Following the large-scale Russian invasion in February 2022, policymakers and humanitarian actors urgently sought to anticipate displacement flows within Ukraine. However, existing internal displacement data systems had not been adapted to contexts as dynamic as a full-fledged war marked by uneven trigger events. A year and a half later, policymakers and practitioners continue to seek forecasts, needing to anticipate how many internally displaced persons (IDPs) can be expected to return to their areas of origin and how many will choose to stay and seek a durable solution in their place of displacement. This article presents a case study of an anticipatory approach deployed by the International Organization for Migration (IOM) Mission in Ukraine since March 2022, delivering nationwide displacement figures less than 3 weeks following the invasion alongside near real-time data on mobility intentions as well as key data anticipating the timing, direction, and volume of future flows and needs related to IDP return and (re)integration. The authors review pre-existing mobility forecasting approaches, then discuss practical experiences with mobility prediction applications in the Ukraine response using the Ukraine General Population Survey (GPS), including in program and policy design related to facilitating durable solutions to displacement. The authors focus on the usability and ethics of the approach, already considered for replication in other displacement contexts.

In this work, we focus on stochastic modeling for sustainable systems and introduce the family of r-modified reliability systems. This new family generalizes classical reliability systems studied in the literature by considering the components in the system to exhibit a kind of dependence that relaxes the component operating requirements and provides energy and resource efficiency. From a theoretical viewpoint, such a dependence is modeled with the use of a modified binary sequence. We then derive the reliability of two members of the family, i.e., the r-modified-k-out-of-n:F system and the r-modified-consecutive-k-out-of-n:F system, under different assumptions on the component reliabilities by using a variety of approaches, including Markov chains, combinatorial methods, and simple probabilistic arguments. We finally give some examples of real-life systems wherein the developed models and results are applicable and present the corresponding numerical results.

Governments all over the world are struggling to control the spiralling costs of healthcare – the UK government is no exception. Its long-term strategy includes a much greater focus on prevention: to keep people as healthy and productive as possible for longer. This paper asks whether a greater focus on prevention is a possible lifeline for the National Health Service (NHS) as is often claimed, but it also examines other benefits to society. After considering various examples of prevention and the metrics used to measure their effectiveness, we use tobacco consumption as a case study to evaluate the costs to the public purse and to wider society. We give further examples, including obesity, but in less depth. We find that whilst there are significant benefits to public expenditure, including the NHS, in both cases, these are dwarfed by wider benefits to society both in terms of tangible economic benefits and improved well-being. We offer several suggestions for improving our understanding of the effectiveness of prevention policies in general and how the Actuarial profession can contribute to this debate.

This paper investigates time-varying risk sharing between annuity buyer and provider. It explores Pareto optimal (PO) and viable Pareto optimal (VPO) risk-sharing designs, in which the share of the reserve deviation transferred to the policyholder varies over time. The optimization problem, based on a weighted average of mean-variance preferences, results in a complex quartic objective function. Such optimization problems are difficult to solve, and checking their convexity is known to be NP-hard. A heuristic method is introduced to simplify the problem, providing a closed-form solution that closely approximates the numerical results. The paper also highlights factors influencing the existence of VPO designs, with age playing a critical role, thereby suggesting the suitability of these designs as retirement products.

Investigating risk factors for mpox’s infectious period is vital for preventing this emerging disease, yet evidence remains scarce. This study aimed to identify risk factors associated with the duration of mpox infectiousness among mpox cases in Vietnam. The primary outcome was the duration of the mpox infectiousness, defined between symptom onset and the first negative test result for the mpox virus. Fine and Gray’s regression models were employed to assess the associations between the infectious period and several risk factors while accounting for competing risks of death by mpox. Most mpox cases recovered within 30 days. Patients with HIV or treated at multiple facilities for mpox had lower incidence rates of cleared infection compared to those who were HIV-negative or treated at a single facility. In regression models, patients with mpox symptoms of rash or mucosal lesions (sub-distribution hazard ratios = 0.62, 95% confidence interval = 0.46–0.83), ulcers (0.57, 0.41–0.80), or fever (0.62, 0.46–0.83) had significantly prolonged infectious periods than those without such symptoms. Our findings provided insights for managing mpox cases, especially those vulnerable to prolonged infectious periods in settings with sporadic cases reported.

