HANOVER INSURANCE GROUP, INC.

The following is our discussion and analysis of the results of operations by business segment. The operating results are presented before interest expense, income taxes and other items which management believes are not indicative of our core operations, including realized gains and losses, as well as unrealized gains and losses on equity securities, and the results of discontinued operations.

The following table summarizes the results of operations for the periods indicated:

Operating income before interest expense and income taxes was $933.0 million for the year ended December 31, 2025, compared to $650.1 million for the year ended December 31, 2024, an improvement of $282.9 million. This increase was primarily due to lower catastrophe losses, improvements in current accident year underwriting results in our Personal Lines segment, higher net investment income, and earned premium growth. These improvements were partially offset by higher current accident year losses in our Core Commercial segment. Pre-tax catastrophe losses were $276.3 million for the year ended December 31, 2025, compared to $375.9 million for the year ended December 31, 2024, a decrease of $99.6 million. The catastrophe losses in 2025 were due to severe convective storms across multiple states, and to a lesser extent, the California Palisades and Eaton wildfires. The higher catastrophe losses in 2024 were primarily due to several convective storms across multiple states, with wind and hail damage representing the majority of reported losses, primarily in the Midwest, and, to a lesser extent, the impacts of hurricanes Helene and Beryl. The improved Personal Lines current accident year underwriting results were primarily due to the benefit of earned pricing outpacing loss trends in both homeowners and personal automobile lines, and moderated loss frequency trends, particularly in homeowners property and automobile collision coverages. The increase in Core Commercial current accident year losses was primarily in our commercial automobile and workers’ compensation lines of business, partially offset by lower current accident year losses in our commercial multiple peril line of business.

Net premiums written increased $238.5 million for the year ended December 31, 2025, compared to the year ended December 31, 2024. The increase in net premiums written was primarily due to renewal price increases and new business.

PREMIUM PRODUCTION AND UNDERWRITING RESULTS

The following tables summarize premiums written on a gross and net basis, net premiums earned and, catastrophe loss, loss and LAE (including catastrophe losses), expense, and combined ratios for our Core Commercial, Specialty and Personal Lines segments. Loss and LAE, catastrophe loss and combined ratios shown below include prior year reserve development. These items were not meaningful for our Other segment.

The following tables summarize net premiums written, and loss and LAE and catastrophe loss ratios by line of business for the Core Commercial, Specialty and Personal Lines segments. Loss and LAE and catastrophe loss ratios include prior year reserve development.

The following tables summarize GAAP underwriting results for our Core Commercial, Specialty, Personal Lines and Other segments and reconciles them to operating income (loss) before interest expense and income taxes.

Core Commercial net premiums written were $2,273.7 million for the year ended December 31, 2025, compared to $2,195.5 million for the year ended December 31, 2024. The $78.2 million increase in net premiums written was primarily driven by renewal price increases and an increase in new business.

Core Commercial underwriting profit for the year ended December 31, 2025 was $52.2 million, compared to $114.8 million for the year ended December 31, 2024, a decrease of $62.6 million. Catastrophe losses for the year ended December 31, 2025 were $113.4 million, compared to $77.2 million for the year ended December 31, 2024, an increase of $36.2 million. Net favorable development on prior years’ loss reserves for the year ended December 31, 2025 was $7.1 million, compared to $17.7 million for the year ended December 31, 2024, a decrease of $10.6 million.

Core Commercial current accident year underwriting profit, excluding catastrophes, was $158.5 million for the year ended December 31, 2025, compared to $174.3 million for the year ended December 31, 2024. The $15.8 million decrease in underwriting results was primarily due to higher current accident year losses, partially offset by lower expenses and earned premium growth. Higher current accident year losses were primarily due to our commercial automobile and workers’ compensation lines of business, partially offset by lower losses in our commercial multiple peril line.

We continue to manage underwriting performance through rate actions, risk selection and mitigation, pricing segmentation, specific underwriting actions and targeted new business growth. Our ability to achieve overall rate increases is affected by many factors, including regulatory activity and the competitive pricing environment, particularly within the workers’ compensation line.

Specialty net premiums written were $1,441.5 million for the year ended December 31, 2025, compared to $1,373.9 million for the year ended December 31, 2024. The $67.6 million increase in net premiums written was primarily due to renewal price increases and an increase in new business.

Specialty underwriting profit for the year ended December 31, 2025 was $197.3 million, compared to $175.2 million for the year ended December 31, 2024, an increase of $22.1 million. Catastrophe losses for the year ended December 31, 2025 were $34.1 million, compared to $37.5 million for the year ended December 31, 2024, a decrease of $3.4 million. Net favorable development on prior years’ loss reserves for the year ended December 31, 2025 was $56.8 million, compared to $46.2 million for the year ended December 31, 2024, an increase of $10.6 million.

Specialty current accident year underwriting profit, excluding catastrophes, was $174.6 million for the year ended December 31, 2025, compared to $166.5 million for the year ended December 31, 2024. The $8.1 million increase in underwriting results was primarily due to earned premium growth.

