Property Reinsurance Ratemaking

1
Property Reinsurance Ratemaking

• Sean Devlin
• Reinsurance Boot Camp
• on Pricing Techniques
• July 29, 2005

2
Agenda

• Background
• ELR determination
• Primary Price
• Experience Rating
• Exposure Rating
• Weighting of Methods
• Catastrophe Loads and Issues
• Conversion of Loss Cost to Pricing
• Summary and Questions

3
Background

• My past experience, particularly
• AmRe 3 years of leading Finite, National and
  Specialty business pricing
• GE 3 years leading Global Property Product
  Pricing
• What I have seen
• Common mistakes,
• Emerging exposures
• Worst and best of the market
• The most complex treaties
• Management of a global portfolio and its effect
  on strategy and pricing

4
Exposure Rating

5
ELR Determination

• Foundation of Exposure Rating
• Which ELR to use?
• Must match your curve in exposure rating
• Preference Eliminate cat as much as possible
• Options for ELR
• Full LR
• No cat whatsoever
• Exclude certain cats
• Methodology
• Equivalent to primary ratemaking, except
• Need for factors to back out certain cats to
  match exposure curve, if the match isnt already
  made

6
ELR Calculation - Per Risk/Pro Rata

• Determining your ELR
• Breakout components
• Basic LR very stable small, non-cat events
• Risk LR losses subject to a per risk Layer
• Breakout into layers, like per risk rating
• Appropriate blend of experience exposure
• Small Cat LR(s) experience rate vs. model
• Modeled Cats
• More reasons for breakout?
• Inuring reinsurance or contract features
• Understand the drivers of the ELR
• Appropriate targets for quoting business

7
ELR Determination

• Trend Parameters
• Cost of contracting labor
• Size of homes increasing
• Deductible impacts on frequency and severity
• Data shifts in and out of ES market
• Excess business
• Non-standard classes
• Demand surge

8
Note on Primary Price

• Price Monitoring Reports
• Typically created to measure price lift circa
  2000
• Know what is (isnt) captured
• Filed rate changes
• Schedule modification factors
• Experience modification factors
• SIR/Limit
• Terms and conditions
• New business
• Test for bias
• Trend or shift in adjusted loss ratios
• Discuss with client changes
• More important for high capacity eaters

9
Note on Primary Price

• Effect of missing uncaptured price
• Typically underestimated the magnitude of change
• Softening Cycle
• Underestimating decreased rates
• Underestimating reserves
• Calendar year results lag true results
• Delays recognition of results
• Softening prolonged damage is slowly realized
• Hardening Cycle
• Underestimating increased rates
• Overestimating reserves
• Calendar year results lag true results
• Delays recognition of results
• Hardening prolonged success is slowly realized

10
Primary Price (contd)

11
Primary Price (contd)

Uncaptured Rate change
12
Primary Price (contd)

Actual peak of soft market
Calendar Year results understated during soft
market
Should be hardening here
13
Exposure and Experience Rating

14
Experience Rating

• Premium Side
• Same as pro rata, mostly
• Splitting up business into exposed and not
  exposed
• In split business, parameters may be different
• Exiting class? Reflect all premium affected if
  excl.
• Loss Side
• Capping at policy limits TIV and loss both
  trend
• Losses should be on same basis as exposure
  rating
• Reflective of per risk definition READ the
  slip
• Two methods to calculate burning cost
• Empirical - weighted
• Fit distribution
• Split quoted layers into sub layers to add
  credibility

15
Exposure Rating Loss Curves (contd)

• General Considerations
• ELR must reflect the data underlying loss curve
• Understanding of the data and assumptions is key
• Assumptions of the loss curves
• Data in exposure profiles
• What curves to use
• PSOLD
• Lloyds curves
• Salzmann curves
• Ludwig curves
• Curves created by reinsurers

16
Exposure Rating Loss Curves (contd)

• PSOLD
• Becoming a standard
• Most recent data
• Only one that varies my AOI
• Has the most variables
• More on this later
• Lloyds curves
• Reversals exist
• A premium calculator for facultative
• Source unknown
• Curves created by reinsurers
• Old data
• Source unknown in some cases

17
Exposure Rating Loss Curves (contd)

• Salzmann curves
• 1960 Cov A Fire Losses Only
• Varied by protection construction classes
• Not recommended by Salzmann herself
• Use was to describe first loss scales
• Ludwig curves
• 1984-88 data to update the Salzmann paper
• Based on Hartford Insurance Co. data
• HO - all coverages, all perils
• HO - varies by protection/construction
• CP - small commercial data
• CP - varies by occupancy class

