Munich Reinsurance America, Inc. Basics of CAT Modeling Webinar. June 4, 2009

Transcription

1 Munich Reinsurance America, Inc. Basics of CAT Modeling Webinar June 4, 2009

2 Agenda Introduction Carl Hedde Tropical Cyclone Basics Mark Bove Basics of Catastrophe Modeling Mark Bove Hurricanes, Society, and Insurance Carl Hedde 1

3 Question and Answer Process You may submit a question at any time during the session. Find the Chat function in the right hand side of the screen. Send questions to Host.

4 Tropical Cyclone Basics

5 Tropical Cyclone Basics What is a Tropical Cyclone? A nonfrontal low-pressure system that develops over tropical or subtropical waters. Has organized convection (i.e., thunderstorm activity) and definite cyclonic (counterclockwise in northern hemisphere) surface wind circulation. Derives most of its energy from the release of latent heat. Naming conventions Tropical Depression Sustained winds below 39 mph and closed a surface circulation. (Storm is given a number) Tropical Storm Hurricane, Typhoon, Cyclone Sustained winds between mph. (Storm is given a name) Sustained winds above 73 mph. 4

6 Tropical Cyclone Basics Saffir-Simpson Hurricane Wind Scale Category Winds (mph, 1 minute average) Description Example U.S. Storms Damaging winds expected Lili (2002) Widespread damage expected Gustav (2008) Extensive damage expected Ivan (2004) Devastating damage expected Charley (2004) 5 >155 Catastrophic damage expected Andrew (1992) New scale is experimental in 2009, central pressure and surge estimates have been removed due to wide variability in pressure and surge associated with a given wind speed. Due to uncertainty in wind measurements, tropical cyclone winds reported operationally in 5 mph increments. Maximum sustained wind given for a storm usually only covers a very small area. The force of wind on structures increases as the cube of the wind speed, so damage potential does not increase linearly with wind speeds. Category 3, 4, and 5 hurricanes are considered major hurricanes, which account for about 20% of all U.S. landfalls, ll but about 80% of all insured losses due to tropical cyclones. 5

7 Tropical Cyclone Basics Tropical Cyclone Structure: Nature s Heat Engine Outflow ( Exhaust ) Energy Release ( Cylinders ) Ocean ( Fuel ) Source: NOAA 6

8 Tropical Cyclone Basics Tropical Cyclone Development Requirements Water temperatures of ~80 F to a depth of at least 150 feet. Be at least 300 miles away from the equator. Light upper atmospheric winds. Moist, unstable atmosphere at low and mid-level altitudes. A weak pre-existing weather disturbance. 7

9 Tropical Cyclone Basics ºF Wind shear common 86º 77º 68º 59º 50º Hurricane development not possible near Equator 41º Data Source: U.S. Naval Research Labs, Stennis, MS. 32º

10 Tropical Cyclone Basics Atlantic Hurricane Tracks, Map: NOAA Coastal Services Center 9

11 Tropical Cyclone Basics Causes of Tropical Cyclone Lysis Moving over colder water. Moving over land / interaction with nearby land. Wind shear. Dry air intrusion / Saharan dust layers. Interaction with or absorption into an extratropical cyclone. Extratropical transition 10

12 Tropical Cyclone Basics Extratropical Transition Process of tropical cyclone becoming an extratropical (frontal) cyclone. Usually induced when hurricanes interact with colder, drier air, jet streams, or extratropical weather systems, such as a cold front. Floyd as Tropical Low Floyd as Transitioning Low Source: NOAA 11

13 Tropical Cyclone Basics Primary Hurricane Hazards: Wind Damage due to force of wind (direct) Damage due to impact of wind-driven driven objects (indirect) Charley Wilma Photos: Munich Reinsurance America 12

14 Tropical Cyclone Basics Primary Hurricane Hazards: Storm Surge Onshore rush of sea or lake water caused by the high winds associated with a landfalling cyclone and secondarily by the low pressure of the storm. Typically accompanied with extensive wave damage and erosion along immediate coastline. Katrina Photo: Munich Reinsurance America 13

15 Basics of Catastrophe Modeling

16 Basics of Catastrophe Modeling Catastrophe models have many different functions Develop catastrophe pricing loads by postal code or rating territory. Track exposure levels in key catastrophe regions. Calculate the effects of deductibles and other risk mitigation techniques. Quantify potential of catastrophic loss by peril. Evaluate the impact and cost of reinsurance protection. Estimate potential losses of real world events. Satisfy regulatory requirements. 15