We continue to manage underwriting performance through rate actions, risk selection and mitigation, pricing segmentation, specific underwriting actions and targeted new business growth. Our ability to achieve overall rate increases is affected by many factors, including regulatory activity and the competitive pricing environment.

Personal Lines net premiums written were $2,606.9 million for the year ended December 31, 2025, compared to $2,514.2 million for the year ended December 31, 2024. The $92.7 million increase in net premiums written was primarily due to renewal price increases, higher retention, and increased new business.

Net premiums written in the personal automobile line of business for the year ended December 31, 2025 were $1,489.9 million, compared to $1,463.6 million for the year ended December 31, 2024, an increase of $26.3 million. Personal automobile policies in force decreased by 3.1% since December 31, 2024. Net premiums written in the homeowners and other lines of business for the year ended December 31, 2025 were $1,117.0 million, compared to $1,050.6 million for the year ended December 31, 2024, an increase of $66.4 million. Homeowners policies in force decreased by 2.6% since December 31, 2024.

Personal Lines underwriting profit for the year ended December 31, 2025 was $243.7 million, compared to an underwriting loss of $4.1 million for the year ended December 31, 2024, a favorable change in underwriting results of $247.8 million. Catastrophe losses for the year ended December 31, 2025 were $128.8 million, compared to $261.2 million for the year ended December 31, 2024, a decrease of $132.4 million. Net favorable development on prior years’ loss reserves for the year ended December 31, 2025 was $7.2 million, compared to $4.9 million for the year ended December 31, 2024, an increase of $2.3 million.

Personal Lines current accident year underwriting profit, excluding catastrophes, was $365.3 million for the year ended December 31, 2025, compared to $252.2 million for the year ended December 31, 2024. The $113.1 million increase in underwriting results was primarily due to lower current accident year losses in our homeowners and personal automobile lines and, to a lesser extent, earned premium growth. The improved Personal Lines current accident year results were primarily due to the benefit of earned pricing outpacing loss trends in both homeowners and personal automobile lines, and moderated loss frequency trends, particularly in homeowners property and automobile collision coverages.

We obtained pricing increases of approximately 14% in our homeowners line and 9% in our personal automobile line during 2025 and believe that our ability to obtain pricing increases, although moderating, will continue. Consistent with our expectations, PIF declined in 2025, driven by lower policies available to renew year over year from prior margin recapture actions taken. Additionally, our Personal Lines net premiums written may be affected by price competition and the regulatory and overall macroeconomic environment. These factors may also affect our ability to maintain and improve underwriting results.

Other

Our Other segment had an operating profit of $6.2 million for the year ended December 31, 2025, compared to an operating loss of $0.5 million for the year ended December 31, 2024, an improvement of $6.7 million, due primarily to an increase in net investment income.

RESERVE FOR LOSSES AND LOSS ADJUSTMENT EXPENSES

Overview of Loss Reserve Estimation Process

We maintain reserves for our insurance products to provide for our ultimate liability for losses and loss adjustment expenses (our “loss reserves”) with respect to reported and unreported claims incurred as of the end of each accounting period. These reserves are estimates, taking into account past loss experience, modified for current trends, as well as prevailing economic, legal and social conditions. Loss reserves represent our largest liability.

Management’s process for establishing loss reserves is comprehensive and involves input from multiple functions throughout our organization, including actuarial, finance, claims, legal, underwriting, distribution, and business operations management. The process incorporates facts currently known, as well as the current, and in some cases, the anticipated, state of the law and coverage litigation. Based on information currently available, we believe that the aggregate loss reserves at December 31, 2025 were adequate to cover claims for losses that had occurred as of that date, including both those known to us and those yet to be reported. However, as described below, there are significant uncertainties inherent in the loss reserving process. Our estimate of the ultimate liability for losses that had occurred as of December 31, 2025 is expected to change in future periods as we obtain further information, and such changes could have a material effect on our results of operations and financial position.

Our loss reserves include case estimates for claims that have been reported and estimates for claims that have been incurred but not reported (“IBNR”) at the balance sheet date. They also include estimates of the expenses associated with processing and settling all reported and unreported claims, less estimates of anticipated salvage and subrogation recoveries. Our loss reserves are not discounted to present value.

Case reserves are established by our claim personnel individually, on a claim by claim basis, and based on information specific to the occurrence and terms of the underlying policy. For some classes of business, average case reserves are used initially. Case reserves are periodically reviewed and modified based on new or additional information pertaining to the claim.

Our ultimate IBNR reserves are estimated by management and our reserving actuaries on an aggregate basis for each line of business or coverage for loss and loss expense liabilities not reflected within the case reserves. The sum of the case reserves and the IBNR reserves represents our estimate of total unpaid losses and loss adjustment expenses.

We regularly review our loss reserves using a variety of industry accepted analytical techniques. We update the loss reserves as historical loss experience develops, additional claims are reported and resolved, and new information becomes available. Net changes in loss reserves are reflected in operating results in the period in which the reserves are changed.

