The statistical estimation and forecasting of mortality is a critical component of population forecasting and input to the planning process of governments and financial agencies. In this talk I will describe methodologies that we have developed for providing mortality estimation and forecasting for medium- and long-term planning. The COVID-19 pandemic has intensified interest in robust methods for estimating excess mortality. We can use the long-term mortality trends from the annual forecasting model to establish a realistic baseline. Our methodology then focuses on modelling short-term weekly mortality fluctuations relative to the long-term trajectory captured by the annual model. Using data from England and Wales, we identify suitable dynamic harmonic regression models, which are then embedded within a Bayesian framework that coherently accounts for all sources of uncertainty.  This allows for well-calibrated estimates of cumulative excess deaths during the early stages of the pandemic, which we compare with alternative estimates

Jon Forster

Kristan Romano: Kristian.Romano.1@warwick.ac.uk
Gianmarco del Pino: gianmarco.del-pino@warwick.ac.uk