You are on page 1of 14

986 51.

BÜHLMANN CREDIBILITY: BASICS

Coverage of this material in the three syllabus options


This material is required. It is covered in all three syllabus reading options.

Exercises

51.1. Aggregate losses for various risks in a portfolio have the following probability distribution:

Number of Loss Loss


Risks Amount Probability Amount Probability
Risk 1 60 100 0.9 500 0.1
Risk 2 30 200 0.9 800 0.1
Risk 3 15 200 0.8 1000 0.2
Risk 4 10 300 0.6 600 0.4
Risk 5 5 300 0.4 700 0.6
Determine the expected value of the process variance for this portfolio of risks.

51.2. Two urns have balls with numbers. The number of balls with each number is as follows:

Number on ball 0 1 2 x
Urn 1 5 3 1 1
Urn 2 2 2 4 4
x is a positive number.
Urn 1 is twice as likely to be selected as urn 2. The variance of the hypothetical means of the values of
the balls in the urns is 0.5.
Determine x.

51.3. [4B-S90:35] (1 point) The underlying expected loss for each individual insured is assumed to be
constant over time. The Bühlmann credibility assigned to the pure premium for an insured observed for
one year is 12 .
Determine the Bühlmann credibility to be assigned to the pure premium for an insured observed for
3 years.

(A) 1/2 (B) 2/3 (C) 3/4 (D) 6/7


(E) Cannot be determined

51.4. [4B-S91:25] (2 points) Assume that the expected pure premium for an individual insured is constant
over time. The Bühlmann credibility for two years of experience is equal to 0.40.
Determine the Bühlmann credibility for three years of experience.
(A) Less than 0.500
(B) At least 0.500, but less than 0.525
(C) At least 0.525, but less than 0.550
(D) At least 0.550, but less than 0.575
(E) At least 0.575

C/4 Study Manual—17th edition Exercises continue on the next page . . .


Copyright ©2014 ASM
EXERCISES FOR LESSON 51 987

51.5. [4B-F92:19] (1 point) You are given the following:


1
• The Bühlmann credibility of an individual risk’s experience is 3 based upon 1 observation.
• The risk’s underlying expected loss is constant.
Determine the Bühlmann credibility for the risk’s experience after four observations.

(A) 1/4 (B) 1/2 (C) 2/3 (D) 3/4


(E) Cannot be determined

51.6. [4B-S93:3] (1 point) You are given the following:


X is a random variable with mean m and variance v.
m is a random variable with mean 2 and variance 4.
v is a random variable with mean 8 and variance 32.
Determine the value of the Bühlmann credibility factor Z after three observations of X.
(A) Less than 0.25
(B) At least 0.25, but less than 0.50
(C) At least 0.50, but less than 0.75
(D) At least 0.75, but less than 0.90
(E) At least 0.90

51.7. [4B-F93:25] (1 point) You are given the following:


(i) A random sample of losses taken from policy year 1992 has sample variance s 2  16.
(ii) The losses are sorted into 3 classes, A, B, and C, of equal size.
(iii) The sample variances for each of the classes are:
2
sA 4 s B2  5 s C2  6

Estimate the variance of the hypothetical means.


(A) Less than 4
(B) At least 4, but less than 8
(C) At least 8, but less than 12
(D) At least 12, but less than 16
(E) At least 16

51.8. [4B-S95:2] (2 points) You are given the following:


• The Bühlmann credibility of three observations is twice the credibility of one observation.
• The expected value of the process variance is 9.
Determine the variance of the hypothetical means.

(A) 3 (B) 4 (C) 6 (D) 8 (E) 9

C/4 Study Manual—17th edition Exercises continue on the next page . . .


