Lebanon does not have a national mortality table that reflects its demographic and health conditions. Despite ongoing changes in mortality patterns driven by economic crises, political instability, and social changes, outdated foreign tables such as AM80 remain in use in the insurance and public sectors. This dependency introduces significant risks in actuarial calculations, policy design, and long-term planning. This study addresses this gap by building a mortality table specifically adapted to the Lebanese insurance context, together with a first estimation of population-level mortality. In the absence of any official mortality database, we collaborated directly with local insurance companies to access and organize internal records of insured lives. These data, which represent one of the few available structured sources of mortality information in the country, form the core of our analysis. We apply actuarial methods to estimate age-specific death rates and life expectancy and benchmark the results against national and international references to assess consistency and range. By offering a locally grounded, data-driven alternative to imported mortality assumptions, this work fills a critical statistical need. The resulting table supports more accurate forecasting, pricing, and demographic modeling, with applications across insurance, pensions, and public health planning in Lebanon.

This article introduces a blockchain-based insurance scheme that integrates parametric and collaborative elements. A pool of investors, referred to as surplus providers, locks funds in a smart contract, enabling blockchain users to underwrite parametric insurance contracts. These contracts automatically trigger compensation when predefined conditions are met. The collaborative aspect is embodied in the generation of tokens, which are distributed to surplus providers. These tokens represent each participant’s share of the surplus and grant voting rights for management decisions. The smart contract is developed in Solidity, a high-level programming language for the Ethereum blockchain, and deployed on the Sepolia testnet, with data processing and analysis conducted using Python. In addition, open-source code is provided and main research challenges are identified, so that further research can be carried out to overcome limitations of this first proof of concept.

This paper explores the challenges that insurance purchasers face when selecting from multiple distribution channels and their impact on insurance demand optimality. It investigates the choice between independently assessing various insurance contracts or relying on potentially dishonest intermediaries. We use a lab experiment mimicking real-world insurance distribution channels with varying informational frameworks, where information is gathered in exchange for search costs. Search costs deter some individuals from thorough exploration, preventing them from finding optimal contracts. However, suboptimality in choices also arises for those exploring extensively, due to choice overload. Lastly, some individuals prefer to delegate to intermediaries, but intermediaries are likely to deviate from optimal advice due to their own financial incentives. Overall, we find that brokers yield the highest rate of optimal choices, but these choices also come at a higher cost.

This study addresses a research gap in quantitative modeling framework and scenario analysis for the risk management of stable value fund wraps, a crucial segment of the U.S. financial market with over USD 400 billion in assets. In this paper, we present an asset–liability model that encompasses an innovative approach to modeling the assets of fixed-income funds coupled with a liability model backed by empirical analysis on a unique data set covering 80

The Devylder–Goovaerts conjecture is probably the oldest conjecture in actuarial mathematics and has received a lot of attention in recent years. It claims that ruin with equalized claim amounts is always less likely than in the classical model. Investigating the validity of this conjecture is important both from a theoretical aspect and a practical point of view, as it suggests that one always underestimates the risk of insolvency by replacing claim amounts with the average claim amount a posteriori. We first state a simplified version of the conjecture in the discrete-time risk model when one equalizes aggregate claim amounts and prove that it holds. We then use properties of the Pareto distribution in risk theory and other ideas to target candidate counterexamples and provide several counterexamples to the original Devylder–Goovaerts conjecture.

In this paper, we investigate the behavioral and statistical characteristics of cash flows for stable value funds provided by numerous U.S. employee benefit plans. We analyze participant-initiated aggregated cash flow data, representing approximately 80% of the market for large employer plans with stand-alone stable value wraps within a 401(k) offering. By leveraging this unique dataset and contextualizing the 401(k) ecosystem, we examine numerous behavioral lapse hypotheses. Our findings highlight key behavioral lapse hypotheses for modeling lapses and generating risk scenarios. We demonstrate that cash flows exhibit medium- to long-term non-monotonic trends. Factors within the plan sponsor’s ecosystem, such as employment growth, default 401(k) plan options, and the introduction of new investment options, significantly impact participant cash flow behavior indirectly. Moreover, we find that flight-to-safety behavior plays a dominant role during global market crises. Although the risk of mass lapses due to reputational issues is observed, their probability of occurrence is low. Other behavioral hypotheses discussed in the literature, such as the moneyness hypothesis, are found to be less prevalent in this context.

