The following discussion should be read in conjunction with Item 1A Risk Factors, Item 6 Selected Financial Data and Item 8 Financial Statements and Supplementary Data of this Form 10-K.

The preparation of Consolidated Financial Statements in conformity with GAAP requires us to make estimates and assumptions that affect the reported amounts of assets and liabilities and the disclosure of contingent assets and liabilities at the date of the Consolidated Financial Statements and the amount of revenues and expenses reported during the period. Actual results may differ from those estimates.

Our Consolidated Financial Statements and accompanying notes have been prepared in accordance with GAAP applied on a consistent basis. We continually evaluate the accounting policies and estimates used to prepare the Consolidated Financial Statements. In general, our estimates are based on historical experience, evaluation of current trends, information from third-party professionals and various other assumptions that are believed to be reasonable under the known facts and circumstances.

The accounting estimates discussed below are considered by us to be critical to an understanding of our Consolidated Financial Statements as their application places the most significant demands on our judgment. Note A to the Consolidated Financial Statements included under Item 8 should be read in conjunction with this section to assist with obtaining an understanding of the underlying accounting policies related to these estimates. Due to the inherent uncertainties involved with these types of judgments, actual results could differ significantly from our estimates and may have a material adverse impact on our results of operations, financial condition, equity, business, and insurer financial strength and corporate debt ratings.

Insurance reserves are established for both short and long-duration insurance contracts. Short-duration contracts are primarily related to property and casualty insurance policies where the reserving process is based on actuarial estimates of the amount of loss, including amounts for known and unknown claims. Long-duration contracts are primarily related to long term care policies and are estimated using actuarial estimates about morbidity and persistency as well as assumptions about expected investment returns and future premium rate increases. The reserve for unearned premiums represents the portion of premiums written related to the unexpired terms of coverage. The reserving process is discussed in further detail in the Reserves-Estimates and Uncertainties section below.

Future policy benefit reserves for our long term care policies are based on certain assumptions, including morbidity, persistency, inclusive of mortality, discount rates and future premium rate increases. The adequacy of the reserves is contingent upon actual experience and our future expectations related to these key assumptions. If actual or expected future experience differs from these assumptions, the reserves may not be adequate, requiring us to add to reserves.

A prolonged period during which interest ratesinvestment returns remain at levels lower than those anticipated in our reserving discount rate assumption could result in shortfalls in investment income on assets supporting our obligations under long term care policies, which may also require an increase to our reserves. In addition, we may not receive regulatory approval for the level of premium rate increases we request.

These changes to our reserves could materially adversely impact our results of operations, financial condition and equity. The reserving process is discussed in further detail in the Reserves - Estimates and Uncertainties section below.

Exposure exists with respect to the collectibility of ceded property and casualty and life reinsurance to the extent that any reinsurer is unable to meet its obligations or disputes the liabilities we have ceded under reinsurance agreements. An allowance for uncollectible reinsurance is recorded on the basis of periodic evaluations of balances due from reinsurers, reinsurer financial strength rating and solvency, industry

experience and current and forecast economic conditions. Further information on our reinsurance receivables is in Note G to the Consolidated Financial Statements included under Item 8.

Additionally, exposure exists with respect to the collectibility of amounts due from policyholders related to insurance contracts, including amounts due from insureds under high deductible policies and retrospectively rated policies. An allowance for uncollectible insurance receivables is recorded on the basis of periodic evaluations of balances due from insureds, currently as well as in the future, historical business default data, management's experience and current and forecast economic conditions.

If actual experience differs from the estimates made by management in determining the allowances for uncollectible reinsurance and insurance receivables, net receivables as reflected on our Consolidated Balance Sheets may not be collected. Therefore, our results of operations, financial condition or equity could be materially adversely affected. Further information on our process for determining the allowances for uncollectible reinsurance and insurance receivables is in Note A to the Consolidated Financial Statements included under Item 8.

