BONUS MALUS

SYSTEM
Bonus-malus system as an experience rating scheme
Bonus-malus system in few countries
Markov chain – discrete time
Properties associated with the stationary distribution of
bonus-malus system
Motor Insurance

Comprehensive cover
Third party, fire & theft cover
Third party cover
Death or injury to third Loss or damage to third Damage to your own Damage to your own
party in an accident. party’s car or property. vehicle due to fire or vehicle due to accident.
. theft.

STQA4113/NM BONUS MALUS SYSTEM 2

 Motor Tariff Third Party,
Fire & Theft
(RM)

75% for

 Motor insurance is mandatory insurance protection. premium using

calculation

 The premiums which insurance companies can charge

method for
comprehensive

consumers have been regulated by a tariff structure.

 This has resulted in a growing gap between premiums collected and claims paid out by the
insurers.
 This has led to the Motor Third Party business becoming unsustainable.

Motor Tariff:
 Tariffs are fixed price lists that determine the premium rates which insurance companies can
charge consumers for insurance products
 Insurance companies are not allowed to vary the prices chargeable on an insurance policy
 Basic rating factors considered were scope of insurance, cubic capacity of vehicle and estimated
value of vehicle (or sum insured, whichever is lower).
 The final premium to be paid is adjusted by the policyholder’s claim experience, or equivalently,
his NCD entitlement.

STQA4113/NM BONUS MALUS SYSTEM 3

 De-tariffication/Liberalization of Motor Tariff
 Liberalisation of the tariff means removing the tariff structure or
freeing up the fixed premium rates so that insurance companies are
able to charge premiums that commensurate or correspond to the
risk behaviour of the consumers.
 Different insurance companies can charge different rates for the
same risk behaviour based on their business risks models and
strategies.
 Pricing of motor insurance protection is more reflective of risks
behaviour of the consumers.
 Premium pricing is determined by individual insurers and takaful
operators, and the consumers are able to enjoy a wider choice of
motor insurance products at competitive prices.
 premiums are based on broader risk factors:
 cubic capacity of vehicle, estimated value of vehicle
 age of vehicle, age of driver, safety and security features of vehicle,
 geographical location of vehicle and traffic offences of driver Starting 1st July 2017

STQA4113/NM BONUS MALUS SYSTEM 4

Understanding Motor Insurance

Premium Excess

Exclusions
STQA4113/NM BONUS MALUS SYSTEM 5
 Bonus Malus System

 Also known as: no-fault discount, merit rating, experience rating or no-claim discount (NCD)
 penalizing insureds who are responsible for one or more claims by a premium surcharge (malus),
and rewarding insureds with a premium discount (bonus) if they do not have any claims.
 Insurers use bonus-malus systems for two main purposes:
 to encourage drivers to drive more carefully in a year without any claims, and
 to ensure insureds to pay premiums proportional to their risks based on their claims experience via an experience
rating mechanism.

 NCD is an experience rating system commonly used in motor insurance & health insurance.
 It represents an attempt to categorize insureds into homogeneous groups who pay premiums
based on their claims experience.
 Depending on the rules in the scheme, new policyholders may be required to pay full premium
initially, and obtain discounts in the future years as a result of claim-free years.

STQA4113/NM BONUS MALUS SYSTEM 6

 Bonus Malus System
 Insurance premium is usually derived from an annual frequency of claims, which is modelled by
using statistical data.
 The annual frequency of the claims is calculated from the number of the claims on a contract.
 They depend on many factors that are believed to have an impact on the expected cost of future
claims, such as the car characteristics (vehicle body, vehicle age) and the profile of the driver (age,
gender, driving history).
 Hidden factors that difficult to be taken into account (such as driving capacity, swiftness of reflexes,
aggressiveness behind the wheel, knowledge of the highway code, drug abuse) are revealed by
the number of claims reported by the policyholders over the successive insurance periods
 For these reasons, insurance companies approach to individualization of risk and they use the
Bonus-Malus system.
 Then the premium amount paid by the policyholder depends on the risk factors but also on the
claims history

STQA4113/NM BONUS MALUS SYSTEM 7

 NCD Discount at Renewal
Insurance Period Private Commercial Motorcycle

NCD Systems in Malaysia 1st year

2nd year
0%
25%
0%
15%
0%
15%
3rd year 30% 20% 20%
4th year 38.3% 25% 25%
 No Claim Discount (NCD) is a type of motor insurance 5th year 45% 25% 25%

discount given to you: 6th year onwards 55% 25% 25%

 for not making any insurance claim (being a safe driver) in

the past 12 months AND 5
 for not having any insurance claim by third party made
4
against your car insurance policy
 NCD plays a big role in determining how much you have to 3
pay for motor insurance as it offers discount up to 55% of
2
your base motor insurance or motor takaful coverage.
 NCD structure from the Motor Tariff remains ‘unchanged’ 1
and continue to exist, and is ‘transferable’ from one insurer,
0
or from one takaful operator, to another.
Transition Diagram for NCD Classes (Malaysia)

