Consideration of Bias in Chain Ladder Estimates

The chain ladder method may be the most commonly used and well‐known approach for estimating ultimate claims. As it is most often employed, the same development pattern is used to project each accident year and its results are generally considered by practitioners to be valid for each accident year. It is the author’s contention that, under this application, the chain ladder method will produce biased projections of the ultimate claims for a single accident year. This paper identifies the sources of the bias and provides the actuary with a tool to understand and compensate for a portion of the bias.