Recommendations Concerning the Terrorism Section of A.M. Best s Supplemental Rating Questionnaire. February 20, 2004

Transcription

1 Recommendations Concerning the Terrorism Section of A.M. Best s Supplemental Rating Questionnaire February 20, 2004

2 INTRODUCTION A.M. Best Company s recent additions to the Supplemental Rating Questionnaire (SRQ), due April 1, 2004, require you to demonstrate the extent of your company s exposure to terrorism risk and to disclose how you currently manage that exposure. At ISO and AIR, we fully understand the information being requested by A.M. Best and the alternative responses that you may be considering. Using the AIR terrorism model s portfolio exposure management and loss analysis tools, we have been helping insurers address these issues for the past two years. Immediately following the events of September 11, 2001, AIR began developing a comprehensive terrorism model in conjunction with a team of counter-terrorism experts. The result is a model that is currently being used by insurers, reinsurance intermediaries and reinsurers to quantify the financial impact of conventional and chemical, biological, radiological and nuclear (CBRN) attacks, whether carried out by domestic or by international terrorist groups. Best practice in terrorism risk management has moved well beyond simple exposure concentration analysis. With the AIR model, you can discover your largest exposures and largest concentrations of exposures, estimate the impact of specific (deterministic) attacks on a concentration at risk, and perform a complete probabilistic loss analysis of your portfolio to generate an exceedance probability (EP) curve. The newly introduced terrorism section of the A.M. Best SRQ is intended to determine the progress insurers have made in managing their terrorism exposure. While the questionnaire does provide insurers with the option to report simple exposure concentrations, more sophisticated responses incorporating risk concentration and modeled loss analyses are favored. After consultation with A.M. Best, AIR has developed this white paper to describe in detail how AIR s suite of tools and services can provide comprehensive answers to questions 42, 43, and 44 of the SRQ. This white paper has been reviewed by A.M. Best and found to be an acceptable approach to meeting A.M. Best s needs. AIR can work with your company to develop comprehensive responses to the terrorism section of the questionnaire by the April 1 deadline. We will also be happy to discuss with you how applications such as CATStation and CLASIC/2 can be used not only to develop your responses to the questionnaire, but also to significantly enhance your overall terrorism risk management practices. CONFIDENTIAL 1

3 DEVELOPING COMPREHENSIVE RESPONSES TO THE NEW TERRORISM SECTION OF THE A.M. BEST SUPPLEMENTAL RATING QUESTIONNAIRE Below are some recommended guidelines for developing comprehensive responses to Questions 42, 43 and 44 of the A.M. Best SRQ. QUESTION 42: Largest Single Insured At Risk. Insureds At Risk are defined as properties such as trophy buildings; major government, state and municipal buildings; nuclear power plants; bridges and tunnels; stadiums; tourist attractions; and any other insured properties which you consider to be at risk or highly susceptible to terrorist acts Definition of Question A.M. Best is asking insurers to identify the top five At Risk locations, considering both property and workers compensation exposure, based on Total Gross Losses. A.M. Best defines maximum occurrence Total Gross Losses as the sum of the maximum occurrence gross losses for all TIV, Workers Compensation (WC) and other related coverages (such as business interruption or allied lines coverages) the insurer writes at each property. More specifically, the maximum occurrence gross losses for TIV and WC are defined as: TIV: Maximum amount payable by insurer for all property coverages subject to policy limits WC: Maximum losses payable assuming death benefits for all insured employees of that insurer in that building Description of Solution AIR believes that it is critical to use a credible, expert-developed database of at-risk properties to answer this question. AIR has the most comprehensive, well-validated target database currently available. Over the past two years, AIR has worked to create, crosscheck, and maintain a Landmark Database of more than 300,000 at-risk locations across the U.S., featuring a subset of approximately 30,000 locations considered to be at higher risk, that are consistent with the A.M. Best definitions and are also consistent with the Homeland Security Department s identified potential targets. AIR s Landmark Database, which is illustrated in Figure 1, was developed based on input provided by a panel of terrorism experts, and includes the range of property types suggested by A.M. Best as representative of at-risk properties. CONFIDENTIAL 2

