PART I

                In the ordinary course of business, we rely on other insurance companies to share risks through reinsurance. A large portion of the reinsurance is put into effect under contracts known as treaties and, in some instances, by negotiation on each individual risk (known as facultative reinsurance). We have quota share, excess of loss and catastrophe reinsurance contracts that protect against losses over stipulated amounts arising from any one occurrence or event. The arrangements allow us to pursue greater diversification of business and serve to limit the maximum net loss on catastrophes and large risks. Reinsurance is subject to certain risks, specifically market risk, which affects the cost of and the ability to secure these contracts, and credit risk, which is the risk that our reinsurers may not pay on losses in a timely fashion or at all. The following table illustrates, through premium volume, the degree to which we have utilized reinsurance during the past three years. For an expanded discussion of the impact of reinsurance on our operations, see Note 5 to our audited consolidated financial statements included in our 2008 Financial Report to Shareholders, attached as Exhibit 13 and incorporated by reference herein.

Specialty Insurance Market Overview

                The specialty insurance market differs significantly from the standard market. In the standard market, insurance rates and forms are highly regulated, products and coverage are largely uniform with relatively predictable exposures, and companies tend to compete for customers on the basis of price. In contrast, the specialty market provides coverage for risks that do not fit the underwriting criteria of the standard carriers. Competition tends to focus less on price and more on availability, service and other value-based considerations. While specialty market exposures may have higher insurance risks than their standard market counterparts, we manage these risks to achieve higher financial returns. To reach our financial and operational goals, we must have extensive knowledge and expertise in our markets. Most of our risks are considered on an individual basis and restricted limits, deductibles, exclusions and surcharges are employed in order to respond to distinctive risk characteristics.

                We operate in the excess and surplus insurance market and the specialty admitted insurance market.

Excess and Surplus Insurance Market

                The excess and surplus market focuses on hard-to-place risks. Excess and surplus eligibility allows us to underwrite nonstandard market risks with more flexible policy forms and unregulated premium rates. This typically results in coverages that are more restrictive and more expensive than in the standard admitted market. The excess and surplus lines regulatory environment and production model also effectively filters submission flow and matches market opportunities to

3

our expertise and appetite.  In 2008, the excess and surplus market represented approximately $27 billion, or 5 percent, of the entire $504 billion domestic property and casualty industry, as measured by direct premiums written. Our excess and surplus operation wrote gross premiums of $321.6 million, or 47 percent, of our total gross premiums written.

Specialty Admitted Insurance Market

                We also write business in the specialty admitted market. Most of these risks are unique and hard to place in the standard market, but for marketing and regulatory reasons, they must remain with an admitted insurance company. The specialty admitted market is subject to greater state regulation than the excess and surplus market, particularly with regard to rate and form filing requirements, restrictions on the ability to exit lines of business, premium tax payments and membership in various state associations, such as state guaranty funds and assigned risk plans. For 2008, our specialty admitted operations wrote gross premiums of $359.6 million representing approximately 53 percent of our total gross premiums written for the year.

Business Segment Overview

                Our segment data is derived using the guidance set forth in Statement of Financial Accounting Standards (SFAS) 131, “Disclosures about Segments of an Enterprise and Related Information.”  As prescribed by the pronouncement, reporting is based on the internal structure and reporting of information as it is used by management.  The segments of our insurance operations are casualty, property and surety.  For additional information, see Note 11 to our audited consolidated financial statements included in our 2008 Financial Report to Shareholders, attached as Exhibit 13 and incorporated by reference herein.

Casualty Segment

General Liability

                Our general liability business consists primarily of coverage for third party liability of commercial insureds including manufacturers, contractors, apartments and mercantile. Net premiums earned from this business totaled $140.9 million, $167.9 million and $180.0 million, or 25 percent, 26 percent, and 28 percent of consolidated revenues for 2008, 2007, and 2006, respectively.

Commercial and Personal Umbrella Liability

                Our commercial umbrella coverage is principally written in excess of primary liability insurance provided by other carriers and, to a modest degree, in excess of primary liability written by us. The personal umbrella coverage is written in excess of the homeowners and automobile liability coverage provided by other carriers, except in Hawaii, where some underlying homeowners’ coverage is written by us. Net premiums earned from this business totaled $65.1 million, $66.3 million and $64.7 million, or 12 percent, 10 percent, and 10 percent of consolidated revenues for 2008, 2007, and 2006, respectively.

Commercial Transportation

                Our transportation insurance facility in Atlanta provides automobile liability and physical damage insurance to local, intermediate and long haul truckers, public transportation risks and equipment dealers. In early 2005, we expanded our focus to include other types of commercial automobile risks. We also offer incidental, related insurance coverages, including general liability, commercial umbrella and excess liability, and motor truck cargo. The facility is staffed by highly experienced transportation underwriters who produce business through independent agents and brokers nationwide. Net premiums earned from this business totaled $46.7 million, $49.1 million, and $48.3 million, or 8 percent of consolidated revenues for 2008, 2007, and 2006.

Executive Products

                We provide a variety of professional liability coverages, such as directors’ and officers’ (D&O) liability insurance, employment practices liability and other miscellaneous professional liability coverages, for a variety of low to moderate classes of risks. We tend to focus on smaller accounts, avoiding the large account sector which is generally more sensitive to price competition.  Our target accounts include publicly traded companies with market capitalization below $5 billion (where we are writing part of the traditional D&O program), Clause 1 (also known as “Side A” coverage, which is direct liability coverage for the individual directors and officers), private companies, nonprofit organizations, and sole-sponsored and multi-employer fiduciary liability accounts.  We successfully transitioned from primarily writing high layers of excess D&O for publicly traded companies to writing more Clause 1 coverage.  Additionally, we are having success rounding out our portfolio by writing more fiduciary liability coverage, primary and excess D&O coverage for private companies and non-profit organizations.  Net

4

premiums earned from this business totaled $13.8 million, $12.0 million, and $13.0 million, or 2 percent of consolidated revenues for 2008, 2007, and 2006.

Specialty Program Business

                We offer program business in a variety of areas, which are typically multiple coverages combined into a package or portfolio policy.  Our program coverages include: commercial property, general liability, inland marine, and crime. We rely primarily on program administrators as sources for this business.    Net premiums earned from this business totaled $31.4 million, $29.4 million, and $25.5 million for 2008, 2007, and 2006, respectively. These amounts represent 6 percent, 5 percent, and 4 percent of consolidated revenues for 2008, 2007, and 2006, respectively.

Other

                We offer a variety of other smaller programs in our casualty segment, including deductible buy-back, at-home business, and employer’s excess indemnity. Net premiums earned from these lines totaled $15.5 million, $18.7 million, and $16.6 million, or 3 percent of consolidated revenues for 2008, 2007, and 2006.

Property Segment

Commercial

                Our commercial property coverage consists primarily of excess and surplus lines and specialty insurance such as fire, earthquake and “difference in conditions,” which can include earthquake, wind, flood and collapse coverages, and inland marine.  We provide insurance for a wide range of commercial and industrial risks, such as office buildings, apartments, condominiums, and certain industrial and mercantile structures. We also write boiler and machinery coverage under the same management as commercial property.  Net premiums earned from commercial property business totaled $85.3 million, $92.6 million, and $91.5 million, or 15 percent, 14 percent, and 14 percent of consolidated revenues for 2008, 2007, and 2006, respectively.

