The claims reserves form the main part of the liabilities of a general insurance company. They exist in order that money is set aside to pay claims for business that has been written (and premium income received). It is often the case that delays occur in the payment of claims, and it is important that the company takes account of this, and has the money required to pay the claims whenever they occur. In recent years the emphasis has been on the possible variability of these reserves, which may change as the estimates change of the amounts that will have to be paid in claims (or the actual payments themselves turn out to be different from what they were predicted to be). For capital modelling and solvency purposes, it is this uncertainty that is often the focus of attention - not only is it necessary for companies to set aside the amount that they expect to have to pay, they also have to make some provision to recognise that the actual amount they have to pay may be higher than expected.

There are many methods for claims reserving, of which one of the most well-known is the chain-ladder technique. When this is applied in practice, it is often the case that the estimates obtained directly from the data are not used without some adjustments and alterations being made. These changes are usually done using knowledge of the characteristics of the business and what has happened in the past, and what may change in the future. For example, it may be the case that special circumstances have led to a certain pattern in the delays of claims being paid in the past that are not applicable for the future. The use of this type of information can be straightforward when deterministic methods are used to set the reserve, but it is more problematic when using stochastic methods to assess the uncertainty associated with the reserve. This is where Bayesian methods can be useful, and this presentation discusses how these methods can be used for claims reserving. There are a number of papers associated with this, which are available as pdf downloads below.