4 Figure 1: AIR Landmark Database Distribution By working with AIR, an insurer can overlay their portfolio on our Landmark Database, as shown in Figure 2, to identify which properties should be reported in the SRQ as being atrisk. AIR can identify your largest at-risk insureds and provide you a completed table for question 42, including the requested Maximum Occurrence Gross Losses and Pre-Tax Net Losses. Figure 2: Sample Insurers Exposures and AIR Landmarks, Midtown Manhattan CONFIDENTIAL 3

5 QUESTION 43: Largest Single Insured (TIV). TIV is defined as total insured value. Definition of Question A.M. Best is asking insurers to identify the top five single insured locations considering both property and workers compensation exposure based on maximum occurrence Total Gross Losses. Again, A.M. Best defines maximum occurrence Total Gross Losses as the sum of the maximum occurrence gross losses for all TIV, Workers Compensation (WC) and other related coverages (such as business interruption and allied lines coverages) the insurer writes at each property. More specifically, the maximum occurrence gross losses for TIV and WC are defined as: TIV: Maximum amount payable by insurer for all property coverages subject to policy limits WC: Maximum losses payable assuming death benefits for all insured employees of that insurer in that building Description of Solution This question is similar to question 42, except that it asks insurers to identify their largest single insured location without respect to whether the location is considered at-risk. AIR has the ability to identify these largest single insureds across lines of business, as illustrated in Figure 3, to assist you in completing the table in response to question 43. This can be achieved by mapping multi-line exposures to identify accumulations. Figure 3: Identification of Largest Single Insured CONFIDENTIAL 4

6 QUESTION 44: Zonal Aggregate Exposure. Zonal Aggregate exposure identifies your rating unit s largest aggregate concentration of risk using specified geocoded areas, loss scenarios, and loss distributions. Definition of Question A.M. Best is looking for modeled losses for the insurer s top five zonal aggregations of exposure. Insurers are to identify their top five zonal aggregations based on modeled Total Gross Losses, as opposed to questions 42 and 43, which ask for maximum occurrence losses at single sites. The zonal aggregation includes all property, workers compensation, and other related coverages exposures within a radius determined by the insurer. Thus it includes multiple locations in addition to multiple lines of business. They ask insurers to report modeled gross losses based on deterministic loss scenarios (select weapon type). Zones can be defined as centered either on the insurer s locations or on potential targets. The insurer can choose either method. The Radius Used should reflect significant damage possibilities from a conventional weapon selected by the insurer. This weapon selection is what is meant by loss scenarios. The modeled loss should reflect a weapon damage model, referred to as loss distribution, which recognizes the decrease in damageability with increased distance from the point of detonation. Description of Solution AIR modeling will help insurers identify their zonal aggregations of exposure and the corresponding potential terrorism losses as described in Question 44 of the SRQ. AIR modeling capabilities support the full range of acceptable approaches to this question. A simple approach is to identify the largest zonal accumulations of exposure. A.M. Best permits insurers to select the radius with which they will identify zonal concentrations. Since the SRQ only considers potential losses from conventional weapons, one approach would be to accumulate risk within a radius appropriate to the weapon type, which AIR can help select. Figure 4 demonstrates the ability of AIR software to identify and accumulate exposure concentrations within the selected radius of a selected location. AIR can provide exposure accumulation centered on either the landmark targets or the insurer s locations. CONFIDENTIAL 5

7 Figure 4: Identification of Zonal Aggregate Exposure Concentrations A.M. Best prefers that insurers go beyond zonal exposure concentrations, and report modeled losses on a zonal aggregation basis. AIR s terrorism model can be utilized to run specific, or deterministic, loss scenarios against the exposures in all of these zones. Weapon damage functions are probabilistic estimates of damageability based on theoretical and empirical information. Damage probability decreases with the distance from the detonation, but is subject to complex interactions. Damageability varies with the size of the explosive, measured in equivalent pounds of TNT, and with the building construction type. Figure 5 illustrates mean damage ratios versus distance for combinations of weapon sizes and construction types. Figure 5: Weapon Damage Functions (Loss Distributions) M ea n Da m ag e Ra tio Mean Damage Ratio Distance (ft) (ft) CONFIDENTIAL 6

