SIDLEY AUSTIN LLP |BEIJING GENEVA SAN FRANCISCO
787 SEVENTH AVENUE |BRUSSELS HONG KONG SHANGHAI
SIDLEY AUSTIN LLP NEW YORK, NY 10019 |CHICAGO LONDON SINGAPORE
- ----------------| 212 839 5300 |DALLAS LOS ANGELES TOKYO
SIDLEY | 212 839 5599 FAX |FRANKFURT NEW YORK WASHINGTON, DC
|
|
|FOUNDED 1866
October 19, 2006
By Federal Express
Mr. Frank Wyman
Division of Corporation Finance
Securities and Exchange Commission
100 F Street, N.W.
Washington, D.C. 20549
Re: PXRE Group Ltd.
File No. 1-15259
Dear Mr. Wyman:
Our client, PXRE Group Ltd. ("PXRE" or the "Company"), has instructed us
to provide this letter to you to respond to your inquiry regarding how the
Company proposes to expand its future public disclosures in order to encompass
the additional information and disclosures contained in our letter to the
Commission dated August 31, 2006.
The Company intends to include additional disclosures in its Quarterly
Report on Form 10-Q for the nine months ended September 30, 2006 on the status
of its reinsurance business and certain of its critical accounting policies,
all as described to you in our letter dated August 31, 2006. These proposed
disclosures, attached as Exhibit A hereto, are substantially similar to the
information we provided to you in the letter dated August 31, 2006.
The Company intends to include additional disclosures in its Annual
Report on Form 10-K for the year ended December 31, 2006 concerning its
probabilistic risk limits and retrocessional reinsurance buying strategies.
These proposed disclosures, attached as Exhibit B hereto, are substantially
similar to the information we provided to you in the letter dated August 31,
2006. The Company also intends to include the disclosures outlined in Exhibit
A in its Annual Report on Form 10-K for the year ended December 31, 2006.
The registrant has authorized us to acknowledge on its behalf that:
o the registrant is responsible for the adequacy and accuracy of the
disclosure in the filings;
o staff comments or changes to disclosure in response to staff comments do
not foreclose the Commission from taking any action with respect to the
filings; and
SIDLEY AUSTIN LLP
- ----------------|
SIDLEY |
Mr. Frank Wyman
October 19, 2006
Page 2
o the registrant may not assert staff comments as a defense in any
proceeding initiated by the Commission or any person under the federal
securities laws of the United States.
Please contact Nancy Corbett at 212-839-5880 with any questions or
comments regarding this matter. Thank you for your time and attention.
Very truly yours,
/s/ Nancy Corbett
------------------
Nancy Corbett
SIDLEY AUSTIN LLP
- ----------------|
SIDLEY |
Mr. Frank Wyman
October 19, 2006
Page 3
EXHIBIT A
Status of Reinsurance Business
- ------------------------------
As of January 1, 2006, more than 75% of our business (by premium volume)
was subject to contractual provisions allowing clients additional rights upon
a decline in PXRE's ratings or capital. As of September 30, 2006,
approximately XX% of our in-force business as of January 1, 2006 has either
been cancelled or non-renewed and we anticipate that this percentage will
increase. In order to manage our peak zone catastrophe exposures, the Company
has also selectively allowed extra-contractual cancellations on certain
contracts that do not contain cancellation provisions triggered by rating
downgrades.
In the wake of the downgrades, we did not renew any of the reinsurance
contracts that expired on June 30, 2006, nor did we write any new reinsurance
contracts during the July 1, 2006 renewal season. The next major renewal date
in the property catastrophe reinsurance market is January 1, 2007. Unless we
are able to implement a strategic alternative that would allow us to provide
clients with an acceptably rated counterparty, we do not anticipate being able
to renew any of our remaining reinsurance contracts or being able to
underwrite any material amount of new reinsurance business.
The Company had $XX.X million in net premiums earned during the quarter
ended September 30, 2006, as compared to $68.8 million in the comparable prior
year period. The Company had $XX.X million in net premiums earned during the
nine months ended September 30, 2006, as compared to $231.7 million in the
comparable prior year period. Given the rate of cancellations and our impaired
ability to renew our existing reinsurance contracts and underwrite new
reinsurance contracts, we expect significant decreases in net premiums earned
in future quarters as compared to prior quarters in 2006 as well as to the
same quarters in the prior year. The reductions in premium income will likely
have a material adverse effect on our future operating results, liquidity and
financial condition. Net premiums earned were our primary source of revenue in
2005, accounting for 93% of our revenue. In 2005, revenue from non-premium
sources was not sufficient to offset operating expenses and interest expenses.
We therefore expect to incur net operating losses in future periods unless we
are successful in executing a strategic alternative other than runoff. If such
operating losses were to occur, this would result in a decline in our
shareholders' equity. However, we currently expect that our portfolio of short
duration high credit quality fixed income securities, totaling $X.X billion as
of September 30, 2006, will provide sufficient liquidity to meet the currently
foreseen needs of our counterparties.
