This paper uses longitudinal data analysis techniques to study the problem of non-life loss reserving. This forecasts a liability for claims which have occurred, but have not yet been settled and closed. The data is a publicly available set of information from several hundred insurance companies over a period of twenty year.