel of Fitch Inc., is a member of the board at Assured Guaranty Ltd.

This piece was originally published in The American Banker on January 9, 2009.