Marine

                In 2005, we launched a new marine insurance division.  Marine coverages include hull, cargo and protection and indemnity (P&I), as well as inland marine coverages including builders’ risks, contractors’ equipment and other “floater” type coverages.  In May 2007, the marine division added specialty cargo coverage that focuses on high-tech and life sciences risks.  In March 2008, the marine division added a yacht program.  In 2008, 2007 and 2006, marine net premiums earned totaled $48.2 million, $32.9 million and $16.8 million, or 9 percent, 5 percent and 3 percent, respectively, of consolidated revenues.

Other

                We offer a variety of other smaller programs in our property segment, including a limited amount of homeowners and dwelling fire insurance in Hawaii.  We have reduced our Hawaii wind exposure through more restrictive underwriting over the last 18 months.

                In July 2007, we launched a new division focusing on facultative reinsurance.  The division is responsible for underwriting property facultative reinsurance for insurance companies utilizing reinsurance intermediaries.

                Net premiums earned from the above coverages totaled $13.4 million, $12.9 million, and $14.3 million, or 2 percent of consolidated revenues for 2008, 2007, and 2006.

Surety Segment

                Our surety segment specializes in writing small-to-large commercial and small contract surety coverages, as well as those for the energy (plugging and abandonment of oil wells), petrochemical, and refining industries. We offer miscellaneous bonds, including license and permit, notary, and court bonds.  In September 2008, we launched a new Fidelity Division focusing on fidelity and crime coverage for commercial insureds and select financial institutions.  These bonds are written through independent agencies as well as regional and national brokers. Net earned premium totaled $68.4 million, $62.7 million, and $59.5 million, or 12 percent, 10 percent and 9 percent of consolidated revenues for 2008, 2007, and 2006, respectively.

5

Competition

                Our specialty property and casualty insurance subsidiaries are part of an extremely competitive industry that is cyclical and historically characterized by periods of high premium rates and shortages of underwriting capacity followed by periods of severe competition and excess underwriting capacity. Within the United States alone, approximately 2,400 companies, both stock and mutual, actively market property and casualty coverages. Our primary competitors in our casualty segment are, among others, Ace, Arch, James River, Landmark, Navigators, USLI, Great West, Lancer, National Interstate, Chubb, Philadelphia, Great American, Travelers and CNA. Our primary competitors in our property segment are, among others, Ace, Lexington, Arch, Crum & Forster, Travelers and Markel. Our primary competitors in our surety segment are, among others, Ace, Arch, HCC, CNA, Safeco, North American Specialty, Travelers and Hartford. The combination of coverages, service, pricing and other methods of competition vary from line to line. Our principal methods of meeting this competition are innovative coverages, marketing structure and quality service to the agents and policyholders at a fair price. We compete favorably in part because of our sound financial base and reputation, as well as our broad geographic penetration into all 50 states, the District of Columbia, Puerto Rico, the Virgin Islands and Guam. In the property and casualty area, we have acquired experienced underwriting specialists in our branch and home offices. We have continued to maintain our underwriting and marketing standards by not seeking market share at the expense of earnings. We have a track record of withdrawing from markets when conditions become overly adverse. We offer new coverages and new programs where the opportunity exists to provide needed insurance coverage with exceptional service on a profitable basis.

Ratings

A.M. Best ratings for the industry range from ‘‘A++’’ (Superior) to ‘‘F’’ (In Liquidation) with some companies not being rated. Standard & Poor’s ratings for the industry range from ‘‘AAA’’ (Extremely strong) to ‘‘R’’ (Regulatory Action). Moody’s ratings for the industry range from “Aaa” (Exceptional) to “C” (Lowest).  The following table illustrates the range of ratings assigned by each of the three major rating companies that has issued a financial strength rating on our insurance companies:

Publications of A.M. Best, Standard & Poor’s and Moody’s indicate that ‘‘A’’ and ‘‘A+’’ ratings are assigned to those companies that, in their opinion, have achieved excellent overall performance when compared to the standards established by these firms and have a strong ability to meet their obligations to policyholders over a long period of time. In evaluating a company’s financial and operating performance, each of the firms reviews the company’s profitability, leverage and liquidity, as well as the company’s spread of risk, the quality and appropriateness of its reinsurance, the quality and diversification of its assets, the adequacy of its policy and loss reserves, the adequacy of its surplus, its capital structure, its risk management practices and the experience and objectives of its management. These ratings are based on factors relevant to policyholders, agents, insurance brokers and intermediaries and are not directed to the protection of investors.

6

At December 31, 2008, the following ratings were assigned to our insurance companies:

For A.M Best, Standard & Poor’s and Moody’s, the financial strength ratings represented above are affirmations of previously assigned ratings.  A.M. Best, in addition to assigning a financial strength rating, also assigns financial size categories.  During 2008, RLI Insurance Company, Mt. Hawley Insurance Company and RLI Indemnity Company, collectively referred to as RLI Group, were assigned a financial size category of “X” (adjusted policyholders’ surplus of between $500 and $750 million).  As of December 31, 2008, the policyholders’ statutory surplus of RLI Group totaled $678.0 million.

RLI Corp’s existing $100 million of senior notes maturing in 2014 maintains a Standard & Poor’s rating of  “BBB+”, Moody’s “Baa2”, and a Fitch rating of “BBB.”

Reinsurance

                We reinsure a portion of our property and casualty insurance exposure, paying or ceding to the reinsurer a portion of the premiums received on such policies. Earned premiums ceded to non-affiliated reinsurers totaled $172.3 million, $227.1 million, and $264.5 million in 2008, 2007, and 2006, respectively. Insurance is ceded principally to reduce net liability on individual risks and to protect against catastrophic losses. While reinsurance does not relieve us of our legal liability to our policyholders, we use reinsurance as an alternative to using our own capital to fund losses.  Retention levels are adjusted each year to maintain a balance between the growth in surplus and the cost of reinsurance. Although reinsurance does not legally discharge an insurer from its primary liability for the full amount of the policies, it does make the assuming reinsurer liable to the insurer to the extent of the insurance ceded.

Reinsurance is subject to certain risks, specifically market risk (which affects the cost of and the ability to secure reinsurance contracts) and credit risk (which relates to the ability to collect from the reinsurer on our claims). We purchase reinsurance from a number of financially strong reinsurers. We evaluate reinsurers’ ability to pay based on their financial results, level of surplus, financial strength ratings, and other risk characteristics.  A reinsurance committee, comprised of senior management, approves our security guidelines and reinsurer usage.  Each of the top 10 largest reinsurers (listed below and ranked based on amounts recoverable) are rated “A-” or better by A.M. Best and Standard and Poor’s rating services. Additionally, more than 91 percent of our reinsurance recoverables are due from companies rated “A” or better by A.M. Best and Standard & Poor’s rating services.

The following table sets forth the ten largest reinsurers in terms of amounts recoverable, net of collateral we are holding from such reinsurers, as of December 31, 2008. Also shown are the amounts of written premium ceded to these reinsurers during the calendar year 2008.

