The Hartford 2008 Annual Report Download - page 82

Table of Contents
Frequency / Severity methods. Historical data is used to develop claim count development patterns and those patterns are
applied to the number of current reported claims to estimate ultimate claim counts. Estimated ultimate claim counts are
multiplied by an estimated average severity (i.e., an average cost per claim) to calculate estimated ultimate losses. Average
severity is estimated by fitting historical severity data to a trend line and making assumptions about how the current
environment would affect claim severity. In making assumptions about the current environment, industry data is used where
such data is available and appropriate.
The advantage of frequency / severity techniques is that frequency estimates are generally easier to predict and external
information can be used to supplement internal data in making severity estimates.
Expected Loss Ratio method. Loss ratios for prior accident years are used to determine the appropriate expected loss ratio for
the current accident year after applying anticipated changes in rates, pricing and loss costs. The current accident year
expected loss ratio is multiplied by earned premium to calculate estimated ultimate losses.
Expected Loss Ratio techniques are useful for early periods of maturity on long-tailed lines of business, where very little
paid or reported loss information is available.
Bornhuetter-Ferguson method. This method is a combination of the expected loss ratio method and the paid development or
reported development method, where the paid or reported loss development method is given more weight as an accident year
matures.
Berquist-Sherman method. This method is used in cases where historical development patterns may be inappropriate for use
in estimating ultimate losses of recent accident years. Under this method, the pattern of historical reported losses is adjusted
for changes in case reserve adequacy and the pattern of historical paid losses is adjusted for changes in claim settlement
rates.
For all lines of business, variations of the above methods are used. Examples of variation within the paid and reported
development methods include:
• The accident period used may vary (e.g., year, quarter, or month);
• The Company may analyze the data by coverage (e.g., bodily injury separate from property damage);
• There may be adjustments for unusual loss activity;
• For ALAE, the Company uses patterns of the relationship between paid ALAE and paid losses.
Examples of variation within the frequency /severity methods include:
• For one sub-set of professional liability business, management estimates frequency, not through historical claim count
development, but through an analysis of the securities class actions filed and policy listings;
• For some methods, management projects severity on only open claims;
• In the commercial liability lines, the Company performs the frequency / severity technique only on claims over a
certain size;
For each line of business, certain methods are given more influence than other methods. The discussion below gives a
general indication of which methods are preferred by line of business. Because the actuarial estimates are generated at a
much finer level of detail than line of business (e.g., by distribution channel, coverage, accident period), this description
should not be assumed to apply to each coverage and accident year within a line of business. Also, as circumstances change,
the methods that are given more influence will change. For example, for Personal Lines auto liability claims, reported
development techniques are currently given less emphasis in making estimates for recent accident years because case
reserving practices have been changing in the recent past. If case reserving practices become more stable, reported
development techniques may be given more weight.
Property and Auto Physical Damage. These lines are fast-developing and paid and reported development techniques are
used. The Company performs and relies primarily on reported development techniques and frequency/severity and
Bornhuetter-Ferguson techniques for the most immature accident months.
Auto Liability — Personal Lines. For auto liability, and bodily injury in particular, the Company performs a greater number
of techniques than it does for property and auto physical damage, including paid and reported development methods,
frequency/severity approaches, and Berquist-Sherman techniques. The Company generally uses the reported development
method for older accident years and the frequency/severity and Berquist-Sherman methods for more recent accident years.
Recent periods are heavily influenced by changes in case reserve practices and changing disposal rates; the
frequency/severity techniques are not affected as much by these changes and the Berquist-Sherman techniques specifically
Source: HARTFORD FINANCIAL S, 10-K, February 12, 2009