Insurance Nexus · research and knowledge platform; GRA does not lend, invest, insure, rate or advise
Insurance Nexus: domain, methods and instruments
Insurance risk is the uncertainty an insurer accepts when it promises to pay for future losses: how often they occur, how large they are and how they accumulate. Each instrument below is illustrative, educational and non-binding.
Domain definition
What Insurance Nexus studies
Insurance risk is the uncertainty an insurer accepts when it promises to pay for future losses: how often they occur, how large they are and how they accumulate. Insurance Nexus, The Global Risks Alliance's platform for the sector, explains how that risk is written, priced, reinsured and supervised, and where losses fall outside cover. It shares industry understanding and does not underwrite, price or advise.
Insurance Nexus studies how risk becomes insured loss and how insured loss spreads. Insured loss spreads through carriers, brokers, reinsurers and capital markets, through households and firms left without cover, and through supervisors, pools and public budgets that carry what markets do not. Its pages set out the mechanisms, data, methods and decisions behind underwriting, catastrophe exposure, health and life risk, climate and protection gaps, with every value labelled by its evidence. It shares industry understanding and does not underwrite, price or advise.
| 术语 | 定义 |
|---|---|
| Insurance risk | The risk that claims and related expenses differ from what was assumed when policies were priced and reserved, through frequency, severity, timing or accumulation. |
| 保护缺口 | The share of economic loss from an event or risk that is not covered by insurance and falls on households, firms or public budgets. |
| 再保险 | Insurance bought by insurers to transfer part of their risk to another carrier, limiting the effect of large or accumulated losses on their capital. |
| Solvency capital requirement | The capital a supervisor requires an insurer to hold so that it can absorb severe losses; Solvency II calibrates it to a one-year horizon at a stated confidence level. |
| 巨灾模型 | Software that combines hazard, exposure and vulnerability modules to simulate many possible events and estimate the distribution of insured loss. |
| Own Risk and Solvency Assessment | An insurer's own assessment of its risks and the capital it needs over its business plan horizon, required under IAIS ICP 16 and Solvency II. |
Financial system context
The parts of the financial system it covers
-
市场结构
保险公司、经纪人、再保险公司、管理代理人、相互保险组织、塔卡夫运营方、共保体和资本市场工具,以及风险和责任如何在它们之间传递。
-
承保与定价
风险因素、费率结构、数据局限,以及风险分类和定价方面的职业和监管边界。
-
财产与巨灾
自然和人为巨灾的风险暴露、累积、巨灾建模,以及财产与意外险风险的组合视角。
-
健康、寿险与长寿风险
Mortality, morbidity, longevity and pandemic risk, and the boundary between private cover and social protection.
-
理赔与韧性
Claims handling after catastrophes, fraud, cyber and the operational continuity of the systems policyholders depend on.
-
保障缺口
未投保损失所占比例、其持续存在的原因,以及它如何传导给家庭、贷款机构和公共预算。
-
气候与自然
Changing physical hazard, nature-related risk and the conditions under which risks stay insurable at an affordable price.
-
保险科技与 AI
销售、承保和理赔中的数据、模型和自动化,以及对其影响客户的方式的治理。
风险分类体系
Risks, topic by topic
-
保险市场结构
保险将风险从无力承担的一方转移给能够汇集或持有风险的一方。
-
承保与定价风险
Underwriting decides which risks an insurer accepts and on what terms.
-
财产与意外险
财产与意外险承保有形资产的损坏以及对他人的责任。
-
健康险与人寿险
健康与人寿保险承担疾病、残疾、死亡以及寿命超过储蓄可支撑年限的风险。
-
理赔与运营韧性
理赔是检验保险承诺的时刻。
-
保险保障缺口
保障缺口是指没有任何保险覆盖的经济损失部分。
-
气候与自然风险
气候变化改变着保险人所承保危害的频率、强度和位置,而自然丧失则移除了湿地、红树林和森林等曾经减轻损害的缓冲。
-
Insurtech and ai
数据、自动化和机器学习如今已渗透到保险的每一个环节:分销、承保、定价、理赔和欺诈检测。
-
塔卡福(伊斯兰互助保险)风险
Takaful 是依据符合伊斯兰商法的互助原则组织的保险。
Risk-transmission mechanisms
How these risks develop and spread
| 主题 | 机制 | 运作方式 |
|---|---|---|
| 保险市场结构 | Risk passes along a chain | A household or firm cedes risk to a primary insurer, which cedes part to reinsurers, which retrocede part again or transfer it to investors through catastrophe bonds and collateralised vehicles. Each link has its own capital, contract terms and counterparty risk. |
| 保险市场结构 | 能力随周期配置 | After large losses or capital shocks, capacity shrinks and prices and terms harden. As capital returns, they soften. Reinsurance renewals concentrated on a few dates transmit the cycle quickly into primary pricing and availability. |
| 承保与定价风险 | Risk factors and classification | 费率因子按预期损失对投保人分组。某一变量可能预测效果很好,却违法、不公平,或是受保护特征的代理变量,因此分类既是统计问题,也是法律和伦理问题。 |