Copyright ©2014 ASM
988 51. BÜHLMANN CREDIBILITY: BASICS

51.9. [4B-S95:16] (1 point) Which of the following will DECREASE the Bühlmann credibility of the cur-
rent observations?
1. Decrease in the number of observations
2. Decrease in the variance of the hypothetical means
3. Decrease in the expected value of the process variance

(A) 1 (B) 2 (C) 3 (D) 1,2 (E) 1,3

51.10. [4B-F95:2] (1 point) The Bühlmann credibility of five observations of the loss experience of a single
risk is 0.29.
Determine the Bühlmann credibility of two observations of the loss experience of this risk.
(A) Less than 0.100
(B) At least 0.100, but less than 0.125
(C) At least 0.125, but less than 0.150
(D) At least 0.150, but less than 0.175
(E) At least 0.175

51.11. [4B-S96:3] (1 point) Given a first observation with a value of 2, the Bühlmann credibility estimate
for the expected value of the second observation would be 1. Given a first observation with a value of 5,
the Bühlmann credibility estimate for the expected value of the second observation would be 2.
Determine the Bühlmann credibility of the first observation.

(A) 1/3 (B) 2/5 (C) 1/2 (D) 3/5 (E) 2/3

51.12. [4B-F96:10] (2 points) The Bühlmann credibility of n observations of the loss experience of a single
risk is 1/3. The Bühlmann credibility of n + 1 observations of the loss experience of this risk is 2/5.
Determine the Bühlmann credibility of n + 2 observations of the loss experience of this risk.

(A) 4/9 (B) 5/11 (C) 1/2 (D) 6/11 (E) 5/9

51.13. [4B-F96:20] (3 points) You are given the following:


• The number of claims for a single risk follows a Poisson distribution with mean θµ.
• θ and µ have a prior probability distribution with joint density function

f ( θ, µ )  1, 0 < θ < 1, 0 < µ < 1.

Determine the value of Bühlmann’s k.


(A) Less than 5.5
(B) At least 5.5, but less than 6.5
(C) At least 6.5, but less than 7.5
(D) At least 7.5, but less than 8.5
(E) At least 8.5

51.14. 50 observations of experience are made under an insurance coverage. Bühlmann credibility of
75% is given to this experience.
Determine the least number of observations needed to obtain 90% credibility.

C/4 Study Manual—17th edition Exercises continue on the next page . . .


Copyright ©2014 ASM
EXERCISES FOR LESSON 51 989

51.15. 50 observations of experience are made under an insurance coverage. Initially, 80% credibility is
given to this experience. You then change the assumptions for this coverage so that the variance of the
hypothetical means is doubled.
Determine the revised Bühlmann credibility assigned to the experience.

51.16. [4B-F97:5] (2 points) You are given the following:


• A portfolio of independent risks is divided into two classes.
• Each class contains the same number of risks.
• The claim count probabilities for each risk for a single exposure period are as follows:

Class Probability of 0 Claims Probability of 1 Claim


1 1/4 3/4
2 3/4 1/4
• All claims incurred by risks in Class 1 are of size u.
• All claims incurred by risks in Class 2 are of size 2u.
A risk is selected at random from the portfolio.
Determine the Bühlmann credibility for the pure premium of one exposure period of loss experience
for this risk.
(A) Less than 0.05
(B) At least 0.05, but less than 0.15
(C) At least 0.15, but less than 0.25
(D) At least 0.25, but less than 0.35
(E) At least 0.35

51.17. [4B-S98:2] (1 point) You are given the following:


• The number of claims for a single insured follows a Poisson distribution with mean λ.
• λ varies by insured and follows a Poisson distribution with mean µ.
Determine the value of Bühlmann’s k.

(A) 1 (B) λ (C) µ (D) λ/µ (E) µ/λ

51.18. [4B-F98:19] (2 points) You are given the following:


• Claim sizes follow a gamma distribution with parameters α and θ  0.5.
• The prior distribution of α is assumed to be uniform on the interval (0, 4).
Determine the value of Bühlmann’s k for estimating the expected value of a claim.

(A) 2/3 (B) 1 (C) 4/3 (D) 3/2 (E) 2

C/4 Study Manual—17th edition Exercises continue on the next page . . .


Copyright ©2014 ASM
990 51. BÜHLMANN CREDIBILITY: BASICS

Use the following information for questions 51.19 and 51.20:

You are given the following:


• Claim sizes follow a Pareto distribution with parameters θ and α  3.
• The prior distribution of θ has density function

f ( θ )  e −θ , 0 < θ < ∞.

51.19. [4B-S99:5] (2 points) Determine the expected value of the process variance.