17 Basics of Catastrophe Modeling Catastrophe models utilize the following risk-specific information Location specific information Address Building Construction and Occupancy Number of stories Financial information Policy Deductible Policy and Coverage Limits Value of Location Year of Construction 16

18 Basics of Catastrophe Modeling There are four basic components to all catastrophe models, regardless of the peril being modeled. Stochastic ti Event Set Intensity Calculation Damage Estimation Loss Evaluation 17

19 Basics of Catastrophe Modeling Stochastic Event Set A set of discrete events covering all the important combinations of location, size, and other pertinent characteristics of the peril in question, as well as the (annual) probability of each event. Account for all locations that are impacted by peril of interest. A Stochastic Event Set should: Have a sufficient number of events to cover all possible sizes and intensities that are possible in a given location. Consider the probability of a given event occurring. Have enough events to produce statistically stable loss results. 18

20 Basics of Catastrophe Modeling Intensity Calculation A set of rules to calculate the intensity of each event at every site of interest (individual location, insurance portfolio, etc.). Be based on established physics of the hazard being modeled. Intensity Calculations should: Should be calculated at every location of interest. Should consider environmental influences at or around site of interest. To calculate event intensities at each location of interest, the model requires Geocoding and the relevant Geospatial Hazard Databases for the peril being modeled. Both of these tools are included in the model. 19

21 Basics of Catastrophe Modeling Intensity Calculation Geocoding Computer program that transforms location information contained in exposure data (building number, street, city, state, and postal code) into coordinates (usually latitude and longitude) that the risk modeling software can process. The most desirable resolution depends on the peril, usually the more detailed the better. 555 College Road East Princeton, NJ = ºN, ºW Map: Munich Reinsurance America 20

22 Basics of Catastrophe Modeling Intensity Calculation Geospatial Hazard Databases Databases containing information on the local environmental and/or physical factors that can influence an event s intensity at the site. The relevant information depends on the peril; with respect to tropical cyclones: topography and surface roughness. Source: USGS 21

23 Basics of Catastrophe Modeling Damage Estimation Computational models to estimate the damage level of each structure, given its vulnerability to the particular peril and the event s intensity at the site. Damage Estimation should: Consider several types of construction and building occupancy ( primary modifiers ). Consider other construction techniques that either increase or decrease vulnerability to the peril in question (i.e. secondary modifiers ). Consider the initial quality of construction, including the adoption and enforcement of building codes, as well as the age and the maintenance of the building. Have different damage/vulnerability functions for buildings (structural and nonstructural elements), contents, and time element. Use a consistent measure to correlate event intensity with building damage. Account for so-called secondary hazards, such as storm surge and demand surge. 22

24 Basics of Catastrophe Modeling Loss Evaluation Evaluate insured losses, given the damage level & values, as well as the applicable insurance and reinsurance structures, such as deductibles, limits, attachment points, treaty cessions, etc. Determine how building damage translates into insurance loss. Determine whether or how to account for: Loss Evaluation should: Demand surge Price gouging Inflated claims Aggressive construction contractors Questionable claims settlement 23

25 Basics of Catastrophe Modeling Part of model Stochastic Event Set Geocoding Engine Intensity Calculation Geospatial Hazard Database Damage Estimation Loss Evaluation 24

26 Basics of Catastrophe Modeling Provided by user Part of model Stochastic Event Set Exposure Data Geocoding Engine Intensity Calculation Geospatial Hazard Database Insurance Structure Damage Estimation Loss Evaluation 25

27 Basics of Catastrophe Modeling 1. Event Definition 2. Intensity Calculation Location Size Type (Probability) tensity at Site Event In int dist Source-to-Site Distance 4. Loss Evaluation 3. Damage Estimation Probability Dens sity Deductible Ground-Up Loss = DF x Repl. Value Limit Repl. Value Damage Facto or mean DF int The standard deviation is usually estimated as a function of the mean Ground-Up Loss Event Intensity at Site 26

28 Basics of Catastrophe Modeling Creating Event Loss Distributions Involves evaluating Combine coverages (building, contents, t the sum of correlated time element) to arrive at a site random variables and (account) loss distribution for each applying the appropriate event. deductibles and limits. Combine sites to yield a policy loss distribution for each event. Combine policies i to produce a portfolio loss distribution for each event. Creating Annual Loss Distributions Combine all event loss distributions to arrive at an annualized probability distribution. By evaluating the probability that the loss from any one (two, three etc.) event(s) can exceed any given threshold. h 27

