ACTL 2003 Stochastic Models For Actuarial Applications Final Examination
ACTL 2003 Stochastic Models For Actuarial Applications Final Examination
ACTL 2003
STOCHASTIC MODELS
FOR ACTUARIAL APPLICATIONS
FINAL EXAMINATION
1. TIME ALLOWED - 3 HOURS.
= Y2,"" Y.n
Yn ] = pe- A(Yl+Y2+'+Yn) ,
where 0 < p < 1, .\ 2: O. The cumulative number of accidents suffered by the driver over the
time from period 1 up to period n is
n
(i) (3 marks) Verify that the Markov property holds for X 1 ,X2 , ... ,Xn , ... and explain
why the sequence Y 1 , Y 2 , ... , Yn , .. . does not form a Markov chain.
(ii) (1 mark) Write down the transition matrix of the Markov chain X.
(iii) Determine, being careful to explain your reasons in each case:
(a) (1 mark) whether the Markov chain X is time-homogeneous
(b) (1 mark) whether it is irreducible
(c) (1 mark) whether it admits a stationary probability distribution
(iv) (1 mark) Starting from the state X t = j, calculate the probability of suffering no
further accident for the next n successive periods.
(v) Suppose you are provided with full claims records for a number of a company's policy
holders.
(a) (1 mark) Describe a method for estimating the parameters .\ and p
(b) (3 marks) Explain how to test the assumption that the probability of an accident
depends only on the cumulative number of accidents, X n , and does not have a
direct dependence on n.
Question 2 (5 marks)
Cars pass a point on the highway at a Poisson rate of one per minute. If 5 percent of the
cars on the road are vans, then
(i) (1 mark) What is the probability that at least one van passes by during an hour?
(ii) (3 marks) Given that ten vans have passed by in an hour, what is the expected
number of cars to have passed by in that time?
(iii) (1 mark) If 50 cars have passed by in an hour, what is the probability that five of
them were vans?
Level 0 no discount
Level 1 10% discount
Level 2 25 % discount
Level 3 40 % discount
in which the discount level is increased by 1 year after a year with no claims (unless
it is already at 3), or reduced by 1 if one or more claims were made during the year
(unless it is already at 0). In each year there is a 20% chance that the driver suffers
an accident entailing costs greater than the $100 excess on the policy.
(a) (1 mark) Denote by D n the driver's discount level in year n.
{D n : n 2: O} may be modelled as a Markov chain.
Explain why
(b) (1 mark) Assuming that the driver always claims for each incident, calculate the
long-run proportion of years for which D takes each of the value 0,1,2 and 3.
(c) (3 marks) The size of a claim (cost less excess) may be assumed exponentially
distributed with mean $250. The driver believes that there is some level C such
that it is more economical not to claim when the claim size would be less that
C. Describe how the driver might use simulation to investigate the most suitable
value of C.
(ii) A motor insurance company assumes that a holder of a provisional driver's licence will
make claims according to a Poisson process with rate X per year, where X is not fixed
but is determined randomly for each driver according to the density function
f(x) = 2e- 2x
(x> 0)
(a) (2 marks) Describe how to simulate an observation X from the the density f
using a single pseudo-random variable U assumed uniformly distributed on [0,1].
(b) (3 marks) Explain how, given the value X generated in before, you would use a
sequence U1 , U2 , . .. of uniform pseudo-random variables to simulate the number
of claims made in two six-month periods by a provisional driver with mean claim
rate X per year.
4
Question 4 (9 marks)
Marital status is considered using the following time-homogeneous, continuous time Markov
Jump process:
the transition rate from unmarried to married is 0.1 per annum
the divorce rate is equivalent to a transition rate of 0.05 per annum
the mortality rate for any individual is equivalent to a transition rate of 0.025 per
annum, independent of marital status
The state space of the process consists of five states: Never Married (NM), Married (M),
Widowed (W), Divorced (DIV) and Dead (D).
Px is the probability that a person currently in state x, and who has never previously been
widowed, will die without ever being widowed.
(i) (1 mark) Construct the generator matrix for this process.
(ii) (2 marks) Show, by general reasoning or otherwise, that PNM equals PDIV .
(iii) (3 marks) Demonstrate that:
P NM
= - +1
PM = -
X PM
+-
X P DIV
(iv) (1 mark) Calculate the probability of never being widowed if currently in state N M.