(A) 3/8 (B) 3/4 (C) 3/2 (D) 3 (E) 6

51.20. [4B-S99:6] (2 points) Determine the variance of the hypothetical means.

(A) 1/4 (B) 1/2 (C) 1 (D) 2 (E) 4

51.21. [4B-F99:18] (2 points) You are given the following:


• Partial Credibility Formula A is based on the methods of limited fluctuation credibility, with 1,600
expected claims needed for full credibility.
• Partial Credibility Formula B is based on Bühlmann’s credibility formula with a k of 391.
• One claim is expected during each period of observation.
Determine the largest number of periods of observation for which Partial Credibility Formula B yields
a larger credibility value than Partial Credibility Formula A.
(A) Less than 400
(B) At least 400, but less than 800
(C) At least 800, but less than 1,200
(D) At least 1,200, but less than 1,600
(E) At least 1,600

51.22. For a disability coverage, there are 3 rating classes. The number of insureds in each class, and the
mean frequency of claims in each class, is as follows:
Rating Class Number of Insureds Mean Frequency of Claims
A 400 0.20
B 300 0.30
C 300 0.35
Claim frequency for each insured follows a Poisson distribution, with a mean equal to the mean fre-
quency of claims for the class.
An insured is selected at random.
Determine the Bühlmann credibility given to 10 observations from this insured.

C/4 Study Manual—17th edition Exercises continue on the next page . . .


Copyright ©2014 ASM
EXERCISES FOR LESSON 51 991

51.23. [4-S00:3] You are given the following information about two classes of business, where X is the
loss for an individual insured:

Class 1 Class 2
Number of insureds 25 50
E[X] 380 23
E[X 2 ] 365,000 —
You are also given than an analysis has resulted in a Bühlmann k value of 2.65.
Calculate the process variance for Class 2.

(A) 2,280 (B) 2,810 (C) 7,280 (D) 28,320 (E) 75,050

51.24. [4-S01:6] You are given:


(i) The full credibility standard is 100 expected claims.
(ii) The square-root rule is used for partial credibility.
You approximate the partial credibility formula with a Bühlmann credibility formula by selecting a
Bühlmann k value that matches the partial credibility formula when 25 claims are expected.
Determine the credibility factor for the Bühlmann credibility formula when 100 claims are expected.

(A) 0.44 (B) 0.50 (C) 0.80 (D) 0.95 (E) 1.00

51.25. [C-S05:11] You are given:


(i) The number of claims in a year for a selected risk follows a Poisson distribution with mean λ.
(ii) The severity of claims for the selected risk follows an exponential distribution with mean θ.
(iii) The number of claims is independent of the severity of claims.
(iv) The prior distribution of λ is exponential with mean 1.
(v) The prior distribution of θ is Poisson with mean 1.
(vi) A priori, λ and θ are independent.
Using Bühlmann’s credibility for aggregate losses, determine k.

(A) 1 (B) 4/3 (C) 2 (D) 3 (E) 4

51.26. You are given the following:


(i) Losses for a given policyholder follow a gamma distribution with parameters α and θ. θ does
not vary by policyholder.
(ii) α is a random variable with mean 10.
(iii) The Bühlmann credibility of one observation is 0.25.
Determine Var ( α ) .

(A) 5/3 (B) 20/9 (C) 5/2 (D) 3 (E) 10/3

C/4 Study Manual—17th edition Exercises continue on the next page . . .


Copyright ©2014 ASM
992 51. BÜHLMANN CREDIBILITY: BASICS

Use the following information for questions 51.27 and 51.28:

You are given the following:


• For an individual risk in a population, the number of claims for a single exposure period follows a
Poisson distribution with parameter λ.
• For the population, λ is distributed according to an exponential distribution:

h ( λ )  5e −5λ λ > 0.

• An individual risk is randomly selected from the population.


• After two exposure periods, one claim has been observed.

51.27. [4B-F94:25] (2 points) For the selected risk, determine the expected value of the process variance.

(A) 0.04 (B) 0.20 (C) 0.29 (D) 5.00 (E) 25.00

51.28. [4B-F94:26] (3 points) Determine the density function of the posterior distribution of λ for the
selected risk.

