Correspondence

CORRESP 1 filename1.htm Correspondence

June 4, 2010

VIA FEDERAL EXPRESS AND EDGAR

Mr. Jim B. Rosenberg

Division of Corporation Finance

Securities and Exchange Commission

100 F Street, N.E.

Washington, D.C. 20549


Re:		AXIS Capital Holdings Limited
		Form 10-K for Fiscal Year Ended December 31, 2009
		File No. 001-31721

Dear Mr. Rosenberg:

We are responding to the comment letter (the “Comment Letter”) from the staff (the “Staff”) of the Securities and Exchange Commission (the “Commission”), dated May 11, 2010, concerning AXIS Capital Holdings Limited’s (the “Company”) Form 10-K for the fiscal year ended December 31, 2009. For the convenience of the Staff, we have repeated your comments (displayed in italics) immediately before our responses.

Based on our consideration of the Staff’s comments, and as set forth in our responses below, we intend to include disclosures similar to those proposed herein for future annual and/or quarterly (as applicable) filings.

Management’s Discussion and Analysis of Financial Condition and Results of Operation

Critical Accounting Estimates

Reserve for losses and loss expenses

Potential Volatility in Our Estimates, page 99

The second paragraph indicates that, in establishing IBNR, management records additional IBNR for uncertainties relating to the timing of the emergence of claims. We think that this disclosure may be better suited in the discussion of IBNR reserving methodology beginning on page 93. If you agree, please consider deleting your current discussion of additional IBNR in the second paragraph on page 99 or revise it to explain how it relates to the volatility displayed in the tables that follow. Also, please revise the IBNR reserving methodology disclosure to describe the additional IBNR and to disclose:

•

the amount of additional IBNR included in IBNR in total and by line of business as provided in the table under General beginning on page 90; and

•

the specific facts and circumstances that caused you to record the additional IBNR in total and for each line.


Page		1

In addition, please integrate a discussion of the amounts and reasons for changes in the additional IBNR in your prior period reserve development disclosure beginning on page 53.

Response

Beginning with our Form 10-K for the year ending December 31, 2010, we will revise our disclosure similar to that proposed in Exhibit A, which has been marked to show changes from the disclosure contained in our December 31, 2009 Form 10-K.

We agree that disclosures related to management’s consideration of qualitative factors in establishing IBNR point estimates on page 99 may be better situated as part of the discussion of our IBNR reserving methodology beginning on page 93. As such, in Exhibit A we have moved this disclosure from the section on the potential volatility of our loss reserve estimate on page 99 to the loss reserve methodology section beginning on page 93.

Upon review of the Staff’s comment, we acknowledge that the previous wording on page 99 with respect to our recognition of “additional IBNR for uncertainties relating to the timing of the emergence of claims” may have unintentionally implied that we maintain separate and distinct “additional IBNR” provisions in response to such qualitative uncertainties. However, this is not the case. Rather, when selecting our single point best estimate, we consider quantitative actuarial analyses, which rely to a certain extent on industry benchmarks, in conjunction with more qualitative factors such as the uncertainties surrounding the timing on the emergence of claims. We do not, however, record a specific or separate provision for these uncertainties. In Exhibit A, we have enhanced our disclosure on page 93 to clarify the approach used by management in selecting their best estimate of ultimate losses.

Notes to Consolidated Financial Statements

Note 5: Investments

d) Net Realized Investment Gains (Losses), page 150

2.	Please revise your disclosure of your other-than-temporary credit-related impairments to include the significant inputs, including quantification of these inputs, used to measure the amount of your credit losses. Please see ASC 320-10-50-8A.

Response

Beginning with our Form 10-Q for the quarter ending June 30, 2010, we will revise our disclosure similar to those proposed in Exhibit B, which has been marked to show changes from the disclosure contained in our December 31, 2009 Form 10-K.


Page		2

We also intend to enhance our critical accounting estimate disclosure for OTTI in the Management’s Discussion and Analysis by providing similar quantitative disclosures as in Exhibit B, beginning with our Form 10-K for the year ending December 31, 2010.

* * *

In addition, the Company acknowledges the following:

The Company is responsible for the adequacy and accuracy of the disclosure in the filings.

Staff comments or changes to disclosure in response to Staff comments in the filings reviewed by the Staff do not foreclose the Commission from taking any action with respect to the filing.

The Company may not assert the Staff’s comments as a defense in any proceeding initiated by the Commission or any person under the federal securities laws of the United States.

If you have any questions regarding the responses to these comments or require additional information, please contact me at (441) 405-2695.

Very truly yours,

/s/ David B. Greenfield

David B. Greenfield

Executive Vice President and Chief Financial Officer


cc:		Mr. Mark Parkin, Deloitte & Touche LLP
		Mr. Stephen Kuzyk, Deloitte & Touche
		Mr. Gary Horowitz, Simpson Thacher & Bartlett LLP


Page		3

EXHIBIT A

RESERVE FOR LOSSES AND LOSS EXPENSES

General

We believe the most significant judgment made by management is the estimation of our reserve for losses and loss expenses, which we also refer to as loss reserves. We are required by U.S. GAAP to establish loss reserves for the estimated unpaid portion of the ultimate liability for losses and loss expenses (“ultimate losses”) under the terms of our policies and agreements with our insured and reinsured customers. Our loss reserves comprise the following components:

•

Case reserves - cost of claims that were reported to us but not yet paid, and

•

IBNR reserves - anticipated cost of claims incurred but not reported.

