Dataset of an actual life-risk insurance portfolio

The foundation of the insurance business is built on data, the latter being one of the most valuable assets of any insurer. In fact, the risk structure to which an insurance company is exposed can actually be deduced by reviewing its customer database. It is not surprising, therefore, that access to real insurance datasets is very limited. This paper introduces and describes a dataset corresponding to a cross-section extraction of a real life-risk insurance portfolio. The dataset contains information on 76,102 policies and a total of 15 variables, including the capital at risk, the genders and dates of birth of the insured, and the effective and renewal dates of their policies. This dataset can be used both in teaching and in research. Combined with external life tables, the data available in the dataset can be used to build and compare pricing systems, to evaluate marketing strategies, in portfolio analysis, for calculations required by Solvency II regulations or for market benchmarking analysis. For example, the data from this dataset have been used in Pavía and Lledó [1] to compare the classic pricing methodology based on annual life tables with a new pricing methodology based on life tables with less than annual periodicity Pavía and Lledó [2], specifically quarterly, and in Lledó et al. to demonstrate the impact that using a new methodology to manage catastrophic risks in life insurance would have in terms of solvency capital requirements.