The IBNR reserve includes both a provision for claims that have occurred but have not yet been reported to us, some of which may not yet be known to the insured, and a provision for future development on reported claims. IBNR represents a significant proportion of our total net loss reserves (i.e., net of estimated reinsurance recoverables), particularly for long-tail liability classes. In fact, approximately 56% of our aggregate net loss reserves at December 31, 2025 were for IBNR losses and loss expenses.

Critical Judgments and Key Assumptions

We determine the amount of our net loss reserves based on an estimation process that is complex and considers information from both company specific and industry data, as well as general economic and other information. The estimation process utilizes a combination of objective and subjective information, the blending of which requires significant professional judgment. There are various assumptions required, including future trends in frequency and severity of claims, operational changes in claim handling and case reserving practices, and trends related to general economic and social conditions. Informed judgments as to our ultimate exposure to losses are an integral component of our loss reserve estimation process.

There is greater inherent uncertainty in estimating insurance reserves for certain types of property and casualty insurance lines, particularly liability lines, where a longer period of time may elapse before a definitive determination of ultimate liability and losses may be made (sometimes referred to as “long-tail” business). In addition, the technological, judicial, regulatory and political climates involving these types of claims are continuously evolving. There is also greater uncertainty in establishing reserves with respect to business that is new to us, particularly new business which is generated with respect to newly introduced product lines, by newly appointed agents, or in geographies in which we have less experience in conducting business. In each of these cases, there is less historical experience or knowledge, and less data upon which we can rely. A combination of business that is both new to us and has longer development periods provides even greater uncertainty in estimating insurance reserves. In addition, in recent periods, we have experienced extensions of the “tails” in certain lines of business as the full value of claims are presented later than had been our historical experience.

We regularly update our reserve estimates as new information becomes available and additional events occur which may impact the resolution of unsettled claims. Reserve adjustments are reflected in the results of operations as adjustments to losses and LAE. Often, these adjustments are recognized in periods subsequent to the period in which the underlying policy was written and the loss event occurred. When these types of subsequent adjustments affect prior years, they are described separately as “prior year reserve development.” Such development can be either favorable or unfavorable to our financial results and may vary by line of business. As discussed below, estimated loss and LAE reserves for claims occurring in prior years, in the aggregate, developed favorably by $70.4 million and $67.4 million for the years ended December 31, 2025 and 2024, respectively, although there was some significant variance by segment. Additionally, our estimated loss and LAE reserves for catastrophe claims occurring in prior years developed favorably by $26.0 million for the year ended December 31, 2025 and $42.0 million for the year ended December 31, 2024. There can be no assurance that current loss and LAE reserves will be sufficient.

We regularly review our reserving techniques, our overall reserving position and our reinsurance. Based on (i) our review of historical data, legislative enactments, judicial decisions, legal developments in impositions of damages and policy coverage, political attitudes and trends in general economic conditions, (ii) our review of per claim information, (iii) our historical loss experience and that of the industry, (iv) the nature of policies written by us, and (v) our internal estimates of required reserves, we believe that adequate provision has been made for loss reserves. Given the inherent complexity of our loss reserve estimation process and the potential variability of the assumptions used, the actual emergence of losses will vary, perhaps substantially, from the estimate of losses included in our financial statements, particularly in those instances where settlements or other claim resolutions do not occur until well into the future. Our net loss reserves at December 31, 2025 were $5.9 billion. Therefore, a relatively small percentage change in the estimate of net loss reserves would have a material effect on our results of operations. Similarly, a one percentage point change in the aggregate loss and LAE ratio resulting from a change in reserve estimation is currently projected to have an approximate $62 million impact on operating income, based on 2025 full year premiums.

The major causes of material uncertainty relating to ultimate losses and LAE (“risk factors”) generally vary for each line of business, as well as for each separately analyzed component of the line of business. In some cases, such risk factors are explicit assumptions of the estimation method and in others, 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 substantially 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 affect multiple lines of business. Examples include changes in claim handling and claim reserving practices, changes in claim settlement patterns, claims complexion, regulatory and legislative actions, increased attorney involvement in claims, court actions or delays in legal proceedings, social inflation and legal system abuse dynamics, 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 line of business. Individual risk factors are subject to interactions with other risk factors within line of business components. Thus, risk factors can have offsetting or compounding effects on required reserves.

Inflation generally increases the cost of losses covered by insurance contracts. The effect of inflation varies by product. Our insurance premiums are established before the amount of losses and LAE and the extent to which inflation may affect such expenses are known. Consequently, we attempt, in establishing rates and reserves, to anticipate the potential impact of inflation in the projection of ultimate costs. For example, we monitor, and continue to experience, increases in medical costs, wages, and legal defense costs, all of which are key considerations in setting reserve assumptions for workers’ compensation, bodily injury and other liability lines. We are also monitoring the continued advancements in technology and design found in automobiles and homes, and the increased claims settlement costs that result from labor shortages, and repairs or replacement of such equipment impacted by supply chain disruptions, which could lead to material shortages and elevated prices of building materials. Estimated increases are reflected in our current reserve estimates, but continued increases are expected to contribute to increased losses and LAE in the future.