Loss reserves also include an estimate of the expense associated with settling claims, including legal and other fees and the general expenses of administering the claims adjustment process. The following tables show our unpaid net loss reserves segregated between case reserves and IBNR and by segment at December 31, 2009 and 2008:


	2009						2008
At December 31,	Gross		Ceded		Net		Gross		Ceded		Net
Case Reserves	$	1,895,297	$	369,459	$	1,525,838	$	2,055,027	$	408,754	$	1,646,273
IBNR		4,668,836		1,011,599		3,657,237		4,189,756		905,797		3,283,959

Total	$	6,564,133	$	1,381,058	$	5,183,075	$	6,244,783	$	1,314,551	$	4,930,232


Page		4

INSURANCE


	2009								2008
At December 31,	Gross			Ceded		Net			Gross		Ceded		Net
Property
Case reserves	$	318,418		$	144,401	$	174,017		$	519,871	$	222,230	$	297,641
IBNR		197,022			61,053		135,969			263,068		84,757		178,311

		515,440			205,454		309,986			782,939		306,987		475,952


Marine
Case reserves		156,909			20,952		135,957			249,869		12,510		237,359
IBNR		155,008			39,544		115,464			147,192		25,599		121,593

		311,917			60,496		251,421			397,061		38,109		358,952


Aviation
Case reserves		28,729			—		28,729			30,024		—		30,024
IBNR		47,322			1,548		45,774			55,131		2,389		52,742

		76,051			1,548		74,503			85,155		2,389		82,766


Credit and Political Risk
Case reserves		(40,928	)		—		(40,928	)		13,788		—		13,788
IBNR		215,031			—		215,031			108,830		—		108,830

		174,103			—		174,103			122,618		—		122,618


Professional Lines
Case reserves		264,735			109,700		155,035			225,878		95,036		130,842
IBNR		1,305,842			478,844		826,998			1,144,247		417,199		727,048

		1,570,577			588,544		982,033			1,370,125		512,235		857,890


Liability
Case reserves		172,954			94,406		78,548			139,950		78,978		60,972
IBNR		681,638			392,695		288,943			649,223		347,642		301,581

		854,592			487,101		367,491			789,173		426,620		362,553


Insurance Total
Case reserves		900,817			369,459		531,358			1,179,380		408,754		770,626
IBNR		2,601,863			973,684		1,628,179			2,367,691		877,586		1,490,105

Total	$	3,502,680		$	1,343,143	$	2,159,537		$	3,547,071	$	1,286,340	$	2,260,731


Page		5

REINSURANCE


	2009						2008
At December 31,	Gross		Ceded		Net		Gross		Ceded		Net
Catastrophe, Property and Other
Case reserves	$	370,653	$	—	$	370,653	$	437,072	$	—	$	437,072
IBNR		467,281		—		467,281		478,985		—		478,985

		837,934		—		837,934		916,057		—		916,057


Credit and Bond
Case reserves		88,476		—		88,476		58,303		—		58,303
IBNR		131,038		—		131,038		81,316		—		81,316

		219,514		—		219,514		139,619		—		139,619


Professional Lines
Case reserves		185,586		—		185,586		124,735		—		124,735
IBNR		688,832		790		688,042		653,018		3,313		649,705

		874,418		790		873,628		777,753		3,313		774,440


Motor
Case reserves		229,132		—		229,132		165,470		—		165,470
IBNR		190,199		—		190,199		155,065		—		155,065

		419,331		—		419,331		320,535		—		320,535


Liability
Case reserves		120,633		—		120,633		90,067		—		90,067
IBNR		589,623		37,125		552,498		453,681		24,898		428,783

		710,256		37,125		673,131		543,748		24,898		518,850


Reinsurance Total
Case reserves		994,480		—		994,480		875,647		—		875,647
IBNR		2,066,973		37,915		2,029,058		1,822,065		28,211		1,793,854

Total	$	3,061,453	$	37,915	$	3,023,538	$	2,697,712	$	28,211	$	2,669,501

Case Reserves

For reported losses, management primarily establishes case reserves based on the amounts reported from insureds or ceding companies. Case reserves are established on a case by case basis within the parameters of coverage provided in the insurance and reinsurance contracts. The method of establishing case reserves for reported losses differs among our segments.

With respect to our insurance operations, we are notified of insured losses by brokers and insureds and record a case reserve for the estimated amount of the ultimate expected liability arising from the claim. The estimate reflects the judgment of our claims personnel based on general reserving practices, the experience and knowledge of such personnel regarding the nature of the specific claim and, where appropriate, advice of counsel, loss adjusters and other relevant consultants.

The reserving process for our reinsurance operations is more complicated than for our insurance operations. For reported losses, we generally establish case reserves based on reports received primarily from brokers and also from ceding companies. With respect to contracts written on an excess of loss basis, we typically are


Page		6

notified of insured losses on specific contracts and record a case reserve for the estimated amount of the ultimate expected liability arising from the claim. With respect to contracts written on a pro rata basis, we typically receive aggregated claims information and record a case reserve based on this information. However, our pro-rata reinsurance contracts typically require that pre-defined large losses must be separately notified so that these losses can be adequately evaluated.

