Phd in SCHOOL OF STATISTICAL SCIENCES

Thesis title: Innovative approaches to mortality gaps: emerging trends and methodological contributions

The study of mortality remains one of the central fields of demographic research, yet also one of its most dynamic fields. The continuous rise in life expectancy, often celebrated as one of humanity’s greatest achievements, coexists with persistent inequalities in survival across social groups, sexes, regions, and causes of death. Understanding and modelling such disparities is a crucial challenge for both scientific inquiry and public policy, especially in the context of rapid population ageing and growing heterogeneity. This doctoral dissertation aims to deepen our understanding of how mortality inequalities can be analysed, modelled, and forecasted in a coherent way. Its general goal is twofold: first, to contribute to the empirical understanding of mortality gaps; and second, to develop statistical tools that improve the representation of differences in mortality across populations. To this end, the thesis adopts an integrated approach that combines classical demographic methods with techniques drawn from adjacent disciplines such as applied statistics and econometrics. The research has an explicitly exploratory nature: rather than forming a single analytical continuum, the three papers that constitute the core of the thesis represent independent yet complementary investigations. Each employs distinct analytical strategies, statistical techniques, and data sources, reflecting the conviction that the complexity of mortality processes demands a plurality of methodological perspectives. The first paper introduces a Bayesian Age–Period–Cohort (APC) model for analysing the gender gap in mortality, which embeds the structural dependence between male and female mortality rates and produces more coherent forecasts than traditional formulations. The second presents a stochastic model based on the Skellam distribution, designed to describe directly the difference between two Poisson-distributed death counts. Applied to Italian data on the gap between deaths from circulatory diseases and cancer, the two leading causes of death, it delivers more accurate forecasts than classical benchmark models. The third paper examines “Deaths of Despair” (alcohol-, drug-, and suicide-related mortality) in Italy, showing that the trajectories of the three causes do not follow a common long-term trend and display marked regional heterogeneity. Taken together, the three studies position this dissertation as a point of entry into the scientific study of mortality inequalities, showing that modelling mortality gaps remains an area of substantial potential. The first two papers provide methodological contributions that enhance the coherence and predictive performance of existing models, while the third employs statistical modelling as an interpretive tool, clarifying the empirical meaning of Deaths of Despair in the Italian context. Future research will focus on developing coherent multi-population forecasts and on further investigating mortality inequalities by building upon the methods explored and refined throughout this work.