Description: In this paper, we investigate the dynamics of the Lee-Carter mortality index parameter kt. Specifically, we perform statistical hypothesis tests to examine whether the mortality indexes for Canada, England and Wales and the United States are best described by stochastic trends difference stationary models or deterministic trends trend stationary models. The empirical results favor broken-trend stationary models over difference stationary models, which are used in most previous applications of the Lee-Carter method. The results also give strong statistical evidence that the rates of mortality decline for the three populations have significantly accelerated in mid-1970s.

Description: Discussion of Living to 100 Session 6A: Projection and Statistical Modeling of Mortality at Late Age

Description: This is the informal discussion transcript for Session 4B: Mortality Projections from the January 4-6, 2017 Living to 100 Symposium in Orlando.

Description: Discussion of “Health Expectancy” paper and reply from author, Faye Albert. Discussion of Human Behavior: An Impediment to Future Mortality and Discussion of Health, Wealth and Wisdom - Living Long, Living Well: An Actuary Muses on Longevity

Description: Discussion on Living to 100 Session 4B. Papers presented: Typology and Review of Measures of Human Aging, Longevity and Superlongevity with Applications to US Data and Some Implications for US Public Programs and, Mortality Measurements at Advanced Ages: A Study of the Social Security Master Death File

Description: In this paper, we study a survival regression model known as Cox's proportional Hazards model. We assume that the data available are already grouped and develop a minimum distance estimation procedure that is also applicable when the data are also left or right censored. Asymptotic properties of the estimators are established and the numerical implementation of the method based on an iterated reweighted least squares algorithm is discussed.

Description: The Lee and Carter LC method is getting widely adopted for long-run forecasts of age specific mortality rates. That popularity is first due to the model simplicity. Moreover, the LC model has given successful results for various countries e.g. U.S., Chile, G7 countries. However, some difficulties arose when using population data from certain countries such as UK and Australia. In the present paper, the LC model is applied to data from four Nordic countries: Denmark, Finland, Norway and Sweden. The Singular Value Decomposition SVD, the Weighted Least Square WLS and the Maximum Likelihood estimate MLE are used. These approaches give satisfactory results. The appropriate fitting period needs, however, to be well chosen. The properties of the model's parameters are studied using a bootstrap simulation.

Description: Discussion on Living to 100 Session 4A. Papers presented: Inference for Logistic-type models for the Force of Mortality and, Predictive Modeling for Advanced Age Mortality

Description: In this paper, we investigate the dynamics of the Lee-Carter mortality index parameter kt. Specifically, we perform statistical hypothesis tests to examine whether the mortality indexes for Canada, England and Wales and the United States are best described by stochastic trends difference stationary models or deterministic trends trend stationary models. The empirical results favor broken-trend stationary models over difference stationary models, which are used in most previous applications of the Lee-Carter method. The results also give strong statistical evidence that the rates of mortality decline for the three populations have significantly accelerated in mid-1970s.

Description: Recurrence relations arise naturally in life contingencies for 2 reasons - the aggregate law of mortality, which relates the distribution of the future life times of 2 insureds, if one of them and the other is that most life insurance products can be constructed as a portfolio of 2 basic product types. We use the above simple idea and the extra freedom provided by the fractional age distribution to reveal a structure behind some common recursions. The structure has obvious pedagogical value. More interestingly, a software based on it has been implemented which facilitates actuarial computing.