Changes from the July 13, 2006 Letter are in Bold and Italics

Attachment 1

CRITICAL ACCOUNTING ESTIMATES

The Company considers its most significant accounting estimates to be those applied to claim and claim adjustment expense reserves and related reinsurance recoverables, and investment impairments.

Claim and Claim Adjustment Expense Reserves

Claim and claim adjustment expense reserves (loss reserves) represent management’s estimate of ultimate unpaid costs of losses and loss adjustment expenses for claims that have been reported and claims that have been incurred but not yet reported. Loss reserves do not represent an exact calculation of liability, but instead represent management estimates, generally utilizing actuarial expertise and projection techniques, at a given accounting date. These loss reserve estimates are expectations of what the ultimate settlement and administration of claims will cost upon final resolution in the future, based on the Company’s assessment of facts and circumstances then known, review of historical settlement patterns, estimates of trends in claims severity and frequency, expected interpretations of legal theories of liability and other factors. In establishing reserves, the Company also takes into account estimated recoveries, reinsurance, salvage and subrogation. The reserves are reviewed regularly by a qualified actuary employed by the Company.

The process of estimating loss reserves involves a high degree of judgment and is subject to a number of variables. These variables can be affected by both internal and external events, such as changes in claims handling procedures, economic inflation, legal trends and legislative changes, among others. The impact of many of these items on ultimate costs for loss and loss adjustment expenses is difficult to estimate. Loss reserve estimation difficulties also differ significantly by product line due to differences in claim complexity, the volume of claims, the potential severity of individual claims, the determination of occurrence date for a claim and reporting lags (the time between the occurrence of the policyholder event and when it is actually reported to the insurer). Informed judgment is applied throughout the process. The Company continually refines its loss reserve estimates in a regular ongoing process as historical loss experience develops and additional claims are reported and settled. The Company rigorously attempts to consider all significant facts and circumstances known at the time loss reserves are established. Due to the inherent uncertainty underlying loss reserve estimates including but not limited to the future settlement environment, final resolution of the estimated liability will be different from that anticipated at the reporting date. Therefore, actual paid losses in the future may yield a materially different amount than currently reserved—favorable or unfavorable.

Because establishment of loss reserves is an inherently uncertain process involving estimates, currently established reserves may change. The Company reflects adjustments to reserves in the results of operations in the period the estimates are changed.

Changes from the July 13, 2006 Letter are in Bold and Italics

Attachment 1

There are also risks which impact the estimation of ultimate costs for catastrophes. For example, the estimation of reserves related to hurricanes can be affected by the inability of the Company and its insureds to access portions of the impacted areas, the complexity of factors contributing to the losses, the legal and regulatory uncertainties and the nature of the information available to establish the reserves. Complex factors include, but are not limited to: determining whether damage was caused by flooding versus wind; evaluating general liability and pollution exposures; estimating additional living expenses; the impact of demand surge; infrastructure disruption; fraud; the effect of mold damage and business interruption costs; and reinsurance collectibility. The timing of a catastrophe’s occurrence, such as at or near the end of a reporting period, can also affect the information available to the Company in estimating reserves for that reporting period. The estimates related to catastrophes are adjusted as actual claims emerge.

A portion of the Company’s loss reserves are for asbestos and environmental claims and related litigation which aggregated $4.79 billion at December 31, 2005. While the ongoing study of asbestos claims and associated liabilities and of environmental claims considers the inconsistencies of court decisions as to coverage, plaintiffs’ expanded theories of liability and the risks inherent in complex litigation and other uncertainties, in the opinion of the Company’s management, it is possible that the outcome of the continued uncertainties regarding these claims could result in liability in future periods that differs from current reserves by an amount that could be material to the Company’s future operating results and financial condition. See the preceding discussion of “Asbestos Claims and Litigation” and “Environmental Claims and Litigation.”

The Company acquired SPC’s runoff health care reserves in the merger, which are included in the General Liability product line in the table below. SPC decided to exit this market at the end of 2001 and ceased underwriting new business as quickly as regulatory considerations allowed. SPC had experienced significant adverse loss development on its health care loss reserves both prior to and since its decision to exit this market. The Company continues to utilize specific tools and metrics to explicitly monitor and validate its conclusions with regard to these reserves since management believed that its traditional statistics and reserving methods needed to be supplemented in order to provide a more meaningful analysis. The tools developed track three primary indicators which influence those conclusions and include: newly reported claims; reserve development on known claims; and the “redundancy ratio,” which compares the cost of resolving claims to the reserve established for that individual claim. These three indicators are related such that if one deteriorates, improvement on another is necessary for the Company to conclude that further reserve strengthening is not necessary. The Company’s current view is that it has recorded a reasonable reserve for its medical malpractice exposures as of December 31, 2005.

Changes from the July 13, 2006 Letter are in Bold and Italics

Attachment 1

Claims and claim adjustment expense reserves by product line were as follows:

	2005						2004
(at December 31, in millions)	Case		IBNR		Total		Case		IBNR		Total
General liability	$	8,198	$	12,251	$	20,449	$	8,445	$	12,232	$	20,677
Property	1,987		1,050		3,037		1,534		1,359		2,893
Commercial multi-peril	2,448		2,901		5,349		1,979		2,216		4,195
Commercial automobile	2,792		1,885		4,677		2,817		1,966		4,783
Workers’ compensation	8,816		6,374		15,190		8,313		6,658		14,971
Fidelity and surety	1,240		673		1,913		1,216		845		2,061
Personal automobile	1,470		1,138		2,608		1,484		1,219		2,703
Homeowners and personal—other	709		987		1,696		470		523		993
International and other	3,033		3,055		6,088		2,934		2,774		5,708

Property-casualty	30,693		30,314		61,007		29,192		29,792		58,984
Accident and health	74		9		83		76		10		86

Claims and claim adjustment expense reserves	$	30,767	$	30,323	$	61,090	$	29,268	$	29,802	$	59,070

The $2.02 billion increase in property-casualty claims and claim adjustment expense reserves since December 31, 2004 reflected the impact of the significant catastrophe losses incurred in the second half of 2005 and the $830 million net charge to increase asbestos reserves in the fourth quarter of 2005. These increases in reserves were partially offset by claim and claim expense payments from runoff operations, loss payouts for the third quarter 2004 hurricanes, and favorable non-catastrophe loss experience.

Property-casualty claims and claim adjustment expense reserves at December 31, 2004 increased by $24.51 billion over year-end 2003, primarily as a result of the merger with SPC and reserve charges recorded subsequent to the merger. Of the increase in 2004, $19.50 billion resulted from the addition of the acquired reserves, and $2.39 billion, net of reinsurance, was due to net unfavorable prior year reserve development. For discussion of the components of net unfavorable prior year reserve development, see note 9 to the consolidated financial statements.

Asbestos and environmental reserves are included in the General liability, Commercial multi-peril lines and International and other lines in the summary table. Asbestos and environmental reserves are discussed separately, see “Asbestos Claims and Litigation”, “Environmental Claims and Litigation” and “Uncertainty Regarding Adequacy of Asbestos and Environmental Reserves”.

Changes from the July 13, 2006 Letter are in Bold and Italics

Attachment 1

General Discussion

The process for estimating the liabilities for claim and claim expenses begins with the collection and analysis of claim data. Data on individual reported claims, both current and historical, including paid amounts and individual claim adjuster estimates, are grouped by common characteristics (“components”) and evaluated by actuaries in their analyses of ultimate claim liabilities by product line. Such data is occasionally supplemented with external data as available and when appropriate. The process of analyzing reserves for a component is undertaken on a regular basis, generally quarterly, in light of continually updated information.

Multiple estimation methods are available for the analysis of ultimate claim liabilities. Each estimation method has its own set of assumption variables and its own advantages and disadvantages, with no single estimation method being better than the others in all situations and no one set of assumption variables being meaningful for all product line components. The relative strengths and weaknesses of the particular estimation methods when applied to a particular group of claims can also change over time. Therefore, the actual choice of estimation method(s) can change with each evaluation. The estimation method(s) chosen are those that are believed to produce the most reliable indication at that particular evaluation date for the claim liabilities being evaluated.

In most cases, multiple estimation methods will be valid for the particular facts and circumstances of the claim liabilities being evaluated. This will result in a range of reasonable estimates for any particular claim liability. The Company uses such range analyses to back test whether previously established estimates for reserves at the reporting segments are reasonable, given subsequent information. Reported values found to be closer to the endpoints of a range of reasonable estimates are subject to further detailed reviews. These reviews may substantiate the validity of management’s recorded estimate or lead to a change in the reported estimate.

