Pricing Excess US Workers Compensation

1
Pricing Excess US Workers Compensation

• July 16, 2007

CARe Meeting
Jose Couret London Underwriting Center
2
Outline

• Terminology
• Exposure Rating
• Experience Rating
• Impact of Benefit Changes
• Questions

3
Terminology
4
TerminologyWorkers Compensation

• Workers Compensation Policy
• Coverage A Workers Compensation
• Coverage B Employers Liability
• Workers compensation coverage is statutory
  benefits are unlimited.
• Employers Liability Losses generally make up less
  than 1 of the loss dollars (excluding F-classes
  and Maritime Classes).
• Not separately rated
• Limited, but increased limits available

5
TerminologyState Versus Federal Benefits

• Benefits are state-specific however, some
  classes are subject to the Federal Act.
• F-Class longshoremen, harbor workers, etc.
• http//www.dol.gov/esa/owcp/dlhwc/lstable.htm
• Federal benefits tend to be higher
• Maximum weekly indemnity benefit is higher for F
  classes.
• Currently, wage replacement is capped at
  1,114.44 /week
• Wages relatively high
• Federal benefits subject to escalation
• Ratio of loss assessments to pure loss is much
  higher.
• Important if assessments are part of UNL
• Other acts Merchant Marine Act (Jones Act) /
  FELA, FECA

6
Terminology Premium

• Manual PremiumPayroll in Hundreds x Manual Rate
• Exceptions (e.g. per capita classes)
• Standard Premium includes experience modification
  but is prior to the application of adjustments
  such as retrospective rating plan adjustments,
  schedule rating, and premium discounts.
• Note Unit Statistical Plan definition may vary.
• Net Premium is used to refer to the premium
  actually charged as opposed to the manual
  premium or modified premium. Does not mean net of
  reinsurance in this context.
• Industry rate changes sometimes refer to
  changes in bureau rates or standard premium.
  During a soft market the industry rate changes
  may be flat while actual prices are plummeting.

7
Terminology Premium On-Level Factor

• Net Premium On-Level Factor (AY)
• Standard Premium On-Level Factor (AY)
• x Ratio of Net to Standard (Future)
• / by Ratio of Net to Standard (AY)

8
Exposure Rating
9
Exposure RatingDiscussion

• Exposure rating is essentially a two-step process
• step 1 come up with a ground-up loss ratio or
  loss by state and hazard group
• step 2 allocate ground-up expected losses by
  layer
• Within a given state, the manual rates are
  grossed up to the same permissible loss ratioso
  the expected loss ratio (to manual premium) is
  state-specific, not hazard group specific.
• The percent allocation of loss by layer varies by
  HG, the more severe hazard groups having the
  higher excess loss costs.
• Rating bureaus provide excess loss factors that
  specify the percent of ground-up loss in excess
  of a given percentage.
• NCCI (20K/year)
• WCIRBonline.org (California)
• LERs
• Other Independent Rating Bureaus

10
Exposure RatingUseful Data Sources (NCCI)

• Annual Statistical Bulletin
• Must have!
• NCCI Calendar-Accident Year Underwriting Results
  (by State)
• Free download at NCCI.com
• Latest accident year now available is 2005
• Source of default loss ratios by state
• Need to adjust for subsequent rate level changes
  and trend
• NCCI State Advisory Forum Presentations
• Free downloads at NCCI.com
• Great source of rate adequacy information
• crucial information with workers compensation
  system stakeholders
• NCCI Status of Rate Revision
• 500/year
• Latest bureau rate level information for NCCI
  states

11
Exposure RatingDiscussion

• A key element of the excess percentage is the
  frequency of loss by injury type. Fatalities and
  permanent disabilities cost more than other
  injury types so when they have high relative
  frequency, more of the claims cost arises from
  large losses.
• Relative Frequency claim count for the injury
  type divided by the claim count for temporary
  total.
• Relative frequency for the more serious injury
  types should increase as one moves from a lower
  hazard group to a higher hazard group.
• Fatal
• Permanent Total
• Major Permanent Partial

12
Exposure RatingAn ELF is a Weighted Average of
the ELFs by Injury Type
13
Exposure RatingVariation in Excess Loss Costs
Within and Between Hazard Groups Sample State,
4m XS 1M
14
Exposure RatingRating by Class Versus Rating by
Hazard Group

• Hazard group approach accurate for most treaty
  business, since most books are sufficiently
  diversified within hazard groups.
• Hazard group approach is probably not
  sufficiently refined for facultative pricing or
  specialty books.

