Basic Ratemaking

魏長賢,, FCAS, ASA

魏長賢

Ch
Chang-
Chang -Hsien
H i Wei
W i
 Academy
– Fu-
Fu-Jen Catholic University in Applied Math
– Feng Chia University in Statistics And Actuarial Science
– University of Illinois at Urbana-
Urbana-Champaign (UIUC) in International
Finance
 Working
g Experience
p
– CIGNA Life Insurance Company - Taiwan Branch, 1 yr
– Ming Tai Fire & Marine Insurance Company, 1 yr
– Fu Bon Insurance Company, 3.5 yrs
– S Paull Travelers
St. l Company,
C 6.0
6 0 yrs
 Technology Actuarial, Associate Actuary
 Personal Line Actuarial, Actuary
– Zurich Insurance Company,
p y, Taiwan,, 1 yr
y
– Want Want Union Insurance Company, 2 yrs
– KPMG, Taiwan
 Certificate
– Fellow of Casualty Actuarial Society (FCAS), 2005
– Associate of Society of Actuaries (SOA), 1994
B i Ratemaking
Basic R t ki
 Authors
– Geoff Werner, FCAS, MAAA
– Claudine
Cl di Modlin
M dli , FCAS,
Modlin, FCAS MAAA
 Version 3, January 2010
 Chapters
– Chapter 1 : Introduction
– Chapter 2 : Rating Manuals
– Chapter 3 : Ratemaking Data
– Chapter 4 : Exposures
– Chapter 5 : Premium
– Chapter 6 : Losses and LAE
– Chapter 7 : Other Expenses and Profit
– Chapter 8 : Overall Indication

Chapter 1
Introduction
Chapter 1 Introduction -
Basic Insurance Terms
 Price = Cost + Profit
– For many non
non--insurance goods and services, the
production cost is known before the product is sold
– For insurance, the ultimate cost of an insurance policy
is not known at them of the sale
 Exposure
– The basic unit of risk that underlies the insurance
premium
[Note] The premium is calculated based on a given rate
per unit of risk exposed
– Written Exposure
 The total exposure arising from policies issued during a specific
period of time, such as a calendar year or quarter
– Earned Exposure
 The portion of the written exposure for which coverage has
already been provided as of a certain point of time

Chapter 1 Introduction -
Basic Insurance Terms
– Unearned Exposure
 The portion of the written exposures for which coverage has not
yet been provided as of that point in time
– In-
In-Force Exposure
 The number of insured units that are exposed to loss at a given
point in time
– Will be discussed further in Chapter 4
 Premium
– The amount the insured pays for insurance coverage
– Same concepts of written, earned, unearned, and in
in--
force as those used in exposure
– Will be discussed further in Chapter5
Chapter 1 Introduction -
Basic Insurance Terms
 Claim
– The demand that an insured makes to the insurer for
indemnification
d f under
d an insurance policy
l
[Note] The individual making a claim is called a claimant
– Various dates associated with a claim
pure IBNR pipeline claim settlement
case procedure

Accident Report Opening Closing Reopen Resettlement

Date Date Date Date Date Date

Chapter 1 Introduction -
Basic Insurance Terms
 Loss
– The amount of compensation
p paid
p or payable
p y to the
claimant
– Paid Loss
 Those amounts that have been paid to claimants
– Case Reserve
 An estimate of the amount of money required to ultimately settle
that claim
[Note] The case reserve excludes any payments already made
– Reported Loss ( or Case Incurred Loss)
 Paid Loss + Case Reserve
Ch t 1 - Introduction
Chapter I t d ti
– Ultimate Loss
 The amount of money required to close and settle all claims for a
defined g
group
p of policies
p
 The aggregate sum of reported losses across all known claims may
not equal the ultimate loss for many years
– The reported losses on existing claims may change over time
– At any point in time, there may be unreported claims
– IBNR R
Reserve ( IIncurred
dbbutt N
Nott R
Reported)
t d)
 The amount estimated to ultimately settle the unreported claims
 Usually refer to pure IBNR
– IBNER Reserve ( Incurred but Not Enough Reported)
 The difference between the aggregate reported losses at the time the
losses are evaluated and the aggregate amount estimated to
ultimately settle these reported claims
 a.k.a. Development on Known Claims or Case Emergence
[Note] Broad IBNR includes pure IBNR and IBNER and all
other reserves, but not case reserve
[[Note]] Estimate Ultimate Loss = Reported
p Loss + IBNR +
IBNER

Chapter 1 Introduction -
Basic Insurance Terms
 Loss Adjustment Expense (LAE)
– The expense
p occurred byy an insurer to close a claim
– ALAE (Allocated Loss Adjustment Expense)
 Claim-related expenses that are directly attributable to a specific
Claim-
claim
 Example, fees associated with outside legal counsel hired to
defend a claim can be directly assigned to a specific claim
– ULAE (Unallocated Loss Adjustment Expense)
 Claim--related expenses that cannot be directly assigned to a
Claim
specific claim
 Example, salaries of claims department
 Chapter 1 Introduction -
Basic Insurance Terms
 Underwriting Expenses
– The expenses for acquiring and servicing the policies
 C
Commission
i i and d Brokerage
B k
 Other Acquisition
 General Expense
 Taxes, Licenses, and Fees
– Commission and Brokerage
 Amount paid to insurance agents or brokers as compensation for
generating business
 Typically,
Typically paid as a percentage of premium written
– Other Acquisition
 Expenses other than commissions and brokerage expense paid to
acquire business
 Example,
Example advertisements
– General Expense
 The remaining expenses associated with the insurance operations and
any other miscellaneous costs
– Taxes, Licenses, and Fees
 All taxes and miscellaneous fees paid, but excluding income taxes

Chapter 1 Introduction –
F d
Fundamental l IInsurance
Equation
 Fundamental Insurance Equation
– Price = Cost + Profit
– Premium = Losses + LAE + UW Expenses + UW Profit
– The goal of ratemaking is to assure that the
fundamental insurance equation is appropriately
balanced
 Rates should be set so that the premium is expected to cover all
g underwriting
costs and achieve the target g profit
p
– Two key points to consider
 Ratemaking is prospective
 Balance should be attained at the aggregate and individual levles
Chapter 1 Introduction –
F d
Fundamental l IInsurance
Equation
 Ratemaking is prospective
– Insurance is a p
promise to provide
p compensation
p in the
event a specific loss event occurs during a defined
period in the future
– Using relevant historical experience to estimate the
future expected costs
[Note] This doesn’t mean actuaries are setting premium to recoup
past losses
– It is important to recognize that adjustments will be
necessary to convert the experience into that which will
p
be expected in the future when the rates will be in
effect
 Rate changes
 Operational changes
 Inflationary pressures
 Law changes

Chapter 1 Introduction –
F d
Fundamental l IInsurance
Equation
 Overall and individual balance
– Overall level
 Ensures that the total premium for all policies is sufficient to
cover the total expected losses and expenses and to provide for
the targeted profit
– Individual
I di id l level
l l
 A policy that presents significantly higher risk of loss should have
a higher premium than a policy that represents a significantly
lower risk of loss
Chapter 1 Introduction -
Basic Insurance Ratios
 Frequency
– A measure of the rate at which claims occur
– Frequency = # of Claims / # of Exposures
[Note] Earned Exposure should be used for the denominator
– Help measure the effectiveness of specific underwriting
actions
[Example 1-
1-1] If there are 100,000 written
exposures, and d 80% off the
th written
itt exposures are
earned during 2010. What is the frequency for
the year 2009 if 40,000 claims occurred during
the year?
[Sol] Earned Exposure = 80,000
Frequency = 40,000 / 80,000 = 0.5

Chapter 1 Introduction -
Basic Insurance Ratios
 Severity
– A measure of the average
g cost of claims
– Severity = Total Losses / # of Claims
– Can be
 Paid Severity = Paid Losses / # of Claims Closed
 Reported Severity = Reported Losses / # of Claims Reported

– Help measure the loss trends

[Example 1 1--2] Continue with Example 1
1--1,
1 if the
total losses are $200,000 thousand for the
40,000 claims. What is the severity?
[Sol] Severity = $200,000 / 40,000 = $5 thousand
Chapter 1 Introduction -
Basic Insurance Ratios
 Pure Premium (PP)
– As known as Loss Cost,, or Burning
g Cost
– A measure of the average loss per exposure
– PP = Total Losses / # of Exposures
= Frequency x Severity
[Note] Ultimate losses and earned exposures should be used
– Help measure the overall loss cost trends
[Example
[E l 1-
1-3] Continue
C i with
i h Example
E l 1-
1-2,
2 please
l
determine the pure premium.
[Sol] PP = $200
$200,000
000 / 80
80,000
000 = 0.5
0 5 x $5 thousand
= $2.5 thousand

Chapter 1 Introduction -
Basic Insurance Ratios
 Average Premium (PP)
– Average
g Premium = Total Premium / # of Exposures
p
[Note] Premium and exposures should be on the same basis (both
written or earned)
– Changes
g in average
g premium
p mayy indicate
 Rate change
 Change in the mix of business, e.g. deductible shift, insured limit
shift, risk shifts, and etc
[Example 1-
1-4] Continue with Example 1-
1-1, please
determine the average written premium if the
written premium is $300
$300,000
000 thousand.
thousand
[Sol] Average Premium = $300,000 / 100,000 = $3
thousand
Chapter 1 Introduction -
Basic Insurance Ratios
 Loss Ratio (LR)
– A measure of the p
portion of each premium
p dollar used to
pay losses
– LR = Total Losses / Total Premium
= Pure Premium / Average Premium
[Note] Ultimate losses and earned premium should be used
– Help measure the adequacy of the rates
[Example 1-
1-5] Continue with the examples above,
above
assuming the written premium is earned at the same
proportion
p p as written exposure,
p , Please determine the
loss ratio.
[Sol] Earned Premium = $300,000 x 80% = $240,000
LR = $200,000
$200 000 / $240,000
$240 000 = $2.5
$2 5 / $3 = 83.33%
83 33%

Chapter 1 Introduction -
Basic Insurance Ratios
 Loss Adjustment Expense Ratio (LAE Ratio)
– LAE Ratio = Total LAE / Total Losses
[Note] LAE includes both ALAE and ULAE
 Underwriting Expense Ratio (UW Expense Ratio)
– A measure
meas e of the po
portion
tion of each p
premium
emi m dolla
dollar used
sed
to pay for underwriting expenses
– UW Expense Ratio = Total UW Expenses / Total
P
Premium
i
– Two categories
 Variable expense, e.g. commissions, licenses and fees
 Fixed expense, e.g. salaries, office rental
[Note] The written premium is used to measure the variable
expense ratio, and the earned premium is used to measure the
fixed expense ratio
Chapter 1 Introduction -
Basic Insurance Ratios
 Operating Expense Ratio (OER)
– A measure of the p portion of each premium
p dollar used
to pay for loss adjustment and underwriting expense
– OER = UW Expense Ratio + Total LAE / Total Earned
Premium
[Note] LAE includes both ALAE and ULAE
– Used to monitor operational expenditures
 Combined Ratio
– A primary measure of the profitability of the book of
business
– Combined Ratio = Loss Ratio + OER
[Note] Loss ratio should not include LAE

Chapter 1 Introduction -
Basic Insurance Ratios
 Renewal Ratio
– Retention ratio is used in this book
– A measure of the rate at which existing insureds renew
their policies upon expiration
– Renewal Ratio = # of Policies Renewed / # of Potential
Renewal Policies
 Close Ratio
– A measure of the rate at which prospective insureds
accept a new business quote
– Close Ratio = # of Accepted
p Quote
Q / # of Quotes
Q
– Is used to determine the competitiveness of rates for
new business
Chapter 2
Rating Manuals

Ch t 2 Rating
Chapter R ti Manuals
M l
 Rating manual usually includes
– Rules
– Rate pages
– Rating algorithm
– Underwriting
Unde iting g guidelines
idelines
Chapter 2 Rating Manuals –
Rules
 Rules
– Typically
yp y contain qualitative
q information that is needed
to understand and apply the quantitative rating
algorithms
 For example,
p a primary
p y residence mayy be defined in the rule for a
homeowner policy
– To be an aid in calculating premium

Chapter 2 Rating Manuals –

Rate Pages
 Rate pages usually include
– Base rates
– Rating tables
– Fees
 Base Risk
– A specific risk pre
pre--defined by the insurer
– Represents a set of risk characteristics that are most
common or target market
common,
 Base Rate
– The rate applied to the base risk
– Not
N t th
the average ratet
– The rate for risks other than base risk is determined by
modifying the base rate by a series of multipliers or
addends or some unique mathematical expression
Chapter 2 Rating Manuals –
Rating Algorithms
 Describes in detail how to combine the various
components
p in the rules and rate pages
p g to
calculate the premium charged for any risk
 May include
– The order in which rating varaibles should be
considered
– Multiplicative or additive
– Maximum and minimum premium

Chapter 2 Rating Manuals –

Underwriting Guidelines
 Underwriting guidelines may be used to specify
– Decisions to accept,
p , decline,, or refer (to
( senior
underwriter) risks
– Company placement
– Schedule rating credits/debits
Chapter 2 Rating Manuals –
Example：
E
Example
l ：Residential
R id ti l Fire
Fi
Insurance
 Rules
– 使用性質種類定義
使用性質種類定義::
使用性質種類 說明
1.住宅 家庭手工副業，仍照住宅 費率計費。
2 公共宿舍
2.公共宿舍 一 附著建物 雖為各別 屋頂，仍適用本項費率。
一、附著建物，雖為各別 屋頂 仍適用本項費率
二、全部單身宿舍，雖有 十人以上而不各別舉炊者，不以
公共宿舍論。
三、一部分為眷屬宿舍， 一部分為單身宿舍
者 仍應按間併入計
者，仍應按間併入計
算居住家數。
3.連幢住宅 住宅區之連幢住宅十一間以上者，按本項費率計費，建築
等級以該一連接 建築之外牆為準
建築之外牆為準。倘其中
倘其中 有行號商店或
其他使用性質，仍應併入計算間數。 至各該行號商店等，
仍應 按其本身使用性質單獨計費，如各別費率低於本項
費率時，並仍按本項費率 計費。

– 公寓式樓房、非公寓式樓房、高樓大廈定義
– 自動滅火設備定義

Chapter 2 Rating Manuals –

Example：
E
Example
l ：Residential
R id ti l Fire
Fi
Insurance
 Rate pages
– Base Rate
使用性質種類
特一等 特二等 頭等 二等 三等
1.住宅
0.141 0.167 0.202 0.312 0.412
2.公共宿舍
0.267 0.320 0.384 0.591 0.784
3.連幢住宅
0.362 0.438 0.522 0.808 1.070

– 高樓加費

建築物樓層別 加費比率

15~24層樓 10%

25層樓(含)以上 15%
Chapter 2 Rating Manuals –
Example：
E
Example
l ：Residential
R id ti l Fire
Fi
Insurance
– 營業加費
同層或他層(含地下層)有營業行為者，住宅按其本身基本費率加25％
1 公寓式樓房
1.公寓式樓房 營業加費
營業加費。
同層或他層(含地下層)有營業行為者，住宅按其本身基本費率加25％
2.非公寓樓房
營業加費。
同層或他層(含地下層)有營業行為者，住宅按其本身基本費率加50％
3.高樓大廈 營業加費。

– 消防設備減費

減費項目 減費比率
火警自動警報設備 建築物備有火警自動警報設備 5%
室內消防栓 建築物備有室內消防栓設備 5%
自動滅火設備 整棟建築物之有效防護 A、各層樓之有效防護筗
20%
範圍超過建築物總面積 圍達該樓層面積100％者
90％者。 B、各層樓之有效防護範
圍達該樓層面積90％以 10%
上者
整棟建築物之有效防護 C、各層樓之有效防護筗
C 各層樓之有效防護筗
範圍未達建築物總面積 圍達該樓層面積100％者 10%
90％者。

Chapter 2 Rating Manuals –

Example：
E
Example
l ：Residential
R id ti l Fire
Fi
Insurance
– 附加費用率
 保險代理人： 44.5%
保險代理人：44.5%
 保險經紀人：
保險經紀人：44 44.5%
44 5%
保險經紀人：44.5%
 保險業務員： 44.5%
保險業務員：44.5%
直接業務((要保人直接採購或投保
 直接業務 要保人直接採購或投保))：33.3%
 其他通路：
其他通路：4444.5%
44 5%
其他通路：44.5%
 Rating algorithm
 Total Premium＝
Premium＝Amount of Insurance × Base Rate×
Rate×(1+
(1+高樓加費+營
高樓加費+
業加費--消防設備減費
業加費 消防設備減費)) /(1
/(1--附加費用率
附加費用率))。
Chapter 3
Ratemaking Data

Chapter 3 Ratemaking Data –

Internal Data
 Two types
– Risk information
 Exposures
 Premium,
 Claim counts
 Losses
– Accounting information
 Underwriting
g expense
p
 ULAE
Chapter 3 Ratemaking Data –
Risk Data
 Two databases
– Policy Database
– Claim
Cl i Database
b
 Policy Database
– Policy identifier
– Risk identifier
 One policy may have multiple risks, e.g. two cars, two locations of
insured
– Relevant dates
 Original effective date
 Original expiration date
 Date of amendment
– Premium
– Exposure
– Characteristics
[Note] If there is midterm adjustments, the characteristics
corresponding to each adjustment must be recorded