| 承保与定价风险 | 数据局限与漂移 | 定价依赖过往索赔数据,而相关风险暴露在不断变化:维修和医疗成本的通胀、新增建筑存量、诉讼趋势以及不断变化的气候。过于陈旧或稀少的经验会使下一年的定价失准。 |
| 财产与意外险 | 危害遇上风险暴露 | Loss arises where a hazard footprint overlaps insured values whose vulnerability depends on construction, occupancy, age and local building codes. Exposure data quality often drives model error as much as the hazard does. |
| 财产与意外险 | Accumulation across a portfolio | Policies that look independent share one event. Accumulation control tracks sums insured by zone and peril, and catastrophe models convert them into event loss tables and exceedance probabilities. |
| 健康险与人寿险 | Longevity trend risk | 死亡率的改善需历经数十年累积。对趋势的小幅误判会在年金和养老金负债中不断放大,而误差只有在经验逐步显现后才会被察觉。 |
| 健康险与人寿险 | Morbidity and medical cost | Claims for illness and disability depend on incidence, recovery, treatment practice and the price of care. New therapies, ageing and changing diagnosis shift costs faster than tariffs and reserves adjust. |
| 理赔与运营韧性 | 巨灾理赔激增 | A single event can produce more claims in days than a normal year. Adjuster capacity, contractor supply and access to damaged areas set the pace, and delays raise both costs and customer harm. |
| 理赔与运营韧性 | Systems and third parties | Policy, claims and payment systems depend on cloud providers, software vendors, outsourced administrators and repair networks. A cyber incident or vendor failure can halt claims payment across a market. |
| 保险保障缺口 | 保险不足与未投保 | Some risks are not insured at all. Others are insured for less than their value, or with exclusions and deductibles that leave most of the loss with the owner. Both widen the gap. |
| 保险保障缺口 | 可负担性与可获得性 | 风险上升使保费超出家庭的支付能力,而在风险被判定为不可保的地方,保险公司可能退出。低收入、非正式产权和分销缺失使保障遥不可及。 |
| 气候与自然风险 | 不断变化的危害基线 | Warmer air and seas intensify heavy rain, heat, drought, wildfire weather and coastal flooding. Models calibrated on the past record understate risk if the baseline moves faster than they are recalibrated. |
| 气候与自然风险 | Exposure growth in hazard zones | 在洪泛区、海岸和野火与城市交界地带的开发,会在危害上升的地区抬高保险价值。风险暴露的增长对损失趋势的影响往往大于危害本身的变化。 |
| Insurtech and ai | 模型不透明与漂移 | 复杂模型难以向客户、监管机构和董事会解释。随着数据和行为的变化,模型表现会发生漂移,错误可能在无人察觉之前就扩散至整个业务组合。 |
| Insurtech and ai | 代理歧视 | Rich external data, from credit information to telematics and online behaviour, can stand in for protected characteristics. A model that never sees a protected attribute can still produce disparate outcomes. |
| 塔卡福(伊斯兰互助保险)风险 | Separation of funds | The participants' risk fund, the participants' investment fund and the operator's shareholder fund are kept apart. Losses and expenses must be allocated between them by rules set in the contract and supervised. |
| 塔卡福(伊斯兰互助保险)风险 | Operator models | 在 wakalah 模式下,运营方赚取费用;在 mudarabah 模式下,运营方分享投资利润;混合模式则兼具两者。每种模式给运营方带来不同的激励,也可能与参与者产生利益冲突。 |
研究问题
Who decides, and what they need to know
| Decision holder | 提问 | Evidence the platform's pages describe |
|---|---|---|
| 保险监管机构 | Whether capacity, concentration or reliance on a few reinsurers threatens policyholder protection in their market | 市场结构统计、分出比例和交易对手集中度,按来源标注 |
| Appointed and pricing actuaries | Whether proposed rates are adequate, not excessive and not unfairly discriminatory | 经验数据、趋势选择以及各费率变量的影响,并说明各项背后的假设 |
| Insurer and reinsurer chief risk officers | 按灾种和区域应承担多少巨灾风险暴露,以及应购买多少再保险 | Exceedance probability curves with model, vendor and data uncertainty shown |
| Life insurers and pension fund trustees | 应采用何种死亡率改善和长寿假设,以及是否转移长寿风险 | 经验研究、国家趋势以及改进尺度之间的差异 |
| Insurer boards and senior managers | Which important business services must stay within impact tolerance, and what investment that requires | 服务映射、依赖证据以及对照容忍度的情景测试结果 |
| 财政部门 | 预算中应自留多少灾害损失,以及应预先融资或转移多少 | Loss exceedance curves for public assets and contingent liabilities, with the cost and basis risk of each instrument |
| 保险与再保险公司 | How climate change affects underwriting appetite, pricing, reinsurance and capital over the planning horizon | Climate-conditioned exceedance curves, ORSA scenario analysis and sensitivity to hazard and exposure assumptions |
| Insurer boards and model risk committees | Which AI systems to deploy, with what oversight, testing and human review | 模型清单、验证与公平性测试、漂移监测和事件历史 |
| 伊斯兰保险(Takaful)运营方的董事会 | 应采用何种运营方模式,如何设定费用,以及如何为赤字提供资金 | 基金层面的结果、qard 历史、费用和开支分摊,以及 retakaful 条款 |
These are the decisions of the named institutions. The platform describes the evidence they use; it takes none of these decisions and advises on none.