8 A.M. Best would like to see weapon choices that are appropriate to the target type. AIR has a range of conventional bombs and airplane crash damage functions as well as indications of their likelihood of use in different target environments. Figure 6 identifies the conventional weapons that can be chosen in an AIR analysis. Figure 6: Deterministic Terrorism Loss Report Detailed modeling is necessary to assess losses at each affected property and combine the losses based on the underlying probability distribution around the resulting mean damage ratios. Figure 7 provides an illustration of a deterministic loss analysis on a concentration of exposure in downtown San Francisco. Figure 7: Deterministic Terrorism Loss Report Property Address Loss Total for all 10 Locations in Policy 22,419, XXX Lincoln Way, San Francisco CA USA XXX Sacramento St., San Francisco CA USA 302, XXX Fulton St., San Francisco CA USA XXX Battery St., San Francisco CA USA 4,543, XXX Mason St., San Francisco CA USA XXX Fulton St., San Francisco CA USA XXX Pine St., San Francisco CA USA 11,633, XXX California St., San Francisco CA USA 4,259, XXX Sacramento St., San Francisco CA USA 1,680, XXX Irving St., San Francisco CA USA 0 By working with the AIR team, we can develop your deterministic losses and provide you a completed table for Question 44, including the requested Modeled Gross Losses and Pre-Tax Net Losses. Going forward, AIR s software applications enable you to conduct the analyses required to complete Question 44 internally. In addition, AIR can provide written support on how we developed the loss results, as required in part II of Question 44. CONFIDENTIAL 7

9 Net Losses The A.M. Best SRQ asks for pre-tax net loss information to reflect the impact of reinsurance and TRIA. Here, insurers should be considering the amount of terrorism coverage purchased for certified terrorism events (as defined by the TRIA) and the amount being provided for non-certified events in existing policies. Exposure information should reflect the appropriate coverages and exclusions. Probabilistic Modeling At this point in time, A.M. Best is not requiring submission of probabilistic loss analysis results. Probabilistic analysis involves estimating the frequency and severity of attacks. However, probabilistic analysis may be requested in the future, and may be submitted as supplementary material. Probabilistic loss analysis should be considered a component of prudent terrorism risk analysis. A probabilistic approach recognizes the impact of the differences between target locations and the relative attractiveness to the terrorists of different target types and targets in different regions. Insurance Services Office (ISO) has used probabilistic modeling results provided by AIR to develop advisory loss costs for cities and states across the United States. The result is a tiered set of cities. As shown in Figure 8, there are four Tier 1 cities, which have the highest advisory loss costs, and five Tier 2 cities, which have a lower advisory loss cost. The rest of the country is Tier 3, which has the lowest advisory loss costs. Based on these tiers, and territorial refinements within these cities, the ISO advisory loss costs have been filed and accepted for use in all U.S. jurisdictions. Indication in the SRQ of the ISO tier for each of the reported locations would add value to your responses. Figure 8: ISO Advisory Loss Cost Tiers CONFIDENTIAL 8

10 Beyond the requirements of the SRQ, a company may choose to run a fully probabilistic loss analysis on the full or partial exposure portfolio. The result would be full exceedance probability curves, similar to those provided for natural catastrophes, as shown in Figure 9. Figure 9: Probabilistic Terrorism Loss Report Aggregate Losses Property Address EV SD 20 yr 50 yr 100 yr 250 yr 500 yr 1000 yr Total 154,004 1,864, ,688 2,839,663 13,391,391 24,356,693 32,398, XXX Lincoln Way, San Francisco CA USA , XXX Sacramento St., San Francisco CA USA 10, , , ,212 1,268,017 3,391, XXX Fulton St., San Francisco CA USA , XXX Battery St., San Francisco CA USA 39, , ,991,168 1,991,968 5,359,587 16,000, XXX Mason St., San Francisco CA USA , XXX Fulton St., San Francisco CA USA 27, , , ,089 4,619,642 14,000, XXX Pine St., San Francisco CA USA 15, , , ,741 1,781,385 4,987, XXX California St., San Francisco CA USA 37, , ,866,720 1,866,720 5,024,613 15,000, XXX Sacramento St., San Francisco CA USA 21, , , ,498 3,292,450 5,499, XXX Irving St., San Francisco CA USA , CONCLUSION The AIR terrorism loss estimation model has been used by numerous insurers over the past two years to quantify and manage their terrorism exposure for property, workers compensation and other lines of business. Our software tools and consulting services are well suited to help your company quickly develop the best possible responses to the new terrorism section of the A.M. Best SRQ. Key components of our approach include: AIR s comprehensive database of potential at-risk locations Full set of loss distributions (damage functions) for a wide range of loss scenarios (weapons) AIR would be pleased to discuss in more detail how we can assist you with your responses to the SRQ, either on a service basis or through the use of our software applications. CONFIDENTIAL 9