Chapter 3 Ratemaking Data –

Risk Data
[Example 3-
3-1]
– Policyy A is written on 1/1/2010,
/ / , with an annual
premium of $1,100. The home is located in Territory 1
and the insured has a $250 deductible. The policy
remains unchanged g for the full term of the policy
p y
– Policy B is written on 4/1/2010, with an annual
premium of $600. The home is located in Territory 2
and the insured has a $ $250 deductible. The policy
p y is
canceled on 12/31/2010
– Policy C is written on 7/1/2010, with an annual
premium of $1,000. The home is located in Territory 3
and the insured has a $500 deductible. On 1/1/2011,
the insured decrease the deductible to $250. The full
annual term p premium after the deductible change g is
$1,200
 Chapter 3 Ratemaking Data –
Risk Data
[Sol]

Policy Database

Policy Ori. Eff. Date Ori. Exp. Date Transaction Ded. Terr Written WP
Eff. Date Exposure
A 1/1/2010 12/31/2010 1/1/2010 250 1 1 1,100
B 4/1/2010 3/31/2011 4/1/2010 250 2 1 600
B 4/1/2010 3/31/2011 12/31/2010 250 2 -0.25 -150
C 7/1/2010 6/30/2011 7/1/2010 500 3 1 1,000
C 7/1/2010 6/30/2011 1/1/2011 500 3 -0.5 -500
C 7/1/2010 6/30/2011 1/1/2011 250 3 0.5 600

Chapter 3 Ratemaking Data –

Risk Data
 Claim Database
– Policyy identifier
– Risk identifier
– Claim identifier
– Claimant identifie
identifier
– Event identifier
– Relevant loss dates
 Report date
 Date of loss payment
 Date of reserve change
 Date of claim status change
– Claim status
 Open, Closed
 Reopen, Reclosed
Chapter 3 Ratemaking Data –
Risk Data
– Paid loss
– Case reserve
– ALAE
 ULAE is handled elsewhere
– Salvage & Subrogation
– Type of injury
– Cause of loss

Chapter 3 Ratemaking Data –

Risk Data
[Example 3
3--2]
– Policy A: A covered loss occurs on 1/10/2010. The claims is
reported to the insurance company on 1/15/2010
1/15/2010, and an initial
case reserve of $10,000 is established. An initial payment of
$1,000 is made on 3/1/2010, with a corresponding $1,000
reduction in the case reserve. A final payment of $9,000 is made
on 5/1/2010, and the claim is closed
– Policy C: A covered loss occurs on 10/1/2010, is reported on
10/15/2010, and a case reserve of $18,000 is established. The
insurer makes a payment of $2,000 on 12/15/2010, and reduces
the case reserve to $17
$17,000.
000 An additional payment of $7$7,000
000 is
made on 3/1/2011, and the case reserve is reduced to $15,000.
The claim is closed on 3/1/2012, when the insurer makes a final
payment of $15,000 and receives a $1,000 salvage recovery by
selling damaged property
– Policy C: A second loss occurs on 2/1/2011. The claim is
reported on 2/15/2011, and an initial reserve of $15,000 is set.
On 12/1/2011, the company pays a law firm $1,000 for fees
related to handling of the claim.
claim The claim is closed on that date
with no loss payments made
 Chapter 3 Ratemaking Data –
Risk Data
[Sol]

Claim Database
Policy Claim Accident Report Trans. Claim Paid Case Paid S&S
Number Date Date Date Status Loss Reserve ALAE

A 1 1/10/2010 1/15/2010 1/15/2010 Open 0 10,000 0 0

A 1 1/10/2010 1/15/2010 3/1/2010 Open 1,000 9,000 0 0

A 1 1/10/2010 1/15/2010 5/1/2010 Closed 9,000 0 0 0

C 2 10/1/2010 10/15/2010 10/15/2010 Open 0 18,000 0 0

C 2 10/1/2010 10/15/2010 12/15/2010 Open 2 000

2,000 17 000
17,000 0 0

C 2 10/1/2010 10/15/2010 3/1/2011 Open 7,000 15,000 0 0

C 2 10/1/2010 10/15/2010 3/1/2012 Closed 15,000 0 0 1,000

C 3 2/1/2011 2/15/2011 2/15/2011 Open 0 15,000 0 0

C 3 2/1/2011 2/15/2011 12/1/2011 Closed 0 0 1,000 0

Chapter 3 Ratemaking Data –

Accounting Information
 Underwriting Expenses
– Expenses
p incurred in the acquisition
q and servicing
g of
the policies
– Includes
 General expenses
 Other acquisition expenses
 Commissions and brokerage
 Taxes,, licenses,, and fees
 ULAE
Chapter 3 Ratemaking Data –
Data Aggregation
 4 ways to aggregate data for different types of
analyses
– Calendar
C l d year (CY)
– Accident year (AY)
– Policy year (PY)
– Report
R t year (RY)
 Calendar Year
– Consider all premium and loss transactions that occur
d i the
during th 12 month th calendar
l d year
– All premium and exposures are fixed at the end of the CY
– CY data is available quickly once the CY ends
– Main disadvantage
d d off CY
C is the
h mismatchh in timing
between premium and losses
 Premium earned during CY come from policies in force during that
y
year
 Losses may include payments and reserve changes on claims from
policies issued years ago

Chapter 3 Ratemaking Data –

Data Aggregation
 Accident Year
– Considers losses for accidents that have occurred during a
12--month period, regardless of the policy issued date
12
– Premium and exposures are defined as the same as CY
aggregation
– Reported losses consist of loss payments made plus case
reserves only for those claims that occurred during that
year
– Reported losses can and often do change at the end of AY
 Additional claims are reported and paid
 Reserves are changed
– Future development on those known losses needs to be
estimated
– Better match of premium and losses than CY
 Losses on accidents occurring during the year are compared to
premium earned on policies during the same year
Chapter 3 Ratemaking Data –
Data Aggregation
 Policy Year
– Considers all premium and loss transaction on policies that
were effective during a 12
12--month period
– Premium and exposures are not fixed until after the
expiration date
– Reported losses for PY consist of payments made plus
case reserves only for those claims covered by policies
effective during the year
– Reported losses can and often do change at the end of AY
 Additional claims are reported and paid
 Reserves are changed
– Best match between losses and premium
 Losses on policies effective during the year are compared with
premium earned on those same policies
– Data takes longer to develop than both CY and AY
 For a product with an annual policy term, premium are not fully
earned until 24 months after the start of the policy year

Chapter 3 Ratemaking Data –

Data Aggregation
 Report Year
– Considers losses for accidents that are reported
p during
g
a 12-
12-month period, regardless when the claim occurred
– Premium and exposures are defined as the same as CY
aggregation
– Reported losses consist of loss payments made plus
case reserves only for those claims that are reported
during that year
– Reported losses can and often do change at the end of
AY
 Additional claims are reported and paid
 Reserves are changed
Chapter 4
Exposures

Chapter 4 Exposures –
Exposures
 The basic unit that measures a policy’s exposure
to loss
 Serves as the basis for the calculation of
premium
– Base rates are typically expressed as a rate per
exposure
Chapter 4 Exposures –
Criteria for Exposure Bases
 A good exposure base should meet the following
three criteria
– Directly proportional to expected loss
– Practical
– Consider any preexisting exposure base established
within the industry
 Proportional
p to Expected
p Loss
– All else being equal, the expected loss of a policy with
two exposures should be twice the expected loss of a
similar p
policyy with one exposure
p
 Doesn’t mean that the exposure base is the only item by which
losses may systematically vary
– Should be responsive to any change in exposure to risk

Chapter 4 Exposures –
Criteria for Exposure Bases
– Example, for homeowners insurance, number of house years vs.
amount of insurance
 Practical
– Should be objective
– Should be relatively easy and inexpensive to obtain and verify
– Avoid from manipulating
p g exposure
p information
– Example, for auto insurance, annual miles vs. the number of car-
car-
year
– Example, for product liability, number of product in use vs.
number of product sold vs.
vs annual sales
 Historical Precedence
– Better exposure base may be discovered over time
– Any change in an exposure base should be carefully considered
 May lead a large premium swings for individual insureds
 Will require a change in the rating algorithm, and thus rating systems
 Historical data may need to be adjusted for analyses
Chapter 4 Exposures –
Criteria for Exposure Bases
 Exposure vs. Rating Variable
– Both relate to loss potential
– Both could be used in the rating structure
– Only one exposure base, but could have many rating
variables in the rating structure
– Exposure base has linear and continuous relationship to
expected losses; rating variable could be not

 Auto Liability  Fire Insurance

 Territory R.V.  Territory R.V.
 Age
g R.V.  Use R.V.
 Gender R.V.  Construction type R.V.
 Limits R.V.  Fire protection R.V.
 # of cars E.B.
EB  Value of house R.V.
RV
 # of house E.B.

Chapter 4 Exposures –
Aggregation
[Example 4-
4-1] Assuming the effective date = written date

Policy Effective Date Expiration Date Exposure

A 1/1/2010 12/31/2010 1
B 4/1/2010 3/31/2011 1
C 7/1/2010 6/30/2011 1
D 10/1/2010 9/30/2011 1
E 1/1/2011 12/31/2011 1
F 4/1/2011 3/31/2012 1
G 7/1/2011 6/30/2012 1
H 10/1/2011 9/30/2012 1
I 1/1/2012 12/31/2012 1
Chapter 4 Exposures –
Aggregation
 CY
– Written Exposure
 The
Th exposures arising
i i from
f policies
li i issued
i d during
d i theth marked
k d area
[Example 4-
4-1] Written Exposure for CY 2010= Policies A, B, C, D = 4
Written Exposure for CY 2011= Policies E, F, G, H = 4
 Exposures may be negative if there is any midterm adjustment, e.g. cancel
[Example 4-
4-1] If policy C is cancelled on 4/1/2011
Written Exposure for CY 2010= Policies A, B, C, D = 4
Written Exposure for CY 2011= Policies C, E, F, G, H = -0.25+4=3.75

A B C D E F G H I

1/1/2010 1/1/2011 1/1/2012 1/1/2013

Chapter 4 Exposures –
Aggregation
– Unearned Exposure
 The portion of the written exposures for which coverage has not been
provided as of a certain point in time
 Uniform assumption is usually made for most line of businesses
– The probability of a claim is evenly distributed during the whole policy term
[[Example
p 4 4--1]] Unearned Exposure
p at 12/31/2011=
/ / Policies F,, G,, H
=1 x 0.25 + 1 x 0.5 + 1 x 0.75 = 1.5
Unearned Exposure at 12/31/2010 = Policies B, C, D
+ 1 x 0.25 + 1 x 0.5 + 1 x 0.75 = 1.5

A B C D E F G H I

1/1/2010 1/1/2011 1/1/2012 1/1/2013

Chapter 4 Exposures –
Criteria for Exposure Bases
– Earned Exposure
 The portion of the written exposures for which coverage has already been
provided during the marked area
 Uniform assumption is usually made for most line of businesses
– The probability of a claim is evenly distributed during the whole policy term
 Method 1
– Directly
l determine
d the
h portion that
h has
h been
b earned
d
 Method 2
– For a single policy, Earned Exposure = Written Exposure – Unearned Exposure
– For a book of business Earned Exposure = Written Exposure – Unearned Exposure at the end
off year + U
Unearned d Exposure
E att th
the b
beginning
i i off the
th year
A B C D E F G H I

1/1/2010 1/1/2011 1/1/2012 1/1/2013

Chapter 4 Exposures –
Aggregation
[Example 4
4--1]
Method 1 : Earned Exposure for CY 2011 = Policies B ~ H
=1 x 0.25
0 25 + 1 x 0.5
0 5 + 1 x 0.75
0 75 + 1 x 1
+ 1 x 0.75 + 1 x 0.5 + 1 x 0.25
=4
Method 2 : Earned Exposure for CY 2011
= 4 + 1.5 – 1.5
=4

A B C D E F G H I

1/1/2010 1/1/2011 1/1/2012 1/1/2013

Chapter 4 Exposures –
Aggregation
– In-
In-force Exposure
 The number of exposures that are exposed to having a claim at a given
point in time
 The exposures should be annualized
[Example 44--1] In
In--force Exposure at 7/1/2011= Policies D, E, F, G = 4

A B C D E F G H I

1/1/2010 1/1/2011 1/1/2012 1/1/2013

Chapter 4 Exposures –
Aggregation
 PY
– Written Exposure
p
 The exposures with effective dates during the year
[Example 4
4--1] Written Exposure for PY 2011 at 12/31/2011= Policies E, F, G,
H=4
 Exposures may be changed over time
[Example 4
4--1] If policy G is cancelled on 4/1/2012
Written Exposure for PY 2011 at 12/31/2012 =4 – 0.25=3.75

A B C D E F G H I

1/1/2010 1/1/2011 1/1/2012 1/1/2013

Chapter 4 Exposures –
Aggregation
– Unearned Exposure
 The portion of the written exposures for which coverage has not been
provided as of a certain point in time
 Uniform assumption is usually made for most line of businesses
[Example 4
4--1] Unearned Exposure for PY 2011 at 12/31/2011
= Policies F,
F GG, H =1 x 0.25
0 25 + 1 x 0.5
0 5 + 1 x 0.75
0 75 = 1.5
15
Unearned Exposure for PY 2011 at 12/31/2012 = 0

A B C D E F G H I

1/1/2010 1/1/2011 1/1/2012 1/1/2013

Chapter 4 Exposures –
Criteria for Exposure Bases
– Earned Exposure
 The portion of the written exposures for which coverage has already been
provided
 Uniform assumption is usually made for most line of businesses
 For any given PY, the earned exposure will be equal to written exposure
after 24 months if the policy term is 1 year
 Method 1
– Directly determine the portion that has been earned
 Method 2
– Earned Exposure = Written Exposure – Unearned Exposure

A B C D E F G H I

1/1/2010 1/1/2011 1/1/2012 1/1/2013

Chapter 4 Exposures –
Aggregation
[Example 4
4--1]
Method 1 : Earned Exposure for PY 2011 at 12/31/2011= Policies E ~ H
=1
1 x 1 + 1 x 0.75 + 1 x 0.5 + 1 x 0.25 = 2.5
Earned Exposure for PY 2011 at 12/31/2012 = Policies E ~ H
=4
Method 2 : Earned Exposure for PY 2011 at 12/31/2011
=4
4 – 1.5
1 5 = 2.5
25
Earned Exposure for PY 2011 at 12/31/2012
=4-0

A B C D E F G H I

1/1/2010 1/1/2011 1/1/2012 1/1/2013

Chapter 4 Exposures –
Aggregation
[Sol] Summary of Example 4-
4-1

Policy Effective Date Expiration Date Exposure

A 1/1/2010 12/31/2010 1
B 4/1/2010 3/31/2011 1
C 7/1/2010 6/30/2011 1
D 10/1/2010 9/30/2011 1
E 1/1/2011 12/31/2011 1
F 4/1/2011 3/31/2012 1
G 7/1/2011 6/30/2012 1
H 10/1/2011 9/30/2012 1
I 1/1/2012 12/31/2012 1
Chapter 4 Exposures –
C l l i off Bl
Calculation Blocks
k off
Exposures
 Due to data availability, insurers may not
determine the earned exposures by each policy
 Instead, an approximation may be used
 1/2 method
– Assuming all policies are 1-
1-year
 Auto, Homeowner, and PA insurances may apply
 Engineering, Cargo, and Consumer Loan insurances may not
apply
– Uniform Assumptions
 The probability of a claim is evenly distributed during the whole
policy term
– Engineering and Consumer Loan insurances may not be true
 Policies are uniformly written during the year
– Equivalent to the assumption that all policies are written on 7/1 of the year
– For the insurances with large businesses such as Auto may be true
– Earned Exposure = Written Exposure last year x ½
+ Written Exposure this year x ½

Chapter 4 Exposures –
C l l i off Bl
Calculation Blocks
k off
Exposures
 1/8 method
– All assumptions
p used in 1/2
/ method applies
pp
 Policies are uniformly written during each quarter
– Equivalent to the assumption that all policies are written at mid
mid--date of each
quarter

– Earned Exposure is determined as below

1/8
3/8
5/8
7/8
7/8
5/8
3/8
1/8

1/1/2010 1/1/2011 1/1/2012 1/1/2013

/ / 0 3
Chapter 4 Exposures –
C l l i off Bl
Calculation Blocks
k off
Exposures
 1/24 method
– All assumptions
p used in 1/2
/ method applies
pp
 Policies are uniformly written during each month
– Equivalent to the assumption that all policies are written at 15th of each month