(A) 7e −7λ (B) 5λe −7λ (C) 49λe −7λ (D) 108λ 2 e −6λ (E) 270λ 2 e −6λ

Additional released exam questions: C-S07:6,21

Solutions
51.1. This exercise is mostly a showcase of the Bernoulli shortcut.

v1  (0.9)(0.1)(500 − 100) 2  14,400


v2  (0.9)(0.1)(800 − 200) 2  32,400
v3  (0.8)(0.2)(1000 − 200) 2  102,400
v4  (0.6)(0.4)(600 − 300) 2  21,600
v5  (0.4)(0.6)(700 − 300) 2  38,400
1
60 (14,400) + 30 (32,400) + 15 (102,400) + 10 (21,600) + 5 (38,400)  31,500
 
v 120

51.2. Let µ i be the hypothetical mean of urn i. Then

µ1  0.5 (0) + 0.3 (1) + 0.1 (2) + 0.1x  0.5 + 0.1x


µ2  16 (0) + 16 (1) + 26 (2) + 26 ( x )  5
6 + 13 x

The variance of these two hypothetical means is expressed using the Bernoulli shortcut.
2
1 2 1 7 2
 !
2
a  ( µ1 − µ2 ) ( )( )  + x  0.5
3 3 3 30 9

Now we solve for x.


2
1 7
+  2.25

3 30 x

C/4 Study Manual—17th edition


Copyright ©2014 ASM
EXERCISE SOLUTIONS FOR LESSON 51 993

1 7
3 + 30 x  1.5
10 + 7x  45
35
x 7  5

1
51.3. First we back out the Bühlmann k from Z  2 when n  1:

n 1
Z 
n+k 2
1 1

1+k 2
k1

Then we calculate Z when n  3.


n 3 3
Z   (C)
n+k 3+1 4

51.4. The Bühlmann credibility factor Z  n+k .


n
Here we are given that n  2 and Z  0.40, so

2
 0.4
2+k
k3

If we now set n  3, then Z  3/ (3 + 3)  0.5 . (B)


One of the rare cases where the answer is at the bound of a range.
51.5. The Bühlmann credibility factor Z  n+k .
n
Here we ar given that n  1 and Z  13 , so

1 1

1+k 3
k2

4
If we now set n  4, then 4+2  2
3
. (C)
51.6. m is the hypothetical mean of X and v is the process variance. The expected value of the process
variance is the expected value of v, or 8. The variance of the hypothetical means is the variance of m, or
4. Bühlmann’s k  va  48  2. Then

n 3
Z   0.6 (C)
n+k 3+2

51.7. The total variance is 16 and the expected value of the process variances is (1/3)(4 + 5 + 6)  5. The
expected value of the process variances plus the variance of the hypothetical means adds up to the total
variance, by the conditional variance formula:

Var ( X )  E[Var ( X | I ) ] + Var (E[X | I])  EPV + VHM

So the variance of the hypothetical means is 16 − 5  11 . (C)

C/4 Study Manual—17th edition


Copyright ©2014 ASM
994 51. BÜHLMANN CREDIBILITY: BASICS

51.8. Since Z  n/ ( n + k ) , the first condition implies

3 1
!
2
3+k 1+k
3 + 3k  6 + 2k
k3

Since k  v/a, the second condition implies

9
3
a
a 3 (A)

n k
51.9. Z   1− . Decreasing n will decrease the denominator, increase the fraction, and
n+k n+k
decrease 1 minus the fraction, so the first statement is true. (It is also intuitively obvious.)
n
Z  . Decreasing a will increase k, increase the denominator, and decrease the fraction, so
n + v/a
the second statement is true. Decreasing v will decrease k, decrease the denominator, and increase the
fraction, so the third statement is false. (D)
51.10. As usual, we back out k from Z  n/ ( n + k ) . Here we are given that Z  0.29 and n  5.

