Title: CARe Reinsurance Boot Camp on Pricing Techniques
1CARe Reinsurance Boot Camp on Pricing Techniques
Robert S. Yenke, ACAS August 9, 2007
2 - The views and opinions expressed in this
presentation are solely my own, and do not
necessarily reflect the views or opinions of
Odyssey America Reinsurance Corporation.
3Outline
- Workers Compensation Background
- Quota Share Treaties
- Excess of Loss Treaties
- Catastrophe Treaties
- Facultative
4Workers Compensation Background
- Started in U.S. in early 1900s, before Social
Security, before Federal Income Tax Withholding - Benefits were low
- Early 1970s National Commission of State Workers
Compensation Laws - Frequency declines, severity increases
5Workers Compensation Background
- Independent Bureau states California, Delaware,
Indiana, Massachusetts, Michigan, Minnesota, New
Jersey, New York, North Carolina, Pennsylvania,
Texas, Wisconsin - Monopolistic state Funds North Dakota, Ohio,
Washington, West Virginia, Wyoming
6Workers Compensation Background
- Extremely long tail
- Annuity-type benefits for Survivors and Permanent
Total claims - Benefits defined by state law, not by courts
- Indemnity and Medical benefits
- Benefits, to some degree, vary by state
- No policy limits
- Essentially unlimited medical coverage
- No-Fault system
7Quota Share Treaties
- Does the treaty cover losses up to a dollar
amount, such as the first million, or the ceding
companys Net? - If Net, what happens if the ceding company does
not renew its Excess of Loss treaties? What if
there is a loss above the top of the ceding
companys reinsurance?
8Quota Share Treaties
- Quota Share up to a dollar limit - What is the
cost of Excess of Loss reinsurance treaties? - Compare Premium Net of Reinsurance Cost to
Expected Losses at dollar limit - Keep Actuaries Honest
9Quota Share Treaties
- Is the ceding company writing Retrospectively
Rated policies? If so, how are premium
adjustments accounted for. - Benefit level changes for states in analysis need
to be included in estimate of projected loss
ratio - Use state specific trends if ceding company is
predominately in only one or a small number of
states - In addition to filed rate changes, need to obtain
impact of Schedule Rating, Group Discounts, etc.
10Quota Share Treaties
- Standard Quota Share Treaty issues including
Ceding commission versus actual expenses, Impact
of Sliding Scale Commission or Profit Commission - What is considered ALAE vs. ULAE?
- What is included in definition of Subject
Premium, e.g. Expense Constant? - For smaller companies and start-up operations you
can obtain state specific loss and ALAE ratios
from bureaus usually a little old
11Quota Share Treaties
- Quota Share of Excess WC Insurance
- Who is handling the ground-up claims?
- Reinsurer is relying on Excess Insurer to audit
original claims adjustors. There could be many
different entities handling the claims scary!
12Excess of Loss Treaties
- Typical layers of 4M xs 1M, 3M xs 2M
- Working layers of 750K xs 250K, 500K xs 500K
are less common than they used to be - Occasionally there are requests for lower layers,
but in most cases the primary company decides to
keep the layer Net - Some layers are unusual, such as 6M xs 1M, or
8.75M xs 1.25M
13Excess of Loss Treaties
- Many treaties have free and unlimited
reinstatements - Some treaties have a limited number of free
reinstatements, often expressed as a maximum
aggregate recoverable e.g. 4M xs 1M with four
reinstatements 20M aggregate cap - Usually the aggregate cap is set high, so there
is a low a probability that it will be exceeded,
still nice to have - Some treaties have Annual Aggregate Deductibles
14Excess of Loss Treaties
- Most treaties are flat rated
- Some treaties have adjustable features such as
Swing Rating, Profit Commissions or Reinstatement
Premiums - The long loss reporting patterns make the number
of potential premium or commission adjustments
very large
15Excess of Loss Treaties Claims
- WC Injury Types
- Death
- Permanent Total
- Major Permanent Partial
- Minor Permanent Partial
- Temporary Total
- Medical Only
- Only the first three impact most Excess of Loss
Treaties
16Excess of Loss TreatiesClaims
- Some claims are recognized quickly as high cost,
e.g. multiple person catastrophes - Many claims are take years and years to develop
into excess layer - Injured Worker Mortality how different from
standard population mortality?