STQA4113/NM BONUS MALUS SYSTEM 8

 NCD System in Singapore
5
NCD Discount at Renewal
4
Insurance Period Private Commercial Motorcycle
1st year 0% 0% 0% 3
2nd year 10% 10% 10%
2
3rd year 20% 15% 15%
4th year 30% 20% 20% 1
5th year 40% 20% 20%
0
6th year onwards 50% 20% 20%
Transition Diagram for NCD Classes (Singapore)

Following the Motor Claims Framework (MCF), some insurers may

penalise your NCD by about 10% for private cars, and 5% for
commercial cars and motorcycles, if you fail to file an accident
report within the stipulated 24-hour timeframe.

STQA4113/NM BONUS MALUS SYSTEM 9

 NCD System in Brazil NCD System in Australia
NCD Discount at Renewal NCD Discount at Renewal NCD NCD
following 1 following
Insurance Period Discount claim more than
Insurance Period Discount
st 1 claim
1 year 0%
st
2
nd
year 10% 1 year 0% 0% 0%
nd
rd
3 year 15% 2 year 15% 0% 0%
th 6 rd
3 year 20% 0% 0%
4 year 20%
th
th
5 year 25% 5 4 year 25% 15% 0%
th
30% 5 year 30% 20% 0%
6th year 4
7th year onwards 45%
3
2

0
Transition Diagram for NCD Classes (Brazil)

STQA4113/NM BONUS MALUS SYSTEM 10

NCD System in UK NCD System in India
NCD Discount at Renewal NCD Discount at Renewal

Insurance Discount Insurance Period Discount

st
Period 1 year 0%
st
1 year 0% nd
2 year 20%
nd
2 year 25% rd
3 year 25%
rd
3 year 35% th
4 year 35%
th
4 year 45% th
5 year 45%
th
5 year 55% th
6 year onwards 50%
5
6th year 60%
4
7th year onwards 77%
Thailand: 7 discount levels 3
2
Italy: 18 discount levels Taiwan: 9 discount levels
1
Switzerland & Germany: 22 discount levels 0
Transition Diagram for NCD Classes
STQA4113/NM BONUS MALUS SYSTEM (India) 11
 NCD is determined by 3 elements
 These determine the transfer from one class to another when the
number of claims is known.
 An insured enters the system in the initial class when he applies for
Premium scale insurance and throughout the entire driving lifetime, the transition
NCD

rules are applied upon each renewal to determine the new class.
 Depending on rules, new policyholders may be required to pay full
Initial class premium initially and obtain discounts in future as a result of the
claim free years.

EFFECTS OF THE NCD

Transition rules  drivers become more responsible about their vehicles and when
driving
 reduces the cost of insurance cover
 discourages small claims
 an improvement in risk classification by allowing insurance
companies to make a bad risk pay more and a good one to pay
less
STQA4113/NM BONUS MALUS SYSTEM 12
 Transition Diagram & Matrix
- 3 by 3 NCD system (Scenario 1)
Class Discount level Premium level The rules for moving between levels are:
 If no claims are made during a year, the policyholder
0 0% 100% moves up to the next discount level or remains at the
maximum level.
1 25% 75%  If one or more claims are made during the year, the
policyholder moves down by one level or remains at the
2 40% 60% 0% discount level.

Transition probability from state 𝑖 (at time 𝑛)

𝑝01 𝑝12
𝑝22 to state 𝑗 (at time 𝑛 + 1) is called a one-step
𝑝00
transition probability, and is denoted by
0 1 2 𝑝𝑖𝑗 𝑛, 𝑛 + 1 = 𝑃𝑟 𝑋𝑛+1 = 𝑗|𝑋𝑛 = 𝑖
𝑝10 𝑝21
 p00 p01 p02   p00 p01 0 
   
Transition matrix P   p10 p11 p12    p10 0 p12 
p
 20 p21 p22   0 p21 p22 
STQA4113/NM BONUS MALUS SYSTEM 13
 Transition Diagram & Matrix
- 3 by 3 NCD system (Scenario 2)
Class Discount level Premium level The rules for moving between levels are:
 If no claims are made during a year, the policyholder moves
0 0% 100% up to the next discount level or remains at the maximum
level.
1 25% 75%  If one or more claims are made during the year, the
policyholder moves down to or remains at the 0% discount
2 40% 60% level (lowest level), no matter what level he is at.