15
Exposure RatingUse of Payroll in Exposure Rating

• Are ground-up loss ratio picks reasonable?
• Rates/Loss Costs for NCCI states are available in
  electronic format.
• Cost for non-affiliates is 22K/year
• Available (free) for many independent states on
  bureau web sites.
• Also available through SilverPlume
• Apply bureau loss cost rates to accounts payroll
  distribution by state and class.
• Result should be consistent with loss ratio
  assumption.
• Bureau loss cost rates typically are not pure
  loss
• include allocated and unallocated loss adjustment
  expense and, sometimes, loss-based assessments.
• Use of payroll weights as a proxy for expected
  loss weights in exposure rating understates
  excess loss costs since the average rate
  increases with hazard group.

16
Exposure RatingMaximum Any One Live (MAOL)

• Overwhelming majority of serious workers
  compensation occurrences involve a single
  claimant.
• Bureau statistical plans collect data on
  multi-claim occurrences impacting the same
  policy. Do not pick up clash between different
  policies that may be in the reinsurance treaty.
• Surprisingly little difference between bureau per
  claim and per occurrence excess loss factors.
  Not intended to incorporate terrorism or certain
  non-ratable exposures.
• Construction classes have higher incidence of
  multi-claim occurrences (anecdotal).
• What fraction of the 5M XS 5M layer loss is
  eliminated by a 5M MAOL?

17
Exposure RatingMiscellaneous

• Coding of premium to governing class (the class
  with the most payroll) may distort exposure
  rating.
• A single policy may have exposure in multiple
  classes.
• Should be manual or standard premium by state and
  class
• Excess Loss Factors are produced by rating
  bureaus for Retrospective Rating, not reinsurance
  pricing.
• Non-ratable classes with catastrophic experience
  not assigned a hazard group.
• Example 0766 is non-ratable component of 4766
  EXPLOSIVES OR AMMUNITION MFG.
• Ratio of ALAE to loss should reflect company
  experience.
• California bureau (WCIRB.com) plan has nine
  hazard groups. A mapping to four hazard groups
  is available however, these are NOT equivalent
  to NCCI hazard groups.

18
Experience Rating
19
Experience RatingExcess Loss Reporting Patterns

• Key driver company case reserving practices with
  respect to medical inflation
• Do medical case reserves incorporate a medical
  inflation assumption?
• Company discounting practice
• Does the company have a Catastrophic Claims Unit?
• Practice of settling fatal claims at present
  value reduces exposure in excess layers.
• Fatal and pt claims close for less than the
  carried reserve even if reserves are inadequate.

20
Experience RatingRatio of Undiscounted to
Discounted Annuities
21
Experience RatingExcess Loss Reporting Patterns

• At the May 11, 2007 AIS Research Workshop, NCCI
  released a study of excess loss and claim count
  reporting patterns based on Call 31.
• https//www.ncci.com/ncci/media/pdf/AIS_2007_Exces
  s_Loss_Development.pdf
• https//www.ncci.com/ncci/media/pdf/AIS_2007_WC_Ap
  pendix.pdf
• Statutory excess of various attachment points.
• Can be used in combination with ELFs to derive
  development patterns for limited layers
• Layer LDF (ELF_at_AP
  ELF_at_(APL)) )
  
• (ELF_at_AP/LDF_at_AP -
  ELF_at_(APL)/LDF_at_(APL)
• Easy to build a spreadsheet tool to derive
  patterns for any layer.

