Comparison Of Six Stochastic Mortality Models And Forecasting Mortality - Report Example

These models have been useful in the development of statistical foundations. Most of the methods used in the forecasting process are:

1. Extrapolative, implying that the forecasts utilize the regularities found in the trends over time and the age patterns. Yao, Titus, & MacDonald, (2001) argued that the relatively simple aggregate measure extrapolations such as the life expectancy and the traditional an included in this particular approach.
2. Mortality forecast; which uses the structural models of mortality. This approach is referred to as the explanation approach. The explanation approach uses the mortality data from the causes of death. It is, therefore, possible to identify the exogenous variables and one can also get to know the variables by the use of the explanation approach (Brunelli, Morgan-Hughes, Refai, Salati, Sabbatini, & Rocco, 2007).
3. Expectation approach; the forecasts in this model have their basis on the expert’s subjective opinions which have different degrees of formality.

The models vary according to the main elements that include various sources of randomness in the driving mortality improvement, assumptions of smoothness, and age. According to Yang, Yue, & Huang, (2010), some studies have been made in the comparisons of various models. The models are strictly compared to the variants of Lee-Cartel. This model out of sample performance forecast was compared to; the Booth–Maindonald–Smith variants and the Lee–Miller to a multifactor model in a new class. The comparison of these models was done by Ullah and Hyndaman (2007).  

According to Whipple, Link, Garrison, & Fogarty, (2000) over the last three consecutive years, CMI compared some models to others. The Lee- Carterodel was compared to the P-the splines models, Haberman model, and Renshaw model.

The common mortality models are the P-splines models, Haberman model, and Renshaw model. These models have been analyzed into various features according to recent writers. Some authors in recent years pay much focus to qualitative and quantitative comparisons of stochastic mortality models. In these models, several criteria have been applied in the mortality rate forecast, these criteria include implementation of ease-transparency, quality fit, parsimony a d quality fit which is given by the Bayes information.