 p00 p01 p02   p00 p01 0 

   
𝑝01 𝑝12 P   p10 p11 p12    p10 0 p12 
p  p p22 
 20 p 21 p22   20 0
0 1 2
𝑝00 𝑝22  1  p01 p01 0 
 
𝑝10 𝑝20
  1  p12 0 p12 
1  p 
 22 0 p 22 

STQA4113/NM BONUS MALUS SYSTEM 14

 Transition Diagram & Matrix
- 3 by 3 NCD system (Scenario 3)
Class Discount level Premium level The rules for moving between levels are:
 If no claims are made during a year, the policyholder moves
0 0% 100% up to the next discount level or remains at the maximum
level.
1 25% 75%  If one claim is made during the year, the policyholder
moves down by one level or remains at the 0% discount
2 40% 60% level.
 If two or more claims are made during the year the
policyholder moves down to or remains at the 0% discount
level, no matter what level he is at.
𝑝01 𝑝12
 p00 p01 0 
0 1 2  
P   p10 0 p12 
𝑝00 𝑝10 𝑝21 𝑝22
p p22 
 20 p21
𝑝20

STQA4113/NM BONUS MALUS SYSTEM 15

 Transition Diagram & Matrix
- 4 by 4 NCD system
Class Discount level Premium level The rules for moving between levels are:
 If no claims are made during a year, the policyholder moves
0 0% 100% up to the next discount level or remains at the 60% discount
level.
1 25% 75%  If one claim is made during the year, the policyholder
moves down by one level or remains at the 0% discount
2 40% 60% level.
 If two or more claims are made during the year the
3 60% 40% policyholder moves down two discount levels or remains at
the 0% discount level.
𝑝01 𝑝12 𝑝23  p00 p01 0 0 
 
 p10 0 p12 0 
0 1 2 3 P
𝑝00 𝑝10 𝑝21
 p20 p21 0 p23 
𝑝32 𝑝33
 
𝑝20 𝑝31  0 p31 p32 p33 

STQA4113/NM BONUS MALUS SYSTEM 16

 Transition Diagram & Matrix
- 5 by 5 NCD system
The rules for moving between levels are:
 If no claims are made during a year, the policyholder moves up to the next discount level or remains at the 60%
discount level.
 If one claim is made during the year, the policyholder moves down by one level or remains at the 0% discount
level.
 If two or more claims are made during the year the policyholder moves down two discount levels or remains at the
0% discount level.
𝑝01 𝑝12 𝑝23 𝑝34
𝑝44

0 1 2 3 4
𝑝00 𝑝10 𝑝21 𝑝32 𝑝43
 p00 p01 0 0 0 
𝑝20 𝑝31 𝑝42  
 p10 0 p12 0 0 
P   p20 p21 0 p23 0 
 
 0 p31 p32 0 p34 
 0 p44 
 0 p42 p43
STQA4113/NM BONUS MALUS SYSTEM 17
 Transition Diagram & Matrix
- 6 by 6 NCD system (Malaysia)
The rules for moving between levels are:
 If no claims are made during a year, the policyholder moves up to the next discount level or remains at the 60%
discount level.
 If one more claims are made during the year, the policyholder moves down to or remains at the 0% discount level.

𝑝01 𝑝12 𝑝23 𝑝34 𝑝45

𝑝55

0 1 2 3 4 5
𝑝00 𝑝10 𝑝20 𝑝30 𝑝50
𝑝40

 p00 p01 0 0 0 0 
 
 This set up is highly biased against good drivers, as a driver with only one claim is  p10 0 p12 0 0 0 
p 0 0 p23 0 0 
treated the same as a driver with two or more claims. P   20 
 Meaning the scheme is not ideal/suitable.  p30 0 0 0 p34 0 
 p40 0 0 0 0 p45 
 A driver causing an accident would not claim unless he believes that payback for loss  
would offset the possible loss of NCD  p50 0 0 0 0 p55 

STQA4113/NM BONUS MALUS SYSTEM 18

 Transition probability, 𝑝𝑖𝑗
 𝑝𝑖𝑗: probability of moving from state Example: 3 by 3 NCD system (scenario 1)
𝑖 to state 𝑗 𝑝01 𝑝12
𝑝00 𝑝22
 Pr(no claim) = 𝑝
0 1 2
 Pr one or more claims = 1 − 𝑝 = 𝑞 𝑝10 𝑝21

 p00 p01 0  q p 0  0.1 0.9 0 

    If 𝑝 = 0.9  P   0.1 0 
P   p10 0 p12    q 0 p  0.9 
 0 p22   0 p   0 0.1 0.9 
 p21 q  

