Login
- Document type:
- Masterarbeit
- Author(s):
- Euthum, Maximilian
- Title:
- Multi-Population Mortality Models - A Comparison via a Socio-Economic Index of Deprivation on Italian Population
- Translated title:
- Sterblichkeitsmodellierung mehrerer europäischer Populationen- Ein Vergleich.
- Abstract:
- In recent years, modelling of human mortality rates became increasingly important in actuarial science. Mortality is not the same for everyone and depends on several factors beyond the normally chosen criterion of country affiliation. Hence, in this thesis the aim is to model mortality of a region looking at different socio-economic groups. Here, Italy is subdivided in nine units based on socio-economic indicators which are summarised in an index of multiple deprivation. Based on this index, mortality rates are modelled for all nine groups via multi- population mortality models which respect time series structure of mortality rates. On the one hand, models as Li \& Lee, Kleinow and Plat are implemented, on the other hand recurrent neural networks represented by LSTMs and GRUs are applied on the data. The goal of this thesis is to compare these models through several error measures and residual analysis. First, the models are fitted on training data ranging from 1982 to 2014, then, forecasting is performed for the years 2015 to 2018. The strength of this approach lies in the fact that mortality rates are modelled simultaneously for different groups, and hence respects some basic underlying mortality pattern in the whole Italian population, and allows for differences which may appear for different socio-economic units.
- Supervisor:
- Prof. Dr. Matthias Scherer
- Advisor:
- Prof. Dr. Matthias Scherer
- Year:
- 2020
- University:
- Technische Universität München
- Faculty:
- Fakultät für Mathematik
- TUM Institution:
- Lehrstuhl für Finanzmathematik
- Commencing Date:
- 15.11.2020
BibTeX