Market Update

After the January renewals, we have witnessed dramatic price increases for Catastrophe coverage in certain areas of the country and an overall reduction in available market capacity. The market remains unstable, with little consensus among reinsurers on terms, and price fluctuations on a weekly, sometimes even daily basis. Unlike the aftermath of Northridge, 9/11 or the record-setting 2004 hurricane season, this is the first retraction in capacity since 1993, following Andrew. The retraction seen there lasted just one year. In Holborn’s view, the conditions experienced today are attributable to increased capital requirements from Lloyd’s, rating agencies and reinsurers’ managements, as opposed to impairments in reinsurers’ capital bases, which were generally either not reduced by the 2005 storms, or have since been replenished. This is particularly the case at Lloyd’s.

Holborn has had several broker teams in the market in recent weeks to discuss mid-year renewals with domestic, Bermuda and London reinsurers, traveling with clients with significant exposures in Florida, California and the Northeast. In addition, the Holborn modeling team recently conducted an in-depth discussion with Paul Nunn, Lloyd’s director of modeling. Reinsurers in all markets are giving increased attention this year to catastrophe exposure in the New York City area, Long Island, and eastern Massachusetts and these regions are experiencing the impact of that greater focus. While reinsurers require more capital for Catastrophe exposures in all significant industry accumulations, the effect is (moderately) less extreme in California and the Gulf Coast, where the changes are incremental additions to existing requirements. Florida is facing treaty capacity shortages for supply and demand reasons, in addition to the upcoming model changes and the higher capital standards that have recently been imposed.

2005 Results at Lloyd’s

Despite recording a strong performance in the first half of 2005, the combined impact of hurricanes Katrina, Wilma and Rita, totaling £3.3Bn (approximately $5.7Bn) net, resulted in the Lloyd’s market reporting a small overall GAAP loss of £103Mn and a calendar year combined ratio of 111.8%.

The breakdown of Lloyds’s net losses relating to the 2005 US hurricane seasons is:

Katrina	£2,209Mn
Rita	602Mn
Wilma	498Mn
Total	£3,309Mn

According to Luke Savage, Finance Director and Acting Chief Executive at Lloyds, “The industry continues to see good opportunities at Lloyd’s and many of our biggest insurers have expanded their businesses here in 2006. Capital providers responded in a disciplined way…, putting £1.2Bn of new money into the market. Capacity also increased in line with new business opportunities to £14.8Bn, a 7% rise on the previous year."

Lloyd’s Realistic Disaster Scenarios (RDSs)

The RDSs began about ten years ago, as a reaction to Andrew. Initially, they were subjective and were largely used for syndicates’ internal purposes. In this way, while all syndicates looked at the same market loss amount, they used different values of windspeed, Richter scale, etc. About three years ago, Lloyd’s decided to become “prescriptive” about the loss scenarios used, and to use them as metrics for evaluating syndicate capital adequacy, exposure and capacity levels. This converted the RDSs to a relative measure, requiring equivalent meanings across syndicates and scenarios, and across the three modeling firms.

Katrina demonstrated that even the new, more specific RDS figures were not being calculated equivalently by the individual syndicates. Like A.M. Best, Lloyd’s discovered that not all syndicates had “turned on all of the model switches”. The RDSs now specifically include demand surge, Flood, APD, Life/PA, FFEQ, and Marine and Aviation, especially at ports and airports. Based on this broader view, the new industry numbers appear to be higher. However, the scenario is the same physical storm as last year in Miami, and actually a weaker storm than last year’s assumption for Tampa.

Lloyd’s also discovered that syndicates’ geo-coding wasn’t very detailed outside of the treaty market. Direct and Fac business have a noticeable problem in this regard and binding authorities have a major problem.

The New RDSs

Lloyd’s made a conscious decision that all US hurricane RDS scenarios should stress the system to the same degree, even if the probabilities aren’t exactly the same. This approach points out the inconsistency of requiring enough capital to survive a $100 billion Miami loss, while succumbing to a $51 billion Houston loss, even though both might be 1-in-100 likelihoods. The US hurricane scenarios are for storms in Miami, Tampa, and in Houston together with offshore platforms, and a two-storm scenario involving both the Northeast and Charleston. Each totals about $100Bn of Property loss plus some Auto and non-rig Marine. The California earthquake scenarios are smaller, but are equivalent stresses at $65Bn, each. New Madrid is tested against both $40Bn and $90Bn events, but only the answer from the $40Bn scenario is currently considered. In total, there are ten US scenarios (the categories of New York Terror and New Madrid each have two) that all syndicates must report.

The two-event scenario involving the Northeast and Charleston caused the most reaction from the syndicates back to Lloyd’s Risk Management department. The reason for the two storms assumed this year is a concern that most 2005 industry loss warranty retrocessions were “one-shot” covers, and last year’s combined scenario of an earthquake and a storm would not reveal the weakness of a nonreinstatable wind cover. That change may have helped contribute to the recent decline in retrocessional placements.