STQA4113/NM BONUS MALUS SYSTEM 19

 Transition probability, 𝑝𝑖𝑗
Example: 3 by 3 NCD system (scenario 2) Example: 3 by 3 NCD system (scenario 3)
Given 𝑝 = 0.9, Pr one or more claims = 0.07 &
Given 𝑝 = 0.9 Pr(two or more claims ) = 0.03
𝑝01 𝑝12 𝑝01 𝑝12

0 1 2 0 1 2
𝑝00 𝑝10 𝑝21 𝑝22
𝑝00 𝑝22
𝑝10 𝑝20
𝑝20

 p00 p01 0   0.1 0.9 0   p00 p01 0   0.1 0.9 0 

 

P   p10 0
  
p12    0.1 0 0.9  P   p10 0 p12    0.1 0 0.9

 
p 
p22   0.03 0.07 0.9 
p
 20 0 p22   0.1 0 0.9   20 p21

STQA4113/NM BONUS MALUS SYSTEM 20

 Example: 4 by 4 NCD system (levels
Transition probability, 𝑝𝑖𝑗 of discount, 0%, 25%, 50%, 60%)

The number of claims made by a policyholder in a year is assumed to follow a Poisson distribution with
mean 0.30
𝑝01 𝑝12 𝑝23
0 1 2 3
𝑝00 𝑝10 𝑝21 𝑝32 𝑝33
 1  e e 0 0 
𝑝20 𝑝31   
 1  e 0 e 0 
e  n P
Pr  # claims   1  e    e  e 0 e 
n!   

 
e   0  0 1  e   e e e 
Pr  no claims    e    e  0.3
0!  0.259 0.741 0 0 
e 1  
Pr 1 claims  
0.259 0 0.741 0
  e    0.3e  0.3 P 
1!  0.037 0.222 0 0.741 
 
Pr  more than 1 claim   1  e  0.3  0.3e  0.3  0 0.037 0.222 0.741 
STQA4113/NM BONUS MALUS SYSTEM 21
 Stationary distribution
 An insured enters the system in the initial class when he applies for insurance
 At time 0 (transient state) :
  0
  0 , 1 ,  2 ,...,  n1   1,0,0,...,0 
%
 At time 𝑡:
 t 
  0 ,  1 ,  2 ,...,  n 1  &  0   1   2  ...   n 1  1
%
 At time 1:
1  0

 P

% %
A system is in a steady state if the variables, 𝜋𝑗 (called state variables) which define the
behaviour of the system are unchanging in time.
 P
% %
STQA4113/NM BONUS MALUS SYSTEM 22
 Example 𝑝00
𝑝01 𝑝12
𝑝22
- 3 by 3 NCD system (Scenario 0 𝑝10
1 𝑝21
2
1)
 For good driver, the probability of 1 accident in 1 year is 0.1  3   0.0190 0.1710 0.8100 
%
 0.1 0.9 0    4   0.0190 0.0981 0.8829 
  &      0 ,  1 ,  2   1, 0, 0  %
0
P   0.1 0 0.9   5   0.0117 0.1054 0.8829 
 0 0.1 0.9  % %
  

1  0  2 1  10   0.0110 0.0989 0.8901

 At time 1:   P  At time 2:   P %
% % % %  11   0.0110 0.0989 0.8901
%
 0.1 0.9 0   0.1 0.9 0  Steady state (stationary):

 1  1, 0, 0   0.1 0 0.9    2   0.1 0.9 0   0.1 0 0.9   0  0.0110,  1  0.0989 &  2  0.8901
% %  0 0.1 0.9 
 0 0.1 0.9    110 good drivers at level 0
 
  2   0.1 0.09 0.81
 1
  0.1 0.9 0 
10000 good
989 good drivers at level 1
% drivers
%
8901 good drivers at level 2

STQA4113/NM BONUS MALUS SYSTEM 23

 Example 𝑝00
𝑝01 𝑝12
𝑝22
- 3 by 3 NCD system (Scenario 0 𝑝10
1 𝑝21
2
1)
 For good driver, the probability of 1 accident in 1 year is 0.25

 0.25 0.75 0  Steady state:


P   0.25 0

0.75  &    0 ,  1 ,  2   1, 0, 0 
 0
 0  %  0  0.0769,  1  0.2308 &  2  0.6923
 0.25 0.75 
1  0
 At time 1:   P  3   0.1094 0.3281 0.5625 
% %0 % 385 bad drivers at level 0

