UNIVERSITY EXAMINATIONS, 2019/2020

FIRST YEAR EXAMINATIONS FOR THE DEGREE OF MASTER OF SCIENCE IN

ACTUARIAL SCIENCE

SAC 603 – ADVANCED RISK MATHEMATICS

DATE: TIME: .

Answer Question 1 and Two other Questions.

Question 1: (30 marks)

(a) (i) Explain briefly factors which lead to incurred but not yet reported (IBNR) claims in short-
term insurance contracts.
(ii) Explain in detail how the basic chain-ladder technique is used to estimate the
outstanding IBNR claims provisions in general insurance.
(5 marks)

(b) A decision maker’s utility function is given by u ( x) = - exp(-5 x) . The decision maker has two
random economic prospects available. The outcome of the first, denoted by X , has a normal
distribution with mean 5 and variance 2. The second prospect, denoted by Y , is normally
distributed with mean 6 and variance 2.5. Which prospect will be preferred? (6 marks)

(c) (i) Write down the general form of a statistical model for a claims run-off triangle defining
all terms used. (4 marks)

(ii) The table below shows the cumulative incurred claims on a portfolio of insurance
policies.

Accident Year Delay Year

2014 2,748 3,819 3,991
2015 2,581 4,014
2016 3,217
 The company decides to apply the Bornhuetter-Ferguson method to calculate the
reserves, with the assumption that the Ultimate Loss Ratio is 85%.

Calculate the reserve for 2016, if the earned premium is 5,012 and the paid claims are
1,472. (8 marks)

(d) A No Claims Discount (NCD) system has 3 levels of discount

Level 0 no discount
Level 1 discount = p
Level 2 discount = 2 p

where 0 < p < 0.6 .

The probability of a policyholder not making a claim each year is 0.9.
In the event of a claim, the policyholder moves to, or remains at level 0. Otherwise, the
policyholder moves to the next higher level (or remains at level 2). The premium paid in level 0
is £1,000.

Derive an expression in terms of p for the average premium paid by a policyholder once the
steady state has been reached. (7 marks)

Question 2: (20 marks)

(a) An exponential utility of wealth is given by

u(W ) = -e- AW

where A is a positive constant. His initial wealth is W0 > 1 and he is faced with a random
loss uniformly distributed on [0,1] .

Calculate the expression for the maximum insurance premium, G , he is prepared to pay to
buy complete insurance. Explain how G depends on the initial level of wealth and show that
it is larger than the expected loss. (8 marks)
(b) The following details are given concerning the claims on a general insurance company over
the period 2003 – 2008.

Year Number of
Claims (incl. Payments (KES’000) in development year
IBNR)
0 1 2 3 4 5
2003 6203 192 251 153 145 98 0
2004 6372 205 280 195 150 102
2005 6505 230 345 230 212
2006 6512 288 410 275
2007 7523 398 563
2008 8250 530

Assuming you have no further information, use the chain-ladder technique to estimate the
outstanding claim provision required at the end of 2008, making no allowance for investment
earnings. (12 marks)

Question 3: (20 marks)

(a) A decision maker’s utility function is given by u ( x) = x . The decision maker has wealth of
a = 10 and faces a random loss Y with a uniform distribution of (0,10) . What is the
maximum amount that this decision maker will pay for complete insurance against the
random loss? (5 marks)

(b) The table below shows cumulative claims paid on a portfolio of insurance policies.

Accident Development Year

Year 0 1 2 3
2012 240 281.4 302 305
2013 260 320 322
2014 270 312.9
2015 276

All claims are fully run off by the end of development year 3.

(i) Calculate the total reserve for outstanding claims using the basic chain ladder.
(8 marks)
An actuary is considering modeling the future claims assuming that individual development
factors are lognormally distributed with the following parameters:

Development Year
Accident Year 0 to 1 1 to 2 2 to 3
µ 0.171251 0.035850 0.008787
σ 0.032148 0.045606 0.046853
 (ii) Show that under these assumptions the cumulative development factor to ultimate is
also lognormally distributed. (3 marks)
(iii) Calculate a 99% upper confidence limit for the outstanding claims relating to the 2015
accident year. (4 marks)

Question 4: (20 marks)

An Actuarial student is using Empirical Bayes Credibility Theory Model 2 to calculate credibility
premiums for a group of insurers. She has analyzed the data for six different insurers, using 10
years of past data for each insurer. She has obtained the following figures:

6 10

∑∑ P ij
= 1,498; P * = 18.24
i=1 j=1

E[m(θ )] = 4.00; var[m(θ )] = 42.1; E !" s 2 (θ )#$ = 62.8

She has just received the following information relating to a 7th insurer (Insurer I), and she wishes
to update her estimates using the past ten years of claims data for Insurer I given in the table
below:

Year i 1 2 3 4 5
Aggregate
claims Yi 100 85 90 102 109
Volume Pi 22 24 26 20 25

Year i 6 7 8 9 10
Aggregate
claims Yi 106 128 132 150 131
Volume Pi 30 29 35 40 36

(i) Calculate her updated estimates for E[m(q )], E éë s 2 (q ) ùû and var[m(q )] , and
hence find the credibility premium for Insurer I for the coming year, given that Insurer I
is expected to have a volume figure for the coming year of 38. (17 marks)

(ii) The student also needs a credibility estimate for Insurer K, one of the six insurers
included in the original analysis. She knows that, for Insurer K, å YKj = 986 and
åP Kj = 327 . Explain whether the credibility premium for Insurer K (based on the full
analysis of the seven insurers) will be greater or less than the corresponding figure for
Insurer I (per unit of risk volume), and give reasons for your answer. (3 marks)
Question 5: (20 marks)
On 1 January 2015 an insurance company sold a large number of special house contents
insurance policies incorporating a no claims discount. The premium payable by each policyholder
on that date is denoted P . The premium payable at the start of each subsequent year may be P ,
1.25P or 1.5P , depending on the policyholder’s claims record since 1 January 2015. The
premium payable at the start of a particular year after 2003 will be
P if the policyholder did not make a claim in the previous year,
1.25P if the policyholder made one claim in the previous year and this is the only claim made by
this policyholder in the previous three years, or since 1 January 2015, whichever is shorter,
1.5P if the policyholder made at least one claim in the previous year, and the policyholder has
made at least two claims in the previous three years, or since 1 January 2015, whichever
is shorter.

The initial premium P is calculated by equating the expected present value of the premium
income and the claims and expenses outgo from this group of policyholders from the start of 2015
to the end of 2017. There were no further policies sold after 1 January 2015.

To calculate P the insurance company makes the following assumptions:

(1) the number of claims per policy per year has a Poisson distribution with parameter 0.5;
(1) 20% of those expected to pay a 50% loading do not renew their policies, otherwise there
are no withdrawals;
(2) the average claim amount in 2014 was £100;
(3) claim amounts increase at 10% per annum compound and are payable halfway through
the year;
(4) expenses are incurred at the start of each year and are 10% of the premium income in
the first year, 11% of the premium income in the second year and 12.1% of the premium
income in the third year;
(5) interest is allowed for at the rate of 9% per annum.

(i) By considering each of the three years, derive an expression for the total expected premium
income. (13 marks)
(ii) Similarly, derive the total expected claims and expenses outgo during the period and hence
calculate the initial annual premium P . (7 marks)