Key indicators
What is measured, in which unit, by whom
| 指标 | 单位 | 节奏 | 来源类型 | 证据标签 |
|---|---|---|---|---|
| Gross written premium by line and market | 每年货币金额 | 年度 | Supervisory statistical returns | 已发布统计 |
| 保险渗透率 | premium as a percent of GDP | 年度 | 保费统计与国民账户的比对 | 估算 |
| 损失率 | incurred losses as a percent of earned premium | 每季度 | 法定财务报表 | 已发布统计 |
| 综合成本率 | losses and expenses as a percent of earned premium | 每季度 | 法定财务报表 | 已发布统计 |
| 年均损失 | 每年货币金额 | 年度 | 巨灾模型输出 | 模型估算 |
| Probable maximum loss at stated return period | 货币 | 年度 | Catastrophe model, occurrence and aggregate exceedance | 模型估算 |
| Age-specific mortality rate | 每千名暴露人口的死亡人数 | 年度 | Civil registration and insured experience | 观测 |
| Actual-to-expected mortality ratio | 比率 | 年度 | 保险公司经验研究 | 估算 |
| 理赔结算时间 | days from notification to payment | 每月 | Insurer claims systems | 已报告 |
| 每起事件的巨灾理赔数量 | 主张 | 每个事件 | 行业损失估算服务和保险公司报告 | 估算 |
| Natural catastrophe protection gap | uninsured share of economic loss, percent | 年度 | Catastrophe loss records | 估算 |
| Insured share of economic loss by event | 百分比 | 每个事件 | Catastrophe loss records | 估算 |
| Average annual loss under climate scenario | 每年货币金额 | 每个情景 | Catastrophe model with adjusted hazard | 模型估算 |
| Change in exceedance probability loss | 相对基线的百分比变化 | 每个情景 | 气候条件化巨灾模型 | 模型估算 |
| AI systems in use by business function | 按功能统计的系统数量 | 调查周期 | 对保险公司的监管调查 | 已报告 |
| 差异结果比率 | 各群体之间结果发生率之比 | 每次模型审查 | 公平性测试 | 估算 |
| Takaful contributions by line | 每年货币金额 | 年度 | 监管统计 | 已发布统计 |
| Takaful(伊斯兰互助保险)渗透率 | 缴款占 GDP 的百分比 | 年度 | 对照国民账户的贡献 | 估算 |
Indicators carry a name, a unit and the kind of evidence behind them. This page publishes no value for them: values belong to their sources, with their own dates and definitions.
方法论
The methods the platform's work rests on
-
Actuarial math
精算数学是保险的核心学科:它将赔付经验、风险暴露和假设转化为保费、准备金和资本。在本平台上,它界定了如何估计预期损失及其不确定性,以及专业判断和标准在何处设定边界。
-
极值
极值理论决定了损失分布尾部的形状:罕见事件的损失可能有多大,糟糕年份中损失如何累积,以及超出记录范围后这些估计的不确定性如何扩大。资本、再保险和巨灾定价都以它为基础。
-
Probability statistics
Probability and statistics provide the language of insurance: frequency and severity, dependence, credibility and uncertainty. They make assumptions explicit, test them against experience, and show how much confidence any estimate of loss deserves, from a single portfolio's claim frequency to a national estimate of how many households are uninsured.
-
气候科学
Climate and atmospheric science explains how the hazards that drive insured loss are changing: storms, floods, heat, drought, wildfire and sea level. On this platform it supplies the hazard evidence and scenarios on which catastrophe models, stress tests and insurability studies are conditioned, with their uncertainty.
-
Ai ml
Artificial intelligence and machine learning are used in insurance to read images and documents, detect fraud, triage claims, enrich exposure data and segment risk. On this platform they are treated both as methods and as a source of risk to be governed, tested and explained.
数据来源
证据的来源
| 来源 | 它提供什么 | 许可类别 |
|---|---|---|
| OECD 全球保险统计 | 各国的保费、赔付、投资与市场结构 | 免费,OECD 条款 |
| IAIS 全球保险市场报告(GIMAR) | Sector-wide monitoring of risks and trends from the global monitoring exercise | Public, IAIS terms |
| EIOPA 保险统计数据 | 《偿付能力 II》报告按国家和业务线汇总 | 公开,欧盟再利用条款 |
| NAIC 市场份额报告和统计汇编 | 按保险集团和险种划分的美国保费和市场份额 | 公开报告;详细数据受 NAIC 许可约束 |
| Swiss Re Institute sigma | 按地区划分的全球保险保费、渗透率和密度 | 专有研究,报告可免费阅读 |
| Lloyd's annual report and market statistics | 劳合社市场的承保能力、辛迪加业绩和授权承保权限 | 公开报告 |
| Artemis deal directory | Catastrophe bond and insurance-linked securities transactions | 专有,按条款可免费阅读 |
| NAIC 电子费率和表格备案系统(SERFF)公共访问 | Rate and rule filings with supporting actuarial memoranda, by state | 公开记录,各州条款 |
| Marsh 全球保险市场指数(Marsh Global Insurance Market Index) | Commercial insurance rate movement by region and line | 专有;公布的摘要可免费阅读 |
| Gallagher Re 1st View | 主要续保日的再保险续保定价与条款 | 专有,已发布报告可免费阅读 |
| EIOPA consumer trends report | Pricing practices, fair value and conduct concerns across EU markets | 公开,欧盟再利用条款 |
| 国家统计机构 | Price indices for repair, construction, medical and legal costs | 公开,条款因机构而异 |
| Casualty Actuarial Society and Society of Actuaries research | Ratemaking methods, experience studies and practice notes | Public research, terms per society |
| Verisk Property Claim Services(PCS) | Industry insured loss estimates for catastrophe events, used in index triggers | 专有,授权许可 |
| Munich Re NatCatSERVICE | 含经济损失和保险损失的自然灾难事件记录 | 专有;公布汇总统计数据 |
| 瑞士再保险研究院 sigma 自然巨灾报告 | 按地区和灾因划分的保险与经济巨灾损失 | 专有研究,报告可免费阅读 |
| Oasis Loss Modelling Framework | Open framework for running catastrophe models with exchangeable hazard, vulnerability and exposure | 开源,BSD 许可 |
| NOAA IBTrACS 与 Storm Events Database | Tropical cyclone tracks and severe weather event records | 公有领域 |
| Copernicus ERA5 和应急管理服务 | Reanalysis hazard fields and rapid event footprints | 免费开放,须署名 |
| Global Earthquake Model (GEM) hazard and risk maps | Open seismic hazard and risk models | Creative Commons, terms per product |
| OpenFEMA National Flood Insurance Program data | 美国洪水保单与理赔 | 公有领域 |
| 人类死亡率数据库 | Detailed national mortality and population data for life tables | 注册后免费,采用 Creative Commons 署名许可 |
| 联合国《世界人口展望》 | Population estimates and projections, mortality and fertility by country | Creative Commons attribution, IGO(知识共享署名,政府间组织) |
| 世界卫生组织全球卫生支出数据库 | Health spending by source, including out-of-pocket and voluntary insurance | Creative Commons non-commercial, IGO(知识共享非商业性使用,政府间组织版) |
Instruments explained for research
Analytical instruments
Each instrument runs a named, cited method on stated teaching values, shows its uncertainty and limits, and keeps its data in a table. All of them are illustrative, educational and non-binding: none quotes a premium, a rate, a price, a rating or an eligibility, and none describes a real institution, portfolio or place.