Our fixed maturity and equity securities are carried at fair value on the balance sheet. Fair value represents the price that would be received in a sale of an asset in an orderly transaction between market participants on the measurement date, the determination of which may require us to make a significant number of assumptions and judgments. Securities with the greatest level of subjectivity around valuation are those that rely on inputs that are significant to the estimated fair value and that are not observable in the market or cannot be derived principally from or corroborated by observable market data. These unobservable inputs are based on assumptions consistent with what we believe other market participants would use to price such securities. Further information on our fair value measurements is in Note C to the Consolidated Financial Statements included under Item 8.

We account for income taxes under the asset and liability method. Under this method, deferred income taxes are recognized for temporary differences between the financial statement and tax return basis of assets and liabilities. Any resulting future tax benefits are recognized to the extent that realization of such benefits is more likely than not, and a valuation allowance is established for any portion of a deferred tax asset that management believes will not be realized. The assessment of the need for a valuation allowance requires management to make estimates and assumptions about future earnings, reversal of existing temporary differences and available tax planning strategies. If actual experience differs from these estimates and assumptions, the recorded deferred tax asset may not be fully realized resulting in an increase to income tax expense in our results of operations. In addition, the ability to record deferred tax assets in the future could be limited, resulting in a higher effective tax rate in that future period.

The level of claim reserves we maintain represents our best estimate, as of a particular point in time, of what the ultimate settlement and administration of claims will cost based on our assessment of facts and circumstances known at that time. Reserves are not an exact calculation of liability but instead are complex estimates that we derive, generally utilizing a variety of actuarial reserve estimation techniques, from numerous assumptions and expectations about future events, both internal and external, many of which are highly uncertain. As noted below, we review our reserves for each segment of our business periodically, and any such review could result in the need to increase reserves in amounts which could be material and could adversely affect our results of operations, financial condition, equity, business and insurer financial strength and corporate debt ratings. Further information on reserves is provided in Note E to the Consolidated Financial Statements included under Item 8.

We maintain loss reserves to cover our estimated ultimate unpaid liability for claim and claim adjustment expenses, including the estimated cost of the claims adjudication process, for claims that have been reported but not yet settled (case reserves) and claims that have been incurred but not reported (IBNR). IBNR includes a provision for development on known cases as well as a provision for late reported incurred claims. Claim and claim adjustment expense reserves are reflected as liabilities and are included on the Consolidated Balance Sheets under the heading “Insurance Reserves.” Adjustments to prior year reserve estimates, if necessary, are reflected in results of operations in the period that the need for such adjustments is determined. The carried case and IBNR reserves as of each balance sheet date are provided in the Segment Results section of this MD&A and in Note E to the Consolidated Financial Statements included under Item 8.

As discussed in the Risk Factors discussion within Item 1A, there is a risk that our recorded reserves are insufficient to cover our estimated ultimate unpaid liability for claims and claim adjustment expenses. Given the unprecedented nature of the event, a particularly high level of uncertainty exists as to the potential impact on insurance losses related to the COVID-19 pandemic, mitigating actions and consequent economic crisis. Unforeseen emerging or potential claims and coverage issues are also difficult to predict and could materially adversely affect the adequacy of our claim and claim adjustment expense reserves and could lead to future reserve additions.

In addition, our property and casualty insurance subsidiaries also have actual and potential exposures related to A&EP claims, which could result in material losses. To mitigate the risks posed by our exposure to A&EP claims and claim adjustment expenses, we completed a transaction with NICO under which substantially all of our legacy A&EP liabilities were ceded to NICO effective January 1, 2010. See Note E to the Consolidated Financial Statements included under Item 8 for further discussion about the transaction with NICO, its impact on our results of operations, the deferred retroactive reinsurance gain and the amount of remaining reinsurance limit.

In developing claim and claim adjustment expense (loss or losses) reserve estimates, our actuaries perform detailed reserve analyses that are staggered throughout the year. The data is organized at a reserve group level. A reserve group can be a line of business covering a subset of insureds such as commercial automobile liability for small or middle market customers or it can be a particular type of claim such as construction defect. Every reserve group is reviewed at least once during the year, but most are reviewed more frequently. The analyses generally review losses gross of ceded reinsurance and apply the ceded reinsurance terms to the gross estimates to establish estimates net of reinsurance. In addition to the detailed analyses, we review actual loss emergence for all products each quarter.

