Longevity risk significantly impacts the reserve adequacy ratio of annuity issuers, thereby reducing product profitability. Effectively managing this risk has thus become a priority for insurance companies. A natural hedging strategy, which involves balancing longevity risk through an optimised portfolio of life insurance and annuity products, offers a promising solution and has attracted considerable academic attention in recent years. In this study, we construct a realistic portfolio scenario comprising annuities and life insurance policies across various ages and genders. By applying Cholesky decomposition, we transform the portfolio into an uncorrelated linear model. Our objective function minimises the variance in portfolio value changes, allowing us to explore the impact of mortality on longevity risk mitigation through natural hedging. Using actuarial mathematics and the Bayesian MCMC algorithm, we analyse the factors influencing the hedging effectiveness of a portfolio with minimised variance. Empirical findings indicate that the optimal life-to-annuity ratio is influenced by multiple factors, including gender, age, projection period, and forecast horizon. Based on these findings, we recommend that insurance companies adjust their business structures and actively pursue product innovation to enhance longevity risk management.

This paper presents an actuarially oriented approach for estimating health state utility values using an enhanced EQ-5D-5L framework that incorporates demographic heterogeneity directly into a Generalised Linear Model (GLM). Using data from 148 patients with Stage IV non-small cell lung cancer (NSCLC) in South Africa, an inverse Gaussian GLM was fitted with demographic variables and EQ-5D-5L domain responses to explain variation in visual analogue scale (VAS) scores. Model selection relied on Akaike Information Criterion, Bayesian Information Criterion, and residual deviance, and extensive diagnostic checks confirmed good calibration, no overdispersion, and strong robustness under bootstrap validation. The final model identified age, gender, home language, and financial dependency as significant predictors of perceived health, demonstrating that utility values differ meaningfully across demographic groups. By generating subgroup-specific estimates rather than relying on uniform value sets, the framework supports more context-sensitive cost-effectiveness modelling and fairer resource allocation. Although developed in the South African NSCLC setting, the methodology is generalisable and offers actuaries and health economists a replicable tool for integrating population heterogeneity into Health Technology Assessment, pricing analysis, and value-based care.

An increasing number of reports highlight the potential of machine learning (ML) methodologies over the conventional generalised linear model (GLM) for non-life insurance pricing. In parallel, national and international regulatory institutions are accentuating their focus on pricing fairness to quantify and mitigate algorithmic differences and discrimination. However, comprehensive studies that assess both pricing accuracy and fairness remain scarce. We propose a benchmark of the GLM against mainstream regularised linear models and tree-based ensemble models under two popular distribution modelling strategies (Poisson-gamma and Tweedie), with respect to key criteria including estimation bias, deviance, risk differentiation, competitiveness, loss ratios, discrimination and fairness. Pricing performance and fairness were assessed simultaneously on the same samples of premium estimates for GLM and ML models. The models were compared on two open-access motor insurance datasets, each with a different type of cover (fully comprehensive and third-party liability). While no single ML model outperformed across both pricing and discrimination metrics, the GLM significantly underperformed for most. The results indicate that ML may be considered a realistic and reasonable alternative to current practices. We advocate that benchmarking exercises for risk prediction models should be carried out to assess both pricing accuracy and fairness for any given portfolio.

The goal of the Paris Agreement is to prevent global temperatures from rising by more than 2°C above pre-industrial levels and pursue efforts to limit them to 1.5°C above pre-industrial levels. This requires a significant reduction in global greenhouse gas emissions and achieving net zero emissions by 2050. Portfolio alignment metrics are forward-looking metrics intended to help investors understand whether their investment portfolios are on track to meet the Paris Agreement goals. They also aim to encourage capital flows towards activities needed for a net zero transition. Since 2020, several metrics have been put forward by industry groups and explored in technical papers. Companies and actuaries have been exploring the practicalities of these metrics and starting to incorporate them into investment reporting and design. But this has not been without key challenges. The Net Zero and Implications for Investment Portfolios working party aims to help actuaries improve their understanding of what net zero means for an investment portfolio and what the key mechanisms are to achieve this, as well as key challenges to date and the outlook for development.