The exact boundary points of these ranges are more qualitative than quantitative in nature, as no clear line of demarcation exists to determine when the set of underlying assumptions for an estimation method switches from being reasonable to unreasonable. As a result, the Company does not believe that the endpoints of these ranges are or would be comparable across companies. In addition, potential interactions among the different estimation assumptions for different product lines make the aggregation of individual ranges a highly judgmental and inexact process.

Property casualty insurance policies are either written on a claims made or on an occurrence basis. Policies written on a claims made basis require that claims be reported during the policy period. Policies that are written on an occurrence basis require that the insured demonstrate that a loss occurred in the policy period, even if the insured reports the loss many years later.

Changes from the July 13, 2006 Letter are in Bold and Italics

Attachment 1

Most general liability policies are written on an occurrence basis. These policies are subject to substantial loss development over time as facts and circumstances change in the years following the policy issuance. The use of the occurrence form accounts for much of the reserve development in asbestos and environmental exposures, and it is also used to provide coverage for construction general liability, including construction defect. Occurrence based forms of insurance for general liability exposures require substantial projection of various trends, including future inflation and judicial interpretations and societal litigation dynamics, among others.

A basic premise in most actuarial analyses is that past patterns demonstrated in the data will repeat themselves in the future, absent a material change in the associated risk factors discussed below. To the extent a material change affecting the ultimate claim liability is known, such change is quantified to the extent possible through an analysis of internal company and, if available and when appropriate, external data. Such a measurement is specific to the facts and circumstances of the particular claim portfolio and the known change being evaluated. Significant structural changes to the available data, product mix or organization can materially impact the reserve estimation process.

Informed management judgment is applied throughout the reserving process. This includes the application, on a consistent basis over time, of various individual experiences and expertise to multiple sets of data and analyses. In addition to actuaries, individuals involved with the reserving process also include underwriting and claims personnel as well as other company management. Therefore, it is quite possible and, generally, likely that management must consider varying individual viewpoints as part of its estimation of loss reserves. It is also likely that during periods of significant change, such as a merger, consistent application of informed judgment becomes even more complicated and difficult.

The variables discussed above in this general discussion have different impacts on reserve estimation uncertainty for a given product line, depending on the length of the claim tail, the reporting lag, the impact of individual claims and the complexity of the claim process for a given product line.

Product lines are generally classifiable as either long tail or short tail, based on the average length of time between the event triggering claims under a policy and the final resolution of those claims. Short tail claims are reported and settled quickly, resulting in less estimation variability. The longer the time before final claim resolution, the greater the exposure to estimation risks and hence the greater the estimation uncertainty.

A major component of the claim tail is the reporting lag. The reporting lag, which is the time between the event triggering a claim and the reporting of the claim to the insurer, makes estimating IBNR inherently more uncertain. In addition, the greater the reporting

Changes from the July 13, 2006 Letter are in Bold and Italics

Attachment 1

lag the greater the proportion of IBNR claims to the total claim liability for the product line. Writing new products with material reporting lags can result in adding several years worth of IBNR claim exposure before the reporting lag exposure becomes clearly observable, thereby increasing the risk associated with pricing and reserving such products. The most extreme example of claim liabilities with long reporting lags are asbestos claims.

For some lines, the impact of large individual claims can be material to the analysis. These lines are generally referred to as being low frequency/high severity, while lines without this “large claim” sensitivity are referred to as “high frequency/low severity”. Estimates of claim liabilities for low frequency/high severity lines can be sensitive to the impact of a small number of potentially large claims. As a result, the role of judgment is much greater for these reserve estimates. In contrast, high frequency/low severity lines tend to have much greater spread of estimation risk, such that the impact of individual claims are relatively minor and the range of reasonable reserve estimates is narrower and more stable.

Claim complexity can also greatly affect the estimation process by impacting the number of assumptions needed to produce the estimate, the potential stability of the underlying data and claim process and the ability to gain an understanding of the data. Product lines with greater claim complexity, such as for certain surety and construction exposures, have inherently greater estimation uncertainty.

Actuaries have to exercise a considerable degree of judgment in the evaluation of all these factors in their analysis of reserves. The human element in the application of actuarial judgment is unavoidable when faced with material uncertainty. Different experts will choose different assumptions when faced with such uncertainty, based on their individual backgrounds, professional experiences and areas of focus. Hence, the estimate selected by the various actuaries may differ materially from each other.

Lastly, significant structural changes to the available data, product mix or organization can also materially impact the reserve estimation process. The merger of TPC and SPC resulted in the exposure of each other’s actuaries and claim departments to different products, data histories, analysis methodologies, claim settlement experts, and more robust data when viewed on a combined basis. This impacted the range of estimates produced by the Company’s actuaries, as they reacted to new data, approaches, and sources of expertise to draw upon. It also resulted in additional levels of uncertainty, as past trends (that were a function of past products, past claim handling procedures, past claim departments, and past legal and other experts) may not repeat themselves, as those items affecting the trends change or evolve due to the merger. This also increased the potential for material variation in estimates, as experts can have differing views as to the impact of these frequently evolutionary changes. Events such as mergers increase the inherent uncertainty of reserve estimates for a period of time, until stable trends reestablish themselves within the new organization.

Changes from the July 13, 2006 Letter are in Bold and Italics

Attachment 1

Risk factors

The major causes of material uncertainty (“risk factors”) generally will vary for each product line, as well as for each separately analyzed component of the product line. In a few cases, such risk factors are explicit assumptions of the estimation method and in most cases, they are implicit. For example, a method may explicitly assume that a certain percentage of claims will close each year, but will implicitly assume that the legal interpretation of existing contract language will remain unchanged. Actual results will likely vary from expectations for each of these assumptions, resulting in an ultimate claim liability that is different from that being estimated currently.

Some risk factors will affect more than one product line. Examples include changes in claim department practices, changes in settlement patterns, regulatory and legislative actions, court actions, timeliness of claim reporting, state mix of claimants, and degree of claimant fraud. The extent of the impact of a risk factor will also vary by components within a product line. Individual risk factors are also subject to interactions with other risk factors within product line components.

The effect of a particular risk factor on estimates of claim liabilities cannot be isolated in most cases. For example, estimates of potential claim settlements may be impacted by the risk associated with potential court rulings, but the final settlement agreement typically does not delineate how much of the settled amount is due to this and other factors.

The evaluation of data is also subject to distortion from extreme events or structural shifts, sometimes in unanticipated ways. For example, the timing of claims payments in one geographic region will be impacted if claim adjusters are temporarily reassigned from that region to help settle catastrophe claims in another region.

While some changes in the claim environment are sudden in nature (such as a new court ruling affecting the interpretation of all contracts in that jurisdiction), others are more evolutionary. Evolutionary changes can occur when multiple factors affect final claim values, with the uncertainty surrounding each factor being resolved separately, in step-wise fashion. The final impact is not known until all steps have occurred.

Sudden changes generally cause a one-time shift in claim liability estimates, although there may be some lag in reliable quantification of their impact. Evolutionary changes generally cause a series of shifts in claim liability estimates, as each component of the evolutionary change becomes evident and estimable.

Changes from the July 13, 2006 Letter are in Bold and Italics

Attachment 1

Actuarial methods for analyzing and estimating claim and claim adjustment expense reserves.

The principal estimation and analysis methods utilized by the Company’s actuaries are the paid development method, the case incurred development method(1), the Bornhuetter-Ferguson (BF) method, and average value analysis combined with the reported claim development method. The BF method is usually utilized for more recent accident periods, with a transition to other methods as the underlying claim data becomes more voluminous and therefore more credible. These are typically referred to as “traditional” actuarial methods. (See Glossary for an explanation of these methods.)

While these are the principal methods utilized throughout the Company, those evaluating a particular component for a product line have available to them the full range of methods developed within the casualty actuarial profession. The Company’s actuaries are also continually monitoring developments within the profession for advances in existing techniques or the creation of new techniques that might improve current and future estimates.

Some components of product line reserves are susceptible to relatively infrequent large claims that can materially impact the total estimate for that component. In such cases, the Company’s actuarial analysis generally isolates and analyzes separately such large claims. The reserves excluding such large claims are generally analyzed using the traditional methods described above. The reserves associated with large claims are then analyzed utilizing various methods such as:

· Estimating the number of large claims and their average values based on historical trends from prior accident periods, adjusted for the current environment and supplemented with actual data for the accident year analyzed to the extent available.

· Utilizing individual claim adjuster estimates of the large claims, combined with continual monitoring of the aggregate accuracy of such claim adjuster estimates. (This monitoring may lead to supplemental adjustments to the aggregate of such claim estimates.)

· Utilizing historic longer-term average ratios of large claims to small claims, and applying such ratios to the estimated ultimate small claims from traditional analysis.

· Ground-up analysis of the underlying exposure (typically used for asbestos and environmental).