Most of our business can be characterized as long-tail. For long-tail business, it will generally be several years between the time the business is written and the time when all claims are settled. Our long-tail exposures include commercial automobile liability, workers' compensation, general liability, medical professional liability, other professional liability and management liability coverages, assumed reinsurance run-off and products liability. Short-tail exposures include property, commercial automobile physical damage, marine, surety and

warranty. Specialty, Commercial and International contain both long-tail and short-tail exposures. Corporate & Other contains run-off long-tail exposures.

Various methods are used to project ultimate losses for both long-tail and short-tail exposures.

The paid development method estimates ultimate losses by reviewing paid loss patterns and applying them to accident or policy years with further expected changes in paid losses. Selection of the paid loss pattern may require

For many reserve groups, paid loss data for recent periods may be too immature or erratic for accurate predictions. This situation often exists for long-tail exposures. In addition, changes in the factors described above may result in inconsistent payment patterns. Finally, estimating the paid loss pattern subsequent to the most mature point available in the data analyzed often involves considerable uncertainty for long-tail products such as workers' compensation.

The incurred development method is similar to the paid development method, but it uses case incurred losses instead of paid losses.  Since the method uses more data (case reserves in addition to paid losses) than the paid development method, the incurred development patterns may be less variable than paid patterns. However, selection of the incurred loss pattern typically requires analysis of all of the same factors described above. In addition, the inclusion of case reserves can lead to distortions if changes in case reserving practices have taken place, and the use of case incurred losses may not eliminate the issues associated with estimating the incurred loss pattern subsequent to the most mature point available.

The loss ratio method multiplies earned premiums by an expected loss ratio to produce ultimate loss estimates for each accident or policy year. This method may be useful for immature accident or policy periods or if loss development patterns are inconsistent, losses emerge very slowly or there is relatively little loss history from which to estimate future losses. The selection of the expected loss ratio typically requires analysis of loss ratios from earlier accident or policy years or pricing studies and analysis of inflationary trends, frequency trends, rate changes, underwriting changes and other applicable factors.

The Bornhuetter-Ferguson method using paid loss is a combination of the paid development method and the loss ratio method. This method normally determines expected loss ratios similar to the approach used to estimate the expected loss ratio for the loss ratio method and typically requires analysis of the same factors described above. This method assumes that future losses will develop at the expected loss ratio level. The percent of paid loss to ultimate loss implied from the paid development method is used to determine what percentage of ultimate loss is yet to be paid. The use of the pattern from the paid development method typically requires consideration of the same factors listed in the description of the paid development method. The estimate of losses yet to be paid is added to current paid losses to estimate the ultimate loss for each year. For long-tail lines, this method will react very slowly if actual ultimate loss ratios are different from expectations due to changes not accounted for by the expected loss ratio calculation.

The Bornhuetter-Ferguson method using incurred loss is similar to the Bornhuetter-Ferguson method using paid loss except that it uses case incurred losses. The use of case incurred losses instead of paid losses can result in development patterns that are less variable than paid patterns. However, the inclusion of case reserves can lead to distortions if changes in case reserving have taken place, and the method typically requires analysis of the same factors that need to be reviewed for the loss ratio and incurred development methods.

The frequency times severity method multiplies a projected number of ultimate claims by an estimated ultimate average loss for each accident or policy year to produce ultimate loss estimates. Since projections of the ultimate number of claims are often less variable than projections of ultimate loss, this method can provide more reliable results for reserve groups where loss development patterns are inconsistent or too variable to be relied

on exclusively. In addition, this method can more directly account for changes in coverage that affect the number and size of claims. However, this method can be difficult to apply to situations where very large claims or a substantial number of unusual claims result in volatile average claim sizes. Projecting the ultimate number of claims may require analysis of several factors, including the rate at which policyholders report claims to us, the impact of judicial decisions, the impact of underwriting changes and other factors. Estimating the ultimate average loss may require analysis of the impact of large losses and claim cost trends based on changes in the cost of repairing or replacing

Stochastic modeling produces a range of possible outcomes based on varying assumptions related to the particular reserve group being modeled. For some reserve groups, we use models which rely on historical development patterns at an aggregate level, while other reserve groups are modeled using individual claim variability assumptions supplied by the claims department. In either case, multiple simulations using varying assumptions are run and the results are analyzed to produce a range of potential outcomes. The results will typically include a mean and percentiles of the possible reserve distribution which aid in the selection of a point estimate.