22
Experience RatingExcess Loss Reporting Patterns

• Consider a layer with attachment point AP and
  Limit L.
• The reporting pattern for statutory XS AP is a
  weighted average of the patterns for
• L XS AP
• Statutory XS (AP L)
• Question What if the average report lag for
  Statutory XS (AP L) is shorter than that for L
  XS AP? (This seems to be the case in NCCIs
  study.)
• Answer The average report lag for the layer L XS
  AP will be greater than that for Statutory XS AP
• Illustration (fictitious numbers) AP2M, L1M
• Statutory XS 2M
• Report lag 7 years
• ELF7
• Statutory XS 3M
• Report lag 6 years
• ELF5
• 1M XS 2M
• Report lag(7 x7 5 x 6)/(7 - 5) 9.5 years
• ELF2

23
Experience RatingSpeculation on Excess
Development Patterns

• Perhaps, the loss potential of very large claims,
  the kinds of claims likely to breach a 3M layer,
  is to some extent recognized early on.
• Characterized by catastrophic medical costs
• Other claims eventually over a long period of
  time creep their way into excess layers.
• Not characterized by catastrophic medical costs
  or potential not recognized
• Excess development driven in part by unwinding of
  the discount.
• Such claims would perhaps eventually make it to
  1M, but not 3M.
• The report lag for the lower layer would be
  lengthened by these slow developing claims but
  not the higher layer.
• Unfortunately, development to a 22nd report not
  available below 2M.

24
Experience RatingLarge Loss Severity Trend

• Using ground-up severity trends as a proxy for
  large loss trend in experience rating seemingly
  has an upward bias.
• NCCI estimates that differential frequency trend
  accounted for perhaps 2-3 of the trend in the
  average cost per case.
• For a discussion of this effect, see NCCIs 2006
  State of the Line Report.
• In many states, indemnity benefits for PT and
  Fatal do not escalate.
• Should consider changes in the maximum weekly
  benefit over the trend period.
• Is inflation on attendant care more like wage
  inflation than medical inflation?
• Second Injury Funds

25
Experience RatingLarge Loss Severity Trend
In the above illustration, ground-up trend of
6.47 overstates the true large loss trend of
5. The effect results from differential
frequency trend by injury type. NOTE For
illustration only, not based on real data.
26
Impact of Benefit Changeson Shape of
Size-of-Loss Distribution
27
Impact of Law Amendment FactorsGround-Up Versus
Excess

• Law amendment factors are used to adjust
  historical losses to prospective benefit levels.
  Since rating bureaus derive these factors to
  estimate the impact on ground-up loss costs of a
  law change, we must be extremely careful when we
  apply these factors sometimes unknowingly --in
  pricing excess business.
• For example, some law amendment factors reflect a
  shift of losses from one injury "bucket" to
  another. For example, a law that restricts
  access to permanent total benefits by tightening
  the definition of "permanent total" will have the
  effect of shifting claims from the permanent
  total category to the less generous major
  permanent partial category.
• The expected dollars of pt loss will decrease
  significantly as a result of such a law change
  while the dollars of major permanent partial
  indemnity loss will increase by some
  less-than-offsetting amount.
• Overall, there will be a decrease.
• It is appropriate to apply such law amendment
  factors to ground-up losses in pricing
  proportional business but perhaps incorrect for
  non-proportional business.

28
Impact of Law Amendment FactorsGround-Up Versus
Excess (Illustration)

• Before Law Change
• 100 pt cases with an average severity of
  1,000,000.
• Of these, 20 were borderline pt cases with an
  average severity of 400,000.
• The remaining 80 claims had an average severity
  of 1,150,000. (1,000,000 .2 x 400,000 .8 x
  1,150,000)
• 700 major permanent partial cases with an average
  severity of 200,000.
• Effect of the Law Change
• Shifts the 20 borderline pt claims to major pp.
• The average severity for the 20 claims will be
  300,000 under the less generous pp benefits.

29
Impact of Law Amendment FactorsGround-Up Versus
Excess (Illustration)

• The number of pt claims drops 20.
• The average severity for pt claims increases from
  1,000,000 to 1,150,000 (15).
• The dollars of PT loss decreases by 8, from
  100,000,000 to 92,000,000.
• The number of major pp claims increases by 20
  claims (about 2.9).
• The average severity for major pp claims
  increases by 1.4 (to 202,778)
• The dollars of major pp loss increase by
  6,000,000.
• Overall, losses are decreased by 2,000,000 due
  to the law change.
• There may be no effect excess of 1,000,000.

30
Impact of Law Amendment FactorsConclusion

• Reinsurers who best predict the differential
  impact of law changes by layer will have a
  competitive advantage.
• Sources for Information on Law Amendment Factors
• Annual Statistical Bulletin (ASB)
• WCIRB has extensive information.
• Other rating bureaus
• GOOGLE

31
Questions