– The earned exposure percentages will be

Written Assumed Effective Earned % for Assumed Effective Earned % for
Written Month
Month Date CY 2011 Date CY 2011
Jan., 2010 1/15/2010 1/24 Jan, 2011 1/15/2011 23/24
Feb 2010
Feb, 2/15/2010 3/24 Feb 2011
Feb, 2/15/2011 21/24
Mar, 2010 3/15/2010 5/24 Mar, 2011 3/15/2011 19/24
Apr, 2010 4/15/2010 7/24 Apr, 2011 4/15/2011 17/24
May, 2010 5/15/2010 9/24 May, 2011 5/15/2011 15/24
Jun 2010
Jun, 6/15/2010 11/24 Jun 2011
Jun, 6/15/2011 13/24
Jul, 2010 7/15/2010 13/24 Jul, 2011 7/15/2011 11/24
Aug, 2010 8/15/2010 15/24 Aug, 2011 8/15/2011 9/24
Sep, 2010 9/15/2010 17/24 Sep, 2011 9/15/2011 7/24
O t 2010
Oct, 10/15/2010 19/24 O t 2011
Oct, 10/15/2011 5/24
Nov, 2010 11/15/2010 21/24 Nov, 2011 11/15/2011 3/24
Dec, 2010 12/15/2010 23/24 Dec, 2011 12/15/2011 1/24

Chapter 5
Premium
Chapter 5 Premium –
Premium Aggregation
 All concepts for exposures will be applied to
those for premium
p

Chapter 5 Premium –
Adjustment to Premium
 For loss ratio method, both premium and losses
are required
q
– LR = Losses / Earned Premium
 In the ratemaking process, historical data is
usually
ll used
d
 Chapter 1 “Ratemaking
“Ratemaking is Prospective”
Prospective”
 Three basic adjustments
– Bring the historical premium to the rate level currently
in effect
– Develop premium to ultimate levels if the premium is
still changing
– Project
j the historical premium
p to the premium
p level
expected in the future
Chapter 5 Premium –
Current Rate Level
 The rate levels during the experience period may
not be the same
– The current rate level may not be the same as historical
rate levels
 The adequacy of current rate level is reviewed,
not that of historical rate level
 The premium in the experience periods should be
adjusted to the current rate level
 Th adjusted
The dj t d premiumi is
i called
ll d on-
on-level
l l premium
i

Chapter 5 Premium –
Current Rate Level
[Example 5-
5-1]
– The written premium of a policy written in
2006 is $1,000
– The rate has been increased byy 10% in 2007
– What is the on-
on-level premium at the end of
2008
[Sol]
The current rate should be 10% higher than the
rate in 2006.
The premium would be 10% higher.
Th on-
The on-level
l l premium
i = $1,000
$1 000 × 1.1
1 1 = $1,100
$1 100
Chapter 5 Premium –
Current Rate Level
 Two methods
– Extension of exposures method
– Parallelogram method
 Extension
o of
o Exposures
po u
– Rerate each policy to restate the historical premium to the
amount that would be charged under the current rates
– Most
M t accurate t
– Difficulties
 Doing
o g Manually
a ua y iss impossible
poss b e
 Rating software is needed
 The rating structure could be changed
 The policy detail information is required

Chapter 5 Premium –
Current Rate Level
[Example 5-
5-2] Please determine the on-
on-level premium for
the policy which was effective on 3/1/2011, and had 10
class Y exposures
exposures. Assuming
– Premium = Exposure x Base Rate x Class Relativity+ Fee
– The historical rates are shown below
Overall Ave. Relativities
Rate Level Group Eff. Date Rate Change Base Rate X Y Z Policy Fee
1 Initial 900 1 00 0.60
1.00 0 60 1.10
1 10 1 000
1,000
2 7/1/2010 5.0% 950 1.00 0.60 1.10 1,000
3 1/1/2011 10.0% 1,045 1.00 0.60 1.10 1,100
4 4/1/2012 -1.0% 1,045 1.00 0.70 1.05 1,090

[Sol] On
On--Level Premium = 10 x 1,045 x 0.7 + 1,090
= 8,405
Even though the actual premium charged
=10 x 1,045 x 0.6 +1,100 = 7,370
Chapter 5 Premium –
Current Rate Level
 Parallelogram Method
– Not most accurate
accurate, but used most often in
practical
– Simple geometric relationship
– Uniform assumptions
 The probability of a claim is evenly distributed during the
whole policy term
– That is, risk is uniformly disposed over policy period
 Policies are uniformly written during the experience period
– Rate change
g historyy must be available
– Length of policy is important
 1 year, in general
 If not 1 year, same approach can be applied

Chapter 5 Premium –
Current Rate Level
[Example 5-
5-3] Continue with the Example 5- 5-2, please
on-level factor for the yyears 2011
determine the on-
and 2012
[Sol]
 Rate Indexes

2010 2011 2012 2013

1.0000 1.0500 1.1550 1.1435

7/1 1/1 4/1

+5.0% +10.0% -1.0%
Chapter 5 Premium –
Current Rate Level
 Average Earned Rate
– CY 2011
Ave ERate = (½
(½ × 0.5 × 0.5) × 1.0000 + (1 - ½ × 0.5 × 0.5 - ½ × 1 ×
1) x 1.0500 + (½
(½ × 1 × 1) x 1.1550 = 1.0963
– CY 2012
Ave ERate = (1 - ½ × 0.75 × 0.75) × 1.1550 + (½
(½ × 0.75 × 0.75) x
1.1435 = 1.1518

 O -Level
On-
On L l Factor
F t
– Current rate (at the end of year 2012) = 1.1435
– On
On--Level Factor = Current Rate / Ave ERate
F2011 = 1.1435 / 1.0963 = 1.0431
F2012 = 1.1435 / 1.1518 = 0.9928

Chapter 5 Premium –
Current Rate Level
 If the rates have been moving upward, the on-
on-
gg than one and
level factor should be bigger
decreased
 If the rates have been moving downward, the
on--level
on l l factors
f t should
h ld be
b smaller
ll than
th one and
d
increased
 Historical rate change
– Announced industrial rate change may be not
appropriate for an individual company
– Company
C
Company’ ’s rate
t change
h should
h ld be
b determined
d t i d using
i
company’’s mix of business
company
– In-
In-Force premium should be used to determin the mix
off b
business
i
Chapter 5 Premium –
Current Rate Level
[Example 5-
5-4]
(1) (2) (3)=(2)/(2)Total (4)

In-Force Premium Rate

Use Mix of Business
as of 7/1/2005 Change

A 1,000 14.3% 12.5%

B 2,000 28.6% 15.2%

Known
C ,
1,500 21.4% 6.5%

D 2,500 35.7% -8.0%

Total 7,000 4.7%

Σ(3) × (4)

Chapter 5 Premium －
Premium Development
 The ultimate amount of premium for the
experience
e pe e ce peperiod
od may
ay be u
unknown
o at tthe
e ttime
eoof
the analysis
– Incomplete year of data is used
 if policy year is used,
used the adjustment to ultimate is
needed
– Premium audits are used for the line of business
 The actual premium is determined when the actual
exposure is known at the end of the policy term, e.g.
product liability, workers compensation
 most of line of business do not utilize the p
premium audits
 Analyze historical patterns of premium to
estimate the development factor
Chapter 5 Premium －
Premium Development
 Premium Audits
– The actual premium is determined when
the actual exposure
p is known at the end
of the policy term
– E.g. product liability, workers
compensation

Chapter 5 Premium －
Premium Development
[Example 5-
5-5]
– A workers compensation
p carrier writes one policy
p y perp
month in 2011
– Estimated premium for each policy is booked at policy
inception for $500,000
– Premium on every policy develops upward by 8% at the
first audit, six months after the policy expires
[Sol]
Premium for PY 2011：
2011：
 On 12/31/2012
12/31/2012：
：6 x 500,000 x 1.08 + 6 x 500,000 = 6,240,000
 On 12/31/2013：
12/31/2013：12 x 500,000 x 1.08 = 6,480,000
Premium development factor from 24 months to 36 months
(after the start of the policy)= 6,480,000 / 6,240,000 =
1.0385
Chapter 5 Premium －
Premium Trend
 Average premium level can change over time due to
changes in the characteristics of the policies written
– Rating characteristic
 Gender: increase in the number of male policyholders may
increase the premium for auto
 Val e of home
Value home: as in
increase
ease of the value
al e of home
home, the a
average
e age
premium for homeowner policy may increase
– Deductible change
 It is important to adjust the historical premium to the
level expected during the future time period
 Need to distinguish one-
one-time event or a shift that is
expected to continue in the future
 Need to judgmentally incorporate any additional shifts
that are reasonably expected to happen in the future

Chapter 5 Premium －
Premium Trend
 Approaches to measure the premium trend
– Examine how ppremium distribution byy individual
rating
ti variable
i bl have
h shifted
hift d over time
ti
 Deductible shift
 Model Year shift
– Examine all premium shifts simultaneously
 Average premium per exposure, not per policy
 Both earned premium and written premium can be used
– Written premium is a leading indicator of trends
 Premium should be on-
on-leveled
– If not, the impact of rate change will be double counted
 Quarterly data is better if available
– Any
A reasonable
bl statistics
i i approach
h can be
b used
d
 Linear model
 Exponential model
 Any others
h
Chapter 5 Premium －
Premium Trend
[Example 5-
5-6]
=(2)/(3) =(4)/ Prior Year(4)
(1) (2) (3) (4) (5)
WP @ Current Average WP @
Q
Quarter Rate Level Written Exposures
p Current Rate Level Annual Change
g
2009Q1 323,189 453 713
2009Q2 328,325 458 717
2009Q3 333,502 463 720
2009Q4 338 722
338,722 468 724
2010Q1 343,667 472 728 2.1%
2010Q2 348,696 477 731 2.0%
2010Q3 353,027 481 734 1.9%
2010Q4 358,099 485 738 2.0%
2011Q1 361,755 488 741 1.8%
2011Q2 367,654 493 746 2.0%
2011Q3 372,305 497 749 2.1%
2011Q4 377,253 501 753 2.0%

Chapter 5 Premium –
Premium Trend
 One--Step Trending
One
– Trend factor = (1+Trend )trend period
– Trend period: the average written date in experience
period to average written date in the future
– Trended Premium = Premium ×Trend Factor
[Example 5 5--7] What is the trend period for CY 2009
EP, if the new rate will be effective 1/1/2013 for
one year?
2008 2009 2010 2011 2012 2013 2014

1/1/2009 average 4.5 years 6/30/2013 average

written date for CY written date for rate
2006 EP effective 1/1/2013
Chapter 5 Premium –
Premium Trend
 Two--Step Trending
Two
– Will be used if actuary believes the historical
premium trend is different from the future
trend
– Step 1
 Trend from the average written date in the
hi
historical
i l period
i d to the
h average written
i date
d off
the latest period
Latest
L t t Average
A WP att Current
C t Rate
R t Level
L l
Current Trend Factor 
Historical Average EP at Current Rate Level

Chapter 5 Premium –
Premium Trend
– Step 2
 Projected Trend factor = (1+Trend )trend period
 T d Period:
Trend P i d from
f the
th average written
itt ddate
t off th
the
latest period to the average written date in the future
[Example
p 5 5--8] Following
g the example
p 5 5--7

2008 2009 2010 2011 2012 2013 2014

1/1/2009 average 2.875 years 1.625 years 6/30/2013 average

11/15/2011
written date for CY written date for
average written
2006 EP date of the CY rate effective
1/1/2013
2010Q4 WP
Chapter 5 Premium –
Premium Trend
– Total Premium Trend Factor=
Current Trend Factor × Projected Trend Factor
[Example 5
5--9]
Please determine the trended CY 2009 EP at
current rate level using the information given
below if the new rate will be effective
1/1/2013
/ /
(1) CY 2009 EP at Current Rate Level 1,440,788
(2) CY 2009 Earned Exposure 1,947
(3) 2011 Q4 Ave. WP at Current Rate Level 753
(4) Selected future trend 2 0%
2.0%

Chapter 5 Premium –
Premium Trend
[Sol]
(1)/(2)= (5) CY 2009 Ave. EP at Current Rate Level 740
(3)/(5)= (6) Current Trend Factor 1.0176
[Note1.] (7) Trend Period 1.6250
(1+(4))^(7)
(1+(4))^(7)= (8) Projected Trend Factor 1 0327
1.0327
(6)×(8)= (9) Total Premium Trend Factor 1.0508
(1)×(9)= (10) Trended CY 2009 EP at Current Rate Level 1,514,036

[Note]
[Link] pperiod is from 11/15/2011 to 6/30/2013
Chapter 6
Losses and LAE

Chapter 6 Losses and LAE －

Aggregation
 C.Y.
– All lloss transactions
i occur during
d i the h 12-
12-
month calendar year
– Incurred (Reported) Losses
= Losses paid during the year
+ Loss reserves at the end of the year
- Loss reserves at the beginning of the year
[[Note]] Depending
p g on the purpose,
p p , loss
reserves may or may not include IBNR
– Incurred losses are fixed at the end of C.Y.
Chapter 6 Losses and LAE －
Aggregation
 A.Y.
– All loss transactions for claims that occur
during the year being evaluated
– Incurred (Reported) Losses
= Losses paid + loss reserves as of the
valuation date for all claims due to accident
occurred in the 12 months period
[Note] Depending on the purpose, loss
reserves mayy or mayy not include IBNR
– Incurred losses may change over time after
the end of the year

Chapter 6 Losses and LAE －

Aggregation
 P.Y.
– All loss transactions on policies that were
effective during the year being evaluated
– Incurred (Reported) Losses
= Losses paid + loss reserves as of the valuation
date for all claims arising from policies effective
d ing the 12 months period
during pe iod
[Note] Depending on the purpose, loss reserves
may or may not include IBNR
– Incurred losses may change over time after the
end of the year
Chapter 6 Losses and LAE －
Aggregation
 R.Y.
– All loss transactions for claims that are
reported during the year being evaluated
– Incurred (Reported) Losses
= Losses paid + loss reserves as of the
valuation date for all claims due to accident
occurred in the 12 months period
[Note] In this aggregation, there is not pure
IBNR
– Incurred losses may change over time after
the end of the year

Chapter 6 Losses and LAE －

Aggregation
[Example 6
6--1]
– Please determine the loss ratio for CY, AY, and PY
2009, 2010, and 2011 using the following data below

Policy # Policy Effective Date Written Premium Date of Loss Report Date Transaction Date Incremental Payment Case Reserve
A001 2009/7/1 18,000 2009/11/1 2009/11/19 2009/11/19 0 10,000
2010/2/1 1,000 9,000
2010/9/1 7,000 2,500
2011/1/15 3,000 0
A002 2009/9/1 16,800 2010/2/14 2010/2/14 2010/2/14 5,000 10,000
2010/11/1 8,000 4,000
2011/3/1 1,000 0
Chapter 6 Losses and LAE －
Aggregation
[Sol] P.Y.
@2009/12/31
PY EP Paid Losses Case Reserve_End Reported Losses Loss Ratio
2009 14 600
14,600 0 10 000
10,000 10 000
10,000 68 49%
68.49%

@2010/12/31
PY EP Paid Losses Case Reserve_End Reported
p Losses Loss Ratio
2009 34,800 21,000 6,500 27,500 79.02%
2010 0 0 0 0 #DIV/0!

@2011/12/31
PY EP Paid Losses Case Reserve_End Reported Losses Loss Ratio
2009 34,800 25,000 0 25,000 71.84%
2010 0 0 0 0 #DIV/0!
2011 0 0 0 0 #DIV/0!