29 Basics of Catastrophe Modeling Losses can be produced for a variety of financial perspectives. The output of the catastrophe model contains the losses by event by financial perspective. Ground Up Loss (Loss before policy limits/deductibles) Gross Loss (Loss after application of limits and deductibles) Financial perspectives Treaty Loss (Loss to individual treaties) Net Loss (Loss after limits, deductibles, and treaties) Reinsurance Loss (Loss suffered by the reinsurer) 28

30 Basics of Catastrophe Modeling Exceedance Probability Curve eedance ity of Exce Annual Probabili 1/N Loss ($) N-yr Loss 29

31 Basics of Catastrophe Modeling Example of output from a catastrophe model Event ID Perspective Loss Probability 1 Gross 5,000, % 2 Gross 20,000, % 3 Gross 10,000, % 0.1% 30

32 Basics of Catastrophe Modeling When assessing catastrophe risk, the two main measures of loss are: Return Period Losses Return period losses are a measure of ultimate risk, and are often used in risk management and accumulation control decisions. Average Annual Loss Average Annual Loss is a measure of expected loss, and provides the starting point for pricing the catastrophe component of treaties. 31

33 Basics of Catastrophe Modeling Return Period Losses Potential catastrophe losses are often expressed in terms of an exceedence probability. bilit The return period is the inverse of the exceedence probability (ex: 1% EP = 1 in 100 year loss). This is often referred to as a PML (Probable Maximum Loss) This represents the loss level that we would expect to surpass, on average, once every n years. The larger the return period, the more conservative the loss estimate. This method of loss estimation does not attempt to quantify the potential loss estimate for the n th year. Return Period Losses are depicted graphically with the EP curve. 32

34 Basics of Catastrophe Modeling Construction of EP Curve Step 1: Sort the events in descending order of loss Event ID Perspective Loss Probability 2 Gross 20,000, % 1 Gross 10,000,000, 0.1% 3 Gross 5,000, % 33

35 Basics of Catastrophe Modeling Construction of EP Curve Step 2: Aggregate the probabilities Event ID Perspective Loss Probability Accumulated Probability 2 Gross 20,000, % 0.3% 1 Gross 10,000, % 0.4% 3 Gross 5,000, % 1.0% 34

36 Basics of Catastrophe Modeling Construction of EP Curve Step 3: Graph the losses and accumulated probabilities for each event 35

37 Basics of Catastrophe Modeling Construction of EP Curve EP Curves and Return Periods The Return Period is 1 / Accumulated Probability ID Persp. Loss Probability Accumulated Probability Return Period 2 Gross 20,000, % 0.3% Gross 10,000, % 0.4% Gross 5,000,000, 0.6% 1.0% 100 In this example, the 1/250 year is $10M. 249 years out of 250 losses are expected to be less than 10M. 36

38 Basics of Catastrophe Modeling Average Annual Loss (AAL) Also referred to as expected loss or pure premium. Calculated as sum (Loss * Event Probability) for all events. Event ID Perspective Loss Probability 2 Gross 20,000, % 1 Gross 10,000, % 3 Gross 5,000,000, 0.6% AAL = (20M *.3%) + (10M *.1%) + (5M *.6%) = 150K 37

39 Basics of Catastrophe Modeling Average Annual Loss (AAL) AAL of 150K represents the amount of loss expected in a given year. This measure is more useful for pricing than risk management. For risk management purposes, the 150K average has little significance. This company has 0 losses 99 years out of 100 and the 100 th year suffers losses of at least 5M. Graphically, the AAL represents the area under the EP curve. 38

40 Hurricanes, Society, & Insurance Photo: Munich Reinsurance America

41 Question and Answer Process You may submit a question at any time during the session. Find the Chat function in the right hand side of the screen. Send questions to Host.

42 Hurricanes, Society, & Insurance Coastal Exposures: A Rapidly Growing Risk Historical Losses due to natural catastrophes in the United States are increasing i dramatically. Insured losses are about nine times greater during the past ten years than in the 1980s. Most of the increase of losses is due to coastal exposures vulnerable to tropical cyclones. Population The U.S. coastal population grew by about 40 million from (U.S. Census Bureau). Values Property values along the coast are increasing. Building Units The number of building units are increasing. Climate Change Potential of increased storm severity and/or frequency. 41

43 Hurricanes, Society, & Insurance Economic and Insured Property Losses Due to Significant United States Natural Catastrophes (All values 2008 Dollars) Significant is defined as at least $1 billion in economic losses or 50 fatalities are directly attributed to the event. Source: Munich Re Topics Geo, 2008 U.S. Edition 42

