Related Experiment Videos
Applying Lee-Carter under conditions of variable mortality decline
Heather Booth1, John Maindonald, Len Smith
1Demography and Sociology Program, Australian National University. heather.booth@anu.edu.au
Population Studies
|January 30, 2003
Summary
Modified mortality forecasting improves Australian life expectancy predictions. The enhanced Lee-Carter method better accounts for age distribution and non-linear time trends, reducing forecast errors significantly.
Area of Science:
- Demography
- Actuarial Science
- Biostatistics
Background:
- The standard Lee-Carter method assumes a constant age structure and linear time trends for mortality forecasting.
- Its application to Australian data is challenged by non-linear time trends and evolving age components.
Observation:
- Australian mortality data exhibit significant departures from linearity in time trends.
- The age component in Australian mortality has shown changes over time, violating the invariance assumption.
Findings:
- A modified Lee-Carter method, adjusting the time component for the age distribution of deaths and optimizing the fitting period, improves forecast accuracy.
- This modification better meets the invariance assumption for Australian data.
- The enhanced model yields higher forecast life expectancy and a 50% reduction in forecast error compared to the original method and official projections.
Implications:
- The modified Lee-Carter method offers more reliable mortality forecasts for Australia.
- Improved demographic projections can inform public health policy, pension planning, and resource allocation.
- Further research could explore incorporating age-time interactions for even more nuanced forecasting.