Analyzing Reserves: Devil Is In The Details

The U.S. insurance industry recognized approximately $52 billion in past and future asbestos costs at the end of 2003. This is an increase of 50 percent compared to the $35 billion of net loss and expense payments plus reserves recognized at the end of 2001.

The appropriateness of the associated reserve charges totaling $8 billion in 2002 and estimated as $8 billion in 2003?depends upon the level of detail that can be, or has been, reflected in the ground-up exposure-based modeling analysis (see sidebar).

The deeper understanding obtained from gathering the required information and considering each of the elements of the ground-up analysis significantly enhances the overall reserve estimate.

The process can be split into two major sections: estimation of losses for known accounts, and estimation of a provision for unallocated IBNR (not related to a specific account). The provision for known accounts encompasses all future losses for both the modeled and extrapolated accounts, including a provision for non-products losses. The unallocated IBNR should reflect unidentified accounts as well as a provision for unexpected additional emergence on known accounts.

Known Accounts

Companies with strong data storage and collection systems should be able to extract the policy level information for their insured asbestos defendants.

Once these policies have been identified, the calculation of the respective ground-up ultimate loss and expense can be performed, reflecting the current plaintiff filing levels and average settlement values as well as projections of future claim filings, inflation and expense-to-indemnity ratios.

The resulting estimates can then be allocated to policy year and compared to the available coverage for each defendant as identified by the insurer to estimate the reserve across all modeled defendants.

Due to the significant number of potential peripheral defendants, estimates for this group are often estimated for a credible sample size and then extrapolated to the remainder of the group. The sample is often selected on a stratified basis, where the strata consider the defendants? expected loss potential based on reported loss experience or exposed limits. Groups with the greatest exposure are typically sampled more heavily in order to minimize the additional uncertainty introduced through the extrapolation process.

While most asbestos claims relate to products losses, some defendants are also seeking reimbursement for operations claims from their insurers. Several defendants (for example, those with insulation products) have exhausted their products aggregate limits and have sought additional operations coverage. Some have already reached settlements with their insurers on operations claims.

But additional defendants conducted activities that may potentially result in non-products losses. These additional defendants may not yet have filed operations claims, because they may believe that the products liability coverage will be sufficient to cover all of the liabilities, and because the standards of proof are lower in products liability actions. While most major defendants with operations claims have been identified, it is important to review the sampled peripheral defendants for this possibility.

IBNR Losses

The IBNR associated with a ground-up analysis should incorporate a provision for defendants that have not yet filed asbestos claims with their insurer. It may also include a provision for deterioration in liability estimates for the known accounts.

The provision for future insureds with asbestos claims is based on the historical number of newly identified policyholder defendants emerging by calendar year. Meanwhile, the provision for possible deterioration in the liability estimates should consider the sensitivity of the projections to changes in the modeling parameters.

The distribution of IBNR by component has not typically been disclosed by insurers, shedding little light on how much of their IBNR provision relates to the uncertainty in their estimates.

Uncertainty in Reserve Estimates

Asbestos reserve reviews generally produce a wider range of estimates than most lines of casualty insurance. The greater level of uncertainty is driven by many factors, both at the macro level affecting the asbestos litigation as a whole, and at the micro level affecting parameters specific to an insurer's ground-up reserve estimate.

While there is little an insurer can do to control the former, the latter can be evaluated by testing the sensitivity of various parameters used in the insurer's ground-up study (see second sidebar).

While the basic list of factors is not new, the recent deterioration in the litigation environment and corresponding reserve estimates has refocused attention on how various insurers reflect the changes to the environment through specific factors in their reserve analyses. Four micro-level factors are discussed in detail below:

Treatment of the Historical Data:

When reviewing defendant-specific historical data, it is important to appropriately consider recent increased filing levels in certain jurisdictions (such as Mississippi), as well as disease-specific claim settlement values (both past and present). A change from a high concentration of malignant claims to non-malignant claims could significantly distort projections for a given defendant, if it is not interpreted correctly.

Adequacy of the policy search:

Additionally, the capabilities of an insurer to identify all policies for major defendants may have a material impact on results, since each additional policy discovered has the potential of a dollar-for-dollar increase in the insurer's asbestos-related reserve indications (or more if expenses are payable in addition to the stated limits).

Estimated reinsurance recoveries:

Another parameter that may affect the insurer's net asbestos reserves is the anticipated reinsurance recoverable. Ideally, an insurer will model both direct and ceded losses, reflecting the actual terms of its reinsurance protections. However, if the reinsurance program is too complicated to model, the insurer may analyze the relationship of direct, ceded and net losses by category of defendant. A further enhancement would be to analyze the expected reinsurance recoveries for a sample of defendants in each category in order to estimate ceded reserve level. Recently, ceded amounts have received greater scrutiny.

? Provision for possible deterioration:

Lastly, the insurer may incorporate a provision for additional deterioration to account for some of the macro-level conditions that might affect their liabilities in the future. This can be accomplished by testing the sensitivity of the projections to various parameters to determine an overall range of results. For many insurers, the carried IBNR reserve incorporates a provision for this additional uncertainty, providing a cushion for unanticipated deterioration in results.

Important Unknowns

Quantification of the effects of bankruptcies further complicates the reserving process. As more defendant companies file for bankruptcy, the remaining solvent defendants see higher settlement demands due to the decreasing shares of settlement awards by the bankrupt companies.

Many recent bankruptcy filings have been "pre-packaged" meaning that the debtor negotiates its reorganization plan with its creditors and the plan is voted upon prior to filing. Under these pre-packaged bankruptcies, the insurers generally have no representation in the negotiations, and have often been faced with accelerated and elevated demands for payment from the defendants.

A second complicating factor is the increasing number of newly identified peripheral defendants that are just beginning to access their insurance coverage. Litigation against these newer defendants is immature and, in many cases, coverage has not been fully identified. The unallocated IBNR provision needs to account for potential deterioration in the reserve estimates associated with these new defendants.

The overall effect of all of these issues is still uncertain. Pre-packaged bankruptcies are of significant concern to insurers, although there have been some encouraging developments. (For example, a judge denied confirmation of the AC&S pre-packaged bankruptcy filing, finding that it was not proposed in good faith and unfairly favored one claimant over another).

Continued increases in plaintiff filings and settlement demands could materially affect the reserves posted by insurers and reinsurers. Much depends on the outcome of federal reform efforts or the ability of various jurisdictions to enact individual reforms.

In the meantime, insurers should closely monitor their asbestos reserves, comparing actual loss emergence to the ultimate provisions for known defendants and the level of newly identified defendants accessing coverage for the first time to the IBNR provision developed in the reserve review. Continual attention and re-review of important parameters are required in order to assure that reserves are adjusted appropriately.

Mike Angelina is a principal in the Philadelphia office of the Tillinghast business of Towers Perrin (mike.angelina@towersperrin.com). Jenni Biggs is a principal in the St. Louis office (jenni.biggs@towersperrin.com).

Reproduced from National Underwriter Edition, June 25, 2004. Copyright 2004 by The National Underwriter Company in the serial publication. All rights reserved.Copyright in this article as an independent work may be held by the author.