 0.25 0.75    4   0.1094 0.2227 0.6680 

 1  1, 0, 0   0.25 0 0.75 
%
 5   0.0830 0.2490 0.6680 
5000 bad drivers 1154 bad drivers at level 1
%  0
 0.25 0.75  %
 3462 bad drivers at level 2
 1
  0.25 0.75 0 
%  11   0.0770 0.2309 0.6921
 2 1 %
 At time 2:   P  12   0.0769 0.2308 0.6923 
% % %
  2
  0.25 0.1875 0.5625 
%
STQA4113/NM BONUS MALUS SYSTEM 24
 Stationary distribution
 A stationary distribution: a probability distribution that remains unchanged as time
progresses into the future.
 It is represented by a row vector 𝝅 = 𝜋0 , 𝜋1 , 𝜋2 , … , 𝜋𝑘 , with
 𝜋𝑗 = σ𝑖 𝜋𝑖 𝑝𝑖𝑗
 0 ≤ 𝜋𝑗 ≤ 1
 σ 𝜋𝑗 = 1
 𝜋𝑗 = σ𝑖 𝜋𝑖 𝑝𝑖𝑗 can be written as 𝝅 = 𝝅𝑷
 Once the Markov Chain has reached the stationary state (steady state), its probability
distribution will stay stationary over time

STQA4113/NM BONUS MALUS SYSTEM 25

 Stationary distribution
 Using 𝝅 = 𝝅𝑷,
 p00 p01 0   p00 1  p00 0 
 0  1  2    0  1  2   p10 0
 
p12    0  1  2   p10 0

1  p10 
 0 p22   0  
 p21  p 21 1 p21 

 0   0 p00   1 p10 (1)  1 1  p10 

From (3) :  2   2 1  p21    1 1  p10    2 
p21
 1   0 1  p00    2 p21 (2) From (1) :  0 
 1 p10
1  p00 
 2   1 1  p10    2 1  p21  (3)
 0  1   2  1
 0  1   2  1 (4)  1 p10  1 1  p10 
 1  1
1  p00  p21
STQA4113/NM BONUS MALUS SYSTEM 26
  The number of claims made by a policyholder in a year is
Example – refer slide #21 assumed to follow a Poisson distribution with mean 0.30
 levels of discount: 0%, 25%, 50%, 60%.

 0.259 0.741 0 0 
𝝅 = 𝝅𝑷  
 0.259 0.741 0 0  0.259 0 0.741 0
  P 
0.259 0 0.741 0  0.037 0.222 0 0.741 
 0  1  2  3    0  1  2 3    
 0.037 0.222 0 0.741   0 0.037 0.222 0.741 
 
 0 0.037 0.222 0.741 
0.741
(1)  0  0.259 0  0.259 1  0.037 2 From (4) :  3   2  2.861 2
1  0.741
(2)  1  0.741 0  0.222 2  0.037 3
0.365
(3)  2  0.741 1  0.222 3 Substitute  3 into (3):  2  0.222  2.861 2   0.741 1   1   2  0.492 2
0.741
(4)  3  0.741 2  0.741 3 0.164
From (1) :  0   2  0.222 2
(5)  0   1   2   3  1 0.741
From (5) :0.222 2  0.492 2   2  2.861 2  1
4.575 2  1   2  0.219
 0  0.049,  1  0.108 &  3  0.627

STQA4113/NM BONUS MALUS SYSTEM 27

 Example
 A motor insurer operates a NCD system with the following levels of discount, 0%, 25%, 50%, 75%. The rule
governing a policyholder’s discount level based upon the number of claims made in the previous year is as
follows:
 Following a year with no claims, the policyholder moves up one level of discount or remains at the 75%
level.
 Following a year with one claim, the policyholder moves down one level of discount or remains at the 0%
level.
 Following a year with two or more claims, the policyholder moves down two discount levels (subject to a
limit of the 0% discount level).
 Policyholders at different discount levels are found to have different claims experience. The number of claims
made within a year is said to have a Poisson distribution with parameter , as follows
Discount level: 0% 25% 50% 75%
 0.29 0.22 0.18 0.10
a) Write down the transition matrix.
b) Calculate the ratio of policyholders at each discount level when the system reaches a steady state.

STQA4113/NM BONUS MALUS SYSTEM 28

 Solutions
 = 0.29  = 0.22

 At discount level 0% :  At discount level 25% :

p p
q 0 1 2 3 0 1 2 3

q
 Pr 𝑛𝑜 𝑐𝑙𝑎𝑖𝑚𝑠 = Pr 𝑋 = 0  Pr 𝑛𝑜 𝑐𝑙𝑎𝑖𝑚𝑠 = Pr 𝑋 = 0
= 𝑒 −0.29 = 0.7483 = 𝑒 −0.22 = 0.8025