Chapter 6 Losses and LAE －

Aggregation
A.Y.
@2009/12/31
AY EP Paid Losses Case Reserve_End Reported Losses Loss Ratio
2009 14 600
14,600 0 10 000
10,000 10 000
10,000 68 49%
68.49%

@2010/12/31
AY EP Paid Losses Case Reserve_End Reported
p Losses Loss Ratio
2009 14,600 8,000 2,500 10,500 71.92%
2010 20,200 13,000 4,000 17,000 84.16%

@2011/12/31
AY EP Paid Losses Case Reserve_End Reported Losses Loss Ratio
2009 14,600 11,000 0 11,000 75.34%
2010 20,200 14,000 0 14,000 69.31%
2011 0 0 0 0 #DIV/0!
 Chapter 6 Losses and LAE －
Aggregation
C.Y.
Policy A001 Policy A002

CY EP Paid Losses Change in Case Reported Losses EP Paid Losses Change in Case Reported Losses

2009 9,000
9 000 0 10 000
10,000 10 000
10,000 5 600
5,600 0 0 0
2010 9,000 8,000 -7,500 500 11,200 13,000 4,000 17,000
2011 0 3,000 -2,500 500 0 1,000 -4,000 -3,000
Total 18,000
, 11,000
, 11,000
, 16,800
, 14,000
, 14,000
,

Total
CY EP P id Losses
Paid L Ch
Change in
i Case
C R
Reported
t d Losses
L L R
Loss Ratio
ti
2009 14,600 0 10,000 10,000 68.49%
2010 20,200 21,000 -3,500 17,500 86.63%
2011 0 4,000 -6,500 -2,500 #DIV/0!
Total 34,800 25,000 25,000 71.84%

Chapter 6 Losses and LAE －

Adjustments to Losses
 Loss adjustments includes
– Remove extraordinary events
– Restate losses to the benefit and cost
levels expected in the future
– Loss trend
– Develop immature losses to ultimate
Chapter 6 Losses and LAE －
Extraordinary Losses
 Extraordinary Losses include
– Large
g individual losses
– Catastrophe losses
 Large individual losses
– If actual shock losses are included
included, indicated rates may
increase immediately after a year with shock losses
– May exclude
 The portion above some predetermined threshold
 The entire losses if the losses are excess of the threshold
– A provision will be added to reflect the expected shock
long-
losses based on a longg-term view
 Indicated rates will be underestimated if not do so

Chapter 6 Losses and LAE －

Extraordinary Losses
– Selection of the threshold
 Basic limit is one of the choice
– The rate level indication is the rate for the basic limit
– The premium used must be adjusted to the basic limit
– ILF will then be determined for the limits other than the
basic
b i limit
li it
 A point at which the losses are extraordinary and their
inclusion causes volatility in the rates
– Including as many losses as possible and minimizing the
volatility
– Examining the size of distribution and set the threshold at a
given percentile, such as the 99th percentile
– When capping the historical losses, it is better to
use the trended reported losses
 Chapter 6 Losses and LAE －
Aggregation
[Example 6
6--2]
Threshold : 1 000 000
1,000,000
=(3)-1000000×(2) =(1)-(4) =(4)/(5)
(1) (2) (3) (4) (5) (6)
AY Reported Losses Number of Excess Claims Ground-up Excess Losses Losses Excess of Threshold Non-Excess Losses Excess Ratio
1996 118,369,707
118 369 707 5 66,232,939
232 939 11,232,939
232 939 117,136,768
117 136 768 11.1%
1%
1997 117,938,146 1 1,300,000 300,000 117,638,146 0.3%
1998 119,887,865 3 3,923,023 923,023 118,964,842 0.8%
1999 118,488,983 0 0 0 118,488,983 0.0%
2000 122,329,298
, , 7 12,938,382
, , 5,938,382
, , 116,390,916
, , 5.1%
2001 120,157,205 3 3,824,311 824,311 119,332,894 0.7%
2002 123,633,881 0 0 0 123,633,881 0.0%
2003 124,854,827 1 3,000,000 2,000,000 122,854,827 1.6%
2004 125,492,840 0 0 0 125,492,840 0.0%
2005 127 430 355
127,430,355 6 13 466 986
13,466,986 7 466 986
7,466,986 119 963 369
119,963,369 6 2%
6.2%
2006 123,245,269 3 4,642,423 1,642,423 121,602,846 1.4%
2007 123,466,498 0 0 0 123,466,498 0.0%
2008 129,241,078 10 17,038,332 7,038,332 122,202,746 5.8%
2009 , ,
123,302,570 0 0 0 123,302,570
, , 0.0%
2010 123,408,837 3 4,351,805 1,351,805 122,057,032 1.1%
Total 1,841,247,359 42 70,718,201 28,718,201 1,812,529,158 1.6%

1+(6)Total = (7) Excess Loss Factor 1.016

Chapter 6 Losses and LAE －

Ch
Changes iin C
Coverage or
Benefit Levels
 Coverage or benefit levels may change due to
– Changes initiated by insurers
– Changes in law
 Any changes in coverage or benefit levels should be reflected
in the historical loss data
– May
M haveh both
b h direct
di and
d indirect
i di impacts
i
– Difficult to measure the indirect impacts
 Need to understand how the changes impact the losses
– Losses
L arising
i i from
f the
h policies
li i effective
ff i after
f a certain
i date
d
– Losses arising form the accidents occurred after a certain date
– Loss payments made after a certain date
 T
Two approaches
h
– Restate each historical claim to future benefit levels
 Most accurate, but cumbersome
– Measure the average effect
Chapter 6 Losses and LAE －
Ch
Changes iin C
Coverage or
Benefit Levels
[Example 6
6--3]
Current: 5,000
Proposed: 3,000

= Min((1),3000) = (2)/(1)
(2)/(1)-11
(1) (2) (3)
Losses Capped at
Claim # Current Level Proposed Level Effect of Change
1 1 100
1,100 1 100
1,100 0 0%
0.0%
2 2,350 2,350 0.0%
3 3,700 3,000 -18.9%
4 4,100 3,000 -26.8%
5 5 000
5,000 3 000
3,000 -40.0%
40 0%
6 5,000 3,000 -40.0%
Total 21,250 15,450 -27.3%

Chapter 6 Losses and LAE －

Ch
Changes iin C
Coverage or
Benefit Levels
[Example 66--4]
 The compensation rate is 66.7% of the worker’s pre-
pre-injury
wage
 The sate average weekly wage (SAWW) is currently $1,000
 The minimum indemnity benefit remains at 50% of the SAWW
 The
Th maximumi benefit
b fit is
i decreasing
d i from
f 100% off the
th SAWW
to 83.3% of the SAWW
 The distribution of workers according to how their wages
compare to the SAWW is as follows:
Ratio to SWAA # of Workers Total Weekly Wages
< 50% 7 3,000
5
50% - 75%
5 24 16,252
6 5
75% - 100% 27 23,950
100% - 125% 19 23,048
125% - 150% 12 16,500
150% > 11 17,250
Total 100 100,000
Chapter 6 Losses and LAE －
Ch
Changes iin C
Coverage or
Benefit Levels
[Sol]
 Workers with wages less than 75.0% of SAWW will be compensated
at the minimum level
– 75.0% = 50.0% / 66.7%
 Workers with wages more than 150.0% of SAWW will be
compensated
p at the current maximum level
– 150.0% = 100.0% / 66.7%
 Workers with wages more than 125.0% of SAWW will be
compensated at the proposed maximum level
– 125.0%
125 0% = 83.3%
83 3% / 66.7%
66 7%
 The current benefits and proposed benefits are shown below

Chapter 6 Losses and LAE －

Ch
Changes iin C
Coverage or
Benefit Levels
(1) (2) (3) (4) (5)

R ti to
Ratio t SWAA # off W
Workers
k T t l Weekly
Total W kl Wages
W C
Current
t Beefits
B fit P
Proposed
dBBenefits
fit
< 50% 7 3,000 3,500 3,500
50% - 75% 24 16,252 12,000 12,000
75% - 100% 27 23,950 15,967 15,967
100% - 125% 19 23,048 15,365 15,365
125% - 150% 12 16,500 11,000 10,000
150% > 11 17,250 11,000 9,167
Total 100 100,000 68,832 65,999

(6) Benefit Change Effect -4.1% =(5)Total / (4) Total-1

[[Note]]
1. (4)= (2)×1000×50% For (1)<75%
(3)×66.7% For Others
(2)×1000×100% For (1)>150%
2
2. (5)=
(5) (2)×1000×50% For (1)<75%
(3)×66.7% For Others
(2)×1000×83.3% For (1)>125%
Chapter 6 Losses and LAE －
Ch
Changes iin C
Coverage or
Benefit Levels
 If the change is on policies effective on and after
7/1/2010
– For AY loss data
 Proposed benefit level: 0.959
 Average Historical benefit level: 1.000 x (1-
(1- 0.125) + 0.959 x
0 125 = 0.995
0.125 0 995
 Adjustment Factor = Proposed Benefit Level / Average Historical
Benefit Level = 0.964

2010 2011

1.000 0.959

7/1 -4.1%

Chapter 6 Losses and LAE －

Ch
Changes iin C
Coverage or
Benefit Levels
– For PY loss data
 Proposed benefit level: 0.959
 Average Historical benefit level: 1.000 x (1-
(1- 0.5) + 0.959 x 0.5 =
0.979
 Adjustment Factor = Proposed Benefit Level / Average Historical
Benefit Level = 0.979

2010

1.000 0.959

7/1 -4.1%
Chapter 6 Losses and LAE －
Ch
Changes iin C
Coverage or
Benefit Levels
 If the change is on losses arising from accidents on and
after 7/1/2010
– For AY loss data
 Proposed benefit level: 0.959
 Average Historical benefit level: 1.000 x (1-
(1- 0.5) + 0.959 x 0.5 =
0.979
0 979
 Adjustment Factor = Proposed Benefit Level / Average Historical
Benefit Level = 0.979
2010

1.000 0.959

7/1 -4.1%

Chapter 6 Losses and LAE －

Ch
Changes iin C
Coverage or
Benefit Levels
– For PY loss data
 Proposed benefit level: 0.959
 Average Historical benefit level: 1.000 x 0.125 + 0.959 x (1-
(1-
0.125) = 0.964
 Adjustment Factor = Proposed Benefit Level / Average Historical
Benefit Level = 0.995

2010

0.959

1.000

7/1 -4.1%
Chapter 6 Losses and LAE －
Loss Development
 Why need to develop the losses to ultimate
– The existing loss data may not completely
develop
– Some losses haven’t
haven t been reported
– If not project to ultimate, the rate indication
will be underestimated
 Methods to project the ultimate losses
– Loss Development
p Method ( Chain Ladder
Method, Link Ratio Method)
– Bornhuetter-
Bornhuetter-Ferguson Method (BF Method)

Chapter 6 Losses and LAE －

Loss Development
 Loss Development Method
– Assumption
 Claims move from unreported to reported-
reported-and-
and- unpaid
to paid in a pattern that is sufficiently consistent
– Loss Triangle
 Data are arrayed by AY and development age
– Development Age = Valuation Date – Accident Date
Chapter 6 Losses and LAE －
Loss Development
– Loss Development Factor (LDF)
 Age-to
Age-to--Age LDF (a.k.a Link Ratio, Age LDF, or Age Factor,
or LDFx,x+1, hereafter)
– The relationship between two consecutive valuations
– The development of losses from one valuation to the
followed valuation
– LDFx,x+1 = Loss
L att x+1
1 / LLoss att x
[Note] LDFult,ult+1 = 1
 Age to--Ultimate LDF (Ultimate LDF, or LDF, or LDFx,
Age--to
hereafter))
– The development of losses from one valuation to ultimate
– LDFx = ult 1 LDF
 t,t 1
t x
– 1/LDF represents the % of ultimate that has been
developed
– Ultimate Loss
 Ultimate Loss for valuation of x = Loss at x × LDFx

Chapter 6 Losses and LAE －

Loss Development
[Example 6
6--5]
 Please determine the ultimate losses using the
data below
C
Case-Incurred
I dLLosses andd ALAE
Unit: '000
Accident As of 12/31
Year 2003 2004 2005 2006 2007 2008
2003 2,116 3,129 3,543 3,707 3,854 3,929
2004 2,316 3,527 3,993 4,182 4,339
2005 2,744 4,052 4,593 4,797
2006 3,130 4,589 5,230
2007 3,625 5,381
2008 3,920
Chapter 6 Losses and LAE －
Loss Development
[Sol]
 Step 1: Transfer to Loss Triangle

Case-Incurred
Case Incurred Losses and ALAE Triangle
Unit: '000
Accident As of 12/31
Year 12 24 36 48 60 72
2003 2 116 3,129
2,116 3 129 3,543
3 543 3,707
3 707 3,854
3 854 3,929
3 929
2004 2,316 3,527 3,993 4,182 4,339
2005 2,744 4,052 4,593 4,797
2006 3,130 4,589 5,230
2007 3 625 5,381
3,625 5 381
2008 3,920

Chapter 6 Losses and LAE －

Loss Development
 Step 2: Determine the Age-
Age-to-
to-Age LDF and
Age--to
Age to--Ultimate LDF
Age LDF & Ultimate LDF

Accident As of 12/31
Year 12-24 24-36 36-48 48-60 60-72 72-Ult
2003 1.4785 1.1326 1.0463 1.0396 1.0194 1.0000
2004 1.5230 1.1320 1.0474 1.0375
2005 1.4768 1.1336 1.0444
2006 1.4661 1.1397
2007 1.4841

Selected 1.4800 1.1350 1.0450 1.0385 1.0200 1.0000

Ult LDF 1.8594 1.2564 1.1069 1.0593 1.0200 1.0000
Chapter 6 Losses and LAE －
Loss Development
 Step 3: Determine the Ultimate Losses

Unit: '000
AY Age
A L andd ALAE Ult LDF Ulti
Loss Ultimatet
2003 72 3,929 1.0000 3,929
2004 60 4,339 1.0200 4,426
2005 48 4,797 1.0593 5,082
2006 36 5,230 1.1069 5,790
2007 24 5,381 1.2564 6,760
2008 12 3,920 1.8594 7,288
Total 27,595 33,274

Chapter 6 Losses and LAE －

Loss Trend
 Loss Trend
– The loss will change due to inflation, jury award, legal
environment, and etc.
– Including frequency and severity trend
 Most often do the two trends separately
p y
 The combine the two trends to determine the pure premium
trend
 Frequency Trend
– More strict traffic law may reduce the bodily injury (BI) frequency
– Court decision may increase the liability frequency
 Severity Trend
– Inflation is a nature cause of the severity
– Court decision may increase the severity of the liability
– Improvement of the medical equipment may increase the severity
of BI
Chapter 6 Losses and LAE －
Loss Trend
– Data used
 For more stable,, short-
short-tailed lines of business,, CY paid
p
losses are better
 For more volatile, long-
long-tailed lines of business, AY
reported losses are better
 Using
U i 12-
12-monthth rolling
lli data
d t can smooth th outt the
th effect
ff t off
seasonality
 Catastrophe losses are normally excluded
– May consider any external information
information, such as price
price--
index
– The trend in the experience period and future may
be diffe
different
ent
 Based on actuaries’
actuaries’ judgment
 Use two steps
p to reflect the different trends

Chapter 6 Losses and LAE －

Loss Trend
[Example 6-
6-6]
 Please determine the trends for frequency, severity, and pure
premium
CY (Year Ending Quarter) Earned Exposure Closed Claim Count Paid Losses
2009Q1 131,911 7,745 8,220,899
2009Q2 132,700 7,785 8,381,016
2009Q3 133 602
133,602 7 917
7,917 8 594 389
8,594,389
2009Q4 135,079 7,928 8,705,108
2010Q1 137,384 7,997 8,816,379
2010Q2 138,983 8,037 8,901,163
2010Q3 140,396 7,939 8,873,491
2010Q4 140,997 7,831 8,799,730
2011Q1 140,378 7,748 8,736,859
2011Q2 139,682 7,719 8,676,220
2011Q3 138,982 7,730 8,629,925
2011Q4 138 984
138,984 7 790
7,790 8 642 835
8,642,835
2012Q1 139,155 7,782 8,602,105
2012Q2 139,618 7,741 8,535,327
2012Q3 139,996 7,720 8,466,272
Q
2012Q4 140,141
, 7,691
, 8,412,159
, ,
2013Q1 140,754 7,735 8,513,679
2013Q2 141,534 7,769 8,614,224
2013Q3 141,800 7,755 8,702,135
2013Q4 142,986 7,778 8,761,588
 Chapter 6 Losses and LAE －
Loss Trend
[Sol]
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
CY (Year Closed Claim
Ending Quarter) Earned Exposure Count Paid Losses Frequency Annual Change Severity Annual Change Pure Premium Annual Change
2009Q1 131,911 7,745 8,220,899 0.0587 1,061.45 62.32
2009Q2 132,700 7,785 8,381,016 0.0587 1,076.56 63.16
2009Q3 133,602 7,917 8,594,389 0.0593 1,085.56 64.33
2009Q4 135,079 7,928 8,705,108 0.0587 1,098.02 64.44
2010Q1 137,384 7,997 8,816,379 0.0582 -0.9% 1,102.46 3.9% 64.17 3.0%
2010Q2 138,983 8,037 8,901,163 0.0578 -1.4% 1,107.52 2.9% 64.04 1.4%
2010Q3 140,396 7,939 8,873,491 0.0565 -4.6% 1,117.71 3.0% 63.20 -1.7%
2010Q4 140 997
140,997 7 831 8,799,730
7,831 8 799 730 0 0555
0.0555 -5.4%
5 4% 1 123 70
1,123.70 2 3%
2.3% 62 41
62.41 -3.2%
3 2%
2011Q1 140,378 7,748 8,736,859 0.0552 -5.2% 1,127.63 2.3% 62.24 -3.0%
2011Q2 139,682 7,719 8,676,220 0.0553 -4.4% 1,124.01 1.5% 62.11 -3.0%
2011Q3 138,982 7,730 8,629,925 0.0556 -1.6% 1,116.42 -0.1% 62.09 -1.8%
2011Q4 138,984 7,790 8,642,835 0.0560 0.9% 1,109.48 -1.3% 62.19 -0.4%
2012Q1 139,155 7,782 8,602,105 0.0559 1.3% 1,105.38 -2.0% 61.82 -0.7%
2012Q2 139,618 7,741 8,535,327 0.0554 0.3% 1,102.61 -1.9% 61.13 -1.6%
2012Q3 139,996 7,720 8,466,272 0.0551 -0.9% 1,096.67 -1.8% 60.48 -2.6%
2012Q4 140,141 7,691 8,412,159 0.0549 -2.1% 1,093.77 -1.4% 60.03 -3.5%
2013Q1 140 754
140,754 7 735 8,513,679
7,735 8 513 679 0 0550
0.0550 -1.7%
1 7% 1 100 67
1,100.67 -0.4%
0 4% 60 49
60.49 -2.2%
2 2%
2013Q2 141,534 7,769 8,614,224 0.0549 -1.0% 1,108.79 0.6% 60.86 -0.4%
2013Q3 141,800 7,755 8,702,135 0.0547 -0.8% 1,122.13 2.3% 61.37 1.5%
2013Q4 142,986 7,778 8,761,588 0.0544 -0.9% 1,126.46 3.0% 61.28 2.1%