In deciding whether to provide treaty reinsurance, we carefully review and analyze the cedant’s underwriting and risk management practices to ensure appropriate underwriting, data capture and reporting procedures. We undertake an extensive program of cedant audits, utilizing outsourced legal and industry expertise when necessary. This allows us to review a cedant’s claims administration to ensure that its claim reserves are consistent with reinsured exposures, are adequately established and are properly reported in a timely manner, and to verify that claims are being handled appropriately. For those losses where we receive contract-specific loss notifications, our claims department evaluates each notification and, as discussed above, may record additional case reserves if claims are not considered to be adequately reserved by the ceding company. This requires considerable judgment. At December 31, 2009, additional case reserves were $145 million, or 8% of our total case reserves compared to $106 million, or 5%, respectively at December 31, 2008.

IBNR

IBNR reserves are necessary due to time lags between when a loss occurs and when it is actually reported and settled. This is often referred to as the “claim-tail”. Reporting lags arise from the use of intermediaries to provide loss reports, complexities in the claims adjusting process and other related factors. IBNR reserves are calculated by projecting our ultimate losses on each class of business and subtracting paid losses and case reserves.

Unlike case reserves, IBNR is generally calculated at an aggregate level and cannot usually be directly identified as reserves for a particular loss or contract ~~(see ‘Specific Reserving Issues – Catastrophe losses’ for discussion of specific IBNR provisions)..~~ Our loss and premium data is aggregated by exposure class and by accident year (i.e. the year in which losses were incurred).

The evaluation process to determine our ultimate losses involves the collaboration of our underwriting, claims, internal actuarial, legal and finance departments, and includes various segmental committee meetings, culminating with the approval of a single point best estimate by senior management in our Group Reserving Committee. In conjunction with the reserving methodology outlined below, the evaluation process for establishing a single point best estimate takes into consideration qualitative factors that create uncertainties to the ultimate cost of the claims settlement process. These qualitative factors include, but are not limited to: the timing of the emergence of claims, volume and complexity of claims, social and judicial trends, potential severity of individual claims, and extent of internal historical loss data versus industry information due to our relatively short operating history. Management does not record a specific provision for any of these uncertainties. The evaluation process also includes consultation with an independent actuarial firm. The work performed by the actuarial firm is an important part of the reserving process and we compare our recorded claims and claim expense reserves to those estimated by the actuarial firm to determine whether our ~~estimates are~~ single point best estimate is reasonable.

On an annual basis, our independent actuarial firm performs work for the purpose of issuing an actuarial opinion on the reasonableness of our loss reserves for each of our operating subsidiaries. The actuarial opinions are required to meet various insurance regulatory requirements. The actuarial firm discusses its conclusions with management and presents its findings to our Board of Directors.

Reserving Methodology:

We primarily use the following actuarial methods in our reserving process:

•

Initial expected loss ratio method (“IELR”): This method calculates an estimate of ultimate losses by applying an estimated loss ratio to an estimate of ultimate earned premium for each accident year. The estimated loss ratio is based on pricing information and industry data and is independent of the current claim experience to date. This method is appropriate for classes of business where the actual


Page		7

paid or reported loss experience is not yet mature enough to override our initial expectations of the ultimate loss ratios.

•

Bornhuetter-Ferguson (“BF”): The BF method uses as a starting point an assumed IELR and blends in the loss ratio implied by the claims experience to date by using benchmark loss development patterns. This method is generally appropriate where there are few reported claims and a relatively less stable pattern of reported losses.

•

Loss development (Chain Ladder): This method uses actual loss data and the historical development profiles on older accident years to project more recent, less developed years to their ultimate position. This method is appropriate when there is a relatively stable pattern of loss emergence and a relatively large number of reported claims.

The basis of our selected single point best estimate on a particular line of business is often a blend of the results from two or more methods (e.g. weighted averages). Our estimate is highly dependant on actuarial and management judgment as to which method(s) is most appropriate for a particular accident year and class of business. Our methodology changes over time as new information emerges regarding underlying loss activity and other factors.

Our Key Reserving Assumptions:

Implicit in the actuarial methodologies utilized above are two critical reserving assumptions; the selected IELR for each accident year and the expected loss development profiles. We regularly monitor these assumptions and, at each quarter end, undertake a full actuarial review. Any adjustments that result from this review are recorded in the quarter in which they are identified. The historic loss information we use is also assumed to be indicative of future loss development and trends.

The IELR selections in our insurance segment are primarily developed using industry benchmarks with varying degrees of weight given to our own historical loss experience. We also give consideration to a number of other factors, including exposure trends, rate adequacy on new and renewal business, ceded reinsurance costs, changes in claims emergence and our underwriters’ view of terms and conditions in the market environment. In our reinsurance segment, our IELR selections are based on a contract by contract review which incorporates information provided by clients together with estimates provided by our underwriters and actuaries concerning the impact of changes in pricing, terms and conditions and coverage. Our estimate of the impact of these changes includes assumptions which consider, among other things, the market experience of our independent actuarial firm.