We are also defendants in various litigation matters, including putative class actions, which may seek punitive damages, bad faith or extra-contractual damages, legal fees and interest, or claim a broader scope of policy coverage or settlement and payment obligations than our interpretation. Resolution of these cases is often highly unpredictable, could result in extending the time to settle, and could involve material unanticipated damage awards. We have experienced, and others in the industry have reported, increased attorney involvement in claims resulting in delayed claim settlements, legal system abuse dynamics, and a trend toward higher valued settlements and litigation, all of which contribute to uncertainty regarding reserve estimates.

Loss and LAE Reserves by Line of Business

Reserving Process Overview

Our loss reserves include amounts related to short-tail and long-tail classes of business. “Tail” refers to the time period between the occurrence of a loss and the final settlement of the claim. The longer the time span between the incidence of a loss and the settlement of the claim (i.e., a longer tail), the more the ultimate settlement amount may likely vary from our original estimate.

Short-tail classes consist principally of automobile physical and property damage, homeowners property, commercial property, and marine business. For these property coverages, claims are generally reported and settled shortly after the loss occurs because the claims relate to tangible property and are more likely to be discovered shortly after the loss occurs. Consequently, the estimation of loss reserves for these classes is generally less complex. However, the estimation of loss reserves for these classes is more complex when there are rapid changes in the ultimate timing or future cost to settle the claims.

While we estimate that approximately half of our written premium is in, what we would characterize as, shorter-tail classes of business, most of our loss reserves relate to longer-tail liability classes of business. Long-tailed classes include commercial general liability, third-party coverage, automobile liability, personal and commercial umbrella, and workers’ compensation. For many liability claims, significant periods of time, ranging up to several years or more, may elapse between the occurrence of the loss, the discovery and reporting of the loss to us and the settlement of the claim. As a result, loss experience in the more recent accident years for long-tailed liability coverage has limited statistical credibility because a relatively small proportion of losses in these accident years (the calendar years in which losses are incurred) are reported claims and an even smaller proportion are paid losses. Liability claims are also more susceptible to litigation and can be significantly affected by changing contract interpretations, the legal, political and social environments, the risk and expense of protracted litigation, legal system abuse dynamics, and inflation. Consequently, the estimation of loss reserves for these coverages is more complex and typically subject to a higher degree of variability and uncertainty compared to short-tailed coverages.

Most of our indirect business from our ongoing involuntary and run-off voluntary pools is assumed long-tailed casualty reinsurance. Reserve estimates for this business are therefore subject to the variability caused by extended loss emergence periods. The estimation of loss reserves for this business is further complicated by delays between the time the claim is reported to the ceding insurer and when it is reported by the ceding insurer to the pool manager and then to us, and by our dependence on the quality and consistency of the loss reporting by the ceding company and actuarial estimates by the pool manager. These reserving factors also apply to our discontinued assumed accident and health reinsurance pools and arrangements that are included in our liabilities of discontinued businesses (See “Risk Factors” in Part I – Item 1A for further discussion).

A review of loss reserves for each of the classes of business in which we write is conducted regularly, generally quarterly. This review process takes into consideration a variety of trends that impact the ultimate settlement of claims. Where appropriate, the review includes a review of overall payment patterns and the emergence of paid and reported losses relative to expectations.

The loss reserve estimation process relies on the basic assumption that past experience, adjusted for the effects of current developments and likely trends, is an appropriate basis for predicting future outcomes. As part of this process, we use a variety of analytical methods that consider experience, trends and other relevant factors. IBNR reserves are generally calculated by first projecting the ultimate cost of all claims that have been reported or expected to be reported in the future and then subtracting reported losses and loss expenses. Reported losses include cumulative paid losses and loss expenses plus case reserves. Within the loss reserving process, standard actuarial methods which include: (1) loss development factor methods; (2) expected loss methods (Bornheutter-Ferguson); and (3) adjusted loss methods (Berquist-Sherman), are given due consideration. These methods are described below:

•

Loss development factor methods generally assume that the losses yet to emerge for an accident year are proportional to the paid or reported loss amount observed to date. Historical patterns of the development of paid and reported losses by accident year can be predictive of the expected future patterns that are applied to current paid and reported losses to generate estimated ultimate losses by accident year.

•

Bornheutter-Ferguson methods utilize the product of the expected ultimate losses times the proportion of ultimate losses estimated to be unreported or unpaid to calculate IBNR. The expected ultimate losses are based upon current estimates of ultimate losses from prior accident years, adjusted to reflect expected earned premium, current rating, claims cost levels and changes in business mix. The expected losses, and corresponding loss ratios, are a critical component of Bornheutter-Ferguson methodologies and provide a general reasonability guide.

•

Berquist-Sherman methods are used for estimating reserves in business lines where historical development patterns may be deemed less reliable for more recent accident years’ ultimate losses. Under these methods, patterns of historical paid or reported losses are first adjusted to reflect current payment settlement patterns and case reserve adequacy and then evaluated in the same manner as the loss development factor methods described above. When the adequacy of case reserves change, the Berquist-Sherman incurred method may be deemed more reliable than the reported loss development factor method. Likewise, when the settlement patterns change, the Berquist-Sherman paid method may be deemed more reliable than the paid loss development factor method.

