Feasibility of Singular Spectrum Analysis in the Field of Forecasting Mortality Rate

Rahim Mahmoudvand,Fatemeh Alehosseini,Mohammad Zokaei

doi:10.6339/jds.201310_11(4).0010

Abstract

The Lee-Carter model and its extensions are the most popular methods in the field of forecasting mortality rate. But, in spite of introducing several different methods in forecasting mortality rate so far, there is no general method applicable to all situations. Singular Spectrum Analysis (SSA) is a relatively new, powerful and non parametric time series analysis that its capability in forecasting different time series has been proven in the various sciences. In this paper, we investigate the feasibility of using the SSA to construct mortality forecasts. We use the Hyndman-Ullah model, which is a new extension of Lee-Carter model, as a benchmark to evaluate the performance of the SSA for mortality forecasts in France data sets.

Highlights

Mortality rate is of the most prominent demographic indicators that national statistical offices and United Nations routinely consider in their reports
We demonstrate the feasibility of Singular Spectrum Analysis (SSA) for forecasting mortality rates using one application involving demographic data age specific mortality rates
Annual French mortality rates (1899-2001) for single years of age were obtained from the Human Mortality Database

Summary

Introduction

Mortality rate is of the most prominent demographic indicators that national statistical offices and United Nations routinely consider in their reports. Mortality rates have many applications in governmental agencies and insurance companies. Insurance companies and social security systems use the projections of future mortality rates in order to construct life tables and determine annuity prices. We notice that the human mortality rate has declined substantially over the last century by progress in the health systems which cause to additional stress in support-systems for the elderly, such as pension provision. This change affects on the population size and structure, as well as social security systems.

Objectives

Results

Conclusion