Chapter 6 Losses and LAE －

Loss Trend

Exponential Fit
Number of Points Frequency Severity Pure Premium
20 -1.7% 0.5% -1.2%
16 -1.3%
1.3% -0.1%
0.1% -1.4%
1.4%
12 -0.7% -0.2% -0.9%
8 -1.3% 1.2% -0.1%
6 -0.9% 2.5% 1.6%
4 -1 4%
-1.4% 3 3%
3.3% 1 9%
1.9%
Chapter 6 Losses and LAE －
Loss Trend
 Trend Period
– Average
g accident date in experience
p period
p to average
g
accident date in the future
[Note] The loss trend period is different from the premium
trend period, but same length of the trend
[Example 6-6-7] What is the loss trend period for AY 2009
losses, if the new rate will be effective 1/1/2013 for one
year?
2008 2009 2010 2011 2012 2013 2014

4.5 years 1/1/2014 average accident

7/1/2009 average accident
date for AY2009 date for rate effective
1/1/2013

Chapter 6 Losses and LAE －

Loss Trend
 Effects of Limits on Severity Trend
– The insured limit for liability insurance will have a impact
on severity trend
– Higher layer, larger impact of trend
– Higher
g e limit,, higher
g e trend
e d
– Different effects among layers
X: unlimited loss size
L: lower
o e bou
bound d of
o layer
aye
U: upper bound of layer
T: severity increase rate
Original Loss Size Rate of Increase in Layer
X≦L Undefined
L < X ≦ U/(1+T) (X(1+T)--L)/(X-
(X(1+T) L)/(X-L)
L)--1=XT/(X
1=XT/(X--L)
U/(1+T) < X ≦ U (U--L)/(X
(U L)/(X--L)
L)--1=(U
1=(U--X)/(X
X)/(X--L)
U<X 0
Chapter 6 Losses and LAE －
Loss Trend
[Example 6
6--8]
 Please determine the trend for each layer
y if there is a trend of
10% on all claims

Distribution of Loss Amount by

y Layer
y
15,000 25,000 50,000
Original First excess of excess of excess of
Claim Loss Amount 10 000
10,000 10 000
10,000 25 000
25,000 50 000
50,000
1 5,000 5,000 0 0 0
2 20,000 10,000 10,000 0 0
3 40 000
40,000 10 000
10,000 15 000
15,000 15 000
15,000 0
4 70,000 10,000 15,000 25,000 20,000
Total 135,000 35,000 40,000 40,000 20,000

Chapter 6 Losses and LAE －

Loss Trend
[Sol]
Distribution of Loss Amount by Layer
15,000 25,000 50,000
10% of Trend First excess of excess of excess of
Claim Loss Amount 10,000 10,000 25,000 50,000
1 5,500 5,500 0 0 0
2 22 000
22,000 10 000
10,000 12 000
12,000 0 0
3 44,000 10,000 15,000 19,000 0
4 77,000 10,000 15,000 25,000 27,000
Total 148,500 35,500 42,000 44,000 27,000
Increase % 10.00% 1.43% 5.00% 10.00% 35.00%
 Chapter 6 Losses and LAE －
L
Loss D
Development
l vs. L
Loss
Trend
 Overlap Fallacy
– A double counting of loss development and trend
– Loss development is to reflect the development
from occurrence to settlement
– Trend is to reflect the change from the historical
occurrence to future occurrence

Chapter 6 Losses and LAE －

Loss Development vs. Loss Trend

Trend
Loss Development

Occurrence Settlement

Occurrence Settlement

Experience Period Exposure Period

1/1/2010 1/1/2011 1/1/2012 1/1/2013 1/1/2014 1/1/2015

Chapter 6 Losses and LAE －
Loss Adjustment Expense
 The expense occurred by an insurer to close a claim
 Including
– ALAE (Allocated Loss Adjustment Expense)
 Claim-related expenses that are directly attributable to a specific claim
Claim-
 Example fees associated with outside legal counsel hired to defend a
Example,
claim can be directly assigned to a specific claim
– ULAE (Unallocated Loss Adjustment Expense)
 Claim-related expenses
Claim- p that cannot be directlyy assigned
g to a specific
p
claim
 Example, salaries of claims department

 Definition has been changed

g since 1998
– ALAE → DCC (Defense and Cost Containment)
– ULAE → A&O (Adjusting and Other)

Chapter 6 Losses and LAE －

Loss Adjustment Expense
 Treatment of LAE
– ALAE
 In general, ALAE is included in losses, but not
always
 Usually vary by the dollar amount of each claim
 Incorporated with a % of reported losses

– ULAE
 In g
general,, ULAE is usuallyy included in expense
p
 Usually vary by the number of claims reported
– Study how adjusters spend their time
Chapter 6 Losses and LAE －
Loss Adjustment Expense
 Simple Approach to incorporate ULAE
– Assumption
 ULAE track with loss plus ALAE dollars
consistently over time
– Determine
D t i a paid-
paid
id-paid
id ratio
ti
 Paid
Paid--Paid ratio
= CY p paid ULAE / CY p paid Losses + ALAE
– Determine the ULAE
 Paid
Paid--Paid ratio × Trended Ultimate Losses

Chapter 6 Losses and LAE －

Loss Adjustment Expense
[Example 6-
6-9]
=(2)/(1)
(1) (2) (3)
CY Paid Losses and ALAE Paid ULAE ULAE Ratio
2008 913,467 144026 15.8%
2009 1,068,918 154170 14.4%
2010 1 234 240
1,234,240 185968 15 1%
15.1%
Total 3,216,625 484,164 15.1%

ULAE Factor = 1.151

Chapter 7
Other Expenses and Profit
f

Chapter 7 Other Expenses and

Profit－
Profit－Underwriting Expenses
 Commission and Brokerage
– Amount paid to insurance agents or brokers as
compensation for generating business
– Typically, paid as a percentage of premium written
 Other Acquisition
– Expenses other than commissions and brokerage expense
paid to acquire business
– Example,
Example advertisements
 General Expense
– The remaining expenses associated with the insurance
operations and any other miscellaneous costs
 Taxes, Licenses, and Fees
– All taxes and miscellaneous fees paid, but excluding
i
income taxes
t
Chapter 7 Other Expenses and
Profit－
Profit－Underwriting Expenses
 Underwriting expenses can be divided into two
types
yp
– Fixed Expense
 The expenses are the same for each risk or policy, regardless of the
size of the premium
 Th expenses is
The i a constant
t td dollar
ll amountt for
f eachh risk
i k or policy,
li
regardless of the size of the premium
 Examples
– Overhead costs
– Electricity
El t i it
– Rental
– Variable Expense
 The expenses vary directly with premium
 The expense is a constant percentage of the premium
 Examples
– Premium taxes
– Commissions

Chapter 7 Other Expenses and

Profit－
Profit－Expense Ratio
 Expense Ratio = Expenses / Premiums
– The premiums can be either the written premium or the
earned premium
– Written Premium
 Used when expenses are incurred at the inception of the policy
 Variable expenses usually use the written premium
– Earned Premium
 Used when expenses are incurred throughout the policy
 It reflects
fl t ththe gradual
d l paymentt off expenses th
thatt can b
be proportional
ti l
to the earning of premium over the policy term
 Fixed expenses usually use the earned premium
– Impacts if the choice of premiums is not appropriate
 The impact will be relatively little for a stable company,
 The expense ratio will be overestimated if the earned premium is
used to measure the variable expense for a growing company
– Earned premium is less than written premium
– The commissions and brokerage are high
Chapter 7 Other Expenses and
Profit－
Profit－Simple Example
 Fundament Equation
– Premium = Losses + LAE +UW Expenses + UW Profit
 P ≡ Premium
 L ≡ Losses QT ×P
 EL ≡ LAE
 EF ≡ Fixed Expense
V×P
 V ≡ Variable Expense Ratio
 QT ≡ UW Profit EF
– P = L + EL + (EF + V × P) + QT × P
– P = (L + EL + EF) / (1 - V – QT )
L + EL

Chapter 7 Other Expenses and

Profit－
Profit－Simple Example
 Simple Example
– L + EL = 180
– EF = 20
QT ×P=12.5
– V = 15%
– QT = 5% V × P=37.5

EF=20
P = (180+20) / (1
(1--15%
15%--5%) = 250
– V = 250 × 15% = 37.5
– QT = 250 × 5% = 12.5 L + EL=180
Chapter 7 Other Expenses and
P
Profit
Profit－
fit－Methods
M th d tto incorporate
i t
expense provisions
 Approaches to derive expense provisions for
ratemaking
– All Variable Expense Method
– Premium-
Premium-based Projection Method
– Exposure/Policy-
Exposure/Policy-based Projection Method
 All Variable Expense Method
– Traditional approach
 But still widely used
– Does not differentiate between fixed and variable
underwriting expenses
– Treats all expenses as variable
– Assumes that expenses ratios during the projected period
will be consistent with the historical expenses ratios

Chapter 7 Other Expenses and

P
Profit
Profit－
fit－Methods
M th d tto incorporate
i t
expense provisions
[Example 7-
7-1]
Items 2013 2014 2015 Total Selected
(1) Written Premium 1,532,091 1,981,109 2,801,416 6,314,616
(2) Earned Premium 1,440,166 1,756,600 2,391,263 5,588,028
(3) Commissions and Brokerage 183,851 198,111 294,149 676,111
(4) Other Acqusition 72,009 104,707 142,072 318,788
(5) General Expense 115,213 151,068 215,214 481,494
(6) Taxes, Licenses, and Fees 30,642 39,622 56,028 126,292
(7) Commissions and Brokerage Ratio 12.0% 10.0% 10.5% 10.7% 10.5%
(8) Other Acquisition Ratio 4 7%
4.7% 5 3%
5.3% 5 1%
5.1% 5 0%
5.0% 5 0%
5.0%
(9) General Expense Ratio 8.0% 8.6% 9.0% 8.6% 8.6%
(10) Taxes, Licenses, and Fees Ratio 2.0% 2.0% 2.0% 2.0% 2.0%
(11) Total Underwriting Expense Ratio 26.7% 25.9% 26.6% 26.4% 26.2%
– Potential
P t ti l di
distortions
t ti
 Underestimate the premium need for a small risk
 Overestimate the premium need for a large risk
[Note] Companies that usese this approach
app oach may
ma implement a premium
p emi m
discount structure that reduces the expenses loading
Chapter 7 Other Expenses and
P
Profit
Profit－
fit－Methods
M th d tto incorporate
i t
expense provisions
[Example 7-
7-2]

=((1)+(3))/(1-(2)) =(3)/250 =(2)+(5) =(1)/(1-(6))

(1) (2) (3) (4) (5) (6) (7)
Loss Variable Fixed Correct Fixed Total All Variable
Cost Expense Expense Premium Expense Ratio Expense Ratio Expense Method
Small Size 135 20.0% 20 194 8.0% 28.0% 188
Medium 180 20.0% 20 250 8.0% 28.0% 250
Large
g 225 20.0% 20 306 8.0% 28.0% 313

Chapter 7 Other Expenses and

P
Profit
Profit－
fit－Methods
M th d tto incorporate
i t
expense provisions
 Premium--based Projection Method
Premium
– Schofield
Schofield, D.,
D “Going
Going From a Pure Premium to a Rate
Rate,” CAS
Study Note, 1998
 [Link]
– Recognizes the fixed and variable expenses separately
– Like All Variable Expense Method, assumes that expenses
ratios during the projected period will be consistent with
the historical expenses ratios
Chapter 7 Other Expenses and
P
Profit
Profit－
fit－Methods
M th d tto incorporate
i t
expense provisions
[Example 7-
7-3]
Expense Ratio
Itmes Total Ratio Fixed Portion Fixed Variable
(1) Commissions and Brokerage Ratio 10.5% 0.0% 0.0% 10.5%
(2) Other Acquisition Ratio 5.0% 0.0% 0.0% 5.0%
(3) General Expense Ratio 8 6%
8.6% 75 0% 6.5%
75.0% 6 5% 2 2%
2.2%
(4) Taxes, Licenses, and Fees Ratio 2.0% 0.0% 0.0% 2.0%
(5) Total Underwriting Expense Ratio 26.2% 6.5% 19.7%

(6) Projected
P j d Average
A Premium
P i per Exposure
E 200
(7) Projected Fixed Expense per Exposure 13

– Potential
P t ti l di
distortions
t ti
 Recent rate changes can impact the historical expense ratio
 Significant differences in average premium can distort the fixed
expenses

Chapter 7 Other Expenses and

P
Profit
Profit－
fit－Methods
M th d tto incorporate
i t
expense provisions
 Exposure/Policy--Based Projection Method
Exposure/Policy
– Diana Childs and Ross Currie,
Currie “Expense
Expense Allocation in
Insurance Ratemaking,” 1980
 [Link]
– Variable expenses are treated the same way as the
Premium--Based Projection Method
Premium
– Historical fixed expenses are divided by historical
exposures or policy count rather than premium
Chapter 7 Other Expenses and
P
Profit
Profit－
fit－Methods
M th d tto incorporate
i t
expense provisions
[Example 7-
7-4]
3-Yr Total 3-Yr Expense
Items Expense Fixed Portion Fixed Variable
(1) Commissions and Brokerage Ratio 676,111 0.0% 0 676,111
(2) Other Acquisition Ratio 318,788 0.0% 0 318,788
(3) General Expense Ratio 481,494 75.0% 361,121 120,374
(4) Taxes, Licenses, and Fees Ratio 126,292 0.0% 0 126,292
(5) Total Underwriting Expense Ratio 1,602,685 361,121 1,241,564

(6) 33-Yr
Yr Total Earned Exposure 85,000
85 000
(7) Fixed Expense per Exposure 4.25
(8) 3-Yr Total Written Premium 6,314,616
(9) Variable Expense Ratio 19.7%

Chapter 7 Other Expenses and

Profit－
Profit－Trending Expenses
 No need to trend the variable expense
– Since the variable expenses will automatically change as the
premium changes
 Fixed expense should be trended
– The fixed expense may increase over time due to inflationary
pressures
– Trend from the average date the policies are earned in the
historical period to the average earned date in the projection
period
[Example 7-
7-5]
2011 2012 2013 2014

2.5 years 1/1/2014 average earned

7/1/2011 average earned date for rate effective
date for 2011 1/1/2013
Chapter 7 Other Expenses and
P
Profit
Profit－
fit－Underwriting
U d iti Profit
P fit
Provision
 Total Profit = UW Profit + Investment Income
 Underwriting Profit
– UW Profit = Premium – Losses – LAE – UW Expenses
 Investment Income
– Including
I l di
 Investment income on capital
 Investment income on policyholder-
policyholder-supplied funds
– Investment income on capital
 Capital funds belong to the owners of the insurance company
 There is substantial disagreement as to whether this should be
returned to insureds
– Investment income on policyholder
policyholder--supplied funds
 Policyholder-supplied funds includes unearned premium reserves and
Policyholder-
loss reserves
 Fixed expenses usually use the earned premium

Chapter 7 Other Expenses and

Profit－
Profit－Permissible Loss Ratio
 Variable Permissible Loss Ratio (VPLR)
– VPLR = 1 – V – QT
– The percentage of each premium dollar that is intended to
pay for the projected loss and LAE and projected fixed
expense
 Permissible Loss Ratio (PLR)
– PLR = 1 – F - V – QT
– The percentage of each premium dollar that is intended to
pay for the projected loss and LAE
Chapter 8
Overall Indication

Ch t 8 Overall
Chapter O ll Indication
I di ti
 Two basic approaches
– Pure Premium Method
– Loss Ratio Method
 Pure Premium Method vs. Loss Ratio Method
– Mathematically equivalent
 the results are the same if same data and consistent assumptions
are used
– Differences
Pure Premium Method Loss Ratio Method
Based on exposure Based on premium
Existing rate is not required Existing rate is required
O L l premium
On-Level pr mi m iis nott required
r ir d O L l premium
On-Level pr mi m iis rrequired
ir d
Produces indicated rates Produces indicated rate changes
Chapter 8 Overall Indication
Indication－
－
Pure Premium Method
 Indicated Average Rate Per Exposure
– Rate = Pure Premium + LAE + UW Expenses + Target Profit %
 R ≡ Average Rate per Exposure
 PP ≡ Pure Premium
 EL ≡ LAE
 EF ≡ Fixed Expense QT ×P
 V ≡ Variable Expense Ratio
 QT ≡ Target
g Profit %
V×P
– R = PP + EL + (EF + V × P) + QT × R EF
– R = (PP + EL + EF) / (1 - V – QT )

PP + EL

Chapter 8 Overall Indication –

Loss Ratio Method
 Determine the rate change %
– Compare
p the projected
p j loss ratio (or
( experience
p loss ratio,,
ELR) to the permissible loss ratio (PLR, or target loss
Ratio)
 If the projected loss ratio is higher than the permissible loss
ratio,
ratio the current rate is too low and should be increased
 Indicated Change Factor (A) = ELR / PLR
[Note] The numerator can include ALAE, ULAE, or/and fixed
expense, as long as the denominator uses the same
definition
 Determine the rate
– R = A × R0
Where R : indicated rate
R0 : current rate
Chapter 8 Overall Indication –
Loss Ratio Method
[Example 8
8--1] Please determine the rate change % using the
information below.
– Projectedd loss
l and
d LAE ratio = 65%
– Projected fixed expense ratio = 6.5%
– Variable expense ratio = 25%
– Target
T t profit
fit % = 10%
[Sol 1]
PLR = 1 - 6.5% - 25% - 10% = 58.5%
A = ELR / PLR = 65% / 58.5%
8 = 111.1%
Rate change % = A – 1 = 11.1%
[Sol 2]
VPLR = 1 - 25% - 10% = 65%
A = (ELR + Fixed expense ratio) / VPLR
= (65% + 6.5%)
6 5%) / 65% = 110%
0%
Rate change % = 110% - 1 = 10%
 Credibility

鍾孟芳
Kris Chung, FCAS
November 2, 2012

Outlines

 Introduction
 Methods for Calculation of Credibility
 Desirable Qualities of a Complement
p of Credibility
y
 Methods For Developing Complements of
Credibility - First Dollar Ratemaking
 Methods For Developing Complements of
Credibility - Excess Ratemaking

2
 Introduction

General Concept
 Set rate levels so that rates are “adequate, reasonable, and not
unfairly
f i l di
discriminatory”
i i ”
 Adequate: Not too low
 Reasonable: Not too high
 Not unfairly discriminatory: Allocation of overall rate to individuals
is based on cost justification

 At various steps in the ratemaking process, the concept of credibility

is introduced

 The credibility of data is commonly denoted by the letter “Z”

4
 Definitions of Credibility

 Common vernacular (Webster):

 “Credibility” = the quality of being credible
 “Credible”
Credible = believable
 So, credibility is “the quality of being believable”

 In actuarial circles:
 Credibility is “a measure of the predictive value in a given
application that the actuary attaches to a particular body of
experience”
-- L.H. Longley-Cook

 Refers to the degree of believability; a relative concept

Why Do We Need Credibility?

