Носко В.П. - Эконометрика. Введение в регрессионный анализ рядов - 2002

..
2002
1.
2. . ARMA
2.1. .
2.2.
2.3.
2.4.
2.5. (
)
2.6. ARMA,
3. ARMA
3.1. ARMA
3.2.
3.3.
4.
4.1.
4.2.
4.3.
4.4.
5.
5.1. ARMA
5.2. TS
DS
5.3. TS DS ARMA. .
6. TS DS
6.1.
6.2.
6.3. -
6.4.
6.5. DGP SM
6.6. .
6.7.
6.8.
6.8.1.
6.8.2.
6.8.3. .
6.8.4. DF-GLS
6.8.5. (KPSS)
6.8.6. ( )
6.9. , TS DS
6.9.1.
6.9.2.
6.9.3. TS DS
6.9.4.
6.10.
6.10.1.
6.10.2.
7.
7.1.
7.2. .
7.3. .
7.4.
8.
8.1.
8.2.
,
[ (2000)],

yt = 1 xt1+ 2 xt2+ + p xtp + t , t = 1, 2, , n ,
, xt1, xt2, , xtp , t
= 1, 2, , n , , 1, 2, , n ()
,
(
).

.

( );
;
.

.

:
, ,
, , ,
,
t - ;
,
t - ;

, F- ;

,
.
,
n ,

xt1, xt2, , xtp , t = 1, 2, , n ;
( ) t , t
= 1, 2, , n , ,
y1, y2, , yn , .
,

,
. , ,

n

.

, ,
, , ,

.

. ,
( )
.
,
.
,
, ..
(, , , ..).

, ,
,
.

y 0 = 0,
yt = yt 1 + t , t = 1, 2, , n ,
1, 2, , n ,
2.
,
xt t
, .. xt = yt 1.
, (
),
,
n
y y
t
t =2
n
y
t =2
t 1
2
t 1
, ,

t :
x1 = 0 ,
x2 = y1 = 1 ,
x3 = y2 = y1 + 2 = 1 + 2 ,
x4 = y3 = y2 + 3 = ( 1 + 2 ) + 3 = 2 1 + 2 + 3 ,
xn = yn 1 = n 2 1 + n 3 2 + + n 1 .
,
, t- F- ,
,
t- F-, .
, = 1,
( ),
yt = yt 1 + t , t = 1, 2, , n , y 0 = 0,
( )

. ,
.
[White (1958)],
,
.
TS (trend stationary)
(
) DS (difference stationary) ,
.
DS ,

. ,

, ,
Dickey D.A., Fuller W.A, Granger C.W.J., Hansen B.E., Johansen S.,
Juselius K., Perron P., Phillips P.S.B., Sims C.A., Stock J.H., Watson M.W.J.
[Maddala G.S.,
Kim In-Moo (1998)]; . [Hatanaka M. (1996)].

,
,
. , [ (2000)],

. ,
- .
,
, ,
, ,
( , ),
.

.
,
.
. . .. www.iet.ru
1.

,

yt = 1xt1+ 2 xt2+ + p xtp + t , t = 1, 2, , n ,
, xt1, xt2, , xtp ,
t = 1, 2, , n , , 1, 2, , n ()
,
(
).

.

: ,
, , ,
,
,
t - ;
,
t - ;

F- ;

,
.
,
n ,

xt1, xt2, , xtp , t = 1, 2, , n ;
( )
t ,
t = 1, 2, , n , ,
y1, y2, , yn , .
,

,
.
.
, (
) .

,
.
www.iet.ru/mipt/2/text/curs_econometrics.htm
. . .. www.iet.ru
, -
p :
y = X + ,

y = (y1, ..., yn)T - n ,
X (np)- n , n > p,
= ( 1, 2, , p)T - ,
= ( 1, 2, , n)T - () n .
X p , XTX
, (XTX) 1 ,

= (XTX) 1XTy .

E( ) = E ((XTX) 1XT(X + )) = E ((XTX) 1XTX ) + E ((XTX) 1XT ) = + E (XTX)
1 T
X ).
X , E ((XTX) 1XT ) = (XTX) 1XT E ( ) = 0,
E( ) = ,
.. .
(,
) , E ((XTX)
1 T
X ) 0,
E( ) ,
, , ,
.
,
,
, ,
( ).

A
s () xt1, xt2, , xtp t
s,
1, 2, , n ,

2 > 0. ( t ~
i.i.d. N (0, 2). i.i.d. independent, identically distributed.)

E ((XTX) 1XT ) = E ((XTX) 1XT) E( ) = 0,
. . .. www.iet.ru

.
( )
.

X ;
( ).
,
X
X.

, :
| X ~ N , 2 (X T X )
).
N , 2 X T X
) X . ,
, .
, j- :
2
T
j | X ~ N j , (X X )j j ,
1
(X T X ) jj j- (X T X ) ,
1
( X X )j1j
T
X ~ N (0, 1) .
S2/2 , S2 = RSS/(n p), RSS

, - (n p) ,
S2/2 | X ~ 2(n p) .
, t- H0: j = *j
t=
j *j
S ( X T X )j1j
(
)
=
j
*
j
( X T X )j1j
S2 2
,
, H0
t- t- (n p)
,
t | X ~ t(n p) .
X .
, X , t
H0: j = *j
t(n p) .
. . .. www.iet.ru

F- .
A
.
A
| X ~ N(0, 2In) , In ( n n) .

xt = (xt1, xt2, , xtp)T
.
p t-
B
s () xt1, xt2, , xtp t
s;
t ,
t ~ i.i.d., E( t) = 0, D( t) = 2 > 0 E( t4) = 4 < ;
E(xt xtT) = Qt , (1/n)( Q1 + + Qn) Q
n , Q ;
E(xit xjt xkt xst) < i, j, k, s ;
(1/n)( x1 x1T + + xn xnT) Q .
,
, A.
n - () t
S2, t- F-,
, .
,
, ,
, [Hamilton (1994)].
(n) n , Xn
n , S n2 , tn , Fn S2 , t , F , n .
, B, n
n ( (n) ) N (0 , 2 Q 1),
n ( S n2 2 ) N (0, 4 4),
tn N (0 , 1),
qFn 2(q) , q .
, ..
, , n
, .
T
2
1
2
1
(n) N ( , Q n), (n) N ( , (Xn Xn) ) ,
( ),
. . .. www.iet.ru
S n2 N (2 , (4 4) n),
tn N (0, 1),
qFn 2(q).
B n
, t(n p) t-
( N (0 , 1)) F(q, n p) F- ( 2(q)
qFn), (
, ).
,
, n
tn Fn .
C
, , ,
t | X ~ i.i.d. ,
X
n- N (0 , 2 V) ;
V nn.
V , ;
V 1 .
(nn)- P , V 1 = PTP . P ,
* = P .
E (*) = 0 ( X)
*
Cov (*| X ) = E (**T | X ) = E (P (P )T | X )
= P E ( T | X ) PT = P 2 V PT .
V = (V 1) 1 = (PTP) 1 ,
Cov (*| X ) = P 2 V PT = 2 P(PTP) 1PT = 2 In .
P
y = X + ,
:
Py = PX + P ,
*
y = X* + * ,
*
y = Py , X* = PX , * = P .

*
| X ~ N (0 , 2 In) ,
,
A. , , A,
y* = X* + * .
,
* = (X*TX*) 1 X*T y* = (XTPTPX) 1 XTPTPy = (XT V 1X) 1 XT V 1y
. . .. www.iet.ru
, .. E(*) = , (
X)
Cov ( * | X ) = 2(X*TX*) 1 = 2(XT V 1X) 1.
(GLS
generalized least squares).
y* = X* + *
, t- F-.
V ,
, , V = V() ,
, .
,
GLS * = (XT V 1X) 1 XT V 1y
V = V(0) (V(0) )
n 0 . ,
V( n ),

.
, ,
.
,
,
.
. . .. www.iet.ru
2. . ARMA
2.1. .
(time series)
, .
(, , ..),
, x1, ..., xn
t = 1, , n .

, x1, ..., xn
, ,
X1, ..., Xn ,
F(v1, v2, , vn) = P{ X1 < v1, X2 < v2, ... , Xn < vn }.

,

X1, ..., Xn

p( x1, x2, , xn).

,
,
.
.
xt , t = 1, , n , (
), m ( m < n)
X t1 , K , X tm , X t1 + , K , X t m + , t1,, tm
, , 1 t1, , tm n 1 t1+, , tm+ n.
,
. , m = 1
xt ,
Xt t, , t
(, , ), :
E (Xt) = D(Xt)= 2.
,
xt,
.
, ,
, , ,
, .
Xt Xt+

Cov( X t , X t + )
,
Corr(Xt , Xt +) =
D( X t ) D( X t + )
Cov(Xt , Xt +) = E [(Xt E(Xt))(Xt + E(Xt+))] .

. . .. www.iet.ru
xt , Cov(Xt , Xt+) t
; () :
() = Cov(Xt , Xt +) .
,
D(Xt) = Cov(Xt , Xt ) (0) .
,
Corr(Xt , Xt +) ; (),

() = Corr(Xt , Xt +) = () (0) .
, (0) = 1.
xt
x1, x2, , xn .
xt ,
E(Xt) ,
D(Xt) 2,
Cov(Xt , Xt +) = () t .
, ,
( ,
).
,
. ,
,
/ . (
.) , ,
, , , E( X t3 ) t .
xt , t = 1, , n, ,
X1, ... , Xn n- .
.
, xt , (
) ,
( ).
, xt E(Xt) , D(Xt) 2 () = Corr(Xt , Xt+).
()
,
( ).
() = Cov(Xt , Xt +) .
()
(). , ..
. 1
+1; (0) = 1. , xt , () = (),

.
() .
,
. , xt
. . .. www.iet.ru
c ,
.
xt
(xt + c)
,
,
X 1, ..., X n n+1 : , (0), (1), , (n 1) (
, (0), (1), , (n 1)). , ,
, . ,

, ,
.
, , ,
.
2.2.
( ,
) xt ,
E(Xt) 0, D(Xt) 2 > 0
() = 0 0.
, t s Xt Xs ,
t s, .
, Xt , X 1, ...,
X n N(0, 2),
, .. Xt ~ i.i.d. N(0, 2).
.
, , X1, ... ,
Xn , ,
, .. Xt
/ (
).
, ,
- t s Xt Xs
.
(NOISE) D(Xt) 0.04.
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
50
100
150
200
250
300
350
400
450
NOISE
. . .. www.iet.ru

, .
, ,
,
.
,
, t .
,
, , , ,
() - 1984 ( ).

4
3
2
1
0
-1
-2
-3
50
100
150
200
250
DOW-JONES_TEMP
, ,
(

xt
),
.
2.3.

( ).
:
Xt = a Xt 1 + t , t = 1, , n,
t ,
2, X0 , a 0
.

E(Xt) = a E(X t 1),
E(Xt) =
0 t = 0, 1, , n.
,
Xt = a X t 1 + t = a (a Xt 2 + t1) + t = a2 Xt2 + a t1 + t = =
= a t X0 + a t 1 1 + a t2 2 + + t ,
Xt1 = a Xt2 + t1 = a t1 X0 + a t2 1 + a t3 2 + + t1 ,
Xt2 = a Xt3 + t2 = a t2 X0 + a t3 1 + a t4 2 + + t2,
X1 = a X0 + 1.
. . .. www.iet.ru
X0 1, 2,
, n, ,
Cov(X0, 1) = 0, Cov(X1, 2) = 0, , Cov(Xt2, t1) = 0, Cov(Xt1, t) = 0
D(Xt) = D(aXt1 + t) = a2D(Xt1) + D(t), t = 1, , n.

, ,
D(X0) = D(Xt) = X2 t = 1, , n,
:
X2 = a2X2 + 2.
a2 < 1, ..
a < 1.
X2
X2 = 2 (1 a2) .
,
Cov(Xt , Xt +) = Cov(a t X0 + a t1 1 + a t2 2 + + t ,
a t+X0 + a t+1 1 + a t+2 2 + + t+) =
= a2t+D(X0) + a (1 + a2 + + a2(t1))2 =
= a [a2t2 (1 a2) + (1 a2t)2 (1 a2)] = [a (1 a2)]2,
Corr(Xt , Xt +) = a ,
..
, .
, ,
Xt = a Xt1 + t , t = 1, , n,
,
a<1;
X0 1, 2, ,
n ;
E(X0) = 0 ;
D(X0) = 2 (1 a2) .
Corr(Xt , Xt +) = () = a .
( ) ,
.
yt E(Yt) = , ,
Xt = Yt :
Yt = a (Yt1 ) + t , t = 1, , n,

Yt = aYt1 + + t , t = 1, , n,
= (1 a) .

, .
. . .. www.iet.ru
Xt (
), ,
(1) = E(Xt Xt1) = E [(a Xt1 + t) Xt1] = a (0),

(1) = (1) (0) = a ,
a > 0, 1,
,
. a < 0
, , ,
.

, Xt = a Xt1 + t 2 = 0.2
a = 0.8 ( ) a = 0.8 ( ).
1.5
1.5
1.0
1.0
0.5
0.5
0.0
0.0
-0.5
-0.5
-1.0
-1.0
-1.5
50
100 150 200 250 300 350 400 450

a = 0.8
50
100 150 200 250 300 350 400 450

a = - 0.8
.
X0 = x0 ,
Xt = a Xt1 + t Xt
, ,

, , Xt = a Xt1 + t a 2.
, t = 0
, x0
. Xt a
< 1. .
,
t = 1, x0
.
Xt = a t x0 + a t1 1 + a t2 2 + + t ,
E(Xt) = a t x0 + a t1E(1) + a t2E(2) + + E(t) = a t x0 ,
D(Xt) = (a2(t1) + a2(t2) + + 1) 2 = [(1 a2t) (1 a2)]2
= 2 (1 a2) [a2t (1 a2)]2,
Cov(Xt , Xt +) = Cov(Xt a t x0 , Xt + a t+ x0) = a (1 + a2 + + a2(t1))2
= a (1 a2t)2 (1 a2) ,
Xt ,
Cov(Xt , Xt +) t .
, a < 1, t
E(Xt) 0 , D(Xt) 2 (1 a2), Cov(Xt , Xt +) a [2 (1 a2)],
. . .. www.iet.ru
.. t
Xt , Cov(Xt , Xt +) ,
.
, a < 1
, Xt = aXt1 + t X0 = x0 .
Xt ,
t 1
X t = a t x0 + a k t k
, a<1,
k =0
~
X t = a k t k .
k =0
~
X t X t = a t x 0 + a k t k ;
k =t
a x0 0 E a t k
t
k =t
= 2 a 2 k 0 .
k =t
~
, X t Xt ; Xt
~
~
X t t . Xt X t
, X0 a .
~
X t

~
E X t = E a k t k = a k E ( t k ) = 0 ,
k =0
k =0
( )
2

~
D( X t ) = D a k t k = a k D( t k ) = 2 a 2 k = 2 ,
1 a
k =0
k =0
k =0
~ ~
,
Cov( X t , X t + ) = E a k t k a k t + k = a a 2 k E t2k = a
2
1
a
0
0
0
k
k
k
=
=
=
~
X t ( ). ,
( )
1
~
X t 1 = a k t k ,
a k =1
~
~
aX t 1 + t = a k t k = X t ,
k =0
~
.. X t
~
~
X t = aX t 1 + t .
~
~
X t 1 , X t 2 , ... ,
~
~
t X t 1 , X t 2 , , .. t
t
. . .. www.iet.ru
~
(). , X t
, ,
AR(1).

xt, Xt = a Xt1 + t = 0.2
a x0.
x0=2; a=0.5
x0=0; a1=0.5
2.5
2.5
2.0
2.0
1.5
1.5
1.0
1.0
0.5
0.5
0.0
0.0
-0.5
-0.5
10
20
30
40
50
60
70
80
90
100
10
20
30
40
50
60
70
80
90
100
70
80
90
100
x0=2; a=0.9
x0=0; a=0.9
2.5
2.5
2.0
2.0
1.5
1.5
1.0
1.0
0.5
0.5
0.0
0.0
-0.5
-0.5
10
20
30
40
50
60
X
70
80
90
100
10
20
30
40
50
60
X
Xt = a Xt1 + t
. p (
AR(p))
Xt = a1 Xt1 + a2 Xt2 + + ap Xtp + t , ap 0,
t D(t) = 2 .
, Cov(Xts, t) = 0 s > 0; ,
t () ,
t t .
, t
p
t ,
, , , t ,
Xt .

L (lag operator),

L Xt = Xt1 ;
. . .. www.iet.ru

B (backshift operator).
k , Lk ,

Lk Xt = Xtk .
a1 Xt1 + a2 Xt2 + + ap Xtp

(a1L + a2 L2 + + ap Lp) Xt ,
, p- ,
a(L) Xt = t ,
a(L) = 1 (a1L + a2 L2 + + ap Lp).

, ,
a(z) = 0
( ) z 1. (
, AR(1) a(z) = 1 a z , a(z) = 0 z
= 1/a , z > 1 a <
1. ) a(L) Xt = t
Xt =
1
t =
a( L)
b j t j ,
j =0
<,
j=0
, , ,

E(Xt) = E b j t j = b j E ( t j ) = 0 .
j =0
j =0
AR(p)

a(L) ( Xt ) = t ,
a(L) Xt = + t ,
= a(L) = (1 a1 a2 ap) = a(1).

a(L) ( Xt ) = t
1
t .
Xt = +
a ( L)
, AR(p) a(L) Xt = + t ,
,
,
(1 a1 a2 K a p )
. . .. www.iet.ru
10
AR(1) a(L) = 1 aL , ( ,
)
Xt = (1/ (1 aL)) t = (1 + aL + a2L2 + ) t = t + at1+ a2t2 + .
,
(k) = Corr(Xt , Xt+k) = a k , k = 0, 1, 2, .
0 < a < 1 ( (k) k = 0, 1, 2, )
;
1 < a < 0 .
AR(1) a = 0.8 :
1.0
1.5
0.8
1.0
0.6
0.5
0.4
0.0
0.2
-0.5
0.0
-1.0
5
10
15
20
a = 0.8
25
30
35
10
15
20
25
30
35
a = - 0.8
AR(p) p > 1 ,
( ) a(z) = 0.
k (k)
Ak, = 1/zmin zmin
a(z) = 0, ,
A k . A > 0
, a1 , , ap .
Xtk (k >0) ,
AR(p), ,
(k) = a1 (k1) + a2 (k2) + + ap (kp) , k > 0 .

(0),
(k) = a1 (k1) + a2 (k2) + + ap (kp) , k > 0 .

, p , aj
p ,
(. 3.1
3.2).
. AR(2)
Xt = 4.375 + 0.25Xt1 0.125Xt2 + t .
a(z) = 0
1 0.25z + 0.125z2 = 0, z2 2z + 8 = 0,
z 1,2 = 1 i 7 .
, .
. . .. www.iet.ru
11
= / (1 a1 a2) = 4.375/(1 0.25 + 0.125) = 5,

5.
.
p = 2,
(k) = 0.25 (k1) 0.125 (k2), k > 0 .
, (0) = 1. (1)
(1) = 0.25 (0) 0.125 (1) = 0.25 0.125 (1),
:
(1) = 0.25 / (1 + 0.125) = 2/9 = 0.222.
:
(2) = 0.25 (1) 0.125 (0) = 0.25 * 0.222 0.125 = 0.069,
(3) = 0.045, (4) = 0.003, (5) = 0.005 ..
, ,

.
.
10
1.2
1.0
0.8
6
0.6
0.4
0.2
2
0.0
0
-0.2
0
RHO
50
100 150 200 250 300 350 400 450

AR(2)
2.4.

q (MA(q)). ,
Xt = t + b1 t1 + b2 t2 + + bq tq , bq 0 ,
t .
.
, ,
Xt = t + b1 t1 + b2 t2 + + bq tq ,
..
Xt = + t + b1 t1 + b2 t2 + + bq tq .
q MA(q)
( moving average).
q = 0 = 0 . q = 1 ,
Xt = t + bt1
.
D(Xt) = (1 + b2)2 , E [(Xt )(Xt1 )] = b2 , E [(Xt )(Xtk )] = 0, k > 1,
. . .. www.iet.ru
12
Xt
E(Xt) = 0 , D(Xt) = (1 + b2)2 ,
1 + b 2 2 , k = 0 ,
(k) = b 2 ,
k = 1,
0 ,
k > 1.

k = 0,
1 ,
2
(k) = b (1 + b ) , k = 1 ,
0 ,
k > 1,
..
.
.
, b > 0, b < 0. ,
MA(1) b > 0 , , ,
MA(1) b < 0 , ,
. , MA(1)
(1) 0.5 ,
.. ,
AR(1) ( a ,
1).
MA(q)
Xt = b(L) t ,
b(L) = 1 + b1L + + bq Lq .

(k) = E [(Xt )(Xtk
qk
2
b j b j + k , 0 k q ,
)] = j = 0
0,
k >q,
MA(q)
,
X2 = (1 + b12 + + bq2)2

qk
q 2
b , k = 0, 1, K , q ,
b
b

k = j = 0 j j + k j = 0 j
0,
k = q + 1, q + 2, K .
q
(.. q).
,
,
. . .. www.iet.ru
13
,
,
.

, bj = a j , 0 < a < 1, , t
,
MA()
Xt =
t j =
j =0
j =0
j t j
< .
j=0
,
AR(1)
Xt = a Xt1 + t , a < 1,
.. MA() AR(1).
, AR(p)
MA():
Xt = +
1
t = + b j t j = + b(L) t ,
a ( L)
j =0
b(L) =
j
j L =
j =0
1
a ( L)
bj < .
j=0
.
a) MA(1) b = 0.8 E(Xt) = 6 , ..
Xt = 6 + t + 0.8 t1 .
(1) = 0.8/(1+ 0.82) = 0.488.
b) MA(1) b = 0.8 E(Xt) = 6
(1) = 0.8/(1+0.82) = 0.488.

1.0
1.0
0.8
0.8
0.6
0.6
0.4
0.4
0.2
0.2
0.0
0.0
-0.2
-0.2
-0.4
-0.4
-0.6
-0.6
0
b = 0.8
b = - 0.8
2 = 1:
. . .. www.iet.ru
12
12
10
10
14
0
50
100 150 200 250 300 350 400 450
50
100 150 200 250 300 350 400 450
b = 0.8
b = - 0.8
c) MA(2) Xt = 5 + t 0.75t1 + 0.125 t2 :

(1) = (b0b1 + b1b2)/(b02 + b12 + b22)
= ( 0.75 0.750.125)/(1 + 0.752 + 0.1252) = 0.535,
(2) = 0.125/1.578 = 0.079.
.
1.2
10
0.8
0.4
0.0
-0.4
-0.8
0
0
50
rho
100 150 200 250 300 350 400 450

MA(2)
2.5. (
)
Xt ,
, p q AR A
ARMA(p, q) (autoregressive moving average, mixed
autoregressive moving average). , Xt
ARMA(p, q),
p
Xt =
a j Xt j +
j =1
t j ,
ap 0 , bq 0 ,
j =0
t b0 = 1.

a(L) Xt = b(L) t ,
a(L) b(L) , AR(p)
MA(q). ,
ARMA(p, q),
Xt =
a
j =1
(Xt j ) +
t j .
j =0
ARMA(p, q) E(Xt) = .
, a(z) = 0
z 1.
. . .. www.iet.ru
15
,
MA()
Xt =
c j t j , c0 = 1,
j =0
cj < ,
j=0
Xt = c(L) t ,
c(z) =
c z
j=0
b( z )
.
a( z)
b(z) = 0
z1
( ),
Xt AR()
Xt =
(Xt j ) + t ,
j =1
d(L)(Xt ) = t ,
d(z) = 1
d
j =1
zj =
a( z )
.
b( z )
, ARMA(p, q)
,

.
ARMA(p, q)
, AR(p) MA(q). , k > q
a(L) Xt = b(L) t ,
a(L) Xt = t . , ARMA(1, 1)
(k) = a1 (k 1) k = 2, 3, ,
Xt = a1 Xt1 + t . , , (1) a1 .
ARMA
. ARMA(p1, q1) Xt ARMA(p2, q2) Yt
, Zt = Xt + Yt , , Zt
ARMA(p, q) ,
p = p1 + p2 ,
q = p1 + q2 , p1 + q2 > p2 + q1 ,
q = p2 + q1 , p2 + q1 > p1 + q2 .
, p q
. ( , aX(z) aY(z) ,
Xt Yt , .)
, AR(1), ,
ARMA(2, 1).
.
, AR,
ARMA .
. . .. www.iet.ru
16
, AR,
MA. ,
MA MA .
, , AR(p)
, ,
(.. MA(0)). ARMA(p, p).
, ARMA(p, q) Xt
,
AR(). , ,
AR(p), , .
,
ARMA, AR MA . ,
, , ,
( )
. ARMA
.
2.6. ARMA,
,
ARMA, , ,

.

(SAR(1))
Xt = a4 Xt4 + t , a4 < 1
(SMA(1))
Xt = t + b4 t4 .

(k) = a4k/4 k = 4m, m = 0, 1, 2, ,
(k) = 0 k > 0.

(0) = 1, (4) = b4 , (k) = 0 k > 0.
SAR(1) a4 = 0.8
SMA(1) b4 = 0.8.
6
-2
-2
-4
-4
10
20
30
40
50
60
X_SAR
70
80
90
100
10
20
30
40
50
60
70
80
90
100
X_SMA
. . .. www.iet.ru
17
, ,
ARMA((1, 4), 1)
Xt = a1 Xt1 + a4 Xt4 + t + b1 t1
ARMA(1, (1,4))
Xt = a1 Xt1 + t + b1 t1+ b4 t4 .
C
a1 = 2/3, a4 = 1/48, b4 = 1/5 a1 = 0.4, b1 = 0.3,
b4 = 0.8 .
4
-2
-2
-4
-4
10
20
30
40
50
60
70
ARMA((1, 4), 1)
80
90
100
10
20
30
40
50
60
70
80
90
100
ARMA(1, (1,4))
, a(z) = 0 1 2/3z + 1/48

z = 0, .. z4 32z + 48 = 0; z1 = 2, z2 = 2, z3 = 2 + i8, z4 = 2
i8
,
. a(z) = 0 1 0.4 z = 0;
z = 2.5 > 1, .
,
, ,
(1 a1L)Xt = (1+ b1L)(1+ b4L4) t ,
(1 a1L)(1 a4L4) Xt = (1+ b1L) t .

Xt = a1 Xt1 + t + b1 t1+ b4 t4 + b1b4 t5 ,

Xt = a1 Xt1 + a4 Xt4 a1a4 Xt5 + t + b1 t1 .

1 4 (.. t1 t4),
1 4 (.. Xt1 Xt4 ). ,

Xt = a1 Xt1 + t + b1 t1+ b4 t4 + b5 t5 ,
Xt = a1 Xt1 + a4 Xt4 + a5 Xt5 + t + b1 t1
c b5 = b1b4 , a5 = a1a4 .
( , ),

,
(parsimony model). , - ,
(
), .
ARMA , ,
[Enders (1995)].
4
. . .. www.iet.ru
18
. . .. www.iet.ru
3. ARMA
, x1, x2, , xT
ARMA,
, :
1. ;
2. ;
3. .

ARMA, .. p q.
,

.
a1, a2, , ap, b1, b2, , bq .

.
, ,
X1, X2, ,

.

(misspecification tests).
, ,
, ,
.. , .
, ,

, .
ARMA(p, q),
Xt ~ ARMA(p, q). ,
AR(p), Xt ~ AR(p), MA(q), Xt ~ MA(q).
3.1. ARMA
ARMA

(ACF)
(PACF) (ACF autocorrelation function, PACF partial
autocorrelation function) , ARMA.
ARMA
.
( MA )
.
. . .. www.iet.ru
(PACF) Xt .
part(k) k
Xt Xt+k , Xt+1 ,
, Xt+k1 .
, part(k) Xtk
Xt1, , Xtk ,
Xt . , (., ,
[Hamilton (1994)]), part(k) ak
k
(s) = a1 (s1) + a2 (s2) + + ak (sk) , s = 1, 2, , k ,

(s1) a1 + (s2) a2 + + (sk) ak = (s) , s = 1, 2, , k ,
, a1 , a2 , , ak , (1k), , (k1)
.
k k , ,
PACF
part(0) = 1,
part(1) = (1),
1
(1)
(1) (2) (2) 2 (1)
part(2) =
,
=
1
(1)
1 2 (1)
(1) 1
(1)
part(3) = (1)
1
(2)
(2) (1) (3)
1
(1)
(1) (2)
(1)
1
(1) ,
(2) (1)
1
1
1
(1)
(2) K (1)
K (2)
(1)
1
(1)
K (3)
(2)
(1)
1
M
M
M
O
M
part(k) =
(k 1) (k 2) (k 3) K (k ) .
1
(1)
(2) K (k 1)
K (k 2)
(1)
1
(1)
K (k 3)
(2)
(1)
1
M
M
M
O
M
(k 1) (k 2) (k 3) K
1

part(k)

, (1), (2), ..., (k).
. . .. www.iet.ru
, Xt AR(p),
part(p) 0 ,
part(k) = 0 k > p.
PACF
ARMA(p, q) q
> 0.
, ACF q MA(q).
, PACF p AR(p).
ACF PACF
ARMA(p, q) p 0, q 0.
,
(k) part(k)
ACF,
r (k ) =
1
T k
T k
(x
t =1
1
T
)(xt +k )
(x
t =1
(k )
, k = 1,..., T 1 ,
(0)
= x =
(k ) =
1 T
xt = E(Xt) ,
T t =1
1 T k
(xt )(xt +k ) (k),
T k t =1
PACF,

rpart(k) . ,
part(k) (s) r(s).
, , part(k) Xtk
Xt1, , Xtk ,
Xt . ,

Xt = a1 Xt1 + a2 Xt2 + + ak Xtk + ut
( ut
ut = Xt (a1 Xt1 + a2 Xt2 + + ak Xtk) ,
- ).
ak rpart(k) .
Xt ARMA(p,q) E (X t4 ) < ,
, (k ) , r(k) rpart(k) ,
(k), (k) part(k), . (. [Hamilton (1994), p.199].) r(k)
rpart(k) (k) part(k),
(k) part(k). , k =k1 k
=k2 , , (k1) = 0 part(k2) = 0, , , r(k1)
0 rpart(k2) 0,
. ,

.
. . .. www.iet.ru
, ACF PACF -
.
ACF PACF
ARMA.
ACF PACF
.
ACF
PACF
(k) = 0 k 0
part(k) = 0 k
0
MA(0)
AR(1)
(k) =
part(1) = a1
part(k) = 0, k 2
(k) =
part(1) = a1
part(k) = 0, k 2

MA(1)
b1 > 0
k = 1;
k >
1
MA(1)
b1 < 0
k = 1;
k >
1
MA(q)
k
p
ARM
A(1, 1)
1;
a1 > 0
(1)
(a1+ b1)
ARM
A(1, 1)
1;
a1 < 0
(1)
(a1+ b1)
,
a1 > 0
AR(1)
a1k
a1 < 0
AR(p)
a1k
ARM
A(p, q)
;
part(1) > 0
;
part(k) < 0 k
1

1;
part(1) = (1)
1;
part(1) = (1);
part(k)

(1), k > 1
,
. . .. www.iet.ru
1)

q
p
SAR(1
,
; ;

SMA(
,
;
;
AR(p) MA(q)
r(k) rpart(k) ,
(k) part(k)
r(k) rpart(k). ,
,
(.., t ).
Xt MA(q), n
q
1
D(r (k )) 1 + 2 2 ( j ) k > q ,
T
j =1
,
(k), k > q . ,
limT E(r(k)) = (k) .
, T k > q r(k)
. ,
H0 : Xt MA(q)
, ,
r (k ) >
2
T
1 + r 2 ( j)
j =1
k > q . 0.05.
, q = 0, Xt ~ MA(0) , H0: Xt

2
r (k ) >
, k > 0.
T
(2) Xt AR(p), T k > p
rpart(k)
rpart(k) N (0, T 1) ( D(rpart(k)) T 1 ).
, H0: Xt ~ AR(p)
2
rpart (k ) >
, k > p,
T
, 0.05.
. . .. www.iet.ru
,
ACF
PACF, 2/ T . ,
0.95, r(k), Xt ,
rpart(k), Xt ~ AR(p). . ,
, 0.05,
H0 k .
, , ?
, .
T = 499 x1, x2, , x499.

() k = 1, 2, , 36.
A
CF
0.019
0
.038
0
0.019
0
0
.079
.021
.017
0.083
.035
0.049
.069
.041
0.014
0.035
.034
0.064
.032
0.057
0.053
.011
.034
.029
0.042
.013
.046
0.006
0
.021
0
.034
0.057
0
.012
0.044
0.018
0.075
0.057
0
.042
0
.022
0.035
0
.032
0.018
0.085
0.031
0
.028
0
.034
0.015
0.115
0
.099
0
.034
0
.073
0.053
0.036
0
.069
5
-
0.051
P
ACF
0
.053
.061
0.044
.062
A
CF
.126
-
0
.009
0.024
0.071
0.049
.017
0
.035
0.047
.102
0.083
P
ACF
0
3
0.014
0.083
.074
CF
0.019
0.013
ACF
0
.064
0
.055
ACF, , 2/T = 0.0895

r(13) = 0.102. ,
H0: Xt ? PACF,
, , .
(T = 499),
rpart(k), k = 1, 2, .
Bk , , rpart(k) 2/T .
0.05.
( 36) rpart(k), k = 1, 2, , 36,
P2 = C362 (0.05)2(1 0.05)362 = (3635/2) 0.0520.9534,
. . .. www.iet.ru
lg(P2) = lg(630) + 2lg(0.05) + 34lg(0.95) = 0.560.

