Professional Documents
Culture Documents
Siddhartha Chattopadhyay
Department of Humanities and Social Sciences
IIT Kharagpur
Autumn 2018
Suppose, fyt gTt=1 and fzt gTt=1 are two stationary time series
process that depends on each other
consumption and income
investment and interest rate
Structural VAR (1)
BXt = Γ0 + Γ1 Xt 1 + εt (2)
where,
1 b12 b
B = , Γ0 = 10
b21 1 b20
γ11 γ12 y ε
Γ1 = , Xt = t , εt = yt
γ21 γ22 zt εzt
and,
0 σ2y 0
V ( εt ) = E εt εt =
0 σ2z
Xt = A0 + A1 Xt 1 + et (3)
where,
Note,
0 0 σ21 σ12
E ( et ) = , V ( et ) = E et et =Σ=
0 σ21 σ22
σ2y + b12
2 σ2
z
σ21 = (1 b 12 b 21 )2
σ2z + b212 σ2
y
σ22 = (1 b 12 b 21 )2
b21 σ2y + b12 σ2z
σ12 = σ21 = (1 b 12 b 21 )2
et serially uncorrelated
) cov (e1t , e1t j ) = cov (e2t , e2t j) = 0, 8j 6= 0
n (n 1)
2
b
e2t = εzt
b
e1t = b12 εzt + εyt
σ2y + b12
2 σ2
z
b21 =
σ (1 b 12 b 21 )2
σ2z + b212 σ2
y
b22 =
σ (1 b 12 b 21 )2
b21 σ2y + b12 σ2z
b12 = σ
σ b21 = (1 b 12 b 21 )2
Diagonalization of A1 gives,
Aj1 = V Λj V 1
0.7 0.2
Xt = X + et
0.2 0.7 t 1
1 0.8
B =
0 1
Λ = diag (0.5, 0.9) )both variables are stationery and system stable
Yields:
(n 1 ) (n 1 )
yt +n Et (yt +n ) = ∑ φ11 (j ) εyt +n j + ∑ φ12 (j ) εzt +n j
j =0 j =0
(n 1 ) (n 1 )
zt +n Et (zt +n ) = ∑ φ21 (j ) εyt +n j + ∑ φ22 (j ) εzt +n j(6)
j =0 j =0
0 (n 1 )
12
B
B ∑ φ11 (j ) εyt +n j C
C
j =0
E (yt +n Et (yt +n )) 2
= EB
B (n 1 )
C
C
@ A
+ ∑ φ12 (j ) εzt +n j
j =0
0 (n 1 )
12
B
B ∑ φ21 (j ) εyt +n j C
C
j =0
E (zt +n Et (zt +n )) 2
= EB
B (n 1 )
C
C (7)
@ A
+ ∑ φ22 (j ) εzt +n j
j =0
(n 1 ) (n 1 )
σ2y (n) = σ2y ∑ φ11 (j )2 + σ2z ∑ φ12 (j )2
j =0 j =0
(n 1 ) (n 1 )
σ2z (n) = σ2y ∑ φ21 (j )2 + σ2z ∑ φ22 (j )2 (8)
j =0 j =0
σ2y (n 1 ) σ2z
(n 1 )
σ2y (n) j∑ 2 (n ) ∑
2
1 = φ 11 ( j ) + φ12 (j )2
=0 σ y j =0
σ2y (n 1 ) σ2z
(n 1 )
σ2z (n) j∑ 2 (n ) ∑
2
1 = φ 21 ( j ) + φ22 (j )2 (9)
=0 σ z j =0