A New Fourier Approach under the Lee-Carter Model for Incorporating Time-Varying Age Patterns of Structural Changes

The prediction of future mortality improvements is substantial importance for areas such as population projection, government welfare policies, pension planning and annuity pricing. Lee-Carter model one the widely applied models proposed to capture predict trend in reductions. However, some studies have identified presence structural changes historical data, which makes forecasting performance sensitive calibration period. Although attention has been paid investigating time or period effects shifts, potential time-varying age patterns are often overlooked. This paper proposes a new approach that applies Fourier series with parameters sensitivity factor study evolution effects. Since modelling separated from effects, can incorporate these two sources into predictions. Our backtesting results suggest shifts present not only index over time, but also those variations at different ages.