44 Hurricanes, Society, & Insurance Insured U.S. Tropical Cyclone Losses The current 5-year average ( ) insured tropical cyclone loss is $28.3 bn Munich Re Group Source: Property Claims Service, MR NatCatSERVICE, NFIP 43

45 Hurricanes, Society, & Insurance Top 10 U.S. Property Natural Catastrophe Losses, Original Insured Dollar Loss $45.0 $40.0 $40.5 $35.0 $ $25.0 $20.0 $15.0 $10.0 $5.0 $4.2 $4.6 $5.3 $7.1 $7.5 $10.3 $11.5 $12.5 $15.5 $0.0 Hurricane Hugo Hurricane Hurricane Rita Hurricane Ivan Hurricane Hurricane Hurricane Ike Northridge (1989) Frances (2004) (2005) (2004) Charley (2004) Wilma (2005) (2008) Earthquake (1994) Hurricane Hurricane Andrew (1992) Katrina (2005) Data Source: PCS. Losses in original dollar amounts. Does not include offshore losses, aviation losses or losses to the NFIP. 44

46 Hurricanes, Society, & Insurance Estimated Top 10 U.S. Natural Catastrophe Losses , Insured Losses As If Occurred Today $90.0 $80.0 $78.0 $80.0 $70.0 $ $50.0 $40.0 $30.0 $27.0 $28.0 $31.0 $36.0 $41.0 $42.0 $49.0 $53.0 $20.0 $10.0 $0.0 Galveston Southwest Hurricane (1915) Florida (1944) Great New Ft. Lauderdale Hurricane Hurricane Great Great Galveston San Francisco Great Miami England Hurricane (1947) Katrina (2005) Andrew (1992) Okeechobee Hurricane (1900) Earthquake Hurricane (1926) Hurricane (1938) Hurricane (1928) (1906) Data Sources: RMS, PCS. Modeled Losses to property, contents, and BI for personal, commercial, mobile home and auto lines. Includes Storm Surge, Demand Surge, and large loss amplification. 45

47 Hurricanes, Society, & Insurance U.S. Coastal Population Change by County, % of the population in Atlantic coastal states lives in coastal counties. This is equal to 19% of the U.S. population. Image: NOAA 46

48 Hurricanes, Society, & Insurance Percentage of State Population in Atlantic Coast Counties Total Coastal Population: 56.4 million Map: Munich Re America. Data: U.S. Census Bureau 47

49 Hurricanes, Society, & Insurance Breakdown of U.S. Coastal Population by Region 47% 17% 14% 22% TX AL: 7.9 million FL: 12.3 million GA DE: 9.4 million NJ ME: 26.9 million Map: Munich Re America. Data: U.S. Census Bureau 48

50 Hurricanes, Society, & Insurance Total Value of Coastal Exposures (2004 Dollars, Billions) Source: Insurance Information Institute 49

51 Hurricanes, Society, & Insurance Impact of Climate Change Increased sea surface temperature may lead to increased hurricane severity and frequency. 50

52 Hurricanes, Society, & Insurance Economic Reasons for Excessive Coastal Property Exposures Property Owners Low cost of living; low real estate prices & rapid appreciation; no state income tax in some locales; low property tax; rapid job growth. Government-run insurers (e.g., CPIC, NFIP) provide implicit subsidies by selling insurance at below-market prices with few underwriting restrictions. s Government aid; tax deductions; litigation recovery for uninsured losses. No fear of death and injury. Local Zoning & Permitting Authorities Residential construction creates jobs; attracts wealth; increases tax receipts, stimulates commercial construction & permanent jobs; develops infrastructure. Increases local representation in state legislature & political influence. Property and infrastructure damage costs shifted to others (state and federal taxpayers, policyholders in unaffected areas, and nonproperty policyholders. Developers Coastal development is a high-margin business. Financial interest and risk reduced to zero after sale. 51

53 Hurricanes, Society, & Insurance Example: Impact of Coastal Property on Insured Losses Residental Home in Florida Concrete Block Exterior Walls Wood Frame Gable Roof Wind-Resistant Shingles Insured Values $500,000 Building $250,000 Contents $150,000 Loss of Use Photo: Munich Re America 52

54 Hurricanes, Society, & Insurance Example: Impact of Coastal Property on Insured Losses Modified Home Parameters Location On Coast 10 Miles Inland Roof Anchors Toe Nailing Metal Anchors Construction Date Miamii 53