For many exposures, especially those that can beare considered long-tail, a particular accident or policy year may not have a sufficient volume of paid losses to produce a statistically reliable estimate of ultimate losses. In such a case, our actuaries typically assign more weight to the incurred development method than to the paid development method. As claims continue to settle and the volume of paid loss increases, the actuaries may assign additional weight to the paid development method.  For most of our products, even the incurred losses for accident or policy years that are early in the claim settlement process will not be of sufficient volume to produce a reliable estimate of ultimate losses. In these cases, we may not assign much, if any, weight to the paid and incurred development methods. We may use the loss ratio, Bornhuetter-Ferguson and/or frequency times severity methods.  For short-tail exposures, the paid and incurred development methods can often be relied on sooner, primarily because our history includes a sufficient number of years to cover the entire period over which paid and incurred losses are expected to change. However, we may also use the loss ratio, Bornhuetter-Ferguson and/or frequency times severity methods for short-tail exposures.

For other more complex reserve groups where the above methods may not produce reliable indications, we use additional methods tailored to the characteristics of the specific situation.

The reserve analyses performed by our actuaries result in point estimates. Each quarter, the results of the detailed reserve reviews are summarized and discussed with senior management to determine management's best estimate of reserves. Senior management considers many factors in making this decision. Our recorded reserves reflect our best estimate as of a particular point in time based upon known facts and circumstances, consideration of the factors cited above and our judgment. The carried reserve differsreserves differ from the actuarial point estimate as discussed further below.

The key assumptions fundamental to the reserving process are often different for various reserve groups and accident or policy years. Some of these assumptions are explicit assumptions that are required of a particular method, but most of the assumptions are implicit and cannot be precisely quantified. An example of an explicit assumption is the pattern employed in the paid development method. However, the assumed pattern is itself based on several implicit assumptions such as the impact of inflation on medical costs and the rate at which claim professionals close claims. As a result, the effect on reserve estimates of a particular change in assumptions typically cannot be specifically quantified, and changes in these assumptions cannot be tracked over time.

Our recorded reserves are management's best estimate. In order to provide an indication of the variability associated with our net reserves, the following discussion provides a sensitivity analysis that shows the approximate estimated impact of variations in significant factors affecting our reserve estimates for particular types of business. These significant factors are the ones that we believe could most likely materially affect the reserves. This discussion covers the major types of business for which we believe a material deviation to our reserves is reasonably possible. There can be no assurance that actual experience will be consistent with the current assumptions or with the variation indicated by the discussion. In addition, there can be no assurance that other factors and assumptions will not have a material impact on our reserves.

The three areas for which we believe a significant deviation to our net reserves is reasonably possible are (i) professional liability, management liability and surety products; (ii) workers' compensation; and (iii) general liability.

Professional liability, management liability and surety products include U.S. professional liability coverages provided to various professional firms, including architects, real estate agents, small and mid-sized accounting firms, law firms and other professional firms. They also include directors and officers (D&O), employment practices, fiduciary, fidelity and surety coverages, and medical liability. The most significant factor affecting reserve estimates for these liability coverages is claim severity. Claim severity is driven by the cost of medical care, the cost of wage replacement, legal fees, judicial decisions, legislative changes and other factors. Underwriting and claim handling decisions, such as the classes of business written and individual claim settlement decisions, can also affect claim severity. If the estimated claim severity increases by 9%, we estimate that net reserves would increase by approximately $400 million. If the estimated claim severity decreases by 3%, we estimate that net reserves would decrease by approximately $150 million. Our net reserves for these products were approximately $4.4$4.6 billion as of December 31, 2019.2020.