18
Exposure Rating (contd)

• What is in the companies profile?
• Limits dont assume, ask if unsure
• Business interruption and/or contents included?
• Policy limit
• Location limit
• PML
• MFL
• Key location
• Limits or values for layered business
• ITV issues
• Other coverages
• Excess policies
• Subscription business

19
Exposure Rating (contd)

• What is in the companies profile (contd)?
• Any perils excluded?
• Homeowners
• Form (HO-2,3,4,5,6)
• Coverage A only or all coverages
• Farmowners
• Multiple diverse buildings on a farm
• One TIV
• Smell test for reasonability, especially
• Order of magnitude of some TIV
• Premium allocation

20
Exposure Rating - PSOLD 2004

• PSOLD
• Data from 1992-2002
• Can separate business by
• Occupancy 22 groups, diff. strongest btw.
• Manufacturing
• Non-manufacturing
• HPR
• Little differences within these groups
• State just distribution of business in a state
• Gross or Net of Deductible
• Include/Exclude Cats gt100M industry loss
• Coverage BGI, BGII, special, all
• Include/Exclude WTC
• Include/Exclude Business Interruption

21
Exposure Rating (contd)

• Issues With PSOLD
• Not all segments represented evenly by PSOLD
• Loss history is thin for some groups
• Based on 1.8M occurrences, after scrubbing
• Losses above 5M in the database are thin
• of losses gt 5M is 421
• of losses gt 10M is 243
• Refer to a list of large industry losses for
  more input
• Blanket policies small amount of database
• US business only applicable abroad?
• HO US homes are built out of cardboard
• Factory in US similar to one in UK?
• Main street business in US same as France?

22
Exposure Rating (contd)

• Application of PSOLD
• Occupancy classes
• 22 groups, diff. strongest btw.
• Manufacturing
• Non-manufacturing
• HPR
• Little differences within these groups
• May need to enter TIV profile by class
• HPR business is usually higher in limit
• BOP type bussiness usually smaller
• Excess Policies
• Subscription business

23
Exposure Rating (contd)

• Subscription and Excess Policies
• Participation on a single layer policy
• Insured writes 20 of a policy of 5M
• Reinsurance layer is 500K xs 500K
• Layer is really 25 of the loss 2.5M xs 2.5M
• Losses above the 5M limit is not relevant to
  layer
• Pure Excess Policies
• SIRs are important
• Limit TIV or a hard cap
• Blanket policies are common allocation issues
• 10M indivisible premium on 10 locations

24
Exposure Rating (contd)
Subscription Market Layers of 50x50 and 50x100

100x50 reinsurance layer 37.5M from 25 of
150xs350 12.5M unexposed if hard cap of 500M
50x50 reinsurance layer 25M from 25 of
100xs250 25M from 50 of 50x200
25
Exposure Rating (contd)

• Dont Trust the Black Box
• Check the output for reasonability
• Contract Match
• Definition of risk
• One building (possibly less)
• Multiple buildings at one location
• Entire policy
• Company has sole determination
• Exposure profiles
• Loss curve
• Dual trigger contracts cat and risk combined
• Scope of coverage
• READ THE SLIP

26
Weighting of Methods

• General Considerations
• Actual vs. Expected counts to layer
  (significant)
• Actual Needs to be adjusted for volume
• Severity differences may need to subdivide
  layer
• Make sure that both methods reflect the same
  risk
• No loss no weight to experience? Not
  necessarily
• Deficiencies in exposure data or curves
• Past experience indicative of future
• Do not be afraid of splitting quoted layer into
  parts

27
Catastrophe PerilPer RiskPro RataCat XL

28
Vendor Models What to Use?

• Major modeling firms
• AIR
• EQE
• RMS
• Other models, including proprietary
• Options in using the models
• Use one model exclusively
• Use one model by territory
• Use multiple models for each account

29
Vendor Models What to Use? (Contd)

• Use One Model Exclusively
• Benefits
• Simplify process for each deal
• Consistency of rating
• Lower cost of license
• Accumulation easier
• Running one model for each deal involves less
  time
• Drawbacks
• Cant see differences by deal and in general
• Conversion of data to your model format

30
Vendor Models What to Use? (Contd)

• Use One Model By Territory
• Detailed review of each model by territory
• Territory examples (EU wind, CA EQ, FL wind)
• Select adjustment factors for the chosen model
• Benefits
• Simplify process for each deal
• Consistency of rating
• Accumulation easier
• Running one model involves less time
• Drawbacks
• Cant see differences by deal
• Conversion of data to your model format

31
Vendor Models What to Use? (Contd)

Use One Model By Territory An Example
32
Vendor Models What to Use? (Contd)

• Use Multiple Models
• Benefits
• Can see differences by deal and in general
• Drawbacks
• Consistency of rating?
• Conversion of data to each model format
• Simplify process for each deal
• High cost of licenses
• Accumulation difficult
• Running one model for each deal is time
  consuming

