Scribd
Upload
47 views10 pages

The University of New South Wales Month of Examination - November 2005

The document describes a final exam for a stochastic models course. It contains 9 questions related to Markov chains, stochastic processes, and time series analysis. Students are given 3 hours to complete the exam which covers a range of probabilistic and statistical topics.

Uploaded by

Bob
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
47 views10 pages

The University of New South Wales Month of Examination - November 2005

The document describes a final exam for a stochastic models course. It contains 9 questions related to Markov chains, stochastic processes, and time series analysis. Students are given 3 hours to complete the exam which covers a range of probabilistic and statistical topics.

Uploaded by

Bob
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 10

p.,>:.

THE UNIVERSITY OF NEW SOUTH WALES


MONTH OF EXAMINATION - NOVEMBER 2005
FINAL EXAMINATION
ACTL2003

Stochastic Models for Actuaries


1. TIME ALLOWED - 3 HOURS

2. TOTAL NUMBER OF QUESTIONS - 9 QUESTIONS


3. ANSWER ALL QUESTIONS
4. ALL QUESTIONS ARE NOT OF EQUAL VALUE
5. THIS PAPER MAY BE RETAINED BY THE CANDIDATE
6. CANDIDATES MAY BRING THEIR OWN CALCULATORS OR HAND HELD COMPUTERS
7. CANDIDATES MAY BRING THE TEXT "FORMULAE AND TABLES FOR ACTUARIAL EXAMINATIONS" INTO THE EXAMINATION. IT MUST BE WHOLLY UNANNOTATED.
8. THE FOLLOWING MATERIAL WILL BE PROVIDED BY THE ENROLMENT AND EXAMINATIONS SECTION: STATISTICAL TABLES.
ANSWER EACH QUESTION STARTING ON A NEW PAGE.
ANSWER PART A QUESTIONS AND PART B QUESTIONS ON SEPARATE BOOKLETS.
ALL ANSWERS MUST BE WRITTEN IN INK. EXCEPT WHERE THEY ARE EXPRESSLY REQUIRED, PENCILS MAY BE USED ONLY FOR DRAWING, SKETCHING OR GRAPHICAL WORK.

PART A
Question 1. [9 marks]
Consider a homogeneous Markov chain with state S

,-1 I1
o

p=

{1,2,3} and transition matrix

(i) [2 marks] Calculate the 3-step transition matrix.


(ii) [4 marks] Calculate for each of the following initial conditions, the probability that the chain
will be in state 3 when it is observed at time n = 3 given that:
(a) [1 mark] the chain is in state 1 at time zero
(b) [1 mark] the chain is in state 1 at time zero and in state 2 at time 1

*'

(c) [2 marks] the probabilities of being in states 1, 2 and 3 at time zero are given by
and
respectively.

H,

(iii) [3 marks] How would your anSwers to (a), (b) and (c) change if the time of the observation
= 300 instead of n = 37

were n

Question 2. [10 marks]


A no claims discount system for motor policies has 3 discount levels: 0%, 25% and 40%. The
rules for moving between these levels are as follows: If the policyholder makes no claims in a
year, he moves to the next higher discount level, or remains at 40%. If one claim is made,
the policyholder stays at the same discount level. In the event of more than one claim, the
policyholder moves to 0% discount. The full premium is $100.
A portfolio consists of "good" drivers and "bad" drviers, who make an average of 0.5 and 2.0
claims each year (respectively). It is assumed that the number of claims for each driver has a
Poisson distribution.

For good drivers and bad drivers separately, assuming that the steady state has been reached
estimate:

(i) [6 marks] the proportion of policyholders expected at each discount level.

(ii) [2 marks] the average premium paid.


(iii) [2 marks) comment on the differences between the good and bad drivers.

Question 3.

[12 marks]

The night before her exam on Stochastic Models for Actuarial Applications, a student has the
following nightmare. She dreams that she is sitting in a lecture on Stochastic Models for Actuarial
Applications that lasts for eternity. The student is listening in a state of confusion to what is
being said 1 except for certain moments of inspiration which arrive as a Poisson process with
rate.\. Immediately as the student receives inspiration, she begins writing down what she has
understood. The amount of time she will spend writing is typically exponentially distributed
with parameter f.l. Throughout the duration of the never ending lecture, the student has left her
mobile phone left switched on (unfortunately the phone has a battery with infinite life). Calls
arrive at the phone as a Poisson process with rate 1]. The student always answers the phone
irrespective of what she is doing (writing or listening). In any small interval of time of length
dt, the probability that the conversation on the phone will end is vdt + 0 (dt). By the end of
each phone call, the student cannot remember what she was doing when the phone rang and thus
resumes listening in a state of confusion.
Let X(t) be the three state Markov chain that describes whether the student is listening, writing
or talking at time t into the dream. Find the stationary distribution of X.