 P
Property / casualty
l insurance
i costs (l
(losses)) are iinherently
h l stochastic
h i

 Losses are fortuitous events

 Any given insured may or may not have a claim in a given year

 The size of the claim can vary significantly

 Data can be viewed as an observation of a result

 Only one estimate of the “truth”
truth of the probability of having a
claim and the distribution of sizes of claims

 S how
So h much
h can we b
believe
li our d
data?
t ?

6
History of Credibility in Ratemaking
 The CAS was founded in 1914, in p
part to help
p make rates for a new line
of insurance – workers compensation – and credibility was born out the
problem of how to blend new experience with initial pricing

 Early pioneers:
 Mowbray (1914) -- how many trials/results need to be observed
before I can believe my data?
 Albert Whitney (1918) -- focus was on combining existing estimates
and new data to derive new estimates:
New Rate = Credibility*Observed Data + (1-Credibility)*Old Rate

 Perryman (1932) -- how

P h credible
dibl iis my d
data
t if I h
have lless th
than
required for full credibility?

Methods for
C l l ti off C
Calculation Credibility
dibilit

8
Methods of Incorporating Credibility
Limited Limit the effect that random
Fluctuation fluctuations in the data can have
(有限變動法) on an estimate

“Classical credibility”

Greatest Make estimation errors as

y
Accuracy small as possible
(最大精確法)
又稱 Bühlmann Approach “Least Squares Credibility”
“Empirical Bayesian Credibility”
Bühlmann Credibility
Bühlmann-Straub Credibility

Limited Fluctuation Credibility Description

 “A dependable
p [[estimate]] is one for which the
probability is high, that it does not differ from the [truth]
by more than an arbitrary limit.”
-- Mowbray (1916)

 Alternatively, the credibility Z, of an estimate T, is

defined by the probability P, that it within a tolerance k%,
off the
th true
t value
l

 Goal: Determine how much data one needs before

assigning it with full credibility (Z = 1)

10
 Limited Fluctuation - Variables
 T: Estimate the data that we want to test for credibility
y (e.g.
( g loss ratio))
 Z: Credibility, which is between 0 and 1
 k: Tolerance for error (e.g. the observation is within k = 5% of the
mean)
 P: Probability that the observation is within k% of the mean
 using the standard Normal distribution (e.g. P = 90% zp = 1.645)

11

Limited Fluctuation Formula - Derivation

 New estimate =
(Credibility)*(Observed Experience) + (1-Credibility)*(Prior Estimate)

E2 = Z
Z*T
T + (1
(1-Z)*E1
Z) E1

Add and
subtract E2 = Z*T
Z T+Z
Z*E[T]
E[T] – Z
Z*E[T]
E[T] + (1-Z)*E1
(1-Z) E1
Z*E[T]

regroup E2 = (1
(1-Z)*E1
Z)*E1 + Z*E[T] + Z*(T–E[T])
Z*(T E[T])

Stability Truth Random Error

12
 Limited Fluctuation Credibility Formula for Z
 The observed experience is considered fully credible when the probability (p)
is high that the observed experience will not differ significantly from the
expected
t d experience
i b
by more th
than some arbitrary
bit amountt (k)
(k).

P {Z(T E[T]) < kE[T]} = P

Pr{Z(T-E[T])

-or-
or Pr{T < E[T] + kE[T]/Z} = P

E[T] + kE[T]/Z = E[T] + zpVar[T]1/2

(assuming T~Normally)

-so- kE[T]/Z [ ]1/2

[ ] = zpVar[T]

 Z = kE[T]/(zpVar[T]1/2)

13

Limited Fluctuation Formula for Z (continued)

 If we assume
 we are measuring g an insurance p process that has Poisson frequency,
q y, and
 Severity is constant or severity doesn’t matter

 Then E[T] = number of claims (N), and E[T] = Var[T], so:

Z = kE[T]/zpVar[T]1/2 becomes:

Z = kE[T]/zpE[T]1/2 = kE[T]1/2 /zp = kN1/2 /zp

 Solving for N (# of claims for full credibility, i.e., Z=1):

N = ((zp/k))2

■ An actuary may regard the loss experience as fully credible if there is a 90% probability
that the observed experience is within 5% of its expected value. This is equivalent to saying
there is a 95% probability that the observed losses are no more than 5% above the mean.
In the Standard Normal table, the 95th percentile is 1.645 standard deviations above the
mean; therefore, the expected number of claims needed for full credibility is:

14
Limited Fluctuation
Fluctuation– Standards for Full Credibility
 Claim counts required for full credibility based on the
previous derivation:

Number of
k
Claims

P 2.5% 5.0% 7.5% 10%

90% 4,330 1,082 481 271

95% 6,147 1,537 683 384

99% 10,609 2,652 1,179 663

15

Limited Fluctuation
Fluctuation– Standards for Full Credibility
 Exposure counts required for full credibility based on the
previous
i d
derivation:
i ti

Number
N b off Number
N b off
Claims for Projected Exposures
P K
Full Frequency for Full
Credibility Credibility
90% 5% 1,082 10% 10,820

95% 5% 1,537 10% 15,370

99% 5% 2,652 10% 26,520

16
 Limited Fluctuation Credibility Formula 2
 Relaxing
g the assumption
p that severity
y doesn’t matter,,
 Let “data” = T = aggregate losses = frequency x severity = N x S
 then E[T] = E[N]E[S]
 and Var[T] = E[N]Var[S] + E[S]2Var[N]

 Plugging these values into the formula

Z = kE[T]/zpVar[T]1/2

and solving for N (@ Z=1):

N = (zp/k)2{Var[N]/E[N] + Var[S]/E[S]2}

17

Limited Fluctuation Credibility Formula 2

N = (zp/k)2{Var[N]/E[N]+ Var[S]/E[S]2}

Think of this as an adjustment factor to the full

This term is just credibility standard that accounts for relaxing the
the full assumptions about the data.
credibility
standard derived
earlier
The term on the left is derived The term on the right is the
from the claim frequency square of the c.v. of the severity
distribution and tends to be distribution and can be
close to 1 (it is exactly 1 for significant
significant.
Poisson).
18
 Limited Fluctuation – Partial Credibility

 Given a full credibility standard based on a number of claims Nfull,

what is the partial credibility of data based on a number of claims N
th t is
that i lless th
than Nfull?
 Z = (N / Nfull)1/2
 Square root rule
 Designed such that the partial credibility Z will be inversely
proportional to the standard deviation of the partially credible
d t
data

19

Limited Fluctuation – Increasing Credibility

 Under the square root rule, credibility Z can be increased by

 Getting more data (increasing N)

 Accepting a greater margin of error (increasing k)

 Conceding
g to smaller P = being
g less certain ((decreasing
g zp)

 Based on the formula

Z = (N/ Nfull)1/2
Z = [N/(zp/k)2]1/2
Z = k*N1/2/z
/ p
20
Limited Fluctuation – Advantages and Weaknesses
 The strength
g of limited fluctuation credibility
y
 It is the most commonly used and is therefore generally accepted
 The data required for this approach is readily available
 The computations are very straightforward

 Weakness of this approach

 The derivation involves making several simplifying assumptions that
may not be true in practice (e.g., no variation in the size of losses)
 Typical
T i l use off th the fformula
l bbased
d on th
the P
Poisson
i model
d l iis iinappropriate
i t
for most applications
 Partial credibility formula – the square root rule – only holds for a normal
approximation off the underlying distribution off the data but insurance
data tends to be skewed

21

Limited Fluctuation – Example 1

 Calculate the loss ratios, given that the expected loss
ratio
ti iis 75%
75%, and
d using
i the
th square roott rule
l

Loss
Ratio Claims
2002 67% 535 Example: 78.6% =
2003 77% 616 81%(0 60) + 75%(1-0
81%(0.60) 75%(1 0.60)
60)
2004 79% 634
2005 77% 615
2006 86% 686 Credibility at: Weighted Indicated
1,082 5,410 Loss Ratio Rate Change
3 year 81% 1,935 100% 60% 78.6% 4.8%
5 year 77% 3,086 100% 75% 76.5% 2.0%

Example: 1.020 =
76.5%/75% -1
1

22
 Limited Fluctuation – Example 2
Given a current territory factor of 1.08, determine the indicated territory
factor with 5 years of data
data. Use the square root rule and the limited
fluctuation formula for pure premium. Assume a Poisson frequency
distribution and severity coefficient of variation of 1.5.

23

Limited Fluctuation – Example 2

N = ((zp/k))2{{Var[N]/E[N] [ ] [ ] 2}
[ ] [ ] + Var[S]/E[S]
 Remember, with a Poisson distribution, Var(N) = E(N), so the second term is 1.
 Th third
The thi d tterm iis th
the square off th
the coefficient
ffi i t off variation,
i ti which
hi h iis 1 52.
1.5
 Now we just need to select the confidence levels.

 N = (zp/k)2{Var[N]/E[N] + Var[S]/E[S]2}
If we want to be within 5% of the true value 90% of the time, the value for
(z / k)2 is 1 082 Plugging into the formula:

N(claims) = 1,082 * ( 1 + 1.52 ) = 3,516.5

 Assuming the 5-year statewide frequency is 0.2:

N(exposures) = 3,516.5 / 0.2 = 17,582.5

24
Limited Fluctuation – Example 2
 To show the impact of our selection of an exposure standard instead
of a claims standard.
standard

25

Limited
ted Fluctuation
uctuat o – Example
a pe2
 Determine what the indicated territorial factor, assuming
g 15% for fixed
expenses.

 The final indicated territorial factor is (156% / 80%)*0.85 + 0.15 = 1.81

26
Greatest
G eatest Accuracy
ccu acy C
Credibility
ed b ty
Which data exhibits more credibility?

27

Greatest Accuracy Credibility

Which data exhibits more credibility?

S1

E
S2

C
D

28
 Greatest Accuracy Credibility
Average “within” class variance =
“Expected Value of Process Variance” =
Class loss costs p
per exposure...
p or EVPV; denoted s2

0 A B C D 
E
Higher credibility:
less variance within,
within
more variance between
Variance between the means =
yp
“Variance of Hypothetical Means”
or VHM; denoted t2

A B E C D 
0 Lower credibility:
y
more variance within,
less variance between
29

Bühlmann Credibility

Description
p

 Bühlmann credibility is commonly referred to as least

squares credibility since the goal of this approach is
to minimize the square of the error between the
estimate
ti t andd the
th true
t expected
t d value
l off the
th quantity
tit
being estimated.

30
 Bühlmann Credibility - Derivation
 Suppose you have two independent estimates of a quantity, x and y, with
squared
d errors off u and
d v respectively
ti l

 We wish to weight
g the two estimates together
g as our estimator of the
quantity:
a = zx + (1-z)y

 The squared error of a is

w = z2 u + (1-z)2v

 Find Z that minimizes the squared error of “a” a – take the derivative of w
with respect to z, set it equal to 0, and solve for z:
dw/dz = 2zu + 2(z-1)v = 0
Z = u/(u+v)
/( )
31

Bühlmann Credibility - Derivation

Using the formula that establishes that the least squares value for
Z is proportional to the reciprocal of expected squared errors:

Z = (n/s2)/(n/s2 + 1/ t2) =

( s2/t2)
= n/(n+

= n/(n+k)
k = s2/t2

32
Bühlmann Credibility Formula
 New estimate =
(Credibility)*(Observed Experience) + (1-Credibility)*(Prior Mean)

 Credibility =

N: the number of observations

K: expected value of the process variance devided by the variance of
the hypothetical means
* Ratio K can be described more simply as the ratio of average risk
variance and the variance between risks

 Credibility, can be increased by:

 Increasing n = get more data
 decreasing s2 = less variance within classes, e.g., refine data
categories
 increase t2 = more variance between classes

33

Sensitivity of credibility

34
 Comparison of Credibility Values

 For a relatively small number of observations, the Buhlmann credibility

estimate is closest to the classical credibility estimate when K = 5,000
 As the number of observations gets larger
larger, the Buhlmann credibility
estimate is closest to the classical credibility estimate when K = 1,500
35

Comparison of Credibility Values

 Many practitioners using classical credibility make simplifying

assumptions—for
assumptions for example, they ignore the variation in the size of
losses and assume that the risks in the subject experience are
homogeneous.

 If these same assumptions are made with least squares credibility,

then VHM = 0 (this is because all of the exposures have exactly the
same claim distribution). When VHM = 0, then Z = 0 and no
credibility is assigned to the observed experience.

36
Limited Fluctuation – Partial Credibility

100%
90%
80%
dibility

70%
60%
Cred

50% (N/1082)^.5
40% N/N+191
30%
20%
00

00

00

00

00

00
10

30

50

70

90

110
Number of Claims N

37

Assumptions for Bühlmann Credibility

 (1-Z) is applied to the prior mean

 The risk parameters and risk process do not shift over time

 The expected value of process variance of the sum of N

observations increases with N

 The variance of the hypothetical

yp means of the sum of N
observations increases with N

38
 Bühlmann Credibility - Example

 The observed pure premium is 200 based on 21 observations

 Th expected
The t d value
l off th
the process variance
i iis 2
2.00
00
 The variance of hypothetical means is 0.50
 Th prior
The i mean iis 225

K = 2.00 / 0.50 = 4.00

Z = 21 / (21+4.00) = 0.84
Credibility-weighted
Credibility weighted Pure Premium = 0.84x200
0 84x200+0
0.16x225
16x225 = 204

39

Bühlmann Credibility – Advantages and Weaknesses

 Main advantages
▫Is generally accepted
▫It is based on relative variances or volatility of the data
▫There is no such thing as full credibility

 Main disadvantage of this approach

▫ The difficulty of the determination of the expected value of the
process variance and the variance of the hypothetical means
▫ The Credibility Parameter K, is a property of the entire set of data.
So, for example, if a data set has a small, volatile class and a large,
stable class
class, the credibility parameter of the two classes would be
the same.

40
 Desirable Qualities of a
C
Complement
l t off Credibility
C dibilit

41

According to ASOP 25

The actuary should use care in selecting the related

experience that is to be blended with the subject
experience. Such related experience should have
frequency, severity, or other determinable characteristics
that may reasonably be expected to be similar to the
subject experience. If the proposed related experience
does not or cannot be adjusted to meet such criteria, it
should not be used. The actuary should apply credibility
procedures that appropriately reflect the characteristics of
both the subject experience and the related experience

42
 Fundamental Principles –
What should the actuary consider

 Practical issues
Readily available and easy to calculate
 Competitive market issues
Unbiased and accurate
 Regulatory issues
L i l relationship
Logical l ti hi tto b
base statistic
t ti ti
 Statistical issues
Need to consider all types of error that make up
prediction error

43

Desirable qualities for a complement of Credibility

 Accurate
 U bi
Unbiased d
 Statistically independent from the base statistic
 A il bl
Available
 Easy to compute
 Logical relationship to base statistic

44
 Methods For Developing
p
Complements of Credibility
y
- First Dollar Ratemaking

45

Methods for developing complements for

first dollar ratemaking

 Loss costs of a larger group that includes the group being

rated
 Loss costs of a larger related group
 Rate change from the larger group applied to present rates
 Harwayne’s method
 Trended present rates
 Competitor’s rates

46
Loss costs of a larger group that includes
the group being rated
 Complements of credibility for Group A, Class 1

47

Loss costs of a larger group that includes

the group being rated
Evaluation
 The
Th complement l t is
i bbased
d on a greater
t volume
l off d
data
t
so it’s likely to have a lower process variance than the
subject experience
 If the complement excludes the subject experience, then
it’ss likely to be independent
it
 If the subject experience is included, it should not
dominate the groupg p
 It’s available and easy to calculate

48
 Loss costs of a larger related group
Method Description
 Use the loss costs of a separate but similar, large group
 For example, a homeowners insurer may use the contents loss experience
from the owners forms to supplement the contents experience for the
condos form
Evaluation
 It is generally biased though the magnitude and direction of bias are
unknown
 If the actuary can adjust the related experience to match the exposure to
loss in the subject experience, the bias can be reduced
 In the example mentioned above, the actuary needs to consider how the
exposure tot loss
l for
f condos
d differs
diff ffrom owneddhhomes andd adjust
dj t the
th
experience accordingly
 Since the complement
p does not contain the subject
j experience,
p , this lack of
dependence may make it a better choice than the first method described
49

Rate Change from the Larger Group

Applied to Present Rates
Method Description
 Usingg the rate change
g indicated for a larger
g g group
p
applied to the current loss cost of the subject
experience, rather than using the larger group’s loss
costs
t directly
di tl tto reduce
d th
the bi
bias

Evaluation
 Assuming the rate changes are relatively small, this
complement
l t is
i lik
likely
l tto b
be accurate
t over th
the llong tterm
 The data for this method is most likely readily available,
and the calculations are very straightforward
50
Harwayne’s
Harwayne s Method

Method Description
 Harwayne’s Method is used when the subject experience
andd related
l t d experience
i h
have significantly
i ifi tl diff
differentt
distributions, and the related experience requires
adjustment before it can be blended with the subject
experience.