5
 0.29
5+k
5 − 1.45 3.55
k 
0.29 0.29
2 0.58
Z   0.140 (C)
2 + 3.55/0.29 4.13

51.11. Bühlmann credibility is a linear function of the mean of the observations; in fact

PC  Z X̄ + (1 − Z ) µ

Here we are given

1  2Z + (1 − Z ) µ
2  5Z + (1 − Z ) µ

Subtracting the first equation from the second, we see that 3Z  1, so Z  1/3 . (A)
51.12. From n observations,
n 1
 ⇒ n + k  3n
n+k 3
From n + 1 observations,
n+1 2

n+1+k 5
Substituting n + k  3n and solving for n,

n+1 2

3n + 1 5
5n + 5  6n + 2

C/4 Study Manual—17th edition


Copyright ©2014 ASM
EXERCISE SOLUTIONS FOR LESSON 51 995

n3
k6
Calculating the credibility factor for n + 2  5 observations,

5 5
 (B)
5+6 11

51.13. In this exercise, the classes are continuous rather than discrete. Since the number of claims is Pois-
son, the hypothetical mean is θµ and process variance is also v ( θ, µ )  θµ. Both θ and µ are uniformly
distributed on [0, 1] and are independent, so the expected value of the process variance is the product of
the expected values of θ and µ:
1 1 1
! !
v  E[v ( θ, µ ) ]  E[θ] E[µ]  
2 2 4
The variance of the hypothetical means can be calculated by calculating the expected value of the squares
of the hypothetical means and then subtracting the square of the overall expected value:
a  E[θ 2 µ2 ] − E[θµ]2
For a uniform distribution, the second moment is 31 , and by independence the expected value of the prod-
uct is the product of the expected values, so
1 1 1
! !
E[θµ]  E[θ] E[µ]  
2 2 4
1 1 1
! !
2 2 2 2
E[θ µ ]  E[θ ] E[µ ]  
3 3 9
!2
1 1 7
a − 
9 4 144
v 144 36
k    5.14 (A)
a 4·7 7
51.14. First we back out k.
50 12.5 50
Z  0.75 ⇒ 37.5 + 0.75k  50 ⇒ k  
50 + k 0.75 3
Now we solve for n such that Z  0.9.
n
Z  0.9 ⇒ 0.9n + 15  n ⇒ n  150
n + 50
3

51.15.
50
 0.8
50 + v/a
v
 12.5
a
v
 6.25
2a
50 8

50 + 6.25 9

C/4 Study Manual—17th edition


Copyright ©2014 ASM
996 51. BÜHLMANN CREDIBILITY: BASICS

51.16. Let µ i be the hypothetical mean of Class i and v i the process variance of Class i.

µ1  34 u
µ2  24 u

We use the Bernoulli shortcut to calculate the variance of the hypothetical means.
2
1 1 3 2 u2
! !
a u− u 
2 2 4 4 64
We also use the Bernoulli shortcut to calculate the process variance of each class.

1 3 2 3 2
! !
v1  u  u
4 4 16
3 1 12 2
! !
v2  4u 2  u
4 4 16
1 3 2 12 2 15 2
 
v u + u  u
2 16 16 32
v 15
k  (64)  30
a 32
1 1
Z   0.032 (A)
1 + 30 31
51.17. λ is both the hypothetical mean and the process variance, since for a Poisson the two are the same.
Therefore,

a  Var ( λ )  µ
v  E[λ]  µ
µ
k  1 (A)
µ

51.18.
1 16 1
! !
1
a  Var (0.5α )  Var ( α ) 
4 
4 12 3
1
v  E[0.25α]  0.25 E[α] 
2
1/2 3
k  (D)
1/3 2

51.19. The process variance is the variance of a Pareto, or

Var ( X | θ )  E[X 2 | θ] − E[X | θ]2


!2
2θ 2 θ
 −
( α − 1)( α − 2) α−1
θ 2 3θ 2
 θ2 − 
4 4
The prior distribution of θ is exponential with mean 1, whose second moment is 2, so the expected value
of 3θ 2 /4 is 3 (2) /4  3/2 . (C)

C/4 Study Manual—17th edition


Copyright ©2014 ASM
EXERCISE SOLUTIONS FOR LESSON 51 997

51.20. The hypothetical mean is θ/ ( α − 1)  θ/2. The variance of θ, which is exponential with mean 1,
is 1, so the variance of θ/2 is 1/4 . (A)