7

We utilize both treaty and facultative reinsurance coverage for our risks. Treaty coverage refers to a reinsurance contract that is applied to a group or class of business where all the risks written meet the criteria for that class.  Facultative coverage is applied to individual risks as opposed to a group or class of business. It is used for a variety of reasons including supplementing the limits provided by the treaty coverage or covering risks or perils excluded from treaty reinsurance.

                Much of our reinsurance is purchased on an excess of loss basis. Under an excess of loss arrangement, we retain losses on a risk up to a specified amount and the reinsurers assume any losses above that amount. We may choose to participate in the reinsurance layers purchased by retaining a percentage of the layer.  It is common to find conditions in excess of loss covers such as occurrence limits, aggregate limits and reinstatement premium charges. Occurrence limits cap our recovery for multiple losses caused by the same event.  Aggregate limits cap our recovery for all losses ceded during the contract term.  We may be required to pay additional premium to reinstate or have access to use the reinsurance limits for potential future recoveries during the same contract year.  Our property and surety treaties tend to include reinstatement provisions.

We analyze our reinsurance covers in conjunction with our three segments: property, casualty and surety.

                In the property segment, the reinsurance structure is divided into three categories: commercial property, catastrophe earthquake and catastrophe other than earthquake, which could include such events as hurricanes, windstorm, hailstorms, explosions, severe winter weather, fires, etc.

Commercial Property Reinsurance

Our commercial property treaty renews annually on January 1.  In 2009, 2008, 2007, and 2006, for most risks, we retain the first $1.0 million in losses.  We purchase treaty reinsurance for the next $9.0 million in limit.  We retain $600,000 within the treaty.  In 2006, we retained $1.0 million within the treaty.

The marine treaty renews May 1 annually.  For marine exposures, we retain the first $1.0 million in losses.  Reinsurance covers 100 percent of the next $39.0 million.  For the 2007 treaty, we retained the first $1.0 million in losses and purchased coverage for 100 percent of the next $29.0 million.  For the 2006 treaty, we retained the first $0.5 million in losses.  We purchased $19.5 million of reinsurance and retained $0.4 million within the treaty structure.

Property Reinsurance- Catastrophe (CAT) Coverage

                Our property catastrophe reinsurance reduces the financial impact a catastrophe could have on our property segment.  Catastrophes involve multiple claims and policyholders.  Reinsurance limits purchased fluctuate due to changes in the number of policies we insure, reinsurance costs, insurance company surplus levels, and our risk appetite.  In addition, we monitor the expected rate of return for each of our catastrophe lines of business.  At high rates of return, we grow the book of business and may purchase additional reinsurance depending on our capital position.  As the rate of return decreases, we shrink the book and may purchase less reinsurance.  For the last three years, we have maintained a $25.0 million first dollar retention.  For 2009, we purchased $350.0 million in limit above our retention.  We purchased $375.0 million in limit in 2008 which decreased to $350.0 million at July 1, 2008.  In 2007, we purchased $475.0 million in catastrophe reinsurance which decreased to $425.0 million at July 1, 2007.  These CAT limits are in addition to the per-occurrence coverage provided by facultative and other treaty coverages.  We have participated in the catastrophe layers purchased by retaining a percentage of each layer throughout this period.  Our participation has varied based on price and the amount of risk transferred by each layer.

8

                Our property catastrophe program continues to be on an excess of loss basis.   It attaches after all other reinsurance has been considered.   Although covered in one program, limits and attachment points differ for California earthquakes and all other perils. The following charts use information from our catastrophe modeling software to illustrate our net retention resulting from particular events that would generate the listed levels of gross losses:

                These tables were generated using theoretical probabilities of events occurring in areas where our portfolio of currently in-force policies could generate the level of loss shown. Actual results could vary significantly from these tables as the actual nature or severity of a particular event cannot be predicted with any reasonable degree of accuracy.  Reinsurance limits are purchased based on the anticipated losses to large events.  The largest losses shown above are unlikely to occur based on the probability of those events occurring.  However, there is a remote chance that a larger event could occur.  If the actual event losses are larger than anticipated, we could retain additional losses above the limit of our catastrophe reinsurance.

                Our catastrophe program includes one prepaid reinstatement for the first two layers of coverage, up to $100 million, for a catastrophe other than California earthquake.  If a loss does occur, reinstatement must be purchased for the remaining limits.  For a California earthquake, there is a prepaid reinstatement for the $50.0 million excess $50.0 million layer (placed at 77 percent) and a reinstatement must be purchased for the remaining limits.

                We continuously monitor and quantify our exposure to catastrophes, including earthquakes, hurricanes, terrorist acts, and other catastrophic events.  In the normal course of business, we manage our concentrations of exposures to catastrophic events, primarily by limiting concentrations of exposure to acceptable levels and by purchasing reinsurance.  Exposure and coverage detail is recorded for each risk location.  We quantify and monitor the total policy limit insured in each region.  In addition, we use third party catastrophe exposure models and an internally developed analysis to assess each risk and ensure we include an appropriate charge for assumed catastrophe risks.  Catastrophe exposure modeling is inherently uncertain due to the model’s reliance on an infrequent observation of actual events and exposure data, increasing the importance of capturing accurate policy coverage data.  The model results are used both in the underwriting analysis of individual risks, and at a corporate level for the aggregate book of catastrophe-exposed business. From both perspectives, we consider the potential loss produced by individual events that represent moderate-to-high loss potential at varying return periods and magnitudes. In calculating potential losses, we select appropriate assumptions, including but not limited to loss amplification and storm surge.  We establish risk tolerances at the portfolio level based on market conditions, the level of reinsurance available, changes to the assumptions in the catastrophe models, rating agency capital constraints, underwriting guidelines and coverages, and internal preferences.  Our risk tolerances for each type of catastrophe, and for all perils in aggregate, change over time as these internal and external conditions change.

Casualty Reinsurance

                Our 2009 casualty reinsurance includes both excess of loss treaties and quota share treaties, as was the case in 2008, and 2007.  Annually on January 1, we place a Combined Casualty Treaty that incorporates coverage for a majority of our casualty lines.  In 2009, we added transportation to the treaty effective May 1, 2009.  Our first dollar retention on those lines of

9

business covered ranges from $0.5 million to $1.0 million.  Considering our participation in the reinsurance layers placed, our retention on a full limits loss ranges from $0.5 million to $1.5 million.  Maximum reinsurance limits purchased are $10.0 million inclusive of our retention for the last three years.  With respect to our 2008 Combined Casualty Treaty, we retained the initial $0.5 million to $1.0 million in loss.  Our total retention considering participation in the reinsurance layers purchased ranged from $1.3 million to $1.4 million depending on the type of policy.  This was also the case in 2007.  For our transportation coverage, the separate treaty renews on May 1 annually.  In 2008, we retained the first $0.5 million in loss.  We purchased between $1.5 million and $4.5 million of reinsurance limit depending on the type of risk for the last three years.  Our total retention inclusive of treaty participation ranged from $0.6 million and $1.0 million.  We also retained the first $0.5 million in loss in our 2007 treaty.  Our total retention inclusive of treaty participation ranged from $0.6 million and $0.9 million.  For the 2006 treaty, our first dollar retention was $0.5 million and we placed 100 percent of the reinsurance limits purchased.