Predicting the evolution of mortality rates plays a central role for life insurance and pension funds. Various stochastic frameworks have been developed to model mortality patterns taking into account the main stylized facts driving these patterns. However, relying on the prediction of one specific model can be too restrictive and lead to some well documented drawbacks including model misspecification, parameter uncertainty and overfitting. To address these issues we first consider mortality modelling in a Bayesian Negative-Binomial framework to account for overdispersion and the uncertainty about the parameter estimates in a natural and coherent way. Model averaging techniques are then considered as a response to model misspecifications. In this paper, we propose two methods based on leavefuture-out validation which are compared to the standard Bayesian model averaging (BMA) based on marginal likelihood. An intensive numerical study is carried out over a large range of simulation setups to compare the performances of the proposed methodologies. An illustration is then proposed on real-life mortality datasets which includes a sensitivity analysis to a Covid-type scenario. Overall, we found that both methods based on out-of-sample criterion outperform the standard BMA approach in terms of prediction performance and robustness.

We study the problem of approximating the copula and copula density function from a sequence of transformed moments. In particular, when frequency moments of an underlying bivariate distribution are available, the uniform convergence of the reconstructed copula and the rate of approximation of the copula density function are obtained. Finally, the accuracies of the approximation and estimation are illustrated in a simulation study.

This paper proposes a method for quantifying the basis risk present in index-based insurance. It applies when the inherent uncertainty is represented by a randomly scaled variable. This turns out to be a reasonable assumption in a number of practical situations. Several properties of such a variable are first briefly studied. Their order in the s -convex sense is discussed and the associated extreme distributions are obtained to generate the worst situations. In each scenario, the basis risk consequences are then assessed using a penalty function that takes into account the risk tolerances of the protection buyer. Basis risk limits for a fixed budget can also be set. The proposed approach is illustrated by a few simple examples.

This paper introduces quantitative risk management and scenario analysis framework for asset-liability management of stable value fund wraps. Stable value funds are guaranteed return employee benefit investment options, with currently over $400 billion USD of assets under management. These investment schemes are synthesized as such to transform a volatile fund return into a more stable one and to buffer against short-term market fluctuations.Having the main objective of understanding risk scenarios from the insurer's perspective, we argue that the simplest model for participants' withdrawal, while still maintaining sufficient flexibility to detect tail scenarios, is a two-state regime-switching model. We present such a model and attempt to identify risk scenarios that may result in substantial losses for the insurer. We conclude that there are two scenarios to consider: the inflationary scenario and the yield spike scenario. We also back-test these scenarios historically during the financial crisis, the 1980 inflation period, the Great Depression period, and the 2020 pandemic.Finally, we apply the model to estimate the tail risk and study the impact of the fund's characteristics, e.g. duration and rating. Furthermore, we observe that under normal conditions, while our model suggests that there is a convex tail risk, the U.S. regulatory formulaic risk-based capital is almost always zero, which may infer an incomplete picture of U.S. insurers' actuarial risk and profitability metrics.

The major economic and health consequences of COVID-19 called for various protective measures and mass vaccination campaigns. A previsional model was used to predict the future impacts of various measure combinations on COVID-19 mortality over a 400-day period in France. Calibrated on previous national hospitalization and mortality data, an agent-based epidemiological model was used to predict individual and combined effects of booster doses, vaccination of refractory adults, and vaccination of children, according to infection severity, immunity waning, and graded non-pharmaceutical interventions (NPIs). Assuming a 1.5 hospitalization hazard ratio and rapid immunity waning, booster doses would reduce COVID-19-related deaths by 50–70% with intensive NPIs and 93% with moderate NPIs. Vaccination of initially-refractory adults or children ≥5 years would half the number of deaths whatever the infection severity or degree of immunity waning. Assuming a 1.5 hospitalization hazard ratio, rapid immunity waning, moderate NPIs and booster doses, vaccinating children ≥12 years, ≥5 years, and ≥6 months would result in 6212, 3084, and 3018 deaths, respectively (vs. 87,552, 64,002, and 48,954 deaths without booster, respectively). In the same conditions, deaths would be 2696 if all adults and children ≥12 years were vaccinated and 2606 if all adults and children ≥6 months were vaccinated (vs. 11,404 and 3624 without booster, respectively). The model dealt successfully with single measures or complex combinations. It can help choosing them according to future epidemic features, vaccination extensions, and population immune status.

