City Research Online

Two-population Mortality Forecasting: An Approach Based on Model Averaging

De Mori, L., Haberman, S. ORCID: 0000-0003-2269-9759, Millossovich, P. ORCID: 0000-0001-8269-7507 & Zhu, R. ORCID: 0000-0002-9944-0369 (2024). Two-population Mortality Forecasting: An Approach Based on Model Averaging. Risks, 12(4), doi: 10.3390/risks12040060


The analysis of residual life expectancy evolution at retirement age holds great importance for life insurers and pension schemes. Over the last 30 years, numerous models for forecasting mortality have been introduced, and those that allow us to predict the mortality of two or more related populations simultaneously are particularly important. Indeed, these models, in addition to improving the forecasting accuracy overall, also enable evaluation of the basis risk in index-based longevity risk transfer deals. This paper implements and compares several model averaging approaches in a two-population context. These approaches generate predictions for life expectancy and the Gini index by averaging the forecasts obtained using a set of two-population models. In order to evaluate the eventual gain of model averaging approaches for mortality forecasting, we quantitatively compare their performance to that of the individual two-population models, using a large sample of different countries and periods. The results show that, overall, model averaging approaches are superior both in terms of mean absolute forecasting error and interval forecast accuracy.

Publication Type: Article
Additional Information: This article has been accepted for publication in Risks by MDPI and it will be available online at
Publisher Keywords: model averaging, mortality forecasting, two-population models, life expectancy, Gini index
Subjects: H Social Sciences > HA Statistics
Departments: Bayes Business School
Bayes Business School > Actuarial Science & Insurance
SWORD Depositor:
[thumbnail of paper.pdf]
Text - Accepted Version
Download (563kB) | Preview


Add to AnyAdd to TwitterAdd to FacebookAdd to LinkedinAdd to PinterestAdd to Email


Downloads per month over past year

View more statistics

Actions (login required)

Admin Login Admin Login