: P2 = 0.275,
PACF 36 .
ACF,

r(k), k = 1, 2, , Xt ( MA(q) q 1
). ACF

P1 = C361 (0.05)(1 0.05)35 ,
lg(P1) = 0.780, : P1 = 0.166.
,
ACF PACF .
, , r(k)
EVIEWS, : (k )
T k , T . , (k)
.
ACF
PACF Q-,
,
.

Q-.
( ) [Box, Pearce (1970)]

Q =T
(k ) .
k =1

r(1), r(2), , r(M) , Xt , ,
T T r(k) N (0, 1), T r2(k) [N (0, 1)]2 = 2(1) .
(, Xt . [ (1974)].)
, T
Q ~ 2(M).
H0 .
0.05,

Q > 20.95(M).
P- Q
M = 1, 2, . M H0
, P- 0.05.
, ,
2(M) T .
[Ljung,
Box (1979)]
r 2 (k )
,
k = 1 (T k )
M
Q = T (T + 2)
. . .. www.iet.ru
( T ) 2(M),
T ,
. EVIEWS (Econometric Views)
P-,
2(M).

Q-
. P- (Prob) Q-
.
Prob
Prob
13
.670
.292
.066
0
.037
.044
.044
.033
.187
.065
.061
.076
.084
0
0
0
0
0
.037
.146
.072
.360
.438
7
0
.539
.063
0
6
.455
.077
0
.049
.349
0
6
.045
0
.348
0
.061
4
0
.243
25
.064
.873
Prob
.096
0
.049
.056
.064
.099
.119
.119
P-, M = 14, 15, 17 22, 0.05,

M
H0: Xt ,
M P- , 0.05, H0
M .
- , ,
, M , .

[Kwan (1996a,b)].
. . .. www.iet.ru
,
ARMA
.
,
.
, .
.
AR , .. ,
Xt AR(k)
k
kj
(Xt j ) = t , ak0 = 1,
j =1
k , k
[Akaike (1973)].
, k ,

AIC(k ) = ln k2 + (2 k T )
T ,
k2 t AR
k- . k2 k-

k
s = ak j s j ,
j =1
a k j , j = 1, , k , ak j
, x , t ,
k
t = a k j (xt j x ) ,
j=0
k2
k2 =
1
T
2
t
t =1
,
k0
()
. , ,

ln k2 + k cT ,
cT = O(T 1 ln T) (.. cT T , T 1
ln T ).

SIC [Schwarz (1978)],
lnT
.
SIC = ln k2 + k
T
[Hannan, Quinn
(1979)], cT = 2ck T 1lnlnT , c > 1,
. . .. www.iet.ru
10
2cln ln T
,
T
k0 T .

T
.
HQ = ln k2 + k
AR(2)
Xt = 1.2 Xt-1 0.36 Xt-2 + t .
a(z) = 0
1 1.2 z + 0.36 z2 = 0
z = 5/3 > 1, ,
.
C t = 1, 2, , 500
.
8
-4
-8
50
100 150 200 250 300 350 400 450 500

PRIM1
:
A
ACF
PACF
C
PAC Q-Stat Prob
.|*******
.|*******
***|.
.|******
.899
.732
.|.
.|****
.561
.|.
.|***
.409
.|**
.|.
.277
.|*
.|.
.167
.|*
.|*
.095
.|.
.|.
.045
.|.
.|.
0
4
0
.899 06.25 .000
0
6
0.396 75.97 .000
0
8
0.005 34.98 .000
0
9
0.027 19.59 .000
0
9
0.048 58.40 .000
0
9
0.015 72.59 .000
9
0
0
.071 77.15 .000
0
9
0.045 78.19 .000
0
0
9
. . .. www.iet.ru
.014
.|.
.|.
.|.
.|.
.|.
.|.
11
.011
78.29 .000
0
9
0 0.001 .020 78.29 .000
0
0
9
1 .003 .055 78.30 .000
9
0
0
2 .019 .001 78.49 .000
2/T = 0.089 PACF,

k = 1, 2. ,
, H0: Xt ~ AR(2).

, AR 4-, 3-, 2- 1-
, AR
.
p=
1
p
=2
p=
3
p=
4
3.0
2.9
2.9
2.
83264
19244
24441
91800
3.1
2.9
2.9
2.
00148
53116
66846
94336
AR(2).
IC
IC
AR , ,
, ARMA(p0, q0) ( p0, q0)
a(L) Xt = b(L) t ,
(p0, q0),
(. [Kavalieris (1991)]).

AR(k)
k
a
j=0
k j
X t j = t , ak 0 = 1 ,
a k j , j = 1, , k,

k
k (t ) = a k j xt j , a k 0 = 1 .
j=0
k
. ,
. (,
,
.)
Xt Xt j , j = 1, , p, Xt k (t
j), j = 1, , q.
. . .. www.iet.ru
12
aj , .. a k j , j = 1, , p, b j bj , j = 1, , q.
, a(z), b(z)
p
j=0
j=0
a( z ) = a j z j , b( z ) = b j z j ,

= b 1 ( L) a ( L) x ,
t
~ p2, q =
1
T
t =1
, , ARMA
, t = b 1 ( L) a ( L) xt
.

p0 , q0
( ~p, q~ ) ,
ln T
SIC ~ p2 , q = ln~ p2, q + ( p + q )
.
T
, SIC ~ p2, q p q , p p0 , q
q0 , ( ~
p, q~ ) .
3.2.
() ARMA, ..
p, q
ARMA(p, q), ,
. ,
.
(, AR(1)),
, ()
, .

, . ,
(k) r(k)

(k) . ,
AR(p), a1, , ap p

p
(k ) = a j (k j ), k = 1,K, p ,
j =1
. . .. www.iet.ru
13

(1), , (p)
( ) r(1), , r(p)
.
MA(q) (q > 0)
, . 2.5. MA(1)
Xt = t + bt1 , t = 1, , T .
x1, x2, , xT , 1, 2,
, T () 0 :
1 = X1 b0 ,
2 = X2 b1 = X2 b(X1 b0) = (X2 ) b(X1 ) + b2 0,
T = XT bT 1 = (XT ) b(XT 1 ) + b2(XT 2 )

+(1)T 1 b T 1 (X1 ) + (1)T b T 0 .
( b) ,
x1, x2, , xT 0 ,

Q(b) = 12 + 22 + + T2 ,
b .
,
, , ()
b . ,
.
b 0 ,
. ,
b < 1, b 1.
, 0 = 0 .
0 ,
, 0 = 0,
, 0 ,
. MA(q) q
> 1 : 0 = 1 = = q + 1 = 0 .
, ( EVIEWS)
(backcasting) ,
0 , 1 , , q + 1
.
,
,
.
ARMA
, , [Hamilton
(1994)]. ,
,
.
.
AR(p)
a(L) Xt = + t .
. . .. www.iet.ru
14
,
Xt
(1 a1 a2 K a p )
a1, , ap ,

Xt = + a1 Xt1 + a2 Xt2 + + ap Xtp + t ,
, a1 , ... , a p ,
=
.
(1 a1 a 2 K a p )
, , ,
EVIEWS (Econometric Views), . ,

Xt = (1 a1 a2 ap) + a1 Xt1 + a2 Xt2 + + ap Xtp + t
a1, , ap .
. ,
, MA ,
(NLLS nonlinear least squares)
.
(
, ).
946
0.8
956
0.4
947
0.3
957
0.2
948
1.1
958
0.6
949
0.9
959
0.9
950
1.8
960
0.3
951
1.2
961
0.7
952
1.2
11.4
962
0.6
963
0.7
964
0.5
953
954
1.2
. . .. www.iet.ru
955
0.5
15
0.9
965
:
12.0
11.5
11.0
10.5
10.0
46
48
50
52
54
56
58
60
62
64
, ,
Autocorrelati
on
. |***
Partial
Correlation
. |***
.429
. |***
. |**
.366
. | .
**| .
.059
. | .
. | .
.016
*| .
*| .
0.156
**| .
**| .
0.255
***| .
*| .
0.321
. *| .
. |*
0.133
A
PAC
0
.429
0
.222
0
0.204
0
0.034
0.129
0.195
0.123
.175
Q-Stat Prob
0
.2694
0
.5350
.6255
.6324
.3498
0.393
3.879
0
4.523
4
.039
7
.023
7
0
.054
0
7
.106
8
.138
1
0
.109
0
1
.053
1
.069
,
2/T = 2/20 = 0.447. 0.447
. ,
,
MA(0)
X0 = + t .
, , ACF
1 . ,

AR(1) MA(1). ,
: , ,
. -
,
, , ,

.
. . .. www.iet.ru
16
,
,
, -
, .

,
1940 1945 ( ).
1940 1965 :
12
11
10
9
8
7
40
42
44
46
48
50
52
54
56
58
60
62
64
1942 1944 ..
1940 1965 .
1946 1965 ..
AR(1), , , (1) = a1.

(1) r(1) = 0.429,
a1. ,
AR(1)
, .
Dependent Variable: X
Method: Least Squares
Sample(adjusted): 1947 1965
Included observations: 19 after adjusting endpoints
Convergence achieved after 3 iterations
Variable
Coef
.
Std.
Error
10.8
1451
AR(1)
0.159
261
0.43
0515
R-squared
Pro
b.
67.90
445
0.219
257
0.18
4864
tStatistic
0.0
000
1.963
522
0.0
662
Mean dependent
var
10.
81053
Adjusted
R0.13
S.D. dependent
0.4
squared
6915
var
25434
S.E.
of
0.39
Akaike
info
1.0
regression
5238
criterion
80645
Sum squared
2.65
Schwarz criterion
1.1
resid
5627
80060

a1 .
MA(1), (1) = b1/(1 + b12).
(1) r(1) = 0.429
b1/(1 + b12) = 0.429. 0.567 1.704.
. . .. www.iet.ru
17
MA(1) ;
MA(1) . MA(1)
(backcasting) .
Sample: 1946 1965
Included observations: 20
Backcast: 1945
Variable
Coef
.
Std.
Error
10.8
1379
MA(1)
0.113
355
0.28
0610
R-squared
Pro
b.
95.39
726
0.228
102
0.0
000
1.230
195
0.11
7961
tStatistic
0.2
345
Mean dependent
var
10.
81000
Adjusted
R0.06
squared
8959
var
S.E.
of
0.39
regression
9561
criterion
Sum squared
2.87
resid
3684
Log likelihood
8.977395
Durbin-Watson
1.78
stat
9895
S.D.
dependent
0.4
14094
Akaike
info
1.0
97739
Schwarz criterion
1.1
97313
F-statistic
2.4
07257
Prob(F-statistic)
0.1
38178
b1
. P t- F- b1 .
,
, 20
.
1945
, :
Sample: 1946 1965
Backcast: OFF
Variable
Coef
ficient
10.8
1515
MA(1)
0.27
R-squared
0.229
0.11
R-
0.0
000
1.195
405
0.2
474
Mean dependent
var
0.06
Pro
b.
96.06
431
231
6800
tStatistic
0.112
582
4024
Adjusted
Std.
Error
10.
81000
S.D.
dependent
0.4
. . .. www.iet.ru
squared
7734
var
S.E.
of
0.39
regression
9824
criterion
Sum squared
2.87
resid
7464
Log likelihood
8.990543
Durbin-Watson
1.77
stat
8286
18
14094
Akaike
info
1.0
99054
Schwarz criterion
1.1
98628
F-statistic
2.3
80443
Prob(F-statistic)
0.1
40261
3.3.

, .. ,
( )
.

, H0 ,
, , t
.
ARMA(p, q)
a(L) Xt = b(L) t ,
..
Xt = a1 Xt1 + + ap Xtp + t + b1 t1 + + bq tq ,

a ( L) X = b( L) ,
t
a ( L) = 1 a1 L K a p Lp ,
b( L) = 1 + b1 L + K + bq Lq .
MA ARMA(p, q) ,
a ( L)
t =
Xt ,
b( L )
t a(L) b(L)
a ( L)
b( L) , :
t =
a ( L)
Xt .
b( L)
, , ,
-
.
t
t . C, t ,
.
. . .. www.iet.ru
19
, [Bartlett (1946)] [Box,

Pierce (1970)]
t
T k
r (k ) =

t =1
T
t t+k
2
t
t =1

QBP = T
r (k ) (Q- ).
2
k =1
, QBP ,
2(M p q), , T M ,
(M/T) . ,
QBP > 20.95(M p q).
, T
QBP 2(M p q).

[Ljung, Box (1978)], ,
M ( M + 5)
E (QBP ) ( M p q )
.
2T + 2
M(M + 5)/(2T + 2) ,
,
2
(M p q) .
.

QBP 2(E(QBP)) ,
E(QBP) (
).
D(r(k)) (1/T)
(T k)/(T2 + 2T). Q-
QLB
r2
= T (T + 2)
,
k = 1 (T k )
M
2(M p q), QBP ,

T QLB 2(M p q),
QBP . ,

; M . ,

, t , ,
t .
, D(t). (
,
, , .
t
, .
t .
.)
. . .. www.iet.ru
20
EVIEWS ARMA
P- Q
.
AR(1) A(1)
.
AR(1):
A
A
P
CF
ACF
C
PAC Q-Stat Prob
*| .
*| .
. |**
. |**
*| .
*| .
. |*
. | .
*| .
. | .
*| .
*| .
**| .
**| .
. | .
. | .
. | .
|*
*| .
**| .
. | .
. | .
*| .
*| .
0
0.096 0.096 .2033
0
0
1
.271 .265 .9334 .164
2
0.116 0.078 .2687 .322
0
2
.076 0.008 .4232 .489
2
0.099 0.049 .7024 .609
3
0.125 0.175 .1852 .671
5
0.257 0.260 .3801 .496
0
0
5
.019 .051 .3928 .612
0
0
5
.047 .189 .4826 .705
6
0 0.178 0.259 .8945 .648
0
6
1 .040 0.050 .9748 .728
8
2 0.187 0.132 .9579 .626
MA(1):
A
P
A
CF
ACF
C
PAC Q-Stat Prob
. |*
0
0
0
. |*
.100 .100 .2306
0
0
3
. |*** . |***
.365 .358 .4838 .062
3
. | .
*| .
0.050 0.127 .5491 .170
0
3
. | .
*| .
.058 0.068 .6417 .303
. . .. www.iet.ru
*| .
*| .
*| .
*| .
*| .
**| .
*| .
. |*
. | .
. |*
4
0.139 0.088 .2051 .379
5
0.169 0.182 .1071 .403
7
0.277 0.199 .6991 .261
0
7
0.066 .094 .8603 .345
0
7
0.021 .159 .8772 .446
9
0 0.187 0.302 .4191 .400
0
9
1 .002 0.042 .4192 .493
1
2 0.187 0.104 1.338 .415
.
*| .
**| .
. | .
. | .
.
*| .
21
*| .
P- QLB 0.05,
, t
, . ,
t : P- Jarque Bera ,
, 0.480 0.608.
.
, ,
() .
, t1, , tn X t 1 , K, X t n
.
,
. ,
()
.
[Lomnicki (1961)].
mk =
1 T
(X t X )k ,
T t =1
G1 =
m3
m
, G2 = 42 3 .
3/ 2
m2
m2
, Xt
, T G1 G2

D(G1 ) =
6
T
3 (k ) , D(G2 ) =
k =
24 4
(k ) .
T k =
. . .. www.iet.ru
22
,
(k) r(k).
D (G1 ), D (G2 ),
G1 =
G1
, G2 =
D(G )
1
G2
,
D (G2 )
,
.
, T
(G ) + (G )
2
1
2
2
2 (2) .
, ,
T
G2 .
, (
). ,
(G ) + (G ) , AR
2
1
2
2
MA, .. Xt ,
.
,
Xt , ,
Xt . AR MA
t , , ,
, ,
N(0, 2).
, :
6
24
D(G1 ) = , D(G2 ) =
,
T
T
(.. (k) = 0 k 0),
G
G
G1 = T 1 , G2 = T 2 ,
6
24

G2 G2
= T 1 + 2 ,
24
6
,
Jarque Bera [Jarque, Bera (1980)].
, Jarque Bera
(
), ,
, , , .
-, EVIEWS
T (T K), K
, .
, .
,
.
?
(G ) + (G )
2
1
2
2
. . .. www.iet.ru
23
T (p, q)
ARMA, ,
,
.

, ,
. ,

,
.
MA(1) AR(1)
, , H0 : a1 = 0 AR(1)
H0 : b1 = 0 A(1) . ,
MA(0)
Xt = + t .
, :
Variable
Coef
.
Std.
Error
10.8
1000
R-squared
0.092
594
0.00
0000
tStatistic
Pro
b.
116.7
460
0.0
000
Mean dependent
var
Adjusted
R0.00
squared
0000
var
S.E.
of
0.41
regression
4094
criterion
Sum squared
3.25
resid
8000
Log likelihood
10.23258 stat
10.
81000
S.D.
dependent
0.4
14094
Akaike
info
1.1
23258
Schwarz criterion
1.1
73045
Durbin-Watson
1.1
38735
,
,
AR(1) MA(1). ,
(1) part(1),
, ,
.
, ,
.
,
, ,
.
[Sargan, Bhargava (1983)] . ,
0.05 T= 21 1.069. ,
, MA(0).
MA(0), MA(1) AR(1) .
. . .. www.iet.ru
M
A(0)
M
A(1)
1
.123
IC
A
R(1)
1
.098
24
1
.081
.173
.197
.180
AR(1),
MA(0).
:
,
.
IC
. 3.1
Xt = 1.2 Xt1 0.36 Xt2 + t . ,
, ()
. AR 4, 3, 2 1 AIC SIC
.

.
Variable
Coef
.
Std.
Error
0.00
1015
AR(1)
0.330
958
1.25
AR(2)
tStatistic
0.003
067
0.041
6580
257
0.397095
290
0.041
Pro
b.
0.9
976
30.45
0.0
723
000
9.617188
000

AC
PA
A
CF
C
PAC Q-Stat Prob
.|.
.|.
.|.
.|.
.|.
.|.
.|.
.|.
0.0
0
0.003 0.003 .0042
0
0.005 0.005 .0165
0
0
0
.000 .000 .0165 .898
0
0
0
.036 .036 .6866 .709
. . .. www.iet.ru
.|.
*|.
.|.
|
0
0
1
.037 .037 .3675 .713
5
0.086 0.085 .0736 .280
0
0
5
.005 .005 .0882 .405
5
0.004 0.006 .0977 .531
5
0.002 0.004 .0993 .648
6
0 0.054 0.050 .5887 .582
6
1 0.014 0.008 .6897 .669
0
0
6
2 .019 .011 .8676 .738
*|.
.|.
.|.
.|.
.|.
.|.
.|.
.|.
.|.
.|.
.|.
.|.
.|.
25
P- QLB 0.05, ,
t ,
. t (P-
Jarque Bera 0.616). ,
Xt
,
AR(2) .
,
Variable
Coef
.
AR(1)
AR(2)
Std.
Error
1.25
tStatistic
0.041
6581
215
0.397096
248
0.041
Pro
b.
30.48
0.0
807
000
9.627056
000
0.0
S.E.
of
1.03
Akaike
info
2.9
regression
6707
criterion
1398
Sum squared
533.
Schwarz criterion
2.9
resid
0816
3089
Inverted
AR
.63
.63+.05i
Roots
-.05i
.
AR(2), AR(3) AR(4) P ,
AR(2).
AR(2)
Xt 1
1.26
Xt 2
0.40
Xt 3
Xt 4
. . .. www.iet.ru
26
1.25
0.39
P = 0.87
AR(3)
1.25
0.40
P = 0.72
P = 0.56
AR(4)
,
,
, ,
. AR(2)
.
, , .
, AR(2),
, a(z) = 0 , .. 1 1.2 z + 0.36 z2 = 0 ,
z = 5/3 1.67. ,
, . , ,
, ,
,
. 0.63 0.05i ,
z = 1.58 0.125i . , ,
() a(z) = 0 ,
, , AR(2)
.
. . .. www.iet.ru
4.

, , ,

.
4.1.
1 , ,
,
( A, A, B, C),

.
, .
AR(p)
yt = + a1 yt 1 + a2 yt 2 + + ap yt p + t ,
t , D(t) =2.

yt = xtT + t ,
xt = (1, yt 1, yt 2, , yt p)T , = ( , a1 , a2 , , ap)T .

, A
B. s xt s t ,
s = t 1, t 1 yt 1,
xt . , A B.
, A C.
,
X N(0, 2V ) ()
V ( A ).
xt + 1 = (1, yt , yt 1, , yt p + 1)T xt t
.
, AR(p) , :
D
yt
yt = + a1 yt 1+ a2 yt 2 + + ap yt p + t
(t );
. . .. www.iet.ru
1 a1z a2 z2 ap z
;
t ~ i.i.d., E( t) = 0, D( t) = 2 > 0, E( t4) = 4 < .
= 0
, n
= ( , a1 , a2 , , ap), n ,

n 1/2 ( ) N(0, 2Q 1 ) ,
n
Q ,
yt .
2 Q1

2
Sn (XnTXn n)1 , ,
, n N(,
Sn2(XnTXn ) 1). ( Xn n
.)
,
,
.
,
, ,

. .
yt
yt = + t + t ,
t ~ i.i.d., E( t) = 0, D( t) = 2 > 0, E( t4) = 4 < .

yt = xtT + t , xt = (1, t)T , = (, ),

n = ( n , n ),

: ( ) T 1/2 , ( ) T
3/2
. ,
n n . ,
N(0, 1) . ,
,
.

t-

.
C E
yt
yt = + t + a1 yt 1 + a2 yt 2 + + ap yt p + t (t );
. . .. www.iet.ru
1 a1z a2 z2 ap z p = 0
;
t ~ i.i.d., E( t) = 0, D( t) = 2 > 0, E( t4) = 4 < .
t- qF
( q , F F
)
N(0, 1) 2(q) .
, .
,
,
,
.

y = X + , X = Xn ,
, ,
yt = xtT + t , t = 1, 2, , n ,
xt = (xt1, xt2, , xtp)T

p t- , n
= ( 1 , 2 , , p)T ,
n . (., , [Green (1997)]),

n n :
1 n
plim
xt t = 0
n t =1
[ : plim (n 1 XnT ) = 0];

1 n
plim xt xtT = Q
n t =1
[ : plim(n 1 X nT X ) = Q ] , Q
;
1 n
xt t N 0, 2Q
n t =1
[ : (n 1/2 XnT ) N(0, 2Q)].

( plim ;
.) , n ,
D,
n ( ) N(0, 2Q 1 ).
n
[Mann, Wald (1943)] ( ).

. . .. www.iet.ru
1 n
plim xt xtT = Q
n t =1
[ : plim (n 1 XnT X ) = Q] , Q
,
t ~ i.i.d., E( t) = 0, D( t) = 2 > 0, E t m < m = 1, 2, ,
E(xt t) = 0, t = 1, 2, , n ,

, n
n .
, E(xt t) = 0, t = 1, 2, , n , E( t) = 0, ,
Cov(xt k , t) = 0 k = 1, 2, , p ,
.. t
. E t m < m = 1, 2, ,
, , t .
.
F

yt = xtT + t , t = 1, 2, , n ,
xt = (xt1, xt2, , xtK)T

K t- , n
= ( 1, 2 , , K), n
. :
1
1 n
plim xt xtT = Q .. plim X nT X n = Q ,

n n t =1
n n
Q ,
t ~ i.i.d., E( t) = 0, D( t) = 2 > 0, E t m < m = 1, 2, ,
Cov(xt k , t) = 0 k = 1, 2, , K .
n
n ( ) N(0, 2Q 1 ).
, xt (K-) ,
E(xt) = = const , Cov(xt) = Q , Cov(xt k , xt +s, l) = kl (s)

t, s k, l = 1, 2, , K. ( kl (s) -
k- l- xt , s .
s , kl (s) s , kl (s) k- l- xt .)
, F,
. . .. www.iet.ru
Cov(xt) = Q
xt xtT t = 1, 2, , n .
F ARX
:
yt = a1 yt 1 + a2 yt 2 + + ap yt p + ztT + t ,
zt = (zt1, zt2, , ztM)T , = (1 , 2 , , M)T.

F ,
xt = (yt 1, yt 2 , , y t p, z1 , z2 , , z M)T ,
= (a1 , a2 , , ap , 1 , 2 , , M)T .
:
zt (M-) ;
1 n
plim
z t ztT = QZ ,
n n
t =1
QZ ;
t ~ i.i.d., E( t) = 0, D( t) = 2 > 0, E t m < m = 1, 2, ;

Cov(zt m , t) = 0 m = 1, 2, , M ;
Cov(yt j , t) = 0 j = 1, 2, , p ;
a(z) = 1 a1 z a2 z2 ap zp = 0
.
(. [Green (1993)]) F, n
1/2
n ( ) N(0, 2Q 1).
n
, a(z) = 0,
ARX. ,
(..
t )
(long-run) yt , zt1, zt2, , ztM ,
.
4.2.
ARX ,
(
ADL )
yt = 0 + a1 yt 1 + a2 yt 2 + + ap yt p +
+ (10 x1, t + 11 x1, t 1 + + 1r x1, t r ) +
+ +
+ (s 0 xs, t + s 1 xs, t 1 + + s r xs, t r) + t .
ADL(p,r; s), p
yt , r x1, t , x2, t ,
, xs, t , yt , s
. ,
. . .. www.iet.ru
ij ,
xi, t .
ADL(p,r; s)
a(L) yt = + b1(L) x1, t + + bs(L) xs, t + t ,
a(L) = 1 a1 L a2 L 2 ap L p,
bi(L) = i 0 + i 1L + + i r L r , i = 1, , s .
, yt
1
1
1
1
yt =
b1 ( L) x1, t + K +
bs ( L) xs , t +
+
t ,
a( L)
a( L)
a( L)
a( L)
yt =
1
1
+ c1 ( L) x1, t + K + c s ( L) x s , t +
t ,
a ( L)
a( L)
ci ( L ) =
bi ( L)
.
a ( L)
,
yt
L = 1, t 0.

1
yt =
+ c1 (1) x1, t + K + cs (1) xs , t ;
a(1)
, t :
1
y=
+ c1 (1) x1 + K + cs (1) xs .
a(1)
1(1) , , s(1)
(long-run multipliers).
ADL(1, 1; 1),
(1 1L) yt = + 0 xt + 1 xt 1 + t .
1 < 1
1
1
yt =
+
( x + x + ) ,
(1 1L ) (1 1L ) 0 t 1 t 1 t
..
yt = (1 + 1 + 12 + ) + (1 + 1 L + 12 L2 + )(0 xt + 1 xt 1 + t),
:
yt xt = 0 ,
yt +1 xt = yt xt 1 = 1 + 1 0 ,
yt +2 xt = yt xt 2 = 1 1 + 120 ,
yt +3 xt = yt xt 3 = 121 + 130 ,
. . .. www.iet.ru
.. ,
()
xt
yt . ,
:
yt xt + yt xt 1 + yt xt 2 + yt xt 3 + =
= 0 (1 + 1 + 12 + ) + 1(1 + 1 + 12 + ) =
= (1 1L) 1(0 + 1).
, ,
ADL(1, 1; 1).
,
yt
xt .
, ADL
, .
,
(
. . 4.1, F):
t- N(0,1) .
F F- q
, qF
2 q .

t qF
() ( t-,
F-).

.

t
, .
,
. . .. www.iet.ru
d d = 2,

( t ).
ADL(3, 2; 1)
(1 0.5L 0.1L 2 0.05L 3) yt = 0.7 + (0.2 + 0.1 L + 0.05L 2) xt + t .
y x L =
1 t 0:
(1 0.5 0.1 0.05) y = 0.7 + (0.2 + 0.1 + 0.05) x ,
.. 0.35 y = 0.7 + 0.35 x ,
y=2+x.
xt
= 0.7xt1 + xt , xt ~ i.i.d. N(0, 1), yt ,
ADL(3, 2; 1), t ~ i.i.d. N(0, 1), t

xt .
: x1 = 0, y1 = y2 = y3 = 0.
6
4
2
0
-2
-4
10
20
30
40
50
60
70
80
90
100
, ,
. yt = + xt + t
; yt = 1.789 + 0.577xt +
et , et . :
4
2
0
-2
-4
-6
10
20
30
40
50
60
70
80
90
100
DELTA
,

ACF
PAC
F
C
PAC Q-Stat Prob
. . .. www.iet.ru
0
4
.696 9.981 .000
.
.
0
7
|****
|*
.536
.099 9.868 .000
.
*|
9
|***
.
.364 0.081 3.801 .000
.
. |
9
|**
.
.227 0.056 9.279 .000
.
. |
1
|*
.
.130 0.015 01.10 .000
. |
. |
1
.
.
.057 0.020 01.46 .000
, P 0.0000. , ,
.
.
xt
ACF
PACF
AC
PAC
Q-Stat
Prob
|*****
|*****
.696
. |*****
. |*****
0.686
0.686
48.468
. |***
*| .
0.429
-0.079
67.594
. |*
*| .
0.193
-0.132
71.527
.|.
*| .
0.024
-0.066
71.591
*| .
|.
-0.058
0.003
71.958
*| .
*| .
-0.140
-0.107
74.090
xt AR(1).
yt
ACF PACF
AC
PAC Q-Stat
Prob
.
. |******
0.767
0.767
60.58 0.000
. |*
0.629
0.100
101.75 0.000
.|.
0.494 -0.042
127.37 0.000
.|.
0.399
0.019
144.32 0.000
.|.
0.318 -0.003
155.21 0.000
.|.
0.257
162.38 0.000
0.000
0.000
0.000
0.000
0.000
0.000
|******
|*****
|****
|***
|**
0.004
|**
AR(1).
ADL
, :
yt = + a1 yt 1 + 0 xt + 1 x t 1 + t .
. . .. www.iet.ru
10
ADL(1, 1; 1), :
Dependent Variable: Y
Variable Coefficient Std. Error t-Statistic Prob.
C
Y(-1)
X
X(-1)
0.558588
0.695204
0.208971
0.161690
0.157276
0.066095
0.126135
0.132352
3.55163
10.51828
1.65673
1.22166
0.0006
0.0000
0.1009
0.2249
(P-
AR(1) 0.164),
t (P- Jarque Bera =
0.267), (P- = 0.159),
, ,
,
t- F-.
H0: 0 = 1 = 0 F
P- 0.0032;
2(2) P- 0.0022.
. xt1 ,
P-,
xt , :
Variable
Coefficient Std. Error
C
Y(-1)
X
R-squared
Adjusted R-squared
S.E. of regression
Sum squared resid
Log likelihood
Durbin-Watson stat
0.517868 0.154098
0.719738 0.063131
0.310343 0.095241
0.637523
0.629971
1.039901
103.8138
-142.8251
2.256404
t-Statistic
Prob.
3.360648
11.40064
3.258511
0.0011
0.0000
0.0015
Mean dependent var

S.D. dependent var
Akaike info criterion
Schwarz criterion
F-statistic
Prob(F-statistic)
1.844751
1.709520
2.945962
3.024602
84.42207
0.000000
,
.
x t 1, x t ,
Variable
C
Y(-1)
X(-1)
R-squared
Adjusted R-squared
S.E. of regression
Sum squared resid
Log likelihood
Durbin-Watson stat
0.567821 0.158600
0.692523 0.066673
0.305939 0.100582
0.632818
0.625169
1.046627
105.1611
-143.4634
2.221594
t-Statistic
Prob.
3.580215
10.38690
3.041698
0.0005
0.0000
0.0030
Mean dependent var

S.D. dependent var
Schwarz criterion
F-statistic
Prob(F-statistic)
1.844751
1.709520
2.958857
3.037497
82.72550
0.000000
. . .. www.iet.ru
11

xt1 , . ,
.
,
(1 0.720 L) yt = 0.518 + 0.310 xt + et .
L = 1
et 0, : 0.28 yt = 0.518 + 0.310 xt ,

y = 1.839 + 1.107x .
, , . ,
, ADL(3, 2; 1).
, :
Variable
Coefficient Std. Error t-Statistic
C
Y(-1)
Y(-2)
Y(-3)
X
X(-1)
X(-2)
0.539617
0.590293
0.153936
-0.031297
0.205570
0.191959
-0.024779
R-squared
Adjusted R-squared
S.E. of regression
Sum squared resid
Log likelihood
Durbin-Watson stat
0.643818
0.620072
1.051648
99.53667
-138.8891
1.999213
0.174057
0.105583
0.120517
0.099814
0.129483
0.153608
0.138389
3.100239
5.590795
1.277303
-0.313555
1.587626
1.249666
-0.179053
Mean dependent var

S.D. dependent var
Schwarz criterion
F-statistic
Prob(F-statistic)
Prob.
0.0026
0.0000
0.2048
0.7546
0.1159
0.2147
0.8583
1.882787
1.706160
3.008022
3.193826
27.11327
0.000000
y x
, , :
y = 1.882 +1.300 x ,
, ,
. ,
.
4.3.

