AGEFI Luxembourg - juillet août 2025

Juillet / Août 2025 19 AGEFI Luxembourg Banques / Assurances E conomic losses fromnatural catastro- phes have been rising steadily over the past decades, driven by climate change, urbanisation, and the increasing complexity of global supply chains. Despite significant advances in catastro- phemodeling, data analytics, and visuali- sation tools, many corporations still lack a comprehensive under- standing of their exposure portfolios. When extreme events such as hurricanes, floods, or earthquakes strike, these companies and their cap- tives often find themselves scrambling to assemble frag- mented data sets – highlighting the urgent need for an innovative tech-driven approach to riskmanagement. ArecentanalysisbytheSwissReInstitutefoundthat, following the long-term annual growth trend of 5– 7%,globalinsurednaturalcatastrophe(NatCat)losses may reachUSD145 billion in 2025 (1) . This forecast underscores a pressing reality: without a clear view of where and how assets are at risk, it’s hard for companies to optimise their risk strategies ordeploymitigationmeasureseffectively.Manyrisk managers relyonapatchworkof spreadsheets, lega- cy databases, and departmental reports. These sources often lack standardisation, suffer from data quality issues, and fail to capture evolving NatCat- related hazards such as sea-level rise or changing rainfall patterns. The challenge of fragmenteddata Consider a multinational manufacturing firm with production sites spread across North America, Europe, and Asia. Each region’s operations team maintains its ownasset register, insuranceprogram, and loss history – often using different naming con- ventionsandgeocodingprotocols.Whenattempting to answer a simple question such as “What is our aggregate windstorm exposure in coastal regions?” the risk team must normalise hundreds of data fields, reconcile policy limits, and manually adjust for inconsistencies. The result: lengthy project time- lines, potential errors, and missed opportunities to discuss an optimal risk transfer strategywith insur- ers. Left unchecked, these silos can lead tosignificant miscalculations.Anundercount of replacement val- ues may leave a company underinsured when dis- aster strikes, forcing unexpected capital calls or bal- ance sheet hits. Conversely, overestimating expo- sures can inflate insurance premiums and divert resources fromstrategic investments. Introducing the digital twin approach To address these gaps, a digital twin approach pro- vides a unified, dynamic representation of a corpo- ration’s physical assets and their associated insur- ance coverages. By consolidating asset data, under- writing parameters, and historical loss records into a single platform – such as Swiss Re’s Risk Data & Services (RDS) – risk managers gain an interactive “what-if” environment to assessmultiple scenarios, stress-test assumptions, and compare mitigation strategies side by side. At the core of the RDS platform are five key data domains: 1. Locationdetails : Precisegeocodingof everyasset, down to building footprint or infrastructure seg- ment. 2. Asset values : Replacement costs, business inter- ruption sums insured, and critical thresholds for each location. 3. Buildingcharacteristics : Construction type, occu- pancy, year built, and other factors that influence vulnerability. 4. Historical losses : Credible loss records linked to past events, enabling trend analysis and calibration of model outputs. 5. Policy terms : Coverage layers, sublimits, deductibles, and co-insurance arrangements across programs. By ingesting and standardising these domains, the digital twin serves as a single source of truth. Automated quality checks flag missing addresses, coordinate anomalies, or policy-data mismatches – prompting users to correct or enrich the underlying records before proceeding to analysis. From rawdata to actionable insights Oncedata integrity is ensured, thenext step involves enrichment and ontology building. External data sources – ranging fromopen government GIS feeds to hazard maps – are linked directly to the digital twin, creating relationships between exposures and relevant riskdrivers. In theRDSexample, floodplain boundaries can be overlaid on site polygons, while wind-speed contourmaps connect eachproperty to its expected wind loads. Establishing a common ontology across asset, hazard and policy data unlocks powerful query capabilities. Risk managers can instantlyfilterportfoliosby riskcharacteristics (e.g., all concrete struc- tures within 100 km of the coast) or by insurance terms (e.g. policies with sub- limits below a specified threshold). This level of granularity supportsmore refined pricing, optimised layer placements, and targetedmitigation investments suchas floodproofing or