Federated Learning is a novel method of training machine learning models, pioneered by Google, aimed for use on smartphones. In contrast to traditional machine learning, where data is centralised and brought to the model, Federated Learning involves the algorithm being brought to the data, ensuring privacy is preserved. This paper will demonstrate how insurance companies in a market could use this technique to build a claims frequency neural network prediction model collectively by combining and using all of their customer data, without actually sharing or compromising any sensitive information with each other. A simulated car insurance market with 10 players was created using the freMTPL2freq dataset. It was found that if all insurers were permitted to share their confidential data with each other, they could collectively build a model that achieved 5.57% of exposure weighted Poisson Deviance Explained (% PDE) on an unseen sample. However, if they are not permitted to share their customer data, none of them can achieve more than 3.82% exposure weighted PDE on the same unseen sample. With Federated Learning, they can retain all of their customer data privately and construct a model that achieves a similar level of accuracy to that achieved by centralising all the data for model training, reaching 5.34% exposure weighted PDE on the same unseen sample.

The fast-moving field of data science is increasingly permeating into the health and care actuarial sciences. Given this context, the Institute and Faculty of Actuaries set out to form a “techniques in data science in health and care” working party. This working party was tasked with creating a framework for those actuaries working within the health and care domain that would assist them in determining which techniques are appropriate for a project. The framework presented here was developed through a combination of literature review and synthesis of expert opinion from experienced practitioners from diverse backgrounds. The framework offers a structured, itemised approach, serving as a checklist to ensure that all relevant analytics and decisions are considered and documented. Each itemised topic is covered by a summary providing guidance and relevant references for further reading. The checklist follows the natural workflow of a data analytics project, guiding users through each step to prevent omissions and maintain rigour in both analysis, reporting and peer-review. The framework blends relevant analytics elements from actuarial science, data science and epidemiology. We hope the framework will enhance transparency, reproducibility, and comprehensiveness in the reporting and peer-review of health and care data analytics projects.

Health and care (H&C) actuaries are well positioned to benefit from recent advances in data science as machine learning (ML) techniques have become increasingly transparent and accessible. The ML developments allow actuaries to detect complex nonlinear patterns and interactions that are difficult to capture using traditional generalised linear models (GLMs), without sacrificing the clarity and governance advantages that make GLMs central to actuarial practice. Using a large life insurance data set, we demonstrate and appraise three emerging hybrid approaches: interpretable boosted linear models, XGBoost-informed GLM and an interaction detection workflow. Our findings show that actuaries can improve modelling accuracy, measured by Poisson deviance, by integrating ML insights into traditional modelling techniques, achieving a practical balance of interpretability, expert judgement, and modern analytical innovation.

Health insurers systematically underinvest in prevention. Programme costs are immediate but claims benefits accrue over years, and actuaries have lacked a formal mechanism to translate behavioural intervention evidence into pricing-ready claims adjustments. This paper introduces the Behavioural Adjustment Factor (BAF), a multiplicative actuarial framework that quantifies the claims impact of behavioural interventions by decomposing reach, efficacy, clinical translation, and durability into a single pricing-ready construct. To the best of the author’s knowledge, the BAF is the first actuarial framework to decompose behavioural intervention impact into condition-specific claims projections suitable for pricing and reserving. Drawing on randomised controlled trial evidence, the framework distinguishes interventions that generate reliable claims savings from those that do not. Programme architecture is shown to matter more than incentive magnitude, and the distinction between disease management and general lifestyle programmes emerges as the principal axis along which actuarial expectations should diverge. A worked hypertension example illustrates how the four BAF components combine to produce a defensible claims-adjustment range, and a sensitivity analysis highlights the dominant role of effect persistence. The framework provides confidence intervals, Monte Carlo integration for Solvency II capital modelling, a milestone-based pilot funding structure, and a clear pathway from international evidence to UK-calibrated practice.

This paper overviews an overlapping generations financial cash flow valuation model that evaluates the financial sustainability of English NHS Trusts. It quantifies the financial sustainability constraints related to societal demographic shifts that affect their ability to maintain service delivery quality. The financial model computes a new long-term financial sustainability performance metric based on the notion of the Social Return on Investment (“SROI”). The measure evaluates the financial sustainability of two English acute care hospital foundation Trusts. Significant generational imbalances are identified for both sample NHS Trusts, both within different cohorts of existing generations, and between existing and future generational cohorts. Suggestions are provided for implementing these ideas and expanding actuaries’ expertise and skill sets.