Our executive products group (EPG) treaty renews on July 1 annually.  We have purchased quota share treaties over the last three years and have varied our percentage of participation.  In 2008, our reinsurance limit increased to $25.0 million.  Our maximum retained loss on a policy does not exceed $7.5 million.  In 2007 and 2006, we purchased reinsurance limits up to $20.0 million.  In 2007 and 2006, our maximum retained loss on any EPG policy was $6.0 million and $4.0 million, respectively.

Surety Reinsurance

                Our surety reinsurance treaty renews on April 1 annually.  The treaty is on an excess of loss basis.  In 2008, we purchased reinsurance limits of $49.0 million above our $1.0 million first dollar retention.  Within the reinsurance treaty, we participate up to a maximum retention of $6.3 million for any one principal.  We purchased $34.0 million of reinsurance limit in 2007 above our first dollar retention of $1.0 million.  Our maximum retention considering treaty participation was $8.0 million.  In 2006, we purchased $24.0 million in limit with a $7.0 million maximum retention inclusive our first dollar retention of $1.0 million.  For a majority of risks, our potential net loss did not exceed $2.0 million.

                During 2008, we purchased a separate quota share treaty for our new fidelity product line.  We retain up to $3.8 million in losses with a limit of $25.0 million.  This treaty has a renewal date of March 1, 2010.

Marketing and Distribution

                We distribute our coverages primarily through branch offices throughout the country that market to wholesale and retail brokers and through independent agents.  We also market through agencies and more recently through e-commerce channels.

                                                Broker Business

                The largest volume of broker-generated premium is in our commercial property, general liability, commercial surety, commercial umbrella and commercial automobile coverages. This business is produced through wholesale and retail brokers who are not affiliated with us.

                                                Independent Agent Business

                Our surety segment offers its business through a variety of independent agents. Additionally, we write program business, such as at-home business and personal umbrella, through independent agents. Homeowners and dwelling fire is produced through independent agents in Hawaii. Each of these programs involves detailed eligibility criteria, which are incorporated into strict underwriting guidelines, and prequalification of each risk using a system accessible by the independent agent. The independent agent cannot bind the risk unless they receive approval through our system.

                Underwriting Agents

                We contract with certain underwriting agencies who have limited authority to bind or underwrite business on our behalf.  The underwriting agreements involve strict underwriting guidelines and the agents are subject to audits upon request.  These agencies may receive some compensation through contingent profit commission.

                                                E-commerce

                We are actively employing e-commerce to produce and efficiently process and service business, including package policies for limited service motel/hotel operations, restaurant/bar/tavern operations and at-home businesses, small commercial and personal umbrella risks and surety bonding.

10

Environmental, Asbestos, and Mass Tort Exposures

                We are subject to environmental site cleanup, asbestos removal, and mass tort claims and exposures through our commercial umbrella, general liability, and discontinued assumed casualty reinsurance lines of business. The majority of the exposure is in the excess layers of our commercial umbrella and assumed reinsurance books of business.

                The following table represents inception-to-date paid and unpaid environmental, asbestos, and mass tort claims data (including incurred but not reported losses) as of December 31, 2008, 2007, and 2006:

              Our environmental, asbestos and mass tort exposure is limited, relative to other insurers, as a result of entering the affected liability lines after the insurance industry had already recognized environmental and asbestos exposure as a problem and adopted appropriate coverage exclusions.  During 2008, payment activity was more than we experienced in 2007, which was not surprising given the increased amount of case reserve activity during 2007.  However, we did not observe a corresponding decrease in case reserves, as gross case reserves decreased by only $4.8 million while net case reserves increased by about $0.5 million.  Much of the reported adverse development can be tied to a small number of claims.  Two related asbestos claims from our assumed reinsurance book accounted for $2.7 million in gross emergence and $2.1 million in net emergence with large payments being made.  A mass tort claim from accident year 2005 against an insured hotel involving carbon monoxide discharge accounted for another $1.6 million direct and $1.0 million net.  In light of these developments and a few additional reported claims, we decided to strengthen our incurred but not reported (IBNR) position.

                During 2007, while payment activity was less than we experienced in 2006, we experienced an unusual amount of case reserve activity. About two-thirds of the increase in net reserves was from case reserve adjustments. A $4.8 million case reserve increase was posted due to a reversal of a previous court decision for a claim involving pollution during the late 1980s. In addition, an insurance company we reinsured in the early 1980s and that went into liquidation in 1986, reported a number of claims in 2007. The largest of these involves asbestos. Because of this situation, we posted total net case reserves of $2.9 million. Also, a $2.2 million reserve adjustment was made on a 1983 asbestos related claim because the coverage layers below our excess policy had been exhausted. In addition to these case reserve increases, we made a minor adjustment in our emergence pattern assumptions that resulted in an increase in our total reserve position relative to industry benchmarks.

                While our environmental exposure is limited, the ultimate liability for this exposure is difficult to assess because of the extensive and complicated litigation involved in the settlement of claims and evolving legislation on such issues as joint and several liability, retroactive liability, and standards of cleanup. Additionally, we participate primarily in the excess layers of coverage, where accurate estimates of ultimate loss are more difficult to derive than for primary coverage.

Losses and Settlement Expenses

Overview

Loss and loss adjustment expense (LAE) reserves represent our best estimate of ultimate amounts for losses and related settlement expenses from claims that have been reported but not paid, and those losses that have occurred but have not yet been reported to us. Loss reserves do not represent an exact calculation of liability, but instead represent our estimates, generally utilizing individual claim estimates and actuarial expertise and estimation techniques at a given accounting date. The loss reserve estimates are expectations of what ultimate settlement and administration of claims will cost upon final resolution. These estimates are based on facts and circumstances then known to us, review of historical settlement patterns, estimates of

11

trends in claims frequency and severity, projections of loss costs, expected interpretations of legal theories of liability, and many other factors. In establishing reserves, we also take into account estimated recoveries, reinsurance, salvage, and subrogation. The reserves are reviewed regularly by a team of actuaries we employ.

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, claim personnel, economic inflation, legal trends, and legislative changes, among others. The impact of many of these items on ultimate costs for loss and LAE is difficult to estimate. Loss reserve estimations also differ significantly by coverage due to differences in claim complexity, the volume of claims, the policy limits written, the terms and conditions of the underlying policies, 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. We continually refine our loss reserve estimates as historical loss experience develops and additional claims are reported and settled. We rigorously attempt to consider all significant facts and circumstances known at the time loss reserves are established.

Due to inherent uncertainty underlying loss reserve estimates, including but not limited to the future settlement environment, final resolution of the estimated liability may 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 and unfavorable.

The amount by which estimated losses differ from those originally reported for a period is known as “development.” Development is unfavorable when the losses ultimately settle for more than the levels at which they were reserved or subsequent estimates indicate a basis for reserve increases on unresolved claims. Development is favorable when losses ultimately settle for less than the amount reserved or subsequent estimates indicate a basis for reducing loss reserves on unresolved claims. We reflect favorable or unfavorable developments of loss reserves in the results of operations in the period the estimates are changed.

We record two categories of loss and LAE reserves — case-specific reserves and IBNR reserves.