Background: Compartmental models may help deciding on public health interventions. They were used during the first French COVID-19 lockdown to estimate the reproduction numbers and predict the number of hospital beds required. This study aimed to assess the ability of similar compartmental models to reflect equivalent epidemic dynamics. Methods: The study considered three compartmental models independently designed to describe the COVID-19 outbreak in France. These models were scrutinized and their compartments and parameters expressed in a common framework. The parameters were set alike in the three models according to values taken from the literature. The models were calibrated using a common maximum likelihood function and the same hospitalization data taken from two official public databases. The calibration procedure was repeated over three different periods to compare model abilities to (1) fit over the whole lockdown; (2) predict the course of the epidemic during the lockdown; and, (3) provide a set of profiles to forecast the hospitalization prevalence after the lockdown. The study considered national and regional coverages. Results: The three models were all flexible enough to match hospitalization data during the lockdown, but the numbers of cases in the other compartments differed. The three models failed to predict reliably the number of hospitalizations after the fitting periods at national and regional levels. At the national scale, a refined calibration led to epidemic course profiles that reflected hospitalizations dynamics and corresponded to reproduction numbers coherent with official and literature estimates. This result could not be consistently obtained at the regional scale. Conclusion: Not all predictions were consistent between models. Even over the period used for calibration by fitting to hospitalization cases, important differences remained regarding the prevalence in the other compartments. Prevalence data are needed to further constrain the calibration and perform selection between still divergent models. This underlines strongly the need for repeated prevalence studies on representative samples, stratified by age and regions, which would undergo virological or serological tests.

The outbreak of the SARS-CoV-2 virus, enhanced by rapid spreads of variants, has caused a major international health crisis, with serious public health and economic consequences. An agent-based model was designed to simulate the evolution of the epidemic in France over 2021 and the first six months of 2022. The study compares the efficiencies of four theoretical vaccination campaigns (over 6, 9, 12, and 18 months), combined with various non-pharmaceutical interventions. In France, with the emergence of the Alpha variant, without vaccination and despite strict barrier measures, more than 600,000 deaths would be observed. An efficient vaccination campaign (i.e., total coverage of the French population) over six months would divide the death toll by 10. A vaccination campaign of 12, instead of 6, months would slightly increase the disease-related mortality (+6%) but require a 77% increase in ICU bed–days. A campaign over 18 months would increase the disease-related mortality by 17% and require a 244% increase in ICU bed–days. Thus, it seems mandatory to vaccinate the highest possible percentage of the population within 12, or better yet, 9 months. The race against the epidemic and virus variants is really a matter of vaccination strategy.

In this paper, we extend the non-cooperative one-period game of Dutang et al. (Journal of Operational Research 231(3):702–711, 2013) to model a non-life insurance market over several periods by considering the repeated (one-period) game. Using Markov chain methodology, we derive general properties of insurer portfolio sizes given a price vector. In the case of a regulated market (identical premium), we are able to obtain convergence measures of long run market shares. We also investigate the consequences of the deviation of one player from this regulated market. Finally, we provide some insights of long-term patterns of the repeated game as well as numerical illustrations of leadership and ruin probabilities.

