Announcement: Access the latest SOA updates on COVID-19 at

Due to unusually high email volume in Customer Service, you may experience a longer response time than usual. We apologize for any inconvenience this may cause.

Quebec City Spring Life Meeting: Session 19 Estimating Mortality Risk Using Predictive Modeling

Quebec City Life Spring Meeting: Session 19 Estimating Mortality Risk Using Predictive Modeling

By Steve Craighead


On Monday, June 16, Philip Adams from Munich American Reassurance Company and I spoke regarding the modeling of mortality using predictive modeling. I spoke briefly about various statistical procedures being used in predictive modeling such as Generalized Linear Models (GLM), Generalized Additive Models (GAM) and the use of Projection Pursuit Regression (PPR).

Adams, a new ASA, was the primary speaker and he spoke about his modeling of mortality based on the raw data that underpins the current SOA 2008 VBT. He first discussed the PPR model that was used regarding the modeling of the ultimate tables and then went on to discuss the more complex (but more accurate) GLM model(s) that he constructed regarding the select tables. The last part of his presentation was a comparison of his model results against the prior and current VBT tables.

The session was well attended and there were several questions throughout the talks and at the end especially regarding the differences between the separate models.