A commonly heard criticism of this scenario is that the Charleston part at $35Bn is unrealistically severe, given the characteristics of the loss. The damage factors peak at 12% or about twice the damageability for the New York event. Lloyd’s feels that a large loss could happen outside of the high-value states of Florida, New York and Texas, and it had to be modeled somewhere. (Allstate ignored this fact before leaving Mississippi bare to Katrina.) In the past, Lloyd’s was concerned that syndicates were growing too much outside of where the RDSs were sited, and having a large, fairly random landfall location helps prevent against that tendency.

The New York area landfall was designed to complement the Charleston landfall and create a combined stress to the market equivalent to the $100Bn events in Florida and Texas. It was also consciously placed there because Lloyd’s felt that the high Northeast concentration was not sufficiently appreciated, and to test a significant degree of binding authorities operating along the Jersey shore. Lloyd’s damage assumptions on the New York landfall of this scenarios are:

The New York loss is likely still a tropical storm (prior to undergoing “transitioning”.) As in the 1938 storm, the landfall is in Nassau county, and the right side has the typical greater strength (hitting Suffolk harder than Nassau). But in contrast, the RDS storm stays together better than 1938’s and there are relatively higher damage factors in the City and Jersey shore. That will cause significant losses for syndicates with shares of the high values there. Unlike 1938, the right side doesn’t get “bloated” by the transitioning process and the Boston suburbs are largely spared. Holborn believes the storm was consciously “widened” to test both New Jersey and eastern Long Island, and a more likely category 3 event would be narrower, and locally more intense.

There are two New York City terror attacks. Both are two-ton truck bombs at, respectively, Rockefeller Center and 20 Exchange Place (Hannover Square). These are the only terrorism scenarios modeled worldwide, reflecting the conventional wisdom that Manhattan’s dense values and symbolic importance are the world’s peak insured accumulations of Terrorism exposures. The map below illustrates the damage assumptions for the lower Manhattan attack.

There are three mandatory RDSs outside of the US for Japanese earthquake and wind and northern European wind. All are smaller losses than the US events due to lesser values, and in the case of Japan Earthquake coverage, very limited take-up. Writers of Space, Aviation, “Blue Water” Marine, Political Risk, E&O, etc., may be required to submit up to an additional seven RDSs for these lines. Syndicates with heavy exposures not considered by the specified scenarios (Australian Earthquake, for example) should supply one or two “free-form” RDSs. The industry losses for the mandatory US scenarios are:

How the RDS Results Are Used

Syndicates provide their RDS results to the Lloyd’s Risk Management department, the UK Financial Services Authority (FSA) and the members’ agents. The FSA can use them in consideration of the syndicates’ business plans, although that is driven more by a process similar to the NAIC’s RBC or Best’s BCAR. The members’ agents are now using them to help manage underwriting members’ spreads, and allot capacity, say, to different syndicates with their respective peak RDS events in different places.

Some of the publicly-traded vehicles are now referring to their RDS amounts in their shareholder disclosures. Also, several syndicates are running additional RDSs internally. BRIT reports that they use about 50 different scenarios. Since syndicates now manage to the RDSs, they have become an economic reality. Underwriters must give increased weight to the scenarios and less to probalistic models.

Lloyd’s stated target is that a syndicate should have no modeled RDS amount that exceeds 75% of stamp capacity (that is, their maximum allowed premium volume) on a gross basis or 20% net, which they believe corresponds to an “AAA” capital level. However, Lloyd’s ability to manage any exceptions to the target are limited, because the FSA now also approves the syndicates’ business plans. But Lloyd’s can entirely shut down a syndicate, reduce its stamp, or conditionally approve the plan, subject to changes. So, syndicates with excessive RDSs must negotiate with Lloyd’s Risk Management department.

A common negotiated solution is an agreed phasing down of aggregate liabilities in the over-lined zone. Buying more retrocessions is a theoretical possibility but likely to be impractical in the 2006 market.

Committing more capital is possible, at least for integrated syndicates that own their capital. (Names’ preemption rights make that difficult for syndicates with traditional structures.) Syndicates have also been able to argue for extra leeway. This has been granted in at least two cases where syndicates could demonstrate conservatively calculated RDSs and prove them based on their actual 2004 – 2005 storm performance.

The biggest net capacity cutbacks have come in Direct and Fac and binding authority business. Wellmodeled treaty books had relatively fewer surprises from the RDSs. That is most true in the Florida, Gulf, California and Japan scenarios that are largely in the same places as last year and less so in New York and the northeast. The absence of retrocessional capacity has caused more significant treaty capacity cutbacks on a gross basis.

The RDSs will continue to be reviewed by Lloyd’s for periodic updates. The impact of this update was extreme because of the lessons from Katrina and because of Lloyd’s new philosophy on what the RDSs mean. Lloyd’s expects that changes in the future will be more incremental.