Within a reasonable period of time after a claim is reported, our claim department completes an initial investigation and establishes a case reserve. This case-specific reserve is an estimate of the ultimate amount we will have to pay for the claim, including related legal expenses and other costs associated with resolving and settling a particular claim. The estimate reflects all of the current information available regarding the claim, the informed judgment of our professional claim personnel, our reserving practices and experience, and the knowledge of such personnel regarding the nature and value of the specific type of claim. During the life cycle of a particular claim, more information may materialize that causes us to revise the estimate of the ultimate value of the claim either upward or downward. We may determine that it is appropriate to pay portions of the reserve to the claimant or related settlement expenses before final resolution of the claim. The amount of the individual claim reserve will be adjusted accordingly and is based on the most recent information available.

We establish IBNR reserves to estimate the amount we will have to pay for claims that have occurred, but have not yet been reported to us; claims that have been reported to us that may ultimately be paid out differently than expected by our case-specific reserves; and claims that have been paid and closed, but may reopen and require future payment.

Our IBNR reserving process involves three steps including an initial IBNR generation process that is prospective in nature; a loss and LAE reserve estimation process that occurs retrospectively; and a subsequent discussion and reconciliation between our prospective and retrospective IBNR estimates which includes changes in our provisions for IBNR where deemed appropriate. These three processes are discussed in more detail in the following sections.

LAE represents the cost involved in adjusting and administering losses from policies we issued. The LAE reserves are frequently separated into two components: allocated and unallocated. Allocated loss adjustment expense (ALAE) reserves represent an estimate of claims settlement expenses that can be identified with a specific claim or case. Examples of ALAE would be the hiring of an outside adjuster to investigate a claim or an outside attorney to defend our insured. The claims professional typically estimates this cost separately from the loss component in the case reserve. Unallocated loss adjustment expense (ULAE) reserves represent an estimate of claims settlement expenses that cannot be identified with a specific claim. An example of ULAE would be the cost of an internal claims examiner to manage or investigate a reported claim.

                All decisions regarding our best estimate of ultimate loss and LAE reserves are made by our Loss Reserve Committee (LRC). The LRC is made up of various members of the management team including the chief executive officer, chief operating officer, chief financial officer, chief actuary, general counsel and other selected executives. We do not use discounting (recognition of the time value of money) in reporting our estimated reserves for losses and settlement expenses. Based on current assumptions used in calculating reserves, we believe that our overall reserve levels at December 31, 2008, make a reasonable provision to meet our future obligations.

12

Net loss and loss adjustment reserves by product line at year-end 2008 and 2007 were as follows:

Initial IBNR Generation Process

                Initial carried IBNR reserves are determined through a reserve generation process. The intent of this process is to establish an initial total reserve that will provide a reasonable provision for the ultimate value of all unpaid loss and ALAE liabilities. For most casualty and surety products, this process involves the use of an initial loss and ALAE ratio that is applied to the earned premium for a given period. The result is our best initial estimate of the expected amount of ultimate loss and ALAE for the period by product. Paid and case reserves are subtracted from this initial estimate of ultimate loss and ALAE to determine a carried IBNR reserve.

                For most property products, we use an alternative method of determining an appropriate provision for initial IBNR. Since this segment is characterized by a shorter period of time between claim occurrence and claim settlement, the IBNR reserve is determined by an initial loss percentage applied to the rolling 12 month’s premium earned. No deductions for paid or case reserves are made. This alternative method of determining initial IBNR reacts more quickly to the actual loss emergence and is more appropriate for our property products where final claim resolution occurs quickly.

                We do not reserve for natural or man-made catastrophes until an event has occurred.  Shortly after such occurrence, we review insured locations exposed to the event, model loss estimates of the event based on our own exposures, and industry loss estimates of the event, and we consider our knowledge of frequency and severity from early claim reports to determine an appropriate reserve for the catastrophe.  These reserves are reviewed frequently based on loss reports and appropriate changes to our estimates are made to reflect any new information as it arises.

The initial loss and ALAE ratios that are applied to earned premium are reviewed at least semi-annually. Prospective estimates are made based on historical loss experience adjusted for mix and price change and loss cost inflation. The initial loss and ALAE ratios also reflect some provision for estimation risk. We consider estimation risk by segment and product line. A segment with greater overall volatility and uncertainty has greater estimation risk. Characteristics of segments and products with higher estimation risk, include those exhibiting, but not limited to, the following characteristics:

·           significant changes in underlying policy terms and conditions,

·           a new business or one experiencing significant growth and/or high turnover,

·           small volume or lacking internal data requiring significant reliance on external data,

·           longer emergence patterns with exposures to latent unforeseen mass tort,

·           high severity and/or low frequency,

13

·           operational processes undergoing significant change, and/or

·           high sensitivity to significant swings in loss trends or economic change.

                Following is a table of significant risk factors by major product line.  We distinguish between loss ratio risk and reserve estimation risk.  Loss ratio risk refers to the possible dispersion of loss ratios from year to year due to inherent volatility in the business such as high severity or aggregating exposures.  Reserve estimation risk recognizes the difficulty in estimating a given year’s ultimate loss liability.  As an example, our property catastrophe business (identified below as “Difference in conditions”) has significant variance in year-over-year results; however its reserving estimation risk is relatively low.

14

                The historical and prospective loss and ALAE estimates along with the risks listed are the basis for determining our initial and subsequent carried reserves. Adjustments in the initial loss ratio by product and segment are made where necessary and reflect updated assumptions regarding loss experience, loss trends, price changes, and prevailing risk factors. The LRC makes all final decisions regarding changes in the initial loss and ALAE ratios.

Loss and LAE Reserve Estimation Process

A full analysis of our loss reserves takes place at least semi-annually. The purpose of these analyses is to provide validation of our carried loss reserves. Estimates of the expected value of the unpaid loss and LAE are derived using actuarial methodologies. These estimates are then compared to the carried loss reserves to determine the appropriateness of the current reserve balance.

The process of estimating ultimate payment for claims and claims expenses begins with the collection and analysis of current and historical claim data. Data on individual reported claims including paid amounts and individual claim adjuster estimates are grouped by common characteristics. There is judgment involved in this grouping. Considerations when grouping data include the volume of the data available, the credibility of the data available, the homogeneity of the risks in each cohort, and both settlement and payment pattern consistency. We use this data to determine historical claim reporting and payment patterns which are used in the analysis of ultimate claim liabilities. For portions of the business without sufficiently large numbers of policies or that have not accumulated sufficient historical statistics, our own data is supplemented with external or industry average data as available and when appropriate. For our executive products and marine business, we utilize external data extensively.

In addition to the review of historical claim reporting and payment patterns, we also incorporate an estimate of expected losses relative to premium by year into the analysis. The expected losses are based on a review of historical loss performance, trends in frequency and severity, and price level changes. The estimation of expected losses is subject to judgment including consideration given to internal and industry data available, growth and policy turnover, changes in policy limits, changes in underlying policy provisions, changes in legal and regulatory interpretations of policy provisions, and changes in reinsurance structure.

We use historical development patterns, estimations of the expected loss ratios, and standard actuarial methods to derive an estimate of the ultimate level of loss and LAE payments necessary to settle all the claims occurring as of the end of the evaluation period. Once an estimate of the ultimate level of claim payments has been derived, the amount of paid loss and LAE and case reserve through the evaluation date is subtracted to reveal the resulting level of IBNR.