(VAR vector autoregression).
y1t y2t
:
y1t = 1 + 11.1 y1, t 1 + 12.1 y2, t 1 + 1t ,
y2t = 2 + 21.1 y1, t 1 + 22.1 y2, t 1 + 2t ,
.. , , y1t
y1,t1 , y2,t1 y2t .
1t 2t :
Cov(jt , ls) = 0 t s j , l = 1, 2 ;
Cov(jt , yl, t r) = 0 r 1 j , l = 1, 2 .
. . .. www.iet.ru
12
, 1t 2t
.

y1t y2t .
, ,
. p ,
VAR(p).
k y1t, y2t, , ykt .
p ,

y1t = 1 + 11.1 y1, t 1 + 11.2 y1, t 2 + + 11.p y1, t p +
+ 12.1 y2, t 1 + 12.2 y2, t 2 + + 12.p y2, t p +
++
+ 1k. 1 yk, t 1 + 1k. 2 yk, t 2 + + 1k.p yk, t p + 1t ,
y2t = 2 + 21.1 y1, t 1 + 21.2 y1, t 2 + + 21.p y1, t p +

+ 22.1 y2, t 1 + 22.2 y2, t 2 + + 22.p y2, t p +
++
+ 2k.1 yk, t 1 + 2k.2 yk, t 2 + + 2k.p yk, t p + 2t ,
...
yk t = k + k1.1 y1, t 1 + k1.2 y1, t 2 + + k1.p y1, t p +
+ k2.1 y2, t 1 + k2.2 y2, t 2 + + k2.p y2, t p +
++
+ kk. 1 yk, t 1 + kk. 2 yk, t 2 + + kk.p yk, t p + kt ,
ij.r - yj, t r yi t .
1t , 2t , , kt ,
Cov(jt , ls) = 0 t s j , l = 1, , k ;
Cov(jt , yl, t r) = 0 r 1 j , l = 1, , k ;
Cov(jt , lt) .
C 1t , 2t , , kt t = (1t , 2t , ,
T
kt) ,
y1t, y2t, , ykt .
()
Yt 1 = ( y1, t 1 , y1, t 2 , , y1, t p , , yk, t 1 , yk, t 2 , , yk, t p) .
VAR(1) (k = 2, p = 1):
y1t = 0.6 + 0.7 y1, t 1 + 0.2 y2, t 1 + 1t ,
y2t = 0.4 + 0.2 y1, t 1 + 0.7 y2, t 1 + 2t .
y1t,
y2t t = 2, 3, , 100.
. . .. www.iet.ru
13
y11 = y21 = 0; 1t 2t
, N(0, 0.12).
6
5
4
3
2
1
0
10
20
30
40
50
60
Y1
70
80
90
100
Y2
(y1t y2t).
0.8
0.6
0.4
0.2
0.0
10
20
30
40
50
60
70
80
90
100
Y1-Y2
, :
. ,
y1t y2t
0.4 ( y1t y2t 0.403).
y1t , y2t VAR .
, VAR
, , .
VAR(p) k
yt = + 1 yt 1 + 2 yt 2 + + p yt p + t .
yt = (y1t , y2t , , ykt )T , = ( 1, 2, , k )T , t = (1t , 2t , , kt)T ,

r = ( ij.r ) k k y1, t r , y2, t r , , yk, t r
k .

yt 1 yt 1 2 yt 2 p yt p = + t ,
(Ik 1 L 2 L2 p Lp) yt = + t ,
A(L) yt = + t ,
A(L) = Ik 1 L 2 L2 p Lp .
. . .. www.iet.ru
14
VAR :
k
det(Ik z 1 z2 2 zp p) = 0 (.. det A(z) = 0)
(..
k ).
,
, y 1 , y2 , , y p
.
L = 1 t = 0 .
A(1) yt = ,

yt = A 1(1) .
VAR(1)
y1t = 0.6 + 0.7 y1, t 1 + 0.2 y2, t 1 + 1t ,
y2t = 0.4 + 0.2 y1, t 1 + 0.7 y2, t 1 + 2t .

yt = + 1 yt 1 + t ,
y

0.6
0.7 0.2
, t = 1 t ,
yt = 1 t , = , 1 =
0.4
0.2 0.7
y2 t
2t
A(L) yt = + t ,
1 0 0.7 L 0.2 L 1 0.7 L 0.2 L

,
=

A( L) = I 2 1 L =
0 1 0.2 L 0.7 L 0.2 L 1 0.7 L

6 4
0.3 0.2
.
, A 1 (1) =
A(1) =
0.3
4 6
0.2
det A(z) = 0
1 0.7 z 0.2 z
= 0 ,
det A( z ) =
0.2 z 1 0.7 z
.. (1 0.7 z)2 (0.2 z)2 = 0 , (1 0.9 z)(1 0.5 z) = 0 . z = 1/0.9
z = 1/0.5 1, .. .
()
6 4 0.6 5.2
= .
yt = A 1 (1) =
4 6 0.4 4.8
,
y1t = 5.2, y2t = 4.8,
. . .. www.iet.ru
15
y1t y2t
y1 y2 = 0.4 .
, , , y1t
5.2, y2t
4.8; (y1t y2t) 0.4 .
VAR(1)
.
, , ,
VAR, VAR,
,
( ),
( ).
, - ,
,
. ,
.
, VAR .
VAR
A(L) yt = + B(L) xt + t ,
A(L) = I 1 L 2 L2 p Lp
B(L) .
detA(z) = 0
, ,

yt = A 1(L) + (L) xt + A 1(L) t ,
(L) = A 1(L)B(L) (transfer function).
(L) ;
.
(, , long-run )
, L = 1
t 0. :
yt = A 1(1) + (1) xt .
(1) . ( i ,
j)- cij(1) xjt yit
(. .
4.2).
VAR(1)
VAR
y1t = 0.6 + 0.7 y1, t 1 + 0.2 y2, t 1 + 0.1 x1, t 1 + 0.2 x2, t + 1t ,
y2t = 0.4 + 0.2 y1, t 1 + 0.7 y2, t 1 + 0.2 x1, t + 0.4 x2, t 1 + 2t .
A(L) , ,
. . .. www.iet.ru
16
0.1 L 0.2
0 0.2 0.1 0
,
L =
+
B( L) = B0 + B1 L =
0.2 0.4 L
0.2 0 0 0.4

0.1 0.2
.
B(1) = B0 + B1 =
0.2 0.4

6 4 0.1 0.2 1.4 2.8
,
=
C (1) = A1 (1) B(1) =

4 6 0.2 0.4 1.6 3.2

y1 5.2 1.4 2.8 x1
= +
,
y2 4.8 1.6 3.2 x2
..
y1 = 5.2 + 1.4 x1 + 2.8 x2 ,
y2 = 4.8 + 1.6 x1 + 3.2 x2 .

, x1, t x2, t AR(1) ,
x1, t = 0.7 x1, t 1 + 1t , x2, t = 0.5 x2, t 1 + 2t ; 1t 2t ~ i.i.d. N(0, 1).

x11 = x21 = 0, y11 = y21 = 0 ( 1)
x11 = x21 = 0, y11 = 5.2, y21 = 4.8 ( 2).
:
Variant 1
Variant 2
7
6
6
5
4
4
3
2
2
0
1
10
20
30
40
50
Y1
60
70
80
90
100
10
Y2
20
30
40
50
Y1
60
70
80
90
100
Y2
, -
, ,
.
.
VAR(1) :
y1t = 0.8 y1, t 1 + 0.2 y2, t 1 + 1t ,
y2t = 0.2 y1, t 1 + 0.8 y2, t 1 + 2t .

y1t 0.8 L 0.2 L y1t 1t
=
+ ,
y2t 0.2 L 0.8L y2t 2t
. . .. www.iet.ru
17

1 0.8L 0.2 L
.
A( L) =
0.2 L 1 0.8L

0.2 0.2
,
A(1) =
0.2 0.2
, A 1(1) .
det A(z) = 0 (1 0.8 z)2 (0.2 z)2 = 0, ..
(1 z)(1 0.6 z) = 0.
(1/0.6) 1. , 1,
. ?
.
2
0
-2
-4
-6
-8
10
20
30
40
50
60
Y1
70
80
90
100
Y2
. ,
-
y11 =
y21 = 0.
,
: y11 = y21 = 5.
8
6
4
2
0
-2
10
20
30
40
50
Y1
60
70
80
90
100
Y2
- ,
VAR .
4.4.
. . .. www.iet.ru
18
, ,
ADL(1,1;1)
yt = + a1 yt 1 + 0 xt + 1 x t 1 + t .
, ,
.

= (a1 , 0 ,1) .
,
ADL(1,1;1)
. 9 .
(1) (a1 = 1 = 0): yt = + 0 xt + t .
yt xt ;
yt 1 x t 1 yt .
,
, , ,
t
.
(2) (0 = 1 = 0): yt = + a1 yt 1 + t .
yt yt 1 ; xt
t (t 1) yt .

- ,
,
yt .
(3) (a1 = 0 = 0):
yt = + 1 x t 1 + t .
,
y x
. , ,
1 .
, ,
. , -
y .
(4) (a1 = 1, 1 = 0)
yt = + 0 xt + t ,
( = 1 L , yt = yt yt 1 , xt = xt xt 1),
, , (
).
.
. ,

.
(5) (a1 = 0)
yt = + 0 xt + 1 x t 1 + t
y .
, ,
xt x t 1 .
(6) (1 = 0)
. . .. www.iet.ru
19
yt = + a1 yt 1 + 0 xt + t
x .
, , .
y*t = + xt y ,
yt = yt yt 1
yt yt 1 = (1 )( y*t yt 1) + t , 0 1,
..
yt = (1 ) y*t + yt 1 + t ,
, t , yt
y*t
y .
(, yt , xt .)
yt = yt 1 + (1 )( + xt yt 1) + t = (1 ) + yt 1 + (1 ) xt + t ,
yt = + a1 yt 1 + 0 xt + t ,
= (1 ) , a1 = , 0 = (1 ) .

, x t 1 a1 ,
a1 .
(7) , (0 = 0):
yt = + a1 yt 1 + 1 x t 1 + t .
, ,
xt = x t 1 + ut .
xt ADL(1,1;1) :
yt = + a1 yt 1 + 0 xt + 1 x t 1 + t
= + a1 yt 1 + (0 + 1) x t 1 + (t + 0 ut),
yt = + a1 yt 1 + *1 x t 1 + *t .
( )
0 1 , . ..
,
( ADL(1,1;1)).
(8) (1 = a10):
yt = + a1 yt 1 + 0 xt a10 x t 1 + t .

yt a1 yt 1 = (1 a1) + 0 (xt a1 x t 1) + t .

,

yt = + 0 xt + ut , ut = a1 ut 1 + t , a1 < 1.
(9) ( a1 < 1, 0 +1 = b(1 a1), b 0):
yt = + 0 xt (1 a1)( yt 1 b x t 1) + t ,
yt = 0 xt (1 a1)( yt 1 a b x t 1) + t ,
a = (1 a1), b = ( 0 +1) (1 a1).
. . .. www.iet.ru
20

.

y=a+bx
yt xt
yt 1 a b x t 1
.
() ADL(1,1;1)
yt = + a1 yt 1 + 0 xt + 1 x t 1 + t

yt yt 1 = (1 a1) yt 1 + 0 (xt x t 1) + (0 + 1) x t 1 + t .
a1 < 1 ( ),
yt = + 0 xt (1 a1)( yt 1 ((0 + 1)/(1 a1)) x t 1) + t ,
0 + 1 0 .
, a1 < 1 0 + 1 0
.
.
,
(data generating process DGP).
, -
,
(statistical model SM), , ,
. ,

,

. ,
.
SM DGP,
SM . ,
, SM,
. ,
, ..
, ,
,
.
(n = 100) ADL(1,1,1)
yt = 0.5 yt 1 + 0.2 xt + 0.3 x t 1 + t , t ~ i.i.d. N(0, 0.12),
xt = 0.5 xt 1 + t , t ~ i.i.d. N(0, 0.52),
t t .

. . .. www.iet.ru
21
1.5
1.0
0.5
0.0
-0.5
-1.0
-1.5
10
20
30
40
50
X
60
70
80
90
100
ADL(1,1;1)
:
Variable
C
Y(-1)
X
X(-1)
0.014122
0.555208
0.188567
0.258377
R-squared
Adjusted R-squared
S.E. of regression
Sum squared resid
0.913395
0.910660
0.092824
0.818547
0.009556
0.034143
0.018421
0.020673
t-Statistic
Prob.
1.477773
16.26107
10.23666
12.49808
0.1428
0.0000
0.0000
0.0000
Mean dependent var

S.D. dependent var
Schwarz criterion
0.062869
0.310554
-1.876660
-1.771806
, :
Variable
Y(-1)
X
X(-1)
R-squared
Adjusted R-squared
S.E. of regression
Sum squared resid
Log likelihood
0.565569 0.033621
0.190325 0.018495
0.256578 0.020764
0.911404
0.909558
0.093394
0.837363
95.76965
t-Statistic
Prob.
16.82186
10.29043
12.35668
0.0000
0.0000
0.0000
Mean dependent var

S.D. dependent var
Schwarz criterion
Durbin-Watson stat
0.062869
0.310554
-1.874134
-1.795494
2.218619
.
.
.

t (P- = 0.375 AR(1) 0.165 AR(2)
).
Jarque Bera
(P- = 0.689).
(P- = 0.285).
. . .. www.iet.ru
22
,
.
,
SM 8 .
SM1 ( ): yt = + 0 xt + t .
:
Sample: 1 100
Variable
X
R-squared
Adjusted R-squared
S.E. of regression
Sum squared resid
Log likelihood
0.271208 0.053356
0.174472
0.174472
0.280794
7.805700
-14.37805
t-Statistic
Prob.
5.082965
0.0000
Mean dependent var

S.D. dependent var
Schwarz criterion
Durbin-Watson stat
0.062241
0.309046
0.307561
0.333613
0.839862

.
.
t , ..
.
SM2 : yt = + a1 yt 1 + t .
:
C
Y(-1)
0.013941 0.020679 0.674149 0.5018

0.764874 0.065621 11.65594 0.0000

,
. t ,
. AR(1) P-
0.00003, t
. ,
.
SM3 : yt = + 1 x t 1 + t .
:
C
X(-1)
0.049508 0.019722 2.510238 0.0137

0.455497 0.037291 12.21457 0.0000
. . .. www.iet.ru
23
AR(1) P- 0.0002,
t
.
.
SM4 : yt = + 0 xt + t

Dependent Variable: D(Y)
Variable
Coefficient Std. Error t-Statistic Prob.
C
D(X)
-0.001126 0.021384 -0.052674 0.9581

0.040538 0.033362 1.215078 0.2273
Log likelihood
13.74152
Durbin-Watson stat 1.574116
F-statistic
Prob(F-statistic)
1.476415
0.227286
AR(1) P- 0.029,
t
.
.
SM5 :
yt = + 0 xt + 1 x t 1 + t

C
X
X(-1)
0.046032 0.018096 2.543741 0.0126

0.156214 0.035435 4.408526 0.0000
0.414363 0.035435 11.69370 0.0000
AR(1) P- 0.00000,
t
.
.
SM6 :
yt = + a1 yt 1 + 0 xt + t

C
Y(-1)
X
0.007088 0.015431 0.459354 0.6470

0.753212 0.048925 15.39514 0.0000
0.253493 0.028588 8.867013 0.0000
AR(1) P- 0.012,

t .
.
SM7 : yt = + a1 yt 1 + 1 x t 1 + t .

. . .. www.iet.ru
24
Variable
C
Y(-1)
X(-1)
0.020438 0.013757 1.485648 0.1406

0.517457 0.048968 10.56734 0.0000
0.318058 0.028613 11.11579 0.0000
S.E. of regression 0.133909

Sum squared resid 1.721440
Akaike info criterion -1.153476

Schwarz criterion -1.074836
P- AR(1)
0.499, AR(2)
0.538.
t ,
. Jarque Bera
(P- = 0.937).
(P- = 0.348).
,
. ,
.
:
Variable
Y(-1)
X(-1)
0.532005 0.048276 11.02001 0.0000

0.316252 0.028765 10.99445 0.0000
S.E. of regression 0.134740

Sum squared resid 1.761018
Akaike info criterion -1.150947

Schwarz criterion -1.098520
.
,
.
yt =
a1 yt 1 + 0 xt + 1 x t 1 + t , , xt
( 1.874) ( 1.795)
.
:
SM8
yt = + a1 yt 1 + 0 xt a10 x t 1 + t .
()

Y=C(1)+C(2)*Y(-1)+C(3)*X-(C(2)*C(3))*X(-1)
t-Statistic
Prob.
C(1)
C(2)
C(3)
0.014489
0.749747
0.052577
0.702743
10.68267
1.439066
0.4839
0.0000
0.1534
R-squared
Adjusted R-squared
0.592630
0.584144
0.020617
0.070184
0.036535
Mean dependent var

S.D. dependent var
0.062869
0.310554
. . .. www.iet.ru
S.E. of regression
Sum squared resid
Log likelihood
0.200267
3.850250
20.25069

Schwarz criterion
Durbin-Watson stat
25
-0.348499
-0.269859
1.447077
AR(1) P- 0.0002,

t .
.
SM (
).
SM9 :
yt = + 0 xt (1 a1)( yt 1 b x t 1) + t .
( ):
D(Y) =C(1)+C(2)*D(X) + (C(3)-1)*(Y(-1)-C(4)*X(-1))
t-Statistic
Prob.
C(1)
C(2)
C(3)
C(4)
0.014122
0.188567
0.555208
1.004839
1.477773
10.23666
16.26107
12.86299
0.1428
0.0000
0.0000
0.0000
R-squared
Adjusted R-squared
S.E. of regression
Sum squared resid
Log likelihood
0.816395
0.810597
0.092824
0.818547
96.89465
0.009556
0.018421
0.034143
0.078119
Mean dependent var

S.D. dependent var
Schwarz criterion
Durbin-Watson stat
-0.001100
0.213288
-1.876660
-1.771806
2.248395
P- AR(1) 0.130,
AR(2) 0.318;
t . Jarque Bera
(P- = 0.711).
(P- = 0.380).
,
. , ,
; :
D(Y) =C(2)*D(X) + (C(3)-1)*(Y(-1)-C(4)*X(-1))
t-Statistic
Prob.
C(2)
C(3)
C(4)
0.190325
0.565569
1.028710
10.29043
16.82186
12.82279
0.0000
0.0000
0.0000
R-squared
Adjusted R-squared
S.E. of regression
Sum squared resid
0.812174
0.808261
0.093394
0.837363
0.018495
0.033621
0.080225
Mean dependent var

S.D. dependent var
Schwarz criterion
-0.001100
0.213288
-1.874134
-1.795494
. . .. www.iet.ru
Log likelihood
95.76965
Durbin-Watson stat
26
2.218619
..
yt = 0.190 xt 0.434( yt 1 1.029 x t 1) + et .
.
yt , :
yt = 0.566 yt 1 + 0.190 xt + 0.253 x t 1 + et .

DGP: yt = 0.5 yt 1+ 0.2 xt + 0.3 x t 1+ t
:
yt = 0.565 yt 1+ 0.190 xt + 0.257 x t 1+ t .
xt
,

yt
:
yt = 0.5 yt + 0.2 xt + 0.3 x t y = x .
, , SM,
DGP:
yt = 0.565 yt + 0.190 xt + 0.257 xt y = 1.002 x .
, , SM9 (
):
yt = 0.566 yt + 0.190 xt + 0.253 x t y = 1.021 x .
, SM7 ( )
, :
yt = 0.532 yt + 0.316 x t y = 0.675 x .
,
SM,
DGP. , SM
.
SM1 SM6 SM8 ,
SM7 .
, , ,
SM () DGP,
, DGP, ,
.
DGP (1) (8), SM
ADL(1,1;1) :
yt = + a1 yt 1 + 0 xt + 1 x t 1 + t .
(1) (8)
, ADL(1,1;1)
yt = 0.5 yt 1+ 0.2 xt + 0.3 x t 1+ t .
DGP t ~ i.i.d. N(0, 0.12).
DGP1 :
yt = 0.2 xt + t .
. . .. www.iet.ru
27

Variable
t-Statistic
Prob.
C
Y(-1)
X
X(-1)
-0.004647
0.102848
0.186813
0.000201
-0.451175
1.009966
9.238033
0.007101
0.6529
0.3151
0.0000
0.9943
R-squared
Adjusted R-squared
S.E. of regression
Sum squared resid
0.507795
0.492252
0.102190
0.992068
0.010300
0.101833
0.020222
0.028272
Mean dependent var

S.D. dependent var
Schwarz criterion
0.001230
0.143412
-1.684398
-1.579545
P- AR(1) 0.760,
AR(2) 0.951,
t . Jarque Bera
(P- = 0.733).
(P- = 0.770).
,
. ,
.
yt 1 x t 1 F-, F =
0.738, qF = 2.214. F- F , P 0.532. 2 (3) qF
P- 0.529.
. ,
yt = + 0 xt + t , :
Sample: 1 100
Variable
t-Statistic
Prob.
0.190067
9.852604
0.0000
R-squared
Adjusted R-squared
S.E. of regression
Sum squared resid
0.494995
0.494995
0.101522
1.020359
0.019291
Mean dependent var

S.D. dependent var
Schwarz criterion
0.001935
0.142860
-1.727139
-1.701087

.
, .
DGP2 :
yt = 0.5 yt 1 + t .

Variable
t-Statistic
Prob.
C
Y(-1)
X
X(-1)
-0.004519
0.576756
-0.013220
0.021476
-0.438075
6.855173
-0.652774
1.061719
0.6623
0.0000
0.5155
0.2911
R-squared
0.338422
0.010315
0.084134
0.020253
0.020228
Mean dependent var
-0.007891
. . .. www.iet.ru
Adjusted R-squared
S.E. of regression
Sum squared resid
0.317530
0.102290
0.994011
S.D. dependent var

Schwarz criterion
28
0.123820
-1.682441
-1.577588
P- AR(1) 0.600,
AR(2) 0.773,
t . Jarque Bera
(P- = 0.654).
(P- = 0.956).
xt xt 1
F-, F = 0.641, qF = 1.283.
F- F , P- 0.529.
2 (3) qF P- 0.527.
.
, yt = a1 yt 1 + t ,
:
Variable
t-Statistic
Prob.
Y(-1)
0.575922
6.963585
0.0000
R-squared
Adjusted R-squared
S.E. of regression
Sum squared resid
0.328274
0.328274
0.101482
1.009258
0.082705
Mean dependent var

S.D. dependent var
Schwarz criterion
-0.007891
0.123820
-1.727825
-1.701612

.
t .
DGP3 :
yt = 0.3 x t 1 + t .

Variable
t-Statistic
Prob.
C
Y(-1)
X
X(-1)
-0.005202
0.052373
-0.012475
0.315962
-0.504767
0.966363
-0.614213
15.30433
0.6149
0.3363
0.5405
0.0000
R-squared
Adjusted R-squared
S.E. of regression
Sum squared resid
0.736662
0.728346
0.102236
0.992959
0.010305
0.054196
0.020310
0.020645
Mean dependent var

S.D. dependent var
Schwarz criterion
0.004035
0.196154
-1.683501
-1.578647
P- AR(1) 0.614,
AR(2) 0.868,
t . Jarque Bera
(P- = 0.740).
(P- = 0.804).
xt yt 1
F-, F = 0.577, qF = 1.730.
F- F , P- 0.632.
. . .. www.iet.ru
29
2 (3) qF P- 0.630.
.
, yt = 1 x t 1 + t ,
:
Variable
t-Statistic
Prob.
X(-1)
0.315777
16.35987
0.0000
R-squared
Adjusted R-squared
S.E. of regression
Sum squared resid
0.731866
0.731866
0.101572
1.011044
0.019302
Mean dependent var

S.D. dependent var
Schwarz criterion
0.004035
0.196154
-1.726058
-1.699844

.
t .
DGP4 :
yt = + 0 xt + t .

Variable
t-Statistic
Prob.
C
Y(-1)
X
X(-1)
-0.029026
0.959750
0.184064
-0.173461
-1.636049
39.61303
9.206520
-8.528162
0.1051
0.0000
0.0000
0.0000
R-squared
Adjusted R-squared
S.E. of regression
Sum squared resid
Log likelihood
Durbin-Watson stat
0.944972
0.943234
0.101277
0.974412
88.26661
1.745026
0.017741
0.024228
0.019993
0.020340
Mean dependent var

S.D. dependent var
Schwarz criterion
F-statistic
Prob(F-statistic)
-0.599911
0.425076
-1.702356
-1.597502
543.7988
0.000000
yt 1 ,
, DGP a1 = 1.
. , yt ,
,
0.5
0.0
-0.5
-1.0
-1.5
-2.0
10
20
30
40
50
60
70
80
90
100
. . .. www.iet.ru
30
.
H0: a1 = 1 ,
. ,
yt 1,

yt = + 0 xt + 1 x t 1 + t .
()
, xt
1.5
1.0
0.5
0.0
-0.5
-1.0
-1.5
10
20
30
40
50
60
70
80
90
100
, yt

0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
10
20
30
40
50
60
70
80
90
100
DELTA
yt .
:
Variable
t-Statistic
Prob.
C
X
X(-1)
-0.477336
9.186333
-8.916410
0.6342
0.0000
0.0000
-0.004915 0.010297
0.185224 0.020163
-0.179782 0.020163
, ,
, , xt x t 1
. ,

yt = + 0 xt + 1 x t 1 + t
H0: = 0, 0 = 1. F- F - 2(2) qF = 2F ,
P- 0.876. H0 ,
. . .. www.iet.ru
31
, .. yt = 0 xt + t .
:
Variable
t-Statistic
Prob.
D(X)
11.49045
0.0000
0.182495
0.015882
P- AR(1) 0.328,
AR(2) 0.605;
t . Jarque Bera
(P- = 0.673).
(P- = 0.988). ,
, ,
DGP.
DGP5
yt = + 0.2 xt + 0.3 xt 1 + t .

Variable
t-Statistic
Prob.
C
Y(-1)
X
X(-1)
-0.003282
0.000523
0.181735
0.289502
-0.321584
0.008975
9.097166
11.61638
0.7485
0.9929
0.0000
0.0000
R-squared
Adjusted R-squared
S.E. of regression
Sum squared resid
0.804020
0.797831
0.101218
0.973283
0.010206
0.058294
0.019977
0.024922
Mean dependent var

S.D. dependent var
Schwarz criterion
0.010665
0.225113
-1.703515
-1.598661
P- AR(1) 0.972,
AR(2) 0.826;
t . Jarque Bera
(P- = 0.689).
(P- = 0.433).
H0: = 0, a1 = 0 . F- F-
- 2(2) qF = 2F ,
P- 0.950. H0 ,
= 0, a1 = 0, .. yt = + 0 xt + 1 x t 1 + t .
:
Variable
t-Statistic
Prob.
X
X(-1)
9.185972
14.64736
0.0000
0.0000
0.181461
0.289367
0.019754
0.019756
DGP6
yt = + 0.5 yt 1 + 0.2 xt + t .
:
Variable
t-Statistic
Prob.
. . .. www.iet.ru
C
Y(-1)
X
X(-1)
-0.005099
0.592041
0.188766
0.000973
0.010263
0.071296
0.020280
0.025019
-0.496869
8.304016
9.308077
0.038898
32
0.6204
0.0000
0.0000
0.9691
P- AR(1) 0.904,
AR(2) 0.723;
t . Jarque Bera
(P- = 0.691).
(P- = 0.533).
H0: = 0, 1 = 0. F- F-
- 2(2) qF = 2F ,
P- 0.884. H0 ,
= 0, 1 = 0, .. yt = + a1 yt 1 + 0 xt + t .
:
Variable
t-Statistic
Prob.
Y(-1)
X
0.592666
0.188468
10.45233
9.814814
0.0000
0.0000
R-squared
Adjusted R-squared
S.E. of regression
Sum squared resid
0.690477
0.687286
0.100924
0.988000
0.056702
0.019202
Mean dependent var

S.D. dependent var
Schwarz criterion
0.004719
0.180476
-1.728911
-1.676484
DGP7
yt = 0.5 yt 1 + 0.3 xt 1 + t .

Variable
Coefficien Std. Error
t
t-Statistic
Prob.
C
Y(-1)
X
X(-1)
-0.573100
13.08387
-0.507807
15.65291
0.5679
0.0000
0.6128
0.0000
-0.005886
0.559497
-0.010318
0.316645
0.010271
0.042762
0.020320
0.020229
P- AR(1) 0.701,
AR(2) 0.827;
t . Jarque Bera
(P- = 0.740).
(P- = 0.586).
H0: = 0, 0 = 0. F- F-
- 2(2) qF = 2F ,
P- 0.734. H0 ,
= 0, 0 = 0, .. yt = a1 yt 1 + 1 xt 1 + t .
:
Variable
t-Statistic
Prob.
Y(-1)
13.41452
0.0000
0.561389
0.041849
. . .. www.iet.ru
X(-1)
0.313207
0.019269
16.25422
0.0000
Variable
t-Statistic
Prob.
C
Y(-1)
X
X(-1)
-0.438075
6.855173
9.222532
-3.693770
0.6623
0.0000
0.0000
0.0004
33
DGP8

yt = 0.5 yt 1 + 0.2 xt 0.1 x t 1 + t .
:
-0.004519
0.576756
0.186780
-0.093875
0.010315
0.084134
0.020253
0.025414
P- AR(1) 0.600,
AR(2) 0.773;
t . Jarque Bera
(P- = 0.654).
(P- = 0.682).
:
Variable
t
t-Statistic
Prob.
Y(-1)
X
X(-1)
6.908866
9.251413
-3.741827
0.0000
0.0000
0.0003
0.578309 0.083705
0.186426 0.020151
-0.094531 0.025263
, yt 1 xt 0.108,
.. x t
1 . , H0: 1 = a10 .
,
.
:
H0: 1 = a10 ; H0: 0 = (1 /a1) ; H0: a 1 = (1 / 0) .
P- 2(1)- 0.515, 0.514
0.506, H0 :
. ,
1 = a10 , .. yt = a1 yt 1 + 0 xt
a10 xt 1 + t .
Y =C(1)*Y(-1)+C(2)*X-(C(1)*C(2))*X(-1)
C(1)
C(2)
t-Statistic
Prob.
0.575812
0.182370
6.906747
9.543254
0.0000
0.0000
0.083369
0.019110

yt = 0.576 yt 1 + 0.182 xt 0.105 xt 1 + t .
. . .. www.iet.ru
34
, , DGP
,
(
SM ) ,
DGP. ,
, ,
, DGP,
.

, ..
,
, , ,

.
,
,
.

. , , (
8)
ADL(1,1;1).
yt = + a1 yt 1 + 0 xt + 1 xt 1 + t
, ,
1 = a10 .
, 0 0 ADL(1,1;1)
(1 a1L)yt = 0 1 + 1 L xt + t ,

0
a(L) yt = b(L) xt + t ,
a(L) = 1 a1L , b(L) = 0 1 + 1 L .

0
( = 0.)
a1 =
1
,
0
(1 a1L)yt = 0 (1 a1L) xt + t ,
a(L) b(L) (1 a1L).
,
yt = 0 xt + ut ,
ut =
t
1 a1 L
(1 a1L) ut = t ut = a1 ut 1 + t .
. . .. www.iet.ru
35
,
, COMFAC (common factors).
1 =
a10 . COMFAC
.
yt = 0 xt + ut
, , .

a(L) yt = b(L) xt + t
; , t
,
.
, a(L) b(L) .
,
H0: a1 = 0 ( ). ,

.
.
H0: t , ,
HA: t ~ AR(k) c k p , ..
t =1 t 1 + + p t p + t ,
t ~ i.i.d. j 0. , ,

yt = a1 yt 1 + 0 xt + 1 xt 1 + 1 t 1 + + p t p + t ,
, , H0: 12 = = p2 = 0 HA:
2
1 + + p2 0 . , LM
.
, . 3.1
. (., , [Kwan, Sim
(1996)].)

H1: 1 = a10 HA: 1 a10 . ,
,
, : 1 = a10 , a1 = 1 0
0 = 1 a1 , .
,
.
, , H1
, , , H2: a1 = 0
.
, , =
0.05.
? :
P{ H1, H2}
P{ H1} + P{ H2} =
. . .. www.iet.ru
36
= + = 2 .
, = 0.025,
0.05.
, ,
.
. . .. www.iet.ru
5.

, .
.

(GNP gross national product) 1947 .
1961 . (
60 , ., ).
600
500
400
300
200
48
50
52
54
56
58
60
GNP
. (NONDURABLE)

1974 . 1985 . (
48 , , ):
66000
64000
62000
60000
58000
56000
54000
52000
50000
74 75 76 77 78 79 80 81 82 83 84 85
NONDURABLE
.