SIDLEY AUSTIN LLP
- ----------------|
SIDLEY |
Mr. Frank Wyman
October 19, 2006
Page 4
Critical Accounting Policy Disclosures
- --------------------------------------
Estimation of Losses and Loss Expenses
As a property catastrophe reinsurer, incurred losses are inherently more
volatile than those of primary insurers and reinsurers of risks that have an
established historical pattern of losses. In addition, with respect to insured
events that occur near the end of a reporting period, as well as with respect
to our retrocessional book of business, the significant delay in losses being
reported to insurance carriers, reinsurers and finally retrocessionaires
require us to make estimates of losses based on limited information from our
clients, industry loss estimates and our own underwriting data. Because of the
uncertainty in the process of estimating our losses from insured events, there
is a risk that our liabilities for losses and loss expenses could prove to be
inadequate, with a consequent adverse impact on our earnings and shareholders'
equity in future periods.
In establishing our loss and loss expense liabilities, PXRE records
reserves as management's best estimate of liabilities as of the balance sheet
date. Management believes that the Company's liability for loss and loss
expenses as of September 30, 2006 is adequate. Since year-end 2004, all of our
loss and loss expense estimates have been reviewed annually by an independent
internationally recognized actuarial firm, and their conclusions have not, in
any period, been materially different than those of the Company.
Significant Uncertainties
The most significant uncertainty in our reserves involves our estimates
of catastrophe losses. In reserving for catastrophe losses, our estimates are
influenced by underwriting information provided by our clients, industry
catastrophe models, industry loss estimates and our internal analyses of this
information. This reserving approach can cause significant development from
initial loss estimates in the immediate wake of a catastrophe event due to the
limited information available to us as a reinsurer and retrocessionaire
regarding the actual underlying losses. This process can cause our ultimate
estimates to differ significantly from initial projections. The French storm
Martin that occurred on December 27, 1999 presents an example of the
uncertainty and potential variability associated with reserving for
catastrophe losses. Initially, we based our reserves to a significant degree
on industry estimates of the total loss, which were approximately $1.0 billion
at the time of the loss. In 2001, the final cost was estimated to be $2.5
billion by SIGMA, a widely used industry publication. Our gross loss estimate
at December 31, 1999 for this event was $31.3 million. Our gross loss estimate
for French storm Martin at December 31, 2005 was $68.9 million. The original
industry loss estimate increased by 150%, while our loss estimate increased by
120%.
SIDLEY AUSTIN LLP
- ----------------|
SIDLEY |
Mr. Frank Wyman
October 19, 2006
Page 5
Historically, there has been significant variability in the development
of catastrophe losses during the twelve month period immediately following the
catastrophe event with such variability reducing after the initial twelve
month period. To further illustrate the variability of setting catastrophe
loss estimates, the following chart outlines the changes in the initial loss
estimates for recent costly catastrophes in the United States over a
three-year period based on industry insured loss:
- -----------------------------------------------------------------------------------------------------------------------------------
Recent U.S. Catastrophes
- -----------------------------------------------------------------------------------------------------------------------------------
Subsequent Development After Initial Estimates
- -----------------------------------------------------------------------------------------------------------------------------------
Percent Change from Initial Percent Change from Initial
- -----------------------------------------------------------------------------------------------------------------------------------
PCS Industry Loss Estimates ** PXRE Loss Estimates **
- -----------------------------------------------------------------------------------------------------------------------------------
Date Event @ 6 mos. @ 12 mos. @ 2 yrs @ Final @ 6 mos. @ 12 mos. @ 2 yrs @ 3 yrs @ Final
- -----------------------------------------------------------------------------------------------------------------------------------
Aug-92 Hurricane Andrew * 37% 99% 99% 99% 0% 46% 68% 59% 69%
- -----------------------------------------------------------------------------------------------------------------------------------
Jan-94 Northridge Earthquake 22% 131% 178% 178% 49% 100% 145% 151% 153%
- -----------------------------------------------------------------------------------------------------------------------------------
Sep-98 Hurricane George 16% 16% 16% 16% 8% 63% 41% 28% 39%
- -----------------------------------------------------------------------------------------------------------------------------------
Sep-01 September 11th Events 0% 23% 23% 13% 1% 2% 2% 4% 6%
- -----------------------------------------------------------------------------------------------------------------------------------
Aug-04 Florida Hurricanes 6% 11% 11% 11% 1% 28% 32% NA 32%
- -----------------------------------------------------------------------------------------------------------------------------------
* Hurricane Andrew's initial PXRE loss estimate is at September 30, 1992. We have extrapolated this loss estimate based on the
best available information.
- ------------------------------------------------------------------------------------------------------------------------------------
** Initial estimates are as of quarter-end immediately following an event. PCS stands for Property Claims Services, a division
of the Insurance Services Office (ISO).
- ------------------------------------------------------------------------------------------------------------------------------------
Reserving Methodologies
We establish loss and loss expense liabilities (to cover expenses
related to settling claims, including legal and other fees) to provide for the
ultimate cost of settlement and administration of claims for losses, including
claims that have been reported to us by our reinsureds and claims for losses
that have occurred but have not yet been reported to us.
For reported losses, we establish reserves (liabilities for formally
reported claims) when we receive notice of the claim, which we refer to as
case reserves. It is our general policy to establish liabilities for reported
losses in an amount equal to the liability set by the reinsured. In certain
but infrequent instances, such as the receipt of inconsistent, incorrect or
inadequate supporting documentation, we will conduct an investigation to
determine if the amount established by the reinsured is appropriate or if it
should be adjusted in the form of an additional case reserve.