For workers' compensation, since many years will pass from the time the business is written until all claim payments have been made, the most significant factor affecting workers' compensation reserve estimates is claim cost inflation on claim payments. Workers' compensation claim cost inflation is driven by the cost of medical care, the cost of wage replacement, expected claimant lifetimes, judicial decisions, legislative changes and other factors. If estimated workers' compensation claim cost inflation increases by 100 basis points for the entire period over which claim payments will be made, we estimate that our net reserves would increase by approximately $350 million. If estimated workers' compensation claim cost inflation decreases by 100 basis points for the entire period over which claim payments will be made, we estimate that our net reserves would decrease by approximately $350 million. Our net reserves for workers' compensation were approximately $4.0$3.9 billion as of December 31, 2019.2020.

For general liability, the most significant factor affecting reserve estimates is claim severity. Claim severity is driven by changes in the cost of repairing or replacing property, the cost of medical care, the cost of wage replacement, judicial decisions, legislation and other factors. If the estimated claim severity for general liability increases by 6%, we estimate that our net reserves would increase by approximately $200 million. If the estimated claim severity for general liability decreases by 3%, we estimate that our net reserves would decrease by approximately $100 million. Our net reserves for general liability were approximately $3.4$3.5 billion as of December 31, 2019.2020.

Given the factors described above, it is not possible to quantify precisely the ultimate exposure represented by claims and related litigation. As a result, we regularly review the adequacy of our reserves and reassess our reserve estimates as historical loss experience develops, additional claims are reported and settled and additional information becomes available in subsequent periods. In reviewing our reserve estimates, we make adjustments in the period that the need for such adjustments is determined. These reviews have resulted in our identification of information and trends that have caused us to change our reserves in prior periods and could lead to our identification of a need for additional material increases or decreases in claim and claim adjustment expense reserves, which could materially affect our results of operations, financial condition, equity, business and insurer financial strength and corporate debt ratings, positively or negatively. See discussion within Note E to the Consolidated Financial Statements included under Item 8 for additional information about reserve development and the Ratings section of this MD&A for further information regarding our financial strength and corporate debt ratings.

Our Life & Group segment includes our run-off long term care business as well as structured settlement obligations not funded by annuities related to certain property and casualty claimants. Long term care policies provide benefits for nursing homes, assisted living facilities and home health care subject to various daily and lifetime caps. Generally, policyholders must continue to make periodic premium payments to keep the policy in force and we have the ability to increase policy premiums, subject to state regulatory approval.

We maintain both claim and claim adjustment expense reserves as well as future policy benefit reserves for policyholder benefits for our Life & Group segment. Claim and claim adjustment expense reserves consist of estimated reserves for long term care policyholders that are currently receiving benefits, including claims that have been incurred but are not yet reported. In developing the claim and claim adjustment expense reserve estimates for our long term care policies, our actuaries perform a detailed claim experience studyreserve review on an annual basis. The study reviewsreview analyzes the sufficiency of existing reserves for policyholders currently on claim and includes an evaluation of expected benefit utilization and claim duration. Our recorded claim and claim adjustment expense reserves reflect management's best estimate after incorporating the results of the most recent study. In addition, claim and claim adjustment expense reserves are also maintained for the structured settlement obligations. In developing the claim and claim adjustment expense reserve estimates for our structured settlement obligations, our actuaries monitor mortality experience on an annual basis. Both elementsOur recorded claim and claim adjustment expense reserves reflect management's best estimate after incorporating the results of Life & Groupthe most recent reviews. Claim and claim adjustment expense reserves for long term care policies and structured settlement obligations are discounted as discussed in Note A to the Consolidated Financial Statements included under Item 8.

Future policy benefit reserves representconsist of the active life reserves related to our long term care policies whichfor policyholders that are not currently receiving benefits and represent the present value of expected future benefit payments and expenses less expected future premium. The determination of these reserves requires management to make estimates and assumptions about expected investment and policyholder experience over the life of the contract. Since many of these contracts may be in force for several decades, these assumptions are subject to significant estimation risk.