Loss distribution
Loss exceedance, annual average loss and probable maximum loss
Illustrative, educational and non-binding
How often a loss of each size is exceeded, from a hazard curve and a vulnerability curve. The area under the curve is the annual average loss; the 1 in 100 and 1 in 250 points are the probable maximum losses that capital and reinsurance discussions refer to.
- Loss at the teaching values
- 5 to 95 % of runs
- 25 to 75 % of runs
Read the curve from left to right: frequent, small losses on the left, rare and severe ones on the right. Raising the protection standard removes the frequent losses; widening the spread of extremes lifts the tail.
控制措施
- 典型年度危害强度(所在地)
- 1.00 强度指数
- 极端值的扩散(尺度)
- 强度指数 0.80
- 损失一半价值时的强度
- 4.00 强度指数
- 脆弱性的离散程度
- 0.55
- 危害足迹中的价值占比
- 40 %
- 防护标准(重现期)
- 50 年
结果
- Annual average loss (% of exposed value a year)
- 0.51
- Loss at 1 in 100 years (% of exposed value)
- 24.5
- Loss at 1 in 250 years (% of exposed value)
- 28.4
- Uncertainty band
- 200 runs over the plausible ranges
Data, source and model record
| 重现期(年) | Loss (% of exposed value) |
|---|---|
| 1 | 0.00 |
| 2 | 0.00 |
| 4 | 0.00 |
| 7 | 0.00 |
| 13 | 0.00 |
| 24 | 0.00 |
| 44 | 0.00 |
| 82 | 23.5 |
| 154 | 26.4 |
| 287 | 28.9 |
| 536 | 30.9 |
| 1,000 | 32.5 |
- 来源
- Computed by the estate's hevl model (the network's own records, sha256 e8f84afa4822), the same code the platform's existing simulators run. Methods: Gumbel extreme-value hazard (Gumbel (1958), Statistics of Extremes, Columbia University Press); Lognormal fragility (vulnerability) (FEMA (2003), HAZUS-MH Technical Manual); Exceedance-probability curve and average annual loss (Grossi and Kunreuther (2005), Catastrophe Modeling, Springer); Uncertainty fan (parameter Monte Carlo) (Spiegelhalter, Pearson and Short (2011), Visualizing uncertainty about the future, Science 333); One-at-a-time sensitivity (tornado) (Saltelli and others (2008), Global Sensitivity Analysis: The Primer, Wiley).
- 日期
- Instrument record 1.0.0, 2026-11-16; teaching values from the estate's instrument catalogue (the network's own records).
- 范围
- A hypothetical portfolio of property in one hazard footprint. It names no place, insurer, insured or event, and represents none.
- 单位
- Losses in percent of the total exposed value of the portfolio; return periods in years; probabilities in percent a year.
- Currency
- None: every value is a share (percent), so no currency or exchange rate applies
- 状态
- Illustrative, educational and non-binding: no premium, rate, price, rating, eligibility or advice is given or implied.
Model record
- 建模
- gra.hevl.catastrophe
- 版本
- 1.0.0
- 负责人
- GRA platforms lane (model code: the estate's instrument engine, W13)
- 目的
- To show how a hazard curve and a vulnerability curve combine into a loss exceedance curve, and how annual average loss and probable maximum loss are read from it.
- Risk domain
- Insurance and catastrophe risk
- 假设
- Annual maximum hazard intensity follows a Gumbel distribution. Damage follows a lognormal fragility curve in hazard intensity. Losses below the protection standard are prevented entirely. One footprint, one peril, one event a year at most.
- 输入
- Typical annual hazard intensity (location) (intensity index); Spread of extremes (scale) (intensity index); Intensity at which half the value is lost (intensity index); Dispersion of vulnerability (ratio); Share of value in the hazard footprint (%); Protection standard (return period) (years)
- 数据
- None: teaching values set by the controls; no observed exposure, hazard or claims data.
- 方法
- Gumbel extreme-value hazard (Gumbel (1958), Statistics of Extremes, Columbia University Press); Lognormal fragility (vulnerability) (FEMA (2003), HAZUS-MH Technical Manual); Exceedance-probability curve and average annual loss (Grossi and Kunreuther (2005), Catastrophe Modeling, Springer); Uncertainty fan (parameter Monte Carlo) (Spiegelhalter, Pearson and Short (2011), Visualizing uncertainty about the future, Science 333); One-at-a-time sensitivity (tornado) (Saltelli and others (2008), Global Sensitivity Analysis: The Primer, Wiley)
- 验证
- The code is the estate engine's; this page's copy is checked against it on 300 random parameter sets (the network's own records). It is not calibrated to any portfolio, peril model or loss history.