In addition to the methods described above, various tailored reserving methodologies are used for certain businesses. For example, for some low volume and high volatility classes of business, special reserving techniques are utilized that estimate IBNR by selecting the loss ratio that balances actual reported losses to expected reported losses as defined by the estimated underlying reporting pattern. Also, for some classes with long exposure periods (e.g., construction defect, engineering and surety), earnings patterns plus an estimated reporting lag applied to the Bornheutter-Ferguson initial expected loss ratio are used to estimate IBNR. This is done in order to reflect the changing average exposure periods by policy year (and consequently accident year).

In completing the loss reserve analysis, a variety of assumptions must be made for each line of business, coverage and accident year. Each estimation method has its own pattern, parameter and/or judgmental dependencies, with no estimation method being better than the others in all situations. The relative strengths and weaknesses of the various estimation methods, when applied to a particular class of business, can also change over time, depending on the underlying circumstances. In many cases, multiple estimation methods will be valid for the particular facts and circumstances of the relevant class of business. The manner of application and the degree of reliance on a given method will vary by line of business and coverage, and by accident year based on an evaluation of the above dependencies and the potential volatility of the loss frequency and severity patterns. The estimation methods selected or given weight at a particular valuation date are those that are believed to produce the most reliable indication for the loss reserves being evaluated. Selections incorporate input from claims personnel, pricing actuaries, and underwriting management on loss cost trends and other factors that could affect ultimate losses.

For most classes of shorter-tailed business in our Core Commercial, Specialty and Personal Lines segments, the emergence of paid and incurred losses generally exhibits a relatively stable pattern of loss development from one accident year to the next. Thus, for these classes, the loss development factor method is generally appropriate. For some of the classes of shorter-tailed business, the emergence of paid and incurred losses may exhibit a relatively volatile pattern of loss development from one accident year to the next. In these cases where there is a relatively low level of reliability placed on the available paid and incurred loss data, expected loss methods or adjusted loss methods are considered appropriate for the most recent accident year.

For longer-tailed lines of business, applying the loss development factor method often requires even more judgment in selecting development factors, as well as more significant extrapolation. For those long-tailed lines of business with high frequency and relatively low per-loss severity (e.g., personal automobile liability), volatility will often be sufficiently modest for the loss development factor method to be given significant weight, even in the most recent accident years, but expected loss methods and adjusted loss methods are always considered and frequently utilized in the selection process. For those long-tailed lines of business with low frequency and high loss potential (e.g., commercial general liability and personal and commercial umbrella), anticipated loss experience is less predictable because of the small number of claims and erratic claim severity patterns. In these situations, the loss development factor methods may not produce a reliable estimate of ultimate losses in the most recent accident years since many claims either have not yet been reported or are only in the early stages of the settlement process. Therefore, the loss reserve estimates for these accident years may be based on methods less reliant on extrapolation, such as Bornheutter-Ferguson. Over time, as a greater number of claims are reported and the statistical credibility of loss experience increases, loss development factor methods or adjusted loss methods are given increasing weight.

Management endeavors to apply as much available data as practicable to estimate the loss reserve amount for each line of business, coverage and accident year, utilizing varying assumptions, projections and methods. The ultimate outcome is expected to fall within a range of potential outcomes around this loss reserve estimated amount.

Our carried reserves for each line of business and coverage are determined based on our quarterly loss reserving process. In making the determination, we consider numerous quantitative and qualitative factors. Quantitative factors include changes in reserve estimates in the period, the maturity of the accident year, trends observed over the recent past, the level of volatility within a particular class of business, the estimated effects of reinsurance, including reinstatement premiums, general economic trends, and other factors. Qualitative factors may include legal and regulatory developments, changes in claim handling and case reserving practices, recent entry into new markets or products, changes in underwriting practices or business mix, concerns that we do not have sufficient or quality historical reported and paid loss and LAE information with respect to a particular line or segment of our business, effects of the economy and political outlook, litigation trends and legal system abuse dynamics, perceived anomalies in the historical results, and evolving trends or other factors. In doing so, we must evaluate whether a change in the data represents credible actionable information or an anomaly. Such an assessment requires considerable judgment. Even if a change is determined to be apparent, it is not always possible to determine the extent of the change. As a result, there can be a time lag between the emergence of a change and a determination that the change should be partially or fully reflected in the carried loss reserves. In general, changes are made more quickly to reserves for more mature accident years and less volatile classes of business.