Prior to September 30, 2019, the active life reserves for long term care were based on the actuarial best estimate assumptions established at December 31, 2015 as a result of a reserve unlocking in the fourth quarter of 2015. The September 30, 2018 GPV indicated the carried reserves included a margin of approximately $182 million. The September 30, 20192020 GPV indicated a premium deficiency of $216$74 million and future policyholderpolicy benefit reserves at that date were increased accordingly. As a result, the long term care active life reserves carried as of September 30, 20192020 represent management's best estimate assumptions at that date with no margin for adverse deviation. A summary of the changes as a result of the 20192020 GPV is presented in the table below:

The premium deficiency was primarily driven by changes in discount rate assumptions driven bydue to lower expected reinvestment rates, contemplating both near-term market indications and long-term normative assumptions. The premium deficiencyThis unfavorable driver was also adversely affected by changes in persistency assumptions, primarily from lower projected active life mortality rates. Recognition of margin in earnings subsequent to the 2018 GPV also contributed to the premium deficiency. These unfavorable drivers were partiallysignificantly offset by higher than expectedpreviously estimated rate increases on active rate increase programs, new planned rate increase filings and favorable changes to the underlying morbiditypersistency and expensemorbidity assumptions.

Catastrophes are an inherent risk of the property and casualty insurance business and have contributed to material period-to-period fluctuations in our results of operations and/or equity. We reported catastrophe losses, net of reinsurance, of $179$550 million and $252$179 million for the years ended December 31, 20192020 and 2018.2019. Net catastrophe losses for the year ended December 31, 2020 included $294 million related primarily to severe weather related events, $195 million related to the COVID-19 pandemic and $61 million related to civil unrest. Net catastrophe losses for the year ended December 31, 2019 related primarily to U.S. weather related events. Net catastrophe losses for the year ended December 31, 2018 included $88 million related to Hurricane Michael, $47 million related to the California wildfires and $33 million related to Hurricane Florence. The remaining net catastrophe losses in 2018 resulted primarily from U.S. weather related events.

We generally seek to manage our exposure to catastrophes through the purchase of catastrophe reinsurance and have catastrophe reinsurance treaties that cover property and workers’ compensation losses. We conduct an ongoing review of our risk and catastrophe coverages and from time to time make changes as we deem appropriate. The following discussion summarizes our most significant catastrophe reinsurance coverage at January 1, 2020.2021.

We purchased corporate catastrophe excess-of-loss treaty reinsurance covering our U.S. states and territories and Canadian property exposures underwritten in our North American and European companies. Exposures underwritten through Hardy are excluded. The treaty has a term of JanuaryMay 1, 20192020 to May 1, 2020. The 2019 treaty2021 and provides coverage for the accumulation of covered losses from catastrophe occurrences above our per occurrence retention of $250 million up to $1.0$1.2 billion. Losses stemming from terrorism events are covered unless they are due to a nuclear, biological or chemical attack. All layers of the treaty provide for one full reinstatement.

We also purchased corporate Workers' Compensation catastrophe excess-of-loss treaty reinsurance for the period January 1, 20202021 to January 1, 20212022 providing $275 million of coverage for the accumulation of covered losses related to natural catastrophes above our per occurrence retention of $25 million. The treaty also provides $475 million of coverage for the accumulation of covered losses related to terrorism events above our retention of $25 million. Of this $475 million in Terrorism coverage, $200 million is provided for nuclear, biological chemical and radiation events. One full reinstatement is available for the first $275 million above the retention, regardless of the covered peril. We also purchased a targeted facultative facility to address exposure accumulations in specific peak Terrorism zones.

Our principal reinsurance protection against large-scale terrorist attacks, including nuclear, biological, chemical or radiological attacks, is the coverage currently provided through TRIPRA which has been extendedruns through the end of 2027. TRIPRA provides a U.S. government backstop for insurance-related losses resulting from any “act of terrorism”, which is certified by the Secretary of Treasury in consultation with the Secretary of Homeland Security for losses that exceed a threshold of $200 million industry-wide for the calendar year 2020.2021. Under the current provisions of the program, in 2020,2021, the federal government will reimburse 80% of our covered losses in excess of our applicable deductible up to a total industry program cap of $100 billion. Our deductible is based on eligible commercial property and casualty earned premiums for the preceding calendar year. Based on 20192020 earned premiums, our estimated deductible under the program is $850$820 million for 2020.2021. If an act of terrorism or acts of terrorism result in covered losses exceeding the $100 billion annual industry aggregate limit, Congress would be responsible for determining how additional losses in excess of $100 billion will be paid.