The results of such methodologies are subjected to various reasonability and diagnostic tests, including paid-to-incurred loss ratios, implied incurred-loss-to-earned-premium ratios and non-zero claim severity trends. An actual versus expected analysis is also performed comparing actual loss development in the last six months to expected

(1) The paid and incurred development methods are sometimes also known as “link ratio” methods.

Changes from the July 13, 2006 Letter are in Bold and Italics

Attachment 1

development based on the prior review. Additional analysis may be performed based on the results of these diagnostics, including the investigation of other actuarial methods.

The above is generally utilized to evaluate management’s existing estimate for prior accident periods. For the initial estimate of the current accident year, the available claim data is typically insufficient to produce a reliable indication. Hence, the initial estimate for an accident year is generally based on a loss ratio projection method which uses the earned premium for the current year multiplied by a projected loss ratio. The projected loss ratio is determined through analysis of prior experience periods using loss trend, rate level differences, mix of business changes and other known or observed factors influencing the current accident year relative to prior accident years. The exact number of prior accident years utilized varies by product line component, based on the volume of business for that component and the reliability of an individual accident year estimate.

Management’s estimates

At least once per quarter, Company management meets with its actuaries to review the latest claim and claim adjustment expense reserve analyses. Based on these analyses, management determines whether its ultimate claim liability estimates should be changed. In doing so, it must evaluate whether the new data provided represents credible actionable information or an anomaly that will have no effect on estimated ultimate claim liability. For example, as described above, payments may have decreased in one geographic region due to fewer claim adjusters being available to process claims. The resulting claim payment patterns would be analyzed to determine whether or not the change in payment pattern represents a change in ultimate claim liability.

Such an assessment requires considerable judgment. It is frequently not possible to determine whether a change in the data is an anomaly until sometime after the event. Even if a change is determined to be permanent, it is not always possible to reliably determine the extent of the change until sometime later. The overall detailed analyses supporting such an effort can take several months to perform. This is due to the need to evaluate the underlying cause of the trends observed, and may include the gathering or assembling of data not previously available. It may also include interviews with experts involved with the underlying processes. As a result, there can be a time lag between the emergence of a change and a determination that the change should be reflected in the Company’s estimated claim liabilities. The final estimate selected by management in a reporting period is a function of these detailed analyses of past data, adjusted to reflect any new actionable information.

Changes from the July 13, 2006 Letter are in Bold and Italics

Attachment 1

Discussion of Product Lines

The following section details reserving considerations and common risk factors by product line. There are many additional risk factors that may impact ultimate claim costs. Each risk factor presented will have a different impact on required reserves. Also, risk factors can have offsetting or compounding effects on required reserves. For example, in workers’ compensation, the use of expensive medical procedures that result in medical cost inflation may enable workers to return to work faster, thereby lowering indemnity costs. Thus, in almost all cases, it is impossible to discretely measure the effect of a single risk factor and construct a meaningful sensitivity expectation.

In order to provide information on reasonably possible reserving changes by product line, the historical changes in year-end loss reserves over a one-year period are provided for the U.S. product lines. This information is provided for both the Company and the industry(2) for the nine most recent years, and is based on publicly available data for the reported line(s) that most closely match the individual product line being discussed.

General Liability

General liability is generally considered a long tail line, as it takes a relatively long period of time to finalize and settle claims from a given accident year. The speed of claim reporting and claim settlement is a function of the specific coverage provided, the jurisdiction and specific policy provisions such as self-insured retentions. There are numerous components underlying the general liability product line. Some of these have relatively moderate payment patterns (with most of the claims for a given accident year closed within 5 to 7 years), while others can have extreme lags in both reporting and payment of claims (e.g., a reporting lag of a decade for “construction defect” claims).

While the majority of general liability coverages are written on an “occurrence basis,” certain general liability coverages (such as those covering directors and officers or professional liability) are typically insured on a “claims-made” basis.

General liability reserves are generally analyzed as two components: primary and excess/umbrella, with the primary component generally analyzed separately for bodily injury and property damage. Bodily injury liability payments reimburse the claimant for damages pertaining to physical injury as a result of the policyholder’s legal obligation

(2) These changes were calculated from (net of reinsurance) statutory annual statement data found in Schedule P of those statements, and represent the reported reserve development on the beginning of the year claim liabilities divided by the beginning claim liabilities, all accident years combined, excluding non-defense related claim adjustment expense.

Note that the Company data included history for the entire St. Paul Travelers group – U.S. companies only, whether or not the individual subsidiaries were originally part of the St. Paul or the Travelers. This treatment is required by the statutory reporting instructions promulgated by state regulatory authorities for Schedule P. Comparable data for non-U.S. companies is not available.

Changes from the July 13, 2006 Letter are in Bold and Italics

Attachment 1

arising from non-intentional acts such as negligence, subject to the insurance policy provisions. In some cases the damages can include future wage loss (which is a function of future earnings power and wage inflation) and future medical treatment costs. Property damage liability payments result from damages to the claimant’s private property arising from the policyholder’s legal obligation for non-intentional acts. In most cases, property damage losses are a function of costs as of the loss date, or soon thereafter. In addition, sizable or unique exposures are reviewed separately, such as asbestos, environmental, other mass torts, construction defect, medical malpractice and large unique accounts that would otherwise distort the analysis. These unique categories often require a very high degree of judgment and require reserve analyses that do not rely on traditional actuarial methods.

Defense costs are also a part of the insured costs covered by liability policies and can be significant, sometimes greater than the cost of the actual paid claims. For some products this risk is mitigated by policy language such that the insured portion of defense costs erodes the amount of policy limit available to pay the claim. Such “defense within the limits” policies are most common for “claims made” products. When defense costs are outside of the limits, amounts paid do not erode the policy limits.

This line is typically the largest source of reserve estimate uncertainty in the United States (excluding assumed reinsurance contracts covering the same risk). Major contributors to this reserve estimate uncertainty include the reporting lag (i.e., the length of time between the event triggering coverage and the actual reporting of the claim), the number of parties involved in the underlying tort action, whether the “event” triggering coverage is confined to only one time period or is spread over multiple time periods, the potential dollars involved (in the individual claim actions), whether such claims were reasonably foreseeable and intended to be covered at the time the contracts were written (i.e., coverage dispute potential), and the potential for mass claim actions. Claims with longer reporting lags result in greater inherent risk. This is especially true for alleged claims with a latency feature, particularly where courts have ruled that coverage is spread over multiple policy years, hence involving multiple defendants (and their insurers and reinsurers) and multiple policies (thereby increasing the potential dollars involved and the underlying settlement complexity). Claims with long latencies also increase the potential recognition lag, i.e., the lag between writing a type of policy in a certain market and the recognition that such policies have potential mass tort and/or latent claim exposure.

The amount of reserve estimate uncertainty also varies significantly by component for the General Liability product line. The components in this product line with the longest latency, longest reporting lags, largest potential dollars involved, and greatest claim settlement complexity are Asbestos and Environmental. Components that include latency, reporting lag and/or complexity issues, but to a materially lesser extent than Asbestos and Environmental, include construction defect, medical malpractice, and other mass tort actions. Many components of General Liability are not subject to material latency or claim complexity risks and hence have materially less uncertainty than the previously mentioned components. In general, policies providing coverage with shorter reporting

Changes from the July 13, 2006 Letter are in Bold and Italics

Attachment 1

lags, fewer parties involved in settlement negotiations, only one policy potentially triggered per claim, fewer potential settlement dollars, reasonably foreseeable (and stable) potential hazards/claims and no mass tort potential result in much less reserve estimate uncertainty than policies without those characteristics.

Besides the traditional actuarial methods mentioned in the general discussion section, the company utilizes various report year development methods and S-curves for the Construction Defect components of this product line. The Construction Defect report year development analysis is supplemented with projected claim counts and average values for IBNR claim counts. For components with greater lags in claim reporting, such as excess and umbrella components of this product line, the company utilizes the BF method more heavily than paid and case incurred development.

Examples of common risk factors that can change and, thus, affect the required general liability reserves (beyond those included in the general discussion section) include:

General liability risk factors

Changes in claim handling philosophies

Changes in policy provisions or court interpretation of such provision

New theories of liability

Trends in jury awards

Changes in the propensity to sue, in general with specificity to particular issues

Changes in statutes of limitations

Changes in the underlying court system

Distortions from losses resulting from large single accounts or single issues

Changes in tort law

Shifts in law suit mix between federal and state courts

Changes in claim adjuster office structure (causing distortions in the data)

Changes in settlement patterns (e.g., medical malpractice)

General liability book of business risk factors

Changes in policy provisions (e.g., deductibles, policy limits, endorsements)

Changes in underwriting standards

Product mix (e.g., size of account, industries insured, jurisdiction mix)

Unanticipated changes in risk factors can affect reserves. As an indicator of the causal effect that a change in one or more risk factors could have on reserves for general liability (excluding asbestos and environmental), a 1% increase (decrease) in incremental paid loss development for each future calendar year could result in a 1.6% increase (decrease) in loss reserves.