retrofits. Integrating future climate scenarios Climate change com- pounds the challenge. Traditional loss models rely onhistoricaleventfrequencies and intensities, which may no longer reflect emerging patterns. A forward-lookingmodeling framework incorporates scenarios from the latest climate projections – span- ning low- to high-emissionpathways. By recalibrat- ing hazard frequencies and intensities under each scenario, risk managers can quantify how expected lossesmay shift by 2030, 2050, and beyond. For floods (both fluvial and pluvial), windstorms, and tropical cyclones, users can compare baseline loss projections to future states – enabling strategic decisions on portfolio rebalancing, alternative risk transfer, or resilience upgrades. For instance, a coastal warehouse may see its 1-in-100-year flood depth increase by 20 cmunder a high-end scenario; this insight could justifyelevatingcritical equipment or revisiting policy limits. Mapping linear and complex assets Whilepoint-basedassets suchas buildings are often the focus, many corporations rely on critical infras- tructurenetworks–pipelines,transmissionlinesand transport corridors – that require a different model- ingapproach. Bybreaking linear assets intogranular segments, the digital twin enables the identification of hotspots wheremultiple risk drivers converge. A pipeline crossingmultiple flood zones, for example, mayrequiresegmentedrisktransfersolutionsorsite- specificmitigationmeasures. Similarly, multi-building campuses benefit from subdividing sites into individual assetswithunique risk profiles. A single facility with production, stor- age, and office functions may have vastly different vulnerability curves; treating the entire campus as one point canmask critical exposures. Structured risk assessment for captives Tobringthisalltogether,afour-stepassessmentpro- cess for captives, which are increasingly becoming risk advisors to the core business of their parent cor- porations, is advisable: 1. Digital twin : Building an accurate and com- plete set of data containing exposures, losses, and policies. 2. Risk assessment : Scanning the portfolio of assets for riskhotspots, assessingdiversification, and iden- tifying high-risk locations to single out loss drivers. 3. Risk quantification : Quantifying current risks for informeddecision-makingandassessingfuturemod- eled losses for developing a long-termrisk strategy. 4. Riskmitigation : Risk transfer through traditional or alternative risk transfer solutions and adaptation totherisklandscapewithphysicalriskimprovement measures. Lastly, this increased understanding of risk will allow captives to refine their insurance program through external data and validation. Enhancing negotiation and pricing power Armedwithhigh-fidelity insights, captivemanagers can engage brokers and external insurers from a positionof strength. Byvalidating in-house loss esti- mates against industry benchmarks – such as exter- nal NatCat model outputs or environmental risk indices – captives demonstrate robust analytical rigor. This transparency can lead to more competi- tivepremiumplacements,improvedprogramstruc- ture, and access to alternative riskmarkets. In addition, the enriched data foundation supports deeper cost-of-risk optimisation. By quantifying marginalbenefitsofmitigationmeasures(e.g.,sprin- kler installation, flood barriers), captives can priori- tise investments that yield the highest risk-adjusted returns. Combined with real-time analytics, this approach fosters continuous improvement of cor- porate resilience programs. Conclusion In an era of evolving natural hazards and climate dynamics, traditional riskmanagement approach- es – bound by fragmented data and siloed work- flows – fall short. A digital twin methodology, underpinned by unified data domains, advanced hazard integration, and future scenario modeling, equips captive risk managers with the tools they need to anticipate, quantify, and mitigate expo- sures proactively. By transforming raw exposure data into a living, analysable digital asset, organisations can optimise their risk transfer strategies, enhance negotiation leverage, and drive targeted resilience investments. Ultimately, this approach supports long-term sus- tainability, preserves corporate value, and fortifies stakeholder confidence in an uncertainworld. Philipp LÜRZER, Captive Centre of Excellence Manager, Swiss Re Corporate Solutions Adrien NORULAK, Head Risk Analytics, Swiss Re Risk Data Solutions 1) Hurricanes and earthquakes could lead to global insured losses of USD300billioninapeakyear,findsSwissReInstitute|SwissRe Rethinking Corporate Risk Management with Data and Technology A u cours des 12 derniers mois, 78%des entreprises présentes auBenelux ont été victimes de fraude, contre 69% en 2023/2024. Lemontant des dom- mages augmente également. C'est ce qui ressort de l'enquête annuelle sur la fraudemenée par l'assureur- créditAllianzTrade. C'est la quatrième année consécutive qu'Allianz Trade enquête sur l'évolution de la fraude au Benelux. L'enquête (qui porte à la fois sur la fraude interne et ex- terne) montre que l'impact de la fraude s'est considérablement accru. 