Créer une présentation
Télécharger la présentation

Download

Download Presentation

Introduction to Property Exposure Rating Casualty Actuarial Society Reinsurance Pricing Seminar

364 Vues
Download Presentation

Télécharger la présentation
## Introduction to Property Exposure Rating Casualty Actuarial Society Reinsurance Pricing Seminar

- - - - - - - - - - - - - - - - - - - - - - - - - - - E N D - - - - - - - - - - - - - - - - - - - - - - - - - - -

**August 10, 2009**Introduction to Property Exposure RatingCasualty Actuarial SocietyReinsurance Pricing Seminar Kevin Hilferty, Morristown**PROPERTY Exposure Rating**• Commercial Property • Residential Property • Ocean Marine • Inland Marine**Property Rating – In case I use any of these terms…**A bit of vocabulary Basically the value of the building or the policy limit, whichever is smaller TIV: Total Insured Value TSI: Total Sums Insured PML: Probable Maximum Loss MFL: Maximum Forseeable Loss The largest loss that seems reasonable to expect (this is almost always less than TIV/TSI) Shades of meaning, or a real difference?**Exposure Rating Overview**• We always start with the subject premium • The loss ratio determines the expected ground–up loss • Exposure Rating simply tells us how much of the expected loss will fall into a given layer • Once we have expected loss to the layer, we can break it up into its component frequency and severity • The mechanics of how we do this is different depending on the curve we use**Reinsurance Exposure Rating**• Allocation of Premium/Loss to Layer through use of some generated curve/equation (model of loss) • Based on Industry • Based on Company Data • FLS Based on ????**Why Do We Exposure Rate?**• Exposure Rating can be used to: • Estimate Mean (Expected) Loss (for any layer or limit) • Estimate Reinsurance Price So Can Experience Rating for that Matter!**WHEN Do We Exposure Rate?**When company experience: • Is approximately like Industry • Or another company • Is insufficient • Low volume • New LOB • Is non-credible • Mix changes • Changing profiles**When DON’T We Exposure Rate?**When company: • Experience is not like industry • Info is not available • Company doesn’t provide necessary info • No industry data is available**Exposure Rating by LOB**Although the ideas behind exposure rating never change, the actual mechanics of it differ by LOB • LIABILITY (GL) uses Increased Limits Factors (ILFs) • PROPERTY uses: • First Loss Scales (FLSs), or • Size-of-Loss Curves (PSOLD) • WORKER’S COMP uses Excess Loss Factors (ELFs)**Exposure Rating - Ingedients**• Subject Premium & Loss Ratio • Limit Profile • Curve**Exposure Rating - Issues**• Wind vs Fire vs CAT Loss Ratios??? • Sometimes on a combined basis, sometimes calculated separately • Best to have Cat vs Non-Cat • Why? • Exposure rate is always a non-cat rate • We let the Cat Models (AIR/RMS/EQE) calculate the cat portion of expected loss • Ultimately, you want a loss ratio that excludes modeled causes of loss. • If you only modeled hurricane and quake, you don’t want a loss ratio that excludes winter storm**Property Rating – PureWhy should Reinsurance be priced**differently? Why do we need curves? Building Value = $1M Rate = 20 ¢ per $100 in Value Using a single rate for the entire exposure leaves us in a bit of a bind.... How much went for 500K x 500K ?????? Since reinsurer is responsible for 50% of limit, should he/she/it get 50% of the premium?**Property Rating - Pure**So what are we supposed to do ??? Why can’t the property people use ILFs too?**Property Rating – Bit o’ History**• In the old days, it was believed that: • Virtually all losses were fire losses • Virtually all fire losses were total losses • If so, a single rate makes sense**Property Rating – Bit o’ History**• These days, it is believed that: • For Homeowners • There are lots of total fire losses • But there are a lot of partial losses too • For Commercial Property • There are lots of ways to have losses • Hardly any losses are total • In response, rating methods are different**Some people think E&O behaves more like property**Property Rating – Liab vs Prop • For Liability we think in terms of dollars • e.g. a slip & fall costs $2000 • For Property we think in terms of % of TIV • e.g. a HO claim is for 10% of the TIV • For Liability, loss is independent of limit For Property, loss is dependent on TIV**Property Rating – First Loss Scales**Traditionally, Property has used something called a First-Loss Scale aka Lloyds Scales aka Salzmann Curves aka Ludwig Curves First-Loss Scales give the distribution of loss as a percent of insured value (as opposed to the distribution of loss dollars) This means for property we basically only do allocation of premium based on losses**Property Curves**Where do they come from? • Lloyd’s Scales ???????????????????? • Salzmann Curves “Rating by Layer of Insurance” – Ruth Salzmann, 1963 HO Fire losses only 1960 Accident Year Data from INA • Ludwig Curves “An Exposure Rating Approach to Pricing Property Excess-of-Loss Reinsurance” – Stephen Ludwig, 1991 Hartford HO AY Data, 1984-1988 Hartford Commercial Property database Fire, Wind, Other Retail/Wholesale, Service/Office, Apartment/Condo, Restaurant**Wrinkles to Using First Loss Scales**• Appropriate First Loss Scale • Over 50 First Loss scales • Some are more popular with reinsurers • Different scales are used differently**Wrinkles to Using First Loss Scales**• TIV vs PML vs Other • Salzmann Curves - Bldg losses for Bldg TIV • Ludwig Curves - All losses but Bldg TIV • Some curves apply to PMLs • No consistent definition of PML**Property Rating – First Loss Scales**Interpretation: A layer from 0-10% of TIV should see 25% of the total losses A layer from0-50% of TIV should see 70% of the total losses**Property Rating – First Loss Scales**TIV = $100,000 25% of losses are less than or equal to 10% of TIV. Therefore, 25% of Premium goes to pay the losses for the first 10,000 of building value. (since 10% * 100,000 = 10,000) 60% of the premium goes to pay the losses for the first 40,000 of building value (since 40% * 100,000 = 40,000)**Property Rating – First Loss Scales**TIV = $100,000 10% (= 50% - 40%) of losses are expected to fall in the layer between $20,000 (20% of TIV) and $30,000 (30% of TIV). This also means that if you have a loss, there is a 30% chance more than 50% of the building will be lost. If a there’s a 30% chance that half of a $1M building can get wiped out, does this also mean that there’s a 30% chance that half of a $100M building will be lost?**100K**50K 50K 100K 40K 10K 10K First Loss Scales – Example What premium is needed for a 40K x 10K fac cert? 50% 10% TIV = 100K Prem = 1,000 Loss Ratio = 60% Reins. Expenses = 20% Step 1: We need to know what the retention and the top of the layer are as a % of TIV**Step 2: Calculate Expected Loss**1000 * 60% = 600 Step 3: Look up Ratios on Table 10% 25% of loss 50% 70% of loss Step 4: Multiply E(Loss) by Ratio Difference E(Loss)40x10 = (70% - 25% ) * 600 = 270 Step 5: Gross Up for Reins. Expenses Reins. Prem40x10 = 270/(1 - 0.3) = 386 So insuring 40% of limit for 38.6% of premium First Loss Scales – Example What premium is needed for a 40K x 10K treaty? Prem = 1,000 Loss Ratio = 60% Reins. Expenses = 20%**First Loss Scales – Another ExampleMultiple Locations**What premium is needed for a 500K x 200K treaty? 200K to 700K Put LOTS of these together and you get … a Limits Profile! Loss Ratio = 60% Reins. Expenses = 20%**1M**First Loss Scales – ExamplePolicy with SIR What premium is needed for a 500K x 500K treaty? 1250K 100% of TIV 100% of Loss 500K 750K 60% of TIV (750/1250) 75% of Loss 500K 100% - 75% = 25% 25 % of Total Loss Expected in the Layer 250K Policy Limit = 1M SIR = 250K TIV = 1.25M Prem = 10,000 Loss Ratio = 55% Reins. Expenses = 20% 250K**First Loss Scales – ExamplePolicy with SIR**BUT WHAT IS THE TOTAL LOSS?