The Alpha, Delta, and Omicron variants of the SARS-CoV-2 virus have been deemed as variants of concern (VOCs) by the WHO due to their increased transmissibility, severity of illness, and resilience against treatments. Geographically tracking the spread of these variants can help us implement effective control measures. RNA from 8,154 SARS-CoV-2 positive nasal swab samples from a Central Texas hospital collected between March 2020 and April 2023 were sequenced in Temple, TX. Global and U.S. sequencing metadata was obtained from the GISAID database on 3 April 2023. Using sequencing metadata, the growth rate of Alpha, Delta, and the first subvariant of Omicron (BA.1) were obtained as 0.27, 0.3, and 1.08 each. The average time in days to penetrate the US for Alpha, Delta, and Omicron were 269.2, 326.2, and 27.3 days, respectively. Viral sequencing data can be a useful tool to examine the spread of viruses. Each emerging SARS-CoV-2 variant penetrated cities more rapidly as the pandemic progressed. With a high logarithmic growth rate, the Omicron variant penetrated the US more rapidly as the pandemic progressed.

Inequality is a critical global issue, particularly in the United States, where economic disparities are among the most pronounced. Social justice research traditionally studies attitudes towards inequality—perceptions, beliefs, and judgments—using latent variable approaches. Recent scholarship adopts a network perspective, showing that these attitudes are interconnected within inequality belief systems. However, scholars often compare belief systems using split-sample approaches without examining how emotions, such as anger, shape these systems. Moreover, they rarely investigate Converse’s seminal idea that changes in central attitudes can lead to broader shifts in belief systems. Addressing these gaps, we applied a tripartite analytical strategy using U.S. data from the 2019 ISSP Social Inequality module. First, we used a mixed graphical model to demonstrate that inequality belief systems form cohesive small-world networks, with perception of large income inequality and belief in public redistribution as central nodes. Second, a moderated network model revealed that anger towards inequality moderates nearly one-third of network edges, consolidating the belief system by polarizing associations. Third, Ising model simulations showed that changes to central attitudes produce broader shifts across the belief system. This study advances belief system research by introducing innovative methods for comparing structures and testing dynamics of attitude change. It also contributes to social justice research by integrating emotional dynamics and highlighting anger’s role in structuring inequality belief systems.

We define the generalized equilibrium distribution, that is the equilibrium distribution of a random variable with support in $\mathbb{R}$. This concept allows us to prove a new probabilistic generalization of Taylor’s theorem. Then, the generalized equilibrium distribution of two ordered random variables is considered and a probabilistic analog of the mean value theorem is proved. Results regarding distortion-based models and mean-median-mode relations are illustrated as well. Conditions for the unimodality of such distributions are obtained. We show that various stochastic orders and aging classes are preserved through the proposed equilibrium transformations. Further applications are provided in actuarial science, aiming to employ the new unimodal equilibrium distributions for some risk measures, such as Value-at-Risk and Conditional Tail Expectation.

We show that on every affine Weyl group natural random walks are noise sensitive in total variation.

As the global elderly population expands, the associated risks of longevity intensify, presenting significant challenges to traditional retirement security systems. We study actuarial fairness in tontines under the Volterra mortality framework, integrating long-range dependence mortality models rates with tontine structures. Initially, we establish an optimal tontine model for a homogeneous tontine under this framework. However, according only to individual actuarial fairness can neglect the collective nature of tontines. So we propose a hybrid optimization model that accounts for age and wealth discrepancies affecting payment amounts and the collective fairness. Specially, we first apply the f-value fairness measure in age-heterogeneous tontines for assessing fairness. Our results reveal that while the model ensures actuarial fairness at the group level, relative payments are lower for older age groups. By incorporating dynamic mortality modeling through the Volterra mortality framework, our work demonstrates that this comprehensive scheme significantly enhances the robustness and sustainability of retirement security systems. These findings provide valuable insights for the future integration of dynamic mortality models with innovative retirement income structures.