 In the application of Harwayne’s method, the

complement of credibility is determined using
countrywide data (excluding the base state being
reviewed), but the countrywide data is adjusted to
remove overall differences between states
states.
51

Harwayne’s
Harwayne s MethodThe following example illustrates the steps

necessary to calculate the complement for class 1
of state A.
 First Step, calculate average pure premium for
state A, B, C

 Adjustment factors are calculated by dividing the

average pure premium for state A by the
reweighted
i ht d average pure premiumi ffor B and
dCC:

 The adjusted loss costs for class 1 in states B

and C,

 Complement of credibility for class 1 of state A

52
 Harwayne’s
Harwayne s Method
Evaluation
a uat o
 The complement derived from this method is unbiased
j
as it adjusts for the distributional differences.
 The use of multi-state data generally implies the
complement is reasonably accurate as long as there is
sufficient countrywide data to minimize the process
variance.
 Since
Si th
the subject
bj t experience
i andd related
l t d experience
i
consider data from different states, the complement is
considered mostly independent
independent.
 The data for the complement is usually available but the
computations can be time-consuming
time consuming and complicated.
53

Trended Present Rates

Method Description
 Used when there is no larger group to use for the complement
complement.
 Actuaries may rely on the current rates as the best available proxy for the
indicated rate but two adjustments are made before using the current rates.
 First, insurers do not always implement the rate that is indicated. Thus, the
current rates should be adjusted to what was previously indicated rather
than what was implemented.
 Second, changes in loss cost level may have occurred between the time
the current rates were implemented and the time of the review. Sources of
suchh changes
h may bbe monetary
t iinflation,
fl ti di
distributional
t ib ti l shifts,
hift etc.
t

The actuary should select an appropriate annual loss trend and apply it from
th original
the i i l target
t t effective
ff ti date
d t off the
th currentt rates
t tot the
th target
t t effective
ff ti
date of the new rates.

54
 Trended Present Rates
 Example
• Present
P average rate is
i $200
• The selected annual loss trend is 5%
• The rate change indicated in the last review was 10%, and the target
effective date was January 1, 2011
• The rate change implemented with the last review was 6%, and the
actual effective date was Februaryy 1,, 2011
• The proposed effective date of the next rate change is January 1, 2013
 First, the loss trend length must be measured
The length from the target effective date of the last rate review (January
1, 2011) to the target effective date of the next rate change (January 1,
2013), or two years.
 The complement of Credibility

55

Trended Present Rates

Evaluation
 The accuracy of this complement depends largely on the process
variance of the historical loss costs. That is why it is used primarily for
indications with voluminous data.
 This complement is unbiased since pure trended loss costs are
unbiased (with no updating for more current loss costs).
 This
Thi complement
l t may or may nott be
b independent
i d d t depending
d di on theth
historical experience used to determine the subject experience and
complement.
For example, if the complement comes from a review that used data
from years 2007 through 2010, and the subject experience is based
on data from 2008 throughg 2011,, then the two are not independent.
p
 The data required is readily available, the calculations are very
straightforward, and the approach is easily explainable.

56
 Competitors’ Rates
Competitors
Method Description
 New or small companies
p with small volumes of data often find their own data too
unreliable for ratemaking.
 The actuary may use the competitors’ rates as a complement.
 The rationalization is that if the competitors
p have a much larger
g number of
exposures, the competitors’ statistics have less process error.

Evaluation
 An actuary must consider that competitors’ manual rates are not only based on the
competitors’ loss costs but also reflect marketing considerations, judgment, and
the effects of the regulatory.
 Competitors may also have different underwriting and claim practices than the
subject company, which creates bias that may be difficult to quantify.
 The competitors’ rates will be independent of the company data.
 While the calculations may be straightforward, the data needed for this
complement may be difficult or time-consuming to obtain.
 Even with the potential differences between competitors, the rates of a similar
competitor
tit bbear a llogical
i l relationship
l ti hi andd are generally
ll accepted
t d as a complement
l t
by regulators.
57

Methods For Developing

p
Complements of Credibility
y
- Excess Ratemaking

58
 Methods for developing complements for
excess ratemaking

 Increased limits analysis

 Lower limits analysis

 Limit analysis

59

Increased limits analysis

Method Description
 Actuaries use this method when data is available for ground-up losses
through the attachment point (i.e., losses have not been truncated at any
point below the bottom of the excess layer
p y being gp
priced).
)
 Increased limits factors are used to adjust losses capped at the
attachment point to produce an estimate of losses in the specific excess
layer.
layer


• LA are the losses capped at the attachment point A;

• ILFA is the increased limits factor for the attachment point A;
• ILFA+L is the increased limits factor for the sum of the attachment point
A and the excess insurer’s
insurer s limit of liability L.
L

60
 Increased limits analysis
Calculate the complement of credibility for the excess layer
between $500,000 and $750,000 (i.e., $250,000 of
g in excess of $500,000).
coverage )
 Assume the losses capped at $500,000 are $2,000,000
 The followingg increased limits factors apply:
pp y

61

Increased limits analysis

Evaluation
 If the subject experience has a different size of loss distribution than that
used to develop the increased limits factors, then the results of this
procedure will be biased. This is particularly relevant as the increased
li it factors
limits f t may be
b based
b d on industry
i d t data
d t rather
th than
th theth iinsurer’s
’ own
data.
 The error with this approach
pp is not the p
process error but rather the
parameter error associated with the selection of the increased limits
factors. Thus, the error associated with this estimate tends to be
independent of the error associated with the base statistic.
 This procedure requires increased limits factors—preferably industry
factors—and ground-up losses that have not been truncated below the
attachment point
point. To the e
extent
tent that information is a
available,
ailable the
procedure is practical. In terms of acceptability, however, this estimate is
more logically related to the data below the attachment point (which is
used for the projection) than to the data in the layer, and this may be
controversial. 62
Lower limits analysis
Method Description
p
 If those losses are too sparse to be reliable, the actuary may prefer to
use losses capped at a limit lower than the attachment point. This
lower limit can often be the basic limit
limit.


• Ld are the losses capped at the lower limit, d;

• ILFA is the increased limits factor for the attachment point A;
• ILFd is the increased limits factor for the lower limit, d.
• ILFA+L is the increased limits factor for the sum of the attachment
point A and the excess insurer’s limit of liability L
(i.e., this sum represents the top of the excess layer being priced).

63

Lower limits analysis

Calculate the complement of credibility for the excess layer
between $500,000 and $750,000 (i.e., $250,000 of
g in excess of $500,000).
coverage )
 Assume the losses capped at $250,000 are $1,500,000
 The followingg increased limits factors apply:
pp y

64
 Lower limits analysis
Evaluation
 It iis diffi
difficult
lt tto d
determine
t i whether
h th thithis iis more or lless accurate
t th
than th
the
previously discussed complement.
 Intuitively, y, this complement
p will be more biased as the differences in size
of loss distributions will be exacerbated when using losses truncated at
lower levels.
 On the other hand hand, using losses capped at lower limits may increase the
stability of the estimate.
 Like the previous complement, the error associated with this complement
is generally independent of the error of the base statistic.
 Insurers generally code losses capped at basic limits for statistical
reporting purposes
purposes. If some other lower limit is chosen
chosen, the data may not
be as available. The calculations are no more difficult than the previously
discussed complement, but the criticism of being logically related.

65

Limits analysis
Method Description
 Primary
Pi iinsurers generallyll sellll policies
li i with
ith a wide
id variety
i t off policy
li lilimits.
it
Some of the individual policy limits fall below the attachment point, and
some extend beyond the top of the excess layer.
 For this approach the actuary analyzes the policies at each limit of
coverage separately. The actuary calculates the estimated losses in a
given layer using the premium volume and expected loss ratio in that
layer. Then the actuary performs an increased limits factor analysis on
each first dollar limit’s loss costs separately.


• LR= Total limits loss ratio,

• Pd= Total premium for policies with limit d
• The ILFs have the same meaning as previously discussed.
66
 Limits analysis

The following shows an example of the

calculation of the expected loss for the
layer
y between $ $500,000
, and $$750,000
,
assuming a total limits loss ratio of 60%.

67

Limits analysis
Evaluation
 This complement is biased and inaccurate to the same
extent as the prior two complements.
 It involves relying on the additional assumption that the
expected loss ratio does not vary by limit.
 Because this type of excess loss analysis is typically
undertaken by reinsurers that do not have access to the full
loss distribution
distribution, this may be the only method available
available.
 It is more time-consuming to compute, but the calculations
are straightforward.
straightforward
 It generates the same criticism as the other methods
because it is not based on actual data from the layer being
priced.
68
 Fitted Curves
Method Description
 Actual
A t l loss
l di
distributions
t ib ti can bbe very volatile,
l til especially
i ll iin th
the
tail of the distribution (i.e., the higher losses).
 Actuaries may fit curves to the actual data to smooth out the
volatility and to extrapolate the distribution to higher limits.
 Once the curve is fitted
fitted, the techniques described in Chapter
11 can be used to determine the expected losses in the layer
being gppriced.
 The following formula is used to determine the percentage of
the curve’s total losses that are expected in the excess layer:

69

Fitted Curves
Evaluation
 This complement tends to be less biased and more stable than the
other excess methods, assuming that the fitted curve replicates the
general shape of the actual data.
 This approach tends to be significantly more accurate than the
others when there are relatively few claims in the higher layers.
 Because the curve-fitting process involves the underlying data, it
can be heavily dependent on the existence or non-existence of
larger claims. Thus, the error associated with a complement
developed using this approach will tend to be less independent than
complements determined from the other approaches.
 This approach tends to be the most computationally complex and
requires data that may not be readily available.
 The complement developed using this approach tends to be the
most logically
l i ll related
l d to the
h losses
l in
i the
h llayer than
h theh others
h as the
h
data is more fully used.
 However, the computational complexity may make it difficult to
communicate
communicate.
70
 Werner, G, and Modlin, C., Basic Ratemaking
Chapter 16:
16: Claims
Claims--Made Ratemaking

黃慕淳
2011 11 06
2011.11.06

CONTENT

 Introduction
 Aggregation of losses by report year and report year lag
 Coverage triggers for claims
claims-made
made coverage
 The five principles of claims-made policies
 Issues related to coordinating coverage between claims-made
and occurrence policies

2
 INTRODUCTION

 Loss trends for liability lines during the 1960s and 1970s
□ increased dramatically in severity due to high economic and social
inflation
□ as well as increases in claim frequency
□ especially the case for professional liability insurance including medical
malpractice
• It took several years before insurers realized that their products
were significantly underpriced because of the long-tailed nature of
professional liability
• This delay in recognizing price inadequacy highlights the
significant pricing risk that exists for long-tailed insurance

INTRODUCTION

 The long period between the occurrence of a claim and the settlement of a
claim can be driven by
□ Reporting Lag :
the time between the occurrence date and report date
• relates to pure IBNR – claims that are incurred but not reported
• Ex.: it may take several years for the physician’s error to cause
id ifi bl symptoms
identifiable
□ Settlement Lag :
the time between the report date and settlement date
• relates to IBNER – claims that are incurred but not enough reported
• Ex.: it mayy take manyy yyears for the claim to be ultimately
y settled
due to factors such as the need for ongoing treatment and lengthy
court proceedings

4
 COVERAGE TRIGGERS

 The major difference between claims-made and occurrence coverage is that

the coverage trigger
□ For claims-made policies
• The coverage trigger is the date the claim is reported
• Only needs to project claims reported during next year’s policy
period
□ For occurrence policies
• The coverage trigger is the date the event occurs
• Must consider claims that will be reported many years into the
future

REPORT YEAR AGGREGATION

 Categorizes claims by the year reported and the report lag:

Report Report Lag

Year 0 1 2 3 4
2010 L(2010 0)
L(2010,0) L(2010 1)
L(2010,1) L(2010 2)
L(2010,2) L(2010 3)
L(2010,3) L(2010 4)
L(2010,4)
2011 L(2011,0) L(2011,1) L(2011,2) L(2011,3) L(2011,4)
2012 ( , )
L(2012,0) L(2012,1)
( , ) L(2012,2)
( , ) L(2012,3)
( , ) L(2012,4)
( , )
2013 L(2013,0) L(2013,1) L(2013,2) L(2013,3) L(2013,4)
2014 L(2014,0) L(2014,1) L(2014,2) L(2014,3) L(2014,4)
2015 L(2015,0) L(2015,1) L(2015,2) L(2015,3) L(2015,4)

□ L(2010,0) : a claim that occurs in 2010 and is reported in year 2010

□ L(2012,2) : a claim that is reported in 2012 after a report lag of two year

6
 REPORT YEAR AGGREGATION

 Row
□ claims
l i reportedd in
i a given
i year (i.e.,
(i the h report year))
 Column
□ claims that share the same reporting lag
 Diagonal (top left to bottom right)
□ claims that occurred in the same year (i.e., the same accident year)
N
 Occurrence policies (Y) =  L(Y  i,i)
0
□ Occurrence policies (2010) = L(2010,0) + L(2011,1) + … + L(2014,4)
N
 Claims-made policies (Y) =  L(Y,i)
0
□ Occurrence policies (2010) = L(2010,0) + L(2010,1) + … + L(2010,4)

REPORT YEAR AGGREGATION

 2010 Claims-Made Policies:

Report Report Lag

Year 0 1 2 3 4

2010 L(2010,0) L(2010,1) L(2010,2) L(2010,3) L(2010,4)

2011 L(2011,0) L(2011,1) L(2011,2) L(2011,3) L(2011,4)
2012 L(2012,0) L(2012,1) L(2012,2) L(2012,3) L(2012,4)
2013 L(2013,0) L(2013,1) L(2013,2) L(2013,3) L(2013,4)
2014 L(2014 0)
L(2014,0) L(2014 1)
L(2014,1) L(2014 2)
L(2014,2) L(2014 3)
L(2014,3) L(2014 4)
L(2014,4)
2015 L(2015,0) L(2015,1) L(2015,2) L(2015,3) L(2015,4)

8
 REPORT YEAR AGGREGATION

 2010 Occurrence Policies:

Report Report Lag

Year 0 1 2 3 4

2010 L(2010,0) L(2010,1) L(2010,2) L(2010,3) L(2010,4)

2011 L(2011,0) L(2011,1) L(2011,2) L(2011,3) L(2011,4)
2012 L(2012,0) L(2012,1) L(2012,2) L(2012,3) L(2012,4)
2013 L(2013,0) L(2013,1) L(2013,2) L(2013,3) L(2013,4)
2014 L(2014 0)
L(2014,0) L(2014 1)
L(2014,1) L(2014 2)
L(2014,2) L(2014 3)
L(2014,3) L(2014 4)
L(2014,4)
2015 L(2015,0) L(2015,1) L(2015,2) L(2015,3) L(2015,4)

PRINCIPLES OF CLAIMS-MADE POLICIES

 A claims-made policy should always cost less than an occurrence policy as

long as claim costs are increasing
 If there is a sudden, unpredictable change in the underlying trends, the
claims-made policy priced based on the prior trend will be closer to the
correct price than an occurrence policy based on the prior trend
 If there is a sudden, unexpected shift in the reporting pattern, the cost of a
mature claims-made
claims made policy (i.e.,
(i e a policy that covers claims reported during
the policy period regardless of accident date) will be affected relatively little,
if at all, relative to the occurrence policy
 Claims-made policies incur no liability for IBNR, so the risk of reserve
inadequacy is greatly reduced
 The investment income earned from claims-made policies is substantially
less than under occurrence policies

10
 PRINCIPLES OF CLAIMS-MADE POLICIES

 Example
□ A
Assumptions
i
• Exposure levels are constant
• The average loss cost for Report Year 2010 is $1,000
$1 000
• Loss costs increase by 5% each report year
• An equal number of incurred claims are reported each year and all
claims are reported within five years of occurrence (i.e., 20%
reported each year)
• Loss costs do not vary by report year lag. Also, any trends affecting
settlement lag have been ignored

