Sylvain Goulet & Caesar Balona: The Canadian Institute of Actuaries (CIA) has outsourced the construction of the latest Canadian life insurance table, the CIA2014. The table was graduated using a traditional construction approach, namely Whittaker-Henderson interpolation. Sylvain and his partners at QED in South Africa were asked by the CIA to investigate and develop alternative construction methods, including the use of General Additive Model (GAM) and Neural Network Model (NNM). We will explain the advantages and disadvantages in using these methods. We will further make comparison to the published CIA2014 table. The use of alternative methods, such as a Neural Network Model (NNM), allows the model to construct future mortality tables based on the characteristics of the underlying data. The method also allow the construction of parts of the tables where data may actually be missing. So, the approaches are predictive and not just as a way to construct a static table.

Patrick Nolan: Presentation of an experience study of pension mortality for national pension systems for four Caribbean countries. The experience period spans the calendar years 2011-2018, and the experience data include over 230,000 life-years of exposure and over 7,500 deaths. The analysis includes comparisons to existing mortality bases and analysis of trends by age group and income level.