Property Per-Risk Pricing Current Challenges

1
Property Per-Risk PricingCurrent Challenges

• David R. Clark
• American Re-Insurance Company
• CAS Seminar on Reinsurance June, 2003

2
Property Per-Risk Basics

• Experience Rating (burn cost)
• Exposure Rating
• Layer the overall premium
• Requires Insured Value profile and severity
  curve(s)
• Price other features
• Annual Aggregate Deductible
• Limited Reinstatements

3
Property Per-Risk Basics

• So were done, right?

4
Property Per-Risk A Preliminary Problem

• What is a risk?
• Typical Treaty wording
• The ceding Company shall be the sole judge of
  what constitutes one risk
• (subject to conditions)
• Risk can be decided after loss occurs!

5
Property Per-Risk A Preliminary Problem

• Roughly, risk location
• Do we have information on a per-location basis?
• Are our pricing tools based on per-location data?

6
Property Per-Risk Problems Solutions

• Problems
• No data on Blanket policies
• Little detail in data for other policies
• Prevalence of outdated curves
• Poor price monitors
• Disconnect over PML

7
Property Per-Risk Problem 1

• Blanket Policies often not captured on a
  per-location basis
• The majority of large risks are either
  blanket-rated or specifically-rated
• Blanket policies are in neither our pricing
  models or our TIV profiles

8
Property Per-Risk Problem 1

• Solution
• Need a data standard that includes blanket
  policies
• Per-location detail as included in Declarations
  Page
• Use ISO or Catastrophe Models as platform?

9
Property Per-Risk Problem 2

• Lack of Detail in Insured Value Profiles
• Do not distinguish Building vs Contents
• Do not include Time Element Coverages
• Do not list deductibles
• Do not detail level of coverage
• All Perils vs Named Perils
• Replacement Cost and Insurance-to-Value
• Ordinance or Law provision for Time Element

10
Property Per-Risk Problem 2

• Solution
• Need data standard that includes more information
• Need pricing models to run on detailed file, not
  on summarized TIV profile

11
Property Per-Risk Problem 3

• What severity curve is used?
• Current Data
• ISO PSOLD
• Company-specific, proprietary curves
• Outdated Data
• Lloyds Scales (source date unknown)
• Ludwig (Hartford Ins Grp 1984-1988)
• Salzmann (INA Homeowners, 1960)

12
Property Per-Risk Problem 3

• Why Size Matters

Derived from proprietary American Re-Insurance
study based upon customized ISO data.
13
Property Per-Risk Problem 3

• Solution
• Replace outdated models
• But show impact of new model!
• Incorporate other data sources
• National Fire Protection Association (NFPA)
• Size matters

14
Property Per-Risk Problem 4

• Lack of Consistency in Ceding Company Price
  Monitors
• Critical to experience and exposure rating
• Wide flexibility in charged premium due to
  discretionary pricing factors
• Minimal info on Specifically-Rated risks

15
Property Per-Risk Problem 4

• Principle
• Rate and Price changes are explanatory variables
  for movement in loss cost.
• Consequence We need to test how well they
  explain that movement.

16
Property Per-Risk Problem 4

• Solution
• This is tough and requires discipline
• Double check
• First Principles OnLevel based on rate and
  price changes
• Historical comparison of average premium
• E.G., ISO MarketWatch

17
Property Per-Risk Problem 5

• Difficulty in including Underwriters expertise
• What is a PML?

18
Property Per-Risk Problem 5

• The term PML or probable maximum loss is one
  of the most widely used terms in property
  insurance underwriting. But it represents one of
  the least clear concepts in all insurance.
• John McGuinness - Is Probable Maximum Loss
  (PML) a Useful Concept? PCAS 1969

19
Property Per-Risk Problem 5

• PML is still an ambiguous concept
• Internationally key location
• U.S. Underwriters most likely loss amount
  given that a significant loss event has taken
  place
• U.S. Actuaries 99 percentile (?)

20
Property Per-Risk Problem 5

• Solution
• Follow concept of U.S. Underwriters
• Divide the world into big and small losses
• (small losses lt1 of TIV are 75 of counts)
• Define severity as mix of big and small
• Define PML Eloss big

21
Property Per-Risk Conclusions

• Property Per-Risk Pricing is not a solved
  problem.
• Towards a solution
• Need for Data Standard
• Need to make use of all available data