- 局限性
- No secondary uncertainty in damage, no correlation across locations, no demand surge, no policy terms beyond the cover controls, and no event set: a teaching model of a catastrophe model's three blocks.
- 不确定性
- Parameter uncertainty is shown as a fan: the engine's own Monte Carlo over the plausible ranges of the catalogue (5 to 95 % and 25 to 75 % of runs).
- Approved use
- Education and research: understanding how exceedance curves, annual average loss and probable maximum loss are built and read.
- Prohibited use
- Pricing, quoting or binding cover; reserving; capital or solvency calculation; any statement about a real portfolio, place or insurer.
- Runs on
- Static: computed at build time in the site and in the reader's browser; no server, no data leaves the page.
不确定性
What moves the annual average loss
Illustrative, educational and non-binding
Each input moved alone across its plausible range, the others held at their teaching values: the longer the bar, the more the annual average loss depends on that input. This is where an analyst looks first for data.
- Input at the low end
- Input at the high end
Bars to the right raise the annual average loss, bars to the left lower it. The ranking, not the values, is the lesson.
Annual average loss at the teaching values: 0.51 % of exposed value a year.
Data, source and model record
| 输入 | 范围 | Annual average loss at the low end | Annual average loss at the high end |
|---|---|---|---|
| 防护标准(重现期) | 20.0 to 100 years | 1.03 | 0.28 |
| 损失一半价值时的强度 | 2.80 to 5.60 intensity index | 0.67 | 0.32 |
| 危害足迹中的价值占比 | 30.0 to 50.0 % | 0.38 | 0.63 |
| 极端值的扩散(尺度) | 0.60 to 1.00 intensity index | 0.38 | 0.60 |
| 典型年度危害强度(所在地) | 0.70 to 1.30 intensity index | 0.47 | 0.54 |
| 脆弱性的离散程度 | 0.39 to 0.72 | 0.54 | 0.48 |
- 来源
- Computed by the estate's hevl model (the network's own records, sha256 e8f84afa4822), the same code the platform's existing simulators run. Methods: Gumbel extreme-value hazard (Gumbel (1958), Statistics of Extremes, Columbia University Press); Lognormal fragility (vulnerability) (FEMA (2003), HAZUS-MH Technical Manual); Exceedance-probability curve and average annual loss (Grossi and Kunreuther (2005), Catastrophe Modeling, Springer); Uncertainty fan (parameter Monte Carlo) (Spiegelhalter, Pearson and Short (2011), Visualizing uncertainty about the future, Science 333); One-at-a-time sensitivity (tornado) (Saltelli and others (2008), Global Sensitivity Analysis: The Primer, Wiley). Sensitivity: the engine's tornado() (Saltelli and others, 2008).
- 日期
- Instrument record 1.0.0, 2026-11-16; teaching values from the estate's instrument catalogue (the network's own records).
- 范围
- A hypothetical portfolio of property in one hazard footprint. It names no place, insurer, insured or event, and represents none.
- 单位
- Losses in percent of the total exposed value of the portfolio; return periods in years; probabilities in percent a year.
- Currency
- None: every value is a share (percent), so no currency or exchange rate applies
- 状态
- Illustrative, educational and non-binding: no premium, rate, price, rating, eligibility or advice is given or implied.
Model record
- 建模
- gra.hevl.sensitivity
- 版本
- 1.0.0
- 负责人
- GRA platforms lane (model code: the estate's instrument engine, W13)
- 目的
- To show how a hazard curve and a vulnerability curve combine into a loss exceedance curve, and how annual average loss and probable maximum loss are read from it.
- Risk domain
- Insurance and catastrophe risk
- 假设
- Annual maximum hazard intensity follows a Gumbel distribution. Damage follows a lognormal fragility curve in hazard intensity. Losses below the protection standard are prevented entirely. One footprint, one peril, one event a year at most.
- 输入
- Typical annual hazard intensity (location) (intensity index); Spread of extremes (scale) (intensity index); Intensity at which half the value is lost (intensity index); Dispersion of vulnerability (ratio); Share of value in the hazard footprint (%); Protection standard (return period) (years)
- 数据
- None: teaching values set by the controls; no observed exposure, hazard or claims data.
- 方法
- Gumbel extreme-value hazard (Gumbel (1958), Statistics of Extremes, Columbia University Press); Lognormal fragility (vulnerability) (FEMA (2003), HAZUS-MH Technical Manual); Exceedance-probability curve and average annual loss (Grossi and Kunreuther (2005), Catastrophe Modeling, Springer); Uncertainty fan (parameter Monte Carlo) (Spiegelhalter, Pearson and Short (2011), Visualizing uncertainty about the future, Science 333); One-at-a-time sensitivity (tornado) (Saltelli and others (2008), Global Sensitivity Analysis: The Primer, Wiley)
- 验证
- The code is the estate engine's; this page's copy is checked against it on 300 random parameter sets (the network's own records). It is not calibrated to any portfolio, peril model or loss history.
- 局限性
- No secondary uncertainty in damage, no correlation across locations, no demand surge, no policy terms beyond the cover controls, and no event set: a teaching model of a catastrophe model's three blocks.
