Travelers 2009 Annual Report Download - page 143

evaluation. The estimation method(s) chosen are those that are believed to produce the most reliable
indication at that particular evaluation date for the claim liabilities being evaluated.
In most cases, multiple estimation methods will be valid for the particular facts and circumstances
of the claim liabilities being evaluated. This will result in a range of reasonable estimates for any
particular claim liability. The Company uses such range analyses to back test whether previously
established estimates for reserves at the reporting segments are reasonable, given subsequent
information. Reported values found to be closer to the endpoints of a range of reasonable estimates
are subject to further detailed reviews. These reviews may substantiate the validity of management’s
recorded estimate or lead to a change in the reported estimate.
The exact boundary points of these ranges are more qualitative than quantitative in nature, as no
clear line of demarcation exists to determine when the set of underlying assumptions for an estimation
method switches from being reasonable to unreasonable. As a result, the Company does not believe
that the endpoints of these ranges are or would be comparable across companies. In addition, potential
interactions among the different estimation assumptions for different product lines make the
aggregation of individual ranges a highly judgmental and inexact process.
Property-casualty insurance policies are either written on a claims-made or on an occurrence basis.
Claims-made policies generally cover, subject to requirements in individual policies, claims reported
during the policy period. Policies that are written on an occurrence basis require that the insured
demonstrate that a loss occurred in the policy period, even if the insured reports the loss many years
later.
Most general liability policies are written on an occurrence basis. These policies are subject to
substantial loss development over time as facts and circumstances change in the years following the
policy issuance. The occurrence form, which accounts for much of the reserve development in asbestos
and environmental exposures, is also used to provide coverage for construction general liability,
including construction defect. Occurrence-based forms of insurance for general liability exposures
require substantial projection of various trends, including future inflation and judicial interpretations
and societal litigation dynamics, among others.
A basic premise in most actuarial analyses is that past patterns demonstrated in the data will
repeat themselves in the future, absent a material change in the associated risk factors discussed below.
To the extent a material change affecting the ultimate claim liability is known, such change is quantified
to the extent possible through an analysis of internal company data and, if available and when
appropriate, external data. Such a measurement is specific to the facts and circumstances of the
particular claim portfolio and the known change being evaluated. Significant structural changes to the
available data, product mix or organization can materially impact the reserve estimation process.
Informed judgment is applied throughout the reserving process. This includes the application of
various individual experiences and expertise to multiple sets of data and analyses. In addition to
actuaries, experts involved with the reserving process also include underwriting and claims personnel
and lawyers, as well as other company management. Therefore, management may have to consider
varying individual viewpoints as part of its estimation of loss reserves. It is also likely that during
periods of significant change, such as a merger, consistent application of informed judgment becomes
even more complicated and difficult.
The variables discussed above in this general discussion have different impacts on reserve
estimation uncertainty for a given product line, depending on the length of the claim tail, the reporting
lag, the impact of individual claims and the complexity of the claim process for a given product line.
Product lines are generally classifiable as either long tail or short tail, based on the average length
of time between the event triggering claims under a policy and the final resolution of those claims.
Short tail claims are reported and settled quickly, resulting in less estimation variability. The longer the
131