 Pr #𝑐𝑙𝑎𝑖𝑚 ≥ 1 = 1 – 0.7483  Pr(#𝑐𝑙𝑎𝑖𝑚 ≥ 1) = 1 – 0.8025 = 0.1975

= 0.2517

STQA4113/NM BONUS MALUS SYSTEM 29

 Solutions
 = 0.18  = 0.10

 At discount level 50% :  At discount level 75% :

p
p
0 1 q1 2 3 0 1 2 3
q2 q1
q2
q
 Pr 𝑛𝑜 𝑐𝑙𝑎𝑖𝑚𝑠 = Pr 𝑋 = 0
= 𝑒 −0.18 = 0. 8353  Pr 𝑛𝑜 𝑐𝑙𝑎𝑖𝑚𝑠 = Pr 𝑋 = 0
= 𝑒 −0.10 = 0. 9048
 Pr #𝑐𝑙𝑎𝑖𝑚 = 1 = Pr 𝑋 = 1
= 0.18𝑒 −0.18 = 0.1503  Pr #𝑐𝑙𝑎𝑖𝑚 = 1 = Pr 𝑋 = 1
= 0.10𝑒 −0.10 = 0.0905
 Pr #𝑐𝑙𝑎𝑖𝑚 ≥ 2 = 1 – 0.8353 − 0.1503
= 0.0144  Pr #𝑐𝑙𝑎𝑖𝑚 ≥ 2 = 1 – 0.9048 − 0.0905
= 0.0047
STQA4113/NM BONUS MALUS SYSTEM 30
 Solution: Transition matrix
0% 25% 50% 75% 0% 25% 50% 75%

 p00 p01 p02 p03  0.2517 0.7483 0 0 

p p11 p12 p13  0.1975 
  0 0.8025 0
P
10 
 p20 p21 p22 p23  0.0144 0.1503 0 0.8353 
   
 p30 p31 p32 p33  0.0047 0 0.0905 0.9048 

Steady state:
0% 25% 50% 75% 0% 25% 50% 75%

 p00 p01 p02 p03   0.2517 0.7483 0 0 

p p11 p12 p13   0.1975 
0 0.8025 0
  P       
10

 p20 p21 p22 p23   0.0144 0.1503 0 0.8353 
   
 p30 p31 p32 p33   0.0047 0 0.0905 0.9048 

STQA4113/NM BONUS MALUS SYSTEM 31

0.2517 0  0.1975 1  0.0144 2  0.0047 3   0  (1)
0.7483 0  0.1503 2   1  (2)
0.8025 1  0.0905 3   2  (3)
0.8353 2  0.9048 3   3  (4)
 0  1   2   3  1  (5)
0.8353
From (4),  3   2  8.7742 2
0.0952
From (3),  2  0.8025 1  0.0905  8.7742 2
0.8025
2   1  3.8969 1
0.2059
From (2),  1  0.7483 0  0.1503  3.8969 1
0.7483
1   0  1.8062 0
0.4143
STQA4113/NM BONUS MALUS SYSTEM 32
Steady state:

 0  1   2   3  1
1.8062 0  3.8969 1.8062 0  8.7742 1.8062 0  1
 0  1.8062 0  7.0386 0  61.7579 0  1
71.6027 0  1
  0  0.0140,  1  0.0252,  2  0.0983,  3  0.8625

STQA4113/NM BONUS MALUS SYSTEM 33

  0.1 0.9 0   0.25 0.75 0 
 
PG   0.1 0 0.9  & PB   0.25 
Good Driver vs Bad Driver  0 0.1 0.9 

 0
0 0.75 
0.25 0.75 
  

 A bad driver causes accident 2.5 times as often as a good Proportion of drivers (steady
state) Average
Pr(accident)
driver Full 20% 40% Premium
premium discount discount
 Total expected # of accidents per year = 2250 (5/9 are 0.05 0.0026 0.0499 0.9475 0.6110
caused by bad driver) 0.1 0.0110 0.0989 0.8901 0.6242

 On the average, good drivers would pay about 62.42% of the 0.15
0.2
0.0258
0.0476
0.1461
0.1905
0.8281
0.7619
0.6395
0.6571
full premium, while bad drivers pay 67.69% of the full 0.25 0.0769 0.2308 0.6923 0.6769
premium (probability of causing an accident by a bad driver is 0.3 0.1139 0.2658 0.6203 0.6987
0.35 0.1586 0.2945 0.5469 0.7223
2.5 times higher) 0.4 0.2105 0.3158 0.4737 0.7474
 In aggregate, good drivers pay 64.84% of the total premium 0.45 0.2691 0.3289 0.4020 0.7734
0.5 0.3333 0.3333 0.3333 0.8000
collected. This can be regarded as being highly unfair,
especially when the bad drivers cause 55.56% of the #
Full 20% 40% Total
premium discount discount premium
accidents
Good D 10000 110 989 8901 6242
 This kind of a setup is highly biased against the good drivers. Bad D 5000 385 1154 3462 3385
𝐴𝐺 = 1000 𝐴𝐵 = 1250 64.84%