- 不确定性
- Parameter uncertainty is shown as a fan: the engine's own Monte Carlo over the plausible ranges of the catalogue (5 to 95 % and 25 to 75 % of runs).
- Approved use
- Education and research: understanding how exceedance curves, annual average loss and probable maximum loss are built and read.
- Prohibited use
- Pricing, quoting or binding cover; reserving; capital or solvency calculation; any statement about a real portfolio, place or insurer.
- Runs on
- Static: computed at build time in the site and in the reader's browser; no server, no data leaves the page.
保护缺口
The protection gap on the same loss curve
Illustrative, educational and non-binding
The same ground-up losses read through a simple cover: the share of value insured, a deductible and a limit per insured unit. The shaded area between the two curves is the loss nobody's policy pays: it falls on households, firms, lenders and public budgets.
- Ground-up loss
- 保险损失
- 保护缺口
Lower take-up widens the gap at every return period; a high deductible widens it most for frequent losses; a low limit widens it most in the tail.
控制措施
- Share of exposed value insured (take-up)
- 50 %
- Deductible
- 2.0 % of an insured unit's value
- Policy limit
- 80 % of an insured unit's value
结果
- Annual average loss, ground-up (% of exposed value a year)
- 0.51
- Annual average loss, insured (% of exposed value a year)
- 0.24
- Protection gap (% of the annual average loss)
- 52
- Uninsured loss at 1 in 100 years (% of exposed value)
- 12.6
Data, source and model record
| 重现期(年) | Ground-up loss (%) | Insured loss (%) | Uninsured loss (%) |
|---|---|---|---|
| 1 | 0.00 | 0.00 | 0.00 |
| 2 | 0.00 | 0.00 | 0.00 |
| 4 | 0.00 | 0.00 | 0.00 |
| 7 | 0.00 | 0.00 | 0.00 |
| 13 | 0.00 | 0.00 | 0.00 |
| 24 | 0.00 | 0.00 | 0.00 |
| 44 | 0.00 | 0.00 | 0.00 |
| 82 | 23.5 | 11.4 | 12.2 |
| 154 | 26.4 | 12.8 | 13.6 |
| 287 | 28.9 | 14.0 | 14.8 |
| 536 | 30.9 | 15.0 | 15.8 |
| 1,000 | 32.5 | 15.9 | 16.7 |
- 来源
- Computed by the estate's hevl model (the network's own records, sha256 e8f84afa4822), the same code the platform's existing simulators run. Methods: Gumbel extreme-value hazard (Gumbel (1958), Statistics of Extremes, Columbia University Press); Lognormal fragility (vulnerability) (FEMA (2003), HAZUS-MH Technical Manual); Exceedance-probability curve and average annual loss (Grossi and Kunreuther (2005), Catastrophe Modeling, Springer); Uncertainty fan (parameter Monte Carlo) (Spiegelhalter, Pearson and Short (2011), Visualizing uncertainty about the future, Science 333); One-at-a-time sensitivity (tornado) (Saltelli and others (2008), Global Sensitivity Analysis: The Primer, Wiley). Cover: take-up, deductible and limit applied per insured unit (this page's model, the network's own records).
- 日期
- Instrument record 1.0.0, 2026-11-16; teaching values from the estate's instrument catalogue (the network's own records).
- 范围
- A hypothetical portfolio of property in one hazard footprint. It names no place, insurer, insured or event, and represents none. Hazard and vulnerability are held at the teaching values.
- 单位
- Losses in percent of the total exposed value of the portfolio; return periods in years; probabilities in percent a year.
- Currency
- None: every value is a share (percent), so no currency or exchange rate applies
- 状态
- Illustrative, educational and non-binding: no premium, rate, price, rating, eligibility or advice is given or implied.
Model record
- 建模
- gra.cover.protection-gap
- 版本
- 1.0.0
- 负责人
- GRA platforms lane (model code: the estate's instrument engine, W13)
- 目的
- To show how take-up, deductibles and limits turn a ground-up loss curve into an insured one, and where the protection gap sits on the curve.
- Risk domain
- Insurance and catastrophe risk
- 假设
- Annual maximum hazard intensity follows a Gumbel distribution. Damage follows a lognormal fragility curve in hazard intensity. Losses below the protection standard are prevented entirely. One footprint, one peril, one event a year at most. Insured and uninsured units have the same exposure and damage. Cover terms apply per unit, with no aggregate limits or reinstatements.
- 输入
- Share of exposed value insured (%); Deductible (% of an insured unit's value); Policy limit (% of an insured unit's value)
- 数据
- None: teaching values set by the controls; no observed exposure, hazard or claims data.
- 方法
- Gumbel extreme-value hazard (Gumbel (1958), Statistics of Extremes, Columbia University Press); Lognormal fragility (vulnerability) (FEMA (2003), HAZUS-MH Technical Manual); Exceedance-probability curve and average annual loss (Grossi and Kunreuther (2005), Catastrophe Modeling, Springer); Uncertainty fan (parameter Monte Carlo) (Spiegelhalter, Pearson and Short (2011), Visualizing uncertainty about the future, Science 333); One-at-a-time sensitivity (tornado) (Saltelli and others (2008), Global Sensitivity Analysis: The Primer, Wiley)
- 验证
- The code is the estate engine's; this page's copy is checked against it on 300 random parameter sets (the network's own records). It is not calibrated to any portfolio, peril model or loss history.
- 局限性
- No secondary uncertainty in damage, no correlation across locations, no demand surge, no policy terms beyond the cover controls, and no event set: a teaching model of a catastrophe model's three blocks.