Hepatitis B virus vaccination is currently recommended in Australia for adults at an increased risk of acquiring infection or at high risk of complications from infection. This retrospective cohort study used data from an Australian sentinel surveillance system to assess the proportion of individuals who had a recorded test that indicated being susceptible to hepatitis B infection in six priority populations, as well as the proportion who were then subsequently vaccinated within six months of being identified as susceptible. Priority populations included in this analysis were people born overseas in a hepatitis B endemic country, people living with HIV, people with a recent hepatitis C infection, gay, bisexual and other men who have sex with men, people who have ever injected drugs, and sex workers. Results of the study found that in the overall cohort of 43,335 individuals, 14,140 (33%) were identified as susceptible to hepatitis B, and 5,255 (37%) were subsequently vaccinated. Between 26% and 33% of individuals from priority populations were identified as susceptible to hepatitis B infection, and the proportion of these subsequently vaccinated within six months was between 28% and 42% across the groups. These findings suggest further efforts are needed to increase the identification and subsequent vaccination of susceptible individuals among priority populations recommended for hepatitis B vaccination, including among people who are already engaged in hepatitis B care.

Exchangeable partitions of the integers and their corresponding mass partitions on $\mathcal{P}_{\infty} = \{\mathbf{s} = (s_{1},s_{2},\ldots)\colon s_{1} \geq s_{2} \geq \cdots \geq 0$ and $\sum_{k=1}^{\infty}s_{k} = 1\}$ play a vital role in combinatorial stochastic processes and their applications. In this work, we continue our focus on the class of Gibbs partitions of the integers and the corresponding stable Poisson–Kingman-distributed mass partitions generated by the normalized jumps of a stable subordinator with an index $\alpha \in (0,1)$, subject to further mixing. This remarkable class of infinitely exchangeable random partitions is characterized by probabilities that have Gibbs (product) form. These partitions have practical applications in combinatorial stochastic processes, random tree/graph growth models, and Bayesian statistics. The most notable class consists of random partitions generated from the two-parameter Poisson–Dirichlet distribution $\mathrm{PD}(\alpha,\theta)$. While the utility of Gibbs partitions is recognized, there is limited understanding of the broader class. Here, as a continuation of our work, we address this gap by extending the dual coagulation/fragmentation results of Pitman (1999), developed for the Poisson–Dirichlet family, to all Gibbs models and their corresponding Poisson–Kingman mass partitions, creating nested families of Gibbs partitions and mass partitions. We focus primarily on fragmentation operations, identifying which classes correspond to these operations and providing significant calculations for the resulting Gibbs partitions. Furthermore, for completion, we provide definitive results for dual coagulation operations using dependent processes. We demonstrate the applicability of our results by establishing new findings for Brownian excursion partitions, Mittag-Leffler, and size-biased generalized gamma models.

The continuous random energy model (CREM) was introduced by Bovier and Kurkova in 2004 as a toy model of disordered systems. Among other things, their work indicates that there exists a critical point $\beta_\mathrm{c}$ such that the partition function exhibits a phase transition. The present work focuses on the high-temperature regime where $\beta<\beta_\mathrm{c}$. We show that, for all $\beta<\beta_\mathrm{c}$ and for all $s>0$, the negative s moment of the CREM partition function is comparable with the expectation of the CREM partition function to the power of $-s$, up to constants that are independent of N.

Sharp, nonasymptotic bounds are obtained for the relative entropy between the distributions of sampling with and without replacement from an urn with balls of $c\geq 2$ colors. Our bounds are asymptotically tight in certain regimes and, unlike previous results, they depend on the number of balls of each color in the urn. The connection of these results with finite de Finetti-style theorems is explored, and it is observed that a sampling bound due to Stam (1978) combined with the convexity of relative entropy yield a new finite de Finetti bound in relative entropy, which achieves the optimal asymptotic convergence rate.