STQA4113/NM BONUS MALUS SYSTEM 34

 Proportion of drivers (steady
state) Average
Possible variations Pr(accident)
Full 20% 40%
premium discount discount
Premium

0.05 0.0026 0.0499 0.9475 0.6110

0.1 0.0110 0.0989 0.8901 0.6242
1. Variation in the probability of accident between groups is 0.15 0.0258 0.1461 0.8281 0.6395
possibly not big enough 0.2 0.0476 0.1905 0.7619 0.6571

2. A driver causing an accident would not necessarily claim 0.25 0.0769 0.2308 0.6923 0.6769
0.3 0.1139 0.2658 0.6203 0.6987
unless the belief is that the reimbursement would offset the 0.35 0.1586 0.2945 0.5469 0.7223
possible loss of NCD 0.4 0.2105 0.3158 0.4737 0.7474
3. Severity of accidents (size of claims) should be brought into 0.45 0.2691 0.3289 0.4020 0.7734
0.5 0.3333 0.3333 0.3333 0.8000
consideration
4. Maybe there should be higher number of categories as well
as magnitude of discounts as well as the number of groups
of drivers
Average premium paid increases with the increase in Size of the
probability of accident (refer table), however not at any claim to be
considered
alarming rate.
The stationary distributions of policyholders across the three
levels does not depend on the discount levels.

STQA4113/NM BONUS MALUS SYSTEM 35

 Assumptions in addition to the ones made for the
example above:
 Full premium is RM10000
 Claims size has a log-normal distribution with  = 7 &  = 2.5
 Time value of money is ignored
 All drivers (following an accident) believe that they will not have another accident in the next two years
Based on these assumptions, the decision to make a claim or not shall be dependent on the following two
factors:
 Size of claims
 Level to which the driver currently belongs Potential loss for making a claim (the loss of NCD)
Premium next Premium the
Class Action NCD loss
year year after
When to Claim 10000 8000
claim? 0 4000
Do not claim 8000 6000

A claim shall be made only if the amount of Claim 10000 8000

1 6000
claim is greater than the loss of NCD; otherwise Do not claim 6000 6000
it might be better not to claim Claim 8000 6000
2 2000
Do not claim 6000 6000

STQA4113/NM BONUS MALUS SYSTEM 36

 Probability
 Claims size has a log-normal distribution with  = 7 &  = 2.5

Pr Claim | Accident   Pr  Damage  NCD Loss 

 ln  NCD Loss    
 Pr ln  Damage   ln  NCD Loss    Pr  Z  
  
Class NCD Loss 𝑍 Pr[𝐶𝑙𝑎𝑖𝑚|𝐴𝑐𝑐𝑖𝑑𝑒𝑛𝑡]
0 4000 0.51762 0.3024
1 6000 0.67981 0.2483
2 2000 0.24036 0.4050

Set up new transition matrix and obtain stationary distribution under the assumption that the
claim will be made only if the amount of claim is greater than the loss of NCD suffered by
making the claim
STQA4113/NM BONUS MALUS SYSTEM 37
 New transition matrix
Pr Claim | Good Driver   Pr  Accident | Good Driver   Pr Claim | Accident 
 Transition matrix:  ln x   
X ~ lognormal ( , )  Pr( X  x)    
a b c  a b 0   
PNew

 d e
 
f   d 0 f
 a  0.1 Pr  X  NCDLoss0   0.1 0.3024  0.0302
g i   0 h i 
 h b  0.9  0.1 Pr  X  NCDLoss0 
 0.9  0.1 1  0.3024   0.9698
d  0.1 Pr  X  NCDLoss1   0.1 0.2483  0.0248
f  0.9  0.1 Pr  X  NCDLoss1   0.9  0.1 1  0.2483   0.9752
h  0.1 Pr  X  NCDLoss2   0.1 0.4050  0.0405
i  0.9  0.1 Pr  X  NCDLoss2   0.9  0.1 1  0.4050   0.9595
STQA4113/NM BONUS MALUS SYSTEM 38
 Good driver: new transition matrix & steady state
Pr Claim | Good Driver   Pr  Accident | Good Driver   Pr Claim | Accident 

 Transition matrix:
 0.1 0.3024 1  0.1 0.3024 0   0.0302 0.9698 0 
   
PG   0.1 0.2483 0 1  0.1 0.2483    0.0248 0 0.9752 
 0.1 0.4050 1  0.1 0.4050   0 0.0405 0.9595 
 0

 Stationary distribution, 𝜋 = 𝜋𝑃:  0  0.0302 0  0.0248 1 (1)

 0.0302 0.9698 0   1  0.9698 0  0.0405 2 (2)
 
