Hello, dear friend, you can consult us at any time if you have any questions, add WeChat: daixieit

MAT00061M

MMath and MSc Examinations 2021/22

Statistics for Insurance

1 (of 4) .    (a)    In  an  insurance  company,  two separate  Poisson surplus  processes  are

being managed, where the respective parameters are given in the following table .  In the table, U denotes the initial reserve, A denotes the Poisson parameter, 9 is the security loading and x is the typical claim size variable .

Table:  Parameters for the two Poisson surplus processes

Process     U      A       9              Distribution of x

A        500   20   0 .25    Exponential with mean 100

B        600   20   0 .30

Γ(2, 0

02)

(i)    For Process A, write down the adjustment equation and calculate the adjustment coefficient .                                                             [5]

(ii)    For Process B, write down the adjustment equation and calculate the adjustment coefficient .                                                             [5]

(iii)    Use the Lundberg upper bound as an approximation for the probability

of ruin, and rank the two processes A and B .                               [4]

(b)   The aggregate claim process in a company is modelled by a compound Poisson surplus process, where the Poisson parameter is A = 20, initial reserves are U = 1200, a typical claim is exponentially distributed with mean 50 . The insurance company is considering buying proportional rein- surance where a portion a of every claim is retained .  Let 9 = 0.2 be the insurer’s loading factor and 5 = 0.3 be the reinsurer’s loading factor .

(i)   What is the minimum value of a that the insurance company should consider?                                                                                        [2]

(ii)    Derive the adjustment coefficient for this process and write it as a

function of a .                                                                                 [6]

(a)    Let s = x1 +× × ×+xN  have a compound Poisson distribution, where A = 5

(b)   A portfolio of 200 one-year re insurance policies is summarised in the following table .

Table:  Number of policies by incidence and claim distribution

 

Claim size

distribution

Claim probability

Uniform [0, 48]

Uniform [48, 96]

0.02

80

60

0.04

40

20

For example, there are 140 policies where the chance of a claim being made is 0.02, and for 80 of these 140 policies if a claim is made, it is equally likely to be anywhere in the interval [0, 48] .

(i)    Let s denote total claims from this portfolio during the year .  Using a compound Poisson distribution to model s, determine the security loading factor 9 which is necessary to be 95% sure premiums exceed claims .                                                                                          [14]

(ii)    If the number of policies were to triple in each of the four categories, what would be the necessary security loading factor?                   [8]

3 (of 4) .

(a)   A random sample of 120 claims was observed from a portfolio, where

12d                                   12d

zi  = 9, 000;         zi(2)  = 420, 000,

i91                                   i91

zi  denotes the i-th claim observation .

(i)    Model the data  by  an  exponential distribution with density  9e-a_ (z  > 0) and estimate the positive parameter 9  by the method of moments .                                                                                        [3]

(ii)    Model  the  data  by  a  distribution  with  density  f (z)  =  9ze-a_2 /2 (z > 0) and nd the maximum likelihood estimate of the positive

parameter 9 .                                                                                   [5]

(b)   A binomial generalised linear model with the logistic link function is used by a bank to predict the probability of success of new personal investment packages .  The linear predictor is defined by n = ai + 8i z, where ai  and 8i  are the parameters with the index i denoting types of products (i = 1 for  property”, i = 2 for  equity”, and i = 3 for  bond”), the covariate z (in million pounds) represents the marketing budget for the product. Maximum likelihood estimates for these parameters have been obtained:

i = 1 (property): 1  = .0.045 , 1  = 0.028;

(ii)   What is the predicted success probability of a new  equity” product

with a 2 million pounds budget?                                                    [4]

(iii)   At which budget level will a property” and an equity”-based prod- uct have an equal probability of success?                                      [2]

4 (of 4) .    (a)   The following table gives cumulative paid claims in a motor insurance

scheme over a four-year period, together with annual premium income and estimated loss ratios determined by an underwriter . Assume that the amounts have been adjusted for inflation and all claims are settled by the end of development year 3 .

Table:  Motor insurance scheme: paid claims, premiums and loss ratios

Origin                 Development year

year i

0

1

2

3

Premium

Loss ratio

1998

31,766

48,708

64,551

69,003

76,725

92%

1999

30,043

45,720

59,883

 

77,005

92%

2000

35,819

54,790

 

 

78,520

90%

2001

40,108

 

 

 

83,400

90%

Use the  Bornhuetter-Ferguson  method and the weighted development factors to estimate the amount of reserves which should be set aside for the future claims at the end of 2001 .                                                 [12]

(b)   A claim-size random variable is modelled by a Pareto distribution, i .e ., x ~ Pareto(a, 入) with a = 4 and  = 900 .  A reinsurance arrangement has been made for future years whereby the excess of any claim over 600 is paid by the reinsurer .

(i)    Determine the mean reduction in claim size for the baseline insurance company which is achieved by this arrangement .                           [8]

(ii)    If claim inflation of 10% is expected for the next year and the excess level remains at 600, what will be the expected cost of a claim to the baseline insurance company?                                                          [6]