Papers and reports

This paper considers cross-validation as an objective and risk-based method for selecting the smoothing parameter in a non-parametric graduation. In addition, the relative merits of two kernel estimators are compared in the context of mortality graduation. Finally, it is well known in the statistical literature that the use of theoreticlly superior kernels is not as important as the choice of bandwidth. Our results support this conclusion, suggesting that the focus on such weights is misguided in the actuarial textbooks on moving weighted averages.

A brief description is given of some problems encountered in designing a model office program. The key to an effective and flexible model lies in the structure which the model imposes on the data. An example of a model used for research is described.

A method of graduation by mathematical formula is extended to cover the simultaneous graduation of more than one comparable experience and is applied to the recent UK mortality experience of smoking and non-smoking assured lives.

A brief description is given of some problems encountered in designing a model office program. The key to an effective and flexible model lies in the structure which the model imposes on the data. An example of a model used for research is described.