46%des entreprises parlent d'un 'impact significatif ou important'. Fait remarqua- ble, la perception de la fraude comme un risquecroissantalégèrementdiminué(de 87%à80%).Cependant,l'essordel'IAest perçu comme un risque croissant, même si les entreprises mentionnent également l'IAprincipalementcommeunoutildedé- tectionde la fraude. Une autre tendance frappante est qu'un nombre croissant d'entreprises se disent suffisamment protégées, alors que le nombre de fraudes et l'impact des dom- mages augmentent. Ce 'sentiment de sé- curité'reposesouventsuruneidéefausse, par exemple la croyance erronée que l'as- surancegénéralenon-viecouvrelesconsé- quences de la fraude. La fraude interne reste la plus persistante La fraude commise par les propres em- ployés (fraude interne) reste la forme de fraude la plus courante. 66 % des entre- prises en ont été victimes au cours des 12 derniersmois.Lesquatreprincipauxtypes de fraude interne sont : 1) le vol de mar- chandises,2)lafalsificationdedocuments, 3) le vol d'argent, 4) la collaboration avec des fraudeurs externes. Fraude externe 55%des entreprises ont été confrontées à lafraudeexternel'annéedernière.Ilestin- téressantdenoterquelafraudesurfacture, qui occupait la première place ces der- nières années, est passée en troisième po- sition. La fraude via des partenaires ex- ternes est désormais la forme la plus courante. La deuxième place est occupée parlevolouladestructiondedonnées.La baissedelafraudesurfacturepeuts'expli- querparlaforteaugmentationdel'utilisa- tiondes systèmes de facturation en ligne. Les pourcentages de fraude interne et ex- ternesontnonseulementplusélevésqu'en 2024, mais le montant des dommages augmente également. En 2024, moins de 25 % des organisations avaient subi des pertes supérieures à100.000euros à cause de la fraude interne ; en2025, ce chiffre est passé à 39%. Pour la fraude externe, 44% desorganisationsontsubidesdommages dépassant 100.000 euros. Fraudeur interne : souvent un employé senior Parmi les fraudeurs internes, les em- ployés seniors arrivent toujours en tête (53 %). Ils connaissent parfaitement les systèmes et les processus de l'entreprise. Comme ils sont souvent employés de- puis longtemps, ils jouissent d'une grande confiance de la part de leurs col- lègues et de leurs supérieurs. Cela leur donne toutes sortesde libertés, ycompris l'accès àdes systèmes et des informations confidentiels. Les fraudeurs internes se retrouvent dans tous les départements. Les 'Opérations'arrivententêteavec40% (principalement pour levol demarchan- dises), suivies des achats (26%), ducom- merce (19 %) et des finances (16 %). Le travail à domicile L'enquête montre que le travail à domi- cile continue d'augmenter le risque de fraude interne et externe.Alors que 55% des entreprises reconnaissaient ce risque en 2024, elles sont aujourd'hui 67 %. Le nombre d'entreprises ayant pris desme- sures préventives est également passéde 61 % à 69 %. 75%des entreprises ne signalent pas à la police Ilestdifficilepourlesentreprisesdelutter seules contre la fraude. Elles se tournent de plus en plus vers des parties externes pour obtenir de l'aide - principalement vers des assureurs. De moins en moins d'entreprises signalent les casde fraudeà la police. L'année dernière, plus de 60 % d'entre elles ne l'ont pas fait ; cette année, ce chiffre est passé à 73%. Beaucoup d'ambiguïté sur la façon de s'assurer contre la fraude De plus en plus d'entreprises pensent être assurées contre les conséquences de la fraude. Par exemple, 57 % déclarent avoir une assurance contre la fraude (contre 41%en2024). 44%affirment dis- poserd'unecyber-assurance(contre34%) et56%disposentd'uneassurancerespon- sabilité civile des administrateurs (contre 45 %). Cela montre que les entreprises et lesorganisationsontuneperceptionerro- née de la couverture contre la fraude. Ainsi,45%desorganisationspensentpar exemple que tous les dommages liés à la fraude sont couverts par une assurance non-vie classique, comme une assurance responsabilité civile ou une cyber-assu- rance—ce qui est incorrect. Près de la moitié des organisations pen- sent, à tort, que leur cyber-assurance les protège adéquatement contre la fraude numérique. Cen'est pas le cas. Par exem- ple,lesdommagescausésparunefacture fantôme envoyée par courriel par un fraudeur ne sont généralement pas cou- verts par une cyber-assurance. La fraude touche de plus en plus d'entreprises au Benelux ©AllianzTrade