- 不确定性
- Parameter uncertainty is shown as a fan: the engine's own Monte Carlo over the plausible ranges of the catalogue (5 to 95 % and 25 to 75 % of runs).
- Approved use
- Education and research: understanding how exceedance curves, annual average loss and probable maximum loss are built and read.
- Prohibited use
- Pricing, quoting or binding cover; reserving; capital or solvency calculation; any statement about a real portfolio, place or insurer. Measuring the protection gap of any real market or country.
- Runs on
- Static: computed at build time in the site and in the reader's browser; no server, no data leaves the page.
Reinsurance structure
An excess-of-loss layer on the insured loss
Illustrative, educational and non-binding
A layer that pays the insured loss above a retention, up to a limit. It shows the layer's expected annual loss, the share of the insured annual average loss it takes, and how often it is reached and exhausted. It shows no premium, rate or price, and none can be read from it.
- 保险损失
- Loss in the layer
Raising the retention makes the layer rarer and its expected loss smaller; widening the limit adds expected loss from ever rarer years. The protection standard (1 in 50 years at the teaching values) stops every smaller loss, so the insured loss jumps from zero to about 14 % of value there: a layer retaining less than that is reached in every year the protection fails.
控制措施
- 自留
- 15.00 % of exposed value
- Layer limit
- 4.00 % of exposed value
结果
- Insured annual average loss (% of exposed value a year)
- 0.34
- Expected annual loss to the layer (% of exposed value a year)
- 0.037
- Share of insured annual average loss in the layer (%)
- 11
- Probability the layer is reached (% a year)
- 1.56
- Probability the layer is exhausted (% a year)
- 0.45
Data, source and model record
| 重现期(年) | Insured loss (%) | Loss in the layer (%) |
|---|---|---|
| 1 | 0.00 | 0.00 |
| 2 | 0.00 | 0.00 |
| 4 | 0.00 | 0.00 |
| 7 | 0.00 | 0.00 |
| 13 | 0.00 | 0.00 |
| 24 | 0.00 | 0.00 |
| 44 | 0.00 | 0.00 |
| 82 | 15.9 | 0.92 |
| 154 | 17.9 | 2.95 |
| 287 | 19.6 | 4.00 |
| 536 | 21.1 | 4.00 |
| 1,000 | 22.2 | 4.00 |
- 来源
- Computed by the estate's hevl model (the network's own records, sha256 e8f84afa4822), the same code the platform's existing simulators run. Methods: Gumbel extreme-value hazard (Gumbel (1958), Statistics of Extremes, Columbia University Press); Lognormal fragility (vulnerability) (FEMA (2003), HAZUS-MH Technical Manual); Exceedance-probability curve and average annual loss (Grossi and Kunreuther (2005), Catastrophe Modeling, Springer); Uncertainty fan (parameter Monte Carlo) (Spiegelhalter, Pearson and Short (2011), Visualizing uncertainty about the future, Science 333); One-at-a-time sensitivity (tornado) (Saltelli and others (2008), Global Sensitivity Analysis: The Primer, Wiley). Layer: expected loss to an excess-of-loss layer integrated over the same exceedance curve (Grossi and Kunreuther, 2005; this page's model).
- 日期
- Instrument record 1.0.0, 2026-11-16; teaching values from the estate's instrument catalogue (the network's own records).
- 范围
- A hypothetical portfolio of property in one hazard footprint. It names no place, insurer, insured or event, and represents none. The cover is fixed at 70 % take-up, a 2 % deductible and an 80 % limit.
- 单位
- Losses in percent of the total exposed value of the portfolio; return periods in years; probabilities in percent a year.
- Currency
- None: every value is a share (percent), so no currency or exchange rate applies
- 状态
- Illustrative, educational and non-binding: no premium, rate, price, rating, eligibility or advice is given or implied.
Model record
- 建模
- gra.reinsurance.xl-layer
- 版本
- 1.0.0
- 负责人
- GRA platforms lane (model code: the estate's instrument engine, W13)
- 目的
- To show how an excess-of-loss layer sits on an insured loss curve: its attachment and exhaustion probabilities and its expected loss.
- Risk domain
- Insurance and catastrophe risk
- 假设
- Annual maximum hazard intensity follows a Gumbel distribution. Damage follows a lognormal fragility curve in hazard intensity. Losses below the protection standard are prevented entirely. One footprint, one peril, one event a year at most. One layer, one event a year, no reinstatements, no aggregate features.
- 输入
- Retention (% of exposed value); Layer limit (% of exposed value)
- 数据
- None: teaching values set by the controls; no observed exposure, hazard or claims data.
- 方法
- Gumbel extreme-value hazard (Gumbel (1958), Statistics of Extremes, Columbia University Press); Lognormal fragility (vulnerability) (FEMA (2003), HAZUS-MH Technical Manual); Exceedance-probability curve and average annual loss (Grossi and Kunreuther (2005), Catastrophe Modeling, Springer); Uncertainty fan (parameter Monte Carlo) (Spiegelhalter, Pearson and Short (2011), Visualizing uncertainty about the future, Science 333); One-at-a-time sensitivity (tornado) (Saltelli and others (2008), Global Sensitivity Analysis: The Primer, Wiley)
- 验证
- The code is the estate engine's; this page's copy is checked against it on 300 random parameter sets (the network's own records). It is not calibrated to any portfolio, peril model or loss history.
- 局限性
- No secondary uncertainty in damage, no correlation across locations, no demand surge, no policy terms beyond the cover controls, and no event set: a teaching model of a catastrophe model's three blocks.
