Turning to a U.S. wildfire risk, which was included in one multiperil catastrophe bond that S&P recently rated, Mr. Zuber noted that the rating agency would never rate a bond that only covers wildfires because of the potential for a non-natural factor--arson--to trigger the events.

In the bond that S&P rated, the contribution of wildfire risk to the deal was minimal, he said.

"The probability of attachment for one of the bond tranches (segments or tiers) was on the order of 2 percent," he said. In addition, "even if the modeling had been off by a factor of 1,000 percent," or if the insurer were to suffer wildfire losses 10-times greater than it had in the most recent large event--the $85 million for the October 2007 California wildfires--the attachment point for that tranche would not be breached, he noted.

The rating he was referring to was for East Lane Re II notes issued by Chubb and rated by S&P in early April. S&P assigned a "double-B" to two of the tranches and a "B-minus" to a third.

Mr. Zuber said S&P has also never gotten comfortable with rating cat bonds linked to terrorism risk, again highlighting "the potential for human activity" to impact the exposure as the reason.

In spite of efforts of catastrophe modeling firms to convince him otherwise, "I fail to understand how anyone can model the possibility of someone walking into Grand Central with a backpack and a bomb," he said.

At a separate session, Peter Nakada, managing director of RMS Consulting, noted that another rating agency, Moody's, did rate the first-ever terrorism cat bond (A3) for World Cup soccer games in Germany in 2006.

Referring to S&P's unwillingness to issue terror bond ratings, and more general marketplace skepticism about the ability of modeling firms to quantify the probability of terror attacks, he told the audience that RMS scientist Gordon Woo believes RMS is "better able to estimate the probabilities of terrorist attacks than a hurricane."

"The behind-the-scenes reason" for that is that for every terrorist plot "we hear about in the mainstream media, there are nine others that don't actually get reported [in the media] for security reasons."

Mr. Nakada said his firm has tapped into nonclassified sources of that data, noting that the probabilities RMS determines for modeling purposes are the odds that a plot gets hatched and doesn't get foiled.

Turning to a different issue, both Mr. Nakada and Rodney Clark, managing director for S&P's Financial Institutions group, noted some changes in bond structures--a trend toward so-called "indemnified deals"--which the S&P representatives said can also be problematic from a rating perspective.

Mr. Nakada referred to this as "micro-trend or a blip" in a longer-term trend moving in the reverse direction--away from such deals and toward deals based in external triggers, such as industry loss indexes or parametric triggers (like the wind speed in a hurricane or the amount of shake during an earthquake).

He attributed the re-emergence of "indemnified deals" to the demise of a sidecar market in which insurer-specific risks were ceded to limited life reinsurers.

James Brender, an S&P director, explained that an indemnified deal is one that is triggered by losses specific to the issuer cedent.

"A specific cedent is going to perform differently than the industry as a whole because it can have [geographic] concentrations, favorable or unfavorable, and weaker or stronger selection of risk."

"So a key element of [analyzing] an indemnified cat bond is determining to what extent is the cedent likely to be an unfavorable performer after a major event," he said.

Mr. Clark conceded that S&P has "been slightly an annoyance to some members of the banking community" in the past year, because of its reluctance to rate some "indemnified deals."

Mr. Zuber explained that the problematic deals were ones that came from nonrated entities--those insurance companies for which S&P has not previously issued a financial strength rating.

While he said that a full public rating of the financial strength of an insurer is not required for the insurers to get a rating on an indemnified cat bond it issues, he also said S&P would probably have to know as much about a company's underwriting and risk management as it would know if it did issue such a full-blown rating.

Some nonrated entities "feel as if we would be somehow intruding if we want to get answers to what to me seem like very basic questions," he said. "We're not going to risk our franchise on taking someone's statements at face value" assuring S&P that they know what their doing.

He continued, "There have been a couple of occasions where [insurers] felt as if some of [these] hurdles we've put in front of them were a little too onerous--and I understand that they may have gone to other rating agencies."