33
Model Inputs

• Garbage In gt Garbage Out
• TIV checks/ aggregates
• As-if past events
• Scope of data (e.g. RMS WS, EQ, TO datasets)
• Which territories modeled and not modeled
• Type of country considered for exposures abroad
• Clash between separate zones (US Caribbean)

34
Unmodeled Perils

• Winter storm
• Not insignificant peril in some areas, esp. low
  layers
• 1993 1.75B 14th largest
• 1994 100M, 175M, 800M, 105M
• 1996 600M, 110M, 90M, 395M
• 2003 1.6B
• of occurrences in a cluster?????
• Possible Understatement of PCS data
• Methodology
• Degree considered in models
• Evaluate past event return period(s)
• Adjust loss for todays exposure
• Fit curve to events

35
Unmodeled Perils (contd)

• Flood
• Less frequent
• Development of land should increase frequency
• Methodology
• Degree considered in models
• Evaluate past event return period(s),if possible
• No loss history not necessarily no exposure
• Terrorism
• Modeled by vendor model? Scope?
• Adjustments needed
• Take-up rate current/future
• Future of TRIA exposure in 2006
• Other depends on data

36
Unmodeled Perils (contd)

• Wildfire
• Not just CA
• Oakland Fires 1.7B 15th largest
• Development of land should increase
  freq/severity
• Two main loss drivers
• Brush clearance mandated by code
• Roof type (wood shake vs. tiled)
• Methodology
• Degree considered in models
• Evaluate past event return period(s), if
  possible
• Incorporate Risk management, esp. changes
• No loss history not necessarily no exposure

37
Unmodeled Perils (contd)

• Fire Following
• No EQ coverage No loss potential? NO!!!!!
• Model reflective of FF exposure on EQ policies?
• Severity adjustment of event needed, if
• Some policies are EQ, some are FF only
• Only EQ was modeled
• Methodology
• Degree considered in models
• Compare to peer companies for FF only
• Default Loadings for unmodeled FF
• Multiplicative Loadings on EQ runs

38
Unmodeled Perils (contd)

• Extratropical wind
• National writers tend not to include TO
  exposures
• Models are improving, but not quite there yet
• Significant exposure
• Frequency TX
• Severity May 2003 event of 10B 9th largest
• Methodology
• Experience and exposure Rate
• Compare to peer companies with more data
• Compare experience data to ISO wind history
• Weight methods

39
Unmodeled Perils (contd)

• No Data
• Typically for per risk contracts without
  detailed data
• Typically not a loss driver on per risk treaties
• However, exceptions exist
• Methodology
• Experience and exposure Rate
• Compare to peer companies with modeling
• Develop default loads by layer/location

40
Unmodeled Perils (contd)

• Other Perils
• Expected the unexpected Dave Spiegler article
• Examples Blackout caused unexpected losses
• Methodology
• Blanket load
• Exclusions, Named Perils in contract
• Develop default loads/methodology for an
  complete list of perils

41
Using the Output

• Dont Trust the Black Box
• Data, Data, Data
• Contract Match
• Definition of risk
• Definition occurrence
• Dual trigger contracts
• Scope of coverage
• Modeling of past exposures
• Need to convert to prospective period
• TIV inflation
• Change in exposures
• Know what assumptions were used by modeler

42
Experience Rating Adjustments

Reduce 80 for more credible long term experience
43
Loadings to final EL

• Considerations in final indicated price
• of loss?
• of s?
• Combination of above?
• Target LR, TR, CR?
• Reflect red zone capacity constraints?
• Unused capacity loads
• EL for Layer 100M x 100M is 5M
• EL for Layer 200M x 100M is 5.1M
• Loading for 100M x 200M??????

44
Capacity Charge - Simplistic

45
Conversion to Pricing

• General Considerations
• Create loss distribution even if not needed
• Adjust for treaty features AAD, swing rate,
  etc.
• Understand upside and downside of deal
• Unpriced capacity blown limit, cat on tail
  of curve
• Is the rate on line appropriate
• Red Zone catastrophe utilization
• Treaty correlation to book
• Layered/Subscription business
• Catastrophes
• Soft Factors Dont be biased, though
• Check yourself for naive capital cheap cat
  cover

46
Finishing The Job

47
Pro Rata Example

Determining your Target Loss Ratio
48
Key Takeaways

• Understand the data inputs
• Understand your models and parameters
• Understand strength and weakness of the models
• Proper match to treaty terms READ THE SLIP
• Reflect true primary price
• Rate for everything
• Include the untested and unmodeled exposure
• Work with your underwriter
• Question everything Assume nothing at face
  value
• THINK - Dont Just Go Through The Motions