, ,
. GNP
Autocorrelation
. |*******
. |*******
. |******
. |******
. |******
. |*****
Partial Correlation
. |*******
.|.
.|.
.|.
.|.
.|.
1
2
3
4
5
6
AC
PAC
Q-Stat
Prob
0.946
0.893
0.840
0.791
0.743
0.696
0.946
-0.021
-0.024
0.013
-0.021
-0.022
56.419
107.52
153.55
195.14
232.52
265.90
0.000
0.000
0.000
0.000
0.000
0.000
. . .. www.iet.ru
. |*****
. |*****
. |****
. |****
. |***
. |***
. |***
. |**
. |**
. |**
. |*
. |*
. |*
. |*
.|.
.|.
.|.
*| .
.|.
.|.
.|.
.|.
*| .
.|.
.|.
.|.
.|.
.|.
.|.
.|.
.|.
.|.
.|.
.|.
.|.
.|.
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
0.648
0.599
0.550
0.498
0.442
0.388
0.337
0.291
0.253
0.218
0.182
0.143
0.106
0.069
0.032
-0.003
-0.037
-0.066
-0.030
-0.044
-0.033
-0.052
-0.073
-0.034
-0.002
0.007
0.041
-0.002
-0.034
-0.044
-0.021
-0.039
-0.028
-0.030
-0.021
-0.005
295.41
321.09
343.13
361.57
376.44
388.08
397.06
403.91
409.21
413.22
416.08
417.90
418.92
419.36
419.45
419.45
419.59
420.04
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
NONDURABLE
ACF
. |*******
. |******
. |******
. |*****
. |*****
. |****
. |****
. |***
. |***
. |**
. |**
. |*
. |*
. |*
. |*
.|.
.|.
.|.
.|.
*| .
PACF
. |*******
.|.
.|.
.|.
.|.
.|.
*| .
.|.
*| .
.|.
.|.
.|.
.|.
.|.
.|.
.|.
.|.
*| .
.|.
*| .
AC
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
PAC Q-Stat Prob
0.917 0.917 42.921

0.843 0.014 79.976
0.774 -0.004 111.92
0.704 -0.040 138.99
0.643 0.013 162.09
0.581 -0.041 181.36
0.510 -0.090 196.57
0.439 -0.050 208.14
0.366 -0.066 216.39
0.300 -0.012 222.07
0.246 0.026 225.99
0.196 -0.006 228.56
0.150 -0.013 230.11
0.106 -0.025 230.91
0.072 0.036 231.28
0.041 -0.016 231.41
0.019 0.028 231.43
-0.007 -0.063 231.44
-0.032 -0.018 231.52
-0.062 -0.073 231.85
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
,
ACF PACF
AR(1). ,

Xt = + t + a1Xt 1 + ut .
( ut , t , ut
.)
. . .. www.iet.ru
.1
GNP:
Dependent Variable: GNP
Sample(adjusted): 1947:2 1961:4
Variable
t-Statistic
Prob.
C
T
AR(1)
216.0630
5.269279
0.846976
19.11661
18.70170
11.64665
0.0000
0.0000
0.0000
Inverted AR Roots
11.30237
0.281754
0.072723
.85
: P-
AR(1) 0.0000.
, :
Dependent Variable: GNP
Variable
t-Statistic
Prob.
217.7399
5.221538
1.380274
-0.630066
5.054473
0.140436
0.109452
0.109453
43.07865
37.18089
12.61078
-5.756490
0.0000
0.0000
0.0000
0.0000
.69 -.39i
.69+.39i
t
C
T
AR(1)
AR(2)
Inverted AR Roots
, , 0.7926,

Xt0 = Xt t .
K GNP .
,
Xt = + t + ut :
Variable
t-Statistic
Prob.
C
T
218.4825
5.181995
82.75373
68.84144
0.0000
0.0000
Durbin-Watson stat
0.316211
2.640153
0.075274
Prob(F-statistic)
0.000000
(C) t (T)

(Xt t ) = a1(Xt 1 ( t 1)) + a2(Xt 2 ( t 2)) + ut
(a2 = 0 ).
. AR(1) a1 , AR(2)
a2 .
. . .. www.iet.ru
, ,
,
Autocorrelation
. |******
. |****
. |*.
.*| .
**| .
***| .
***| .
***| .
**| .
**| .
**| .
**| .
Partial Correlation
. |******
****| .
**| .
.|.
.|.
.*| .
.*| .
.*| .
.|.
.*| .
***| .
. |*.
1
2
3
4
5
6
7
8
9
10
11
12
AC
PAC
Q-Stat
Prob
0.836
0.531
0.183
-0.100
-0.272
-0.339
-0.350
-0.332
-0.281
-0.234
-0.234
-0.226
0.836
-0.554
-0.210
0.044
-0.004
-0.082
-0.169
-0.072
0.058
-0.177
-0.321
0.103
44.028
62.115
64.294
64.960
69.949
77.846
86.446
94.332
100.07
104.16
108.32
112.26
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
AR(2).
AR(2) ()
Xt_detrended = Xt 218.4825 5.181995 t :
Variable
Coeff.
Std. Error
AR(1)
AR(2)
1.379966 0.107605
-0.630426 0.107605
t-Statistic
Prob.
12.82435
-5.858722
0.0000
0.0000
,
Xt 218.4825 5.181995 t =
= 1.379966 (Xt1 218.4825 5.181995(t1))
0.630426 (Xt2 218.4825 5.181995(t2)),
Xt = [(1 1.379966 + 0.630426)218.4825

+ 1.3799665.181995 0.6304265.1819952]
+ (1 1.379966 +0.630426) 5.181995 t
+ 1.379966 Xt1 0.630426 Xt2 + et
= 55.338375 +1.297882 t + 1.379966 Xt1 0.630426 Xt2 + et .
,
Xt = + t + a1Xt1 + a2Xt2 + ut
Xt 217.7399 5.221538 t =
= 1.380274 (Xt1 217.7399 5.221538(t1))
0.630066 (Xt2 217.7399 5.221538(t2)),
Xt = [(1 1.380274 + 0.630066)217.7399

+ 1.3802745.221538 0.6300665.2215382]
+ (1 1.380274 +0.630066) 5.221538 t
+ 1.380274 Xt 1 0.630066 Xt 2 + et
= 55.017011 + 1.304298 t + 1.380274 Xt1 0.630066 Xt2 + et ,
. . .. www.iet.ru
,
, .
,
,
Xt = + t + a1Xt1 + a2Xt2 + ut .
,
.
NONDURABLE.
( Xt t )
Autocorrelation
. |******
. |*****
. |***
. |**
. |*
*| .
**| .
***| .
****| .
Partial Correlation
. |******
.|.
**| .
**| .
.|.
*| .
**| .
*| .
**| .
1
2
3
4
5
6
7
8
9
AC
PAC
Q-Stat
Prob
0.793
0.632
0.432
0.219
0.090
-0.067
-0.242
-0.362
-0.510
0.793
0.011
-0.195
-0.193
0.062
-0.152
-0.277
-0.084
-0.211
32.083
52.942
62.887
65.515
65.965
66.218
69.647
77.505
93.500
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
1,
Xt = + t + a1Xt1 + ut . :
Variable
t-Statistic
Prob.
C
T
AR(1)
47962.75
315.1909
0.884803
16.75451
4.122961
10.94727
0.0000
0.0002
0.0000
2862.678
76.44770
0.080824
P-
0.05 M 1 20.
P-, 0.05, AR(1)
, AR(2), AR(3) .. ,
P- Jarque Bera 0.648 ,

Xt 47962.75 315.1909 t =
= 0.884803 (Xt1 47962.75 315.1909 (t1)) + et .
, , .
0.884803
Xt1 ,
0.080824
,
, 0.884803 20.080824
, 1.
.
, , 0.884803
Xt1
(..
AR(1) ),
. . .. www.iet.ru
,
.
,
,
.
80- 20 , .

,
,
.
,
( )
ARMA ,
.
5.1. ARMA
AR(1)
Xt = a1Xt1 + t .
,
1< a1 < 1. Xt
? a1 = 0.5, a1 =
0.7, a1 = 0.9, a1 = 1, a1 = 1.05, a1 = 1.1.
4
-2
-2
-4
-4
5
10
15
20
25
30
35
40
45
50
10
15
20
a1= 0.5
25
30
35
40
45
50
35
40
45
50
a1= 0.7
2
0
-2
0
-4
-2
-6
-4
-8
-6
-10
5
10
15
20
25
30
a1= 0.9
35
40
45
50
10
15
20
25
30
a1= 1
. . .. www.iet.ru
5
0
-5
-10
-15
-20
50
100
150
200
250
300
350
400
450
500
a1= 1
10
50
0
-50
-10
-100
-150
-20
-200
-30
-250
-40
5
10
15
20
25
30
35
40
45
50
a1= 1.05
-300
5
10
15
20
25
30
35
40
45
50
a1= 1.1
X1 X1 = 0

1 , , T ,
, :
3
2
1
0
-1
-2
-3
5
10
15
20
25
30
35
40
45
50
NOISE
.
, c
a1 a1 = 0.5 a1 = 1.1. ,
Xt .
. . .. www.iet.ru

Noise
)
25
0.046
AR(1)
0.5
a1 =
14
0.097
AR(1)
0.7
a1 =
0.191
AR(1)
0.9
a1 =
0.649
AR(1)
1.0
a1 =
3.582
AR(1)
1.05
a1 =
13.511
AR(1)
1.1
a1 =
59.621
a1 a1= 0 ( ) a1= 1
,
,
. a1= 1, 500 ,

Xt = Xt1 + t ,
, , t = 1 Xt
= x1 ( x1 = 0), x1
( x1 ) ,
( ),
.
. . .. www.iet.ru
,
, ,
.
, AR(1) a1= 1:
Xt = Xt 1 + t
random walk).
,

Xt = a1Xt1 + t
a1 = 1.05 a1 = 1.1.
(explosive) AR(1) a1 > 0:
.

- .
AR(1)

Xt Xt1 = a1Xt1 Xt1 + t = (a1 1)Xt1 + t ,
Xt = Xt1 + t ,
Xt = Xt Xt1 , = a1 1.
a1 = 1 = a1 1= 0, Xt Xt
, Xt
() Xt1 = xt1 xt1 0. ,
() Xt1 = xt1 ,
Xt = Xt + Xt1 xt1 .
t (
, ),
. . .. www.iet.ru
10
Xt = xt ,
xt1 . .
a1 > 1 = a1 1 > 0, Xt
() Xt1 = xt1 , E( XtXt1 = xt1) =
xt1 , , xt1. , xt1 > 0,
Xt = xt xt1 ,
xt1 < 0, Xt = xt
xt1 .
,
AR(1) a1 = 1.05 a1 = 1.1.
, 0 < a1 < 1 = a1 1 < 0,
Xt () Xt1 = xt1 ,
E( XtXt1 = xt1) = xt1 , , xt1. ,
xt1 > 0, Xt = xt
xt1 , xt1 < 0, Xt =
xt xt1 .
,
E(Xt) = ( , = 0),
.
a1 > 0,
.
Xt = a1Xt1 + t a1 = 1 a1 =
1.1.
6
30
20
10
2
0
0
-10
-2
-20
-4
-30
5
10
15
20
25
30
35
40
a1= -1
45
50
10
15
20
25
30
35
40
45
50
a1= - 1.1

Xt = Xt1 + t , t = 1, , T ,
X0 = x0 . Xt
Xt = Xt1 + t = (Xt2 + t1) + t = Xt2 + t1 + t = (Xt3 + t2) + t1 + t =
= Xt3 + t2 + t1 + t = ... = X0 + (1 + ...+ t ),
t
X t = X 0 + j .
j =1
:
. . .. www.iet.ru
11
E(XtX0 = x0) = x0 ,
D(XtX0 = x0) = D(1 + ... + t ) = D(1) + ... + D(t ) = tD(1) = t2 .
,
Cov(Xt , Xt1X0 = x0) = E[(Xt x0)(Xt1 x0)X0 = x0] =
= E[(1 + ... + t )(1 + ... + t1 )] = (t 1) 2
x0 ,
Corr ( X t , X t 1 ) =
(t 1) 2
(t 1) 2
=
=
D( X t ) D( X t 1 )
2t 2 (t 1)
t 1
1
= 1 .
t
t
:
t
Corr(Xt , Xt1)
0.707
0.806
0.866
0.894
0.913
0.925
0.935
0.943
10
0.949
.. Xt Xt1 ,
, t .
.
,

x0 .
x0 (
),
. ,
,
, , x0 .
X0 = 0
t
X t = j , t = 1, , T .
j =1
. . .. www.iet.ru
12
(
), :
E(Xt) = 0 , D(Xt) = t2 ,
.
,

, .
,
,
, (,
) .
.

Xt = + t + t , t = 1, , T ,
..
. , ..
Xt = + Xt1+ t , t = 1, , T , X0 = x0 ,

E( Xt) = 0.
Xt
Xt = + Xt1 + t = + (+ Xt2 + t1) + t = 2a+ Xt2 + t1 + t =
= 3a + Xt3 + t2 + t1 + t = = x0 + a t + (1 + ...+ t ),
t
X t = x0 + at + j ,
j =1
Xt .

Xt0 = Xt (+ t) = t ,
.
t
X t0 = X t ( x0 + at ) = j ,
j =1
.
. ,
Xt
Xt = Xt Xt1 ,
. . .. www.iet.ru
13
( ).

Xt = Xt Xt1 = (+ t + t ) (+ (t 1)+ t1 ) = + t t1 ,

Xt = Xt Xt1 = + t .
.

Xt
MA(1)
. MA(0)
.
, ,
.
,
.

. (., ,
[Hamilton (1994), 4 5].) -

,
, 3.
,
Xt = 0 + 1 t + 2 t2 + t
Y t = + t + t2 + Z t ,
Zt - ,
Zt = t + 2t 1 + 3t 2 + + t 1 , t = 1, , T .

Xt0 = Xt (0+ 1 t + 2 t2) = t .

Yt0 = Yt ( + t + t2) = Zt ,

D(Zt) = D(t + 2t 1 + 3t 2 + + t 1) = 2(1 + 2 + + t) = 2 t (t + 1)/2,
.
,
Xt
Xt = 1 2 + 22 t + t t 1 ,
. . .. www.iet.ru
14
:
Xt (1 2 + 22 t) = t t 1
, MA(1) .
Yt ,
Yt = ( + 2 t) + Zt Zt 1 =
= + 2 t + (t + 2t 1 + 3t 2 + + t 1)
(t 1 + 2t 2 + 3t 3 + + (t 1)1) =
= + 2 t + (1 + 2 + + t 1 + t)
:
Yt ( + 2 t) = (1 + 2 + + t 1 + t)
.
, Yt , ..
2Yt , 2 = (1 L)2 = 1 2L + L2 ,
MA(0)
2Yt = 2 + 2 Zt = 2 + Zt 2Zt 1 + Zt 2 =
= 2 + (t + 2t 1 + 3t 2 + + t 1)
2(t 1 + 2t 2 + 3t 3 + + (t 1)1) +
+ (t 2 + 2t 3 + 3t 4 + + (t 2)1) = 2 + t .
Xt ,
2Xt = ( Xt ) = (Xt Xt1) = (Xt Xt1) (Xt 1 Xt 2 ) =
= Xt 2Xt 1+ Xt 2 = 2 2 + t 2t 1 + t 2 ,

MA(2) 22 ,
. , b(z) = 0
1 2z + z2 = 0
z = 1.
, Xt
Yt , Xt
.
.

Xt
f(t) , Xt f(t) . Xt
, ,
,
, TS (TS time stationary).
. . .. www.iet.ru
15
TS ,
.
Xt k, k = 1, 2, ,
Xt

, .. TS ;
k Xt , k- Xt
, ;
k 1Xt , (k 1)-
Xt , TS .
0Xt = Xt , k = 1 .
k I(k) .
Xt k ,
Xt ~ I(k). Xt ~ I(0) ,

TS .
TS Xt = + t + Yt , Yt ,
. Xt
X t = + t + j t j , 0 = 1 ,
j=0
2
j
<,
j =0
t . (

.) ,
:
X t = f (t ) + j ( t j t 1 j ) =
j =0
= + b j t j = + b( L) t ,
j=0
b( L) = b j L j , b0 = 1 , b j = j j 1 , j = 1, 2, K ,
j =0
b(1) = b j = 0 ,
j=0
.. b(z) = 0 .
, Zt ,
. . .. www.iet.ru
16
Zt = + b(L) t , b( L) = b j L j , b0 = 1 ,
j =0

b(1) = 0,
Zt : TS
TS .
k = 1, 2,
, DS (DS difference stationary) .
Xt , DS .
Xt k . k-
.
ARMA(p, q), , Xt ARIMA(p, k, q),
k ARMA(p, q) (ARIMA autoregressive
integrated moving average). p = 0 q = 0,
:
ARIMA(p, k, 0) = ARI (p, k), ARIMA(0, k, q) = IMA( k, q),
ARIMA(0, k, 0) = ARI (0, k) = IMA( k, 0).
, :
Xt = + t + t ~ I(0);
Xt = + Xt1+ t ~ I(1), Xt ARIMA(0, 1, 0);
Xt = 0 + 1 t + 2 t2 + t ~ I(0);
Xt = + t + t2 + t + 2t 1 + 3t 2 + + t 1 ~ I(2), Xt ARIMA(0, 2, 0).
TS , DS
.
, ,

TREND_1 t = 1 + 0.5 t + t , t ~ N(0, 1),
WALKt = 0.5+ WALKt1 + t , t ~ N(0, 0.52), WALK0 = 0:
. . .. www.iet.ru
60
17
50
50
40
40
30
30
20
20
10
10
0
10
20
30
40
50
60
70
80
90 100
10
20
30
40
TREND_1
50
60
70
80
90
100
WALK
:
Dependent Variable: TREND_1
Variable
Coef.
Std. Error
t-Statistic
Prob.
C
T
0.208085
0.003577
3.827226
140.1946
0.0002
0.0000
Std. Error
t-Statistic
Prob.
-3.726248
101.9477
0.0003
0.0000
0.796390
0.501522
Dependent Variable: WALK

Variable
Coef.
C
T
-0.930832 0.249804
0.437818 0.004295
:
3
-1
-1
-2
-2
-3
-3
10
20
30
40
50
60
70
X_DETRENDED
80
90
100
10
20
30
40
50
60
70
80
90
100
WALK_DETRENDED
,
.
.
[Chan, Hayya, Ord (1977)] [Nelson, Kang (1981)]:

, (
)
.
,
:
. . .. www.iet.ru
2.0
1.5
18
1.0
0.5
0
0.0
-2
-0.5
-4
10
20
30
40
50
60
70
80
90
100
-1.0
10
20
X_TREND_DIF
30
40
50
60
70
80
90
100
WALK_DIF
(X_TREND_DIF)
ACF
***| .
.|.
.|.
.|.
. |*
*| .
. |*
**| .
. |*
.|.
PACF
***| .
**| .
**| .
**| .
.*| .
**| .
. |*
*| .
.|.
*| .
AC
PAC
Q-Stat Prob
1 -0.449 -0.449 20.527 0.000

2 -0.045 -0.308 20.736 0.000
3 -0.006 -0.236 20.740 0.000
4 -0.052 -0.266 21.021 0.000
5
0.078 -0.157 21.676 0.001
6 -0.074 -0.221 22.258 0.001

7
0.194 0.073 26.358 0.000
8 -0.216 -0.120 31.473 0.000

9
0.079 -0.047 32.159 0.000
10 -0.056 -0.143 32.513 0.000
(WALK_DIF)
ACF
.|.
.|.
.|.
.|.
.|.
.|.
. |*
*| .
.|.
.|.
PACF
.|.
.|.
.|.
.|.
.|.
.|.
. |*
*| .
.|.
*| .
AC
1
PAC
Q-Stat Prob
0.035 0.035 0.1271 0.721
2 -0.044 -0.045 0.3271 0.849

3 -0.042 -0.039 0.5099 0.917
4 -0.025 -0.024 0.5766 0.966
5
0.065 0.063 1.0215 0.961
6 -0.004 -0.012 1.0231 0.985

7
0.101 0.107 2.1405 0.952
8 -0.173 -0.181 5.4243 0.711

9 -0.041 -0.013 5.6087 0.778
10 -0.057 -0.073 5.9707 0.818
.
,

. . .. www.iet.ru
19
MA(1) . r(1) = 0.449

(1) (1) = b1/(1 + b12), 0.449 = b1/(1 + b12).
: 1.6036 0.6236.
, MA(1) . ,
b1 0.6236.
MA(1),
.
Dependent Variable: X_TREND_DIF

Backcast: 1
C
MA(1)
0.501429 0.004483 111.8447 0.0000

-0.977743 0.015377 -63.58459 0.0000
backcasting ( . . 3.2)
Dependent Variable: X_TREND_DIF

Backcast: OFF
C
MA(1)
0.518807 0.007616 68.12224 0.0000

-1.062852 0.042816 -24.82378 0.0000
backcasting .

b1 , 1, MA(1)
, . ,
,

. ([Slutsky (1937)].
:
Xt = 0.01 t2 + t , t ~ N(0, 52),
Yt = 0.04 t2 + t + 2t 1 + 3t 2 + + t 1 , t ~ N(0, 0.12):
. . .. www.iet.ru
120
120
100
100
80
80
60
60
40
40
20
20
-20
-20
10
20
30
40
50
60
70
80
90
100
10
20
30
40
50
60
70
80
90
20
100
Variable
Coef.
Std. Error
t-Statistic
Prob.
T^2
0.000114
86.93333
0.0000
0.009926
( .)
Variable
Coef.
C
T
T^2
Std. Error
-2.273387 0.621988
-0.119781 0.028427
0.013087 0.000273
t-Statistic
Prob.
-3.655036
-4.213709
47.99344
0.0004
0.0001
0.0000
:
15
10
2
5
0
0
-5
-2
-10
-4
-15
10
20
30
40
50
60
70
80
90
100
10
20
30
40
50
60
70
80
90
100
Y_DETRENDED
X_DETRENDED
X_DETRENDED
ACF
.|.
.|.
.|.
.|.
. |*
.|.
PACF
.|.
.|.
.|.
.|.
. |*
.|.
AC
PAC
Q-Stat Prob
1 0.030 0.030 0.0936 0.760

2 -0.049 -0.050 0.3429 0.842
3 -0.042 -0.039 0.5278 0.913
4 -0.024 -0.024 0.5867 0.965
5 0.071 0.068 1.1208 0.952
6 0.000 -0.008 1.1208 0.981
. . .. www.iet.ru
. |*
*| .
.|.
.|.
. |*
*| .
.|.
*| .
21
7 0.101 0.107 2.2429 0.945

8 -0.171 -0.177 5.4698 0.706
9 -0.039 -0.011 5.6366 0.776
10 -0.050 -0.068 5.9165 0.822
, Y_DETRENDED
ACF
PACF
AC
PAC
Q-Stat Prob
. |********
. |******** 1 0.985 0.985 99.989
2 0.956 -0.507 195.02
. |******* ****| .
3 0.912 -0.314 282.52
. |*******
**| .
4 0.857 -0.196 360.52
. |*******
**| .
5 0.791 -0.136 427.67
|******
*| .
6 0.716 -0.073 483.36
. |******
*| .
7 0.635 -0.044 527.61
. |*****
.|.
8 0.549 -0.067 560.97
. |****
*| .
9 0.458 -0.043 584.49
. |***
.|.
10 0.365 -0.051 599.58
. |***
.|.
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
AR(2) .
Xt Yt :
30
0.4
20
0.3
10
0.2
0.1
0.0
-10
-0.1
-20
-0.2
-30
10
20
30
40
50
60
70
80
90
100
-0.3
10
20
30
X_DIF2
40
50
60
70
80
90
100
Y_DIF2
X_DIF2
ACF
*****| .
. |*
.|.
*| .
. |*
*| .
. |**
**| .
. |*
PACF
*****| .
****| .
**| .
**| .
*| .
**| .
.|.
.|.
.|.
AC
PAC
Q-Stat Prob
1 -0.635 -0.635 40.777 0.000

2 0.128 -0.463 42.440 0.000
3 0.038 -0.294 42.586 0.000
4 -0.064 -0.273 43.014 0.000
5 0.081 -0.141 43.710 0.000
6 -0.137 -0.292 45.710 0.000
7 0.243 0.037 52.065 0.000
8 -0.254 -0.037 59.064 0.000
9 0.160 0.057 61.891 0.000
. . .. www.iet.ru
*| .
.|.
22
10 -0.099 -0.056 62.976 0.000
MA . r(1) = 0.635
(1) MA(1), ..
0.5 (1) 0.5 . ( 0.635 (1)
(1) = b1/(1 + b12) 0.635 b12 + b1 +
0.635 = 0, .)
Y_DIF2
ACF
PACF
.|.
.|.
.|.
.|.
. |*
.|.
. |*
*| .
.|.
*| .
.|.
.|.
.|.
.|.
. |*
.|.
. |*
*| .
.|.
*| .
AC
PAC
Q-Stat Prob
1 0.031 0.031 0.0942 0.759

2 -0.041 -0.042 0.2636 0.877
3 -0.038 -0.036 0.4127 0.938
4 -0.013 -0.013 0.4313 0.980
5 0.073 0.071 0.9925 0.963
6 -0.006 -0.012 0.9958 0.986
7 0.104 0.111 2.1713 0.950
8 -0.172 -0.178 5.3769 0.717
9 -0.031 -0.006 5.4804 0.791
10 -0.066 -0.084 5.9673 0.818
DS DS ;
TS TS ;
TS
ARMA, TS MA
, ;
k- Xt ~ I(k)
; I(k)
ARIMA, k-
MA .
TS
, TS-
, DS-

. AR(1) Xt = a X t 1 + t .
(. . 2.3)
Xt = a t X0 + a t 1 1 + a t 2 2 + + t ,

. . .. www.iet.ru
23
Xt + h = a t + h X0 + a t + h 1 1 + a t + h 2 2 + + ah t + + h+ t .
,
() () t
:
Xt t = 1, Xt +1 t = a, Xt +2 t = a2, , Xt + h t = ah, .
, h
Xt + h t 0 a < 1,
Xt + h t 1 a = 1.
, a > 1 ( ),
Xt + h t ,

.
5.2. TS
DS

() .
,
, (
) TS (trend stationary) , ,
(, )
k-2
DS (difference stationary) .
,
TS
, DS
-
.
,
,
. ,
TS DS ,
TS DS
2
. . .. www.iet.ru
24
,
(
, , ).
, DS, (.
7), , ,
(., ,
[Dweyer, Wallace (1992), Dutt, Ghosh (1999)]),

([Ardeni, Lubian (1991), Dutt(1998)]),

([Johansen, Juselius (1990)], [Hafer, Jansen (1991)], [Funke, Thornton (1999)]).
, DS-

,

, .
.
[Maddala, Kim (1998)], [Enders (1995)], [Hamilton
(1994)], [Hatanaka (1996)]. ( ) ,
: [Hasan (1998)].
: [Metin (1995)]
: [Christiano, Eichenbaum (1990)]; [Murray, Nelson
(2000)].
: [Clark (1989)]; [Woodward, Pillarisetti (1999)].
: [Copeland (1991)], [Kim, Mo (1995)],
[Nadal-De Simone, Razzak (1999)].
: [Milas (1998)].
: [Molana (1994)] .
: [Cheung, Chinn (1996)], [den Haan (2000)].
: [Fama, French (1988)].
TS ,
, , ,
-. DS (
) , -
DS
( ),
. . .. www.iet.ru
25
, -
,
.
TS-
, DS-
.

,

,
- .

, (, ARMA)
, , ..
. ([,
(1974)]) ARIMA
()
. ,
,
, .
[Chan, Hayya, Ord (1977)], [Nelson, Kang (1981)] ,
DS ()
, (
),
,
TS ,
,
, -;
( ). ,
-

. (., , [Hamilton (1994), 4 5].)
, ,
,

, ,
.. (TS
DS). .
,
, , [Maddala, Kim (1998)],

.
,
,
( , )

. . .. www.iet.ru
26
, .

, ,
[Maddala, Kim (1998)], [Enders (1995)], [Hamilton (1994)], [Hatanaka (1996)].
,

TS- DS-.
, 14
( 62 111 ) [Nelson, Plosser
(1982)] ([Perron (1989a)]).
14 TS, ,
, 11 . ,

TS . TS- ,
, .
, ,
, (. [Zivot, Andrews
(1992)]).
, . [Bierens (1997)]. , [Nunes, Newbold, Kuan (1997)]
:
[Zivot, Andrews (1992)]
14 , [Nelson, Plosser (1982)].
,
,
, (TS DS)
.
. . .. www.iet.ru
27
5.3. TS DS ARMA.
.
, Xt
TS ( ) DS (
) .
. ,
, ,
, ( ),
.
, ,

.
,
, .

TS DS ,
, (
) .
TS DS
.
.
, DS TS ,
ARMA ( ).
Xt ARIMA(p, k, q), k-
kXt ARMA(p, q),
,
a*(L) kXt = b(L) t ,
a*(L) b(L) L ,
p q, . , Xt = (1 L) Xt ,
kXt = (1 L)kXt ,
*
a (L) (1 L)k Xt = b(L) t ,
a(L) Xt = b(L) t ,
a(L) = a*(L) (1 L)k (p + k). kXt
, p a*(z) ,
a(z) p k
, , z = 1 k .
, Xt ARMA(p+ k, q),
a(L) k , 1,
1. H0 ,
. . .. www.iet.ru
28
ARMA Xt DS- ( ),
, a(L)
, 1. , ,
a(z) , ..
. H0
(UR unit root hypothesis),
. TS
, ARMA .
, () TS,
DS-.
a(z)
z = 1 b*(z) = 0, b*(L)
L Xt =
b*(z) t Xt = Xt Xt1 Xt .
. . .. www.iet.ru
6. TS DS

.
6.1.
, ,
,
.
TS
xt = + t + a1 xt 1 + t , | a1| < 1.
?
yt , yt = xt t
yt = a1 yt 1 + t .
yt yt 1 , :
xt t = a1(xt 1 (t 1)) + t ,
xt = ( a1 + a1)+ (1 a1) t + a1xt 1 + t ,
= a1 + a1 = (1 a1),
= (1 a1), = ( a1( + )) (1 a1)2 .
, xt
a1 ( + )
t.
+
2
(1 a1 )
(1 a1 )
, = 0 0
.
(1 a1 )
a1 = 1, ,
xt .
xt = + t + xt 1 + t =
= ( + t + t) + ( + ( t 1) + t1) + + ( + + 1) + x0 =
= x0 + ( + /2) t + (/2) t2 + (1 + 2 + + t) .
= = 0
xt = xt1 + t , xt = x0 + (1 + 2 + + t) .
0, = 0
xt = + xt1 + t , xt = x0 + t + (1 + 2 + + t) ,
.. x0 + t .
, 0, 0
xt = + t + xt1 + t , xt = x0 + ( + /2) t + (/2) t2 + (1 + 2 + + t) ,

xt
x0 + ( + /2) t + (/2) t2 .
,
xt = + t + a1 xt1 + t
0, = 0 ,
| a1| < 1 xt ;
. . .. www.iet.ru
a1 = 1 xt
j =1
x0 + t ;
0,
| a1| < 1 xt
a1 ( + )
t;
+
2
(1 a1 )
(1 a1 )
t
a1 = 1 xt
j =1
x0 + ( + /2) t + (/2) t2.

,
xt = + t + a1 xt 1 + t
, , a1 :
ST_1 : a1 = 0.8, = 0, = 0 .
ST_2 : a1 = 0.8, = 0.2, = 0 .
ST_3 : a1 = 0.8, = 0.16, = 0.04 .
WALK_1 : a1 = 1, = 0, = 0 .
WALK_2 : a1 = 1, = 0.2, = 0 .
WALK_3 : a1 = 1, = 0.2, = 0.1 .

, ,
xt = + t + t2 + a1 xt 1 + t :
ST_4 : a1 = 0.8, = 0.12, = 0.13, = 0.01.

6
4
2
0
-2
-4
-6
10
20
30
40
50
60
70
80
90
100
ST_1
0
2
-2
-4
-6
-2
-8
-4
10
20
30
40
50
60
ST_2
70
80
90
100
-10
10
20
30
40
50
60
70
80
90
100
WALK_1
. . .. www.iet.ru
25
15
20
10
15
5
10
0
5
-5
0
-5
-10
10
20
30
40
50
60
70
80
90
100
10
20
30
40
ST_3
50
60
70
80
90
100
70
80
90
100
WALK_2
600
600
500
500
400
400
300
300
200
200
100
100
0
10
20
30
40
50
60
70
80
90
100
ST_4
10
20
30
40
50
60
WALK_3
ST_1, ST_2, WALK_1

ST_3 WALK_2, ST_4 WALK_3 ,
,

.
( )
.
6.2.
,

SM: xt = a1 xt 1 + t , t = 1, , T ,
, , ( ,
DGP data generating process),
DGP: xt = xt 1 + t , t ,
.. a1 = 1, a1 a1
.
. T a1 = 1
(
1 / 2) {[W (1)]2 1}
n(a1 1)
,
1
2
[W (r )] dr
0
W(r)
,
, , .
W(r)
xt = xt 1 + t .
,
. . .. www.iet.ru
W(0) = 0;
(W(r2) W(r1)),, (W(rk) W(r k1))

, 0 r1 < r2 < < rk ; W(s) W(r) ~ N(0, s r) s > r ;
W(r) 1.
, , , W(1) = W(1) W(0) ~ N(0, 1), [W(1)]2

~ 2(1). , T
P{a1 < 1} = P{a1 1 < 0} P{[W (1)]2 1 < 0} = P{ 2 (1) < 1} = 0.68 ,
DGP ( ), DGP: xt = xt 1 + t ,
SM: xt = a1 xt 1 + t a1 < 1 2/3 .
T( a1 1) a1 = 1
T ( );
[Fuller (1976)].
, t ~ N(0, 2), 1 , , T x0
= 0. , x0
T( a1 1) , T( a1 1) .
, T( a1 1), H0: a1 = 1
HA: a1 < 1 5% T( a1 1),
T( a1 1), a1 , a1 ,
T
T( a1 1) a1
0.708
25 7.3
0.846
50 7.7
0.921
100 7.9
0.968
250 8.0
8.0
0.998
500
8.1
, T H0
a1 , 1.
. ,
T H0: a1 = 1 HA: a1 < 1
, a1 1.
, , H0: a1
= 1 HA: a1 < 1
a 1
(t-, t-ratio, t-),
t= 1
s (a1 )
s( a1 ) a1 . , a1 = 1
a1 ,
t- . t- T
T
[Fuller (1976)]. H0: a1 = 1 HA:
a1 < 1 . 5%
, , 5%
t-, t(T 1) (T
1) :
T

1.71
25 1.95
1.68
50 1.95
. . .. www.iet.ru
1.66
100 1.95
1.65
250 1.95
1.65
500 1.95
1.95
( a1 1)/s( a1 ) a1 = 1.
, . 6.1
xt = xt1 + t (WALK_1).
xt = a1 xt1 + t , 50 :
Dependent Variable: WALK_1
Variable
Coeff.
Std. Error
t-Statistic
Prob.
WALK_1(-1)
0.970831
0.035729
27.17224
0.0000
t- (0.970831 1)/ 0.035729 = 0.816

1.95.
5% .

0.970831
a1,
0.846.
, . 6.1 ST_1
xt = 0.8 xt1 + t ( )
xt = a1 xt1 + t ,
50 , :
Dependent Variable: ST_1
AR(1)
0.790557
0.090501 8.735349 0.0000
t- 2.314 < 1.95,

. , a1
0.791 1.
ST_2 AR(1)
xt
= 0.2 + 0.8 xt1 + t
(
1).
50 0.596.
, , ,
, .
H0: xt = xt1 + t ( )

HA: xt = + a1 xt1 + t , a1 < 1, 0.