                Our reserve processes include multiple standard actuarial methods for determining estimates of IBNR reserves. Other supplementary methodologies are incorporated as deemed necessary. Mass tort and latent liabilities are examples of exposures where supplementary methodologies are used. Each method produces an estimate of ultimate loss by accident year. We review all of these various estimates and the actuaries assign weight to each based on the characteristics of the product being reviewed. The result is a single actuarial point estimate by product, by accident year.

The methodologies we have chosen to incorporate are a function of data availability and appropriately reflective of our own book of business.  There are a number of additional actuarial methods that are available but are not currently being utilized because of data constraints or because the methods were either deemed redundant or not predictive for our book of business.  From time to time, we evaluate the need to add supplementary methodologies.  New methods are incorporated if it is believed that they improve the estimate of our ultimate loss and LAE liability.  All of the actuarial methods tend to converge to the same estimate as an accident year matures.  Our core methodologies are listed below with a short description and their relative strengths and weaknesses:

                Paid Loss Development — Historical payment patterns for prior claims are used to estimate future payment patterns for current claims.  These patterns are applied to current payments by accident year to yield expected ultimate loss.

                Strengths :  The method reflects only the claim dollars that have been paid and is not subject to case-basis reserve changes or changes in case reserve practices.

                Weaknesses :  External claims environment changes can impact the rate at which claims are settled and losses paid (e.g., increase in attorney involvement or legal precedent).  Adjustments to reflect changes in payment patterns on a prospective basis are difficult to quantify.  For losses that have occurred recently, payments can be minimal and thus early estimates are subject to significant instability.

15

Incurred Loss Development — Historical case-incurred patterns (paid losses plus case reserves) for past claims are used to estimate future case-incurred amounts for current claims.  These patterns are applied to current case-incurred losses by accident year to yield an expected ultimate loss.

                Strengths :  Losses are reported more quickly than paid, therefore, the estimates stabilize sooner.  The method reflects more information (claims department case reserve) in the analysis than the paid loss development method.

                Weaknesses :  Method involves additional estimation risk if significant changes to case reserving practices have occurred.

                Case Reserve Development — Patterns of historical development in reported losses relative to historical case reserves are determined. These patterns are applied to current case reserves by accident year and the result is combined with paid losses to yield an expected ultimate loss.

                Strengths :  Like the incurred development method, this method benefits from using the additional information available in case reserves that is not available from paid losses only. It also can provide a more reasonable estimate than other methods when the proportion of claims still open for an accident year is unusually high or low.

                Weaknesses :  It is subject to the risk of changes in case reserving practices or philosophy. It may provide unstable estimates when an accident year is immature and more of the IBNR is expected to come from unreported claims rather than development on reported claims.

Expected Loss Ratio — Historical loss ratios, in combination with projections of frequency and severity trends as well as estimates of price and exposure changes, are analyzed to produce an estimate of the expected loss ratio for each accident year.  The expected loss ratio is then applied to the earned premium for each year to estimate the expected ultimate losses.  The current accident year expected loss ratio is also used as the input in the determination of the prospective loss and ALAE ratio used in our initial IBNR generation process.

                Strengths :  Reflects an estimate independent of how losses are emerging on either a paid or a case reserve basis.  Method is particularly useful in the absence of historical development patterns or where losses take a long time to emerge.

                Weaknesses :  Ignores how losses are actually emerging and thus produces the same estimate of ultimate loss regardless of favorable/unfavorable emergence.

                Paid and Incurred Bornhuetter/Ferguson (BF) — This approach blends the expected loss ratio method with either the paid or incurred loss development method.  In effect, the BF methods produce weighted average indications for each accident year.  As an example, if the current accident year for commercial automobile liability is estimated to be 20 percent paid, then the paid loss development method would receive a weight of 20 percent, and the expected loss ratio method would receive an 80 percent weight.  Over time, this method will converge with the ultimate estimated in the respective loss development method.

                Strengths :  Reflects actual emergence that is favorable/unfavorable, but assumes remaining emergence will continue as previously expected.   Does not overreact to the early emergence (or lack of emergence) where patterns are most unstable.

                Weaknesses :  Could potentially understate favorable or unfavorable development by putting some weight on the expected loss ratio.

                In most cases, multiple estimation methods will be valid for the particular facts and circumstances of the claim liabilities being evaluated.  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 weight given to each estimation method will likely change by accident year and with each evaluation.

                The actuarial point estimates typically follow a progression that places significant weight on the BF methods when accident years are younger and claims emergence is immature.  As accident years mature and claims emerge over time, increasing weight is placed on the incurred development method, the paid development method, and the case reserve development method.  For product lines with faster loss emergence, the progression to greater weight on the incurred and paid development methods occurs more quickly.

16

For our long- and medium-tail products, the BF methods are typically given the most weight for the first 36 months of evaluation.  These methods are also predominant for the first 12 months of evaluation for short-tail lines.  Beyond these time periods, our actuaries apply their professional judgment when weighting the estimates from the various methods deployed.

Judgment can supersede this natural progression if risk factors and assumptions change, or if a situation occurs that amplifies a particular strength or weakness of a methodology.  Extreme projections are critically analyzed and may be adjusted, given less credence, or discarded all together.  Internal documentation is maintained that records any substantial changes in methods or assumptions from one loss reserve study to another.

                Our estimates of ultimate loss and LAE reserves are subject to change as additional data emerges. This could occur as a result of change in loss development patterns, a revision in expected loss ratios, the emergence of exceptional loss activity, a change in weightings between actuarial methods, the addition of new actuarial methodologies or new information that merits inclusion, or the emergence of internal variables or external factors that would alter our view.

                There is uncertainty in the estimates of ultimate losses. Significant risk factors to the reserve estimate include, but are not limited to, unforeseen or unquantifiable changes in:

·           loss payment patterns,

·           loss reporting patterns,

·           frequency and severity trends,

·           underlying policy terms and conditions,

·           business or exposure mix,

·           operational or internal process changes affecting timing of recording transactions,

·           regulatory and legal environment, and/or

·           economic environment.

                Our actuaries engage in discussions with senior management, underwriting, and the claims department on a regular basis to attempt to ascertain any substantial changes in operations or other assumptions that are necessary to consider in the reserving analysis.

                A considerable degree of judgment in the evaluation of all these factors is involved in the analysis of reserves. The human element in the application of 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 various qualified experts may differ materially from each other. We consider this uncertainty by examining our historic reserve accuracy and through an internal peer review process.

                Given the substantial impact of the reserve estimates on our financial statements, we subject the reserving process to significant diagnostic testing and reasonability checks. We have incorporated data validity checks and balances into our front-end processes.  Data anomalies are researched and explained to reach a comfort level with the data and results.   Leading indicators such as actual versus expected emergence and other diagnostics are also incorporated into the reserving processes.

Determination of Our Best Estimate

                Upon completion of our full loss and LAE estimation analysis, the results are discussed with the LRC.  As part of this discussion, the analysis supporting an indicated point estimate of the IBNR loss reserve by product is reviewed.  The actuaries also present explanations supporting any changes to the underlying assumptions used to calculate the indicated point estimate.  Quarterly, we also consider the actual loss emergence as compared to the expected loss emergence derived from the last full loss and LAE analyses.  A review of the resulting variance between the indicated reserves and the carried reserves determined from the initial IBNR generation process takes place.  After discussion of these analyses and all relevant risk factors, the LRC determines whether the reserve balances require adjustment.