This research sought to update understanding following improvements to treatment and deepen the understanding of the mortality risk associated with Type 1 or Type 2 diabetes, including relative risk in the presence of co-morbidities. Specifically, we developed a model to provide mortality predictions at a granular level for lives with and without diabetes. The model is tailored for use by the insurance industry to provide an updated source from which to appreciate the risk posed when underwriting people with diabetes. By providing an updated and deeper understanding of mortality risk, the research’s aim is to improve access to insurance for those individuals living with diabetes. The model combines industry standard underwriting risk factors, such as age, gender, deprivation index, body mass index (BMI), smoker status, blood pressure (BP) and cholesterol level, with various co-morbidities related to diabetes. A comprehensive analysis of mortality risk factors, between 2010 and 2019, for people with and without diabetes is undertaken on over 1.2 million records based on Clinical Practice Research Datalink (CPRD), Hospital Episode Statistics (HES) and Office for National Statistics (ONS) death registrations data. Cox proportional hazards models are used to estimate the probability of death, stratified by gender across three distinct populations: Type 1 diabetes, Type 2 diabetes and a general population sample. The model outputs produced are permutations of the following: gender; population split by general sample, Type 1 and Type 2 diabetes; and a time-dependent exponential model and a time-invariant homogeneous model. A Shiny model application allows interaction with the model outputs (https://0jv7e6-scott-reid.shinyapps.io/diabmdl/) and spreadsheets provide additional explanation. Useful insights were obtained through industry discussions on the variation of existing market practice against that implied by the results. Key rating factors were generally aligned with market practice, such as age, BMI, BP, cholesterol and years since a diabetes diagnosis. However, for a few significant mortality risks impacting co-morbidities, the results did not adhere to prior expectations. Exploratory work suggested that the order and sequencing of key co-morbidities for diabetes must be included in future model development.

This paper presents the results of a targeted survey conducted among UK pension consultancies, insurers and reinsurers to explore current actuarial practices and challenges in modelling the impact on mortality and morbidity of climate perils. The responses reveal several recurring themes, including varying degrees of climate risk integration, plus significant data and methodological limitations. This is against a background of growing regulatory and stakeholder pressure to incorporate climate considerations into actuarial models. While a small number of organisations have made substantive progress, many are at a preliminary stage, relying primarily on qualitative or scenario-based approaches. Respondents identified a pressing need for standardised methodologies, improved data sources and practical guidance to support consistent and credible modelling. The findings highlight a clear need for profession-led guidance, improved access to climate-health data and practical frameworks to support the consistent integration of climate-related mortality and morbidity risks into actuarial work. Respondents expressed strong support for collaborative development of standardised scenarios, illustrative modelling approaches and case studies to bridge the gap between public-health research and actuarial application. The paper concludes with practical recommendations for practitioners and identifies priority areas for future research and professional guidance.

A period of prolonged, stable economic inflation across much of the developed world arguably reduced the relative significance of inflation assumptions in the reserving process. This stability leant validity to one of the inherent assumptions of the Chain Ladder reserving method – that inflation in future years would mirror that observed historically. The effects of the COVID-19 pandemic on global supply chains resulted in a state of persistently high economic inflation hitherto unexperienced since the 1980s. This spike in turn sparked concerns around the adequacy of general insurance claim provisions (loss reserves) and the methods for their assessment amongst insurance boards and their regulators. In this paper, we explore a variety of methods for attempting to incorporate a more volatile and variable state of claims inflation in reserving. These methods are largely those discussed by regulators in response to the 2021 inflation spike. The key findings are the importance of explicitly considering the inflation inherent in a claim cohort; how it compares with views of future inflation; and revising both our views of the future and actual inflation, as observations become available. We further consider the impact of “imperfect knowledge” or parameter uncertainty with regard these inflation assumptions, when setting reserves.

Dependencies are a major theme in capital modelling, and these are often some of the most material and subjective judgements in an internal model. Most internal models rely to some extent on external models, and dependencies between internal and external models are increasingly important, given the growing tendency for risks in one area of the risk profile to impact outcomes in other risk areas; the impact of economic conditions on underwriting performance, and the importance of non-physical climate-change risks are just some examples. This paper provides a novel method, which goes part way to addressing a limitation in apparent standard practice, by enabling the specification of more than one dependence relationship between sets of pre-simulated data. It therefore sets actuaries up to be able to deal with the ever-more correlated world around us.