SM: xt = + a1 xt1 + t .
T( a1 1) ,
SM = 0; T
= 25 P{ a1 < 1} = 0.95.
:
T
T( a1 1) a1 t ()
0.500 3.00
25 12.5
. . .. www.iet.ru
0.734 2.92
50 13.3
0.863 2.89
100 13.7
0.944 2.88
250 14.0
0.972 2.87
500 14.0
14.1
2.86
SM: xt = + a1 xt1 + t WALK_1 (
50 )
t = 2.143 > t = 2.92,
, xt =
xt1 + t , .
ST_2 ( T = 50)
t = 2.245 , ,
. , ,
, a1 = 0.794 0.734.
SM: xt = + a1 xt1 + t ST_1 (c =

0),
a1 = 0.785 > a1 = 0.734, t = 2.298 > t = 2.92,
ST_1 . ,
, ST_1 SM: xt =
a1 xt 1 + t , .
,
( .. )
, ,
. ,
,
.
ST_3
xt = 0.16 + 0.04 t + 0.8 xt 1 + t ,
0.2 t .
WALK_2 0.2
xt = 0.2 + xt 1 + t :
25
15
20
10
15
5
10
0
5
-5
0
-5
-10
10
20
30
40
50
60
ST_3
70
80
90
100
10
20
30
40
50
60
70
80
90
100
WALK_2
, ,

H0: xt = + xt1 + t , 0, ( )

SM: xt = + t + a1 xt1 + t .
. . .. www.iet.ru
a1 t-
a1 = 1 .
DGP: xt = + xt1 + t 0, 5% T( a1 1)
t- :
T
T( a1 1) a1 t ()
0.284 3.60
25 17.9
0.604 3.50
50 19.8
0.793 3.45
100 20.7
0.914 3.43
250 21.3
0.957 3.42
500 21.5
21.8
3.41
SM: xt = + t + a1 xt1 + t
( T = 50).
WALK_2 a1 = 0.858 > a1 = 0.604, t = 2.027 > t = 3.50,
.
ST_3 a1 = 0.733 > a1 = 0.604, t = 2.687 > t = 3.50.
a1 t , WALK_2,
, .
, , ST_3,
, DGP ,
a1 t
SM: xt = + t + a1 xt1 + t
,
DGP: xt = xt1 + t .
, a1
t ,
DGP: xt = + xt1 + t , 0,

SM: xt = + t + a1 xt1 + t
a1 t DGP: xt = xt1 + t
DGP: xt = + xt1 + t , 0.
SM: xt = + t + a1 xt1 + t
DGP: xt = xt1 + t , WALK_1.
, ( DGP: xt = 0.2 + xt1 + t ): a1 = 0.858 >
a1 = 0.604,
t = 2.027 > t = 3.50,
.
, , [MacKinnon (1991)],
t . , , t(p, T)
t- , p
T ,
t(p, T) + 1T 1 + 2T 2 ,
, 1 , 2 , p ,
.
p = 0.01, 0.05, 0.10.
. . .. www.iet.ru
5

1974 . 1985 .

:
Xt 47962.75 315.1909 t = 0.884803 (Xt1 47962.75 315.1909 (t1)) + t ,
Xt = 5804.037 + 36.30898 t + 0.884803 Xt1 + t .

, ,
0.884803 Xt1
(..
AR(1) ),
,
.
.
,

SM: xt = + t + a1 xt 1 + t
H0: xt = + xt 1 + t ,
EVIEWS, :
Test Statistic
-1.425277
1% Critical Value*
5% Critical Value
10% Critical Value
-4.1630
-3.5066
-3.1828
*MacKinnon critical values for rejection of hypothesis of a unit root.
, H0 . ,
, xt = +
t + a1 xt 1 + t , xt = + xt 1 + t . xt =
+ t :
236.8958 80.80998 2.931517 0.0052

Durbin-Watson stat 2.135959

xt = 236.8958 + xt 1 + t .

SM: xt = a1 xt 1 + t
SM: xt = + a1 x t 1 + t
SM: xt = + t + a1 xt 1 + t
a1 t a1 = 1
:
DGP: xt = xt 1 + t ( ),
. . .. www.iet.ru
DGP: xt = + xt 1 + t ( ).
, , SM: xt = + a1 xt 1 + t , xt = xt 1 + t

H0: = 0, a1 = 1.
,
F-
( RSS 0 RSS ) 2
,
1 =
RSS ((T 1) 2)
H0 F- F(2,
T 3) T 3 . , ,
H0 , , , ,
1 H0 ( )
F(2, T 3). [Dickey,
Fuller (1981)]. 1
H0: = 0, a1 = 1. 5% 1 ,
, ( ) 5%
F , F(2, n 3) (., [Hamilton (1994),
.7 Case 2] [Enders (1995), ]):
1 F
T
3.44
25 5.18
3.20
50 4.86
3.10
100 4.71
3.00
250 4.63
3.00
500 4.61
4.59
3.00
WALK_1, ST_2, ST_1 ( 50 ).

SM: xt = + a1 xt1 + t WALK_1, = 0.579, a1 = 0.850,
RSS = 48.0335. = 0, a1 = 1 xt = xt1 ,
T
RSS 0 = ( xt xt 1 ) = 52.7939 ,
2
t=2
(52.7939 48.0335) / 2
= 2.329 < 4.86
1 =
48.0335 /(50 1 2)
H0 .
ST_2 : = 0.181, a1 = 0.777, RSS = 52.6618. = 0, a1 =
1 RSS0 = 59.0547,
(59.0547 52.6618) / 2
= 2.853 < 4.86
1 =
52.6618 /(50 1 2)
H0 .
ST_1 : = 0.042, a1 = 0.785, RSS = 52.7007. = 0, a1
= 1 RSS0 = 58.0671,
1 = 2.662 < 4.86 H0 .
, ,
t- a1 , ,
F-.

. . .. www.iet.ru
10
SM: xt = + t + a1 xt1 + t
DGP: xt = xt1 + t
H0: = = 0, a1 = 1 ,
DGP: xt = + xt1 + t
H0: = 0, a1 = 1 .
F- 2 ; 5%
(. [Enders (1995), ])
1
T
25 5.68
50 5.13
100 4.88
250 4.75
500 4.71
4.68
F- 3 ; 5%
(. [Hamilton (1994), .7 Case 4] [Enders (1995), ]):
1
T
25 7.24
50 6.73
100 6.49
250 6.34
500 6.30
6.25
WALK_1, WALK_2, ST_3.

SM: xt = + t + a1 xt1 + t

H0: = = 0, a1 = 1 , 2;
H0: = 0, a1 = 1 , 3 .
H0: = = 0, a1 = 1
WALK_1: = 0.854, = 0.012, a1 = 0.858, RSS = 46.7158.
RSS0 = 52.7939,
2 = 1.995 < 5.13 H0 .
WALK_2: = 0.711, = 0.040, a1 = 0.858, RSS = 46.7158.
RSS0 = 52.7939,
2 1.995 < 5.13 H0 .
ST_3: = 0.345, = 0.070, a1 = 0.733, RSS = 50.3928, 2 = 1.207 < 5.13
H0 .
H0: = 0, a1 = 1
WALK_1: RSS = 46.7158. RSS0 = 52.7282,
3 = 1.973 < 5.13 H0 .
WALK_2: 3 1.973 H0 .
ST_3: RSS = 50.3928, 2 = 0.711 H0 .
,

. . .. www.iet.ru
11

.
. ,
1 a1 < 0.8, 1
> 1. 2 3 .
, t- T( a1 1).
,
H0 .
,
( )
.
, ,
. ,
.
,
() .

: 50
ST_1 ST_2
DGP: xt = xt1 + t , SM: xt = + a1 xt1 + t ;
ST_3
DGP: xt = + xt1 + t ( DGP: xt = xt1 + t) ,
SM: xt = + t + a1 xt1 + t .
, ,
; T = 100.
( 10%
):
n = 50
n = 100
t
=
3.238
< t = 2.89,
t
=
2.298
>
t
=
2.92,
ST_1

t = 2.245 > t = 2.92,

t = 3.217 < t = 2.89,
ST_2

t
=
3.207
< t = 3.15,
t = 2.687 > t = 3.18.
ST_3

10%
10%
,
.
. . .. www.iet.ru
12
6.3. -

(GNP) 1947 . 1961
. 5
Xt 217.740 5.222 t = 1.380 (Xt1 217.740 5.222(t1))
0.630 (Xt2 217.740 5.222(t2)) + t,
Xt = 55.017 + 1.304 t + 1.380 Xt1 0.630Xt2 + t .

,
?
.
.

SM: xt = + t + a1 xt1 + a2 xt2 + + ap xtp + t .

(#) xt = + t + xt1 + (1 xt1 + + p1 xt p+1 ) + t ,
= a1 + a2 + + ap , j = (aj + 1 + + ap) .
( GNP
xt = 55.017 + 1.304 t + 1.380 Xt1 0.630Xt1 + 0.630Xt1 0.630Xt2 + t =
= 55.017 + 1.304 t + 0.750 Xt1 + 0.630 Xt1 + t .)
, a(z) = 0 z = 1,
p 1 ,
, a1 + a2 + + ap = 1, .. = 1. (.,
, [Hamilton (1994)].) ,
AR(p)
H0: = 1
(#) .
,
T( 1) t- = 1,
(augmented) (#) ( ,
). t- ADF (augmented
Dickey Fuller), DF, AR(1).
, t-
= 1, (#)
(# #) xt = + t + xt1 + (1 xt1 + + p-1 xt p+1 ) + t ,
= 1, H0: = 1 (#) H0: = 0 (# #).
H0: = 1 (#) HA: < 1.
(#) (# #)) HA: < 0 .
, t- H0: = 1 (#)
t- H0: = 0 (# #).
GNP
xt = + t + xt1 + 1 xt1 + t
( ) :
. . .. www.iet.ru
ADF Test Statistic
-4.117782
1% Critical Value*
5% Critical Value
10% Critical Value
13
-4.1219
-3.4875
-3.1718

Augmented Dickey-Fuller Test Equation
Dependent Variable: D(X)
Variable
t-Statistic
Prob.
X(-1)
D(X(-1))
C
@TREND(1947:1)
-4.117782
5.756490
4.272264
4.135949
0.0001
0.0000
0.0001
0.0001
-0.249792
0.630066
56.32136
1.304300
0.060662
0.109453
13.18303
0.315357
: t-
H0: = 0 5% ,
, 1% .
, ,
, ,
, , t
. ,
, ..
,
.

p = p * , ,
t-
= 0 , 1, , p1
.
, p* 1 = 0, t-
t- . ,
p* 1 = p* 2 = 0, F-
F- , ...
.
. p* = 5, :
ADF Test Statistic
-2.873575
1% Critical Value*
5% Critical Value
10% Critical Value
-4.1314
-3.4919
-3.1744

Variable
t-Statistic
Prob.
X(-1)
D(X(-1))
D(X(-2))
D(X(-3))
-2.873575
4.090958
1.228486
-0.133077
0.0060
0.0002
0.2253
0.8947
-0.266169
0.546230
0.183918
-0.020254
0.092626
0.133521
0.149711
0.152201
. . .. www.iet.ru
D(X(-4))
C
@TREND(1947:1)
-0.058683
59.45556
1.397409
0.148061
19.32396
0.482120
-0.396345
3.076779
2.898469
14
0.6936
0.0035
0.0056
t = 2.873575 > 3.1744,

10% . ,
. P-
F- 0.44.
,
, .
- ,
(,
). , xt ARMA(p,
q) q > 0 ?
xt ~ ARMA(p, q) , a(L) xt = b(L) t ,
a(L), b(L) p q, b(L) ,

c(L) xt = t ,
c(L) xt = a(L) b(L) = 1 + 1L + 2L2 + .

(# #)

(# # #) xt = + t + xt 1 + (1 xt 1 + 2 xt 2 + ) + t

. ?
[Said, Dickey (1984)] , ARIMA(p, 1, q)
p q
ARI(p*, 1) p* < 3 T .
(# # #) .
6.4.
,
, [Dickey (1976)],
[Fuller (1976)], [Dickey, Fuller (1979)], [Dickey, Fuller (1981)].
() , xt
DS (DS-);
TS (TS-).
,
(, ).
,
.
(SM , statistical model; DGP
, data generating process).
1) xt (
),
SM: xt = xt 1 + + t + t , t = 2,, T ,
. . .. www.iet.ru
15
DGP: xt = + t , t = 2,, T , 0 .
t ,
.
.
SM
t- t H0 : = 0.
tcrit , ,
DGP
( ). DS- , t < tcrit.
, , ,
[Fuller (1976)], [Fuller (1996)],
T = 25, 50, 100, 250, 500. T ,
, ,
[MacKinnon (1991)].
2) xt (
) ,
SM: xt = xt 1 + + t , t = 2,, T ,
DGP: xt =
t = 2,, T .
SM
t- t H0 : = 0.
tcrit , ,
DGP (
). DS- , t < tcrit . ,
, ,
ax [Fuller (1976)], [Fuller (1996)],
T = 25, 50, 100, 250, 500. T ,
, ,
[MacKinnon (1991)].
3) , xt (
) ,
SM: xt = xt 1 + t , t = 2,, T ,
DGP:
x t =
t = 2, , T .
SM
t- t H0 : = 0.
tcrit , ,
DGP (
). DS- , t < tcrit . ,
, ,
[Fuller (1976)], [Fuller (1996)],
T = 25, 50, 100, 250, 500. T ,
, ,
[MacKinnon (1991)].

. ,
,

, ,
t ,
. . .. www.iet.ru
16
(. [Perron (1988)]). ,
, .

[Dolado, Jenkinson, Sosvilla Rivero (1990)]; . [Enders (1995)].
6.6.
a(L) xt = t
( ) p , a(z) = 0
,
(augmented) .

xtj , j = 1, , p 1,

(1)
p 1
SM: xt = + t + xt 1 + j xt j +
t = p + 1, , T ,
j =1
(2)
p 1
SM: xt = + xt 1 + j xt j +
t = p + 1, , T ,
j =1
(3)
p 1
SM: xt = xt 1 + j xt j +
t = p + 1, , T .
j =1
t t H0 : = 0
tcrit , . DS- , t <
tcrit.
, xt
.
[Said, Dickey (1984)], x1,, xT
ARIMA(p, 1, q) c q > 0, ARI(p*, 1) =
ARIMA(p*, 1, 0) p*< 3 T .
x1,, xT
AR(p) p, p ,
,
.
p=pmax ,
p0 , ,
* ,
, .
, ,
( 10% )
(GS- )
() pmax
(SIC). [Hall (1994)] [Ng, Perron
(1995)] , pmax p0 , ( ) SIC
, GS 0 ;

- .
. . .. www.iet.ru
17
T , ,
[Cheung, Lay (1995)].
(GS SIC),
,

,
. , (GDP)
1870 1994 . (. [Murray, Nelson (2000)]),
pmax = 8, GS- = 6,
SIC = 1.

LM- (. [Holden, Perman (1994)]). ,
, [Taylor (2000)] ,
Ng Perron:

. , ,
.
6.5. DGP SM
: ,
SM: xt = + a1 xt1 + t ,

DGP: xt = + xt1 + t , 0 ( ).
t DGP

, xt1 SM
i =1
(t 1) .
a1
; , ,
, ..
t-, .
0 , T t-
, = 0,
.
WALK_2 0.2
SM: xt = + a1 xt1 + t
Dependent Variable: WALK_2
Variable
C
0.173019 0.160495 1.078031 0.2865
WALK_2(-1) 0.991851 0.045395 21.84958 0.0000
t- t = (0.991851 1)/0.045395
= 0.180 . 5% t- (n p) = (49 2) =
47 1.68, 5% ,
DGP: xt = xt1 + t ( = 0), 2.92,
. ,
. . .. www.iet.ru
18

1.68 2.92.
ST_2 xt = 0.2 + 0.8 xt1 + t
( , 1) T = 50
Dependent Variable: ST_2
C
0.166899 0.159693 1.045128 0.3013
ST_2(-1) 0.793680 0.091904 8.635959 0.0000
t- t = (0.793680 1)/0.091904
= 2.245. 1.68
, 2.92
.
.
ST_1 xt = 0.8 xt1 + t .
xt , , ,
SM: xt = a1 xt1 + t (, DGP
) SM: xt = + a1 xt1 + t (, DGP
).
5% t- ( T = 50)
: 1.95.
, t = 1.68 (
, ), t = 2.92.
SM t = 2.314.
.
SM t = 2.298. ,
DGP , t < t = 1.68,
. , DGP , t > t =
2.92, .
,
(SM) , ,
, .
SM, .
, DGP ,
,
. , 5% ,
DGP .
6.6. .
,
. .
1
DGP:
() xt = xt1 + t ,
1
.
. 6.8.4.
. . .. www.iet.ru
19
() xt = + xt1 + t , 0,
() xt = + t + xt1 + t , 0 .
;
.

SM: xt = + t + a1 xt1 + t , 0 ,
t- H0: a1 = 1 ,
, DGP.
. 6.2, t-
() (), .. , = 0 0.
DGP () 0 , t- ,
t- (, N(0, 1)).
T [Kwiatkowski, Schmidt (1990)].
WALK_3
:
DGP: xt = 0.2 + 0.1 t + xt1 + t .
SM: xt = + t + a1 xt1 + t ;
a = 0.989 t = 0.775 . t-
t = 1.68, .
, DGP = 0 ,
t = 3.50, .
ST_4
DGP: xt = 0.12 + 0.13 t + 0.01 t2 + 0.8 xt1 + t ,
, a =
0.990 t = 0.577 .
, .
6.7.
.( [Dolado, Jenkinson, Sosvilla-Rivero (1990)])

,
(SM) (DGP).
, ,
AR(1), , .
1 ,
,
:
SM:
xt = + t + xt 1 +
t = 2,, T ,
,

DGP: xt = + t , t = 2,, T .
: ().
DS (UR
Unit Root) .
. . .. www.iet.ru
20
H0 : = 0; HA :
< 0. t - ,
,
. , = 0 0.
H0 : = 0 ,
. , H0 : = 0

DGP: xt = + t ( c = 0 0),

DGP: xt = + t + t , 0,
.. t ( ).
2.
1 H0 : = 0 , :
, = 0 ;
0 , H0 : = 0 - ,
DGP = 0, DGP: xt =
+ t + t , 0.
, 2
H0: = 0

SM: xt = + t + xt1 + t ,

DGP: xt = + t + t , 0.
t- (
) [Dickey, Fuller (1981)]. 5%
t t
.
n
( >
0)
2
5
3.25
2.85
5
0
3.18
2.81
1
0
0
3.14
2.79
2
5
3.12
2.79
. . .. www.iet.ru
21
0
5
0
0
3.11
2.78
3.11
2.78
H0: = 0 , ,
= 0 -
, DGP = 0.
H0: = 0 ,
= 0
SM: xt = + t + xt1 + t ,

DGP: xt = + t + t , 0.
t- ( 0) N(0,
1) . T [Kwiatkowski,
Schmidt (1990)]. , = 0 ,
. ,
xt = + t + t , 0.
3,

SM: xt = + t + xt1 + t .
, -
t .
, 3
SM: xt = + xt1 + t
= 0 ( < 0)
SM. t-
( 2).
DGP: xt = t .
, H0: = 0 , (
, 1).
3 = 0 ,
SM . = 0

. . .. www.iet.ru
22
SM: xt = + xt1 + t ,

DGP: xt = t .
t- (
) [Dickey, Fuller (1981)]. 5%
t t
.
n
( >
0)
2
5
2.97
2.61
5
0
2.89
2.56
1
0
0
2.86
2.54
2
5
0
2.84
2.53
5
0
0
2.83
2.52
2.83
2.52
= 0 , ,
= 0 DGP = 0.
= 0 ,
H0: = 0
SM: xt = + xt1 + t ,

DGP: xt = + t 0 .
t- = 0 ,
, H0: = 0

xt = + t 0 . , T
, ,
. . .. www.iet.ru
,
.
23
= 0 ,
, 4 H0: = 0 ,

SM: xt = xt1 + t .
t- H0: = 0
( 1). , :
H0: = 0 ;
H0: = 0 xt = t . (, xt = 1 xt1 + + p1 xt p+1 + t .)
. ,
,
,
.
. ,
.
.
1:
p 1
xt = + t + xt 1 + j xt j + t
j =1
2:
p 1
xt = + xt 1 + j xt j + t
j =1
3:
p 1
xt = + j xt j + t
j =1

1959 1985 ( , 1982 .).
. . .. www.iet.ru
24
2800
2400
2000
1600
1200
800
60 62 64 66 68 70 72 74 76 78 80 82 84
DPI
,
,
. .
1

SM: xt = + t + xt1 + 1 xt1 + + p1 xt p+1 + t .

,
SM: xt = + t + xt1 + t .
(T = 26):
xt = 461.338 + 25.857 t 0.448 xt1 + et ;
t- H0: = 0 t = 2.640. (5%)

DGP: xt = + t , 0 ,
t = 3.59 ( T = 26). t t
.
2
, DGP
DGP: xt = + t + t .

H0: = 0.
= 2.680. (5%)
(HA: > 0) 2.85 H0: = 0 .
3

SM: xt = + xt1 + t ;
DGP
DGP: xt = t .
. . .. www.iet.ru
25
H0: = 0. : xt = 47.069 + 0.00522 xt1 + et ; t t = 0.335. (5%) t = 2.98

.
4
SM: xt = + xt1 + t , DGP: xt = t , H0: = 0. : xt = 47.069 +
0.00522 xt1 + et ; t- t = 1.682 < t = 2.61 H0: = 0
.
5
SM: xt = xt1 + t , DGP: xt = t , H0: = 0. : xt = 0.03070 xt1 +
et ; t- t = 7.987 > t = 1.95 H0: = 0 .
:
xt = xt1 + t .

( , ),
:

.
H0: = 0 2. (,
t-
: = 2.680, t = 2.85.) 2
H0: = 0 , 0, H0:
= 0
SM: xt = + t + xt1 + t , DGP: xt = + t + t , 0.
t N(0, 1) ,
5% t = 1.645.
t = 2.640 (. 1),
, ,
:
xt = 461.338 + 25.857 t 0.448 xt1 + t .
6.8.
6.8.1.
, [Phillips, Perron (1988)],

xt DS H0 : = 0

SM: xt = + t + xt 1 + ut , t = 2,, T ,
, ,
. , , ut
(
),
26
. . .. www.iet.ru
() (
, E u t C < > 2). ,

, .
t-
H0 : = 0 ,
Zt ,
ut . Zt
(long-run)
ut ,

2 = lim T 1 E (u1 + ... + uT )2 .

T
u t ( )
( )
2
xt = xt 1 + + t + ut , t = 2,, T ,
[Newey, West (1987)]
( )
l
j
= 0 + 2 1
l + 1
j =1
= T 1 utut j
t = j +1
j- ut . l T ,
( )
, (l T 1 / 4 ) 0 , 2 (. [Phillips (1987)])
Zt
t .
,
l Newey West ( l
window size). , ..

( ). ,
.
, -
.
( , ,
[Phillips, Perron (1988)], [Schwert (1989)]) -
l.
[Schwert (1989)],
l = [K(T/100)1/4], [a] a, K
4 12 .
l , , EVIEWS, l =
[4(T/100)2/9] ([Newey, West (1994)]).
, l
.
Zt [Fuller (1976)]
[acKinnon(1991)].
, xt IMA(1, q),
q l Newey-West.
2
. . .. www.iet.ru
27
q = 1, xt = t + b1 t 1 , b1 > 0 -
, ,
. , b1 < 0
-
, b1 < 0 (
DS).
GNP
xt = + t + xt1 + 1 xt1 + t
(. . 6.3):
ADF Test Statistic -4.117782
1% Critical Value* -4.1219

5% Critical Value -3.4875
: t-
H0: = 0 5% ,
, 1% .
, AR c p max = 5, :
ADF Test Statistic -2.873575


Variable
Coefficient Std. Error t-Statistic
Prob.
X(-1)
D(X(-1))
D(X(-2))
D(X(-3))
D(X(-4))
C
@TREND(1947:1)
0.0060
0.0002
0.2253
0.8947
0.6936
0.0035
0.0056
-0.266169
0.546230
0.183918
-0.020254
-0.058683
59.45556
1.397409
0.092626
0.133521
0.149711
0.152201
0.148061
19.32396
0.482120
-2.873575
4.090958
1.228486
-0.133077
-0.396345
3.076779
2.898469
t = 2.873575 > 3.1744,

10% . ,
. P-
F- 0.44.
,
,
.
, .
[Newey, West (1994)] l =
[4(T/100)2/9] = 3 ;
PP Test Statistic
-2.871178
Lag truncation for Bartlett kernel: 3

( Newey-West suggests: 3 )
. . .. www.iet.ru
Residual variance with no correction

Residual variance with correction
28
29.28903
54.87482
Phillips-Perron Test Equation

Dependent Variable: D(GNP)
Variable
Coefficient
Std. Error t-Statistic
Prob.
GNP(-1)
C
@TREND(1947:1)
-0.153024
38.33211
0.806326
0.072723 -2.104212 0.0399

15.97824 2.399020 0.0198
0.378145 2.132322 0.0374
,
, [Newey, West (1994)],
,
.

ST_1, ST_2, ST_3: ST_1 ST_2
DGP: xt = xt1 + t , SM: xt = + a1 xt1 + t ;
ST_3
DGP: xt = + xt1 + t ( DGP: xt = xt1 + t) ,
SM: xt = + t + a1 xt1 + t .
DF PP(l), ,
l ,
[Newey, West (1994)].
n = 50
ST_1 DF = 2.298 > t10% = 2.60,
PP(3) = 2.394 > t10% = 2.60,

10%
ST_2 DF = 2.387 > t10% = 2.60,
PP(3) = 2.322 > t10% = 2.60,

10%
n = 100
DF = 3.238 < t5% = 2.89,
PP(4) = 3.399 < t5% = 2.89,

5%
DF = 3.217 < t5% = 2.89,
PP(4) = 3.364 < t5% = 2.89,

5%
DF = 3.207 < t10% = 3.15,
PP(4) = 3.368 < t10% = 3.15,

10%
ST_3 DF = 2.687 > t10% = 3.18,

PP(3) = 2.755 > t10% = 3.18,

10%
, DF PP,

.
. . .. www.iet.ru
29
6.8.2.
[Leybourne (1995)]
DF xt ,
, DFmax .
DFmax
T = 25, 50, 100, 200, 400 ()
.
(). ,
.

.
ST_3 100
DF = 3.207 .
3.352. , 3.207,
5% 3.45, .
5% ( )
3.15, ST_3
5% .
6.8.3. .
[Schmidt, Phillips (1992)] DS (

)
xt = + t + wt ,
wt = wt 1 +
, t = 2,..., T .
, ( = 1 1)
, .
(DS) (TS)
, .
, , l T1/2.
.

[Maddala, Kim (1998), .85].
.
ST_3 100 ,
: 3.12. , 5%
3.06.
5% .
. . .. www.iet.ru
30
6.8.4. DF-GLS
, , ,
[Elliott, Rothenberg, Stock (1996)]. DF-GLS (.
[Maddala, Kim (1998)]) a0 = 0
ydt= a0 ydt1+a1ydt1++ apydtp+error,
ydt - ( . ).
,
DF-GLS.
3.246, 5% 2.89.
5% , ,
-.
6.8.5. (KPSS)
, [Kwiatkowski, Phillips, Schmidt, Shin (1992)],

TS.
= + + .
,
, , ,
. ,
, DS .
LM
.
-, ,
- .
l Newey-West,
l.
,
(. [Schwert (1989)]).
KPSS
[Maddala, Kim (1998), .120-122].
ST_3 100 KPSS l =

3 0.157. ,
TS , ,
. , ,
: 5% 0.146, TS
DS . ,
, - DF-GLS,
TS DS , .
. . .. www.iet.ru
31
6.8.6. ( )
, [Cochrane (1998)],

V
VRk = k
V1
(VR variance ratio),
1
Vk = D( xt xt k ) .
k
xt , VRk = 1 , xt ,
( ), VRk 0 k
.

T / (T k + 1)
VRk .
VRk k = 1,, K
TS DS,
.

VRk :
k
j
VRk = 1 + 2 1
rj ,
k +1
j =1
rj j xt = xt xt1
.
ST_3 ,
,
TS DS ,
. .
1.2
1.0
0.8
0.6
0.4
0.2
5
10
15
20
25
30
35
40
VARRATIO
TS .

WALK_2 :
. . .. www.iet.ru
32
1.3
1.2
1.1
1.0
0.9
0.8
5
10
15
20
25
30
35
40
VARRATIO
, WALK_2 DS
.
6.9. , TS DS
6.9.1.

,
. , ,
,
. , (
) . ,
,
,
DS (.,
, [Ghysels, Perron (1993)]), .

- ([Davidson, MacKinnon (1993)]).

(dummy) D1,,
D12 ( ) D1,, D4 ( ).
,
.
- [Dickey, Bell, Miller (1986)], ,
t
.
6.9.2.
, ,
, , ,
. ,
, , ,
. ,
, [Shiller, Perron (1985)] [Perron (1989b)].
. . .. www.iet.ru
6.9.3.
TS DS
33
TS DS
,
TS, DS, :
H0: TS
H0: TS
H0: DS
1
2
H0: DS
3
4
:
2 DS .
3 TS .
1 -
.
4 (DGP)
TS DS .
6.9.4.
[Fuller (1976)] [Dickey, Fuller (1981)]

,
.
,
.
,

, .
,
, 1.

.

2.
,
,
. . .. www.iet.ru
34
.
: ,
, .
,
, .
, ,
, > 1, [Dickey,
Pantula (1987)]
,
. ,
;
1 ...
AR(2)
a(L) xt = t ,
..
(1 a1L a2L2) xt = t ,
(1 aL)( 1 bL) xt = t ,
a = 1/z1 , b = 1/z2 , z1, z2 a(z) = 0.

, , | z1|, |z2|
1, , | a|, |b| 1.
, xt , ,
:
xt = (a + b) xt 1 abxt 2 + t .
xt 1 :
xt = (a + b 1) xt 1 abxt 2 + t .
xt 1 :
2xt = xt xt 1 = xt 1 + (a + b 1) xt 1 abxt 2 + t =
= (a + b 2) xt 1 + (1 ab) xt 2 + t .
:
2xt = (a + b 2) xt 1 + [ (ab 1) xt 2 + (ab 1) xt 1] (ab 1) xt 1 + t
= (a + b ab 1) xt 1 + (ab 1) xt 1+ t ,

2xt = (a 1)(1 b) xt 1 + (ab 1) xt 1+ t .
, 2
, 1 .
,
a = b = 1 ( ), 2xt = t .
a = 1, |b| < 1 ( ),
2
xt = (b 1) xt 1+ t ,
2xt = xt 1+ t < 0 .
|a| < 1 |b| < 1 ( ),

2
xt = xt 1 + xt 1+ t < 0 < 0.
, , , .
. . .. www.iet.ru
35
,

2xt = + xt 1 + ut
t-
( 1 2, ,
= 0 0). ut
2xt 1 , ...
, 2xt p + 1 .
( = 0) ,

2xt = xt 1 + xt 1+ ut
= 0 < 0.
, xt ,
xt ~ I(1).
AR(2)
:
4
0
-4
-2
-8
-4
-12
-6
-8
10
20
30
40
50
60
ROOT0
70
80
90
100
-16
10
20
30
40
50
60
70
80
90
100
ROOT1
. . .. www.iet.ru
36
2
0
-2
-4
-6
-8
10
20
30
40
50
60
70
80
90
100
ROOT2
, .

SM: 2xt = + xt 1+ t
= 0 < 0. (

.)
ROOT2 ,
2 ( 0),
. T = 100 5%
2.89. t- 1.64;
. ROOT0 ROOT1
, 1 (a = 0),
.
T = 100 5% 1.95.
t- 7.83 ROOT0 5.50
ROOT1; .
. . .. www.iet.ru
37
ROOT0
ROOT1.
SM: 2xt = xt 1 + xt 1+ t

= 0
< 0.
t- 3.89 ROOT0 1.63

ROOT1, = 0 ROOT0
ROOT1.
, DGP ROOT2
, DGP ROOT1 ,
DGP ROOT0 :
DGP ROOT0: xt = 1.1 xt 1 0.3 xt 2 + t ,
(1 0.6L)(1 0.5L) xt = t,
DGP ROOT1:
xt = 1.5 xt 1 0.5 xt 2 + t ,
(1 L)(1 0.5L) xt = t ;
DGP ROOT2: xt = 2 xt 1 xt 2 + t ,
(1 L)2 xt = t .
c , ,
, , ,
[Patterson (2000)].
. . .. www.iet.ru
38
6.10.
6.10.1.
[Perron (1989a)]
DS ,

TB ,
, .
, DS-
,
, , .. DS (., , [Engle, Granger (1991)]).

[Nelson, Plosser (1982)], 13

( 62 111 ) GNP,
(1948 1987 ..). ,
.

SM: xt = + t + xt1 + ut , ut AR(k),
DGP: xt = + ut (c = 0 0),
.
, 13 14
. .

, , ,
DS . , ,
,
14 .
, ,
[Perron (1989a)]. , , GNP 1
1958 . 4 1979 .
3500
3000
2500
2000
1500
1000
58
60
62
64
66
68
70
72
74
76
78
GNP
( )
,
. ,
1974 . 1974
.
. . .. www.iet.ru
39
1958:1 1979:4,
19.086.
1958:1 1973:4 19.852. ,
1975:1 1979:4
31.995.
,
. ,
, 1975 .

, [Perron (1989a)] ,
.
,
,
,
, .

,
, .
A. :
xt = c + DMU t + t + d DTBt + xt 1 + t
B. :
xt = c + DMU t + t + DTS t + xt 1 + t
C. , :
xt = c + DMU t + t + DTt + d DTBt + xt 1 + t
c ,
1 t = TB + 1
DTBt =
;
0
1 t > TB
DMU t =
0 t TB
t TB t > TB
DTS t =
0 t TB
t t > TB
DTt =
.
0 t TB

:
A.
= 1, = = 0, d 0.
B.
= 1, = = 0, 0.
. . .. www.iet.ru
40
C.
= 1, = = 0, d 0, 0.