SIDLEY AUSTIN LLP
- ----------------|
SIDLEY |
Mr. Frank Wyman
October 19, 2006
Page 6
PXRE also records reserves for losses that have been incurred but not
yet reported, which are referred to as IBNR reserves. For IBNR reserves, a
variety of methods have been developed in the insurance industry and are
generally accepted for use in determining the appropriate provision for such
liabilities. In general, these methods involve the extrapolation of reported
loss data to estimate ultimate losses. Our loss calculation methods generally
rely upon a projection of ultimate losses based upon the historical patterns
of reported loss development.
Our methods for establishing loss and loss expense liabilities vary
depending upon the nature of the losses, which can generally be divided into
three categories: (1) non-catastrophe losses in our Catastrophe and Risk
Excess segment, (2) catastrophe losses in our Catastrophe and Risk Excess
segment, and (3) losses in our Exited Lines Segment. The Company generally
considers reserves for each of these categories to be "mature" as follows (1)
for non-catastrophe losses in our Catastrophe and Risk Excess Segment,
generally anywhere between eighteen and thirty-six months after the accident
year, (2) for catastrophe losses in our Catastrophe and Risk Excess Segment,
generally anywhere between six to twelve months after the event occurs, and
(3) losses in our Exited Lines Segment, depending upon the nature of the
underlying business, generally anywhere from five to seven years after the
accident year.
(1) Non Catastrophe Losses
In reserving for non-catastrophe losses in our Catastrophe and Risk
Excess Segment, we use three different methods to estimate IBNR reserves: the
loss ratio method; the incurred Bornhuetter-Ferguson method, which we refer to
as the BF method; and the incurred loss development method, the latter two of
which are "loss development approaches." Initially, at the inception of an
accident year, when there is little reported loss information, we use the loss
ratio method. This method computes IBNR, as the product of an expected loss
ratio and assumed earned premium minus reported loss to date. The expected
loss ratio is primarily established for each line of business based on the
Company's historical loss ratios. In determining the expected loss ratio, we
also consider information provided by our clients and estimates provided by
our underwriters and actuaries concerning the impact of pricing and coverage
changes. As the accident year matures further and reported loss activity
begins to emerge (usually at the end of the second or third quarter following
the loss) we change to the BF method.
The BF method is a premium based method of computing IBNR which blends
the loss ratio method with the loss development method. The BF method computes
IBNR for an accident year as the product of expected loss (earned premium
multiplied by an expected loss ratio) and an expected percentage of unreported
losses. This expected percentage of unreported loss is a function of the loss
development factors developed by our corporate actuaries from our annual
analysis of the Company's historical loss development patterns by line of
business. As experience emerges and accident years mature further (generally
eighteen to thirty-six months after the accident year given the short-tailed
nature of these property reinsurance lines) we
SIDLEY AUSTIN LLP
- ----------------|
SIDLEY |
Mr. Frank Wyman
October 19, 2006
Page 7
transition to the incurred loss development method, which relies solely on our
clients' reported loss information and indicated historical loss development
patterns to estimate ultimate losses.
Other alternate loss reserving methods developed in the insurance
industry include the paid loss development method and the paid
Bornhuetter-Ferguson method. We do not use these methods to establish our IBNR
reserves. We believe that incurred loss methods are more reliable because they
rely on a much larger, more stable and credible source of information, which
is not influenced by the occurrence of one or more large payments. We do,
however, use these alternate loss reserving methods to evaluate the
reasonableness of the IBNR reserves that we establish using the incurred loss
methods.
Our loss development factors and expected loss ratios are monitored
regularly and updated as appropriate but at least once a year. These are the
key assumptions that materially affect our estimates for reserves for losses
and loss expenses. The expected loss ratio is primarily established for each
line of business based on the Company's historical loss ratios. In determining
the expected loss ratio, we also consider information provided by our clients
and estimates provided by our underwriters and actuaries concerning the impact
of pricing and coverage changes. Loss development factors are developed from
our annual analysis of our Company's historical loss development patterns by
line of business. These key assumptions did not materially change from
December 31, 2004 to December 31, 2005.
(2) Catastrophe Losses
In reserving for catastrophe losses in our Catastrophe and Risk Excess
Segment, there is initially little reported loss information in the immediate
wake of a catastrophe event, as was the case with the 2005 Hurricanes,
Katrina, Rita and Wilma ("KRW"). The loss estimation process begins with the
identification of events with characteristics similar to the recent
catastrophe (geographic location, windspeed, damageability etc.), which then
results in a list of the expected losses by contract from our proprietary risk
management system. Third party modeling software is embedded in our
proprietary risk management system.
Concurrently, our underwriting team performs a thorough contract by
contract analysis to identify potential changes to the expected loss estimates
including IBNR by contract. With respect to the 2005 Hurricanes, our
underwriters' estimates were subject to a high level of uncertainty.
Specifically for Hurricane Katrina, this high level of uncertainty arose out
of extremely complex and unique causation and coverage issues, including the
appropriate attribution of losses to wind or flood damage as opposed to other
perils such as fire, business interruption or civil commotion. In order to
address these uncertainties, for contracts exposed to Hurricane Katrina losses
in our most volatile lines of business (retrocessional and direct and
facultative reinsurance) we posted reserves for XX% of occurrence limits as of
December 31, 2005. These lines of business represented approximately 60% of
our gross loss amount on Hurricane Katrina. The underwriters' knowledge of the
clients' underlying books of business is
SIDLEY AUSTIN LLP
- ----------------|
SIDLEY |
Mr. Frank Wyman
October 19, 2006
Page 8
considered in establishing loss estimates by contract. The combination of
catastrophe modeling and underwriting review of affected contracts then forms
the basis of our initial loss estimates for catastrophe losses.
