Overview
Piet de Jong is an actuarial scientist at the Department of Actuarial Studies, Macquarie University (Australia). He co-developed the De Jong-Tickle (DJT) LC(smooth) state-space extension of the Lee-Carter mortality forecasting model, which reformulates LC within a general state-space framework estimated via Kalman filtering.
Key Contributions / Features
- De Jong-Tickle LC(smooth) state-space model (De Jong and Tickle 2006): Generalizes the standard LC model by introducing a B-spline design matrix X that constrains the age effects to follow a smooth quadratic form between knots. This reduces the number of a and b parameters below the number of age groups, imposing across-age smoothness and addressing the jagged age profiles produced by standard LC. The state-space formulation enables maximum-likelihood estimation via Kalman filtering and smoothing, provides exact forecast error calculations, and naturally handles missing data. The fitting period is restricted to 1950+ (consistent with LM and HU) to avoid pre-war outliers.
- Multi-country comparison (Booth, Hyndman, Tickle, and De Jong 2006): Co-authored the 10-country evaluation showing that DJT achieves log death rate MAE of 0.150 (tied for best with HU), but no significant LE advantage over simpler methods. The DJT framework's principal advantage is its ability to extend naturally to multi-population and multi-component modeling — a qualitative benefit not captured by single-population MAE metrics.
Related
Sources