A.
< 1, 0, 0, d = 0
B.
< 1, 0, 0, = 0
C.
< 1, 0, 0, d = 0, 0 .

.
t-
, =T/TB ,
(AO),
, (IO),
. [Perron (1989a)]
2.
, [Perron,
Vogelsang (1993)].
, ,
.
,
, ([Perron
(1989a)], ,
.
zt
,
zt = a1 zt 1 + t ,
yt
yt = f(t) + zt ,
f(t) = 0 t TB f(t) = 0 t > TB . E(zt) = 0, E(yt) = 0 t
TB E(yt) = t > TB . , TB yt
(
).
,
AR(1) yt . ,
yt = f(t) + a1 yt 1 + t , a1 < 1,
f(t) = 0 t TB f(t) = (1 a1) t > TB , 0 .
,
.
. . .. www.iet.ru
41
TB yt .
yt TB ?
:
yt = a1 yt 1 + (f(t) + t) = a1 yt 1 + t .
t = TB + h
yTB + h = y
TB + h
0
TB + h 1
a
k
1
k =0
= y0TB + h +
TB + h 1
t k
a ( f (t k ) + ) =
k
1
t k
k =0
TB + h 1
h1
= y0TB + h + a1k t k `+ a1k (1 a1 ) .
k =0
k =0
( )
E(yt) = 0. h
h 1
lim a1k (1 a1 ) = .
h
k =0
t = TB yt
, .
f(t) t
( ),
. ,
.
, ( ),

.
, .
:
16
12
8
4
0
-4
10
20
30
40
Y1_ADD
50
60
70
80
90
100
Y1_INNOV
. . .. www.iet.ru
42
100
80
60
40
20
0
10
20
30
40
50
Y2_ADD
60
70
80
90
100
90
100
Y2_INNOV
:
140
120
100
80
60
40
20
0
10
20
30
40
50
Y3_ADD
60
70
80
Y3_INNOV
:
120
100
80
60
40
20
0
10
20
30
40
50
60
70
80
90
100
Y4_ADD
[Perron (1989a)].
- 14
,
.
11 14 , ..
-.
,
.

. . .. www.iet.ru
43
, xt = M1, 1 ,
,
.
1995:06 2000:07 . Xt = M1
:
700000
600000
500000
400000
300000
200000
100000
1996
1997
1998
1999
2000
M1

(. [
(2001)]) ,
.
,
1998 1999 ., - 1998
.
,
1998 .,
,
() (AO
).
, TB ,

xt = + t + DTSt + ut ,
DTSt t TB t > TB 0
t . et.
et et 1 e t 1 ,, et p :
et = et 1 +
et j + t ;
j =1
t- H0: = 1
, [Perron,
Vogelsgang (1993), . 249]).
,
.
. . .. www.iet.ru
44
TB = 42, 1998:08.
12 , ..
( 11 ) P- - ( AR(1)
), 0.0002 .
, GS ( )
SIC, ,
( 10%
) .

SIC
P-val
P-val P-val
LM-. White J-B
( 12
)
8
11
10
9*
22.236
1 0.983
2 0.967
22.157
22.089
22.018
21.986
4
5
3
1** ( GS)
7
6
2 ( SIC)
21.974
21.935
21.898
21.837
21.834
21.793
21.782
1 0.590
2 0.844
3 0.954
0.701
0.372
0.040
0.035
0.016
0.006
0.002
0.281
t-
-1.92
0.223
-2.27
-2.60
-2.90
-3.27
0.518
0.184
0.008
0.006
-2.78
-2.59
-2.22
-2.04
-1.37
-1.31
-0.92
,
.
,
10% .
(SIC),
.
P-
(P-values)
LM-
-. , P, ( )
.
P- (White)
.
P- Jarque-Bera
.
t- ()
- ,
( ) .
( 10% )
12
, 8, 11, 10, 9 ().
, , 1 7 12 ;
,
. ,
, 10% ,
, , .
. . .. www.iet.ru
45
,
.

, ,
, .

= TB/T, TB
, , T
. = 42/62 = 0.667. 5%
( )
3.94 ( = 0.6) 3.89 ( =0.7).

.
, 1998:08
1998 .
1. ,
t- .
6.10.2.
14 ,

. [Zivot, Andrews (1992)]
, GNP
1973 . (
- ).
,
().
, (1964 .), ,
80- . ,
1973 ., GNP. ,
,
( ) ,
. ,
, [Perron (1989a)], ,
, .
, Zivot Andrews
( ),
,

t- t H0: = 1;
(tmin), t .
?
, , , (1890 1970).
[Zivot, Andrews (1992)] (A) (. A, B,
C), DTBt .
. . .. www.iet.ru
46
, GNP, 1929 .
( ). t t =
4.95, TB = 40, = 40/81= 0.49. (5%)
t 3.76, .
, , Zivot Andrews
(1929 .), tmin = 4.95. t- .
tmin t
: 5% tmin 5.26.
tmin = 4.95 > 5.26, (H0: = 1)
.

. 11 14
. GNP ( )
(1986 1970).
( ):
,
.
, , , GNP,

(kurtosis): 5.68, 4.658, 4.324,
,
tmin . (
.
3.3)
( )
5%
: 5.86, 5.81 5.86 ( 5.38, 5.33 5.63,
). tmin 5.82, 5.30
5.61, , , .
, ,
, D(t) = , tmin
, 5%
.
[Perron
(1997)] , Zivot, Andrews,
tmin ,
.
(A) (C), ( Zivot, Andrews)
DTBt .
3

= ( 3), 0 .
- , ,
ECONOMETRIC VIEWS, ()
.
. . .. www.iet.ru
47
, [Perron (1997)],
PERRON97 RATS.
IO1
, IO2 ,
, AO , .
:
UR t- = 1;
STUDABS t-
,
( IO1) ( IO2);
STUD t-
, ( IO1)
( IO2);

,
.
( 1)
,
, PERRON97
RATS, ,
[Perron (1997)].
,
,
, .
,
(IO).
PERRON97 :
break date TB = 1999:07; statistic t(alpha=1) = -3.34124
critical values at 1%
5%
10%
for 70 obs.
-6.32
-5.59
-5.29
number of lag retained : 12
explained variable : M1
coefficient
student
CONSTANT
124786.79561
3.33345
DU
-2506239.31872
-3.77751
D(Tb)
40455.79442
2.72347
TIME
9769.03708
3.44839
DT
23866.02686
3.78217
M1{1}
-0.91050
-1.59235
DUt =1 t>TB DUt = 0 t ;

D(Tb)t =1 t=TB+1 D(Tb)t = 0 t ;
DT= t t>TB DTt=0 t ;
(M1{1})t =M1t1.
(,

6
. . .. www.iet.ru
48
: CONST, DU, D(Tb), TIME, DT

M1{1}.)
PERRON97 1999:07,
t-
t=1, . t=1 = 3.341,
5% 5.59, .
, ,
12 GS 10%
.
t d DTt,
, 1998:04. t=1 = 0.547, 5%
5.33; . (
11). ,
t-
DT,
, 1998:04
(UR-).
, (
) (AO).
PERRON97 :
break date TB = 1999:02; statistic t(alpha=1) = -3.59417
critical values at 1%
5%
for 100 obs.
-5.45
-4.83
number of lag retained : 12
explained variable : M1
coefficient
student
CONSTANT
104939.65455
20.48279
TIME
4832.56930
26.73200
DT
14335.07564
21.11189
M1 {1}
-0.75752
-1.54915
10%
-4.48
(,
.

CONST, TIME, DT;
et . et et1
et 1,, et p ).
t=1
1 et 1 .
1999:02, t=1= 3.594 ( 12
), 5% 4.83, UR-
.
, ,
:
1.626
, 3. [Zivot, Andrews (1992)] (
),
,
, UR- .
. . .. www.iet.ru
49
1 1995:06 2000:07,
[ (2001)].
.
()
- ()
-
DF-GLS
KPSS

( )

( )
()
DS
TS

DS

,
, : DS- ,
TS- ;
DS-.
. . .. www.iet.ru
7.

7.1.
(, spurious)
.
,
[ (2000)].
1957 1966
: E -
( -), C - ( .
) H -
( ). :

34.9
35.9
37.9
41.1
43.5
46.7
48.9
52.0
56.1
62.6
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
. .
-
( )
716
478
724
478
797
478
844
481
881
483
946
493
1011
520
1083
528
1157
528
1249
534
:
1400
1200

1000
800
600
400
200
0
. . .. www.iet.ru
, ,

. , ,
E = 2625.5 + 7.131H ,
R 2 = 0.900;
C = 129.30 + 0.350 H ,
R 2 = 0.871; .
E = 23.90 + 19.950C ,
R 2 = 0.993;
C = 0.860 + 0.0498E ,
R 2 = 0.993.
(, , ,
,
, 19.950 0.0498 = 0.993
R 2 .)
,
, ,

, . ,
,

, 0.9
.
, ,
(, - spurious)
. ,
(
) -
.
, , ,
R 2
y x :
R 2 = ryx2 .
,

y x .

Cov ( y , x )
ryx =
,
Var ( y ) Var ( x )
Cov ( y, x) =
Var ( x) =
1
n 1
( yi y )(xi x ) ,
i =1
1
n 1
(x
i =1
Var ( y ) =
1
n 1
(y
i =1
y) ,
2
x) .
2
Var ( x ) Var ( y) , ryx 1 ,

Cov ( y, x ) > 0 .
. . .. www.iet.ru
yi y xi x
y
x , ,
. (

, ,
. , ,
, , , ,
.)
, ryx 1,
Cov ( y, x ) < 0 .
yi y
xi x

y x , ,
, . ( ,
, , , ,
, .)
,

,
.

1t 2t ,
, N(0, 1).

3
-1
-1
-2
-2
-3
-3
5
10
15
20
25
30
EPS_1
35
40
45
50
10
15
20
25
30
35
40
45
50
EPS_2
DGP
DGP : xt = 1 + 0.2 t + 1t ,
yt = 2 + 0.4 t + 2t ,

SM: yt = + xt + t
.
xt yt
. . .. www.iet.ru
25
20
15
10
5
0
5
10
15
20
25
X
30
35
40
45
50
.
:
Sample: 1 50
Variable
t-Statistic
Prob.
C
X
1.553866
1.800255
2.265868
17.47878
0.0280
0.0000
R-squared
Adjusted R-squared
S.E. of regression
Sum squared resid
Log likelihood
Durbin-Watson stat
0.864218
0.861389
2.206028
233.5948
-109.4860
2.150060
0.685771
0.102997
Mean dependent var

S.D. dependent var
Schwarz criterion
F-statistic
Prob(F-statistic)
12.22809
5.925326
4.459442
4.535923
305.5076
0.000000
,
,
.
.
:
Variable
t-Statistic
Prob.
C
T
X
2.037450 0.294861
0.412232 0.028055
-0.054186 0.133658
6.909879
14.69394
-0.405410
0.0000
0.0000
0.6870
R-squared
Adjusted R-squared
S.E. of regression
Sum squared resid
Log likelihood
Durbin-Watson stat
0.975727
0.974694
0.942598
41.75908
-66.44428
2.249075
Mean dependent var

S.D. dependent var
Schwarz criterion
F-statistic
Prob(F-statistic)
12.22809
5.925326
2.777771
2.892492
944.6386
0.000000
,
.
, xt ,
. . .. www.iet.ru
- ,
yt .
xt
Variable
t-Statistic
Prob.
C
T
1.990020
0.401493
7.417403
43.84727
0.0000
0.0000
R-squared
Adjusted R-squared
S.E. of regression
Sum squared resid
Log likelihood
Durbin-Watson stat
0.975642
0.975134
0.934357
41.90511
-66.53155
2.249658
0.268291
0.009157
Mean dependent var

S.D. dependent var
Schwarz criterion
F-statistic
Prob(F-statistic)
12.22809
5.925326
2.741262
2.817743
1922.583
0.000000

. ,
yt = + xt + t , ,
SM 233.59 ,
41.91.
, yt
xt .

, DGP
.
()
.
, ,
. .

DGP: xt = xt 1 + 1t ,
yt = yt 1 + 2t ,
1t 2t , .
10
5
0
-5
-10
-15
-20
10
20
30
40
50
X
60
70
80
90
100
. . .. www.iet.ru
, ,
50 ( 51 100).
SM: yt = + xt + t
:
Sample: 51 100
Variable
t-Statistic
Prob.
C
X
8.616496
0.597513
11.51512
7.707873
0.0000
0.0000
R-squared
Adjusted R-squared
S.E. of regression
Sum squared resid
Log likelihood
Durbin-Watson stat
0.553120
0.543810
2.265356
246.3283
-110.8130
0.213611
0.748277
0.077520
Mean dependent var

S.D. dependent var
Schwarz criterion
F-statistic
Prob(F-statistic)
3.404232
3.354003
4.512519
4.589000
59.41131
0.000000
, DGP yt xt
,
0.553. , ,
:
10
5
0
-5
-10
-15
-20
55
60
65
70
75
X
80
85
90
95
100
, , 100 ,
:
Sample: 1 100
Variable
t-Statistic
Prob.
C
X
1.490206
0.055097
2.242470
0.656086
0.0272
0.5133
R-squared
Adjusted R-squared
S.E. of regression
Sum squared resid
Log likelihood
Durbin-Watson stat
0.004373
-0.005786
3.523613
1216.753
-266.8324
0.061638
0.664538
0.083978
Mean dependent var

S.D. dependent var
Schwarz criterion
F-statistic
Prob(F-statistic)
1.120548
3.513463
5.376648
5.428752
0.430449
0.513306
. . .. www.iet.ru
,
xt 0.0551 0.5975,
51 100.
DGP , ,

xt , .

(0.214 0.062 ).
,
,
( !)
.
.
DGP: xt = xt 1 + 1t , yt = yt 1 + 2t , x0 = 0, y0 = 0 , 1t 2t
, 1t ~ N(0, 12), 2t ~ N(0, 22), Cov(xt , yt) =

0. , T (xt , yt), t = 1, 2, , T ,

SM: yt = xt + ut , ut ~ i.i.d. N(0, u2), Cov(xt , ut) = 0.

T = yt xt xt2 .
t =1
t =1
DGP , T
T ,
.
, SM ,
( DGP !) :
Cov(xt , yt) = Cov(xt , xt + ut) = Cov(xt , xt ) = D(xt),
..
= Cov(xt , yt ) / D(xt).
( DGP) Cov(xt , yt ) = 0, = 0,
( SM) ,
0 .
T
, T t- t H0: = 0
, t , ..
, xt yt
. ,
t , 1 / T t ,
.
(DW), T
DW 0 ,
. . .. www.iet.ru

.
,
.

DGP: xt = xt 1 + 1t , yt = yt 1 + 2t ,
1t 2t ,
, N(0, 1),

SM: yt = + xt + t .
51 100
:
= 0.598, t = 7.708, DW = 0.214.
T

4
2
0
-2
-4
-6
55
60
65
70
75
80
85
90
95
100
RESIDS
.

,
, ,
(
, .).
, , , , ,
(
).
,
.
, ,
,
. ,
.
()
SM: yt = + xt + t 51
100
yt = 8.616 +0.598 xt + e t .
. . .. www.iet.ru
,
, ..
e t = e t 1 + t
t- t H0: = 0,
.
H0 HA: < 0 (
), t < t . t

t + T 1 + T 2 ,
, 1 , 2
[MacKinnon (1991)]. 5%
t 3.3377 + 5.967 T 1 8.98 T 2 ,
T = 50 t = 3.46 . 5%
2.92,
.
t
= 2.01. 5% ,
.
, () ()
xt yt ,
(xt , yt ~ I(1)),
. , yt
xt ?
, xt yt (
xt yt ) . , xt
yt ( xt yt ) .

yt = + xt + ut
, .. T , ,
- xt yt . (.,
, . 7.3.)
, xt yt ,
T
, yt xt
,
DGP: xt = xt 1 + 1t ,
yt = yt 1 + 2t
1t 2t , ..
Cov(1t , 2t) 0, 1t 2t , xt yt
. , ,
yt = xt + ut , ut . yt 1 xt 1 = ut 1 , yt
= xt + ut , 2t = 1t + ut . ut = ut 1 + t ,
t = 2t 1t ~ i.i.d. N(0, 2).

, ut .
. . .. www.iet.ru
10
, x0 = y0 = 0,
Cov(xt , yt ) = Cov(11 + + 1t , 21 + + 2t) = t Cov(11, 21) ,
xt yt , .
,
= (Cov(k 1, s 1)) , k, s = 1, 2.
,
(CRDW cointegrating regression DW):
et = yt T T xt .
T = 50 5% 0.78.
,
.
0.214,
.

1 xt yt ?
:
1.

, ,
SM: yt = + xt + yt 1 + xt 1 + ut ,
ut xt .
:
(a) yt = + yt 1 + xt + ( + ) xt 1 + ut ,
() yt = + yt 1 + ( + ) xt xt + ut .
yt ~ I(1).
()
xt , ; yt 1 , xt 1 ~ I(1), ut
. [Sims, Stock, Watson (1990)],
SM
, . t H0: = 0
N(0,1), ut . , ()
xt ,
; yt 1 , xt ~ I(1), ut .
SM , t H0: = 0
N(0,1), ut .
ut ,
. N(0,1)
. . .. www.iet.ru
11
t-,
ut
, . 6.8.1.
, qF = 2F H0: = = 0
2(2) , SM
,

, . (
.)
2. xt yt ,
..
SM: yt = + xt + ut , ut .

. t-
N(0,1), ut . ut ,
, t-, .
3.
SM: yt = + xt + ut , ut = xt 1 + t , ut ~ i.i.d. N(0, 2).
, , ,

yt yt 1 = (1 ) + (xt xt 1) + t .
1 ( ), , ,

T .
,

DGP: xt = xt 1 + 1t , yt = yt 1 + 2t ,
1t 2t ,
, N(0,1).
50 .
(1)
,
Variable
t-Statistic
Prob.
C
X
Y(-1)
X(-1)
0.812691
-0.116725
17.32664
0.279290
0.4206
0.9076
0.0000
0.7813
0.447271
-0.014458
0.970105
0.033532
0.550358
0.123861
0.055989
0.120061
P- xt xt 1
, xt yt .
(2)
,
. . .. www.iet.ru
Variable
t-Statistic
Prob.
C
D(X)
0.184523 0.117614
-0.033386 0.116361
1.568884
-0.286915
0.1232
0.7754
12
,
1t 2t .
(3)
,
yt yt 1 = (1 ) + (xt xt 1) + t ,
..
yt = * + yt 1 + (xt xt 1) + t .
:
Y=C(1)+C(2)*Y(-1)+C(3)*(X-C(2)*X(-1))
t-Statistic
Prob.
C(1)
C(2)
C(3)
0.217398 0.159423
0.988791 0.035801
-0.027306 0.119276
1.363650
27.61920
-0.228934
0.1792
0.0000
0.8199
R-squared
0.940380
Mean dependent var
3.404232
, yt 1 1,
.
P- 0.367.
, xt yt , ..
xt = xt 1 + 1t , yt = yt 1 + 2t ,

1t ,
, N(0, 1.25),
2t ,
, N(0, 1.25),
Cov(1t, 2s) = 0 t s , Cov(1t, 2t) = 1.
, , , Corr(1t, 2t) = 0.8 .
1t 2t
. . .. www.iet.ru
13
3
2
1
0
-1
-2
-3
10
20
30
40
50
60
NOISE_X
70
80
90
100
NOISE_Y

1t
2t
;
0.789. xt yt ,
.
1t 2t .
10
10
-5
-5
-10
-10
-15
-15
-20
-20
10
20
30
40
50
X
60
70
80
90
100
10
20
30
40
50
60
70
80
90
100
,
;
1t 2t 0.792.
yt = + xt + ut .
, 2.112, 5%
3.46. ,
.
(1)
Variable
t-Statistic
Prob.
C
X
Y(-1)
X(-1)
0.548392
0.718479
0.913556
-0.641522
1.454312
8.987361
13.47210
-7.223976
0.1526
0.0000
0.0000
0.0000
R-squared
0.987574
0.377080
0.079943
0.067811
0.088805
Mean dependent var
-0.957402
, 1t 2t
,
. . .. www.iet.ru
14
xt xt 1. ,
:
Variable
t-Statistic
Prob.
X
Y(-1)
X(-1)
0.695862 0.079341
1.005257 0.025245
-0.707002 0.077447
8.770553
39.82053
-9.128837
0.0000
0.0000
0.0000
R-squared
0.987003
Mean dependent var
-0.957402
.. yt = 1.005 yt 1 + 0.695 xt 0.707 xt 1 + et . yt 1

1; xt xt 1
, DGP.
(2)

Variable
t-Statistic
Prob.
C
D(X)
0.100915
0.694000
1.203508
8.981039
0.2347
0.0000
R-squared
0.626921
0.083851
0.077274
Mean dependent var
0.226857
, DGP,
xt ,
1t 2t , .. xt yt .
, :
Variable
t-Statistic
Prob.
D(X)
9.271155
0.0000
0.709553
0.076533
.. yt = 0.710 xt + et , yt = yt 1 + 0.710 xt 0.710 xt 1 + et .

(3)
, ,
yt = + yt 1 + (xt xt 1) + t ;
:
*
Y=C(1)+C(2)*Y(-1)+C(3)*(X-C(2)*X(-1))
t-Statistic
Prob.
C(1)
C(2)
C(3)
0.329205
0.941984
0.723593
1.314726
16.54170
9.119982
0.1950
0.0000
0.0000
R-squared
0.987410
0.250398
0.056946
0.079341
Mean dependent var
-0.957402
.
:
Y=C(2)*Y(-1)+C(3)*(X-C(2)*X(-1))
. . .. www.iet.ru
C(2)
C(3)
t-Statistic
Prob.
1.014411
0.702102
48.88608
8.970448
0.0000
0.0000
0.020750
0.078268
15
.. yt = 1.014 yt 1 + 0.702 (xt 1.014 xt 1) + e t ,

yt = 1.014 yt 1 + 0.702 xt 0.712 xt 1 + e t
, :
yt = 1.005 yt 1 + 0.695 xt 0.707 xt 1 + et ( 1),
yt =
yt 1 + 0.710 xt 0.710 xt 1 + et ( 2),
yt = 1.014 yt 1 + 0.702 xt 0.712 xt 1 + e t ( 3).
,
:
yt = 0.7 xt + et .

( ):
1.985; P- Jarque Bera 0.344.
, ,
,
. , ,
,
( ) ?
:
(a)
MA
.
(b) 1
,
.
,
, .
7.2. .
yt ~ I(1), xt ~ I(0). yt xt , ..
a b
yt a b xt ~ I(1).
, , yt ~ I(0), xt ~ I(1). a b 0
yt a b xt ~ I(1),
b = 0
yt a b xt ~ I(0),
.
. . .. www.iet.ru
16
yt ~ I(1), xt ~ I(1) .
b
yt b xt ~ I(1),
yt xt , ,
.
, b 0
yt b xt ~ I(0) .
, yt xt , (1,
b)T .
, yt ~ I(1), xt ~ I(1) (
), ()
= (1, 2)T 0 ,
1 xt + 2 yt ~ I(0) .
, = (1, 2)T
xt yt , =
(1, 2)T , 0 . -
, , ,
(1, b)T ( ( a, 1)T ).
, xt , yt ~ I(1), xt , yt
. ,
(xt , yt)T ,
(xt , yt)T = + B(L) t ,
= ( 1, 2 )T , 1 = E(xt ) , 2 = E(yt) ;
t = (1t , 2t )T ,
..
1, 2 , ,
,
E( t) = (0, 0)T , D(1t) = 12 , D(2t) = 22 , Cov(1t , 2t ) = 12 ;
1 0 b11( k ) b12( k ) k
L .
+ ( k )
B( L) =
(k )
0
1
b
b
k = 1 21
22
([Granger (1983)], . [Engle, Granger (1987)])
, I(1) xt yt ( )
(I)
(xt , yt)T = + B(L) t B(1) 1.
(II) xt yt ARMA
A(L) (xt, yt )T = c + d(L) t ,

t , (I),
c = (c1, c2)T , c1 c2 ,
A(L) ,
d(L) ,
. . .. www.iet.ru
17
A(0) = I2 ( 22),
rank A(1) = 1 ( 22- A(1) 1),
d(1) .
, (22)- A(1) ,

(reduced rank VAR).
(II)
q
p
(
)
x
c
a
x
b
y
=
+
+
+
k 1,t k ,
1
1j t j
1j t j
t
k =0
j =1
q
p
y = c + (a x + b y ) +
k 2,t k
2
2j t j
2j t j
t
k =0
j =1
p q .
AR ,
d(L) 1 , p < .
(III)
xt yt
(error correction model ECM)
j =1
k =0
j =1
k =0
xt = 1 + 1 z t 1 + ( 1 j xt j + 1 j yt j ) + k 1,t k ,
yt = 2 + 2 z t 1 + ( 2 j xt j + 2 j yt j ) + k 2,t k ,
zt = yt xt E(yt xt)
,
zt ~ I(0),
12 + 22 > 0.
(II) AR(p) (p < ), ECM
p 1
xt = 1 + 1 z t 1 + ( 1 j xt j + 1 j yt j ) + 1,t ,
j =1
p 1
yt = 2 + 2 z t 1 + ( 2 j xt j + 2 j yt j ) + 2,t ,
j =1
xt , yt ~ I(1) , ECM
.
(xt , yt)T ~ I(1) ( (xt , yt)T

) ECM , xt yt .
(, ECM, zt1,
; zt 1.)
. . .. www.iet.ru
18
xt , yt ~ I(1) , VAR
. ( , xt yt
.)
zt = yt xt , = E(yt xt),
,
t
,
yt xt = 0. xt yt
zt 1 .
zt , ,
, ,
.
1
xt yt ,
xt
yt . yt
xt ,
yt xt
. (
.)
xt , yt ~ I(1) ,
, , .
ECM, 12 + 22 > 0. xt
1 zt 1 yt (.. xt
yt), 2 0. yt 1
zt 1 xt (.. yt
xt), 1 0.
2
xt , yt ~ I(1) wt ~ I(0). k
xt yt k + wt , 0. , xt ~ I(1), xt
xt k . (, xt xt k = xt + xt 1 + + xt k I(0), I(0)-.)
, xt , yt ~ I(1)
() yt = + xt ;
ECM,
.
, , , ECM
( ,
). , .
, .
[Engle, Granger (1987)] ,
yt
xt
yt = + xt + ut .
(K-),

. . .. www.iet.ru
19
zt = yt xt
.
, , (
!)
p 1
xt = 1 + 1 zt 1 + ( 1 j xt j + 1 j yt j ) + t ,
j =1
p 1
yt = 2 + 2 zt 1 + ( 2 j xt j + 2 j yt j ) + wt ,
j =1
(.. VAR(p) xt , yt).

,
( )
(1, )T . (
.) , , ,
ECM, z t 1 ,
,
, z t 1 . (
.) -
.
, ,
zt , .
,
(, 1)T , xt
yt , , ,
.
, ( ) ,
,
. ,
, ,
.
,

.
,
(, ).

DGP: xt = xt 1 + t , yt = 2 xt + t ,
x1 = 0, t t
, ,
. . .. www.iet.ru
20
N(0, 1). xt yt

10
-10
-20
-30
-40
10
20
30
40
50
60
70
80
90
100
(xt , yt)
xt = xt 1 + t ,
yt = 2 xt 1 + t ,
t = t + 2t ~ i.i.d. N(0, 5).
ECM
xt = t ,
yt = (yt 1 2 xt 1) + t = zt + t ,
zt = yt 2 xt ,
xt = 1 zt 1 + t ,
yt = 2 zt 1 + t ,
1 = 0, 2 = 1, 12 + 22 > 0.
, ,
, VAR DGP. ,
ECM
xt = 1 zt 1 + 11xt 1 + 11yt 1 + vt ,
yt = 2 zt 1 + 21xt 1 + 21yt 1 + wt ,
,
(p = 2). 100 .
(I ) yt = + xt + ut .
:
Variable
t-Statistic
Prob.
C
X
-0.006764 0.165007
1.983373 0.020852
-0.040992
95.11654
0.9674
0.0000
R-squared
0.989284
Durbin-Watson stat
2.217786
..
yt = 0.006764 + 1.983373 xt + ut ,

zt = ut = yt + 0.006764 1.983373 xt .
. . .. www.iet.ru
21
, VAR 2,
yt xt
:
,
zt = zt 1 + 1 zt 1 + t .
:
Dependent Variable: D(Z)
Variable
t-Statistic
Prob.
Z(-1)
D(Z(-1))
-1.153515 0.151497
0.038156 0.100190
-7.614088
0.380837
0.0000
0.7042
t = 7.614 5%
3.396 (. [Patterson (2000), 8.7]).
. (
, ,
. t
= 11.423,
.)
, yt xt ,
.
( II) xt :
Variable
t-Statistic
Prob.
C
Z(-1)
D(X(-1))
D(Y(-1))
-0.277810
1.420858
2.482201
-2.218019
0.7818
0.1587
0.0148
0.0290
-0.028016
0.250942
0.639967
-0.258740
0.100847
0.176613
0.257823
0.116654

P-

Variable
t-Statistic
Prob.
D(X(-1))
1.148554
0.2536
0.115141
0.100249
, ,
xt = t ,
xt .
yt ,
Variable
t-Statistic
Prob.
C
Z(-1)
D(X(-1))
D(Y(-1))
-0.283630
-1.727472
2.425311
-1.970459
0.7773
0.0874
0.0172
0.0517
-0.060101
-0.641060
1.313872
-0.482981
0.211899
0.371097
0.541733
0.245111
. . .. www.iet.ru
22
, :
Variable
t-Statistic
Prob.
Z(-1)
D(X(-1))
D(Y(-1))
-1.730381
2.445138
-1.983217
0.0868
0.0163
0.0502
-0.638888 0.369218
1.317763 0.538932
-0.483722 0.243908

zt 1 ,

yt
xt .
, ECM 12 + 22 > 0.
zt 1 xt ,
yt .
yt 1 ,
Variable
t-Statistic
Prob.
Z(-1)
D(X(-1))
-4.767072
1.572671
0.0000
0.1191
-1.186411 0.248876
0.331411 0.210732
xt 1 , yt =
2 zt 1 + wt , ,
Variable
t-Statistic
Prob.
Z(-1)
-5.137760
0.0000
-1.273584 0.247887
H0: 2 = 1 :
Null Hypothesis: C(1)= -1
F-statistic
Chi-square
1.218077
1.218077
Probability
Probability
0.272441
0.269738
, ECM
xt = t , yt = zt 1 + wt ,
zt 1 = yt 1 + 0.006764 1.983373 xt 1 .
zt 1 yt
yt = 0.0068 + 1.983 xt 1 + wt ,

yt = 2 xt 1 + t ,
DGP.
, , wt = yt + zt 1
4.62
( DGP 5.00), t = xt
1.04
( DGP 1.00).
ECM
xt = t , yt = zt 1 + wt ,
. . .. www.iet.ru
23
, yt
: zt 1 , ..
yt 1 ( 0.0068 + 1.983 xt 1) > 0,
yt zt 1
yt . ,
zt 1
yt .
xt zt 1
yt , .. xt yt .
, yt
xt , yt xt .
, ECM Cov(vt, wt) 0,
,
.

DGP: xt = xt 1 + t , yt = 2 xt + t ,
, .
( )
yt = xt + t ,
xt = xt 1 + t ,
(t , t)T ~ i.i.d. N2(0, ) ,

,
. (
.)
, Cov(t , t) = 0, xt
,
.
Cov(t , t) 0, xt
, .. Cov(xt , t) = Cov(xt 1 + t , t) 0.

.
,
.
N y1t , , yN t ,
1. = (1, ... , N)T ,
,
1 y1t + ... + N yN t ~ I(0) ,
, ( );
.
c = E(1 y1t + ... + N yN t),

. . .. www.iet.ru
24
1 y1t + ... + N yN t = c .
t
,
zt = 1 y1t + ... + N yN t c .
zt , , ,
,
, .. .
y1t , , yN t
N 1
zt
. , , ,
y1t = 1 + 2 y2 t + ... + N yN t + ut ,

u = y1t ( + y2 t + ... + yN t),
t
. ,
[MacKinnon (1991)] (. [Patterson (2000), A8.1]).
,
= (1, , , )
2
.

zt = ut .
, , ,
.
= (1, ... , N)T
, 1 y1t + ... + N yN t . ,
y1t , , yN t ( , ) ,
, ,
. ,
.
7.3. .
.
(1) .
. . .. www.iet.ru
25
1 y1t + ... + N yN t .
,
;
N .
(2) y1t , , yN t (, E(yk t) =

0).
(2a) (SM) .

SM: y1t = 2 y2t + ... + N yN t + ut ,

ut = y1 t ( 2 y 2 t + K + N y N t ) ,

ut = ut 1 + 1ut 1 + K + K ut K + t
H0:
= 0 H0: < 0 .
t- t
N .
. . .. www.iet.ru
26
, [Hamilton (1994), Table B.9, Case 1].

.
(2b) (SM) .

SM: y1t = + 2 y2t + ... + N yN t + ut ,

ut = y1 t ( + 2 y 2 t + K + N y N t ) ,

ut = ut 1 + 1ut 1 + K + K ut K + t
H0:
= 0 H0: < 0 .
(2a).
,
[Hamilton (1994), Table B.9, Case 2].
, [MacKinnon (1991), 1 ( no
trend)] [Patterson (2000)].
(3) y2t , , yN t , E(yk t)

0 .
(3a) .

SM: y1t = + 2 y2t + ... + N yN t + ut .
. . .. www.iet.ru
27
(2b), .
,
[Hamilton (1994), Table B.9, Case 3]. T
, [MacKinnon (1991), Table
1 ( with trend)] [Patterson (2000)].
(3b) .