Question 4. [9 marks]
Let Y, = t . B+. t > 0 and Yo = O.
(i) [3 marks] Specify the distribution of Y;, including its mean and variance?

(ii) [4 marks] Compute Cov {y" Y,j.


(iii) [2 marks] Argue that {Y" t 2: O} is a standard Brownian motion process.

Question 5. [15 marks]


Let XI, be a stochastic process satisfying the stochastic differential equation
dX,

= I"(X"

I)dl

+ <7(X" t)dB,

where BI. is a standard Brownian motion and X o > O.

(i) Assuming that I"(X" I)

= I" and <7(X,. I) = <7.

(a) [1 mark] Find the strong solution of the process X,.'


(b)

13

marks] Find E(X,.) , Var(X,) and Cov(X" ,X,,) where 1 12 are fixed times such that
"

0<1 , <1,.

(ii) Assume that I"(X" I) = I"X,. and <7(X" I) = <7X,.


(a) [2 marks] Find the strong solution of the process InX,..
(b) [3 marks] Find E(lnX,.) and Var(lnX,.). Also derive the density function of X,..
(c) [4 marks] Assuming that X t describes the movement of a stock price, find the expression

for
E [K I (XI' > K)]
where K is an constant, T is maturity date and

~:

I(XT > K) = (
is indicator function. Assume that Xo is known.

XT > K )

XT~K

PARTE
Question 6 [15 marks]

(i) !5 marks] Generate an observation from the density function


I

j(x)= (l+x)"

x>O

by applying the inverse transform method and using a pseudo-random number u in (0,1).

(ii) [3 marks] Explain how the procedure used in (1) can be extended to generate an observation
from the density
I

g(x)

-oo<x<oo

= 2(1 + Ixl)"

(iii) [4 marks] Consider the distribution with density having the form
h (x) - _;-;;-,e--;:;;::

- 1r(x' + e')'

-oo<x<oo

where (J is a positive parameter. Show that

-oo<x<oo

h(x) ::; cg(x),

whenever c::::: *(8 + 0- 1 ).


(iv) [3 marks] From (iii), devise a method based on Acceptance Rejection sampling for generating
observations from the density h(x).

Question 7 [10 marks]


A time series plot for the sales of an industrial heater from January 1965 to November 1971 is
shown in Figure 1 below.

"0
is
.c
E

, ~

'.

,. ;:

"~

Jan 65

Jon66

Jan/fT

Jln69

Jon66

Jan70

Jsn71

Year

Figure 1: Chatfield, The Analysis of Time Series, 6th edition.

(i) [2 marks] Comment on the seasonal variation and trend of the sales data.
(ii) [4 marks] Discuss completely an appropriate procedure that will eliminate both seasonal
variation and trend in the data.

(iii) [4 marks] Examining Figure 1~ you hypothesise that eliminating the trend and seasonality
of the data will lead to residuals that are independent and identically distributed. Discuss at
least two simple tests for checking that your hypothesis is correct. For each test, include in your
discussion procedures necessary to carry out such test.

Question 8 [10 marks]


Consider the MA(I) process X t

= Z, + eZ'_l

where {Z" t > aj is white noise with variance ,,'-

(i) [5 marks] Show that autocovariance and autocorrelation functions of this process are defined
by

,,'(I

1x(t+h,t) =

,,'e.

+ e'),

0,

and
1,

Px(h)

= { e/(1 + e'),
a.

h=a
h = 1
Ihl > 1

h=a
h= 1
Ihl >

respectively.

(ii) [2 marks] What is an invertible MA(q) process and for what values of e is {X,j invertible?
(iii) [3 marks] A sequence of observations of {X,j has sample variance "i" = 7.25 and sample
lag-l autocovariance "i, = 2.5. Find an invertible MA(I) process that can be fitted to these data.

Question 9 [10 marks]


A stationary second-order autoregressive process {X t } is defined by

where {Z,j - WN(0,()"2).

(i) [4 marks] Obtain an equation for the autocovariance function i(t), t = 0,1.2 in terms of
'i'j,i= 1.2 and ~(8).S"ft.

(ii) [2 marks] Find the autocorrelation function p(t), t = 0,1,2.


(iii) [4 marks] Suppose that the equations derived in (ii) are used as the basis of an estimation
procedure: estimates for q;l and 1>2 are defined to be the solutions of those equations when p(O)
is replaced by the sample autocorrelations T. Solve these equations for the estimates ;;1 and ~1'

The End

10

You might also like