(v) (2 marks) Suggest two ways in which the model could be made more realistic.
Question 5 (8 marks)
A continuous-time Markov sickness and death model has four states: H (healthy), S (sick),
T (terminally ill) and D (dead). From a healthy state transitions are possible to states S
and D, each at rate 0.05 per year. A sick person recovers his health at rate 1.0 per year;
other possible transitions are to D and T, each with rate 0.1 per year. Only one transition
is possible from the terminally ill state, and that is to state D with transition rate 0.4 per
year.
(i) (2 marks) Define P(t) = {Pij(t) : i,j E H, S, T, D} where Pij(t) denotes the probability of being in state j at time t given that the individual was in state i at time
O. Construct the generator matrix for this process and state the Kolmogorov forward
equation satisfied by the matrix P(t).
(ii) (1 mark) Calculate the probability of being healthy for at least 10 uninterrupted years
given that you are healthy now.
(iii) (3 marks) Let dj denote the probability that a life which is currently in state j will
never suffer a terminal illness. By considering the first transition from state H, show
that dH = ~ + ~ds and deduce similarly that ds = 112 + ~dH' Hence evaluate dH and
ds .
(iv) (2 marks) Write down the expected duration of a terminal illness, starting from
the moment of the first transition into state T. Use the result of (iii) to deduce the
expectation of the future time spent terminally ill by an individual who is currently
healthy.
Question 6 (9 marks)
(i) (1 mark) Define standard Brownian motion Et, t 2::
o.
(iii) (3 marks) Calculate, given the parameters J.1 = 25% p.a., (J = 20% on an annual
basis, the probability that the share price will exceed 45 in four months' time given
that its current price is 38.
(iv) (3 marks) Calculate the probability that the share price will exceed 45 at any stage
during the next four months given that its current value is 38.
y) + e (-yv't-t At) .
At G ( ---ri
y t
).,YG
where y 2:: 0 and G denotes the normalised Gaussian probability distribution function.]
W(O) = O.
(i) (1 mark) Write down an expression for W(t) in terms of a standard Brownian motion
B(t).
(ii) (4 marks) Show that {S(t) : t > O} is a continuous-time martingale.
(iii) Let Ta
inf {t : S (t)
---t
1 as t
---t
00.
= a.
(c) (1 mark) Explain why the fact that E[S(Ta )] of- S(O) does not contradict the
optimal stopping theorem.
(iv) Now suppose instead that a> 1 and define Ta as before
(a) (3 marks) Explain why E[S(Ta )]
< 00].
where {et : t
variance a 2 .
Ik
D:nlk-l!
Ik
= Cov(Xt , X t - k ).
= 0.6Xt _ l + 0.3Xt - 2 + et
10
together with the condition X o = xo, where {et: t = 1,2, ... } is a sequence of independent
variables, each normally distributed with mean 0, variance (}2.
(i) (3 marks) Find an expression for X n in terms of Xo and {et: 1 :S t :S n}. Hence find
the limit as n - 7 00 of the distribution of X n , identifying the condition required for
the distribution to converge.
(ii) (2 marks) Explain why the process X is not weakly stationary.
(iii) (3 marks) Suppose that X o is no longer fixed, but is a random variable, with distribution equal to the limiting distribution found in (i). Show that both the distribution of
X t and the covariance of X t+ s with X t are independent of t, and evaluate the correlation
coefficient Ps = Corr(Xt , Xt+s)'
(iv) (2 marks) Obtain forecasts of X n + 1 and X n +2 based on observations
assuming that 0' and e are known.
x n },
(v) (4 marks) I have a sequence of observations {SI, 82, ... , 8 n } of the closing price of a
share on n successive days. I intend to use time series analysis, involving the process X
considered above, to predict future prices of the share. For each of the models (a)-(d)
indicate whether the process S is 1(0), 1(1) or neither. Discuss which of the models is
likely to be the most suitable for my intended purpose.
(a) St = X t
(b) St
Xt
(c) log St
(d) log St
=
=
+ St-1
Xt
Xt
+ log St-1
(vi) (2 marks) Give two points that I should bear in mind when deciding whether to base
my investment decisions on my model.
END OF EXAMINATION
11