SM: y1t = + t + 2 y2t + ... + N yN t + ut .
, , , (3a),
N , N + 1 .

- .
,
, y1t, y2t , , yN t .
y1t , y2t , y3t , y4t ,

DGP: y1t = y2, t + y3, t + y4, t + 1t ,
y2t = y2, t 1 + 2t ,
y3t = y3, t 1 + 3t ,
y4t = y4, t 1 + 4t ,
1t , 2t , 3t , 4t
, 1 2t , 3t , 4t 2 1t .
T = 200 .
. . .. www.iet.ru
28
60
40
20
0
-20
-40
-60
20
40
60
80
100
120
Y1
Y2
140
160
180
200
Y3
Y4
,
.
. t-,
2.18, 1.78, 0.57, 1.70, . 4 .
,
y1t , y2t , y3t , y4t AR(1) , ..
1.
.
(1)

y1t = y2, t + y3, t + y4, t ,

y1t y2, t y3, t y4, t .

8
6
4
2
0
-2
-4
-6
20
40
60
80
100
120
140
160
180
200
COINT
,
: t-
15.07. .
,
.
(2a) :
Dependent Variable: Y1
Variable
t-Statistic
Prob.
. . .. www.iet.ru
Y2
Y3
Y4
0.996084
0.992550
1.002305
0.009973
0.009578
0.012393
99.88161
103.6296
80.87922
29
0.0000
0.0000
0.0000

Dependent Variable: D(RESID_2A)
Variable
t-Statistic
Prob.
RESID_2A(-1)
-15.17178
0.0000
-1.075552 0.070892
t- 15.17, 5%
3.74 ([Hamilton (1994), Table B.9, Case 1]).
.
(2b) :
Dependent Variable: Y1
Variable
t-Statistic
Prob.
C
Y2
Y3
Y4
0.889279
81.05843
88.04048
77.22129
0.3749
0.0000
0.0000
0.0000
0.332183
1.002583
0.987369
0.999022
0.373542
0.012369
0.011215
0.012937

Dependent Variable: D(RESID_2B)
Variable
t-Statistic
Prob.
RESID_2B(-1)
-15.22764
0.0000
-1.079049 0.070861
t- 15.23 5%
, 4.11 ([Hamilton (1994), Table B.9, Case 2]).
.
(3) y1t , y*1t = y1t + 0.75t ,
y1t :
200
150
100
50
-50
20
40
60
80
Y1
100
120
140
160
180
200
Y1_STAR
. . .. www.iet.ru
30
200
150
100
50
-50
20
40
60
80
100
120
140
Y1_STAR
Y2
160
180
200
Y3
Y4
(3a) :
Dependent Variable: Y1_STAR
Variable
t-Statistic
Prob.
C
Y2
Y3
Y4
4.248195
-14.89224
35.18115
0.775323
0.0000
0.0000
0.0000
0.4391
11.49053
-1.333762
2.856952
0.072630
2.704802
0.089561
0.081207
0.093677
:
15
10
5
0
-5
-10
-15
20
40
60
80
100
120
140
160
180
200
RESID_3A
:

Variable
t-Statistic
Prob.
RESID_3A(-1)
-3.562431
0.0005
-0.119805 0.033630
t- 3.56 5% ,

4.16 ([Hamilton (1994), Table B.9, Case 3]).
.
(3b) :
Dependent Variable: Y1_STAR
Variable
t-Statistic
Prob.
0.778187
0.4374
0.304068
0.390739
. . .. www.iet.ru
@TREND
Y2
Y3
Y4
0.751890
1.008470
0.982658
1.001356
0.007507
0.026468
0.021830
0.015942
100.1621
38.10166
45.01453
62.81247
31
0.0000
0.0000
0.0000
0.0000
:
8
6
4
2
0
-2
-4
-6
-8
20
40
60
80
100
120
140
160
180
200
RESID_3B
,
:
Dependent Variable: D(RESID_3B)
Variable
t-Statistic
Prob.
RESID_3B(-1)
-15.23448
0.0000
-1.079492 0.070859
t- 15.234 5%
, 4.49 ([Hamilton (1994), Table B.9, Case 3]).
.
,
.
, . y1t , y2t , y3t , y4t
, ( ).
. y*1t , y2t , y3t ,
y4t . ,

.
,

,
.
.

.
, ,
(
). ,
.
, ,
, ,
.
,

. . .. www.iet.ru
32
, ..
, . ,
.
. . 7.1 ,

yt = + xt + u t , .. T
- xt
yt . ,
T ,
. , [Entorf (1992)],
.
DGP: xt = x + xt 1 + 1t ,
yt = y + yt 1 + 2t ,
1t 2t , x ,
y 0.
SM: yt = + xt + u t
T , T , T ,
T T y / x .

SM: yt = + xt + t + u t ,
( T ) T y , T
,
.
7.4.
N y1t , , yN t ,
1. = (1, ... , N)T ,
,
1 y1t + ... + N yN t ~ I(0) ,
( );
.
c = E(1 y1t + ... + N yN t),

1 y1t + ... + N yN t = c .
t
,
zt = 1 y1t + ... + N yN t c .
. . .. www.iet.ru
33
zt , , ,
,
, .. .
, , , I(1)
.

y1t , , yN t r , r
. , N ,
r = 1, , N 1 . (,
, r = 0. r = N
, N .)
I(1)
r- ,
.
r

,
,
, .
I(1) y1t , , yN t r
VAR(p) p
(VAR vector autoregression)
A(L) yt = + t ,
yt = (y1t , , yN t )T ,
= (1 , , N )T ,
A(L) = A0 A1L ApLp,
A0 , A1 , , Ap (N N),
A0 = IN ( ),
..
yt = + A1 yt 1 + + Ap yt p + t .
A(1) rank A(1) = r ( N = 2)
VAR ECM ( )
y1 t = 1 + 11 z1, t 1 + K + 1r z r , t 1 +
p 1
+ ( 11, j y1, t j + K + 1N , j y N , t j ) + 1 t ,
j =1
..
y N t = N + N 1 z1, t 1 + K + N r z r , t 1 +
p 1
+ ( N 1, j y1, t j + K + NN , j y N , t j ) + N t ,
j =1
z1,t , ... , zr,t
I(0) , r
(1) , ... (r) ,
(11, ... , N 1)T , , (1 r, ... , N r)T
.

. . .. www.iet.ru
34
yt = + T yt 1 + 1 yt 1 + p 1 yt p + 1 + t ,
1 , , p 1 N N , (N r)-
r . (1) , , (r)
, i j

z1, t 1 = T(1) yt 1 , , zr, t 1 = T(r) yt 1
( t 1 r
y1t , , yN t) y1t , , yN t .
VAR
, (1) , , (r)
. ,

.
I(1) .
r , 0 < r < N ,
I(1) y1t , , yN t
y r + 1, t r +1 vr + 1, t
y r + 1, t v1 t
y1 t 1

M = M + M ,
M = M + C M + M ,
y
v
y

y
N, t
rt r
N N, t
N , t vr t
( ) ,
C = (i j) r (N r) ,
vt = (v1 t , , vN t)T - ( )

E(vt) = 0,
yr + 1, t , , yN , t .
:
y1 t c11 yr + 1, t c1, N r yN, t = 1 + v1 t ,
...
yr t cr 1 yr + 1, t cr, N r yN, t = r + vr t ,

(1) = (1, 0, 0, , 0, c11, , c1, N r )T ,
(2) = (0, 1, 0, , 0, c21, , c2, N r )T ,
...
(r) = (0, 0, 0, , 1, cr 1, , cr, N r )T
. r
y1 t , , yN , t
z1, t = T(1) yt = y1 t c11 yr + 1, t c1, N r yN, t ,
zr, t = T(r) yt = yr t cr 1 yr + 1, t cr, N r yN, t .
v1 t , , vr t vr + 1, t , , vN t , yr
,
+ 1, t , , yN , t
y1 t , , yr t yr + 1, t ,
, yN , t . ci j ,
. . .. www.iet.ru
35
T (ci j ci j ) T . r = 1
y1 t = 1 + c11 y 2, t c1, N 1 yN, t + v1 t ,

(
y 2, t , , yN, t ) ,
, t-
F- (, ),
, v1 t . ,
. 7.1, v1 t
.
ci j r (N 1) (1) , , (r) :
= (1, 0, 0, , 0, c , , c
)T ,
1, N r
11
(1)
...
( r ) = (0, 0, 0, , 1, cr 1 , , cr , N r )T .
(1) , ... , ( r ) ,

z
= T y , ... , z
= T y
.
1, t 1
(1)
t 1
r, t 1
(r )
t 1
ECM
yt = + zt 1 + 1 yt 1 + p 1 yt p + 1 + t ,

z1, t 1
zt 1 = M .
z
r, t 1

, ECM.
, ( c )
, z1, t , ... , zr, t
y1t , , yN t
,
(N +1) y1t , , yN t
t , .
A(1) r , r

. . .. www.iet.ru
36
11 y1t + ... + 1N y1N + 1, N + 1 t ,
r1 y1t + ... + r N y1N + r, N + 1 t .

((N + 1)1)-
(1) = ( 11 , ... , 1N , 1, N + 1)T

..
(r) = ( r1 , ... , r N , r, N + 1)T.

.

(ECM),
, ,
.
I(1)
r ;
.
,

y1t = + 2 y2t + ... + N yN t + ut

( y1t = + 2 y2t + ... + N yN t + N + 1 t + ut). , r > 1,
(1, 2 , , N )T ( (1, 2 , , N , N +1 )T)
. . .. www.iet.ru
37
,
.
r - , r > 1
.

(1) = (1, 0, 0, , 0, c11, , c1, N r )T ,

(2) = (0, 1, 0, , 0, c21, , c2, N r )T ,
...
(r) = (0, 0, 0, , 1, cr 1, , cr, N r )T .
(
) ,

r . (
) y1t , , yN t , (1) ,
, (r) .
( )
,
, .
8.

, [Johansen (1988)].

. .
8.1, ,
r = 1, .. (
) .
N = 2. ,
,
y1t y2t ,
. ,
, ,
(
). ,
.
. . .. www.iet.ru
38

yt = xt + t ,
xt = xt 1 + t ,
t t .
, [West (1988)]:
yt = + xt + ut ,
xt ~ I(1), E(xt) = 0 ( xt
),
ut ~ I(0) ,
.

( , )T.
ut ,
t-, ,
.
t-
T T , :
(X X )
T , S
11
(X X )
T
1
22
T .
X (T2)- (1 xt)
, S2 ut , ut ~ i.i.d.,
S2 =
1 T 2
ut ,
n 2 t =1
ut = yt xt .
, ut ~ i.i.d., t-
(, N(0, 1) ) t- ( T),
S2 .
ut . h = Cov(ut, ut+h)
,
. West ,
S2
ut
(. . 6.8.1),
2 =
h =
, , 2 ,
. , ,
. . .. www.iet.ru
39
h , h = 0, 1, 2, , , , ,
. - , , , ,
ut ,
2
h = Cov(ut, ut+h).
, h h
, () h
2 h .
, ut
MA(q) q , h = 0 |h| > q ,
h h . ,
,
q
h =1
2 = 0 + 2 1
h =
h
h ,
q + 1
1 T
ut ut 1 h . (.,
T t = h +1
, [Hamilton (1994), p. 513]),
q = O(T
1/5
2 ( NeweyWest).
EVIEWS .
:
Newey-West. (, ,
,
.)
. . .. www.iet.ru
40
ut
AR(p) p ,
ut = a1 ut 1 + a2 ut 2 + + ap ut p + t ,
t D(t) =2 ,
2 = 2 (1 a1 a2 ... ap)2 .
2
2 =
(1 a
a 2 K a p )
a1 , a 2 , K , a p a1 , a2 , ... , ap ,
2 =
T
1
2 ,
T p t = p +1
t ut .
, S2 2 t-,
( )
, S .

DGP: yt = 2 xt + ut ,
xt = 1 + xt 1 + vt ,
ut = 0.4 ut 1 + 0.2 ut 2 + t AR(2) ,

t , t ,
: Cov(t , t) = 0.8 :
. . .. www.iet.ru
41
200
150
100
50
0
-50
10
20
30
40
50
60
70
80
90
100
SM: yt = + xt + ut
:
Variable
t-Statistic
Prob.
C
X
-0.398071 0.172093
2.031938 0.007241
-2.313111
280.6336
0.0228
0.0000
R-squared
Adjusted R-squared
S.E. of regression
Sum squared resid
Log likelihood
Durbin-Watson stat
0.998757
0.998745
1.071308
112.4748
-147.7718
1.080957
Mean dependent var

S.D. dependent var
Schwarz criterion
F-statistic
Prob(F-statistic)
37.39809
30.23477
2.995436
3.047539
78755.23
0.000000
,
, DGP yt
.
4
-2
-4
10
20
30
40
50
60
70
80
90
100
RESIDS
AR(2). AR(2) :
Dependent Variable: RESIDS
Variable
t-Statistic
Prob.
RESIDS(-1)
RESIDS(-2)
0.363522
0.205074
3.617344
2.042024
0.0005
0.0439
R-squared
Adjusted R-squared
S.E. of regression
Sum squared resid
0.240364
0.232451
0.941891
85.16727
0.100494
0.100427
Mean dependent var

S.D. dependent var
Schwarz criterion
-0.008547
1.075097
2.738343
2.791098
:
. . .. www.iet.ru
42
= 0.941891/ (1 0.363522 0.205074) = 2.183,

(S ) = 1.071/ 2.183 = 0.491.

t-
P-
t : -2.313111 (P-value = 0.0228)
-1.135738 (P-value = 0.2588)
t : 280.6336 (P-value = 0.0000) 137.791098 (P-value = 0.0000).

.
N = 2
I(1) y1t , , yN t .

y1t = + 2 y2t + ... +
yN t + ut
, ,
,
. , . 7.2,

.
T .

N = 2,

. . .. www.iet.ru
43
y1t = + 2 y2t + ... + N yN t + 1t

y2t = y2, t 1 + 2t ,
...
yN t = y N, t 1 + N t ,
t = (1t , 2t , ... , N t)T N- (.. 1 , 2 , ...
, ,
N-
=(ij) ), 2t , ... , N t
, 1t (
1 j = 0 j = 2, , N ). y2t , , yN t
, = (1, 2 , , N )T
.
( c , 2 2 , , N N)T {y2t , , yN t , t = 1, , T}
N- , , F-
, 2 , ... , N F, t- t-.
,
y1t = + 2 y2t + ... + N yN t + u1t
y2t = y2, t 1 + u2t ,
...
yN t = y N, t 1 + u N t ,
ut =(u1t , u2t , ... , u N t)T N-
( N- ), u2t , ... , u
, u1t
N t
, Cov(u1t, uk s) = 0 k 1 t, s.

. ( ut ,
T NT-
.)

F- t- F- t, S2 u1t
2 2 u1t .

F-
S 2 2
t- S .
. . .. www.iet.ru
44
, -
y2t , , yN t
.
[Stock, Watson (1993)] [Saikonnen (1991)]
,
(lags)
(leads)
leads and lags .) ,

y1t = + 2 y2t + ... + N yN t +
p
(2j y2, t j + ... + Nj yN, t j ) + ut .
j= p
p ( ),
2 , ... , N (,
), t- F, ut .
,
y2t , , yN t .
, ,

.
,
. . .. www.iet.ru
45
y1t 2 y2t ... N yN t ;
(a)
(b) y1t 2 y2t ... N yN t .

(,
(a)
(b),
, .)
DGP: yt = 5 + zt + ut ,
zt = zt 1 + vt ,
z1 = 0,
ut = t + 0.25 t 1 + 0.25 t + 1 + 0.1 t 2 + 0.1 t + 2 + 0.1 t ,
t , t .
ut t , t 1 , t + 1 , t 2 , t + 2 ,
.
DGP (50 ):
12
-4
10
20
30
40
50
Y
60
70
80
90
100
yt zt 50
. .
yt = + zt + t

:
Variable
t-Statistic
Prob.
C
Z
4.851262
1.088870
36.43410
22.88977
0.0000
0.0000
0.133152
0.047570
H0: = 1, t-, , Cov(t ,

zs ) 0 t , s.
-,
et , zt , et ,
. . .. www.iet.ru
46
yt = + zt + t . - Cov(et , zt - i)
i = 0, 1, 2, ; - lag.
- Cov(et , zt + i) i = 0, 1, 2, ;
- lead.
Correlations are asymptotically consistent approximations
e , Z(-i)
e , Z(+i)
. |*********
|.
|.
|.
*| .
*| .
.|.
**| .
.|.
. |*
.|.
. |*********
. |*
*| .
.|.
.|.
.|.
.|.
**| .
.|.
. |*
.|.
lag
lead
0
1
2
3
4
5
6
7
8
9
10
0.9017
-0.0217
-0.0956
0.0064
-0.0510
-0.0824
-0.0171
-0.1858
-0.0292
0.0833
0.0125
0.9017
0.0830
-0.0413
0.0341
0.0118
-0.0228
0.0150
-0.1579
-0.0272
0.0701
0.0216
-
- DGP 7- . ,

zt 7- .
Variable
t-Statistic
Prob.
C
Z
D(Z)
D(Z(-1))
D(Z(-2))
D(Z(-3))
D(Z(-4))
D(Z(-5))
D(Z(-6))
D(Z(-7))
D(Z(1))
D(Z(2))
D(Z(3))
D(Z(4))
D(Z(5))
D(Z(6))
D(Z(7))
4.987362
1.000689
1.006216
0.237875
0.089302
-0.008934
-0.002997
-0.011646
-0.010012
-0.003586
0.262537
0.116863
-0.010921
0.003903
0.021536
-0.008452
0.002945
238.9236
127.9955
76.66298
18.63643
6.971105
-0.722323
-0.241901
-0.956245
-0.839615
-0.308269
19.63226
8.744236
-0.826184
0.294017
1.627644
-0.665583
0.241376
0.0000
0.0000
0.0000
0.0000
0.0000
0.4726
0.8096
0.3423
0.4041
0.7588
0.0000
0.0000
0.4116
0.7696
0.1082
0.5079
0.8100
0.020874
0.007818
0.013125
0.012764
0.012810
0.012368
0.012391
0.012179
0.011925
0.011634
0.013373
0.013365
0.013219
0.013276
0.013232
0.012699
0.012199
: P-
0.252 ( K = 1) 0.427 (K = 2).
. . .. www.iet.ru
47
H0: = 1 t-
;
t = (1.0006891)/ 0.007818 = 0.0081,
H0: = 1 .
()
DGP , I(1)
.
DGP: yt = 5 + xt + ut ,
xt = 1 + xt 1 + vt ,
x1 = 0, ut , vt , .
DGP
120
100
80
60
40
20
0
-20
10
20
30
40
50
X
60
70
80
90
100
yt = + xt + t :
Variable
t-Statistic
Prob.
C
X
5.172117
0.997147
25.62837
289.5306
0.0000
0.0000
0.201812
0.003444
- xt
, .
,
:
Variable
t-Statistic
Prob.
C
X
D(X)
D(X(-1))
D(X(-2))
D(X(-3))
32.88423
1923.269
78.12803
18.92225
7.224562
-0.353864
0.0000
0.0000
0.0000
0.0000
0.0000
0.7245
3.224675
1.000621
1.009585
0.241440
0.093472
-0.004479
0.098061
0.000520
0.012922
0.012760
0.012938
0.012656
. . .. www.iet.ru
D(X(-4))
D(X(-5))
D(X(-6))
D(X(-7))
D(X(1))
D(X(2))
D(X(3))
D(X(4))
D(X(5))
D(X(6))
D(X(7))
0.001328
-0.007856
-0.006808
-0.001175
0.266271
0.120232
-0.007953
0.006576
0.023956
-0.007153
0.003367
0.012648
0.012272
0.011934
0.011513
0.013016
0.012948
0.012873
0.012866
0.012814
0.012294
0.011821
0.104986
-0.640121
-0.570501
-0.102067
20.45699
9.286042
-0.617748
0.511164
1.869580
-0.581813
0.284813
48
0.9167
0.5242
0.5702
0.9190
0.0000
0.0000
0.5388
0.6109
0.0658
0.5626
0.7767
,
H0: = 1 t-
;
t = (1.0006211)/ 0.00520 = 1.194,
H0: = 1 .
. :
Variable
t-Statistic
Prob.
C
@TREND
X
D(X)
D(X(-1))
D(X(-2))
D(X(-3))
D(X(-4))
D(X(-5))
D(X(-6))
D(X(-7))
D(X(1))
D(X(2))
D(X(3))
D(X(4))
D(X(5))
D(X(6))
D(X(7))
31.75848
-0.040686
128.3544
75.49241
18.47841
7.057897
-0.353495
0.097531
-0.633231
-0.565473
-0.106726
19.39718
8.815884
-0.580980
0.499805
1.802394
-0.553156
0.285016
0.0000
0.9677
0.0000
0.0000
0.0000
0.0000
0.7248
0.9226
0.5287
0.5736
0.9153
0.0000
0.0000
0.5632
0.6189
0.0760
0.5820
0.7765
3.223725
-0.000317
1.000938
1.009461
0.241345
0.093380
-0.004552
0.001255
-0.007941
-0.006888
-0.001256
0.266438
0.120397
-0.007809
0.006723
0.024098
-0.007035
0.003469
0.101508
0.007802
0.007798
0.013372
0.013061
0.013231
0.012877
0.012867
0.012541
0.012181
0.011767
0.013736
0.013657
0.013441
0.013451
0.013370
0.012719
0.012172
H0: = 1 , .
.
, .
DGP:
Wt = 5 + t + rwt ,
Vt = 1 + t + 0.5* rwt + 0.1*n2t ,
rwt = rwt 1 + 0.5*n3t ,

. . .. www.iet.ru
49
n2t , n3t
.
50 :
50
40
30
20
10
0
5
10
15
20
25
30
35
40
45
50
Vt = + Wt + t :
Dependent Variable: V
Variable
t-Statistic
Prob.
C
W
-2.657128 0.325116
1.021898 0.011165
-8.172855
91.52908
0.0000
0.0000
Durbin-Watson stat
0.290480
Prob(F-statistic)
0.000000

3
2
1
0
-1
-2
5
10
15
20
25
30
35
40
45
50
RESID_V_W
( 2.22
5% 3.46), Vt Wt
. 1
, Vt
Wt . Vt Wt
.
:
. . .. www.iet.ru
Dependent Variable: V
Variable
t-Statistic
Prob.
C
@TREND
W
-1.493506 0.035792
0.503277 0.007131
0.496897 0.007519
-41.72787
70.57274
66.08810
0.0000
0.0000
0.0000
Durbin-Watson stat
2.275079
Prob(F-statistic)
50
0.000000

0.2
0.1
0.0
-0.1
-0.2
5
10
15
20
25
30
35
40
45
50
RESID_V_W_TREND
( 7.09).
- Wt
e ,W_DIF(-i)
e ,W_DIF(+i)
. | .
. | .
*| .
. | .
*| .
. |*
. |**
. |*
**| .
**| .
. |*
. | .
. |*
. | .
*| .
. |*
. |*
. | .
. |**
. |*
****| .
*| .
lag
lead
0
1
2
3
4
5
6
7
8
9
10
0.0353
-0.0237
-0.0846
0.0052
-0.0776
0.1352
0.1986
0.1093
-0.1751
-0.2456
0.1177
0.0353
0.1217
0.0115
-0.1083
0.1174
0.1018
0.0347
0.1669
0.0614
-0.3565
-0.0421
,
9
Wt .
,
Wt :
@TREND
W
0.497313
0.505495
0.023984
0.025512
20.73533
19.81437
0.0000
0.0000
.
, .
0.5
Wt . Vt 0.5t 0.5Wt
. . .. www.iet.ru
51
-1.3
-1.4
-1.5
-1.6
-1.7
-1.8
5
10
15
20
25
30
35
40
45
50
SER1
.
. Vt Wt ,
. Vt 0.5Wt ,
:
25
20
15
10
5
0
-5
5
10
15
20
25
30
35
40
45
50
SER2
, Vt 0.5t 0.5Wt ,
1.493 0.104.

. , DGP,
Vt 0.5t 0.5Wt = (1 + t + 0.5* rwt + 0.1*n2t) 0.5t 0.5(5 + t + rwt) =
= 1.5 + 0.1*n2t .
. . .. www.iet.ru
8.
8.1.
I(1) y1t , , yN t
yt = (y1t , , yN t)T ,
VAR(p)
A(L) yt = + t ,
A(L) = A0 A1L ApLp,

A0 , A1 , , Ap (N N),
A0 = IN ( ),
..
yt = + A1 yt 1 + + Ap yt p + t .

yt = + 0 yt 1 + 1 yt 1 + p 1 yt p + 1 + t ,
0 = A1 + + Ap IN ,
k = (A k+1 + + Ap) , k = 1, 2, , p 1.
,
0 = A1 + + Ap IN = A(1),
rank 0 = rank A(1).
(.7.4), y1t , , yN t ,
A(1) (rank A(1) < N).
0 . , 0
r = rank 0 = 0, 1, , N .
r = 1, , N 1 VAR
( y1t , , yN t ~ I(1) ).
r = 0 , y1t , , yN t .
r = N ,
N-
, , , (1, 0,
, 0)T, (0, 1, , 0)T, , (0, 0, , 1)T . , y1t , , yN
t .
0 , r = rank 0 ,
y1t , , yN t , ,
.
ECM
.
([Johansen (1988)],
[Johansen (1991)], [Johansen (1992)], [Johansen (1994)], [Johansen, Juselius (1990)]).

. ,
. . .. www.iet.ru
,
.
, I(1) y1t , , yN t
VAR c
rank A(1) = r ,
VAR ECM. ,
1
yt = + 0 yt 1 + 1 yt 1 + p 1 yt p + 1 + t

0 = T ,
(N r)- r . (1) , , (r)
,

z1, t 1 = T(1) yt 1 , , zr, t 1 = T(r) yt 1

( r y1t ,
, yN t) y1t , , yN t .
, t - N- ,
t = (1t , , N t)T N -
Cov(t) = , Cov(k t , j s) = 0 t
s k , j = 1, , N .
, , p
, .
t- F- ( N(0, 1)
t- 2 qF) ,
VAR , , , ,
VAR . ,
VAR , .
, y1t , ,
yN t , t = 1, , T ,
L( , , 0 , 1 , , p 1) , , 0 , 1 , , p 1
r . ,
r ,
r
Lmax (r ) = (T 2 ) ln 1 i , r = 1,K, N ,
i =1
1 , K , N ,
, 1 > 1 > >

> 0 . Lmax (r) , r ,
N
.

.
I(1) . ,
I(2), . , [Johansen (1994a)], [Johansen (1994b)], [Johansen (1995b)].
. . .. www.iet.ru
() H0: r = r* ,
HA: r = r*+1.
r*
Lmax (r ) = (T 2 ) ln 1 i
i =1
r* + 1
Lmax (r + 1) = (T 2 ) ln 1 i .
i =1

max (r*) = 2 (Lmax (r*+1) Lmax (r*)) = (T/2) ln (1 r* + 1 ).
( T ) H0
r* N ; (., , [Patterson
(2000), 14.3 14.7], [Enders (1995), B ] [Hamilton (1994),
.11]).
H0: r = r* , , , .
r* + 1
, r* + 1 ,
*
max (r ) . H0:
r = r* HA: r = r*+1
max (r*), .
() H0: r = r* ,
HA: r > r*.

r*
Lmax (r ) = (T 2 ) ln 1 i
i =1
Lmax ( N ) = (T 2 ) ln 1 i .
i =1

trace (r*) = 2 (Lmax (N) Lmax (r*)) = (T 2 )
ln(1 ).
N
i = r* + 1
( T ) H0
r* N ; (., ,
[Patterson (2000), 14.3 14.7], [Enders (1995), B ] [Hamilton (1994),
.10]).
H0: r = r* , , , .
r* + 1
, ,
*
trace (r ) . H0:
r = r* HA: r > 1
trace (r*), .
, , , , r
. ,
.
,
.
. . .. www.iet.ru
, , , 0.05,
H0: r = 0 HA: r > 0.
, r = 0 . H0: r = 1
HA: r > 1. H0: r = 1 , r
= 1; H0: r = 2 HA: r > 2 ..
r . r
= r0 , T
P{ r = k } 0 , k = 0, 1, , r0 1,
P{ r = r0 } 1 .
, , , .

, .
,
r* N . ,
, /
( , CE cointegrating equation).
, r 5
( , , EVIEWS).
H2(r): ;
CE .
H1*(r): ;
CE , .
H1(r): ;
CE , .
H*(r): ;
CE .
H(r): ;
CE .
r 5
:
H2(r) H1*(r) H1(r) H*(r) H(r) .
, , ,
, , .
-.
,
r H2(r) H1*(r) H1(r) H*(r),

(N r) H1*(r) H1(r) H*(r) H(r) .
, 5 , ,
:
H(0) H(r) H(N)
H*(0) H*(r) H*(N)
H1(0) H1(r) H1(N)
H1*(0) H1*(r) H1*(N)
H2(0) H2(r) H2(N) .
. . .. www.iet.ru
max trace ,
, 5 ,
, ,
.

(AIC) 5(N +1) .
,
. , ,
.
,
.
, 5 ,
,
I(1) .
DGP1 : yt = xt + t ,
xt = xt 1 + t .
:
yt yt 1 = yt 1 + (xt 1 + t ) + t = (yt 1 xt 1) + ut ,
ut = t + t ,
ECM
yt = zt 1+ ut ,
xt = t ,
zt = yt xt ( CE).
yt xt (
), H2(r).
DGP2 : yt = 0 + xt + t ,
xt = xt 1 + t .

yt = zt 1+ ut ,
xt = t ,
zt = yt 0 xt ( CE).
yt xt (
), H1*(r).
DGP3 : yt = 0 + xt + t ,
xt = 0 + xt 1 + t .

yt yt 1 = yt 1 + 0+ (0 + xt 1 + t ) + t =
= (yt 1 0 xt 1) + 0 + ut ,

. . .. www.iet.ru
yt = zt 1 + 0 + ut ,
xt = 0 + t ,
zt = yt 0 xt ( CE).
xt ( ).
H1(r).
DGP4 : yt = 0 + 1t + xt + t ,
xt = 0 + xt 1 + t .
yt yt 1 = yt 1 + 0 + 1t + (0 + xt 1 + t ) + t =
= (yt 1 0 1t xt 1) + 0 + ut ,
yt = zt 1 + 0 + ut ,
xt = 0 + t ,
zt = yt 0 1t xt ( CE).
xt ( ).
H*(r).
DGP5 : yt = 0 + 1t + xt + t ,
xt = 0 + 1t + xt 1 + t .

yt yt 1 = yt 1 + 0 + 1t + (0 + 1 t + xt 1 + t ) + t =
= (yt 1 0 1t xt 1) + 0 + 1 t + ut ,
yt = zt 1 + 0 + 1 t + ut ,
xt = 0 + 1t + t ,
zt = yt 0 1t xt ( CE).
xt ( ).
H(r).
, 5
DGP.
:
= 2, 0 = 5, 1 = 0.2, 0 = 0.2, 1 = 0.01.
t t T = 400
, , 4 1, .
DGP1 : yt = 2 xt + t ,
xt = xt 1 + t .

. . .. www.iet.ru
20
10
0
-10
-20
-30
-40
-50
50
100
150
200
Y1
250
300
350
400
X1

, EVIEWS ( , VAR
2):
Sample: 1 400
Series: Y1 X1
Lags interval: 1 to 1
Data Trend: None
None
Rank or
No Intercept Intercept
No. of CEs No Trend
No Trend
Linear
Intercept
No Trend
Linear
Intercept
Trend
Quadratic
Intercept
Trend
-1526.582
-1433.925
-1432.926
-1526.015
-1433.357
-1432.926
-1526.015
-1433.324
-1430.264
-1525.928
-1433.242
-1430.264
7.691369
7.250880
7.270985
7.698565
7.253051
7.270985
7.698565
7.257911
7.267658
7.708180
7.262521
7.267658
Log
Likelihood
0
1
2
-1526.582
-1434.108
-1434.100
AIC
0
1
2
7.691369
7.246775
7.266836
Schwarz
Criteria
0
1
2
7.731434
7.326905
7.387030
7.731434
7.341026
7.411212
7.758663
7.353213
7.411212
7.758663
7.368089
7.427918
7.788309
7.382716
7.427918
L.R. Test:
Rank = 1
Rank = 1
Rank = 1
Rank = 1
Rank = 2
5 3
(r = 0, 1, 2) :
Lmax (r) (Log

Likelihood), .
(AIC Akaike Information

Criteria), .
(Schwarz Criteria),
.
(Rank or No. of
CEs). 5 5 , . (
. . .. www.iet.ru
H2(r), H1*(r), H1(r), H*(r), H(r).)

,
. ,
r = 1, (H(r))
r = 2.
, H2(1) (
, CE ;
1) ( 7.246775).
( 7.326905).

, .
Sample: 1 400
Test assumption: No deterministic trend in the data
Series: Y1 X1
Eigenvalue
Likelihood
Ratio
5 Percent
Critical Value
1 Percent
Critical Value
0.371673
4.00E-05
184.9642
0.015911
12.53
3.84
16.31
6.51
Hypothesized
No. of CE(s)
None **
At most 1
*(**) denotes rejection of the hypothesis at 5%(1%) significance level

L.R. test indicates 1 cointegrating equation(s) at 5% significance level
(Eigenvalue) 1 , 2 ,
. 184.9642
trace (0), H0: r = 0
HA: r > 0. 5% 1%
trace (0) .
, H0: r = 0
HA: r > 0. . 0.015911
trace (1), H0: r = 1
HA: r > 1. 5% 1%
trace (1). trace (1)
, H0: r = 1 .
1,
.
DGP2 : yt = 5 + 2 xt + t ,
xt = xt 1 + t .