17

As a predominantly excess and surplus lines and specialty insurer servicing niche markets, we believe there are several reasons to carry — on an overall basis — reserves above the actuarial point estimate.  We believe we are subject to above average variation in estimates and that this variation is not symmetrical around the actuarial point estimate.

One reason for the variation is the above-average policyholder turnover and changes in the underlying mix of exposures typical of an excess and surplus lines business.  This constant change can cause estimates based on prior experience to be less reliable than estimates for more stable, admitted books of business.  Also, as a niche market writer, there is little industry-level information for direct comparisons of current and prior experience and other reserving parameters.  These unknowns create greater-than-average variation in the actuarial point estimates.

Actuarial methods attempt to quantify future events. Insurance companies are subject to unique exposures that are difficult to foresee at the point coverage is initiated and, often, many years subsequent. Judicial and regulatory bodies involved in interpretation of insurance contracts have increasingly found opportunities to expand coverage beyond that which was intended or contemplated at the time the policy was issued.  Many of these policies are issued on an “all risk” and occurrence basis.  Aggressive plaintiff attorneys have often sought coverage beyond the insurer’s original intent.  Some examples would be the industry’s ongoing asbestos and environmental litigation, court interpretations of exclusionary language on mold and construction defect, and debates over wind versus flood as the cause of loss from major hurricane events.

We believe that because of the inherent variation and the likelihood that there are unforeseen and under-quantified liabilities absent from the actuarial estimate, it is prudent to carry loss reserves above the actuarial point estimate.  Most of our variance between the carried reserve and the actuarial point estimate is in the most recent accident years for our casualty segment where the most significant estimation risks reside. These estimation risks are considered when setting the initial loss ratio for the product and segment.  In the cases where these risks fail to materialize, favorable loss development will likely occur over subsequent accounting periods.  It is also possible that the risks materialize above the amount we considered when booking our initial loss reserves.  In this case, unfavorable loss development is likely to occur over subsequent accounting periods.

Our best estimate of our loss and LAE reserves may change depending on a revision in the actuarial point estimate, the actuary’s certainty in the estimates and processes, and our overall view of the underlying risks. From time to time, we benchmark our reserving policies and procedures and refine them by adopting industry best practices where appropriate.   A detailed, ground-up analysis of the actuarial estimation risks associated with each of our products and segments, including an assessment of industry information, is performed annually.

Loss reserve estimates are subject to a high degree of variability due to the inherent uncertainty of ultimate settlement values. Periodic adjustments to these estimates will likely occur as the actual loss emergence reveals itself over time. We believe our loss reserving processes reflect industry best practices and our methodologies result in a reasonable provision for reserves as of December 31, 2008.

18

Reserve Sensitivities

There are three major parameters that have significant influence on our actuarial estimates of ultimate liabilities by product.  They are the actual losses that are reported, the expected loss emergence pattern, and the expected loss ratios used in the analyses.  If the actual losses reported do not emerge as expected, it may cause us to challenge all or some of our previous assumptions.  We may change expected loss emergence patterns, the expected loss ratios used in our analysis, and/or the weights we place on a given actuarial method.  The impact will be much greater and more leveraged for products with longer emergence patterns.  Our general liability product is an example of a product with a relatively long emergence pattern.  We have constructed a chart below that illustrates the sensitivity of our general liability reserve estimates for these key parameters.   We believe the scenarios to be reasonable as similar favorable variations have occurred in recent years. In particular, our actual general liability loss emergence in 2008 was very favorable. The numbers below are the resulting change in estimated ultimate loss and ALAE in millions of dollars as of December 31, 2008 as a result of the change in the parameter shown.

There are often significant inter-relationships between our reserving assumptions that have offsetting or compounding effects on the reserve estimate.  Thus, in almost all cases, it is impossible to discretely measure the effect of a single assumption or construct a meaningful sensitivity expectation that holds true in all cases.  The scenario above is representative of general liability, one of our largest, and longest-tailed, products.  It is unlikely that all of our products would have variations as wide as illustrated in the example.  It is also unlikely that all of our products would simultaneously experience favorable or unfavorable loss development in the same direction or at their extremes during a calendar year.  Because our portfolio is made up of a diversified mix of products, there would ordinarily be some offsetting favorable and unfavorable emergence by product as actual losses start to emerge and our loss estimates become more refined.

It is difficult for us to predict whether the favorable loss development observed in 2006 through 2008 will continue for any of our products in the future.  We have reviewed historical data detailing the development of our total balance sheet reserves for each of the last 10 years.  Based on this analysis and our understanding of loss reserve uncertainty, we believe fluctuations will occur in our estimate of ultimate reserve liabilities over time.  Over the next calendar year, given our current exposure level and product mix, it would be reasonably likely for us to observe loss reserve development relating to prior years’ estimates across all of our products ranging from approximately 8 percent ($65 million) favorable to 4 percent ($32 million) unfavorable.

19

                The deviations from our initial reserve estimates appeared as changes in our ultimate loss estimates as we updated those estimates through our reserve analysis process. The recognition of the changes in initial reserve estimates occurred over time as claims were reported, initial case reserves were established, initial reserves were reviewed in light of additional information, and ultimate payments were made on the collective set of claims incurred as of that evaluation date. The new information on the ultimate settlement value of claims is therefore continually updated and revised as this process takes place until all claims in a defined set of claims are settled. As a relatively small insurer, our experience will ordinarily exhibit fluctuations from period to period. While we attempt to identify and react to systematic changes in the loss environment, we also must consider the volume of experience directly available to us, and interpret any particular period’s indications with a realistic technical understanding of the reliability of those observations.

                The table below summarizes our prior accident years’ loss reserve development by segment for 2008, 2007, and 2006:

A discussion of significant components of reserve development for the three most recent calendar years follows:

2008.   During 2008, we experienced favorable loss emergence from prior years’ reserve estimates across all of our segments.  For our casualty segment, we experienced $50.6 million of favorable development, predominantly from the accident years of 2002-2006.  In retrospect, the expected loss ratios initially used to set booked reserves for these accident years have proven to be conservative, which has resulted in loss emergence significantly lower than expected.  This is particularly true for our general liability, personal umbrella, and commercial umbrella products which experienced favorable loss development of $33.1 million, $12.7 million, and $11.8 million, respectively.  The construction class was the largest contributor to the favorable emergence in the general liability product.  In addition, our program business experienced $9.3 million of unfavorable prior years’ loss development during the year, mostly isolated in accident years 2004-2007.   Our experience in the liquor liability class has been particularly adverse.  In the past, we relied on external loss development patterns that have not

20

proven predictive of actual emergence. As a result, this class is being re-underwritten and we implemented a more stringent reserving approach in 2008.

           Our property segment realized $6.6 million of favorable loss development in 2008.  Most of this emergence was in accident years 2005-2007.  The construction and fire products were the drivers of the favorable emergence recording $4.4 million and $4.2 million, respectively.  The construction business has been in run-off for three years and recent experience has been much better than expected with a significant reduction in both frequency and severity of claims.  Only a handful of contracts remain open and we have observed little new activity from this product line.  Our fire product saw significant favorable emergence from the 2007 accident year, as our year-end 2007 reserves have developed more favorably than originally estimated.