Changes from the July 13, 2006 Letter are in Bold and Italics

Attachment 1

Historically, the one-year change in the reserve estimate for this product line over the last nine years has varied from 0% to +17% (averaging +7%) for the Company and -2% to +9% (averaging +4%) for the industry overall, excluding estimated asbestos and environmental amounts. The Company’s year-to-year changes are driven by and are based on observed events during the year. We believe the industry’s range of historical outcomes is illustrative of reasonably possible one-year changes in reserve estimates for this product line, as the high end of the Company’s range of historical adverse development came from certain business that has since been exited. General liability reserves (excluding asbestos and environmental) represent approximately 26% of the Company’s total loss reserves.

The Company’s change in reserve estimate for this product line was +4% for 2005 and +12% for 2004. The 2005 change of +4% was the net result of numerous adjustments for various components. The 2004 change of +12% was principally due to the construction defect and construction wrap-up reserving actions discussed on page 68 of this narrative.

Property

Property is generally considered a short tail line with a simpler and faster claim reporting and adjustment process than liability coverages, and less uncertainty in the reserve setting process (except for more complex business interruption claims). It is generally viewed as a moderate frequency, low to moderate severity line, except for catastrophes and coverage related to large properties. The claim reporting and settlement process for property coverage claim reserves is generally restricted to the insured and the insurer. Overall, the claim liabilities for this line create a low estimation risk, except possibly for catastrophes and business interruption claims.

Property reserves are typically analyzed in two components, one for catastrophic or other large single events, and another for all other events. Examples of common risk factors that can change and, thus, affect the required property reserves (beyond those included in the general discussion section) include:

Property risk factors

Physical concentration of policyholders

Availability and cost of local contractors

For the more severe catastrophic events, “demand surge” inflation, whereby the greatly increased demand for building materials such as plywood far surpasses the immediate supply, leading to short-term material increases in building material costs

Local building codes

Amount of time to return property to full usage (for business interruption claims)

Court interpretation of policy provisions (such as occurrence definition)

Lags in reporting claims (e.g., winter damage to summer homes, hidden damage after an earthquake)

Court or legislative changes to the statute of limitations

Changes from the July 13, 2006 Letter are in Bold and Italics

Attachment 1

Property book of business risk factors

Policy provisions mix (e.g., deductibles, policy limits, endorsements)

Changes in underwriting standards

Unanticipated changes in risk factors can affect reserves. As an indicator of the causal effect that a change in one or more risk factors could have on reserves for property, a 1% increase (decrease) in incremental paid loss development for each future calendar year could result in a 1.2% increase (decrease) in loss reserves.

Historically, the one-year change in the reserve estimate for this product line over the last nine years has varied from -34% to +26% (averaging -3%) for the Company and - -14% to +7% (averaging 0%) for the industry overall. The Company’s year-to-year changes are driven by and are based on observed events during the year. We believe that the industry’s range of historical outcomes is illustrative of reasonably possible one-year changes in reserve estimates for this product line, as the high end of the Company’s range of historical adverse development came from certain business that has since been exited. Property reserves represent approximately 5% of the Company’s total loss reserves.

The Company’s change in reserve estimate for this product line was -34% for 2005 and -18% for 2004. The 2005 change was primarily the result of favorable loss developments for accident year 2004 non-catastrophe losses. The 2004 change was primarily the result of favorable loss developments for accident year 2003 non-catastrophe losses. Although not the case for 2005 and 2004, the reserve estimate for this product line is potentially subject to material changes due to uncertainty in measuring ultimate losses for unprecedented significant catastrophes such as the events of September 11, 2001 and Hurricane Katrina.

Commercial Multi-Peril

Commercial multi-peril provides a combination of property and liability coverage typically for small businesses and, therefore, includes both short and long tail coverages. For property coverage, it generally takes a relatively short period of time to close claims, while for the other coverages, generally for the liability coverages, it takes a longer period of time to close claims.

The reserving risk for this line is dominated by the liability coverage portion of this product, except occasionally in the event of catastrophic or large single losses. The reserving risk for this line differs from that of the general liability product line and the property product line due to the nature of the customer. Commercial multi-peril is generally sold to smaller sized accounts, while the customer profile for general liability and property include larger customers.

Changes from the July 13, 2006 Letter are in Bold and Italics

Attachment 1

See the discussions under the property and general liability product lines with regard to reserving risk for commercial multi-peril.

Unanticipated changes in risk factors can affect reserves. As an indicator of the causal effect that a change in one or more risk factors could have on reserves for commercial multi-peril (excluding asbestos and environmental), a 1% increase (decrease) in incremental paid loss development for each future calendar year could result in a 1.3% increase (decrease) in loss reserves.

Historically, the one-year change in the reserve estimate for this product line over the last nine years has varied from -6% to +2% (averaging -3%) for the Company and -2% to +7% (averaging +3%) for the industry overall. The Company’s year-to-year changes are driven by and are based on observed events during the year. We believe the Company’s range of historical outcomes is illustrative of reasonably possible one-year changes in reserve estimates for this product line. Commercial multi-peril reserves represent approximately 9% of the Company’s total loss reserves.

As discussed above, this line combines general liability and property coverages and it has been impacted in the past by many of the same events as those two lines.

The Company’s change in reserve estimate for this product line was -6% for 2005 and -5% for 2004. The 2005 change was the result of favorable loss developments for accident years 2001 – 2004. The 2004 change was the result of favorable loss developments for accident years 1999 - 2003.

Commercial Automobile

The commercial automobile product line is a mix of property and liability coverages and, therefore, includes both short and long tail coverages. The payments that are made quickly typically pertain to auto physical damage (property) claims and property damage (liability) claims. The payments that take longer to finalize and are more difficult to estimate relate to bodily injury claims. In general, claim reporting lags are minor, claim complexity is not a major issue, and the line is viewed as high frequency, low to moderate severity. Overall, the claim liabilities for this line create a moderate estimation risk.

Commercial automobile reserves are typically analyzed in four components; bodily injury liability, property damage liability, collision claims and comprehensive claims. These last two components have minimum reserve risk and fast payouts and, accordingly, separate risk factors are not presented.

The company utilizes the traditional actuarial methods mentioned in the general discussion above in estimating claim liabilities for this line. This is supplemented with detailed custom analyses where needed.

Changes from the July 13, 2006 Letter are in Bold and Italics

Attachment 1

Examples of common risk factors that can change and, thus, affect the required commercial automobile reserves (beyond those included in the general discussion section) include:

Bodily injury and property damage liability risk factors

Trends in jury awards

Changes in the underlying court system

Changes in case law

Litigation trends

Frequency of claims with payment capped by policy limits

Change in average severity of accidents, or proportion of severe accidents

Subrogation opportunities

Changes in claim handling philosophies

Frequency of visits to health providers

Number of medical procedures given during visits to health providers

Types of health providers used

Types of medical treatments received

Changes in cost of medical treatments

Degree of patient responsiveness to treatment

Commercial automobile book of business risk factors

Changes in policy provisions (e.g., deductibles, policy limits, endorsements, etc.)

Changes in mix of insured vehicles (e.g., long haul trucks versus local and smaller vehicles, fleet risks versus non-fleets)

Changes in underwriting standards

Unanticipated changes in risk factors can affect reserves. As an indicator of the causal effect that a change in one or more risk factors could have on reserves for commercial automobile, a 1% increase (decrease) in incremental paid loss development for each future calendar year could result in a 1.3% increase (decrease) in loss reserves.

Historically, the one-year change in the reserve estimate for this product line over the last nine years has varied from -5% to +9% (averaging +1%) for the Company and -1% to +8% (averaging +3%) for the industry overall. The Company’s year-to-year changes are driven by and are based on observed events during the year. We believe that the Company’s range of historical outcomes is illustrative of reasonable possibly one-year changes in reserve estimate for this product line. Commercial automobile reserves represent approximately 8% of the Company’s total loss reserves.

The Company’s change in reserve estimate for this product line was -5% for 2005 and -2% for 2004. The 2005 change was due to favorable loss developments for accident year 2004. The 2004 change was due to favorable loss developments for accident year 2003.

Changes from the July 13, 2006 Letter are in Bold and Italics

Attachment 1

Workers’ Compensation

Workers’ compensation is generally considered a long tail coverage, as it takes a relatively long period of time to finalize claims from a given accident year. While certain payments such as initial medical treatment or temporary wage replacement for the injured worker are made quickly, some other payments are made over the course of several years, such as awards for permanent partial injuries. In addition, some payments can run as long as the injured worker’s life, such as permanent disability benefits and on-going medical care. Despite the possibility of long payment tails, the reporting lags are generally short, settlements are generally not complex, and most of the liability can be considered high frequency with moderate severity. The largest reserve risk generally comes from the low frequency, high severity claims providing lifetime coverage for medical expense arising from a worker’s injury. Overall, the claim liabilities for this line create a somewhat greater than moderate estimation risk.