Core income increased $134decreased $244 million in 20192020 as compared with 2018.2019. Core income for our Property & Casualty Operations increased approximately $223decreased $356 million primarily due to higher net catastrophe losses and lower net investment income drivenpartially offset by limited partnership and common stock returns and favorableimproved non-catastrophe current accident year underwriting results partially offset by lower favorable net prior period loss reserve development in the current year.results. Core results for our Life & Group segment decreased $152improved $118 million. Life & Group results for the year ended December 31, 2020 include a $59 million driven bycharge related to recognition of a premium deficiency as a result of the third quarter 2020 GPV compared to a $170 million charge in 2019 related to recognition of a premium deficiency as a result of the third quarter 2019 GPV. Core loss for our Corporate & Other segment improved approximately $63 million driven by lower adverse prior year A&EP reserve development.increased $6 million.

Favorable net prior year loss reserve development of $73$20 million and $181$73 million was recorded in 20192020 and 20182019 related to our Specialty, Commercial, International and Corporate & Other segments. Further information on net prior year loss reserve development is in Note E to the Consolidated Financial Statements included under Item 8.

The following discusses the results of operations for our business segments.

Our property and casualty commercial insurance operations are managed and reported in three business segments: Specialty, Commercial and International, which we refer to collectively as Property & Casualty Operations. Specialty provides management and professional liability and other coverages through property and casualty products and services using a network of brokers, independent agencies and managing general underwriters. Commercial works with a network of brokers and independent agents to market a broad range of property and casualty insurance products to all types of insureds targeting small business, construction, middle markets and services to small, middle-market and large businesses.other commercial customers. The International segment underwrites property and casualty coverages on a global basis through a branch operation in Canada, a European business consisting of two insurance companies based in the U.K. and Luxembourg a branch operation in Canada as well as throughand Hardy, our Lloyd's syndicate.

Our operations outside of Property & Casualty Operations are managed and reported in two segments: Life & Group and Corporate & Other. Life & Group primarily includes the results of our long term care business that is in run-off. Corporate & Other primarily includes certain corporate expenses, including interest on corporate debt, and the results of certain property and casualty businesses in run-off, including CNA Re and A&EP. Intersegment eliminations are also included in this segment.

Changes in estimates of claim and claim adjustment expense reserves, net of reinsurance, for prior years are defined as net prior year loss reserve development within this MD&A. These changes can be favorable or unfavorable. Net prior year loss reserve development does not include the effect of any related acquisition expenses. Further information on our reserves is provided in Note E to the Consolidated Financial Statements included under Item 8.

Specialty provides management and professional liability and other coverages through property and casualty products and services using a network of brokers, independent agencies and managing general underwriters. Specialty includes the following business groups:

The following table details the results of operations for Specialty.

Gross written premiums, excluding third party captives, for Specialty increased $181$281 million in 20192020 as compared with 20182019 driven by strong retentionrate and rate.higher new business. Net written premiums for Specialty increased $104$192 million in 20192020 as compared with 2018.2019. The increase in net earned premiums was consistent with the trend in net written premiums.

Core income increased $42decreased $136 million in 20192020 as compared with 20182019 primarily due to higher net catastrophe losses and lower net investment income driven by limited partnership and common stock returns partially offset by lower favorable net priorimproved non-catastrophe current accident year loss reserve development.underwriting results.

Favorable net prior year loss reserve development of $92$61 million and $150$92 million was recorded in 20192020 and 2018.2019. Further information on net prior year loss reserve development is in Note E to the Consolidated Financial Statements included under Item 8.

The following table summarizes the gross and net carried reserves for Specialty.

Corporate & Other primarily includes certain corporate expenses, including interest on corporate debt and the results of certain property and casualty business in run-off, including CNA Re and A&EP.

The following table summarizes the results of operations for the Corporate & Other segment, including intersegment eliminations.

The following table summarizes the gross and net carried reserves for Corporate & Other.