Of the Dodd-Frank Act's sundry requirements, few have caused as much consternation among bank regulators as the mandate to evaluate capital without the help of credit ratings.

However, state insurance regulators aren't struggling with the issue like their banking counterparts. The National Association of Insurance Commissioners replaced ratings-based capital determinations for mortgage-backed and other structured securities with a centralized, third-party loss-modeling system — voluntarily and more than a year ago. The association and others say their method could readily translate to banking.

"There was an awful lot of talk that we should reduce reliance on ratings agencies," said Matti Peltonen, the capital markets bureau chief for the New York State Department of Insurance and a participant in the NAIC's planning process. "We got it done. And it's not designed in any way that's specific to insurance companies. Investments are investments, banks or insurance companies. Risk is risk."

It's not that the banking industry is unaware of the NAIC's approach. A notice of proposed rulemaking on ratings reliance issued last August by the federal banking agencies dispatched a centralized model in two sentences, concluding that "one potential drawback of this approach is excessive reliance on a single third-party assessment of risk."

But Gene Phillips, a former ratings agency analyst, argues that additional oversight could be built in. Adopted as is, he said, the NAIC's modeling-based regulatory capital assessments would be an improvement.

"If you model the stuff, you know what the risks are," said Phillips, who co-founded PF2 Securities Evaluations Inc, a boutique firm that now provides modeling services to banks and private investors. "You're also promoting market liquidity, by reducing ratings reliance and eliminating certain informational asymmetries."

Instead of measuring likelihood of default — which doesn't address how much an insurance company is likely to lose — modeling provides a range of expected losses and a reward for taking timely writedowns. Rather than waiting for mass ratings downgrades, regulators would be able to observe and respond to expected losses on securities in near-real time. Banks could share the price of modeling, instead of bearing the cost of commissioning it on their own. And inspectors in the field would not have to evaluate each institution's structured holdings on an ad hoc basis.

Depending on the price of such a system, it could be positive for smaller banks, said Chris Cole, senior regulatory counsel for the Independent Community Bankers of America.

"I think using expected loss as opposed to probability of default would make for a superior system," he said. "We would like a system like [the NAIC model] for our banks, as long as it was readily available, relatively inexpensive and easily understandable. I think this would be something that might work."

That is not to say there aren't serious objections to the NAIC's approach. Though the confidentiality of banks' holdings would be protected — the entity carrying out the modeling would not be told which banks held which securities — the involvement of third parties like PIMCO and BlackRock in the NAIC's system gave some observers pause. Chris Whalen of Institutional Risk Analytics recalls that, during an NAIC hearing more than a year ago, audience members questioned whether regulators should outsource such a crucial task, a concern Whalen found reasonable.

"I think there's a way of fixing this, but unfortunately everybody's so distracted, they don't even consider it a possibility," he said regarding the banking industry's struggle to transition away from ratings. (Whalen would prefer to see a model in which regulators aggregate banks' own modeling data in a central database.)

And even in a simplified form, there is no doubt that the NAIC's approach would cost more and be more complicated than a ratings-based risk-weighting. Cole noted a widespread sense among the banks the ICBA represents that lawmakers overreacted in abandoning ratings. "I think what we were thinking about was keeping the ratings but having some extra confirmation," he said, though he acknowledged that it was hard to see how a system of partial ratings reliance would work.

Insurance industry regulators said that the industry's experience suggested that such concerns were unnecessary. The NAIC's move came about precisely because of the headache ratings were causing for insurance companies: When mortgage-backed securities started to be downgraded en masse in 2008, insurance companies faced abrupt regulatory capital shortfalls even though it was unlikely that they would take significant losses on many positions.

Insurance companies were "asking for an internal [ratings] upgrade, and we said we can't do that," Peltonen said. "We had to have to create a different system. It took a big push in an incredibly short amount of time."

Using an expected loss model avoided penalizing companies for holding formerly AAA-rated bonds while still recognizing the impairments that threatened them, Peltonen said. The NAIC quickly ruled out requiring companies to model their own holdings because of the cost and supervisory work doing so would create.

"We wouldn't have the capacity to evaluate each company's modeling, because these are thousands of insurance companies," Peltonen said.

The NAIC looked at expanding its own modeling operations and just doing the work itself. But a good portion of the industry already worked with entities like PIMCO, reducing concern about any conflicts from centralized modeling. Insurers "were satisfied that the adequate walls were structured around what we were doing," Peltonen?? said.

From there, the NAIC solicited proposals. After choosing PIMCO in November of 2009, regulators designed a model that provided expected loss numbers for a range of scenarios. The possibilities ranged from rosy to "a Depression-like scenario" in which home prices dropped 59% below peak over the course of a decade. The final expected loss number was a blend of those possibilities.

Representatives of the Federal Reserve, the Federal Deposit Insurance Corp. and the Office of the Comptroller of the Currency either did not speak with American Banker about the NAIC's approach or declined to put their remarks on the record.

PF2's Phillips said he would like to see further discussion of the NAIC's approach. If any parties believe it to be flawed, he said, it would be good for everyone if those faults were clearly articulated.

A centralized modeling system would almost certainly rob boutique firms like PF2 of bank clients. But Phillips said he and his partners at PF2 had expected that at least that portion of their business would last only as long as it took to implement a dependable centralized system. The surprise, Phillips said, is that his company is still going strong.

"Our business was created to bring alternative color to an opaque market which wasn't trading — to satisfy a temporary gap created by the loss of confidence in the accuracy of credit ratings. That disconnect was probably wider and deeper than we had originally anticipated, but it's almost a pity that the gap still exists," he said.

Cole also saw a sense of urgency. "The credit ratings issue is one that I think community banks would get more interested in if they realized what the problems were," he said. "The Basel standardized approach is being held up by the elimination of credit ratings, and most small banks would benefit from Basel II's lower risk weights."

Subscribe Now

Access to authoritative analysis and perspective and our data-driven report series.

14-Day Free Trial

No credit card required. Complete access to articles, breaking news and industry data.