Reserving Process Uncertainties

As stated above, numerous factors (both internal and external) contribute to the inherent uncertainty in the process of establishing loss reserves, including changes in the rate of inflation for goods and services related to insured damages (e.g., medical care, home and automobile repairs, etc.), changes in the judicial interpretation of policy provisions and settlement obligations, changes in the general attitude of juries in determining damage awards, delayed court proceedings, legislative actions, such as expanding liability, coverage mandates or expanding or suspending statutes of limitations which otherwise limit the times within which claims can be made, changes in the extent of insured injuries and claims complexion, changes in the trend of expected frequency and/or severity of claims, changes in our book of business (e.g., change in mix due to new or modified product offerings, new or rapidly expanding geographic areas, etc.), changes in our underwriting practices, and changes in claim handling procedures and/or systems. Regarding our indirect business from involuntary and run-off voluntary pools, we are periodically provided loss estimates by managers of each pool. We adopt reserve estimates for the pools that consider this information and other facts.

In addition, we must consider the uncertain effects of emerging or potential claims and coverage issues that arise as legal, judicial and social conditions, political risks, and economic conditions change. For example, claims which we consider closed may be re-opened as additional damages surface or new liability or damage theories are presented. Also, historically, we have observed more frequent and higher severity in workers’ compensation, bodily injury, umbrella and other liability claims and more credit-related losses (for example, in our surety business) during periods of economic uncertainty or high unemployment. Economic and labor force dynamics have resulted in many experienced workers retiring from their positions and being replaced with less skilled workers, which could result in more workplace accidents. Another example may be changes in driving behavior over time and the impacts to the make-up, complexity and severity of claims in the automobile lines of business. These, and other issues, could have a negative effect on our loss reserves by either extending coverage beyond the original underwriting intent or by increasing the number or size of claim settlements.

As part of our loss reserving analysis, we consider the various factors that contribute to the uncertainty in the loss reserving process. Those factors that could materially affect our loss reserve estimates include loss development patterns and loss cost trends, reporting lags, rate and exposure level changes, the effects of changes in coverage and policy limits, business mix shifts, the effects of regulatory and legislative developments, economic circumstances, the effects of changes in judicial interpretations, the level of attorney involvement in claims, the effects of legal system abuse dynamics, the effects of emerging claims and coverage issues, and the effects of changes in claim handling and claim reserving practices. In making estimates of reserves, however, we do not necessarily make an explicit assumption for each of these factors. Moreover, all estimation methods do not utilize the same assumptions and typically no single method is determinative in the reserve analysis for a line of business and coverage. Consequently, changes in our loss reserve estimates generally are not the result of changes in any one assumption. Instead, the variability will be affected by the interplay of changes in numerous assumptions, many of which are implicit to the approaches used.

For each line of business and coverage, we regularly adjust the assumptions and methods used in the estimation of loss reserves in response to our actual loss experience, as well as our judgments regarding changes in trends and/or emerging patterns. In those instances where we primarily utilize analyses of historical patterns of the development of paid and reported losses, this may be reflected, for example, in the selection of revised loss development factors. In longer-tailed classes of business and for which loss experience is less predictable due to potential changes in judicial interpretations, potential legislative actions, the cost of litigation or determining liability and the ultimate loss, inflation, potential claims, shifting claim settlement patterns due to delayed court proceedings, and other issues, this may be reflected in a judgmental change in our estimate of ultimate losses for particular accident years. Most of the insurance policies we have written over many years are written on an “occurrence” basis, which means we insure specified acts or events which occurred during the covered period, even if claims first arise from such events many years later. For example, the industry has incurred significant losses as a result of claims related from sexual abuse and molestation claims which occurred decades ago and were not known at such time, and in many cases policy limits were available for the years during which such occurrence policies were in place.

The future impact of the various factors that contribute to the uncertainty in the loss reserving process is impossible to predict. There is potential for significant variation in the development of loss reserves, particularly for long-tailed classes of business and classes of business that are more vulnerable to economic, social or political risks.

Reserving Process for Catastrophe Events

The estimation of claims and claims expense reserves for catastrophes is also comprised of estimates of losses from reported claims and IBNR, primarily for damage to property. In general, our estimates for catastrophe reserves are determined on an event basis by considering various sources of available information, including specific loss estimates reported to us based on claim adjuster inspections, overall industry loss estimates, our internal data regarding exposures related to the geographical location of the event and estimates of potential subrogation recoveries. However, depending on the nature of the catastrophe, the estimation process can be further complicated by other impediments. For example, for hurricanes and other severe wind storms and wildfires, complications often include the inability of insureds to promptly report losses, delays in the ability of claims adjusting staff to inspect losses, difficulties in determining whether wind storm losses are covered by our homeowners policy (generally for damage caused by wind or wind-driven rain) or are specifically excluded from coverage caused by flood, challenges in estimating additional living expenses, assessing the impact of demand surge, exposure to mold or smoke damage, and the effects of numerous other considerations. Another example is the complication of estimating the cost of business interruption coverage on Core Commercial and Specialty policies. Estimates for catastrophes which occur at or near the end of a financial reporting period, may be even less reliable since we will have less claims data available and little time to complete our estimation process. In such situations, we may adapt our practices to accommodate the circumstances.