11 APPENDIX A: AIR TERRORISM LOSS ESTIMATION MODEL With the passage of the Terrorism Risk Insurance Act of 2002, all commercial property and casualty insurers were required to cover losses due to international terrorist activity within the United States. For the first time, insurers were required to make difficult pricing decisions regarding terrorism risk. The AIR Terrorism Loss Estimation Model has been helping P&C insurers meet the requirements of the legislation and to determine their reinsurance needs. The fully probabilistic AIR model estimates property and workers compensation losses from possible future terrorist attacks to support pricing and underwriting decisions down to the individual policy level. The AIR model, launched in September 2002, was the first detailed terrorism loss estimation model to provide fully probabilistic loss costs. In fact, the AIR model is used to develop the advisory loss costs filed by ISO. The AIR Terrorism Loss Estimation Model provides the quantitative information that insurers, reinsurers, and corporate risk managers need in order to understand the risks of terrorism and to support decision-making. The AIR model produces estimates of the financial impact of potential acts of terror. The AIR model can be used to: Analyze concentrations of exposures and their proximity to likely targets Examine the effects of deterministic scenarios that affect specific exposures Perform fully probabilistic analyses for company-specific portfolios Support pricing, portfolio management, and overall risk management Analyze correlations of estimated losses across multiple lines of business on an event basis AIR s structured approach provides probabilistic loss calculations for: Property Workers compensation Life Accident & health Attack Location, Frequency and Severity Historical data on terrorist attacks is significantly limited due to the rarity of these events. Uncertainty surrounding the frequency and severity of future terrorist activity is therefore much higher than in the case of natural catastrophes. To develop estimates of the frequency, location and severity of potential future terrorist attacks, AIR employed the Delphi Method, assembling a highly credentialed team with national and international, high-level operational and analytical expertise in counter-terrorism. With input from the expert team, AIR identified the potential types of targets for possible attack. The resulting landmark database consists of over 300,000 potential targets that include commercial, industrial, educational, medical, religious, and governmental facilities. A subset of trophy targets carry a higher probability of major attack. CONFIDENTIAL 10

12 The model analyzes various threats posed by domestic extremists, formal international and state-sponsored terrorist organizations, and loosely affiliated extremist groups. The nature of the selected targets and of the weapons used is a function of the goals and capabilities of the individual groups. A wide variety of weapon types is considered. These include the full range of conventional weapons, including bombs of various sizes, as well as general and commercial aviation crash. Also modeled are the effects of nonconventional weapons, including chemical, biological, radiological, and nuclear (CBRN). Weapons Effects on Structures The AIR terrorism model takes a rigorous engineering-based approach to estimating building damage from weapons effects on both the target and surrounding buildings. These effects are multiple and include pressure and shock waves, fire, and both falling and projectile debris. To model the effects of nonconventional weapons on structures, the AIR terrorism model utilizes the Consequences Assessment Tool Set (CATS), a GIS-based software developed for the Defense Threat Reduction Agency (a unit of the Department of Defense) and the Federal Emergency Management Agency. CATS is capable of simulating various attack types, including chemical agents such as sarin and VX, and biological agents such as anthrax and small pox. Nuclear and radiological attacks using materials such as cesium and cobalt are also modeled. Probability Distributions for Loss and Injuries The AIR terrorism model generates the full range of potential losses to property, workers compensation, life, and accident and health, along with their likelihood of occurrence. The loss calculation takes into account policy coverage structures and, for workers compensation, benefit structures. The AIR terrorism loss estimation model provides the quantitative information that insurers need in order to understand the financial impact of potential acts of terror and to support rating agency filings. CONFIDENTIAL 11