**1M**First Loss Scales – ExamplePolicy with SIR 1250K 100% of TIV Policy Limit = 1M SIR = 250K TIV = 1.25M Prem = 10,000 Loss Ratio = 55% Reins. Expenses = 20% 100% of Loss 500K E(Loss) = Premium * Loss Ratio = 10,000 * 0.55 = 5,500 BUT THIS IS ONLY FOR LOSSES ABOVE 250,000! If 40% of losses are below 250,000, then 5,500 = Total Loss * (1-40%) 5,500/(1-40%) = 9,167 750K 60% of TIV 75% of Loss 500K 250K 20% of TIV 250K 40% of Loss**Calculate Expected Loss in the Layer**9,167 * 25% = 2,292 Gross-up for Reinsurer Expenses 2292 / (1 – 0.2) = 2,865 So insuring 40% of limit for 28.7% of premium First Loss Scales – ExamplePolicy with SIR What premium is needed for a 500K x 500K treaty? Policy Limit = 1M SIR = 250K TIV = 1.25M Prem = 10,000 Loss Ratio = 55% Reins. Expenses = 20%**PSOLD Curves**• 1998 – PSOLD Curves Released, updated every 2 years thereafter • Created to fix assumption of constant loss-to-value ratios across all value ranges • Calculates average severity of loss given policy limit rather than % of value • Separate curves for each of: • 60 value ranges • 22 commercial occupancy classes • Homeowners (new) • Building Only Contents Only Buildings + Contents B + C + BI**PSOLD Curves – Example Calculations**Subject Premium = $75M Loss Ratio = 60% Reinsurer Expenses = 15% What premium is needed for a $3M xs $2M treaty? Expected Loss = $75M x 0.60 = $45M Portion of loss in layer = (15,134 – 14,101) / 16,329 = 0.06326 ($45M x 0.06326) / (1 – 0.15) = $3,349,148**PSOLD Curves**• Buildings and Contents – not an issue • B + C + BI - Watch your Limit Profiles! B + C + BI LOSS B + C LOSS LOSS B + C Policy Limit**PSOLD Curves**• DO NOT INCLUDE BI IN LIMITS PROFILES WHEN RATING WITH PSOLD (Most US Markets) • Overstates Severity of Loss • First-Loss Scales rely on Total Limits Profile (incl. BI)**Per-Location**Bldg vs Cnt vs BI Limit Deductible Premium TIV Participation Account ID Location ID Policy ID Occupancy Property Exposure RatingRequired Data For Premium Allocation to Location, we need premium by account along with all this other stuff… Stacking ALL THIS BY TYPE OF BUSINESS**Per-Location**Bldg vs Cnt vs BI Limit Deductible Premium TIV Participation Account ID Location ID Policy ID Occupancy Property Exposure RatingRequired Data By-Band Limit Range (excl. BI) Average SIR Premium Min & Max TIV (or average) Average Participation Occupancy Distribution Stacking Other Data Used Company Specific First-Loss Scales Perils Covered Protection, Construction (HO) ALL THIS BY TYPE OF BUSINESS**I wish this were a trick question, but this is the kind of**data we often get Limit Profiles with no Premium What premium is needed for a 300K x 200K treaty? Loss Ratio = 60% Reins. Expenses = 20% Total Premium = $500,000 Layer**Limit Profiles with no Premium**What’s wrong? • TIV? • Deductible/Sir? c) Need prem, not # of risks – fatal? Layer Used to be, but we have ways around that now. Still, it’s better to have the premium.**Policy Level Data**• What do you do when your data looks like this? • Need LOCATION LEVEL data • Does every location have the same value and represent the same amount of risk?**Allocation of Premium to Individual Location**• When policies cover multiple locations, it is necessary to allocate the premium to each individual location before exposure rating techniques can be properly applied. • Traditional Methods • By TIV • All Premium Slotted to Highest Limit • By Exposed TIV • Traditional Methods are Wrong • Why?**Allocation of Premium to Individual Location**BY TIV??? Should this location be assigned any premium?**Allocation of Premium to Individual Location**ALL PREMIUM SLOTTED TO HIGHEST LIMIT??? Would assume all locations expose the policy to the same amount of risk! There may be many partial exposures like this one.**Allocation of Premium to Individual Location**BY Exposed TIV??? This location will get less premium Should these three get equal premium? This location won’t get any premium**Allocation of Premium to Individual Location**BY Exposed TIV??? Do they subject the policy to equal risk? Parking Lot Dynamite Factory**Allocate Based on Potential for Loss**SOLUTION Average Severity of loss can be based on First Loss Scales (Lloyds Scales) or PSOLD curves. Parking Lot Average Severity = $1,000 Strip Mall Average Severity = $2,000 Dynamite Factory Average Severity = $5,000 Restaurant Average Severity = $2,000 $50,000 $100,000 $250,000 $100,000**Wrinkles to Using First Loss Scales**• Need the Correct Information • Premium, not number of risks • TIV or PML • Conversion to Pure Loss • Not always clear what to do**SUMMARY**EXPOSURE RATING • Loss Ratios • Exposure Curves • Effect of SIR’s • BI in PSOLD profiles • Premium Allocation