The aim is to highlight some questions that are of practical interest and that we believe could be fruitful areas for further research. Studies in non-life insurance of claims ratio data, show correlation patterns between risk-years and between different classes of insurance. Empirical studies, some progress in international classification of risks, and more work on models would no doubt be helpful for solvency control and methods for insurers' allocation of capital. In particular, we point out the usefulness of more realistic statistical distributions in this context, which is illustrated by analysis of some Swedish data.