Our loss development profiles are primarily developed using industry benchmarks with varying degrees of weight given to our own historical loss experience. Having begun operations in late 2001 and having grown our business substantially since, the credibility of our own loss development profiles have generally been limited. Our development profiles are only adjusted when the weight of our own actual experience becomes sufficiently credible to identify deviations from the market based assumptions. As this happens, we incorporate the experience from these accident years in our actuarial analysis to determine future accident year expected loss ratios, adjusted for the occurrence or lack of large losses, changes in pricing, loss trends, terms and conditions and reinsurance structure.

Reserving by Class of Business:

The weight given to a particular actuarial method is dependent upon the characteristics specific to each class of business, including the types of coverage and the expected claim-tail.

Short-Tail Business:

Short-tail exposures describe classes of business for which losses are usually known and paid within a relatively short period of time after a loss event has occurred. Our short-tail exposures include the majority of the property, terrorism and marine lines of business within our insurance segment, together with the property, catastrophe and crop lines of business within our reinsurance segment.


Page		8

Due to the relatively short reporting development pattern for short-tail lines of business, our estimate of ultimate losses responds quickly to the latest loss data. We therefore typically assign higher credibility to methods that incorporate actual loss emergence, sooner than would be the case for long-tail lines at a similar stage of development.

Although our reserve estimate for short tail business has inherently less uncertainty than longer tail business, it is still subject to significant judgment. For example, because much of our excess insurance and excess of loss reinsurance business has high attachment points, it is often difficult to estimate whether claims will exceed those attachment points. Also, the inherent uncertainties relating to coverage and damage assessment on catastrophe events (see below), together with our typically large line sizes, further add to the complexity of estimating our potential exposure. Additionally, for workers compensation catastrophe reinsurance business, our estimate of ultimate losses requires us to estimate longer term and potentially more variable costs, such as ongoing medical expenses. This business therefore generally has a longer development profile when compared to property catastrophe business.

The reserving process for losses arising from catastrophic events typically involves the determination by our claims department, in conjunction with our underwriters and actuaries, of our exposure and likely losses immediately following an event with subsequent refinement of those losses as our clients provide updated actual loss information. When a catastrophe event occurs, we review our contracts to determine those that could be potentially exposed to the event. We contact brokers and clients to determine their estimate of involvement and the extent to which their programs are affected. We may also use commercial vendor models to estimate loss exposures under the actual event scenario. As part of the underwriting process, we obtain exposure data from our clients, so that when an event occurs we can run the models to produce an estimate of the losses incurred by clients on programs that we insure or reinsure. Typically, we derive our estimate for the losses from a catastrophic event by blending all of the sources of loss information available to us. This estimate is derived by the claims team and, where there are no reported case reserves, we establish a separate provision for IBNR. Natural catastrophe losses were low in 2009 due to the absence on the whole of major catastrophes and a benign North Atlantic hurricane season.

For the 2009 accident year, our short-tail line loss reserves were typically established using the BF method, which, as noted previously, requires initial expected loss ratios by line of business, adjusted for actual experience during the year.

During 2009, we continued to incorporate more of our own historical loss experience within short-tail lines of business. Given our own loss experience has generally been more favorable than we expected, the incorporation of this data into our reserving analysis had the impact of reducing net loss ratios on several lines of business, relative to 2008. Otherwise, for short-tail businesses taken as a whole, our loss trend assumptions for accident year 2009 did not differ significantly from prior years.

In terms of prior accident years, changes to ultimate loss estimates on short-tail lines of business in 2009 were primarily in response to the latest reported loss data rather than any significant changes to underlying actuarial assumptions such as loss development patterns. As discussed in the ‘Underwriting Results – Group – Prior Period Development’ section, we have experienced significant net favorable reserve development on short-tail lines of business during the last three years.

This favorable development partly stemmed from the fact that historically we had relied heavily upon industry-based profiles. Due to the inherent limitations of this, our loss reserves in prior years have also included a provision for reporting delays and other uncertainties specific to our business. These include the inherent delays we expect to arise from obtaining loss information on excess layers of business across our diverse worldwide exposures. Also, within our insurance segment, for certain business, we use managing general agents and other producers, which can delay the reporting of loss information to us. As it has transpired, our actual claims experience in the last three years has been more favorable than we projected, with late reporting being less prevalent than we anticipated.


Page		9

In addition to this broader claims experience, favorable prior period reserve development has also occurred as a result of reductions to individual case reserves following updated loss information received from insureds or ceding companies. Also, for certain specific catastrophe events, we have also taken down our own provisions from prior periods. On contracts that respond to highly visible, major events, we establish IBNR where potential exposure has been identified. However, in a number of instances, mainly within our excess of loss catastrophe reinsurance business, it transpired that claims did not develop to a sufficient level to reach our attachment points.

Medium-Tail Business:

Our medium-tail exposures include the majority of the aviation hull, energy offshore, professional lines and credit and political risk lines of our insurance segment together with most of the trade credit and bond, professional lines and engineering lines of our reinsurance segment. For medium tail business, we generally use the IELR method on more recent accident years and the BF method on older accident years. The significant changes to our reserving for medium tail business in 2009 are described below.