For events designated as catastrophes, we generally calculate IBNR reserves directly as a result of an estimated IBNR claim count and an estimated average claim amount for each event. Such an assessment involves a comprehensive analysis of the nature of the event, of policyholder exposures within the affected geographic area and of available claims intelligence. Depending on the nature of the event, available claims intelligence could include surveys of field claims associates within the affected geographic area, aerial photographs of the affected area, feedback from a catastrophe claims team sent into the area, as well as data on claims reported as of the financial statement date. In addition, loss emergence from similar historical events is compared to the estimated IBNR for our current catastrophe events to help assess the reasonableness of our estimates. However, in some cases, it may be difficult to estimate certain catastrophe losses which are unique and do not have instances of historical precedence, such as the property damage arising from the 2020 riots and civil unrest during the year those events occurred.

Reserving Sensitivity Analysis

The following discussion presents disclosure related to possible variation in net reserve estimates (i.e., net of estimated reinsurance recoverables) due to changes in key assumptions. This information is provided for illustrative purposes only. Many other assumptions may also lead to material reserve adjustments. If any such variations do occur, then they would likely occur over a period of several years and therefore their impact on our results of operations would be recognized during the same periods. It is important to note, however, that there is the potential for future variations greater than the amounts described below and for any such variations to be recognized in a single quarterly or annual period. No consideration has been given to potential correlation or lack of correlation among key assumptions or among lines of business and coverage as described below. As a result, and because there are so many other factors which affect our net reserve estimate, it would be inappropriate to take the amounts described below and simply add them together in an attempt to estimate volatility in total. While we believe these are reasonably possible scenarios, the reader should not consider the following sensitivity analysis as illustrative of a net reserve range.

•

Personal and Commercial Automobile Bodily Injury – loss reserves recorded for bodily injury on voluntary business were $1,039.5 million as of December 31, 2025. A key assumption for bodily injury is the inflation rate underlying the estimated reserve. A five point change (e.g., 3% changed to 8% or -2%) in the embedded inflation rate would have changed total reserves by approximately $183 million, either positive or negative, at December 31, 2025.

•

Personal Automobile Personal Injury Protection Medical Payment – loss reserves recorded for personal injury protection medical payment on voluntary business were $104.7 million as of December 31, 2025, of which approximately 87% relates to Michigan policies. A key assumption for this coverage is the inflation rate underlying the estimated reserve. Given the long reporting pattern for this line of business, an additional key assumption is the amount of additional development required to reach full maturity, thereby reflecting ultimate costs, as represented by the tail factor. A five point change in the embedded inflation rate (e.g., 6% changed to 11% or 1%) and a one point change to the tail factor assumption (e.g., 1% changed to 0% or 2%) would have changed total reserves by approximately $36 million, either positive or negative, at December 31, 2025.

•

Workers’ Compensation – loss reserves recorded for workers’ compensation on voluntary business were $508.9 million as of December 31, 2025. A key assumption for workers’ compensation is the inflation rate underlying the estimated reserve. Given the long reporting pattern for this line of business, an additional key assumption is the amount of additional development required to reach full maturity, thereby reflecting ultimate costs, as represented by the tail factor. A five point change in the embedded inflation rate (e.g., 4% changed to 9% or -1%) and a one point change to the tail factor assumption (e.g., 2% changed to 1% or 3%) would have changed total reserves by approximately $194 million, either positive or negative, at December 31, 2025.

•

Monoline and Multiple Peril General Liability – loss reserves recorded for monoline and multiple peril general liability on voluntary business were approximately $1,282.5 million as of December 31, 2025. A key assumption for monoline and multiple peril general liability is the implied adequacy of the underlying case reserves. A ten point change in case adequacy (e.g., 10% deficiency changed to 0% or 20% deficiency) would have changed total reserves by approximately $175 million, either positive or negative, at December 31, 2025.

•

Specialty Programs – loss reserves recorded for Hanover Programs were $285.5 million as of December 31, 2025. Two key assumptions underlying the actuarial reserve analysis for specialty programs are the inflation rate underlying the estimated reserve for our commercial automobile liability, general liability and workers’ compensation coverages, as well as the tail factor selection for workers’ compensation. A five point change to the embedded inflation rate for the aforementioned coverages (e.g., 4% changed to 9% or -1%), and a one point change in the workers’ compensation tail factor on Hanover Programs (e.g., 0.1% changed to 1.1% or -0.9%) would have changed total reserves by approximately $43 million, either positive or negative, at December 31, 2025.

Carried Reserves and Reserve Rollforward

The following table provides a reconciliation of the gross beginning and ending reserve for unpaid losses and loss adjustment expenses.

The following table summarizes the gross reserve for losses and LAE by line of business and division.

Loss and LAE reserves in our “Other core commercial” lines include monoline general liability, commercial umbrella and monoline property. “Specialty Property & Casualty” includes specialty industrial and commercial property, excess and surplus lines and specialty general liability coverage, and program business. “Professional and Executive Lines” includes management liability, fidelity and crime, professional liability and other property and liability lines for healthcare firms. Loss and LAE reserves in our “Total Other” segment relate to our run-off direct asbestos and environmental business, our run-off voluntary assumed property and casualty reinsurance pools, and run-off product liability business.