Our surety segment experienced $5.1 million of favorable emergence.  Almost all of the favorable emergence is from the 2007 accident year.  Very little observed loss severity in the commercial surety product resulted in $1.7 million of favorable emergence.  Continued improvement in our contract surety loss ratio resulting from past re-underwriting of the business led to $2.5 million of favorable loss reserve development.

2007.   We periodically review our loss reserve estimates and underlying actuarial reserving methodologies in order to assess their accuracy and suitability, and to benchmark our reserving practices against industry best practices. A detailed assessment of recent trends and reserve risk factors was undertaken in 2007. As part of our reviews, we performed a more detailed, ground-up analysis of the actuarial estimation risks associated with each of our products and segments, including an assessment of industry information.

Our analyses also revealed that our quarterly actuarial reserve estimates over recent historical periods had shown a downward trend as a result of a moderating loss trend environment, improvements in policy terms and conditions and a favorable underlying exposure mix that occurred during the hard market period from 2001 through 2004.

Based on this review, we made certain refinements to our reserving methodologies to include a more detailed consideration of the impact of risk factors on total recorded reserves through increased internal dialogue among the claim, underwriting, risk management and actuarial departments, greater transparency of the actuarial process and results, and improved reserving diagnostics. Overall, these enhancements and improved information provide better and faster feedback to management regarding loss development resulting in greater overall confidence in the actuarial estimates. This and the increased stability in our business in the last few years diminished the needed level of carried reserves above the actuarial point estimate. We believe that these reserve methodology enhancements improved the overall accuracy of our best estimate of loss and LAE reserves. Over half of the favorable prior years’ loss development was the result of this detailed assessment and resulting changes in our booked reserves.

Our casualty segment was most impacted by prior years’ loss development realizing a total of $87.4 million of favorable emergence. All casualty products were impacted by the enhanced risk assessment previously mentioned. Our general liability, transportation, personal umbrella, and professional liability products realized favorable development of $42.5 million, $19.6 million, $10.5 million, and $8.1 million, respectively. This favorable emergence was concentrated in accident years 2004-2006. As a result of significant favorable loss development observed over the past several years for our general liability product, we reassessed the expected loss ratios used in our actuarial analysis and subsequently lowered them for the construction classes. For our transportation and personal umbrella products, we reassessed and subsequently lowered the loss development factors in our analysis reflecting our observation that the emergence patterns were more favorable than previously anticipated. Finally, our professional liability products realized actual loss emergence much more favorably than expected.

The property segment realized $6.7 million of favorable prior years’ development. The favorable emergence was realized across almost all of our property products, predominantly in accident years 2005 and 2006. We also executed a favorable reinsurance commutation impacting accident years prior to 2000.

The surety segment realized $11.1 million of favorable prior years’ development. Almost all of the development was the result of the risk reassessment and reflection of significantly lower reserve risk after achieving settlement with the larger banks involved in the CMC litigation (see note 10 to our audited consolidated financial statements included in our 2008 Financial Report to Shareholders, attached as Exhibit 13 and incorporated by reference herein, for more details).

2006.   During 2006, we continued to experience favorable loss development and a reduction in prior years’ loss reserve estimates. Pricing increased substantially and policy terms and conditions became more favorable for most of our products during the 2001-2004 policy years. Many of the improvements in market conditions were difficult to quantify at the time of the original estimate. Our significant growth in premium and exposures made precise quantification of these changes even more challenging. In 2006, losses continued to emerge on the prior accident years much more favorably than we expected

21

when making our original estimates. We experienced favorable development of $43.4 million in aggregate on prior years’ estimates.

Of this decrease to prior years’ loss reserve estimates, approximately $40.0 million occurred in the casualty segment. The development is primarily from our general liability, executive products liability, and employer’s indemnity products. In our general liability product we experienced $25.4 million of favorable development. Most of this development came from the 2004 and 2005 accident years. As part of our normal reserving process, we reviewed the expected loss ratios used in several of our reserving methods. This review confirmed the favorable emergence from 2002-2005 accident years. As a result of this study, the expected loss ratios were reduced for 2004-2006 with the most significant change occurring to the 2005 accident year. Approximately $15.4 million of the favorable general liability development can be attributed to this update in expected loss ratios. The remaining portion of the decrease in prior year’s loss reserve estimate was the result of the continued favorable loss emergence and the natural progression of shifting more weight to our incurred and paid development methods as accident years get older. In our executive products liability business, we experienced $7.4 million of favorable development. Most of this change can be attributed to accident years 2001, 2003, and 2004. The estimates improved as a result of lower than expected loss severity in those accident years. For our employer’s indemnity product, we experienced $5.7 million of favorable development. We experienced significantly less loss emergence than expected for accident years prior to 2003 and benefited from favorable settlements on several claims in accident years 2001-2003.

Overall, our property and surety segments experienced relatively small changes in prior years’ estimates of reserves. However, we experienced $4.2 million of favorable development from 2004 and 2005 hurricane estimates. We also saw $7.2 million of unfavorable development on our construction product that is in runoff. Most of this development came from accident years 2002-2005. The construction emergence pattern revealed itself to be longer than originally anticipated and has not behaved consistent with reporting patterns expected from a property segment. We do not anticipate any further significant deterioration in our estimates.

                The following table presents the development of our balance sheet reserves from 1998 through 2008. The top line of the table shows the net reserves at the balance sheet date for each of the indicated periods. This represents the estimated amount of net losses and settlement expenses arising in all prior years that are unpaid at the balance sheet date, including losses that had been incurred but not yet reported to us. The lower portion of the table shows the re-estimated amount of the previously recorded gross and net reserves based on experience as of the end of each succeeding year. The estimate changes as more information becomes known about the frequency and severity of claims for individual periods.

                Adverse loss and LAE reserve development can be observed in the table for years ending 1999-2002 on a net basis, and 1999-2003 on a gross basis.  This development is related to unexpectedly large increases in loss frequency and severity and unquantifiable expansion of policy terms and conditions that took place in accident years 1997-2001 for our casualty segment.  These causes widely impacted the property and casualty insurance industry during this time as soft market conditions were prevalent.  These factors, combined with our rapid growth during 1999-2002, caused significant estimation risk, and thus had a related impact on our reserve liabilities for those years.

                As the table displays, variations exist between our cumulative loss experience on a gross and net basis, due to the application of reinsurance. On certain products, our net retention (after applying reinsurance) is significantly less than our gross retention (before applying reinsurance).  These differences in retention can cause a significant (leveraged) difference between loss reserve development on a net and gross basis.   Additionally, the relationship of our gross to net retention changes over time. For example, we changed underwriting criteria to increase gross retentions (gross policy limits) on certain products written in 1999 through 2001, while leaving net retention unchanged. These products contained gross retentions of up to $50.0 million, while the relating net retention remained at $0.5 million. Loss severity on certain of these products exceeded original expectations. As shown in the table that follows, on a re-estimated basis, this poor loss experience resulted in significant indicated gross deficiencies, with substantially less deficiency indicated on a net basis, as many losses were initially recorded at their full net retention. In 2002, we reduced our gross policy limits on many of these products to $15.0 million, while net retention increased to $1.0 million. As the relationship of our gross to net retention changes over time, re-estimation of loss reserves will result in variations between our cumulative loss experience on a gross and net basis.

22