Академический Документы
Профессиональный Документы
Культура Документы
Exercise 1
Consider the n-dimension vector time series {Yt }.
Assuming stationarity, show that k = 0k where k = E[(Yt )(Ytk )0 ].
Solution
k = E[(Yt )(Ytk )0 ]
k k k
11 12 1n
k k k
21 22 2n
= .. .. .. ..
.
. . .
k k k
n1 n2 nn
k
ii = cov(yit , yitk ) = cov(yit , yit+k )
k
= ii
k
ij = cov(yit , yjtk ) 6= cov(yit , yjt+k )
k
6= ij
k = E[(Yt )(Yt+k )0 ]
= E[(Ytk )(Ytk+k )0 ]
= E[(Ytk )(Yt )0 ]
0
= [E[(Yt )(Ytk )0 ]]
= 0k
1
Exercise 2
Consider the following VAR(2) model of a n-vector of time series Yt :
ii) Compute the forecast of Yt+3 based on this representation and the corresponding MSE.
iii) Derive the companion form of model (1) and the forecast of Yt+3 in this framework.
Solution
i) Consider the polynomial (L) = In 1 L 2 L2 such that we can rewrite our model as (L)Yt = t .
Then the Wold polynomial (L) = (L)1 can be derived as (L)(L) = In in the following way:
(L)1 (L) = In
(In 1 L 2 L2 ) (0 + 1 L + 2 L2 + 3 L3 + ) = In + 0L + 0L2 + 0L3 +
0 1 0 L 2 0 l2 + 1 L 1 1 L2 2 1 L3 + 2 L2 1 2 L3 2 2 L4 + = In +
At this point we simply need to equate LHS coefficients with RHS coefficients of the same degree:
0 = In
1 0 + 1 = 0 1 = 1
2 0 1 1 + 2 = 0 2 = 1 1 + 2
2 1 1 2 + 3 = 0 3 = 1 2 + 2 1
2 2 1 3 + 4 = 0 4 = 1 3 + 2 2
ii)
2
iii) Rewrite the model as follow:
! ! ! !
Yt 1 2 Yt1 t
= +
Yt1 In 0n Yt2 0n
and define:
!
Yt
t =
Yt1
!
1 2
F =
In 0n
t+3/t = F 3 t
Therefore:
! ! ! ! !
Yt+3 1 2 1 2 1 2 Yt
=
Yt+2 In 0n In 0n In 0n Yt1
! !
1 1 1 + 2 1 + 1 2 1 1 2 + 2 2 Yt
=
1 1 + 2 1 2 Yt1
! !
3 2 2 Yt
=
2 1 2 Yt1
and, if we continue in the substitution of the vector (Yt , Yt1 )0 , we are able to derive exactly the same
formula for Yt+3/t as in the Wold approach
Exercise 3
Consider the following VAR(p) model of a n-vector time series Yt :
A(L)Yt = ut ,
i) Derive the structural VAR (SVAR) representation of series Yt through the Cholesky decomposition of
the variance-covariance matrix .
ii) Explain how many restrictions on the SVAR parameters are required for the identification.
iii) Suppose that you have a VAR(p) for 3 variables: the inflation rate pt , the growth rate yt , and the
real exchange rate rt at which the domestic currency is exchanged for the one of the largest commercial
partner of the considered country. According to an economic reasoning, the three variables at hand
3
are respectively associated with a nominal shock, a real shock and a foreign shock. Explain how the 3
variables should be ordered in the structural VAR assuming that:
Solution
i) Using the Cholesky decomposition = P P 0 we can define the structural shock as t = P 1 ut (note
that both P and P 1 are lower triangular) so that:
0
E[t 0t ] = P 1 E[ut u0t ]P 1
0
= P 1 P P 0 P 1
= In (2)
ii) The number of parameters in the Reduced Form model is given by:
n(n + 1)
2
2
A(L) pn
Dn
B n(n 1)
(L) pn2
n(n1)
Therefore, the difference is 2 which represents the number of restrictions needed in order to
identify the SVAR.
iii) In order to derive the exact order let us first consider the Wold representation od the model
Yt = (L)B 1 t = (L)P t
= (L)t
X
= 0 t + k tk
k
4
with 0 = B 1 = P 6= In and therefore lower triangular. Hence
y1,t p11 0 0 1,t
y2,t = p21 p22 0 2,t +
y3,t pt
1,t realt
2,t = f oreignt
3,t nominalt