In light of the limited loss data available from clients in the wake of
catastrophes, our actuaries use our most current internally generated
catastrophe loss development factors to assess the reasonableness and adequacy
of our catastrophe loss reserves in the immediate wake of a catastrophe. These
significant catastrophe loss development factors reflect the historical
variability of prior catastrophe loss reserves in the industry and PXRE's
specific experience.
The results of this initial process are updated when additional
information is available. This information comes in the form of publicly
available announcements, informal contact with brokers and/or clients,
submission data and formal claim notices. As catastrophic events mature and
reporting loss methods become more credible (usually six to twelve months
after the event) actuarial methods implementing historical patterns can be
assigned more credibility. In evaluating the loss estimates for catastrophic
events, our actuaries utilize our internal database to establish projected
reporting patterns and payment patterns. This database includes data dating
back to the 1980's consisting of well over one hundred catastrophic events, of
which over twenty are hurricanes. Our actuaries also employ industry patterns
from the Reinsurance Association of America, an insurance industry
organization. Using this information, we have developed loss development
factors for significant catastrophes. Our internal significant catastrophe
loss development factors are analyzed as appropriate and at least once per
year to reflect updated industry benchmarks and changes in PXRE's specific
loss history. The volume of reported loss activity in interim quarterly
periods is monitored by the Company to determine consistency with expected
loss activity based on PXRE and industry historical patterns. For individual
storms, PXRE specific loss development factors are applied to reported losses
to estimate IBNR reserves. However these may be weighted with industry factors
or judgmental adjustments for individual significant catastrophes based upon
the nature of the particular catastrophe and the underwriters' knowledge.
This catastrophe reserving process can cause our ultimate estimates to
differ significantly from initial projections. For example, as part of our
year-end closing process for the year ended December 31, 2005, we reassessed
our ultimate liability for losses and loss expenses arising from Hurricanes
Katrina, Rita and Wilma. During the course of our 2005 year-end assessment, we
increased our estimate of the ultimate incurred gross losses and loss expenses
arising from Hurricane Katrina by $214.6 million to $_____ million and from
Hurricane Rita by $48.1 million to $_____ million, in each case as compared to
the gross incurred losses recorded as of September 30, 2005. Our initial loss
estimates for each of the hurricanes were based, in part, on insured industry
loss estimates for each event, catastrophe modeling, preliminary discussions
with clients and a review of potentially exposed contracts by our
underwriters. In our year-end assessment of the liability for the 2005
hurricane losses, we determined that claims reported by clients relating to
Hurricanes Katrina and Rita were significantly higher than expected,
especially
SIDLEY AUSTIN LLP
- ----------------|
SIDLEY |
Mr. Frank Wyman
October 19, 2006
Page 9
following a significant influx of reported claims from late November 2005 to
February 2006. In part, the additional claims arose from a reassessment by
clients of their original loss estimates for the hurricane events. For
example, various clients, who advised our underwriters in the immediate wake
of the hurricanes that they did not expect to experience significant losses to
the reinsurance contracts in the upper layers of their reinsurance programs,
reassessed their losses and submitted notices of claim for the contracts that
they had previously indicated would not be impacted by the catastrophes.
In addition, in reviewing underwriting information provided by clients
during December 2005 as part of the January 1, 2006 renewal process, we found
that certain clients were anticipating higher losses from Hurricanes Katrina
and Rita than had been reported through the formal claims channels.
As of September 30, 2006, our estimate of ultimate incurred gross losses
and loss expenses arising from Hurricanes Katrina, Rita and Wilma is $X,XXX.X
million. As of September 30, 2006, we have paid less than XX% of our net
incurred loss amounts with respect to Hurricanes Katrina, Rita and Wilma.
Accordingly, our estimate of the ultimate liability arising from these
catastrophes is based on preliminary claims notices received from clients,
catastrophe modeling, a review of exposed reinsurance contracts, discussions
with numerous clients and a review of the underwriting information provided by
clients with reinsurance contracts that renewed as of January 1, 2006. As
these events mature, historical reporting patterns, both industry and company
specific, are used to project ultimate net loss. Our estimates fall within a
reasonable actuarial range produced by these methods.
Specifically for Hurricane Katrina, our estimates are subject to a high
level of uncertainty arising out of extremely complex and unique causation and
coverage issues, including the appropriate attribution of losses to wind or
flood damage as opposed to other perils such as fire, business interruption or
civil commotion. The underlying personal lines policies generally contain
exclusions for flood damage; however, water damage caused by wind may be
covered. We expect that causation and coverage issues may not be resolved for
a considerable period of time and may be influenced by evolving legal and
regulatory developments.
Our actual losses from Hurricanes Katrina, Rita and Wilma may exceed our
best estimate as a result of, among other things, the receipt of additional
information from clients, the attribution of losses to coverages that for the
purpose of our estimates we assumed would not be exposed, and inflation in
repair costs due to the limited availability of labor and materials, in which
case our financial results could be further materially adversely affected.
In developing our best estimate for Hurricane Katrina, we have also
assumed flood damage exclusions contained in our cedent's underlying insurance
policies will be effective. We understand that various lawsuits have been
filed seeking to invalidate such flood damage
SIDLEY AUSTIN LLP
- ----------------|
SIDLEY |
Mr. Frank Wyman
October 19, 2006
Page 10
exclusions on various grounds, including a suit filed by the Attorney General
for the State of Mississippi. If such lawsuits were to successfully invalidate
the underlying flood damage exclusions, our liabilities for losses and loss
expenses relating to Hurricane Katrina could prove to be inadequate, with a
consequent adverse impact on our earnings and shareholders' equity in future
periods.