For professional lines business, claim payment and reporting patterns are typically medium to long tail in nature. The underlying business is predominately written on a claims-made basis with the majority of reinsurance treaties written on a risk attaching basis. During 2009, we continued to give weight to our own loss experience, in particular business written on a claims-made basis from accident years 2006 and prior, which has developed a reasonable level of credible data. Generally once the percentage of reported losses in a particular accident year is assumed to have reached 70% of estimated ultimate losses, the selected reserving methodology is gradually transitioned away from sole reliance on the IELR method and, over the course of the next two years, towards sole reliance on the BF method. For more recent accident years, we continue to use the IELR method, although our ultimate loss estimates for accident year 2007 and 2008 are weighted more heavily towards our expected loss exposure to the economic downturn and the sub-prime lending credit crisis. Our reserves for the credit crisis incorporate analyses by our claims personnel, actuaries and underwriters of known notifications of potential loss, as well as a review of accounts that may have exposure to this area, but have not yet provided notice of a claim. During 2009, we strengthened our accident year 2008 professional lines reserves in response to the continued economic downturn and credit crisis.

For accident year 2009 trade credit and bond reinsurance business, we increased our IELRs relative to prior years due to the weaker economic environment. We also strengthened our accident year 2008 reserves in 2009 due to deteriorating loss experience reported by our cedants. For earlier accident years, we continued to give weight to our own loss experience, in particular accident years 2005 to 2007. Our selected reserving methodology is gradually transitioned from a sole reliance on the IELR method to the BF method, generally after two years for trade credit business and after three years for bond reinsurance business.

Claims for credit and political risk business tend to be characterized by their severity risk as opposed to their frequency risk and tend to be heterogeneous in nature. Therefore, claim payment and reporting patterns are anticipated to be volatile. Under the notification provisions of our non-sovereign credit insurance, we anticipate being advised of an insured event within a relatively short time period. Generally, these contracts include waiting periods following the event which specify that the claim payment is due only after specified waiting periods. In some cases, resolution can be achieved during the waiting period. Also of note, a feature of many of these contracts is that after the date we pay a claim, we are generally subrogated to all of the insured’s rights of recovery under the insured agreement. In some situations, we may also receive a transfer or assignment of the insured’s rights. This can lead to the situation where we pay a claim in the short term, but receive a recovery over a longer period of time. We anticipate that this will likely lead to claim reporting patterns that will have a medium development tail. At December 31, 2009, we had negative gross case reserves of $41 million.

During 2009, the level of recoveries due on credit insurance business (within our credit and political risk line of business) increased significantly relative to prior years (see below). At December 31, 2009, our estimate of recoveries due was $340 million, of which $87 million related to recoveries on paid losses, $160 million on case reserves and $93 million on IBNR reserves. The estimate of such recoveries requires significant management judgment. Our estimate requires us to assess the post event circumstances, including any restructuring, liquidation or possession of asset proposals/agreements. The process involves


Page		10

compiling information from several third party sources, including rating agencies, asset valuation specialists and publicly available information.

Our expected current accident year loss ratio for credit and political risk business increased significantly from prior years as a result of increased loss activity emanating from the global financial crisis. The increase included a reserving provision related to one peak credit insurance risk which has been stressed by the global economic downturn. No claim has been presented to date. However, we had been closely monitoring the status of the project and identified an opportunity to work and negotiate with interested parties to settle our exposure. We have reserved for the ultimate amount payable that we believe will eliminate our exposure. In terms of prior accident years, in 2009 we experienced favorable reserve development from credit related business, largely from accident year 2007, and to a lesser extent, accident years 2006 and 2005, driven by the recognition of lower than expected loss activity.

Long-Tail Business:

Long-tail lines of business describe lines of business for which specific losses may not be known for some period. Our long tail exposures include most of the liability business we write in our insurance and reinsurance segment as well as motor reinsurance business. There are many factors contributing to the uncertainty and volatility of long-tail business, including the following:

•

Our historical loss data and experience is generally too immature and lacking in actuarial credibility to place reliance upon for reserving purposes. Instead, we place reliance on industry loss ratios and industry benchmark development profiles that we anticipate reflect the nature and coverage of our business and its future development. Actual loss experience is likely to differ from industry loss statistics that are based on averages as well as loss experience of previous underwriting years;

•

The inherent uncertainty around loss trends, claims inflation (e.g., medical and judicial) and underlying general economic conditions; and

•

The possibility of future litigation, legislative or judicial change that might impact future loss experience relative to prior loss experience relied upon in loss reserve analyses.

For our liability lines of business, we predominately use the IELR method across all accident years. Due to the long-term reporting and settlement period for liability business, additional facts regarding coverage’s written in prior years, as well as actual claims and trends may become known and, as a result, we may be required to adjust our reserves accordingly. During 2009, we experienced net adverse prior period reserve development on E&S liability business, primarily impacting accident year 2007, and to a lesser extent, accident year 2008. We adjusted our loss development profile on these accident years having observed higher than expected frequency and severity of claims emergence on this business over the last twelve months. This was partially offset by net favorable development on E&S umbrella lines, predominately from accident years 2005 and 2006, reflecting the incorporation of more of our own actual experience with respect to reinsurance recoveries on this business.