 0  1  2    0  1  2   0.0248 0 0.9752 
 2  0.9752 1  0.9595 2 (3)
 0.9595 
 0 0.0405
 0  1   2  1
  0  0.0010 ,  1  0.0398,  2  0.9591

STQA4113/NM BONUS MALUS SYSTEM 39

 Bad driver: new transition matrix & steady state
Pr Claim | Bad Driver   Pr  Accident | Bad Driver   Pr Claim | Accident 

 Transition matrix:
 0.25  0.3024 0.75  0.3024 0   0.0756 0.9244 0 
   
PG   0.25  0.2483 0 0.75  0.2483    0.0621 0 0.9379 
 0.25  0.4050 0.75  0.4050   0 0.1013 0.8987 
 0

 Stationary distribution, 𝜋 = 𝜋𝑃:  0  0.0756 0  0.0621 1 (1)

 0.0756 0.9244 0   1  0.9244 0  0.1013 2 (2)
 
 0  1  2    0  1  2   0.0621 0 0.9379 
 2  0.9379 1  0.8987 2 (3)
 0.8987 
 0 0.1013
 0  1   2  1
  0  0.0065,  1  0.0968,  2  0.8967

STQA4113/NM BONUS MALUS SYSTEM 40

Results
10000 Good Driver 5000 Bad Driver

Class 0 Class 1 Class 2 Class 0 Class 1 Class 2

110 989 8901 385 1154 3385

10 399 9591 33 484 4483

Expected Premium  10000 0  8000 1  6000 2

STQA4113/NM BONUS MALUS SYSTEM 41

 Assignment
1 In a No Claims Discount system for health insurance, there are three discount levels, 0%, 20% and
30%. The full annual premium is RM190.
If no claims are made during a year, the policyholder moves up to the next higher level of discount,
or stays at 30%. If one or more claims are made, the policyholder moves down to 0% discount, or
stays at 0%. The probability that a policyholder has one accident in a year is 0.1, and the probability
of more than one accident is negligible. In the event of hospitalisation, the natural logarithm of the
loss has a 𝑁(𝜇, 𝜎2) distribution, with 𝜇 = 4.5 and 𝜎2 = 0.84. The policyholder will only make a
claim when hospitalisation occurs if the loss is greater than the total extra premiums that would
have to be paid over the infinite time horizon, assuming no further hospitalisation occur.
(i) Determine the transition matrix for the Markov chain X1 , X2 , . where Xn is the discount level at
the beginning of year n.
(ii) For a policyholder who pays the full premium in the current year, calculate the expected
premium at the beginning of the next year.

STQA4113/NM BONUS MALUS SYSTEM 42

 Assignment
2 AIG operates a No Claims Discount system for its motor insurance business, with discount levels 0%,
15%, 30% and 50%. The full annual premium is RM500. The rules for moving between discount levels
are:
 If no claims are made during a year, the policyholder moves to the next higher level of discount or
remains at the maximum discount level.
 If one or more claims are made during a year, a policyholder at the 30% or 50% discount level moves
to the 15% discount level and a policyholder at the 0% or 15% discount level moves to, or remains at,
the 0% discount level.
When an accident occurs, the distribution of the loss is exponential with mean RM1,000. In the event of
an accident, a policyholder will claim only if the loss is greater than the total extra premiums (NCD loss)
that would have to be paid over the next three years, assuming that no further accidents occur.
For each discount level, calculate:
(i) the smallest loss for which a policyholder will make a claim.
(ii) the probability of a claim being made in the event of an accident occurring.

STQA4113/NM BONUS MALUS SYSTEM 43

 Assignment
3 Based on AmGeneral’s experience, for each policyholder the probability of having an accident during the year is 0.2.
The insurer operates a no claims discount system {0%, 20% and 50%}. In the event of a claim free year, the
policyholder moves to the next higher level of discount. In the event of a claim during the year, the policyholder
moves in the next year to the next lower level of discount, unless the claim was as a result of drunken behaviour. In
this case, the policyholder moves in the next year to the lowest level of discount. 25% of claims are due to drunken
behaviour. The insurer charges a premium of RM2,500 for policyholders at the 0% discount level. A policyholder
makes a claim following an accident if the cost of repairs is greater than the additional premiums payable at the next
two renewals, assuming no further claims are made.
(i) Calculate the cost of a repair below which the policyholder will not claim, for each level of discount.
(ii) Assuming that the cost of each repair follows a lognormal distribution with parameters  = 6.5 and 2 = 3.5,
calculate the probability that a policyholder makes a claim in the event of an accident, at each level of discount.
(iii) Write down the transition matrix.
(iv) Derive the proportion of policyholders at each level of discount, once a steady state has been reached.

STQA4113/NM BONUS MALUS SYSTEM 44