51.21. By formula A, Z  n/1600, where n is the number of expected claims. By formula B, Z 
n/ ( n + 391) .

n n
r
>
n + 391 1600
n2 n
>
( n + 391) 2 1600
1600n > ( n + 391) 2  n 2 + 782n + 3912
n 2 − 818n + 3912 < 0

818 + 8182 − 4 · 3912 818 + 240
n   529 (B)
2 2
Note that the lower solution for n, 289, is the beginning of the range for which formula B gives larger
credibility than formula A. Formula B gives less credibility below 289, more between 289 and 529, and
less above 529.
51.22.

v  0.4 (0.2) + 0.3 (0.3) + 0.3 (0.35)  0.275


µ  0.275
E[µ (Θ) ]  0.4 (0.22 ) + 0.3 (0.32 ) + 0.3 (0.352 )  0.07975
2

a  0.07975 − 0.2752  0.004125


v 0.275
k 
a 0.004125
10
Z  0.1304
10 + 0.275/0.004125

51.23. There is a 1/3 probability of Class 1 and a 2/3 probability of Class 2. The variance of the hypo-
thetical means is
1 2
! !
a (380 − 23) 2  28,322
3 3
and k  v/a, so
v
2.65 
28,322
v  75,053.3

However, the process variance for Class 1 is 365,000 − 3802  220,600, so

1 2
(220,600) + PV2  75,053.3
3 3
PV2  2279.95 (A)

C/4 Study Manual—17th edition


Copyright ©2014 ASM
998 51. BÜHLMANN CREDIBILITY: BASICS

q
25
51.24. The square-root rule gives Z  100  12 . Then

25 1

25 + k 2
k  25
100
 0.8 (C)
100 + k
51.25. Let S be the random variable for aggregate losses, and as usual N will be the frequency random
variable and X the severity random variable. For this compound process, by the formulas for mean and
variance of a compound Poisson distribution (see equation (14.4) for the latter)
µ ( λ, θ )  E[S | λ, θ]  E[N | λ, θ] E[X | λ, θ]  λθ
v ( λ, θ )  Var ( S | λ, θ )  λ E[X 2 ]  2λθ 2
To calculate the variance of the hypothetical mean λθ, we will calculate the first and second moments
of λθ. Since λ and θ are independent, the moment of the product is the product of the moments. The
distribution of λ is exponential with mean 1, so the second moment of λ is twice its mean squared (see the
distribution tables) or 2 (12 )  2. The distribution of θ is Poisson, so its variance equals its mean, which is
1, and its second moment is the sum of its variance and the square of its mean, or 1 + 12  2. Therefore
E[λθ]  E[λ] E[θ]  1
E ( λθ ) 2  E λ 2 E θ 2  (2)(2)  4
f g f g f g

a  Var ( λθ )  4 − 12  3
The expected value of the process variance v ( λ, θ ) is
v  E[2λθ2 ]  2 E[λ] E[θ 2 ]  (2)(1)(2)  4

Bühlmann’s k is 4/3 . (B)


51.26. The hypothetical mean is αθ and the process variance is αθ 2 . We back out Var ( α ) .
v  E[αθ 2 ]  10θ 2
a  Var ( αθ )  θ 2 Var ( α )
1
0.25  Z 
1+k
10θ 2
k3 2
θ Var ( α )
10
Var ( α )  (E)
3

51.27. The process variance is λ, and its expectation is the mean of the exponential distribution, or 0.20 .
(B)
51.28. α  1, γ  5. α ∗  1 + 1  2, γ∗  5 + 2  7. The constant of the posterior gamma is
1 72
  49
Γ ( α ) θ α Γ (2)

Posterior density is 49λe −7λ . (C)

C/4 Study Manual—17th edition


Copyright ©2014 ASM
QUIZ SOLUTIONS FOR LESSON 51 999

Quiz Solutions
51-1. The hypothetical mean is µ ( M )  0.2M and the process variance is v ( M )  0.2 (0.8) M  0.16M.
The Poisson is shifted by 1, so its mean is E[M]  λ + 1  2 and its variance is Var ( M )  λ  1, since
shifting doesn’t affect variance. Then

v  E[0.16M]  0.32
a  Var (0.2M )  0.04
v 0.32
k   8
a 0.04

C/4 Study Manual—17th edition


Copyright ©2014 ASM

You might also like