Potential Volatility in Our Estimates

While we believe that our loss reserves at December 31, 2009 are adequate, new information, events or circumstances, unknown at the original valuation date, may lead to future developments in our ultimate losses significantly greater or less than the reserves currently provided. The actual final cost of settling claims outstanding at December 31, 2009 as well as claims expected to arise from unexpired period of risk is uncertain. As noted above there are many factors that may cause our reserves to increase or decrease, particularly those related to catastrophe losses and long-tail lines of business.

~~The uncertainty in our reserve estimate~~When considering the potential volatility in our estimated loss reserves, we considered the timing of the emergence of claims to be a significant factor at December 31, 2009. The uncertainty in the timing of the


Page		11

emergence of claims is generally greater for a company like ours that has a limited operating history and therefore relies to a certain extent upon industry benchmarks when establishing our loss reserve estimates. To reduce some of the uncertainty, management performs an analysis of additional factors to be considered when establishing our IBNR, intended to enhance our best estimate beyond quantitative techniques. At December 31, 2009, we recorded additional IBNR for uncertainties relating to the timing of the emergence of claims. Although time lags are incorporated within the actuarial methods discussed above, these rely on industry experience which may not be indicative of our business. For example, the low frequency, high severity nature of much of our business, together with the vast and diverse expanse of our worldwide exposures, may limit the usefulness of claims experience of other insurers and resinsurers for similar types of business.

The tables below quantify the impact on our gross loss reserves of reasonably likely scenarios to the key actuarial assumptions used to estimate our gross reserves at December 31, 2009. The changes to IELR selections represent percentage increases or decreases to our current IELR selections. The change to reporting patterns represents claims reporting that is both faster and slower than our current reporting patterns. The variability factors chosen below are consistent with commonly accepted actuarial practice given our own historical data regarding variability is generally limited. The results show the cumulative increase (decrease) in our reserves across all accident years. For example, if the timing of claims emerged three months faster with no change in IELR for Property, our loss reserves could decrease by approximately $18 million, as illustrated in the following table.

INSURANCE


Reporting Pattern	Initial Expected Loss Ratio
*Property*		5% lower			Unchanged			5% higher

3 months faster	$	(26,051	)	$	(17,643	)	$	(9,235	)
Unchanged		(1,916	)		—			16,985
3 months slower		38,930			50,373			61,816

*Marine*		5% lower			Unchanged			5% higher

3 months faster	$	(15,480	)	$	(10,598	)	$	(5,717	)
Unchanged		(1,623	)		—			9,429
3 months slower		17,913			24,297			30,681

*Aviation*		5% lower			Unchanged			5% higher

3 months faster	$	(9,171	)	$	(7,783	)	$	(6,396	)
Unchanged		(1,800	)		—			1,751
3 months slower		7,899			10,185			12,471

*Credit and Political Risk*		10% lower			Unchanged			10% higher

3 months faster	$	(45,216	)	$	—		$	45,216
Unchanged		(45,216	)		—			45,216
3 months slower		(45,216	)		—			45,216

*Professional Lines*		10% lower			Unchanged			10% higher

6 months faster	$	(212,192	)	$	(70,139	)	$	75,435
Unchanged		(139,972	)		—			159,922
6 months slower		(84,391	)		84,405			251,387

*Liability*		10% lower			Unchanged			10% higher

6 months faster	$	(108,759	)	$	(21,306	)	$	66,147
Unchanged		(87,802	)		—			91,669
6 months slower		(63,601	)		28,385			120,370


Page		12

REINSURANCE


Reporting Pattern	Initial Expected Loss Ratio
*Catastrophe, Property and Other*		5% lower			Unchanged			5% higher

3 months faster	$	(80,840	)	$	(41,684	)	$	(2,529	)
Unchanged		(39,990	)		—			39,990
3 months slower		8,684			49,843			91,001

*Credit and Bond*		10% lower			Unchanged			10% higher

6 months faster	$	(28,480	)	$	(5,722	)	$	17,035
Unchanged		(23,330	)		—			23,330
6 months slower		(17,882	)		6,053			29,988

*Professional Lines*		10% lower			Unchanged			10% higher

6 months faster	$	(103,902	)	$	(29,486	)	$	44,930
Unchanged		(80,111	)		—			80,111
6 months slower		(58,532	)		25,359			110,862

*Motor*		10% lower			Unchanged			10% higher

6 months faster	$	(37,475	)	$	(622	)	$	36,232
Unchanged		(36,916	)		—			36,916
6 months slower		(36,262	)		726			37,714

*Liability*		10% lower			Unchanged			10% higher

6 months faster	$	(77,832	)	$	(183	)	$	77,465
Unchanged		(77,667	)		—			77,667
6 months slower		(77,418	)		276			77,971

It is not appropriate to add together the total impact for a specific factor or the total impact for a specific reserving line as the lines of business are not perfectly correlated. It is important to note that the variations set forth in the tables above are not meant to be a “best-case” or “worst-case” series of scenarios, and therefore, it is possible that future variations may be more or less than the amounts set forth above. While we believe these are reasonably likely scenarios, we do not believe you should consider the above sensitivity analysis an actual reserve range.