Background The outbreak of SARS-CoV-2 virus has caused a major international health crisis with serious consequences in terms of public health and economy. In France, two lockdown periods were decided in 2020 to avoid the saturation of intensive care units (ICU) and an increase in mortality. The rapid dissemination of variant SARS-CoV-2 VOC 202012/01 has strongly influenced the course of the epidemic. Vaccines have been rapidly developed. Their efficacy against the severe forms of the disease has been established, and their efficacy against disease transmission is under evaluation. The aim of this paper is to compare the efficacy of several vaccination strategies in the presence of variants in controlling the COVID-19 epidemic through population immunity. Methods An agent-based model was designed to simulate with different scenarios the evolution of COVID-19 pandemic in France over 2021 and 2022. The simulations were carried out ignoring the occurrence of variants then taking into account their diffusion over time. The expected effects of three Non-Pharmaceutical Interventions (Relaxed-NPI, Intensive-NPI, and Extended-NPI) to limit the epidemic extension were compared. The expected efficacy of vaccines were the values recently estimated in preventing severe forms of the disease (75% and 94%) for the current used vaccines in France (Pfizer-BioNTech and Moderna since January 11, 2021, and AstraZeneca since February 2, 2021). All vaccination campaigns reproduced an advanced age-based priority advised by the Haute Autorit[c] de Sant[c]. Putative reductions of virus transmission were fixed at 0, 50, 75 and 90%. The effects of four vaccination campaign durations (6-month, 12-month, 18-month and 24-month) were compared. Results In the absence of vaccination, the presence of variants led to reject the Relaxed-NPI because of a high expected number of deaths (170 to 210 thousands) and the significant overload of ICUs from which 35 thousand patients would be deprived. In comparison with the situation without vaccination, the number of deaths was divided by 7 without ICU saturation with a 6-month vaccination campaign. A 12-month campaign would divide the number of deaths by 3 with Intensive-NPI and by 6 with Extended-NPI (the latter being necessary to avoid ICU saturation). With 18-month and 24-month vaccination campaigns without Extended-NPI, the number of deaths and ICU admissions would explode. Conclusion Among the four compared strategies the 6-month vaccination campaign seems to be the best response to changes in the dynamics of the epidemic due to the variants. The race against the COVID-19 epidemic is a race of vaccination strategy. Any further vaccination delay would increase the need of strengthened measures such as Extended-NPI to limit the number of deaths and avoid ICU saturation.

Modeling policyholders lapse behaviors is important to a life insurer since lapses affect pricing, reserving, profitability, liquidity, risk management, as well as the solvency of the insurer. Lapse risk is indeed the most significant life underwriting risk according to European Insurance and Occupational Pensions Authority's Quantitative Impact Study QIS5. In this paper, we introduce two advanced machine learning algorithms for lapse modeling. Then we evaluate the performance of different algorithms by means of classical statistical accuracy and profitability measure. Moreover, we adopt an innovative point of view on the lapse prediction problem that comes from churn management. We transform the classification problem into a regression question and then perform optimization, which is new for lapse risk management. We apply different algorithms to a large real-world insurance dataset. Our results show that XGBoost and SVM outperform CART and logistic regression, especially in terms of the economic validation metric. The optimization after transformation brings out significant and consistent increases in economic gains.

Predicting the evolution of mortality rates plays a central role for life insurance and pension funds. Standard single population models typically suffer from two major drawbacks: on the one hand, they use a large number of parameters compared to the sample size and, on the other hand, model choice is still often based on in-sample criterion, such as the Bayes information criterion (BIC), and therefore not on the ability to predict. In this paper, we develop a model based on a decomposition of the mortality surface into a polynomial basis. Then, we show how regularization techniques and cross-validation can be used to obtain a parsimonious and coherent predictive model for mortality forecasting. We analyze how COVID-19-type effects can affect predictions in our approach and in the classical one. In particular, death rates forecasts tend to be more robust compared to models with a cohort effect, and the regularized model outperforms the so-called P-spline model in terms of prediction and stability.

This paper is concerned with properties of Beta-unimodal distributions and their use to assess the basis risk inherent to index-based insurance or reinsurance contracts. To this extent, we first characterize s-convex stochastic orders for Beta-unimodal distributions in terms of the Weyl fractional integral. We then determine s-convex extrema for such distributions, focusing in particular on the cases s = 2, 3, 4. Next, we define an Enterprise Risk Management framework that relies on Beta-unimodality to assess these hedge imperfections, introducing several penalty functions and worst case scenarios. Some of the results obtained are illustrated numerically via a representative catastrophe model.

The insurability of natural disasters has always been an issue faced by the insurers, states, and insured persons. In France, the insurer and the legislator are concerned about the subsidence risks due to several consecutive dry years. More and more open data are provided in France, which allows insurers by geolocating their portfolio to have better knowledge. This knowledge plus the increase in subsidence risks query the insurability of the subsidence risk. Using mostly GLMs, the most common models used in France, this paper shows the improvement of the knowledge subsidence risks. The results bring to the fore the importance of legislative control and the recently enforced new CatNat program, leading authors to question the CatNat fee stagnation.