We study a two-dimensional discounted optimal stopping zero-sum (or Dynkin) game related to perpetual redeemable convertible bonds expressed as game (or Israeli) options in a model of financial markets in which the behaviour of the ex-dividend price of a dividend-paying asset follows a generalized geometric Brownian motion. It is assumed that the dynamics of the random dividend rate of the asset paid to shareholders are described by the mean-reverting filtering estimate of an unobservable continuous-time Markov chain with two states. It is shown that the optimal exercise (conversion) and withdrawal (redemption) times forming a Nash equilibrium are the first times at which the asset price hits either lower or upper stochastic boundaries being monotone functions of the running value of the filtering estimate of the state of the chain. We rigorously prove that the optimal stopping boundaries are regular for the stopping region relative to the resulting two-dimensional diffusion process and that the value function is continuously differentiable with respect to the both variables. It is verified by means of a change-of-variable formula with local time on surfaces that the optimal stopping boundaries are determined as a unique solution to the associated coupled system of nonlinear Fredholm integral equations among the couples of continuous functions of bounded variation satisfying certain conditions. We also give a closed-form solution to the appropriate optimal stopping zero-sum game in the corresponding model with an observable continuous-time Markov chain.

Federated learning (FL) is a machine learning technique that distributes model training to multiple clients while allowing clients to keep their data local. Although the technique allows one to break free from data silos keeping data local, to coordinate such distributed training, it requires an orchestrator, usually a central server. Consequently, organisational issues of governance might arise and hinder its adoption in both competitive and collaborative markets for data. In particular, the question of how to govern FL applications is recurring for practitioners. This research commentary addresses this important issue by inductively proposing a layered decision framework to derive organisational archetypes for FL’s governance. The inductive approach is based on an expert workshop and post-workshop interviews with specialists and practitioners, as well as the consideration of real-world applications. Our proposed framework assumes decision-making occurs within a black box that contains three formal layers: data market, infrastructure, and ownership. Our framework allows us to map organisational archetypes ex-ante. We identify two key archetypes: consortia for collaborative markets and in-house deployment for competitive settings. We conclude by providing managerial implications and proposing research directions that are especially relevant to interdisciplinary and cross-sectional disciplines, including organisational and administrative science, information systems research, and engineering.

Contact tracing is an effective public health policy to put the fast-spreading epidemic under control. The government tracks the contacts of confirmed SARS-CoV-2 cases, recommends testing, encourages self-quarantine, and monitors symptoms of contacts. In developing and less-developed countries with limited resources for widespread SARS-CoV-2 testing, it remains essential to identify and quarantine positive contacts to control outbreaks. Therefore, analysing recall and precision when implementing testing policies for these contacts is necessary. We analysed a contact tracing dataset from a cohort of 827 index patients infected with SARS-CoV-2 and their 14814 close contacts from Jan 2020 to July 2020 in a province in eastern China. We constructed a network from the data and used a Graph Convolutional Network to predict each contact’s infection status. To the best of our knowledge, this is the first method to use population-based contact tracing data for predicting the infection status using graph neural networks. Despite limited information, our model achieves competitive Area Under the Receiver Operating Characteristic Curve (ROC AUC) compared to hospital-onset scenarios. Based on the risk scores, we propose several contact testing policy adaptations that balance resource efficiency and effective pandemic control.

This study explores the potential of applying machine learning (ML) methods to identify and predict areas at risk of food insufficiency using a parsimonious set of publicly available data sources. We combine household survey data that captures monthly reported food insufficiency with remotely sensed measures of factors influencing crop production and maize price observations at the census enumeration area (EA) in Malawi. We consider three machine-learning models of different levels of complexity suitable for tabular data (TabNet, random forests, and LASSO) and classical logistic regression and examine their performance against the historical occurrence of food insufficiency. We find that the models achieve similar accuracy levels with differential performance in terms of precision and recall. The Shapley additive explanation decomposition applied to the models reveals that price information is the leading contributor to model fits. A possible explanation for the accuracy of simple predictors is the high spatiotemporal path dependency in our dataset, as the same areas of the country are repeatedly affected by food crises. Recurrent events suggest that immediate and longer-term responses to food crises, rather than predicting them, may be the bigger challenge, particularly in low-resource settings. Nonetheless, ML methods could be useful in filling important data gaps in food crises prediction, if followed by measures to strengthen food systems affected by climate change. Hence, we discuss the tradeoffs in training these models and their use by policymakers and practitioners.