Workers’ compensation reserves are typically analyzed in three components: indemnity losses, medical losses and claim adjustment expenses.

Examples of common risk factors that can change and, thus, affect the required workers’ compensation reserves (beyond those included in the general discussion section) include:

Indemnity risk factors

Time required to recover from the injury

Degree of available transitional jobs

Degree of legal involvement

Changes in the interpretations and processes of the workers’ compensation commissions’ oversight of claims (1)

Future wage inflation for states that index benefits

Changes in the administrative policies of second injury funds

Medical risk factors

Changes in the cost of medical treatments (including prescription drugs) and underlying fee schedules (“inflation”)

Frequency of visits to health providers

Number of medical procedures given during visits to health providers

Types of health providers used

Type of medical treatments received

Use of preferred provider networks and other medical cost containment practices

Availability of new medical processes and equipment

Changes in the use of pharmaceutical drugs

Degree of patient responsiveness to treatment

Changes from the July 13, 2006 Letter are in Bold and Italics

Attachment 1

(1) These are administrative bodies that evaluate whether or not a given claim for workers’ compensation benefits is valid. Duties include the determination of whether a given injury arose out of the scope of employment, or the determination of the degree of injury where disputes exist.

General workers’ compensation risk factors

Frequency of claim reopenings on claims previously closed

Mortality trends of injured workers with lifetime benefits and medical treatment

Degree of cost shifting between workers’ compensation and health insurance

Workers’ compensation book of business risk factors

Product mix

Injury type mix

Changes in underwriting standards

Unanticipated changes in risk factors can affect reserves. As an indicator of the causal effect that a change in one or more risk factors could have on reserves for workers’ compensation, a 1% increase (decrease) in incremental paid loss development for each future calendar year could result in a 1.3% increase (decrease) in loss reserves.

Historically, the one-year change in the reserve estimate for this product line over the last nine years has varied from -2% to +2% (averaging -1%) for the Company and -3% to +4% (averaging +1%) for the industry overall. The Company’s year-to-year changes are driven by and are based on observed events during the year. We believe that the Company’s range of historical outcomes is illustrative of reasonably possible one-year changes in reserve estimates for this product line. Workers’ compensation reserves represent approximately 25% of the Company’s total loss reserves.

The Company’s change in reserve estimate for this product line was 0% for 2005 and +1% for 2004.

Fidelity and Surety

Fidelity is generally considered a short tail coverage. It takes a relatively short period of time to finalize and settle fidelity claims. The volatility of fidelity reserves is generally related to the type of business of the insured, the size and complexity of the insured’s business operations, amount of policy limit and attachment point of coverage. The uncertainty surrounding reserves for small, commercial insureds is typically less than the uncertainty for large commercial or financial institutions. The high frequency, low severity nature of small commercial fidelity losses provides for stability in loss estimates whereas, the low frequency, high severity nature of losses for large insureds results in a wider range of ultimate loss outcomes. Actuarial techniques that rely on a stable pattern of loss development are generally not applicable to low frequency, high severity policies.

Changes from the July 13, 2006 Letter are in Bold and Italics

Attachment 1

Surety has certain components that are generally considered short tail coverages with short reporting lags, although large individual construction and commercial surety contracts can result in a somewhat longer settlement tail, based on the length and complexity of the construction project or commercial transaction being insured. (Large construction projects can take many years to complete.) The frequency of losses in surety correlates with economic cycles as the primary cause of surety loss is the inability to perform financially. The volatility of surety losses is generally related to the type of business performed by the insured, the type of bonded obligation, the amount of limit exposed to loss and the amount of assets available to the insurer to mitigate losses, such as unbilled contract funds, collateral, first and third party indemnity, and other security positions of an insured’s assets. Certain classes of surety claims are very high severity, low frequency in nature. These can include large construction contractors involved with one or multiple large, complex projects as well as certain large commercial surety exposures. Other claim factors affecting reserve variability of surety include litigation related to amounts owed by and due the insured (e.g., salvage and subrogation efforts) and the results of financial restructuring of an insured.

100

Examples of common risk factors that can change and, thus, affect the required fidelity and surety reserves (beyond those included in the general discussion section) include:

Fidelity risk factors

Type of business of insured

Policy limit and attachment points

Third-party claims

Coverage litigation

Complexity of claims

Growth in insureds’ operations

Surety risk factors

Economic trends, including the general level of construction activity

Concentration of reserves in a relatively few large claims

Type of business insured

Type of obligation insured

Cumulative limits of liability for insured

Assets available to mitigate loss

Defective workmanship/latent defects

Financial strategy of insured

Changes in statutory obligations

Geographic spread of business

Fidelity and Surety book of business risk factors

Changes in policy provisions (e.g., deductibles, limits, endorsements)

Changes from the July 13, 2006 Letter are in Bold and Italics

Attachment 1

Changes in underwriting standards

Unanticipated changes in risk factors can affect reserves. As an indicator of the causal effect that a change in one or more risk factors could have on reserves for fidelity and surety, a 1% increase (decrease) in incremental paid loss development for each future calendar year could result in a 1.2% increase (decrease) in loss reserves.

Historically, the one-year change in the reserve estimate for this product line over the last nine years has varied from -15% to +138% (averaging +19% for all 9 years, or +5% excluding the 2004 value of +138% that resulted from acquired business, including a large construction contractor loss discussed on pages 69 and 70 of this narrative) for the Company and -9% to +24% (averaging +7%) for the industry overall. The Company’s year-to-year changes are driven by and are based on observed events during the year. We believe that the industry’s range of historical outcomes is illustrative of reasonably possible one-year changes in reserve estimates for this product line, as the high end of the Company’s range was due to acquired business in 2004. Fidelity and surety reserves represent approximately 3% of the Company’s total loss reserves.

In general, developments on single large claims (both adverse and favorable) are a primary source of changes in reserve estimates for this product line.

The Company’s change in reserve estimate for this product line was 0% for 2005 and +138% for 2004. The 138% figure resulted from acquired business, including a large construction contractor loss discussed on pages 69 and 70 of this narrative.

Personal Automobile

Personal automobile includes both short and long tail coverages. The payments that are made quickly typically pertain to auto physical damage (property) claims and property damage (liability) claims. The payments that take longer to finalize and are more difficult to estimate relate to bodily injury claims. Reporting lags are relatively short and the claim settlement process for personal automobile liability generally is the least complex of the liability products. It is generally viewed as a high frequency, low to moderate severity product line. Overall, the claim liabilities for this line create a moderate estimation risk.

Personal automobile reserves are typically analyzed in five components: bodily injury liability, property damage liability, no-fault losses, collision claims and comprehensive claims. These last two components have minimum reserve risk and fast payouts and, accordingly, separate factors are not presented.

Examples of common risk factors that can change and, thus, affect the required personal automobile reserves (beyond those included in the general discussion section) include:

Bodily injury and property damage liability risk factors

Trends in jury awards

Changes from the July 13, 2006 Letter are in Bold and Italics

Attachment 1

Changes in the underlying court system and its philosophy

Changes in case law

Litigation trends

Frequency of claims with payment capped by policy limits

Change in average severity of accidents, or proportion of severe accidents

Subrogation opportunities

Degree of patient responsiveness to treatment

Changes in claim handling philosophies

101

No-fault risk factors (for selected states and time periods)

Effectiveness of no-fault laws

Frequency of visits to health providers

Number of medical procedures given during visits to health providers

Types of health providers used

Types of medical treatments received

Changes in cost of medical treatments

Degree of patient responsiveness to treatment

Personal automobile book of business risk factors

Changes in policy provisions (e.g., deductibles, policy limits, endorsements, etc.)

Changes in underwriting standards

Unanticipated changes in risk factors can affect reserves. As an indicator of the causal effect that a change in one or more risk factors could have on reserves for personal automobile, a 1% increase (decrease) in incremental paid loss development for each future calendar year could result in a 1.1% increase (decrease) in loss reserves.

Historically, the one-year change in the reserve estimate for this product line over the last nine years has varied from -9% to +1% (averaging -4%) for the Company and -7% to +3% (averaging - -2%) for the industry overall. The Company’s year-to-year changes are driven by and are based on observed events during the year. We believe that the Company’s range of historical outcomes is illustrative of reasonably possible one-year changes in reserve estimates for this product line. Personal automobile reserves represent approximately 4% of the Company’s total loss reserves.