Other Reinsurance Markets

A.M. Best has made similar capital model changes, looking to understand how reinsurers have developed their catastrophe model estimates, validating the estimates against historical losses, and including estimates for exposure types that may be excluded from the models, such as business interruption, fire following, demand surge, etc. In addition, Best is also explicitly looking to a second event, again, stressing the importance of retrocessional coverage with a reinstatement. Some Catastrophe reinsurers that were surprised by the magnitude of their Katrina losses have voluntarily imposed additional capacity restrictions to avoid further surprises. These changes are causing net capacity reductions similar to those at Lloyd’s.

Prognosis for Later 2006/2007 Renewals

Florida in particular has significantly increased demand for coverage, industrywide, at a time of reduced supply. A.M. Best’s changes to the primary companies’ capital and rating standards are requiring both large and small companies to buy more coverage. Several national writers have purchased additional reinsurance exposed to Florida since January. The smallest companies, especially Florida Homeowner specialists, are finding that the price and supply of coverage are making it difficult or impossible to fill out prudent reinsurance programs. Since late February, it has been apparent that several major reinsurers do not have sufficient aggregate catastrophe capacity available for Florida to renew even their current business. They are selectively non-renewing parts of their portfolios, based largely on a commodity view of the price for their capacity. PXRe and Quanta had significant business that now must be replaced as well. Prices have risen rapidly as a result.

Holborn expects this climate to continue through at least the July 1 renewals. However, the market’s dislocation is already drawing in some new capital. Hedge funds are becoming more active in the retrocessional area, and are talking to large, national writers about additional Catastrophe placements on a Cat bond, swap or fronted basis. This capital, perhaps supplemented with reinsurers’ earnings during 2006, could slow the rate of hardening during the year. It is difficult to forecast whether yearend renewals will face the same percentage increases as mid-year renewals and obviously, the 2006 hurricane season is a “wild card”. However, even assuming good weather, it appears that there will be significant price increases at January 1, 2007 above January 1, 2006 terms, and ceding companies may find capacity shortages restricting their ability to increase limits.

Vendor Models and Lloyd's RDS Support

Lloyd’s has furnished each of the major cat modeling vendors (AIR, EQECAT, and RMS) with details about their RDS scenarios. In turn, each of these three vendors have included these events in their Hurricane and Earthquake modeled loss scenarios for the US, Europe and Japan. Each RDS has a unique event ID in each model, meaning there is a single scenario that each Syndicate can run to model the RDS requirements. These event IDs match up to events with the specified location (landfall or fault rupture) and magnitude associated with the particular RDS scenario. Each event can be run deterministically, and the expected loss and standard deviation reported back to Lloyd’s. All events include demand surge and hurricane events include storm surge.

RMS has created and released additional “footprint” analyses to use within their Terrorism model to assist with the US Terrorism RDS scenarios. Three separate footprints for each scenario are being released: one for Property covered for Terrorism, one for Workers’ Compensation and one for fire following the event on exposures that exclude Terrorism. These footprints can only be run, however, after the annual update for the Terrorism, as the Lloyd’s scenarios need to be included in to the RMS model.

If a syndicate chooses to use modeling software to calculate the RDS returns, Lloyd’s accepts any of the three vendor models. Syndicates, however, are not required to use any model in producing their RDS loss estimates.

Summaries of the 2006 Scenarios

US Scenarios

Scenario	Location	Approximate Magnitude	Maximum Damage	Industry Loss
Two Events	Nassau, NY and Charleston, SC	SS3 SS4	5% 12%	$67Bn $30.75Bn $97.75Bn
Florida Windstorm	Miami-Dade Tampa-Pinellas	SS5 SS5	20% 65%	$103Bn $103Bn
California EQ	Los Angeles San Francisco, San Mateo	7.5 7.7	20% 40%	$74Bn $74Bn
New Madrid	Dunklin County, MO or Mississippi County, MO	7.0 7.5	90% 100%	$45Bn $100Bn
Terrorism	Rockefeller Center Exchange Place	2-ton 2-ton	100% 100%	$14.5Bn* $15.4Bn*
Gulf of Mexico Windstorm	Houston and Offshore	SS5	65%	$102Bn

Damage Maps for Key US Scenarios

Specialty Line Scenarios

Scenario	Location	Description	Effected Lines
Marine event	Prince William Sound	Tanker-Cruise ship collision	P&I*, Hull, Pollution
	or Florida	Cruise ship sinking	P&I, Hull, Pollution
Loss of major complex	North Sea (likely)	Total loss	Rig
Aviation collision	Above major city	Two passenger jets collide	Aviation Hull, Liability, On-ground Property
Major Risk loss	Syndicate’s largest Risk	Total loss	Treaty, Direct, Fac
Satellite		Solar proton flare or Generic design defect	Space Space
Liability Risks	Various	One of several professional liability scenarios
		or Industrial/ Transport disaster	Casualty, Pollution
		or Multiple Public/Products losses	Casualty
		or Collapse of UK housing market	Professional, Mortgage, Unemployment
		or Collapse of recently merged companies	D&O, E&O, FI
Political Risk	SE Asia or South America or Middle East	Various	Political Risk

*Projection and Indemnity coverage is a form of Marine Liability.