Selecting LDFs require some experience and a number of statistical analytics and model distributions can be used to select it for a particular LOB(Line of Business). Each LOB has different features like short tail or long tail, data available, market point of view, and others.
Selecting Loss Development Factors (LDFs)
To select DFs to be used for projecting claims for future years, any of the following can be considered
- average of all the development factors can be taken or
- as a conservative measure, the largest development factor for any development year can be taken or
- the weighted average can be taken.
These all were already explained in the previous exercise.
Selecting loss development factors for unusual cases.
It is also possible that altogether different patterns are selected where they are not applicable. Like a new organization would go for benchmarked factors or may not use any development method at all or a new location for a multinational Insurance company, so they can use development factors of another region with a little tweak or delayed.
You will come to know in upcoming exercises how a new general insurance company usually calculate their IBNR’s when no development factors are available with them.
The selected age-to-age factor (a.k.a. claim development factor or loss development factor) represents the growth anticipated in the subsequent development interval.
Selections are based on a review of the historical claim development data, the age-to-age factors, the various averages of the age-to-age factors, and a review of the prior year’s claim development factor selections.
Selecting Benchmark LDFs
When the credibility of the insurer’s historical experience is limited, there may be a need to supplement the experience with benchmark data.
Possible benchmark includes:
- In case, the insurer is exposed to similar LOBs with matching case handling process
- LDFs from the industry which related to similar products and are comparable
- using LDFs of other geographical location with a delay
When using benchmarks, there may be significant differences between the line of business being analyzed and the benchmark with regard to claims practices, policy coverages, underwriting, geographic mix, claim coding, policyholder deductibles and/or limits, legal precedents, etc.
The following characteristics are considered when selecting DFs for a particular class.
- Smoothened factors
There should be a smooth progression of individual Loss development factors and average factors across development periods.
In case, there is a development period that is adverse, it is generally removed from within the calculation. Refer to Example 2.
A steadily decreasing incremental development from valuation to valuation - Stability of age-to-age factors for the same development period.
As you see in the example, the development for 2-3 and 3-4 years should be almost the same with a little variance within each development interval (i.e. down the columns). - Changed Patterns and Historical experience
There can be changes in claims practices, policy coverages, underwriting, geographic mix, claim coding, policyholder deductibles and/or limits, legal precedents in some years. The factors must be adjusted to involve a drastic change in any of these internal or external/ environmental/ regulatory factors.
Has the insurer’s book of business and insurer operations changed over time?
Have the effects of changes in external factors manifested themselves in the reported claims experience? - The credibility of the experience.
We have read credibility Theory. There may be a time where we provide weights or use experience for a given AY and age.
There may also be a time when some Benchmark DFs from the market may also be considered where credibility is lacking.