55 Hurricanes, Society, & Insurance Example: Impact of Coastal Property on Insured Losses Expected Average Annual Loss Due to Wind Coast Inland Built in 2005, Metal Roof Anchors $3, $699 Built in 2005, Toe Nailing $4,613 $1,078 Built in 1980, Metal Roof Anchors $11,306 $3,862 Built in 1980, Toe Nailing $17,028 $6,259 54

56 Hurricanes, Society, & Insurance Example: Impact of Coastal Property on Insured Losses Loss Due to Wind, 1% Annual Probability of Exceedance (100-Year Return Period) Coast Inland Built in 2005, Metal Roof Anchors $67,180 $14,364 Built in 2005, Toe Nailing $102,095 $21,081 Built in 1980, Metal Roof Anchors $276,411 $104,760 Built in 1980, Toe Nailing $465, $180,

57 Hurricanes, Society, & Insurance Example: Controlling and Managing Risk How do expected losses change as one moves in from the coast? Eleven locations at the same latitude at 1-mile increments from coast to 10 miles inland. Construction types Toe Nailing Metal Anchor Toe Nailing Metal Anchor Miamii 56

58 Hurricanes, Society, & Insurance Change in Expected Annual Loss by Construction: Coast to 10 miles Inland, Wind-Only s ($) nd-up Loss Groun Distance from Coast (Miles) 57

59 Hurricanes, Society, & Insurance Successful Tools for Controlling and Managing g Coastal Risk Improved land use planning Building code improvements & stringent t code enforcement. Improved construction techniques. Incentivize mitigation measures. Actuarially sound, risk-based insurance rates. Remove impediments that impact capital flows dedicated to risk financing. Free market mechanics. 58

60 Hurricanes, Society, & Insurance Unsuccessful Tools for Controlling And Managing g Coastal Risk Actuarially unsound insurance rates Government imposed restraints negatively impacts the availability of insurance. Retroactive interpretation/re-writing of insurance contracts. Weak building codes and poor code enforcement. Poor land use planning. 59

61 Hurricanes, Society, & Insurance Building Wind Vulnerabilities: Poor Performers Large, unprotected openings Flat roofs or gable roofs Large overhangs Irregular shapes (e.g., dormers) Aged wood/structural t timber High rise buildings 60

62 Hurricanes, Society, & Insurance Large, Unprotected Openings Wilma Photo: Munich Re America 61

63 Hurricanes, Society, & Insurance Gable Roofs & Walls Charley Photos: Munich Re America 62

64 Hurricanes, Society, & Insurance Flat Roofs Ivan Photo: Munich Re America 63

65 Hurricanes, Society, & Insurance Strip Malls Large, unprotected openings + flat roofs + large overhangs Charley Photos: Munich Re America 64

66 Hurricanes, Society, & Insurance Roof Irregularities Katrina Photo: Munich Re America 65

67 Hurricanes, Society, & Insurance Building Wind Vulnerabilities: Good Performers Newer, stricter building codes appeared to be effective. Monolithic concrete boxes perform very well. Mitigation measures, such as hurricane shutters and seawalls work! 66

68 Hurricanes, Society, & Insurance An Undamaged New House Charley Photo: Munich Re America 67

69 Hurricanes, Society, & Insurance Concrete Boxes Perform Well Jeanne Photo: Munich Re America 68

70 Hurricanes, Society, & Insurance Seawalls Protect Against Surge and Erosion Jeanne Photo: Munich Re America 69

71 Hurricanes, Society, & Insurance Institute for Business and Home Safety (IBHS) Fortified Homes, Bolivar Peninsula, Texas Ike Photo: Munich Re America

72 Hurricanes, Society, & Insurance 2009 Atlantic Hurricane Season Forecasts Named Storms Hurricanes Major Hurricanes NOAA Colorado State University City College of London Climatology

73 Question and Answer Process You may submit a question at any time during the session. Find the Chat function in the right hand side of the screen. Send questions to Host.

74 Thank you for your time and interest! Copyright 2009 Munich Reinsurance America, Inc. All rights reserved. The Munich Re America name is a mark owned by Munich Reinsurance America, Inc. The material in this presentation is provided for your information only, and is not permitted to be further distributed without the express written permission of Munich Reinsurance America. This material is not intended to be legal, underwriting, financial, or any other type of professional advice. Examples given are for illustrative purposes only. Each reader should consult an attorney and other appropriate advisors to determine the applicability of any particular contract language to the reader's specific circumstances.