PhD Defense Snorre Jallbjørn
Title: Multi-Population Mortality Models and Scenario-Based Projections
Abstract:
This thesis covers five topics related to mortality modelling and forecasting with two overarching themes: (coherent) multi-population models and scenario-based projections. In the first part of the thesis, we study desirable model properties from the practitioner’s point of view. We present a framework in which multiplicative frailty can be used with stochastic mortality models, and we apply the methodology in a case study of the SAINT mortality model used by the Danish Labour Market Supplementary Pension Fund (ATP). Next, we show that cointegration-based mortality models have more to offer than assuring non-diverging forecasts, and we highlight the limitations of cointegration when applied to models only identifiable under certain identification constraints. Following this, we propose a novel approach to analyzing the global properties of the life expectancy sex gap implied by coherent models, using which we challenge the status of coherence as a universally desirable property. In the second part of the thesis, we formulate a generic causal mortality model in the framework of potential outcomes, which facilitates a discussion of interventions and their direct and indirect effects on forecasts. We show by example the assumptions and data needed to operationalize an empirical analysis. Finally, we introduce the basic ideas for how causal models can be estimated and used to answer interventional queries when data consists of density estimates on marginal distributions observed across multiple populations.
Supervisors:
Affiliate Professor Søren Fiig Jarner, heukno, ATP and University of Copenhagen
Professor Niels Richard Hansen, University of Copenhagen
Professor Mogens Steffensen, University of Copenhagen
Assessment committee:
Associate Professor Bo Markussen (chairperson), University of Copenhagen
Associate Professor Malene Kallestrup-Lamb, Aarhus University
Adjunct Professor Christian Bressen Pipper, LEO Pharma and University of Southern Denmark