The Company’s change in reserve estimate for this product line was -9% for 2005 and -6% for 2004. As discussed on page 72 of this narrative, the decrease in both years was primarily driven by declines in claim frequency (for the most recent prior accident years) and in severity resulting from recent claim initiatives.

Changes from the July 13, 2006 Letter are in Bold and Italics

Attachment 1

Homeowners and Personal Lines Other

Homeowners is generally considered a short tail coverage. Most payments are related to the property portion of the policy, where the claim reporting and settlement process is generally restricted to the insured and the insurer. Claims on property coverage are typically reported soon after the actual damage occurs, although delays of several months are not unusual. The claim is settled when the two parties agree on the amount due in accordance with the policy contract language and the appropriate payment is made (or alternatively, the property replacement/repair is performed by the insurer). The resulting settlement process is typically fairly short term, although exceptions do exist.

The liability portion of the homeowners policy generates claims which take longer to pay due to the involvement of litigation and negotiation, but with generally small reporting lags. In addition, reserves related to umbrella coverages have greater uncertainty since umbrella liability payments are often made far into the future.

Overall, the line is generally high frequency, low to moderate severity (except for catastrophes), with simple to moderate claim complexity.

Homeowners reserves are typically analyzed in two components: non-catastrophe related losses and catastrophe loss payments.

Examples of common risk factors that can change and, thus, affect the required homeowners reserves (beyond those included in the general reserve discussion section) include:

Non-catastrophe risk factors

Salvage opportunities

Amount of time to return property to residential use

Changes in weather patterns

Local building codes

Litigation trends

Trends in jury awards

Catastrophe risk factors

Physical concentration of policyholders

Availability and cost of local contractors

Local building codes

Quality of construction of damaged homes

Amount of time to return property to residential use

Homeowners book of business risk factors

Policy provisions mix (e.g., deductibles, policy limits, endorsements, etc.)

Degree of concentration of policyholders

Changes from the July 13, 2006 Letter are in Bold and Italics

Attachment 1

Changes in underwriting standards

Unanticipated changes in risk factors can affect reserves. As an indicator of the causal effect that a change in one or more risk factors could have on reserves for homeowners and personal lines other, a 1% increase (decrease) in incremental paid loss development for each future calendar year could result in a 1.1% increase (decrease) in loss reserves.

Historically, the one-year change in the reserve estimate for this product line over the last nine years has varied from -31% to +3% (averaging -9%) for the Company and -8% to +10% (averaging -2%) for the industry overall. The Company’s year-to-year changes are driven by and are based on observed events during the year. We believe that the Company’s range of historical outcomes is illustrative of reasonably possible one-year changes in reserve estimates for this product line. Homeowners and personal lines other reserves represent approximately 3% of the Company’s total loss reserves.

While this line combines both liability and property coverages, the majority of the reserves relate to property. A source of reserve changes in the past has been updated information related to prior weather-related events.

The Company’s change in reserve estimate for this product line was +3% for 2005 and -31% for 2004. The 2005 change was driven by additional loss development from the 2004 Florida hurricanes. The 2004 change was driven by favorable loss development, particularly from fewer claim counts for accident year 2003.

102

International and other

International and other includes the international product line and other products not discussed above. The principal component of “other” is assumed reinsurance written on an excess-of-loss basis, which may include reinsurance of non-U.S. exposures, and is primarily runoff business.

International and other claim liabilities result from a mix of coverages, currencies and jurisdictions/countries. The common characteristic is the need to customize the analysis to the individual component, and the inability to rely on data characterizations and reporting requirements in the U.S. statutory reporting framework.

Due to changes in the business mix for this line over time, the recently incurred claim liabilities are relatively short term (due to both the products and the jurisdictions involved, e.g., the Republic of Ireland and the United Kingdom), while the older liabilities include some from runoff operations that are extremely long tail (e.g., U.S. excess liabilities reinsured through the London market, and several underwriting pools in runoff). The speed of claim reporting and claim settlement is a function of the specific

Changes from the July 13, 2006 Letter are in Bold and Italics

Attachment 1

coverage provided, the jurisdiction, the distribution system (e.g., underwriting pool versus direct), and the proximity of the insurance sale to the insured hazard (e.g., insured and insurer located in different countries). In particular, liabilities arising from the underwriting pools in runoff may result in significant reporting lags, settlement lags and claim complexity, due to the need to coordinate with other pool members or co-insurers through a broker or lead-insurer for claim settlement purposes.

International and other reserves are generally analyzed by program/pool, country and general coverage category (e.g., U.S. Liability—excess of loss reinsurance, or General Liability — Municipalities — by country.) The business is also generally split by direct versus assumed reinsurance for a given coverage/jurisdiction. Where the underlying insured hazard is outside the United States, the underlying coverages are generally similar to those described under the General Liability and Automobile discussion above, but under a different legal system. Where the underlying hazard is within the U.S., the coverage involved is typically that of General Liability, but on an excess or excess-of-loss reinsurance basis. Excess exposure requires the insured to “prove” not only claims under the policy, but also the prior payment of claims reaching up to the excess policy’s attachment point.

Examples of common risk factors that can change and, thus, affect the required International and other reserves (beyond those included in the general discussion section) include:

International and other risk factors

Changes in claim handling procedures, including those of the primary carriers

Changes in policy provisions or court interpretation of such provision

New theories of liability

Trends in jury awards

Changes in the propensity to sue

Changes in statutes of limitations

Changes in the underlying court system

Distortions from losses resulting from large single accounts or single issues

Changes in tort law

Changes in claim adjuster office structure (causing distortions in the data)

International and other book of business risk factors

Changes in policy provisions (e.g., deductibles, policy limits, endorsements, “claims made” language)

Changes in underwriting standards

Product mix (e.g., size of account, industries insured, jurisdiction mix)

Unanticipated changes in risk factors can affect reserves. As an indicator of the causal effect that a change in one or more risk factors could have on reserves for International and other (excluding asbestos and environmental), a 1% increase (decrease) in incremental paid loss development for each future calendar year could result in a 1.3%

Changes from the July 13, 2006 Letter are in Bold and Italics

Attachment 1

increase (decrease) in loss reserves. International and other reserves (excluding asbestos and environmental) represent approximately 9% of the Company’s total loss reserves.

International and other represents a combination of different product lines, some of which are in runoff. Comparative historical information is not available for international product lines as international operations do not file U.S. statutory reports. Comparative historical information on runoff business is not indicative of reasonably possible one- year changes in reserve estimate for this mix of runoff business. Accordingly, we have not included comparative analyses for International and other.

103

Glossary

Paid loss development method

An actuarial method to estimate ultimate losses for a given cohort of claims such as an accident year/product line component. If the paid-to-date losses are then subtracted from the estimated ultimate losses, the result is an indication of the unpaid losses.

The basic premise of the method is that cumulative paid losses for a given cohort of claims will grow in a stable, predictable pattern from year-to-year, based on the age of the cohort. These age-to-age growth factors are sometimes called “link ratios”.

For example, if cumulative paid losses for a product line XYZ for accident year 2004 were $100 as of 12/31/2004 (12 months after the start of that accident year), then grew to $120 as of 12/31/2005 (24 months after the start), the link ratio for that accident year from 12 to 24 months would be 1.20. If the link ratio for other recent accident years from 12 to 24 months for that product line were also at or around 1.20, then the method would assume a similar result for the most recent accident year, i.e., that it too would have its cumulative paid losses grow 120% from the 12 month to 24 month valuation.

This is repeated for each age-to-age period into the future until the age-to-age link ratios for future periods are assumed to be 1.0 (i.e., the age at which cumulative losses are assumed to have stopped growing).

A given accident year’s cumulative losses are then projected to ultimate by multiplying current cumulative losses by successive age-to-age link ratios up to that future age where growth is expected to end. For example, if growth is expected to end at 60

Changes from the July 13, 2006 Letter are in Bold and Italics

Attachment 1

months, then the ultimate indication for an accident year with cumulative losses at 12 months equals those losses times a 12 to 24 month link ratio, times a 24 to 36 month link ratio, times a 36 to 48 month link ratio, times a 48 to 60 month link ratio.

Advanced applications of the method include adjustments for changing conditions during the historical period and anticipated changes in the future.

Case incurred loss development method

The approach is the same as that described in this glossary under the “paid loss development method”, but based on the growth in cumulative case incurred losses (i.e., the sum of claim-adjustor incurred estimates for claims in the cohort) rather than paid losses. The basic premise of the method is that cumulative case incurred losses for a given cohort of claims will grow in a stable, predictable pattern from year-to-year, based on the age of the cohort.

