Version 1.4 of the Projections Toolkit
Date: Sep 20, 2010
Longevitas Ltd is pleased to announce the production release of v1.4 of the Projections Toolkit. This version contains a number of improvements and new features including:
- Addition of sample paths for drift models
- Addition of sample paths for ARIMA (p,1,q) models.
- User control of knot positioning in models using splines for smoothing.
- User control of optimisation method, including a grid search based on simulated annealing for challenging functional shapes.
- Extension to Excel-based projection explorer to support more charts and a broader range of Excel versions
- Extensions to upper projections limits and smoothing overrides
- TAS-R information added to projections PDF report
- Ability to reuse existing projection models and piggyback projections
- Colour coding for large positive and negative residuals in Excel residuals downloads
- An Excel-based projection explorer, including automatic calculation of reduction factors and improvement rates using your choice of ICA/Solvency II stress level.
- Excel-formatted downloads including a multi-worksheet format for qx stress scenarios.
- Changes to handling of over-dispersion, including addition of over-dispersion parameter to models in the smooth Lee-Carter family.
- Generation of sample paths for smoothed models using penalty projections.
- Addition of drift-model variants for time-series models in the Lee-Carter family.
- Residual file download and large residual weighting.
- 2D P-spline models extended to allow fitting of an overdispersion parameter, allowing for extra-Poisson variation caused by, for example, period effects or duplication
- Addition of base mortality smoothing override to Lee-Carter Fully smoothed and Smooth alpha and beta (time-series) models
- Minimum year boundary lowered to 1900 to support population data from wider number of territories
A range of stochastic projection models supported to investigate model risk:
- Extended Lee-Carter family - time series projections (Original, DDE, Gompertz, smooth alpha and beta)
- Extended Lee-Carter family - penalty projections (Currie Richards 2009, Gompertz, Fully-smoothed)
- 2D P-spline family - penalty projections (2D age-period, 2D age-cohort)
- Projections against any age and year range selection to allow for straightforward back-testing
- Piggyback models to investigate basis risk in smaller portfolios
Insurers and pension schemes are under increasing pressure to understand the uncertainty over future mortality improvements. An important tool for this is a stochastic projection model, which can be complicated to fit. The Projections Toolkit makes it easy for an actuarial advisor to fit a wide variety of projection models to explore the uncertainties over future mortality.
The Projections Toolkit has particular application for life insurers and reinsurers under Solvency II in the EU.