17Excess of Loss TreatiesClaims
- Workers have lower Mortality than general
population - Injured Workers identified as Permanently
Disabled have similar Mortality as population - Sometimes severely injured workers die before
being classified as Permanently Disabled - Medical care for WC injury may help early
diagnoses of other issues - Claims can develop adversely years later
- Family stops taking care of claimant
- Back injuries creep into the layer
18Excess of Loss Treaties Loss Development
- Discounting of case reserves is standard
- Explicit indemnity reserves on lifetime pension
cases discounted 3.5 interest rate common - Implicit medical by not inflating projected
future payments - Mortality assumptions used in setting reserves
Bureau tables are used for Statistical Reporting,
some companies have mortality tables that vary
with injury severity. - Effect of unwinding of discount can be much
larger on an excess layer than the retained layer
19Excess of Loss Treaties Loss Development
- Long payment pattern, explicit discounting of
indemnity reserves, implicit discounting of
medical reserves produces large Excess LDFs - Account specific LDFs often have few claims,
therefore LDFs are volatile for higher layers - RAA gathers data from reinsures every other year
and publishes LDFs by line of business - In the 2005 study, WC Treaty data included
Accident Years from 2004 back to 1958, 46 years!
20Excess of Loss Treaties Loss Development
- My analysis produced a Treaty WC Incurred LDF
from 12 months to ultimate of 11. Therefore,
less than 10 reported after one year - Gross Incurred LDF from 12 months to ultimate is
1.9 - Treaty WC LDF from 156 months to ultimate LDF is
2. Therefore, after 13 years, only 50 of
ultimate losses have been reported, the same as
Gross LDF at 1 year! - The Incurred LDF from 45 to 46 is 1.01, there is
still IBNR at 45 years!
21Excess of Loss Treaties Loss Development
- 2005 RAA Study included Paid Loss Development
data - My analysis of Treaty WC Paid LDFs produced a 12
month to ultimate of 190 approximately 0.5 is
paid at 12 months. The Gross LDF is 4.5 - The Paid LDF to Ultimate reaches 2 at 22 years,
i.e. 50 paid at 22 years, the Gross LDF is 2 at
2 years - The Paid to Case Incurred Ratio at 46 years is
96 - there are still open cases
22Excess of Loss TreatiesImpact on Reinsurer
- An injured worker is expected to live 10 years.
- Weekly indemnity benefits are 500/wk 26,000/yr
- Expected indemnity benefit 260,000
- Discounted value at 3.5 interest 220,000
- Initial stabilizing medical expenses are 150,000
- Annual medical expenses are 50,000/yr
- Expected medical benefit 650,000
- Undiscounted case reserve 260K 650K 910K
- Discounted case reserve 220K 650K 870K
- Expected loss to the 1M xs 1M reinsurer is zero.
23Excess of Loss TreatiesImplicit Discount Effect
- Assume medical expenses are inflating at 6 per
year - Primary company books the ongoing medical loss at
500K, implicitly discounting them at 6/yr. - Total undiscounted ongoing medical expenses are
really 680K instead of the booked 500K - The total undiscounted loss is 260K 150K 680K
1,090K and the 1M xs 1M reinsurer will see 90K of
loss development - If the injured workers life expectancy is 20,
30, 40 or more years, the impact is much larger
24Excess of Loss Treaties Mortality Assumption
Effect
- Instead of a certain 10 year survival, there was
a 50 probability of this worker living only 5
years and 50 of living 15 years. - Losses paid if the claimant lives 5 years 530K
150K (26K50K) 5 - Losses paid if the claimant lives 15 years
1,290K 150K(26K50K) 15 - There is a 50 probability, the 1M xs 1M
reinsurer will see no loss development and 50
probability the reinsurers will see 290K of loss
development
25Excess of Loss Treaties Trend
- The long tail and unlimited medical benefits add
to the difficulty in estimating trends for
Workers Compensation. - In addition, states can and do change the WC
benefits, adding to the uncertainty.
26Excess of Loss TreatiesImpact of Non-uniform
Frequency Trend on Excess of Loss Pricing
- Exposure Rating
- Shape of the distribution changes, with more of
the losses coming from larger claims - Experience Rating
- Measured ground-up severity trend will increase
from the reduced frequency of the smaller claims - Assuming uniform trend by size of loss, the
measured large loss trend will be lower than the
measured ground-up trend - This impact is mitigated by the less-negative
frequency trend
27Excess of Loss TreatiesTrend Example
- Two types of claims, small and large.