Page		13

EXHIBIT B

5.	Investments (Continued)

d)	Net Realized Investment Gains (Losses)

The following table provides an analysis of net realized investment gains (losses):


Year ended December 31,	2009			2008			2007
Gross realized gains	$	181,075		$	107,170		$	31,683
Gross realized losses		(157,102	)		(120,474	)		(17,809	)
Net OTTI recognized in earnings		(337,435	)		(77,753	)		(8,562	)

Net realized losses on fixed maturities and equities		(313,462	)		(91,057	)		5,312

Change in fair value of investment derivatives ⁽¹⁾		(1,032	)		6,650			(82	)

Fair value hedges: ⁽¹⁾
Derivative instruments		(13,655	)		7,248			—
Hedged investments		16,565			(8,108	)		—

Net realized investment gains (losses)	$	(311,584	)	$	(85,267	)	$	5,230

(1)	Refer to Note 9 – Derivative Instruments

The following table summarizes the net OTTI recognized in earnings by asset class:


Year ended December 31,	2009		2008		2007
Fixed maturities:
Corporate debt	$	277,979	$	56,809	$	3,683
Residential MBS		24,594		4,956		509
Commerical MBS		10,843		323		1,024
ABS		2,384		8,615		3,346
Municipals		1,280		394		—

		317,080		71,097		8,562

Equities		20,355		6,656		—

Total OTTI recognized in earnings	$	337,435	$	77,753	$	8,562

Fixed maturities

As disclosed in Note 2 (a), on April 1, 2009, we adopted a new accounting standard which amended the previous OTTI recognition model for fixed maturities. Accordingly, for securities that we intended to sell at the end of each reporting period we recognized the entire unrealized loss in earnings. For the remaining impaired fixed maturities, from April 1, 2009, we have recorded only the estimated credit losses in earnings rather than the entire difference between the fair value and the amortized cost of fixed maturities. Because the new accounting standard does not allow for retrospective application, the OTTI amounts reported in the above table for the year ended December 31, 2009, are not measured on the same basis as prior period


Page		14

amounts and accordingly these amounts are not comparable. Furthermore, the net OTTI recognized in earnings for the first quarter of 2009 includes $26 million of OTTI charges calculated based on the former OTTI recognition model.

In 2009, we incurred OTTI charges of $317 million (2008: $71 million) relating to fixed maturities. These charges primarily relate to an impairment charge of $263 million recorded on our portfolio of MTNs within our corporate debt holdings as well as $20 million due to our intent to sell certain securities during the year ($3 million of corporate debt, $6 million of residential MBS and $11 million of commercial MBS). In 2008, our OTTI charges included $40 million of write-downs on a number of financial institutions, including $32 million related to Lehman Brothers, following its bankruptcy and sale in September, 2008. The balance of the OTTI charges on corporate debt was spread across the portfolio on securities that are unlikely to recover. The OTTI charges on our ABS and MBS portfolio in 2008 were primarily the result of the credit crisis and the higher risk of defaults on certain identified securities.

The following table provides a roll forward of the credit losses, (“credit loss table”), before income taxes, for which a portion of the OTTI was recognized in AOCI:


Year ended December 31, 2009
Beginning balance at April 1, 2009	$	—
Additions for:
Credit losses remaining in retained earnings related to adoption of accounting standard (see Note 2)		45,347
Credit loss impairment recognized on securities not previously impaired		267,770
Change in credit loss estimates on previously impaired securities		1,102
Passage of time on previously recorded credit losses		820
Reductions for:
Intent to sell securities previously impaired		(70,540	)
Securities sold/redeemed during the period (realized)		(82,109	)

Ending balance at December 31, 2009	$	162,390

The credit losses remaining in retained earnings upon adoption of the new accounting standard were $45 million, comprising $33 million of corporate debt securities, $7 million of ABS, $4 million of residential MBS and $1 million of commercial MBS. Corporate debts included $20 million of credit losses related to the bankruptcy of Lehman Brothers.

Credit losses are calculated based on the difference between the amortized cost of the security and the net present value of its projected future cash flows discounted at the effective interest rate implicit in the debt security prior to the impairment. The significant inputs and the methodology used to estimate the credit losses for which a portion of the OTTI was recognized in AOCI were as follows:

Corporate Debt:

Our projected cash flows for corporate debt securities, excluding medium-term notes (MTNs), are primarily driven by our assumptions regarding the probability of default and the timing and amount of recoveries associated with defaults. Our default and recovery ~~rate assumptions~~rates are based on credit rating, credit analysis, industry analyst reports and forecasts, Moody’s historical default data and any other data relevant to the recoverability of the security.~~Additionally, our projected cash flows for MTNs include significant inputs such as future credit spreads and use of leverage over the expected duration of each MTN.~~ At December 31, 2009, the weighted average default rate was 34 % for determining the credit losses on our impaired corporate debt securities. The weighted average recovery rate was nil for these securities. During 2009, we sold corporate debt securities that were previously impaired, resulting in a decrease in credit loss impairments of $21 million, in the above credit loss table.