11

PRINCIPLES OF CLAIMS-MADE POLICIES

 Example
□ A
Assumptions
i
Report Report Lag
Year 0 1 2 3 4
2010 200.00 200.00 200.00 200.00 200.00
2011 210.00 210.00 210.00 210.00 210.00
2012 220.50 220.50 220.50 220.50 220.50
2013 231.53 231.53 231.53 231.53 231.53 +5%
2014 243 10
243.10 243 10
243.10 243 10
243.10 243 10
243.10 243 10
243.10
2015 255.26 255.26 255.26 255.26 255.26
2016 268.02 268.02 268.02 268.02 268.02
2017 281.42 281.42 281.42 281.42 281.42
2018 295.49 295.49 295.49 295.49 295.49

12
 PRINCIPLES OF CLAIMS-MADE POLICIES

 Example
□ Cl i
Claims-Made
d Loss C
Costs
Report Report Lag Claims-made
Year 0 1 2 3 4 Loss Costs
2010 200.00 200.00 200.00 200.00 200.00 $1,000.00
2011 210.00 210.00 210.00 210.00 210.00 $1,050.00
2012 220.50 220.50 220.50 220.50 220.50 $1,102.50
2013 231.53 231.53 231.53 231.53 231.53 $1,157.65
2014 243 10
243.10 243 10
243.10 243 10
243.10 243 10
243.10 243 10
243.10 $1 215 50
$1,215.50
2015 255.26 255.26 255.26 255.26 255.26 $1,276.30
2016 268.02 268.02 268.02 268.02 268.02 $1,340.10
2017 281.42 281.42 281.42 281.42 281.42 $1,407.10
2018 295.49 295.49 295.49 295.49 295.49 $1,477.45

13

PRINCIPLES OF CLAIMS-MADE POLICIES

 Example
□ O
Occurrence Loss C
Costs
Report Report Lag Occurrence
Year 0 1 2 3 4 Loss Costs
2010 200.00 200.00 200.00 200.00 200.00
2011 210.00 210.00 210.00 210.00 210.00
2012 220.50 220.50 220.50 220.50 220.50
2013 231.53 231.53 231.53 231.53 231.53
2014 243 10
243.10 243 10
243.10 243 10
243.10 243 10
243.10 243 10
243.10 $1 105 13
$1,105.13
2015 255.26 255.26 255.26 255.26 255.26 $1,160.39
2016 268.02 268.02 268.02 268.02 268.02 $1,218.41
2017 281.42 281.42 281.42 281.42 281.42 $1,279.33
2018 295.49 295.49 295.49 295.49 295.49 $1,343.29

14
 PRINCIPLES OF CLAIMS-MADE POLICIES

 Principle 1 :
A claims
claims-made
made policy should always cost less than an occurrence policy as
long as claim costs are increasing

15

PRINCIPLES OF CLAIMS-MADE POLICIES

 Principle 1 : (conti.)
□ Occurrence policies
O li i require
i the
h actuary to make k projections
j i about
b the
h
settlement of claims that occur further out into the future
• For a 2011 occurrence policy, it is need to project the ultimate value
of claims that occur in 2011 and may not even be reported until
2015
• F
For a 2011 claims-made
l i d policy,
li iti only
l needsd to project
j theh ultimate
li
cost of claims that will be reported in that year
• For claims
claims-made
made policies, there is a shorter period of time between
coverage trigger and settlement date

16
 PRINCIPLES OF CLAIMS-MADE POLICIES

 Principle 2 :
If there is a sudden,
sudden unpredictable change in the underlying trends,
trends the
claims-made policy priced based on the prior trend will be closer to the
correct price than an occurrence policy based on the prior trend

17

PRINCIPLES OF CLAIMS-MADE POLICIES

 Principle 2 : (conti.)
□ Example:l
Restates the above example assuming the actual loss cost trend by report
year is 7% instead of 5%

Report Report Lag

Year 0 1 2 3 4
2010 200.00 200.00 200.00 200.00 200.00
2011 214.00 214.00 214.00 214.00 214.00
2012 228.98 228.98 228.98 228.98 228.98 +7%
2015 245.01 245.01 245.01 245.01 245.01
…

18
 PRINCIPLES OF CLAIMS-MADE POLICIES

 Principle 2 : (conti.)
□ Cl i
Claims-Made
d Loss C
Costs
Report Report Lag Claims-made
Year 0 1 2 3 4 Loss Costs
2010 200.00 200.00 200.00 200.00 200.00 $1,000.00
2011 214.00 214.00 214.00 214.00 214.00 $1,070.00
2012 228.98 228.98 228.98 228.98 228.98 $1,144.90
2013 245.01 245.01 245.01 245.01 245.01 $1,225.05
2014 262 16
262.16 262 16
262.16 262 16
262.16 262 16
262.16 262 16
262.16 $1 310 80
$1,310.80
2015 280.51 280.51 280.51 280.51 280.51 $1,402.55
2016 300.15 300.15 300.15 300.15 300.15 $1,500.75
2017 321.16 321.16 321.16 321.16 321.16 $1,605.78
2018 343.64 343.64 343.64 343.64 343.64 $1,718.20

19

PRINCIPLES OF CLAIMS-MADE POLICIES

 Principle 2 : (conti.)
□ O
Occurrence Loss C
Costs
Report Report Lag Occurrence
Year 0 1 2 3 4 Loss Costs
2010 200.00 200.00 200.00 200.00 200.00
2011 214.00 214.00 214.00 214.00 214.00
2012 228.98 228.98 228.98 228.98 228.98
2013 245.01 245.01 245.01 245.01 245.01
2014 262 16
262.16 262 16
262.16 262 16
262.16 262 16
262.16 262 16
262.16 $1 150 15
$1,150.15
2015 280.51 280.51 280.51 280.51 280.51 $1,230.66
2016 300.15 300.15 300.15 300.15 300.15 $1,316.81
2017 321.16 321.16 321.16 321.16 321.16 $1,408.99
2018 343.64 343.64 343.64 343.64 343.64 $1,507.62

20
 PRINCIPLES OF CLAIMS-MADE POLICIES

 Principle 2 : (conti.)
□ The
h unexpected d iincrease in
i trendd resulted
l d in
i the
h Report Year 2011 loss
l
cost for the claims-made policy is 1.9%
( = $1,070.00 / $1,050.00 – 1.0)
□ The unexpected increase in trend resulted in the Accident Year 2011 loss
cost for the occurrence policy is 6.1%
( = $1,230.66
$1 230 66 / $1,160.39
$1 160 39 – 1.0)
1 0)
□ An error made in the trend selection for occurrence policies has more of
an impact than for claims-made policies

21

PRINCIPLES OF CLAIMS-MADE POLICIES

 Principle 3 :
If there is a sudden,
sudden unexpected shift in the reporting pattern,
pattern the cost of a
mature claims-made policy (i.e., a policy that covers claims reported during
the policy period regardless of accident date) will be affected relatively little,
if at all, relative to the occurrence policy

22
 PRINCIPLES OF CLAIMS-MADE POLICIES

 Principle 3 : (conti.)
□ Example:l
Instead of 20% of the claims being reported each year, assume that 5%
of the claims are reported one year later than expected, but all claims are
reported within five years ($50 = $1,000 × 5%)
Report Report Lag
Year 0 1 2 3 4
2010 150.00 200.00 200.00 200.00 250.00

– 50.00 +50.00

– 50.00 +50.00

– 50.00 +50.00

– 50.00 +50.00

23

PRINCIPLES OF CLAIMS-MADE POLICIES

 Principle 3 : (conti.)
□ Cl i
Claims-Made
d Loss C
Costs
Report Report Lag Claims-made
Year 0 1 2 3 4 Loss Costs
2010 150.00 200.00 200.00 200.00 250.00 $1,000.00
2011 157.50 210.00 210.00 210.00 262.50 $1,050.00
2012 165.38 220.50 220.50 220.50 275.63 $1,102.51
2013 173.64 231.53 231.53 231.53 289.41 $1,157.64
2014 182 33
182.33 243 10
243.10 243 10
243.10 243 10
243.10 303 88
303.88 $1 215 51
$1,215.51
2015 191.44 255.26 255.26 255.26 319.07 $1,276.29
2016 201.02 268.02 268.02 268.02 335.03 $1,340.11
2017 211.07 281.42 281.42 281.42 351.78 $1,407.11
2018 221.62 295.49 295.49 295.49 369.37 $1,477.46

24
 PRINCIPLES OF CLAIMS-MADE POLICIES

 Principle 3 : (conti.)
□ O
Occurrence Loss C
Costs
Report Report Lag Occurrence
Year 0 1 2 3 4 Loss Costs
2010 150.00 200.00 200.00 200.00 250.00
2011 157.50 210.00 210.00 210.00 262.50
2012 165.38 220.50 220.50 220.50 275.63
2013 173.64 231.53 231.53 231.53 289.41
2014 182 33
182.33 243 10
243.10 243 10
243.10 243 10
243.10 303 88
303.88 $1 115 91
$1,115.91
2015 191.44 255.26 255.26 255.26 319.07 $1,171.70
2016 201.02 268.02 268.02 268.02 335.03 $1,230.30
2017 211.07 281.42 281.42 281.42 351.78 $1,291.80
2018 221.62 295.49 295.49 295.49 369.37 $1,356.40

25

PRINCIPLES OF CLAIMS-MADE POLICIES

 Principle 3 : (conti.)
□ There
h is i no impact
i on the
h loss
l cost estimates
i for
f the
h claims-made
l i d
policies
□ The estimates for the occurrence policies have changed from the
original table
• The Accident Year 2011 loss cost estimate for the occurrence
policies
li i has
h changed
h d by
b 1% ( = ($1,171.70
($1 171 70 / $1,160.39)
$1 160 39) – 1.0)
1 0)

26
 PRINCIPLES OF CLAIMS-MADE POLICIES

 Principle 4 :
Claims made policies incur no liability for IBNR,
Claims-made IBNR so the risk of reserve
inadequacy is greatly reduced

27

PRINCIPLES OF CLAIMS-MADE POLICIES

 Principle 4 : (conti.)
□ For occurrence policies
li i
• need to estimate claims that are incurred but not reported (pure
IBNR) and claims that are incurred but not enough reported
(IBNER)
□ For claims-made policies
• no pure IBNR component; only need to determine an IBNER
reserve
• the risk of reserve inadequacy is greatly reduce

28
 PRINCIPLES OF CLAIMS-MADE POLICIES

 Principle 5 :
The investment income earned from claims
claims-made
made policies is substantially
less than under occurrence policies

29

PRINCIPLES OF CLAIMS-MADE POLICIES

 Principle 5 : (conti.)
□ Claims-made
Cl i d policy
li shortens
h the
h period
i d off time
i between
b collection
ll i off
premium and payment of claim; consequently, funds invested for a
shorter time horizon result in less investment income
□ When determining the target underwriting profit provision for a claims-
made policy, the pricing actuary should take into consideration both the
reduced investment income as well as the reduced pricing risk

30
 COORDINATING POLICIES

 Since occurrence and claims-made policies have different coverage triggers,

insureds converting from one policy type to the other should be cognizant of
coverage overlaps or gaps

31

COORDINATING POLICIES

 Example
□ As shown
A h iin the
h following
f ll i diagram,
di there
h is i overlapping
l i coverage
between the occurrence policy and the claims-made policy for an
insured switches from occ to c-m starting in 2011

Report Report Lag

Year 0 1 2 3 4

2010 L(2010,0) L(2010,1) L(2010,2) L(2010,3) L(2010,4)

2011 L(2011,0) L(2011,1) L(2011,2) L(2011,3) L(2011,4)
2012 L(2012,0) L(2012,1) L(2012,2) L(2012,3) L(2012,4)
2013 L(2013,0) L(2013,1) L(2013,2) L(2013,3) L(2013,4)
2014 L(2014,0) L(2014,1) L(2014,2) L(2014,3) L(2014,4)
2015 L(2015,0) L(2015,1) L(2015,2) L(2015,3) L(2015,4)

32
 COORDINATING POLICIES

 Retroactive Date
□ The
h claims-made
l i d coverage only
l covers claims
l i that
h occur on or after
f theh
retroactive date
□ To provide complete coverage without overlap, the retroactive date
should be coordinated with the expiration of the last occurrence policy
□ The insured can purchase :
• a first-year claims-made policy in 2011 with a retroactive date of January 1,
2011, and
y claims-made ppolicyy with a retroactive date of Januaryy 1,,
• a second-year
2011,
• ,…, and a mature claims-made policy in 2015

33

COORDINATING POLICIES

 Retroactive Date (conti.)

□ The
h first-year
fi claims-made
l i d policy
li will
ill coverL(2011,0)
(2011 0)
The second-year claims-made policy will cover L(2012,0) and L(2012,1)
The mature claims-made policy will cover L(2015,0) to L(2015,5)

Report Report Lag

Year 0 1 2 3 4

2010 L(2010,0) L(2010,1) L(2010,2) L(2010,3) L(2010,4)

2011 L(2011,0) L(2011,1) L(2011,2) L(2011,3) L(2011,4)
2012 L(2012,0) L(2012,1) L(2012,2) L(2012,3) L(2012,4)
2013 L(2013,0) L(2013,1) L(2013,2) L(2013,3) L(2013,4)
2014 L(2014,0) L(2014,1) L(2014,2) L(2014,3) L(2014,4)
2015 L(2015,0) L(2015,1) L(2015,2) L(2015,3) L(2015,4)

34
 COORDINATING POLICIES

 The Step Factor

□ i usedd to recognize
is i theh growthh in
i exposure for
f eachh successive
i claims-made
l i d
policy during the transition
□ is a percentage of the mature claims-made rate
□ The cumulative values of the loss estimates for L(2015,0), L(2015,1), L(2015,2),
L(2015,3) and L(2015,4) expressed as a ratio to the total losses for Report Year
2015 can be used to determine the stepp factors

Claims-Made Year Step Factor

First 40%
Second 70%
Third 85%
F
Fourth
h 95%
Fifth or More 100%

35

COORDINATING POLICIES

 Example
□ S i hi from
Switching f a claims-made
l i d policy
li to an occurrence policy
li ini 2011
□ This situation may creates a coverage gap (i.e., no coverage for claims
that occurred before 2011 but were not reported)

Report Report Lag

Year 0 1 2 3 4

2010 L(2010,0) L(2010,1) L(2010,2) L(2010,3) L(2010,4)

2011 ( )
L(2011,0) L(2011,1)
( ) L(2011,2)
( ) L(2011,3)
( ) L(2011,4)
( )
2012 L(2012,0) L(2012,1) L(2012,2) L(2012,3) L(2012,4)
2013 L(2013,0) L(2013,1) L(2013,2) L(2013,3) L(2013,4)
2014 L(2014,0) L(2014,1) L(2014,2) L(2014,3) L(2014,4)
2015 L(2015,0) L(2015,1) L(2015,2) L(2015,3) L(2015,4)

36
 COORDINATING POLICIES

 Example (conti.)
□ Companies
C i offer
ff an extended
d d reporting
i endorsement
d (or
( tailil coverage )
that covers claims that occurred but were not reported before the
expiration of the last claims-made policy

Report Report Lag

Year 0 1 2 3 4

2010 L(2010,0) L(2010,1) L(2010,2) L(2010,3) L(2010,4)

2011 ( )
L(2011,0) L(2011,1)
( ) L(2011,2)
( ) L(2011,3)
( ) L(2011,4)
( )
2012 L(2012,0) L(2012,1) L(2012,2) L(2012,3) L(2012,4)
2013 L(2013,0) L(2013,1) L(2013,2) L(2013,3) L(2013,4)
2014 L(2014,0) L(2014,1) L(2014,2) L(2014,3) L(2014,4)
2015 L(2015,0) L(2015,1) L(2015,2) L(2015,3) L(2015,4)

37

COORDINATING POLICIES

 The gap in coverage in the case of retirement

□ The
h protection
i isi given
i by
b a tail
il policy
li that
h covers losses
l occurring
i during
d i
the period for which claims-made coverage was in force and that are
reported after the insured’s last claims-made policy expires

Report Report Lag

Year 0 1 2 3 4

2010 L(2010,0) L(2010,1) L(2010,2) L(2010,3) L(2010,4)

2011 ( )
L(2011,0) L(2011,1)
( ) L(2011,2)
( ) L(2011,3)
( ) L(2011,4)
( )
2012 L(2012,0) L(2012,1) L(2012,2) L(2012,3) L(2012,4)
2013 L(2013,0) L(2013,1) L(2013,2) L(2013,3) L(2013,4)
2014 L(2014,0) L(2014,1) L(2014,2) L(2014,3) L(2014,4)
2015 L(2015,0) L(2015,1) L(2015,2) L(2015,3) L(2015,4)

38
 KEY CONCEPTS

 Rationale for claims-made coverage

 Aggregating
A i llosses by
b report year andd report lag
l
 Coverage triggers for claims-made coverage
 Five principles of claims-made
claims made policies
 Coordinating coverage
□ Retroactive date
□ First- and second-year claims-made policies
□ Mature claims-made policies
p
□ Extended reporting endorsement or tail coverage

39