(3) Exited Lines Segment Losses
In reserving for losses on our Exited Lines Segment, there is great
uncertainty due to its composition of casualty and finite businesses, which
produce losses that are slower to be reported and paid than the property
losses of our Catastrophe and Risk Excess Segment. Moreover, given our limited
experience in the casualty and finite businesses, we do not have established
historical loss development patterns that can be used to estimate these loss
liabilities. We must therefore rely on the historical loss development
patterns reported by our clients and industry loss development data in
estimating our liabilities including IBNR. Accident years for this segment
take a longer period of time to "mature" than for our Catastrophe and Risk
Excess Segment, therefore, we primarily utilize the loss ratio method and BF
method to establish our initial Exited Lines loss estimates including IBNR.
When the amount of reported losses become a more reliable means for setting
reserve estimates, generally five to seven years after the relevant accident
year, PXRE places more weight on these reported losses to estimate its loss
reserves including IBNR and less weight on the BF method.
(4) Interim reporting
With respect to the methods outlined above, estimation of IBNR in
interim quarterly periods follows the same process as that done at annual
reporting periods. The incurred loss development factors that are developed at
year-end are interpolated to generate factors applicable for quarterly
evaluation periods. Generally, the expected loss ratios which were selected at
year-end are used in interim periods.
Additional Uncertainties
PXRE has historically been involved in very few disputes with ceding
companies, especially those that enter into contracts that the Company
includes in its catastrophe and risk excess segment; nevertheless contract
disputes in the property casualty reinsurance industry have increased in
recent years.
There is an additional risk of uncertainty in PXRE's estimation of loss
due to the fact that PXRE writes only reinsurance business and no insurance
business. As a result, losses, unearned premiums and premiums written are all
recorded based on reports received from the ceding companies. PXRE does not
receive loss information from the underlying insureds; however, because the
Company's reinsurance business focuses on short-tail lines such as property
SIDLEY AUSTIN LLP
- ----------------|
SIDLEY |
Mr. Frank Wyman
October 19, 2006
Page 11
catastrophe, retrocessional property catastrophe, risk-excess and aerospace,
the delay from the time of the underlying loss to the report date to PXRE is
not as significant a risk as it would be if the Company underwrote a
significant amount of casualty business. However, with respect to insured
events that occur near the end of a reporting period, as well as with respect
to our retrocessional book of business, a delay in losses being reported to
insurance carriers, reinsurers and finally retrocessionaires may require us to
make estimates of losses based on limited information from our clients,
industry loss estimates and our own underwriting data.
We have a system of controls in place that assists us in evaluating the
completeness of the data received from cedents. PXRE derives almost all of its
business from reinsurance intermediaries. As a result, the ceding company
reports claims to the intermediary and the intermediary in turn reports the
data to all the reinsurers included in the underlying program. Controls in
place require that certain claims must be approved by the underwriter or a
member of senior management to validate the reported loss data before recorded
as a case reserve. The underwriter, based on his knowledge and judgment, may
question the broker or ceding company if he did not expect a loss of a certain
magnitude to impact a certain layer. Because many of PXRE's losses are from
events that are well known, such as large hurricanes and earthquakes, the
underwriter may in fact expect losses to pierce certain layers and therefore
would not question the accuracy of such loss reports. If the underwriter does
question the loss data, PXRE may perform audits at the underlying ceding
company in order to determine the accuracy of the amounts ceded. PXRE's risk
management and underwriting systems provide a list of impacted or potentially
impacted contracts by peril and by geographic zone. This assists PXRE in
determining the completeness of losses, as we will contact intermediaries and
the ceding companies for which we believe underlying contracts are impacted
subsequent to an event to request information.
Currently, PXRE does not have any backlog related to the processing of
assumed reinsurance information. When a large loss occurs, the Company shifts
personnel from various functions to assist the claims personnel in the
processing and evaluation of claims data.
Status of Loss Reserves of September 30, 2006
The following charts set forth the allocation of our net liabilities for
loss and loss expenses between case reserves and IBNR reserves:
Net loss and loss expense liabilities as of September 30, 2006 were as
follows:
SIDLEY AUSTIN LLP
- ----------------|
SIDLEY |
Mr. Frank Wyman
October 19, 2006
Page 12
- ----------------------------------- --------------------------- --------------------------- --------------------------
(000's) 2005 and Prior Accident 2006 Accident Year Total
Years
- ----------------------------------- --------------------------- --------------------------- --------------------------
Catastrophe & Risk Excess Segment
- ----------------------------------- --------------------------- --------------------------- --------------------------
Case
- ----------------------------------- --------------------------- --------------------------- --------------------------
IBNR
- ----------------------------------- --------------------------- --------------------------- --------------------------
Total
- ----------------------------------- --------------------------- --------------------------- --------------------------
Exited Lines Segment
- ----------------------------------- --------------------------- --------------------------- --------------------------
Case
- ----------------------------------- --------------------------- --------------------------- --------------------------
IBNR
- ----------------------------------- --------------------------- --------------------------- --------------------------
Total
- ----------------------------------- --------------------------- --------------------------- --------------------------
All Segments
- ----------------------------------- --------------------------- --------------------------- --------------------------
Case
- ----------------------------------- --------------------------- --------------------------- --------------------------
IBNR
- ----------------------------------- --------------------------- --------------------------- --------------------------
Total
- ----------------------------------- --------------------------- --------------------------- --------------------------
On an overall basis, the low and high ends of our selected range of
reasonable net loss reserves are $[XXX] million below and $[XXX] million above
the $[XXX] million best estimate. The range around the overall estimate is not
the sum of the ranges about the component segments due to the impact of
diversification when the reserve levels are considered in total. The impact of
diversification reflects the fact that not all lines with very low
correlations are expected to develop well or poorly at the same time.