Prior Year Development

Conditions and trends that have affected reserve development in the past will not necessarily recur in the future. As discussed under “Reserving Process Overview” in the preceding section, our historical loss experience and loss development patterns are important factors in estimating loss reserves, however, they are not the only factors we evaluate to establish reserves. Therefore, a mechanical application of standard actuarial methodologies in projecting ultimate claims could result in materially different reserves to those held. Accordingly, it is not appropriate to extrapolate future favorable or unfavorable development based on amounts experienced in prior periods.

The following table summarizes prior year (favorable) unfavorable development by segment:

Catastrophe Loss Development

For the year ended December 31, 2025, favorable catastrophe loss development was $26.0 million, primarily due to lower than expected losses related to events from accident years 2023 through 2024, including several convective storms across multiple states, and Winter Storm Elliot. In 2024, prior year catastrophe loss development was favorable $42.0 million, primarily due to lower than expected losses related to events from accident years 2021 through 2023, including several convective storms across multiple states, Winter Storm Elliot and Hurricane Ian.

2025 Loss and LAE Development, excluding catastrophes

In 2025, net favorable loss and LAE development, excluding catastrophes, was $70.4 million. Specialty favorable loss and LAE development of $56.8 million was primarily due to lower than expected losses in our Specialty Property & Casualty division, primarily in our programs and excess and surplus lines businesses, Professional & Executive Lines division, primarily in our claims-made general liability coverages, and in our Marine division. Personal Lines favorable loss and LAE development of $7.2 million was primarily due to lower than expected losses in our homeowners line. Core Commercial favorable loss and LAE development was $7.1 million, with lower than expected losses in the workers’ compensation, general liability, and commercial multiple peril product lines, which was partially offset by higher than expected losses in our commercial automobile and other core commercial lines.

2024 Loss and LAE Development, excluding catastrophes

In 2024, net favorable loss and LAE development, excluding catastrophes, was $67.4 million. Specialty favorable loss and LAE development of $46.2 million was primarily due to lower than expected losses in our general liability-claims-made coverage within our Professional and Executive Lines division, and in our Marine and Surety and Specialty Property & Casualty divisions. Core Commercial favorable loss and LAE development was $17.7 million, with favorability in each of the main product lines: commercial multiple peril, commercial automobile, workers’ compensation and other commercial lines. Personal Lines favorable development of $4.9 million was primarily due to lower than expected losses in our personal automobile line and, to a lesser extent, lower than expected losses in our homeowners line, partially offset by higher than expected losses in other personal lines.

Asbestos and Environmental Reserves

As of December 31, 2025, we had $10.5 million of net asbestos and environmental reserves, comprised of $8.3 million of direct reserves and $2.2 million of assumed reinsurance pool reserves. This compares to net reserves of $10.9 million at December 31, 2024, comprised of $8.7 million of direct reserves and $2.2 million of assumed reinsurance pool reserves.

As of December 31, 2025, we had gross loss and LAE reserves for our assumed reinsurance pool business of $2.2 million compared to $29.9 million in December 31, 2024. These assumed reinsurance pool reserves relate to pools in which we have terminated our participation, however, we continue to be subject to claims related to years in which we were a participant. Results of operations from these pools are included in our Other segment. A significant part of our gross pool reserves at December 31, 2024 was related to our participation in the Excess and Casualty Reinsurance Association (“ECRA”) voluntary pool for our ECRA claim liability participations. During 2025, we completed an insurance business transfer of our ECRA liabilities to a third-party, which novation has fully relieved us of our obligations to ECRA policyholders. This transaction had no significant impact on our results of operations for the year ended December 31, 2025.

We estimate our ultimate liability for asbestos, environmental and toxic tort liability claims, whether resulting from direct business, assumed reinsurance or pool business, based upon currently known facts, reasonable assumptions where the facts are not known, current law, and methodologies currently available. Although these outstanding claims are not believed to be significant, their existence gives rise to uncertainty and are discussed because of the possibility that they may become significant. We believe that, notwithstanding the evolution of case law expanding liability in asbestos and environmental claims, recorded reserves related to these claims are adequate. Nevertheless, the asbestos, environmental and toxic tort liability reserves could be revised, and any such revisions could have a material adverse effect on our results of operations for a particular quarterly or annual period, or on our financial position.

Reinsurance Recoverables

Reinsurance recoverables were $2,011.1 million and $1,994.5 million at December 31, 2025 and December 31, 2024, respectively, of which $62.6 million and $74.2 million, respectively, represent billable recoverables. A reinsurance recoverable is billable after an eligible reinsured claim is paid by an insurer. Billable reinsurance recoverables related to the MCCA were $30.5 million and $39.3 million at December 31, 2025 and December 31, 2024, respectively, and billable non-MCCA reinsurance recoverables totaled $32.1 million and $34.9 million at December 31, 2025 and December 31, 2024, respectively. At December 31, 2025, there were $0.1 million of billed balances outstanding greater than 90 days. At December 31, 2024, there were $0.8 million of billed balances outstanding greater than 90 days.