Impact of longevity risk with best fit mortality forecasting model
Abstract
The data used in this research is from the Human Mortality database for
the United States for the period 2000 - 2009. We first begin by fitting
linear, Makeham, cubic smoothing spline and Lee Cater models to the data
to establish which model best fits the data. Parameters are estimated using
linear regression, graphical method and Singular Value Decomposition
methods for the linear, Makeham and Lee Carter models respectively. By
use of goodness of fit tests conducted using chi square, Cramer Von Mises
criterion, Kolmogorov Smirnov and Anderson Darling tests, a best fit model
to the data is established. Using standard error measures, best forecast
model for the data is identified among the forecast methods, Cubic
smoothing spline, ARIMA (Auto Regressive Integrated Moving Average)
models with and without drift. Mortality rates based on the best fit model
are forecasted to five year horizon using best forecasting method to the
data. These rates are used to check if there has been a decline in mortality
rates over the years. Impact of mortality decline (longevity risk) is then
illustrated by calculation of Actuarial Present Values (APVs) of whole life
annuity of a 60 year old male over the years.
Publisher
University of Nairobi
Description
Thesis