The low and high ends of a range of reasonable net loss and loss expense
liabilities around the best estimate displayed in the table below with respect
to each segment are as follows as of September 30, 2006:
- --------------------------------------------------------------------------------
Net Loss and Loss Expense Liabilities Range at September 30, 2006
- --------------------------------------------------------------------------------
(000's) Low End Best Estimate High End
- --------------------------------------------------------------------------------
Catastrophe and Risk Excess
- --------------------------------------------------------------------------------
Exited Lines
- --------------------------------------------------------------------------------
This range was produced using two different methods, one for the 2005
Hurricanes (KRW) and one for the remainder of our carried loss and loss
expense liabilities.
SIDLEY AUSTIN LLP
- ----------------|
SIDLEY |
Mr. Frank Wyman
October 19, 2006
Page 13
- --------------------------------------------------------------------------------
Net Loss and Loss Expense Liabilities Range at September 30, 2006
- --------------------------------------------------------------------------------
Business Segment Low Best Estimate High
- --------------------------------------------------------------------------------
(000's)
- --------------------------------------------------------------------------------
Catastrophe and Risk Excess: KRW only
- --------------------------------------------------------------------------------
Catastrophe and Risk Excess excl. KRW
- --------------------------------------------------------------------------------
Catastrophe and Risk Excess
- --------------------------------------------------------------------------------
Exited Lines
- --------------------------------------------------------------------------------
Impact of Diversification
- --------------------------------------------------------------------------------
Total
- --------------------------------------------------------------------------------
For the 2005 Hurricanes, the range was calculated by varying key
assumptions regarding PXRE and industry loss reporting patterns. Specifically,
we analyzed the effect on our loss estimates resulting from a one month
acceleration and one month deceleration around the expected historical
reporting patterns. For hurricane catastrophes, the volume of reported loss in
the first year after the occurrence of an event is significant, with greater
than ninety percent of losses reported; thus; assuming faster and/or slower
reporting patterns can produce a wide range of results. A selected range of
$XXX million below and $XXX million above our net carried reserves was
determined in the aggregate for all three events combined, after considering
the sensitivity analysis outlined above, our best loss estimates, and
actuarial judgment. This range of reasonable estimates does not represent a
range of all possible outcomes, but it is intended to reflect the inherent
variability in actuarial reserving methodologies. We believe that results
outside this range are unlikely, and can not assign probabilities with
certainty to any points within this range.
For the remainder of our carried loss and loss expense liabilities, we
used a statistical model to simulate a range of results about our best
estimates assuming probability distributions believed to reasonably reflect
potential for favorable and unfavorable loss development. The simulation
results indicate that a five percent probability exists that the net loss
reserves will be below the Low estimate and a five percent probability exists
that the net loss reserves will be above the High estimate. However, no
assurance can be given that our ultimate losses will not be significantly
different than the simulation.
The low end and high end of the range of reasonable net loss reserves
around the best estimate have decreased from both December 31, 2005 and June
30, 2006 principally because of the reduction in the Company's net loss
reserve amount due to the payment of claims during nine months ended September
30, 2006, as well as the passage of time since the occurrences of Katrina,
Rita and Wilma and the resulting decrease in actuarial variability surrounding
the best estimate.
SIDLEY AUSTIN LLP
- ----------------|
SIDLEY |
Mr. Frank Wyman
October 19, 2006
Page 14
Estimation and Recognition of Premiums
Our premiums on reinsurance business assumed are recorded as earned
evenly over the contract period based upon estimated subject premiums. PXRE's
assumed premium is comprised of both minimum and deposit premium and an
estimate of premium. Minimum and deposit premium is billed and collected in
accordance with the provisions of the contracts and is usually billed
quarterly or semi-annually. A premium estimate is also recorded if the
estimate of the ultimate premium is greater than the minimum and deposit
premium. The final or ultimate premium for most contracts is the product of
the provisional rate and the ceding company's subject net earned premium
income (SNEPI). Because this portion of the premium is reasonably estimable,
the Company records and recognizes it as revenue over the period of the
contract in the same manner as the minimum and deposit premium. The key
assumption related to the premium estimate is the estimate of the amount of
the ceding company's SNEPI, which is a significant element of PXRE's overall
underwriting process. Because of the inherent uncertainty in this process,
there is the risk that premiums and related receivable balances may turn out
to be higher or lower than reported.
The estimated premium receivable included in the total of premiums
receivable for the nine months ended September 30, 2006 of $_____ million is
$121.5 million, including reinstatement premiums of $105.8 million.
We record an allowance for doubtful accounts that we believe
approximates the exposure for all potential uncollectible assets.