- In year 1, small claims have average severity of
100K, while large claims have average severity of
500K. - In year 1, there are an equal number of small and
large claims, say 50 of each claim - Total Losses are 30,000,000
- Average Severity in year 1 is 300K
- (50100k 50500k)/(5050)
28Excess of Loss Treaties Trend Example, continued
- In year 2, there are now 40 small claims
(frequency trend -20), while there are still
50 large claims (0 frequency trend). Total
frequency trend -10 - The average severity for each claim type
increases 10 - Small claim severity 110K
- Large Claim Severity 550K
- Total losses are not 31,900,000 an increase of
6 - But, the measured overall severity is now 354K
- (40110K 50550K)/(4050)
- This is an 18 increase!
29Excess of Loss Treaties Benefit Changes
- Most benefit changes are small, increase in
maximum weekly benefit, change in burial
allowance, etc. - Large changes occur rarely, but sometimes in
quick succession impact by injury type can be
vary - California AB 749 January 1, 2003 PT Benefit
Impact 54, Overall Impact 5 - California AB 227, SB 228 January 1, 2004,
Overall Impact -9 - California SB 899 April 19, 2004 Overall Impact
-20, January 1, 2005 Overall Impact -14 - California January 1, 2006 Fatal Benefit Impact
50, Overall Impact 3
30Excess of Loss Treaties Experience Rating
- Paid losses projections are not usable
- Size of incurred LDFs gives projections low
credibility - Presence or absence of large loss in a recent
period produces large impact on projected losses
to layer if LDF method is used - B-F method, using Cape Cod method to estimate
initial loss cost
31Excess of Loss Treaties Experience Rating
- Higher layers may have no reported losses to
develop. - I do not want to compare a zero from experience
rating to the loss cost obtained from exposure
rating. - Use ratio of exposure rating loss cost of higher
layer to credible layer to apply to experience
rating loss cost of credible layer
32Excess of Loss TreatiesExposure Rating
- Workers Compensation is similar to other lines
- Compute overall expected losses
- Allocate these losses to the layer being priced
by an industry size-of-loss curve
33Excess of Loss TreatiesExposure Rating
- For lines with policy limits, like GL, the rating
bureaus publish ILFs, which are based on
size-of-loss curves - But, WC doesnt have policy limits.
- Reinsurers use Retrospective Ratings Excess Loss
Factors (ELFs), even though they were not
designed for reinsurance
34Excess of Loss TreatiesExposure Rating
- Excess Loss Factors tables are published by
state, typically for loss limits up to 10M, by
Hazard Group - Variations by state are due to differences in
benefit levels - Variations by Hazard Group are due to differences
in the distribution of injury types by Hazard
Group
35Excess of Loss TreatiesExposure Rating
- 2004 ELF changes
- More reliance on state data
- Use empirical data for small claims, fit
distribution for tail - Distributions fit for Fatal, Permanent Total,
Permanent Partial, Temporary Total and Medical
Only - Comparison to Prior ELFs
- Percent of total losses in low layers, less than
1M similar to prior factors, some states
increased, some decreased - Percent of total losses in layers above 1M
dropped significantly in almost every state
36Excess of Loss TreatiesExposure Rating
- Comparison to Prior ELFs
- Refitting current data using prior procedure
produced ELFs much closer to new procedure - Conclusion data, not procedural change, drove
most of the reduction in percent of losses in
high layers - Decline in claim frequency in 1990s changed
theshape of the loss distributions
37Excess of Loss Treaties Exposure Rating How
Many Hazard Groups?
- NCCI and the other bureaus used four Hazard
Groups, however, 95 of the premium was in Hazard
Group II or III. Mining, explosive
manufacturing, longshore, in HG IV. Much HG IV
was in Assigned Risk Pools, so business written
Voluntarily included very little HG IV business
excluded from many treaties except for
incidental. - WCIRB adopted nine Hazard Groups A-I, which were
subsequently updated to J-R. A four Hazard Group
set of LERs was produced but class assignments
did not match NCCI groupings. - NCCI did there own study and selected seven.
38Excess of Loss Treaties Exposure Rating
- Seven Hazard Groups are A through G, collapsible
into four (1, 2, 3 and 4) AB into 1, CD into 2,
EF into 3, G into 4 - Premium by Hazard Group is much more evenly split
Hazard Groups have really increased from 2 to
7 - Improves estimates of losses by layer for
reinsurers and for primary companies, for example
Loss Rated Accounts
39Excess of Loss Treaties Exposure Rating
- Suppose were pricing the 1M xs 1M layer
- Expected Loss Ratio 70
- ELF(1M) 0.10 ELF(2M) 0.06
- Losses in the layer ELF(1M) ELF(2M) 4.0
- 4.0 of the total losses are in this 1M xs 1M
layer - Exposure Loss Cost 70 4.0 2.8
40Excess of Loss Treaties
- Combine experience and exposure loss cost
- Discount?
