Scribd
Upload
481 views2 pages

Quantitative Finance Problems and Solutions

Quantitative Finance problems and solutions regarding analysis of time series, discrete and multivariate modelling.

Uploaded by

Josiah Khor
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
481 views2 pages

Quantitative Finance Problems and Solutions

Quantitative Finance problems and solutions regarding analysis of time series, discrete and multivariate modelling.

Uploaded by

Josiah Khor
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/ 2

FN3142 – Quantitative Finance 1/2

QF ASSIGNMENT 1 – SOLUTIONS
QUESTION 1

(a) Define what is meant by Covariance Stationarity.

(b) Determine whether the following time series are covariance stationary. For each case,
show your working clearly and completely.
(i) Yt  a  b  t  c  t  2 ,  t ~ iid N (0,  2 )
(ii) Yt  a  b  0 ,  t ~ iid N (0,  2 )
(iii) Yt   t   t 1,  t ~ iid N (0,  2 )

Solution

(a) A time series {Yt } is covariance stationary if the following conditions are met:
(i) Its mean is constant for all time periods: E[Yt ]   t
(ii) Its auto-covariance depends on the time-difference only and does not
dependent on time: cov[Yt , Yt  j ]   j t , j  0, 1, 2, 

(b)(i) E[Yt ]  a  b E[ t ]  c E[ t  2 ]  a


cov[Yt , Yt  j ]  cov[a  b  t  c  t  2 , a  b  t  j  c  t  j  2 ]
 cov[b  t  c  t  2 , b  t  j  c  t  j  2 ]
 b 2 cov[ t ,  t  j ]  bc cov[ t ,  t  j  2 ]  bc cov[ t  2 ,  t  j ]  c 2 cov[ t  2 ,  t  j  2 ]
(b 2  c 2 ) 2 if j  0

 bc 2 if | j | 2
0 otherwise

Since both E[Yt ] and cov[Yt , Yt  j ] do not depend on t, {Yt } is covariance stationary.

(b)(ii) E[Yt ]  a  b E[ 0 ]  a


cov[Yt ,Yt  j ]  cov[a  b  0 , a  b  0 ]  b2 cov[ 0 ,  0 ]  b2 2
Since both E[Yt ] and cov[Yt , Yt  j ] do not depend on t, {Yt } is covariance stationary.

(b)(iii) E[Yt ]  E[ t   t 1 ]  E[ t ]  E[ t 1 ]  0


 4 if j  0
cov[Yt , Yt  j ]  cov[  t  t 1 ,  t  j   t  j 1 ]  E[ ( t  t 1 )  ( t  j   t  j 1 )]  
0 otherwise
Since both E[Yt ] and cov[Yt , Yt  j ] do not depend on t, {Yt } is covariance stationary.

QUESTION 2

Let {Yt } be the ARMA(1, 1) process,


Yt  Yt 1   t 1   t , |  | 1, |  | 1,  t ~ WN(0,  2 )
FN3142 – Quantitative Finance 2/2

(a) Show that the first-order autocorrelation of {Yt } is given by


(1   )(   )
1 
1  2   2

(b) For j  2 :  j   j 11

Solution

(a) It is clear that E[Yt ]  0 .

 0  var[Yt ]  var[Yt 1   t 1   t ]  var[Yt 1 ]  var[ t 1 ]  var[ t ]


 2 cov[Yt 1 , t 1 ]  2 cov[Yt 1 ,  t ]  2 cov[ t 1 ,  t ]
  2 var[Yt 1 ]   2 var[ t 1 ]  var[ t ]
 2 cov[Yt 1 ,  t 1 ]  2 cov[Yt 1 ,  t ]  2 cov[ t 1 ,  t ]
  2  0   2 2   2  2 cov[Yt  2   t  2   t 1 ,  t 1 ]
  2  0   2 2   2  2 2
Hence
(1  2   2 ) 2
0 
12

 1  cov[Yt , Yt 1 ]  cov[Yt 1   t 1   t , Yt 1 ]
 cov[Yt 1 , Yt 1 ]  cov[ t 1 , Yt 1 ]  cov[ t , Yt 1 ]
  cov[Yt 1 , Yt 1 ]   cov[ t 1 , Yt 1 ]
   0   cov[ t 1 ,Yt  2   t  2   t 1 ]
   0   cov[ t 1 ,  t 1 ]    0   2
(1  2   2 ) 2  (1  2   2 )   (1   2 ) 2
   2
 
12 12
  2 2   2     2 2  (1   )   (1   ) 2
   
12 12
(1   )(   ) 2
 
12
Thus
 1 (1   )(   )
1  
0 1  2   2
(b) We have
 2  cov[Yt , Yt  2 ] cov[Yt 1   t 1   t , Yt  2 ]
  cov[Yt 1 , Yt  2 ]  cov[ t 1 , Yt  2 ]  cov[ t , Yt  2 ]    1
Thus
 2  1 
2     1   1
0 0 0

Similarly  j   j 1     j 11 j  2

You might also like