The premiums on reinsurance business ceded are recorded as incurred
evenly over the contract period. Certain ceded reinsurance contracts contain
provisions requiring us to pay additional premiums or reinstatement premiums
in the event that losses of a significant magnitude are ceded under such
contracts. Under GAAP, we are not permitted to establish reserves for these
potential additional premiums until a loss occurs that would trigger the
obligation to pay such additional or reinstatement premiums. As a result, the
net amount recoverable from our reinsurers in the event of a loss may be
reduced by the payment of additional premiums and reinstatement premiums.
Frequently, the impact of such premiums will be offset by additional premiums
and reinstatement premiums payable to us by our clients on our assumed
reinsurance business. No assurance can be given however, that assumed
reinstatement and additional premiums will offset ceded reinstatement and
additional premiums. For example, in the case of the September 11, 2001
terrorist attacks, our net premiums earned during 2001 were reduced by $26.3
million as a result of net additional premiums and reinstatement premiums. In
the case of Hurricanes Katrina, Rita and Wilma our net premiums earned were
increased by $43.9 million in 2005 as a result of net additional premiums and
reinstatement premiums.
SIDLEY AUSTIN LLP
- ----------------|
SIDLEY |
Mr. Frank Wyman
October 19, 2006
Page 15
Assumed reinstatement premiums that reinstate coverage are written and
earned at the time the associated loss event occurs. Under the contract terms
of certain of our excess of loss contracts, reinstatement premiums are
mandatorily due PXRE following a ceding company's loss, based on pre-defined
contract terms. Terms principally include a pro-rata amount of the original
contract premium relative to the proportion of the contractual limit exhausted
by the associated loss, without respect to time remaining in the term of the
original limit, that is, the amount due is 100% as to time and pro-rata as to
amount. Less frequently, terms can vary to incorporate a percentage of the
original premium that is more or less than the original premium or can be
pro-rata as to time remaining in the term of the original limit.
Ceded additional premiums are less uniform in their terms than assumed
reinstatement premiums. The single largest additional premium in 2005,
accounting for 80% of the additional premium written and earned for 2005
events, was triggered by an event in excess of $165 million of losses
incurred.
Assumed reinstatement and additional premiums written and earned in the
nine months ended September 30, 2006 and September 30, 2005, as reflected in
the Consolidated Statements of Operations and Comprehensive Operations, were
as follows:
- ------------------------------------------------ -------------------------------
Nine Months Ended September 30,
- ------------------------------------------------ ---------------- --------------
($ in millions) 2006 2005
- ------------------------------------------------ ---------------- --------------
Assumed reinstatement premiums:
- ------------------------------------------------ ---------------- --------------
Reported paid losses and case reserves
- ------------------------------------------------ ---------------- --------------
IBNR
- ------------------------------------------------ ---------------- --------------
Total
- ------------------------------------------------ ---------------- --------------
Ceded additional premiums:
- ------------------------------------------------ ---------------- --------------
Reported paid losses and case reserves
- ------------------------------------------------ ---------------- --------------
IBNR
- ------------------------------------------------ ---------------- --------------
Total
- ------------------------------------------------ ---------------- --------------
Net reinstatement and additional
- ------------------------------------------------ ---------------- --------------
premiums
- ------------------------------------------------ ---------------- --------------
SIDLEY AUSTIN LLP
- ----------------|
SIDLEY |
Mr. Frank Wyman
October 19, 2006
Page 16
Assumed reinstatement premiums receivable associated with our case
reserves and incurred but not reported loss and loss expense liabilities, as
reflected in the Consolidated Balance Sheets, are as follows:
- -------------------------------------------- --------------------------------- ------------------------------
September 30, 2006 September 30, 2005
- -------------------------------------------- ---------------- ---------------- --------------- --------------
($ in millions) $ % $ %
- -------------------------------------------- ---------------- ---------------- --------------- --------------
Assumed reinstatement premiums receivable:
- -------------------------------------------- ---------------- ---------------- --------------- --------------
Case reserves
- -------------------------------------------- ---------------- ---------------- --------------- --------------
IBNR reserves
- -------------------------------------------- ---------------- ---------------- --------------- --------------
Total
- -------------------------------------------- ---------------- ---------------- --------------- --------------
We accrue assumed reinstatement premiums based upon contract terms
applied to the amount estimated to settle ultimate incurred losses. The
primary factor that could affect our estimate of assumed reinstatement
premiums is management's best estimate of ultimate incurred losses. See the
Update on Critical Accounting policies relating to the estimation of loss and
loss expenses with regards to the uncertainty, historical accuracy and
sensitivity of this estimate. While premiums stated relative to limit ("rates
on line") have an effect on the estimate of assumed reinstatement premiums,
those associated with case reserves are based on actual contract rate on line
terms, and those estimates associated with IBNR are based on weighted average
rate on line terms, of the book of business for a given underwriting year by
line of business. Therefore the primary factor that could change our estimate
of assumed reinstatement premium is management's best estimate of ultimate
incurred losses and the mix of treaties along with their respective rate on
lines that ultimately incur losses. Assumed reinstatement premiums receivable
are settled on a net basis when loss payments are made to cedents.
Accordingly, there is an insignificant amount of credit risk associated with
this asset as of any given period end date.
Additional ceded premiums occur less frequently, and were incurred by
the Company in 2005 due to significance of the losses from Hurricanes Katrina
and Wilma. In 2005, these ceded losses were full limit losses, resulting in
additional ceded premiums. These additional ceded premiums are not expected to
vary from the estimate in 2005.