- 不确定性
- Parameter uncertainty is shown as a fan: the engine's own Monte Carlo over the plausible ranges of the catalogue (5 to 95 % and 25 to 75 % of runs).
- Approved use
- Education and research: understanding how exceedance curves, annual average loss and probable maximum loss are built and read.
- Prohibited use
- Pricing, quoting or binding cover; reserving; capital or solvency calculation; any statement about a real portfolio, place or insurer. Any premium, rate on line, price or placement of reinsurance.
- Runs on
- Static: computed at build time in the site and in the reader's browser; no server, no data leaves the page.
情景分析
Scenarios, not forecasts
A scenario is a set of named assumptions about the future, run through a model to see what follows; it is not a prediction. The instruments above move one assumption at a time; scenario work moves several together, with every assumption stated.
Regulatory context and related standards
The frameworks this work is read against
| 框架 | 机构 | 它规定了什么 | Edition or date | Where it applies |
|---|---|---|---|---|
| 保险核心原则与 ComFrame | 国际保险监督官协会(IAIS) | 保险监管的全球基准,以及国际活跃保险集团的共同框架,包括保险资本标准 | ICPs and ComFrame adopted November 2019, with later revisions; Insurance Capital Standard adopted December 2024 | IAIS members, through national law and supervision |
| Solvency II (Directive 2009/138/EC, as amended by Directive (EU) 2025/2) | European Union and EEA, with EIOPA technical standards and guidelines | 针对保险公司和再保险公司的风险资本、治理和披露要求,按一年内两百年一遇损失校准 | Applied from 1 January 2016; amending Directive (EU) 2025/2 published January 2025 | European Union and EEA |
| IFRS 17《保险合同》 | 国际会计准则理事会 | 保险合同负债的计量和列报,包括风险调整和合同服务边际 | Issued May 2017, amended June 2020; effective 1 January 2023 | Jurisdictions that apply IFRS |
| 自身风险与偿付能力评估(ORSA) | IAIS ICP 16、Solvency II、NAIC ORSA 示范法 | 保险公司对其业务规划期内自身风险和资本需求的自我评估,并向监管者报告 | IAIS ICP 16; Solvency II Article 45; NAIC Risk Management and ORSA Model Act (2012) | Global standard; EU and EEA; US states that adopt the model act |
| TCFD 建议与 IFRS S2 气候相关披露 | 金融稳定理事会工作组;国际可持续准则理事会 | 气候相关治理、战略、风险管理、指标和目标的披露,包括情景分析 | TCFD recommendations 2017 (task force disbanded October 2023); IFRS S2 issued June 2023, amended December 2025 | Where a jurisdiction adopts IFRS S2 or requires TCFD-aligned reporting |
| Guidance on climate change materiality and scenarios in the ORSA | 欧洲保险和职业养老金管理局(EIOPA) | 保险公司如何评估气候风险的重要性,并在 ORSA 中运行长期气候情景 | EIOPA Opinion of April 2021 and application guidance of August 2022 | European Union and EEA |
| Sendai Framework for Disaster Risk Reduction | UNDRR | Priorities and targets for understanding disaster risk and investing in resilience, including risk transfer and financial protection | 2015 to 2030 | UN member states (voluntary) |
| InsuResilience 惠及贫困人口原则 | InsuResilience Global Partnership | 气候与灾害风险融资和保险的原则,使贫困和脆弱人群获得公平价值 | As published by the partnership; edition date to be confirmed with the partnership | Voluntary, partnership members |
| 国际精算实务准则与精算实务准则 | 国际精算师协会;精算标准委员会(美国) | 精算工作的职业标准,包括风险分类、巨灾建模和对模型的依赖 | ISAP 1 and later ISAPs (IAA, from 2018); US ASOPs in their current revisions | IAA model standards for adoption by actuarial bodies; United States |
| Takaful(伊斯兰互助保险)治理、偿付能力和风险管理标准 | 伊斯兰金融服务委员会(IFSB) | 塔卡夫和再塔卡夫机构的治理、偿付能力和风险管理原则 | IFSB-8 (2009), IFSB-11 (2010), IFSB-14 (2013) and later standards | IFSB member jurisdictions |
Frameworks change and apply differently by jurisdiction. Check the current version and its applicability with the issuing body before relying on it; nothing here is legal or compliance guidance, and no comparison here is a supervisory determination.
Institutional counterparts
The institutions this work speaks to
- 保险监管机构
- 原保险公司董事会
- Reinsurers and insurance-linked securities investors
- Competition authorities and finance ministries
- Appointed and pricing actuaries
- 保险监管机构和费率监管机构
- Underwriting committees
- 立法者和平等机构
- Insurer and reinsurer chief risk officers
- Reserving actuaries and audit committees
- 建筑规范和土地利用主管部门
- Life insurers and pension fund trustees
- Health ministries and social insurance agencies
- 健康保险公司
- Insurer boards and senior managers
- 行为监管机构和审慎监管机构
研究出版物
Research, methods and sources
Educational resources
Learn the methods
Readiness workflows
金融准备度
Finance readiness is the structured preparation of evidence for financial-sector analysis and due diligence. It is not financing approval: reaching any readiness stage never makes a project, institution or portfolio eligible for funding, cover or credit.
Contribution mechanisms
参与
边界
What this page is not
Insurance Nexus is part of The Global Risks Alliance, which does not accept deposits, extend credit, underwrite or bind insurance, arrange or execute investments, offer securities, broker regulated products, issue credit ratings or give financial-product recommendations. Every instrument on this page is illustrative, educational and non-binding: it quotes no premium, rate, price, rating or eligibility, and it describes no real institution, portfolio or place.