Bornhuetter-Ferguson method

The basic premise of the method is that the historical ratio of additional claim activity to earned premium for a given product line component/age-to-age period is stable and predictable. It implicitly assumes that the actual activity to-date for past periods for that cohort is not a credible predictor of future activity for that cohort, or at least is not credible enough to override the “a priori” assumption as to future activity. It may be applied to either paid or case incurred claim data. It is used most often where the claim data is sparse and/or volatile and for relatively young cohorts with low volumes and/or data credibility.

In illustration, the method may assume that the ratio of additional paid losses from the 12 to 24 month period for an accident year is 10% of the original “a priori” expected losses for that accident year. The original “a priori” expected losses are typically based on the original loss ratio assumption for that accident year, with subsequent adjustment as facts develop.

The ultimate losses equal actual activity to-date plus the expected values for future periods.

Changes from the July 13, 2006 Letter are in Bold and Italics

Attachment 1

Average value method

This actuarial estimation method is used to estimate ultimate losses for a given cohort of claims such as an accident year/product line component. If the paid-to-date losses are then subtracted from the estimated ultimate losses, the result is an indication of the unpaid losses.

The basic premise of the method is that average claim values are stable and predictable over time for a particular cohort of claims. The method is utilized most often where ultimate claim counts are known or reliably estimable fairly early after the start of an accident year and average values are expected to be fairly predictable from one year to the next.

Ultimate losses under the method equal the known or estimated ultimate claim counts times the estimated average value. Estimated ultimate claim counts are frequently based on a claim count development method, essentially the same as the paid and case incurred development methods mentioned above but using claim count rather than claim dollar data. The average values can be based on historical trends from past closed claims, or backed into from estimated ultimate losses divided by estimated ultimate claim counts, or some other approach. When the average values are calculated from ultimate loss estimates, the resulting estimated averages may be supplemented with other data/analyses.

Reported claim development method

An actuarial method to estimate ultimate claim counts for a given cohort of claims such as an accident year/product line component. If the reported-to-date counts are then subtracted from the estimated ultimate counts, the result is an indication of the IBNR counts.

The approach is the same as that described in this glossary under the “paid loss development method”, but based on the growth in cumulative claim counts rather than paid losses. The basic premise of the method is that cumulative claim counts for a given cohort of claims will grow in a stable, predictable pattern from year-to-year, based on the age of the cohort.

S-curve method

A mathematical function which depicts an initial slow change, followed by a rapid change and then ending in a slow change again. This results in an “S” shaped line when depicted graphically. The actuarial application of these curves fit the reported data to-date for a particular cohort of claims to an S-curve to project future activity for that cohort.

Ground-up method

A method to estimate ultimate claim counts for a given cohort of claims such as an accident year/product line component. It involves analyzing the exposure at an individual insured level and then through the use of deterministic or stochastic

Changes from the July 13, 2006 Letter are in Bold and Italics

Attachment 1

scenarios and/or simulations, estimating the ultimate losses for those insured. The total losses for the cohort are then the sum of the losses for each individual insured.

Pro Forma – Note 9

Attachment 2

The fair value adjustments to the acquired claims and claim adjustment expense reserves and reinsurance recoverables as of April 1, 2004, the merger date, are reported as intangible assets and are being amortized over the expected payout period of the acquired reserves. See note 2.

Prior Year Reserve Development

In 2005, estimated claims and claim adjustment expenses for claims arising in prior years totaled a net $260 million, including $325 million of net unfavorable prior year reserve development impacting the Company’s results of operations which excludes $59 million of accretion of discount. In 2005, estimated claims and claim adjustment expenses for claims arising in prior years included $30 million of net favorable loss development on Commercial loss sensitive policies in various lines; however, since the business to which it relates was subject to premium adjustments, there was no impact on results of operations.

In the Commercial segment, net unfavorable prior year reserve development in 2005 was $694 million, including $830 million to strengthen asbestos reserves and $30 million to strengthen environmental reserves, which are discussed in more detail in the following “Asbestos and Environmental Reserves” section. The remaining net favorable prior year reserve development of $166 million was driven by lower ~~than expected~~ frequency and severity for both casualty and property-related lines of business in the Commercial segment, partially offset by reserve strengthening for assumed reinsurance, which is in runoff. Drivers of the reduction in both frequency and severity are attributable to increasingly favorable legal and judicial environments coupled with changes in policy provisions in both the property and casualty businesses. In addition, changes in underwriting and pricing criteria in prior years led to better results than initially expected and were recognized as claims emerged. Company initiatives relating to claims handling that impacted claims staffing and workflows have also contributed to the emergence of favorable severity experience.

The Specialty segment recorded net favorable prior year reserve development of $9 million in 2005, driven by lower ~~than expected~~ severity for property-related exposures that is attributable to a stronger than expected favorable impact of improved changes in policy provisions that resulted from better underwriting and pricing strategies in the prior years.

In the Personal segment, net favorable prior year reserve development in 2005 totaled $360 million~~, driven by lower than expected frequency and severity of losses i~~In the automobile line of business, the improvement was driven by changes in claims handling practices, which allowed case reserves to be set more accurately earlier in the claims settlement process and thus changed the historical loss development patterns. In addition, both industry and internal initiatives to fight fraud in several states caused a decrease in the total number of claims, as well as a change in the historical loss development patterns. In the homeowners line of business, the improvement was driven primarily by a significant decrease in the number of claims, attributable to changes in the marketplace. These changes also resulted in a change in the historical loss development patterns ~~as well as a decline in the frequency of non-catastrophe homeowners’ losses.~~

In 2004, estimated claims and claim adjustment expenses for claims arising in prior years totaled a net $2.40 billion, including $2.39 billion of net unfavorable prior year reserve development impacting the Company’s results of operations, excluding $75 million of accretion of discount. Net unfavorable prior year reserve development included $928 million to strengthen asbestos reserves, reserve adjustments related to the merger of $500 million for construction and $300 million for surety, $290 million to strengthen environmental reserves, $252 million related to a specific construction contractor, $113 million related to the commutation of agreements with a major reinsurer, and other net reserving actions.

Primarily all of the asbestos and environmental charges in 2004 were recorded in the Commercial segment and are discussed in more detail in the following “Asbestos and Environmental Reserves” section. Also included in net unfavorable prior year reserve development in the Commercial segment in 2004 was $38 million related to the commutation of agreements with a major reinsurer along with a strengthening of Gulf reserves, which was more than offset by other favorable prior year reserve development in core Commercial operations due to lower ~~than expected~~ frequency of

Pro Forma – Note 9

Attachment 2

non-catastrophe related losses attributable to better underwriting and pricing strategies in the prior year ~~than originally anticipated~~, the results of which emerged in the 2004 calendar year.

The Specialty segment recorded prior year reserve development of $1.43 billion in 2004, which included the charges to strengthen construction and surety reserves referred to previously, as well as the $252 million charge related to a specific construction contractor and $75 million related to the commutation of agreements with a major reinsurer. Results in 2004 also reflected $150 million of net unfavorable prior year reserve development recorded in TPC’s Construction operation prior to the merger, other reserve increases, and a charge to increase the allowance for estimated amounts due from a co-surety on a specific construction contractor claim.

In the Personal segment, net favorable prior year reserve development in 2004 was $378 million. In the homeowners line of business, the improvement was, driven by a significant decrease ~~lower than expected frequency of non-catastrophe homeowners’ losses~~in the number of claims, attributable to changes in the marketplace, including higher policy deductible levels and fewer small dollar claims. These changes resulted in a change in the historical loss development patterns. In addition, both industry and internal initiatives to fight fraud in several states caused a decrease in the total number of claims, as well as a change in the historical loss development patterns.~~, as well as a reduction in the frequency and severity of losses in the automobile line of business.~~

In 2003, net unfavorable prior year reserve development included in estimated claims and claim adjustment expenses totaled $390 million. That amount included $549 million of net unfavorable development impacting the Company’s results of operations that primarily resulted from $521 million of reserve strengthening at Gulf Insurance Company, a subsidiary that wrote specialty insurance prior to being placed in runoff in 2004. The net 2003 total also included unfavorable development related to American Equity, an operation that was placed in run-off in the second quarter of 2002, and environmental claims.

Those charges were partially offset by net favorable development in other Commercial businesses, principally property coverages, in which the Company experienced lower non-catastrophe-related claim frequency. In 2003, estimated claims and claim adjustment expenses for claims arising in prior years included $42 million of net favorable loss development on Commercial loss sensitive policies in various lines; however, since the business to which it relates was subject to premium adjustments, there was no impact on results of operations.

In addition, Personal recorded $162 million in net favorable prior year reserve development in 2003, principally due to changes in the marketplace. Homeowners were changing their policy provisions (increasing deductible levels) and not submitting small dollar claims in order to avoid rate changes. These changes resulted in a change in the historical loss development patterns. In addition, there was ~~lower than expected non-catastrophe claim frequency in both homeowners and non-bodily injury automobile businesses, and~~ a $50 million reduction in reserves held related to the terrorist attack on September 11, 2001.