- Load for internal and external expenses
- When do you know true result?
41Catastrophe Treaties
- Treaties requiring at least two injured workers,
usually above 5M - Maximum Any One Life (MAOL)
- Example 5M xs 5M, 5M MAOL requires the total
loss from one occurrence to exceed 5M, with no
single injured worker contributing more than 5M - Similarities to Property Cat Treaties
42Catastrophe Treaties
- Reinstatement terms frequently one
reinstatement of limit with a premium expressed
as a percent of original reinsurance premium. - Reinstatement premiums are typically paid when
losses are paid which can be many years from
inception, unlike Property Cat treaties.
43Catastrophe Treaties
- WC Catastrophe Layers extend up to 100M and
frequently higher, far beyond the ELFs. - Pricing is frequently expressed as Rate on Line
Premium / Limit - Rates on Line vary by layer and exposure
- Higher layers have lower Rates on Line
44Catastrophe Treaties
- Exposure is not necessarily correlated to Workers
Compensation Premium the Catastrophe risk is
also related to Payroll - Example two employers with the same WC Premium
but the first has a WC rate 10 times as high as
the second. The second employer has a Payroll 10
times the first. The Catastrophe risk is not
equal.
45Catastrophe Treaties
- Earthquake exposed areas have higher Rates on
Line - Hurricanes are not considered a significant
exposure for Workers Compensation except for
First Responders, most workers are not at work
when the storm arrives - Areas considered at higher risk for Terrorism
have higher Rates on Line
46Catastrophe Treaties
- Similar to Property Cat after Hurricane Andrew,
information has increased - Some submissions include information on
Payroll/Premium/Employee counts by insured, or
zip code, and time of day/shift or maximum
exposed employee counts - Similar to Property Cat, a number of reinsurers
are needed to provide layers requested. Some
participate on lower layers, others on higher
layers.
47Catastrophe Treaties
- As with Property Catastrophe, modeling firms are
analyzing the exposures - As with Property Catastrophe, reinsurers monitor
their aggregate exposure to WC Catastrophe
losses - Potential exists for a Property Catastrophe loss
and a Workers Compensation Catastrophe loss
48Catastrophe Treaties
- Much of the business was written in the London
market or by Life Insurers - Terrorism was not recognized as a risk
- After 9/11, many participants suffered losses and
left the market - Rates on Line increased dramatically, attracting
traditional reinsurers - Coverage was restricted i.e. limits on Terrorism
49Catastrophe Treaties
- Rates on Line have been declining
- Coverage has been expanding more Terrorism
coverage, requests for including NBCR - Fortunately, there have been no significant
losses to this book of business since 9/11 - Unfortunately, a significant Workers Compensation
Catastrophe will eventually happen, whether a
natural Catastrophe or a man-made Catastrophe - As with Property Cat pricing after a large event,
pricing will increase and coverage will be
restricted.
50Facultative
- Individual account pricing.
- By its very nature facultative reinsurance
operates based on adverse selections. Through
careful underwriting and pricing the reinsurers
can establish the proper pricing and terms which
would be profitable to the reinsurer. - The ceding companies will place facultative
reinsurance due to class, exposures, to protect
the treaty, a treaty exclusion or cat exposures.
51Facultative
- Ceding companies are often looking for
reinsurance in Buffer layers such as 750K xs
250K, 500K xs 500K - Underwriters use a combination of loss rating and
manual rating - Loss rating up to a loss limit, usually well
below reinsurance attachment point, then use
ELFs to estimate layer loss cost
52Facultative
- Credibility weight loss rating with manual rating
- Load expected loss cost for internal and external
expenses - In the current market ceding companies want the
WC facultative market to either match their
pricing for the layer or be less than what they
price it for. In the event that this can't be
met then the ceding company would retained the
risk Net.
53Workers Compensation Suggested Reading
- Commentary on the New Hazard Groups by Jose
Couret. Presentation at Spring 2007 CAS Meeting - The 2004 NCCI Excess Loss Factors by Dan Corro
and Greg Engl. CAS Forum 2006 - An Actuarial Note on Workers Compensation Loss
Reserves 25 Years Later by Lee Steeneck. CAS
Forum 1996 - Ratemaking for Excess Workers Compensation
Insurance, by Owen Gleeson. CAS Forum 2001 - Levels of Determinism in Workers Compensation
Reinsurance Commutations by Gary Blumsohn.
Proceedings 1999