The rising life expectancy is one of the biggest challenges the insurance industry has ever faced. This work aims to contribute to a better understanding of the past development of the mortality rates by first, disaggregating the mortality rates by cause of death, and second, by studying the relations between the cause-specific mortality rates through cointegration techniques. This approach helps to complement the current knowledge on cause-of-death mortality dependence that is essential for setting and testing mortality assumptions and scenarios. The present thesis provides answers to the raised question through tree essays, each corresponding to a chapter.
Chapter 1: Short- and Long-Term Dynamics of Cause- specific Mortality Rates Using Cointegration Analysis
This chapter is based on the following article:
S´everine Arnold & Viktoriya Glushko (2021) Short- and Long-Term Dynamics of Cause- Specific Mortality Rates Using Cointegration Analysis, North American Actuarial Jour- nal, https://doi.org/10.1080/10920277.2021.1874421
This paper applies cointegration analysis and vector error correction models to model the short- and long-run relationships between cause-specific mortality rates. We work with the data from five developed countries (USA, Japan, France, England and Wales, and Australia) and split the mortality rates into five main causes of death (Infectious&Parasitic, Cancer, Circulatory diseases, Respiratory diseases, and External causes). We successively adopt the short- and long-term perspective, and analyze how each cause-specific mortal- ity rate impacts and reacts to the shocks received from the rest of the causes. We observe that the cause-specific mortality rates are closely linked to each other, apart from the External causes that show an entirely independent behavior, and hence, could be consid- ered as truly exogenous. We summarize our findings with the aim to help practitioners set more informed assumptions concerning the future development of mortality.
Chapter 2: Cause-Specific Mortality Rates: Common Trends and Differences
This chapter is based on the following article:
S´everine Arnold & Viktoriya Glushko (2021) Cause-specific mortality rates: Common trends and differences, Insurance: Mathematics and Economics, Volume 99, 2021, Pages 294-308, ISSN 0167-6687, https://doi.org/10.1016/j.insmatheco.2021.03.027.
In this paper, we continue to study the past development of cause-specific mortality. We work with the data from five developed countries (USA, Japan, France, England and Wales, and Australia), two sexes, and split the mortality rates into five main groups of causes of death (Infectious&Parasitic, Cancer, Circulatory diseases, Respiratory diseases, and External causes). As it was shown in Arnold and Sherris (2016), these time series of cause-specific mortality rates are cointegrated and so, there exist long-run equilib- rium relationships between them. While the previous research focused on the stationary part of the system of cause-specific mortality rates, in the present paper we study its non-stationary part. For this, we explicitly extract common stochastic trends from the original variables and compare them across the different datasets. By testing cointe- gration assumptions about these trends, we are able to get a better representation and understanding of how cause-specific death rates are evolving. We believe that common patterns emerging from such analysis could indicate a link to more fundamental biological processes such as aging.
Chapter 3: Forecasting Cause-Specific Mortality Rates Using the Insights from the Cointegration Analysis
This chapter is based on the following working paper:
S´everine Arnold & Viktoriya Glushko (2021) Forecasting Cause-Specific Mortality Rates Using the Insights from the Cointegration Analysis. working paper.
Much like the all-cause mortality, cause-specific mortality rates in countries with sim- ilar socio-economic characteristics are likely to follow comparable development patterns. They are also not expected to substantially diverge in the future. We propose to assess the coherence of the past country-specific experiences by the means of the cointegration analysis applied to the mortality time trends extracted by country and cause of death. Indeed, should the time trends of two countries be cointegrated, this would indicate there
existed a long-run stationary relation between them, and so, the mortality patterns of these countries were linked to each other in their long-term development. We analyze the data from five developed Western European countries (France, Italy, Netherlands, Spain, and England and Wales), two sexes, and split the mortality rates into five main groups of causes of death (Infectious&Parasitic, Cancer, Circulatory diseases, Respiratory diseases, and External causes). We observe that while in many cases the cause-specific time trends are indeed cointegrated, this is not always the case in spite of the closeness of the studied countries. Further, once we include the countries having the cointegrated time trends in a multipopulational context, such as the Li-Lee mortality model, the forecast results are improved in comparison with the basic Lee-Carter approach.