For each of the years ended December 31, 2005, 2004 and 2003, changes in allocations between accident years of loss adjustment expenses, pursuant to regulatory reporting requirements, are included in claims and claim adjustment expenses for claims arising in prior years and did not impact results of operations.

Asbestos and Environmental Reserves

At December 31, 2005 and 2004, the Company’s claims and claim adjustment expense reserves included $4.79 billion and $4.57 billion, respectively, for asbestos and environmental-related claims, net of reinsurance.

It is difficult to estimate the reserves for asbestos and environmental-related claims due to the vagaries of court coverage decisions, plaintiffs’ expanded theories of liability, the risks inherent in complex litigation and other uncertainties, including without limitation, those which are set forth below.

Asbestos Reserves. Because each policyholder presents different liability and coverage issues, the Company generally reviews the asbestos exposure presented by each policyholder on a policyholder-by-policyholder basis. In the course of this review, the Company considers, among other factors: available insurance coverage, including the role of any umbrella or excess insurance the Company has issued to the policyholder; limits and deductibles; an analysis of each policyholder’s potential liability; the jurisdictions involved; past and anticipated future claim activity and loss development on pending claims; past settlement values of similar claims; allocated claim adjustment expense;

Pro Forma – Note 9

Attachment 2

potential role of other insurance; the role, if any, of non-asbestos claims or potential non-asbestos claims in any resolution process; and applicable coverage defenses or determinations, if any, including the determination as to whether or not an asbestos claim is a products/completed operation claim subject to an aggregate limit and the available coverage, if any, for that claim. When an estimate of the gross ultimate exposure for indemnity and related claim adjustment expense is determined for a policyholder, the Company calculates, by each policy year, a ceded reinsurance projection based on any applicable facultative and treaty reinsurance, as well as past ceded experience and reinsurance collections. Conventional actuarial methods are not utilized to establish asbestos reserves. The Company’s evaluations have not resulted in any data from which a meaningful average asbestos defense or indemnity payment may be determined.

With respect to asbestos exposures, the Company also compares its historical gross and net loss and expense paid experience, year-by-year, to assess any emerging trends, fluctuations, or characteristics suggested by the aggregate paid activity. Net asbestos losses and expenses paid in 2005 were $399 million, compared with $301 million in 2004. This increase was primarily driven by inclusion of the SPC business for the full year in 2005 and substantial reinsurance recoveries in the third quarter of 2004. Approximately 42% in 2005 and 22% in 2004 of total net paid losses relate to policyholders with whom the Company previously entered into settlement agreements limiting the Company’s liability. The lower percentage experienced in 2004 reflected substantial reinsurance billings during the year. In 2005, gross payments associated with policyholders with settlement agreements totaled $203 million, compared with $199 million in 2004.

The $830 million and $928 million asbestos reserve provisions in 2005 and 2004, respectively, were recorded pursuant to the completion of the Company’s annual ground-up asbestos liability review in the fourth quarter, which included an analysis of exposure and claim payment patterns by policyholder category, as well as recent settlements, policyholder bankruptcies, judicial rulings and legislative actions. The asbestos provision in 2005 resulted, in part, from higher than expected defense costs due to increased trial activity for seriously impaired plaintiffs and prolonged litigation before cases are settled or dismissed. The 2005 provision also considered the January 2006 court decision voiding, on procedural grounds, the previously rendered favorable arbitration decision in the ongoing ACandS litigation (described in more detail in note 17). The asbestos provision recorded in 2004 primarily resulted from an increase in litigation costs and activity surrounding peripheral defendants.

At December 31, 2005, asbestos reserves, net of reinsurance, were $4.36 billion, compared with $3.93 billion as of December 31, 2004. The Company acquired $311 million of net asbestos reserves in the merger with SPC in 2004.

Environmental Reserves. In establishing environmental reserves, the Company evaluates the exposure presented by each policyholder and the anticipated cost of resolution, if any. In the course of this analysis, the Company considers the probable liability, available coverage, relevant judicial interpretations and historical value of similar exposures. In addition, the Company considers the many variables presented, such as the nature of the alleged activities of the policyholder at each site; the allegations of environmental harm at each site; the number of sites; the total number of potentially responsible parties at each site; the nature of environmental harm and the corresponding remedy at each site; the nature of government enforcement activities at each site; the ownership and general use of each site; the overall nature of the insurance relationship between the Company and the policyholder, including the role of any umbrella or excess insurance the Company has issued to the policyholder; the involvement of other insurers; the potential for other available coverage, including the number of years of coverage; the role, if any, of non-environmental claims or potential non-environmental claims, in any resolution process; and the applicable law in each jurisdiction. Conventional actuarial techniques are not used to estimate these reserves. In 2005, the Company recorded a pretax charge of $30 million, net of reinsurance, primarily for declaratory judgment costs. In 2004, the Company recorded a pretax charge of $290 million, net of reinsurance, to increase environmental reserves due to revised estimates of costs related to recent settlement initiatives. Net environmental losses paid were $246 million, $152 million and $155 million for the years ended December 31, 2005, 2004 and 2003, respectively.

As a result of the processes and procedures described above, management believes that the reserves carried for asbestos and environmental claims at December 31, 2005 are appropriately established based upon known facts, current law and management’s judgment. However, the uncertainties surrounding the final resolution of these claims continue, and it is presently not possible to determine the ultimate exposure for asbestos and environmental claims and related litigation. As a result, these reserves are subject to revision as new information becomes available and as claims develop. The continuing uncertainties include, without limitation, the risks and lack of predictability inherent

Pro Forma – Note 9

Attachment 2

in complex litigation, any impact from the bankruptcy protection sought by various asbestos producers and other asbestos defendants, a further increase or decrease in asbestos and environmental claims which cannot now be anticipated, the role of any umbrella or excess policies the Company has issued, the resolution or adjudication of some disputes pertaining to the amount of available coverage for asbestos claims in a manner inconsistent with the Company’s previous assessment of these claims, the number and outcome of direct actions against the Company and future developments pertaining to the Company’s ability to recover reinsurance for asbestos and environmental claims. In addition, the Company’s asbestos-related loss and loss expense experience has been impacted by the exhaustion or unavailability due to insolvency of other insurance potentially available to policyholders along with the insolvency or bankruptcy of other defendants. It is also not possible to predict changes in the legal and legislative environment and their impact on the future development of asbestos and environmental claims. This development will be affected by future court decisions and interpretations, as well as changes in applicable legislation. It is also difficult to predict the ultimate outcome of large coverage disputes until settlement negotiations near completion and significant legal questions are resolved or, failing settlement, until the dispute is adjudicated. This is particularly the case with policyholders in bankruptcy where negotiations often involve a large number of claimants and other parties and require court approval to be effective. As part of its continuing analysis of asbestos reserves, which includes an annual ground-up review of asbestos policyholders, the Company continues to study the implications of these and other developments.

Asbestos and Environmental Reserves. Because of the uncertainties set forth above, additional liabilities may arise for amounts in excess of the current asbestos and environmental reserves. In addition, the Company’s estimate of ultimate claims and claim adjustment expenses may change. These additional liabilities or increases in estimates, or a range of either, cannot now be reasonably estimated and could result in income statement charges that could be material to the Company’s results of operations and financial condition in future periods.

Re:	The St. Paul Travelers Companies, Inc.
	Form 10-K for the Fiscal Year Ended
	December 31, 2005
	File No. 001-10898

	Sincerely,

	Jay S. Benet
	Vice Chairman and
	Chief Financial Officer

(at and for the year ended December 31, in millions)	2005			2004		2003
Claims and claim adjustment expense reserves at beginning of year	$	58,984		$	34,474	$	33,628
Less reinsurance recoverables on unpaid losses	17,538			10,419		10,360
Net balance at beginning of year	41,446			24,055		23,268
Provision for claims and claim adjustment expenses for claims arising in the current year	14,450			12,855		8,554
Estimated claims and claim adjustment expenses for claims arising in prior years	260			2,399		390
Acquisition(1)	—			13,653		—
Total increases	14,710			28,907		8,944

Claims and claim adjustment expense payments for claims arising in:
Current year	4,227			3,998		2,987
Prior years	8,871			7,553		5,170
Total payments	13,098			11,551		8,157
Unrealized foreign exchange (gain) loss	(163		)	35		—
Net balance at end of year	42,895			41,446		24,055
Plus reinsurance recoverables on unpaid losses(1)	18,112			17,538		10,419
Claims and claim adjustment expense reserves at end of year	$	61,007		$	58,984	$	34,474

(at December 31, in millions)	2005		2004
Property-casualty	$	61,007	$	58,984
Accident and health	83		86
Total	$	61,090	$	59,070