For MTNs, our projected cash flows also include significant inputs such as future credit spreads and the use of leverage over the expected duration of each medium-term notes. At December 31, 2009, the range of our significant inputs was as follows:


Page		15


Default rates, per annum		3% - 5%

Recovery rates, per annum		30% - 55%

Collateral spreads , per annum		5.2% - 6.7%

Leveraged duration		6.5 - 8.5 years

During the third quarter of 2009, we recognized $263 million of credit losses on the MTNs ~~as there was considerable uncertainty regarding full recoverability~~. In response to the credit crisis, the MTNs managers reduced their leverage levels which in turn lowered the credit duration of the MTNs. As credit markets recovered and credit spreads tightened in 2009, price appreciation was not as pronounced as the depreciation during 2008 due to the lower credit duration of the MTNs. The tightening of credit spreads was more significant and much quicker than anticipated which has hindered the ability of the MTN managers to reinvest the underlying cash flows at wider credit spreads. Consequently, based on updated cash flow projections, we concluded that we will not fully recover par on these MTNs and therefore we have recorded credit losses on these securities.

~~During 2009, we sold corporate debt securities that were previously impaired, resulting in a decrease in credit loss impairments of $21 million, in the above table.~~ Further, in the fourth quarter of 2009, we made the decision to redeem certain MTNs, of which some redemptions occurred prior to and after December 31, 2009. Accordingly the related credit loss amounts are reported in the above credit loss table as ‘Securities sold/redeemed during the period (realized)’ and ‘Intent to sell securities previously impaired’, respectively.

Residential MBS ~~and Commercial MBS:~~:

For agency residential MBS in an unrealized loss position, we do not impair these securities as these represent AAA rated holdings backed by either the explicit or implicit guarantee of the U.S. government. We ~~utilized models~~believe the risk of loss in this asset class is very remote and linked to ~~determine~~ the ~~estimated~~ overall credit-worthiness of the U.S. government. At December 31, 2009, the fair value of our agency residential MBS was $1.6 billion which included $4 million of gross unrealized losses ~~for structured debt securities. To~~.

For non-agency residential MBS, we estimate discounted cash flows and to the extent it is less than the amortized cost of the non-agency residential MBS we record a credit loss. We project expected cash flows to be collected, ~~we utilized~~by utilizing underlying data from widely accepted third-party data sources ~~as well as~~and incorporate the following significant assumptions: expected defaults, delinquencies, recoveries, ~~foreclosure costs,~~ and prepayments. At December 31, 2009, the fair value of our non-agency residential MBS was $223 million, consisting primarily $136 million of Prime and $70 million of Alt-A MBS. The key assumptions made to estimate the credit loss for potentially impaired Prime and Alt-A MBS at December 31, 2009, were as follows:


	Vintage	Default Rate		Delinquency Rate		Loss Severity Rate		Prepayment Rate
Prime:	2003 and prior	1.1% - 1.2	%	2.3% - 5.1	%	10.5% - 15.3	%	19.1% - 26.5	%
	2004	2.4	%	5.4	%	26.5	%	16.7	%
	2005	1.7	%	4.0	%	25.4	%	14.5	%
	2006	13.6	%	30.6	%	44.9	%	8.1	%
	2007	4.2	%	10.5	%	40.5	%	11.7	%

Alt-A:	2003 and prior	1.4% - 2.4	%	2.8% - 8.4	%	38.9% - 40.5	%	11.3% - 12.1	%
	2004	5.0	%	13.7	%	31.3	%	12.4	%
	2005	4.6	%	13.5	%	33.3	%	5.8	%
	2006	20.9	%	51.6	%	49.7	%	11.3	%
	2007	23.3	%	55.7	%	54.9	%	10.7	%


Page		16

These assumptions require significant management judgment and vary for each structured security based on the underlying property type, vintage, loan to collateral value ratio, geographic concentration, and current level of subordination. ~~For~~We also corroborate our credit loss estimate with the independent investment manager’s credit loss estimate for each structured debt security with a significant unrealized loss position ~~we have also corroborated our principal loss estimate with the independent investment manager’s principal loss estimate~~.

During 2009, based on expected cash flows to be collected, we have recorded credit losses of $14 million on non-agency residential MBS.

Commercial MBS:

Our investment in commercial MBS are highly diversified and rated highly with approximately 79% rated AAA by S&P, with a weighted average estimated subordination percentage of 27% at December 31, 2009. Based on discounted cash flows, the current level of subordination is sufficient to cover the estimated loan losses on the underlying collateral of the commercial MBS.

ABS:

The majority of the unrealized losses on ABS at December 31, 2009 were related to CLO debt tranched securities. We utilized the same internal model as for CLO equity tranched securities (see Note 6 – Fair Value Measurements) to project estimated cash flows to be collected on the various CLO debt tranched securities. ~~The significant inputs used in the model include default and recovery rates and collateral spreads.~~ The weighted average significant inputs used to estimate future discounted cash flows for these securities for potential credit loss at December 31, 2009, were as follows:


	Per annum
Default rate	4.4	%

Recovery rate	50.0	%

Collateral spreads	3.1	%

Our assumptions on default and recovery rates are established based on an assessment of actual experience to date for each CLO debt tranche and review of recent credit rating agencies’ default and recovery forecasts. Based on projected cash flows at December 31, 2009, we do not anticipate credit losses on the CLO debt tranched securities.

Equities

In 2009, we recorded an OTTI charge of $20 million (2008: $7 million) on equities due to the severity and duration of their unrealized loss positions, for which we concluded the forecasted recovery period was uncertain. The recognition of such losses does not necessarily indicate that sales will occur or that sales are imminent or planned.


Page		17