SIDLEY AUSTIN LLP
- ----------------|
SIDLEY |
Mr. Frank Wyman
October 19, 2006
Page 17
EXHIBIT B
Retrocessional Reinsurance Protection
In 2006, the Company purchased $587.1 million of single event
reinsurance and catastrophe bond protection. This compares to $245.5 million
purchased in 2005. Additionally, the amount of coverage available in the event
of a second event in 2006 was $275.0 million, compared to the 2005 amount of
$25.0 million. This increased limit purchased is expected to increase ceded
premiums over prior periods. It should be noted, however, that our risk
management program for our 2006 assumed reinsurance portfolio assumed that we
would have significant gross premium written during 2006. It was assumed that
such premium income would offset losses arising from small or medium
catastrophe events. For example, during 2004, we incurred material losses from
Hurricanes Charley, Frances, Ivan and Jeanne, but still experienced a loss
ratio of 65.9% in our catastrophe and risk excess segment due to the
significant volume of premium written during 2004. If many or all of our
clients exercise their special cancellation rights, we will have significantly
less premium income to offset smaller catastrophe losses.
In purchasing reinsurance coverage to protect PXRE during 2006, as
noted, we assumed that such gross premiums written would be available to
offset smaller losses and; as a result most of the reinsurance coverage
purchased by PXRE is designed to protect PXRE against larger catastrophe
events. Therefore, although our aggregate exposures may be reduced due to
widespread cancellations of our assumed reinsurance contracts, we may be
overexposed to the risk of losses arising from smaller catastrophe events
since our reinsurance protections may not respond to such small catastrophe
events.
Finally, in underwriting our 2006 assumed reinsurance portfolio, we
sought to build a portfolio that was balanced with respect to geographic
region, peril and level of loss exposure. Because we have no control over
which of our clients will cancel their reinsurance contracts, there is a
significant risk that our remaining reinsurance portfolio will be unbalanced
and overexposed to loss for certain geographic areas, perils and loss exposure
levels.
SIDLEY AUSTIN LLP
- ----------------|
SIDLEY |
Mr. Frank Wyman
October 19, 2006
Page 18
Risk Limits
- -----------
We seek to manage our risk of loss through a combination of aggregate
exposure limits, underwriting guidelines that take into account risks, prices
and coverage and retrocessional agreements. As we underwrite risks from a
large number of clients based on information generally supplied by reinsurance
brokers, there is a risk of developing a concentration of exposure to loss in
certain geographic areas prone to specific types of catastrophes. We have
developed systems and software tools to monitor and manage the accumulation of
our exposure to such losses. We have established guidelines for maximum
probabilistic tolerable losses from a single or multiple catastrophic events
based on historical data. However, no assurance can be given that these
maximums will not be exceeded in some future catastrophe.
The purpose of the probabilistic risk limit is to seek to constrain the
maximum amount of net loss the Company could incur from a catastrophe event
with a given mathematical probability. For calendar year 2005 and prior
periods, the Company had established internal risk parameters whereby the
Company did not want to lose more than 25% of its beginning of calendar year
total "capital" ("capital" being defined as the sum of shareholders' equity
and subordinated debt) from a single large catastrophic event after adjusting
the income statement effect of the gross loss from such an event for (a)
reinsurance recoveries and net reinstatement premiums received associated with
the catastrophic event and (b) the expected profitability of the Company for
the calendar year excluding the catastrophic event. As a result of this risk
parameter and the Company's December 31, 2004 capital balance, for calendar
2005, the probabilistic risk limit for a gross loss from a single large
catastrophic event was set at $683 million. This limit served as a cap on the
amount of exposure accepted in a given geographic zone such that a single
event with a remote annual probability would not exceed $683 million. Because
it is probabilistic, the single event probabilistic risk limit does not seek
to be the maximum that could be lost from a particular event. What it does
seek to do, based on internal and third-party catastrophe models, is to
estimate the amount of loss from an event in each zone with a remote annual
probability.
The largest assumption made in calculating the amount of exposure which
could be accepted in compliance with the single event probabilistic risk limit
is the perils that are simulated in the third party catastrophe models. The
next largest assumption is the frequency assumptions for catastrophes with
given physical characteristics. Finally, the exposure information provided by
our clients is subject to change throughout the contract term. We assume that
the information is correct and complete, make assumptions for missing data and
expected changes in the portfolio during the contract term. All of these
assumptions give rise to potential differences between any actual event and
the single event probabilistic risk limit.
The Company has never exceeded its single event probabilistic risk limit
in any event other than Katrina. Several factors contributed to the Company's
exceeding its probabilistic risk limits in Hurricane Katrina. In setting
probabilistic risk limits, we are heavily reliant on third
SIDLEY AUSTIN LLP
- ----------------|
SIDLEY |
Mr. Frank Wyman
October 19, 2006
Page 19
party catastrophe models, which did not include flood damage assessments other
than storm surge associated with wind events in the US. Although we attempt to
account for the flood uncertainty in setting our risk limits, the losses
arising from flood coverage provided by our cedents to their policyholders for
Hurricane Katrina were more widespread than in past hurricanes. In certain
cases, losses were larger than expected for those treaties which saw increases
in cedents exposures in the US Gulf Coast from the time they purchased
reinsurance from us to the time of Hurricane Katrina. Finally, the
unprecedented characteristics of Hurricane Katrina resulted in significant
losses under treaties which would not normally be expected to trigger such
losses in a hurricane.