Techniques for Reserving a Personal Lines Book of Business
The correct method applied to the correct data
The value of any method is dependent on the quality of the underlying data and as such much work can be done prior to analysis to break the data down into homogenous groups based on the class of business, the risk mix of the book and the nature of the claims (as one might imagine, a relatively short-tailed theft claim on a motor policy would be expected to behave in a completely different manner to a bodily injury claim, which could take up to 6 years to settle).
Standard actuarial techniques, such as the Chainladder method, can then be applied to claim numbers, claim payments and incurred amounts to obtain initial projections of claims development, based on the assumption that historical development patterns are representative of how we’d expect the claims to develop in the future. It should be noted that regardless of the amount of work that goes into preparing the data, no method should be applied blindly or without understanding both the drivers and the impact of the projection.
Where ‘large’ claims are encountered, development patterns can become increasingly volatile, reducing the validity of the Chainladder approach. Capping these claims at a suitable level allows the analyst to project a more stable development pattern, with the claims in excess of the cap studied independently and then added back into the full analysis.
What’s driving the results?
Diagnostic measures such as claim frequency, the average cost of claims and the burning cost are used continually to establish the reasonableness of the estimates produced by any method, and reporting these statistics on a detailed basis helps account managers understand the underlying drivers of performance and establish trends within their data. Where data is scarce or the experience immature these measures of performance can be trended directly or used in conjunction with other techniques, such as the Cape Cod or Bornhuetter-Ferguson methods.
Benchmarking
In the cases where historical experience is unavailable or the volume of business small, benchmarking can be employed; either by providing a development pattern to ultimate (of paid claims, for instance) or by selecting expected values for the performance diagnostics mentioned above, based on market knowledge and familiarity with similar books.
Reinsurance and Large Claims
The longer-tailed nature and increased volatility displayed by large claims present a unique problem to any reserving analyst. Our innovative approach at EMB involves analysing large claims on both an aggregate level and on an individual claims level.
Our ‘capping’ technique allows us to independently model claims of similar amounts using standard actuarial techniques. Once a result for the gross ultimate claims cost has been achieved, the problem of then applying a complicated reinsurance programme is presented. Modelling large claims using a stochastic individual claims model within our financial simulation software, Igloo Professional, allows the analyst to more accurately assess the risk relating to these claims, as well as explicitly quantifying the impact of a reinsurance programme and understanding the high volatility relating to these claims.
We can parameterise our stochastic model using results from our aggregate gross claims analysis, benchmarks taken from our 3rd IUA Bodily Injury Study, the historical large claims experience of the account and benchmarks of frequency and severity taken from the behaviour of the attritional claims.
Projecting on an individual claims level allows the user to correctly deal with complicated reinsurance programmes without having to ignore or make assumptions relating to issues such as varying excesses and limits, indexation and aggregate deductibles.
The Benefits of Good Reserving
What EMB can Provide
Motor Market Commentary