. . .. www.iet.ru
20
10
0
-10
-20
-30
-40
-50
50
100
150
200
Y2
250
300
350
400
X2

:
Sample: 1 400
Series: Y2 X2
Data Trend: None
None
Rank or
No. of CEs No Trend
No Trend
Linear
Intercept
No Trend
Linear
Intercept
Trend
Quadratic
Intercept
Trend
-1526.582
-1433.925
-1432.926
-1526.015
-1433.357
-1432.926
-1526.015
-1433.324
-1430.264
-1525.928
-1433.242
-1430.264
7.691369
7.250880
7.270985
7.698565
7.253051
7.270985
7.698565
7.257911
7.267658
7.708180
7.262521
7.267658
Log
Likelihood
0
1
2
-1526.582
-1513.785
-1513.777
AIC
0
1
2
7.691369
7.647159
7.667223
Schwarz
Criteria
0
1
2
7.731434
7.727289
7.787417
7.731434
7.341026
7.411212
7.758663
7.353213
7.411212
7.758663
7.368089
7.427918
7.788309
7.382716
7.427918
L.R. Test:
Rank = 1
Rank = 1
Rank = 1
Rank = 1
Rank = 2
, H*1(1) (
, CE ; 1)
( 7.250880).
( 7.341026).

, .

Series: Y2 X2
Eigenvalue
Likelihood
Ratio
5 Percent
Critical Value
1 Percent
Critical Value
Hypothesized
No. of CE(s)
. . .. www.iet.ru
0.372251
0.005008
187.3130
1.998159
19.96
9.24
24.60
12.97
10
None **
At most 1

187.310 trace (0),

H0: r = 0 HA: r > 0,
. H0: r = 0 HA: r > 0.
1.998159 trace (1),
H0: r = 1 HA: r > 1,
. H0: r = 1 .
1, .
DGP3 : yt = 5 + 2 xt + t ,
xt = 0.2 + xt 1 + t .

140
120
100
80
60
40
20
0
50
100
150
200
Y3
250
300
350
400
X3

:
Sample: 1 400
Series: Y3 X3
Data Trend: None
None
Rank or
No. of CEs No Trend
No Trend
Linear
Intercept
No Trend
Linear
Intercept
Trend
Quadratic
Intercept
Trend
-1531.268
-1438.576
-1432.880
-1526.015
-1433.325
-1432.880
-1526.015
-1433.324
-1430.264
-1525.928
-1433.242
-1430.264
7.714915
7.274253
7.270753
7.698565
7.252889
7.270753
7.698565
7.257911
7.267658
7.708180
7.262521
7.267658
Log
Likelihood
0
1
2
-1531.268
-1506.877
-1504.370
AIC
0
1
2
7.714915
7.612447
7.619950
Schwarz
Criteria
. . .. www.iet.ru
0
1
2
7.754980
7.692577
7.740144
7.754980
7.364399
7.410980
7.758663
7.353051
7.410980
7.758663
7.368089
7.427918
7.788309
7.382716
7.427918
L.R. Test:
Rank = 2
Rank = 2
Rank = 1
Rank = 1
Rank = 2
11
, H1(1)
( , CE ; 1)
( 7.252889).
( 7.353051).
,
:
Sample: 1 400
Test assumption: Linear deterministic trend in the data
Series: Y3 X3
Eigenvalue
Likelihood
Ratio
5 Percent
Critical Value
1 Percent
Critical Value
0.372353
0.002234
186.2692
0.890114
15.41
3.76
20.04
6.65
Hypothesized
No. of CE(s)
None **
At most 1

H0: r = 0 HA: r > 0. H0: r = 1

HA: r > 1.
1, .
DGP4 : yt = 5 + 0.2 t + 2 xt + t ,
xt = 0.2 + xt 1 + t .

250
200
150
100
50
0
50
100
150
200
Y4
250
300
350
400
X4

:
Sample: 1 400
Series: Y4 X4
. . .. www.iet.ru
Data Trend: None

None
Rank or
No. of CEs No Trend
No Trend
Linear
Intercept
No Trend
Linear
Intercept
Trend
Quadratic
Intercept
Trend
-1533.049
-1525.279
-1518.005
-1526.015
-1518.361
-1518.005
-1526.015
-1433.324
-1430.264
-1525.928
-1433.242
-1430.264
7.723863
7.709944
7.698520
7.698565
7.680208
7.698520
7.698565
7.257911
7.267658
7.708180
7.262521
7.267658
12
Log
Likelihood
0
1
2
-1533.049
-1525.311
-1521.201
AIC
0
1
2
7.723863
7.705079
7.704525
Schwarz
Criteria
0
1
2
7.763928
7.785208
7.824720
7.763928
7.800090
7.838747
7.758663
7.780370
7.838747
7.758663
7.368089
7.427918
7.788309
7.382716
7.427918
L.R. Test:
Rank = 2
Rank = 2
Rank = 1
Rank = 1
Rank = 2
, H*(1)
( , CE ;
1) ( 7.257911).
( 7.368089).
,
:
Test assumption: Linear deterministic trend in the data
Series: Y4 X4
Eigenvalue
Likelihood
Ratio
5 Percent
Critical Value
1 Percent
Critical Value
0.372355
0.015260
191.5010
6.120353
25.32
12.25
30.45
16.26
Hypothesized
No. of CE(s)
None **
At most 1

H0: r = 0 HA: r > 0. H0: r = 1

HA: r > 1.
1, .
DGP5 : yt = 5 + 0.2 t + 2 xt + t ,
xt = 0.2 + 0.01 t + xt 1 + t .

. . .. www.iet.ru
13
2000
1500
1000
500
0
50
100
150
200
Y5
250
300
350
400
X5

:
Series: Y5 X5
Data Trend: None
None
Rank or
No. of CEs No Trend
No Trend
Linear
Intercept
No Trend
Linear
Intercept
Trend
Quadratic
Intercept
Trend
-1672.222
-1527.331
-1520.795
-1630.634
-1525.738
-1520.795
-1630.634
-1525.724
-1432.659
-1525.928
-1432.667
-1432.659
8.423224
7.720258
7.712538
8.224289
7.717274
7.712538
8.224289
7.722231
7.279694
7.708180
7.259633
7.279694
Log
Likelihood
0
1
2
-1672.222
-1527.340
-1527.280
AIC
0
1
2
8.423224
7.715279
7.735077
Schwarz
Criteria
0
1
2
8.463289
7.795408
7.855272
8.463289
7.810404
7.852765
8.284386
7.817436
7.852765
8.284386
7.832409
7.439953
7.788309
7.379827
7.439953
L.R. Test:
Rank = 1
Rank = 2
Rank = 2
Rank = 2
Rank = 1
, H(1)
( , CE ;
1) ( 7.259633).
(
7.379827).
,
:
Test assumption: Quadratic deterministic trend in the data
Series: Y5 X5
Likelihood
5 Percent
1 Percent
Hypothesized
. . .. www.iet.ru
Eigenvalue
Ratio
Critical Value
Critical Value
0.374152
3.97E-05
186.5374
0.015819
18.17
3.74
23.46
6.40
14
No. of CE(s)
None **
At most 1

H0: r = 0 HA: r > 0. H0: r = 1

HA: r > 1.
1, .
,
,
.
,
.
DGP: L234t = 0.5W2 t + W3 t + 2W4 t + 1t ,
L23t = W2 t + 0.5W3 t + 2 t ,
W2 t = W2, t 1 + 3 t ,
W3 t = W3, t 1 + 4 t ,
W4 t = W4, t 1 + 5 t ,
1t , 2 t , 3 t , 4 t , 5 t
, , 0.04.
W2 t , W3 t , W4 t
(common trends), ,
() .
L23t W2 t W3 t ,
L234t W2 t , W3 t W4 t .
501 5 DGP
L234t , L23t , W2 t , W3 t W4 t .
.
5
-5
-2
-10
-4
-15
-6
-20
-8
-25
-30
-10
50
100 150 200 250 300 350 400 450 500

L23
50
L234
100 150 200 250 300 350 400 450 500

WALK2
WALK3
WALK4
5 VAR(1) (
)
, 5 .
, L234t
Autocorrelation
Partial Correlation
AC
PAC
Q-Stat Prob
. . .. www.iet.ru
.|.
.|.
.|.
.|.
.|.
.|.
.|.
.|.
.|.
.|.
.|.
.|.
.|.
.|.
.|.
.|.
.|.
.|.
.|.
.|.
1
2
3
4
5
6
7
8
9
10
-0.044
-0.037
0.031
-0.020
0.031
0.036
0.008
0.039
0.003
0.001
-0.044
-0.039
0.028
-0.019
0.031
0.037
0.015
0.041
0.007
0.005
0.9530
1.6395
2.1276
2.3355
2.8073
3.4726
3.5055
4.2674
4.2734
4.2743
15
0.329
0.441
0.546
0.674
0.730
0.748
0.835
0.832
0.893
0.934
VAR(1)
.
,
.
:
Series: L234 L23 WALK2 WALK3 WALK4
Lags interval: No lags
Data Trend: None
None
Rank or
No. of CEs No Trend
No Trend
Linear
Intercept
No Trend
Linear
Intercept
Trend
Quadratic
Intercept
Trend
99.73833
279.4377
457.6821
463.9622
468.0578
470.6760
103.2311
282.9301
461.1675
466.3818
469.2034
470.6760
103.2311
283.2575
462.6611
469.2587
472.1756
473.8191
104.7734
284.7997
464.2006
470.4661
473.0952
473.8191
-0.400556
-1.078063
-1.749727
-1.730772
-1.703044
-1.669382
-0.394502
-1.076025
-1.751677
-1.732457
-1.703628
-1.669382
-0.394502
-1.073323
-1.749643
-1.731963
-1.699501
-1.661924
-0.380616
-1.063453
-1.743777
-1.728780
-1.699178
-1.661924
Log
Likelihood
0
1
2
3
4
5
99.73833
279.2178
457.0419
461.1754
464.5438
464.7799
Akaike
0
1
2
3
4
5
-0.400556
-1.081196
-1.755188
-1.731628
-1.704995
-1.665783
Schwarz
0
1
2
3
4
5
-0.400556
-0.996646
-1.586088
-1.477977
-1.366794
-1.243032
-0.400556
-0.985058
-1.563717
-1.451756
-1.331023
-1.204355
-0.352227
-0.949199
-1.540301
-1.436531
-1.323152
-1.204355
-0.352227
-0.938043
-1.521357
-1.410672
-1.285205
-1.154623
-0.296066
-0.894352
-1.490127
-1.390579
-1.276427
-1.154623
L.R. Test:
Rank = 2
Rank = 2
Rank = 2
Rank = 2
Rank = 2
2.
( ) ,
. . .. www.iet.ru
16
;
2.

W2 t , W3 t W4 t .
Series: WALK2 WALK3 WALK4
Data Trend: None
None
Rank or
No. of CEs No Trend
No Trend
Linear
Intercept
No Trend
Linear
Intercept
Trend
Quadratic
Intercept
Trend
-1.211316
-1.208105
-1.195348
-1.177195
-1.211316
-1.209616
-1.193234
-1.171753
-1.205445
-1.206429
-1.192934
-1.171753
Akaike
0
1
2
3
-1.209358
-1.201837
-1.191311
-1.168162
-1.209358
-1.206417
-1.194747
-1.177195
Schwarz
0
1
2
3
-1.209358
-1.151107
-1.089850
-1.015971
-1.209358
-1.147232
-1.076376
-0.999639
-1.185951
-1.132009
-1.068523
-0.999639
-1.185951
-1.125065
-1.049499
-0.968832
-1.154715
-1.104968
-1.040743
-0.968832
L.R. Test:
Rank = 0
Rank = 0
Rank = 0
Rank = 0
Rank = 0
0,
DGP. ,
,
DGP.
8.2.

( r )
r .
EVIEWS 5 ,
(r = 2), .
Series: L234 L23 WALK2 WALK3 WALK4
Unnormalized Cointegrating Coefficients:
L234
-0.079261
-0.202709
0.001194
-0.002101
-0.000206
L23
-0.198108
0.079211
0.000453
0.001543
0.000771
WALK2
0.236127
0.022787
-0.014625
0.019423
0.011197
WALK3
0.178603
0.161363
-0.034465
-0.024621
-0.009764
WALK4
0.159704
0.406370
0.037834
0.007077
-0.012244

. , , 5
,
. . .. www.iet.ru
17
. 5
, .
, , ..
r = 1,
,
, ..
(-0.079261, -0.198108, 0.236127, 0.178603, 0.159704)T ,
. ,
, .. ,
,
(1, 2.499451, 2.979119, 2.25363, 2.014916)T .
r = 2,
,
, ..
*(1) = ( 0.079261, 0.198108, 0.236127, 0.178603, 0.159704)T
*(2) = ( 0.202709, 0.079211, 0.022787, 0.161363, 0.406370)T.

, , ,

. ,

( )
(, , ..).
, ,
,
, .
r > 1,
q = r 1
(
).

,
, - ()
.
r = 2,
, ,
.
.
, .
, ,
, L234t , L23t ,
W2 t , W3 t W4 t ,
L234t , W2 t, W3 t , W4 t L23t ,
L23t , W2 t, W3 t , W4 t L234t .
,
L234t ,
L23t , ,
, ,

. . .. www.iet.ru
18
(1) = (1, 0, 13 , 14 , 15)T

(2) = (0, 1, 23 , 24 , 25)T .
,
,
(-0.079261, -0.198108, 0.236127, 0.178603, 0.159704)T,
(-0.202709, 0.079211, 0.022787, 0.161363, 0.406370)T,
,
(1, 0, 13 , 14 , 15)T (0, 1, 23 , 24 , 25)T .

EVIEWS :
Normalized Cointegrating Coefficients: 2 Cointegrating Equation(s)
L234
1.000000
L23
0.000000
0.000000
1.000000
WALK2
-0.499995
(0.00541)
-0.991867
(0.00549)
WALK3
-0.993065
(0.00868)
-0.504230
(0.00881)
WALK4
-2.006077
(0.00852)
-0.003537
(0.00865)
,
(1) = (1, 0, 0.499995, 0.993065, 2.006077)T
(2) = (0, 1, 0.991867, 0.504230, 0.003537)T .

L234t = 0.499995W2 t + 0.993065W3 t + 2.006077W4 t,
L23t = 0.991867W2 t + 0.504230W3 t + 0.003537 W4 t,
,
DGP, ,
L234t = 0.5W2 t + W3 t + 2W4 t ,
L23t = W2 t + 0.5W3 t .
, *(1) *(2)
(1) (2) :
*(1) = 0.079261 (1) 0.198108 (2) ,
*(2) = 0.202709 (1) + 0.079211 (2) .

,
.
,
( ).

ECM .
,
, .. .
:
D(L234)
D(L23)
D(WALK2) D(WALK3) D(WALK4)
ecm1 -0.971220 -0.051197 -0.040351 0.005400 0.031996

(0.11006) (0.06585) (0.04248) (0.04468) (0.04231)
(-8.82480) (-0.77748) (-0.94983) (0.12086) (0.75621)
. . .. www.iet.ru
19
ecm2 -0.033474 -1.015363 -0.001894 0.010505 -0.024292

(0.10788) (0.06455) (0.04164) (0.04380) (0.04148)
(-0.31028) (-15.7298) (-0.04549) (0.23985) (-0.58571)
ECM
(L234)t = 0.971220 (ecm1)t 1 0.033474 (ecm2)t 1 + e1t ,
(L23)t = 0.051197 (ecm1)t 1 1.015363 (ecm2)t 1 + e2t ,
(W2)t = 0.040351 (ecm1)t 1 0.001894 (ecm2)t 1 + e3t ,
(W3)t = 0.005400 (ecm1)t 1 + 0.010505 (ecm2)t 1 + e4t ,
(W4)t = 0.031996 (ecm1)t 1 0.024292 (ecm2)t 1 + e5t ,
(ecm1)t = L234t 0.499995W2 t 0.993065W3 t 2.006077W4 t ,

(ecm2)t = L23t 0.991867W2 t 0.504230W3 t 0.003537W4 t .

()
, ,
H0: 25 = 0
,
H0: 12 = 21 = 31 = 32 = 41= 42= 51= 52 = 0,
(ecm1)t 1
(ecm2)t 1 (
).

,
,
EVIEWS
. (,

.)

CATS (Cointegration Analysis of Time Series),
RATS (Regression Analysis of Time Series).

, , [Patterson (2000)].
,
.
o
VAR .
p VAR.
max trace
,
.
. . .. www.iet.ru
20
,
max
trace , (T Np) T ( ).
,
( )
.
,
,
. ,
, .
, ,
,
- (ARFIMA)
(ARCH, GARCH ..), ,
.

( ECM)
( , ).
, I(2) ,
, ,
.
, ,
[Patterson (2000)], [Maddala, Kim (1998)], [Hamilton (1994)], [Mills
(1993)], [ (1972)] [Favero (2001)],
[Clements, Hendry (1998)] [Clements, Hendry (1999)].
1.
., . (1974) . .
. 1, 2. ., .
2.
., . (1972)
. (. .) ., .
3.
.. (2000) . ., .
4.
. (1974) .(. .) ., .
5.
(2001). 34. ., .
6.
Akaike (1973) Information Theory and an Extension of the Maximum Likelihood
Principle, in Petrov B.N. and Cski F. (Eds), Proceedings, 2nd International Symposium on
Information Theory, 267-281. Akadmia Kiado, Budapest.
7.
Andrews D.W.K. (1991) Heteroskedasticity and Autocorrelation Consistent
Covariance Matrix Estimation, Econometrica, 59, 817858.
8.
Ardeni P.G., D. Lubian (1991) Is There Trend Reversion in Purchaising Power
Parity, Europian Economic Review, 35, 1035-1055.
9.
Bartlett M.S. (1946) On the Theoretical Specification of sampling properties of
Autocorrelated Time Series, Journal of the Royal Statistical Society, Series B, 8, 27-41.
10.
Bierens H.J. (1997) Testing the Unit Root with Drift Hypothesis Against Nonlinear
Trend Stationarity, with an Application to the US Price Level and Interest Rate, Journal of
Econometrics, 81, 29-64.
11.
Box G.E.P., Pierce D.A. (1970) Distribution of Residual Autocorrelations in
Autoregressive Integrated Moving Average Time Series Models, Journal of the American
Statistical Association, 65, 1509-1526.
12.
Chan K.H, J.C.Hayya, J.K.Ord (1977) A Note on Trend Removal Methods: The
Case of polynomial versus vatiate differencing, Econometrica, 45, 737-744.
13.
Cheung Y.-W., M.D. Chinn (1996) Deterministic, Stochastic, and Segmented
Trends in Aggregate Output: a Cross-country Analysis, Oxford Economic Papers, 48, 1,
134-162.
14.
Cheung Y.-W., K.S. Lay (1995) Lag Order and Critical Values of a Modified
Dickey-Fuller Test, Oxford Bulletin of Economics and Statistics, 57, 3, 411-419.
15.
Christiano L.J., M. Eichenbaum (1990) Unit Roots in Real GDP: Do We Know,
and Do We Care?, Carnegie-Rochester Conference Series on Public Policy , 32, 7-62.
16.
Clark P.K. (1989) Trend Reversion in Real Output and Unemployment, Journal of
Econometrics, 40,15-32.
17.
Clements M. P., D. F. Hendry (1998) Forecasting Economic Time series, Cambridge
University Press, Cambridge.
18.
Clements M. P., D. F. Hendry (1999) Forecasting Non-stationary Economic Time
series, MIT Press, Cambridge, Massachusetts, London.
19.
Cochrane J.H. (1998) How Big is the Random Walk in GNP?, Journal of Political
Economy, 96, 893-920.
20.
Cogley T. (1990) International Evidence on the Size of the Random Walk in
Output, Journal of Political Economy, 98, 501-518.
21.
Copeland L.S. (1991) Cointegration Tests with Daily Exchange Rate Data, Oxford
Bulletin of Economics and Statistics, 53, 185-198.
22.
Davidson R., J.G. MacKinnon (1993) Estimation and Inference in Econometrics,
Oxford University Press
23.
Dickey D.A. (1976) Estimation and Hypothesis Testing for Nonstationary Time
Series, Ph.D. dissertation, Iowa State University.
24.
Dickey D.A., W.R.Bell, R.B. Miller (1986) Unit Roots in Time Series Models: Tests
and Implications, American Statistican, 40, 12-26.
25.
Dickey D.A. , W.A. Fuller (1979) Distribution of the Estimators for Autoregressive
Time
Series with a Unit Root, Journal of the American Statistical Association, 74, 427
431.
26.
Dickey, D.A., W.A. Fuller (1981) Likelihood Ratio Statistics for Autoregressive
Time Series With a Unit Root, Econometrica, 49, 1057-1072.
27.
Dickey D.A., S. Pantula (1987) Determining the Order of Differencing in
Autoregressive Processes, Journal of Business and Economic Statistics, 15, 455-461.
28.
Dolado H., T. Jenkinson, S. Sosvilla-Rivero (1990) Cointegration and Unit Roots,
Journal of Economic Surveys, 4, 243-273.
29.
Dutt S.D. (1998) Purchasing Power Parity Revisited: Null of Cointegration
Approach, Applied Economic Letters, 5, 573-576.
30.
Dutt S.D., D. Ghosh (1999)
An Empirical Examination of Exchange Market
Efficiency, Applied Economic Letters, 6, 2, 89-91.
31.
Dweyer G.P., Wallace M.S. (1992) Cointegration and Market Efficiency, Journal
of International Money and Finance, 11 318-327.
32.
Elliott G., T.J. Rothenberg, J.H. Stock (1996) Efficient Tests for an Autoregressive
Unit Root, Econometrica, 64, 813-836.
33.
Enders W. (1995) Applied Econometric Time Series, Wiley, New York
34.
Engle R.F., C.W.J. Granger (1987) Co-integration and Error Correction:
Representation, Estimation, and Testing, Econometrica, 55, 251-276.
35.
Engle R.F., C.W.J. Granger (1991) Cointegrated Economic Time Series: An
Overview with New Results, in R.F. Engle and C.W.J. Granger (eds.), Long-Run Economic
Relationships, Readings in Cointegration, Oxford University Press, 237-266.
36.
Entorf H. (1992) Random Walk with Drift, Simultaneous Errors, and Small
Samples: Simulating the Birds Eye View, Institut National de la Statistique et des Etudes
Economiques.
37.
Fama E.F., French K.R. (1988) Permanent and Temporary Components of Stock
Prices, Journal of Political Economy, 96, 246-273.
38.
Favero C. A. (2001) Applied Macroeconometrics, Oxford University Press Inc., NewYork.
39.
Fuller W.A. (1976) Introduction to Statistical Time Series, Wiley, New York.
40.
Fuller W.A. (1996) Introduction to Statistical Time Series, 2nd ed, Wiley, New York
41.
Funke N., J. Thornton (1999) The Demand for Money in Italy, 1861-1988,
Applied Economic Letters, 6, 5, 299-301.
42.
Ghysels E., Perron P. (1992) The Effect of Seasonal Adjustment Filters on Tests
for a Unit Root, Journal of Econometrics, 55, 57-98.
43.
Granger C.W.J. (1983) UCSD Discussion Paper, 83-13a.
44.
Green W.H. (1997) Econometric Analysis. 3rd edition, Prentice-Hall.
45.
den Haan W.J. (2000) The omovement Between Output and Prices, Journal of
Monetary Economics, 46, 1, 3-30.
46.
Hafer R.W., D.W. Jansen (1991) The Demand for Money in the United States:
Evidence from Cointegration Tests, Jounal of Money, Credit, and Banking, 23 (1991), 155168.
47.
Hall A. (1994) Testing for a Unit Root in Time Series with Pretest Data-Based
Model Selection, Journal of Business and Economic Statistics, 12, 451-470.
48.
Hamilton, James D. (1994) Time Series Analysis, Princeton University Press,
Prinseton.
49.
Hannan E.J., Quinn B.G. (1979) The Determination of the Order of an
Autoregression, Journal of the Royal Statistical Society, Series B, 41, 190-195.
50.
Hasan M.S. (1998) The Choice of Appropriate Monetery Aggregate in the United
Kindom, Applied Economic Letters, 5, 9, 563-568.
51.
Hatanaka M. (1996) Time Series-Based Econometrics: Unit Roots and
Cointegration, Oxford University Press.
52.
Holden D., Perman R. (1994) Unit Roots and Cointegration for Economist,
Cointegration for the Applied Economists ( Rao B.B.), Macmillan.
53.
Jarque C., A. Bera (1980) "Efficient Tests for Normality, Homoskedasticity, and
Serial Independence of Regression Residuals," Economics Letters, 6, 255259.
54.
Johansen S. (1988) Statistical Analysis of Cointegration Vectors, Journal of
Economic Dynamics and Control, 12, 231-254.
55.
Johansen S. (1991) Estimation and Hypothesis Testing of Cointegration Vectors in
Gaussian Vector Autoregressive Models, Econometrica, 59, 1551-1580.
56.
Johansen S. (1992) Determination of Cointegration Rank in the Presence of a
Linear Trend, Oxford Bulletin of Economics and Statistics, 54, 383-397.
57.
Johansen S. (1994a) The Role of the Constant Term in Cointegration Analysis of
Nonstationary Variables, Econometric Reviews, 13, 205-219.
58.
Johansen S. (1994b) A Likelihood Analysis of the I(2) model, Scandinavian
Journal of Statistics Johansen S. (1995a) Likelihood-based Inference in Cointegrated Vector
Autoregressive Models, Oxford University Press, Oxford.
59.
Johansen S. (1995) A Statistical Analysis of Cointegration for I(2) Variables,
Econometric Theory, 11, 25-29.
60.
Johansen S., K. Juselius (1990) Maximum Likelihood Estimation and Inferences on
Cointegrationwith applications to the demand for money, Oxford Bulletin of Economics and
Statistics, 52, 169210.
61.
Kavalieris (1991) A Note on Estimating Autoregressive-Moving average Order,
Biometrika, 78, 920-922.
62.
Kim B. J.C., Soowon Mo (1995) Cointegration and the Long-run Forecast of
Exchange Rates, Economics Letters, 48, 3-4, 353-359.
63.
Kwan A.C.C. (1996a) On the Finite-sample Distribution of Modified Portmanteau
Tests for Randomness of a Gaussian Time Series, Biometrika, 83, 4, 938-943.
64.
Kwan A.C.C. (1996b) A Comparative Study of the Finite-sample Distribution of
some Portmanteau Tests for Univariate Time Series Models, Commun. Statist.-Simula, 25,
4, 867-904.
65.
Kwiatkowski D., P.C.B. Phillips, P. Schmidt, Y. Shin (1992) Testing of the Null
Hypothesis of Stationary against the Alternative of a Unit Root, Journal of Econometrics, 54,
159-178.
66.
Kwiatkowski D., P. Schmidt (1990) DickeyFuller Tests with Trend, Commun.
Statist.-Theory Meth., 19, 10, 3645-3656.
67.
Leybourne S.J. (1995) Testing for Unit Roots Using Forward and Reverse DickeyFuller Regressions, Oxford Bulletin of Economics and Statistics, 57, 559-571.
68.
Leybourne S., T. Mills, P. Newbold (1998) Spurious Rejections by Dickey-Fuller
Tests in the Presence of a Break Under Null, Journal of Econometrics, 87, 191-203.
69.
Ljung G., G.E.P. Box (1979) On a Measure of Lack of Fit in Time Series Models,
Biometrika, 66, 255-270.
70.
Lomnicki Z.A. (1961) Tests for Departure from Normality in the Case of Linear
Stochastic Processes, Metrika, 4, 37-62.
71.
Lumsdaine R.L., Kim I.M. (1997) Structural Change and Unit Roots, The Review
of Economics ans Statistics, 79, 212-218.
72.
MacKinnon, J.G. (1991) Critical Values for Cointegration Tests, 13 Longrun Economic Relationships: Readings in Cointegration, edited by R.F.Engle and C.W.J.
Granger, Oxford University Press.
73.
Maddala G.S., In-Moo Kim (1998) Unit Roots, Cointegration, and Structural
Change. Cambridge University Press, Cambridge.
74.
Mann H.B., A. Wald (1943) On Stochastic Limit and Order Relationships, Annals
of Mathematical Statistics, 14, 217-277.
75.
Metin K. (1995) An Integrated Analysis of Turkish Inflation, Oxford Bulletin of
Economics and Statistics, 57, 4, 513-532.
76.
Milas C. (1998) Demand for Greek Imports Using Multivariate Cointegration
Technique, Applied Economics, 30, 11, 1483-1492.
77.
Mills T.C. (1993) The Econometric Modelling of Financial Time Series. Cambridge
University Press, Cambridge.
78.
Molana H. (1994) Consumption and Fiscal Theory. UK Evidence from a
Cointegration Approach, Dundee Discussion Papers, University of Dundey, Dundey,
Scotland.
79.
Murray C.J., C.R. Nelson (2000) The Uncertain Trend in U.S. GDP, Journal of
Monetary Economics, 46, 79-95.
80.
Nadal-De Simone F., W.A. Razzak (1999) Nominal Exchange Rates and Nominal
Interest Rate Differentials, IMF Working Paper WP/99/141.
81.
Nelson C.R., H. Kang (1981) Spurious Periodicity in Inappropriately Detrended
Time Series, Journal of Monetary Economics, 10, 139-162.
82.
Nelson C.R., C.I. Plosser (1982) Trends and Random Walks in Macroeconomic
Time Series, Jornal of Monetary Economics, 10, 139-162.
83.
Newey W., K. West (1987) A Simple Positive Semi-Definite, Heteroskedasticity and
Autocorrelation Consistent Covariance Matrix, Econometrica, 55, 703708
84.
Newey W., K. West (1994) Automatic Lag Selection in Covariance Matrix
Estimation, Review of Economic Studies, 61, 631653.
85.
Ng S., P. Perron (1995) Unit Root Tests in ARMA models With Data-Dependent
Methods for the Selection of the Truncation Lag, Journal of American Statistical
Assosiation, 90, 268-281.
86.
Nunes L.S., Newbold P., C.-M. Kuan (1997) Testing for Unit Roots With Breaks.
Evidence on the Great Crash and the Unit Root Hypothesis Reconsidered, Oxford Bulletin of
Economics and Statistics, 59, 4, 435-448.
87.
Patterson K. (2000) An Introduction to Applied Econometrics: A Time Series
Approach. New York: Sts Martin Press.
88.
Perron P. (1988) Trends and Random Walks in Macroeconomic Time Series: Furter
Evidence from a New Approach, Jounal of Economic Dynamic and Control, 12, 297-332.
89.
Perron P. (1989a) The Great Crash, the Oil Price Shock, and the Unit Root
Hypothesis, Econometrica, 577, 1361-1401.
90.
Perron P. (1989b) Testing for a Random Walk: A Simulation Experiment When the
Sampling Interval Is Varied Advances in Econometrics and Modelling
( B.Ray), Kluwer Academic Publishers, Dordrecht and Boston.
91.
Perron P. (1997) "Further evidence on breaking trend functions in macroeconomic
variables, Journal of Econometrics, 80, 2, 355-385.
92.
Perron P., Vogelsang T.J. (1993) Erratum, Econometrica, 61, 1, 248-249.
93.
Phillips P.C.B. (1987) Time Series Regression with a Unit Root, Econometrica, 55,
277-301.
94.
Phillips P.C.B., P. Perron (1988) Testing for a Unit Root in Time Series
Regression, Biometrika, 75, 335346.
95.
Saikonnen P. (1991) Asymptotically Efficient Estimation of Cointegrated
Regressions, Econometric Theory, 7, 1-21.
96.
Said E., D.A. Dickey (1984) Testing for Unit Roots in Autoregressive Moving
Average Models of Unknown Order, Biometrika, 71, 599607.
97.
Sargan J.D., Bhargava A. (1983) Testing Residuals from Least Squares Regression
for Being Generated by the Gaussian Random Walk, Economertica, 51, N1 153-174
98.
Shiller R.J., Perron P. (1985) Testing the Random Walk Hypothesis: Power versus
Frequency of Observation, Economic Letters, 18, 381-386.
99.
Schmidt P., Phillips P.C.B. (1992) LM Tests for a Unit Root in the Presence of
Deterministic Trends, Oxford Bulletin of Economics and Statistics, 54, 257-287.
100.
Schwarz G. (1978) Estimating the Dimension of a Model, The Annals of Statistics,
16, 461-464.
101.
Schwert G.W. (1989) Tests for Unit Roots: A Monte Carlo Investigation, Journal
of Business and Economic Statistics, 7, 147-159.
102.
Sims C.A., J.H. Stock, M.W. Watson (1990) Inference in Linear Time Series
Models with Some Unit Roots, Econometrica 58, 113-144.
103.
Stock Watson (1993) A Simple Estimator of Cointegrating Vectors in Higher Order
Integrated Systems, Econometrica, 61, 783-820.
104.
Taylor A.M.R. (2000) The Finite Sample Effects of Deterministic Variables on
Conventional Methods of Lag-Selection in Unit-Root Tests, Oxford Bulletin of Economics
and Statistics, 62, 293-304.
105.
West K.D. (1988) Asymptotic Normality, When Regressors Have a Unit Root,
Econometrica, 56, 1397-1417.
106.
White J. S. (1958) The Limiting Distribution of the Serial Correlation Coefficient in
the Explosive Case, Annals of Mathematical Statistics, 29, 1188-1197.
107.
Wirjanto T. S., R.A. Amano (1996) Nonstationary Regression Models with a
Lagged Dependent Variable, Commun. Statist.-Theory Meth., 25, 7, 1489-1503.
108.
Woodward G., R. Pillarisetti (1999) Empirical Evidence on Alternative Theories
of Inflation and Unemployment: a Re-Evaluation for the Scandinavian Countries, Applied
Economic Letters, 6, 1, 55-58.
109.
Zivot E., Andrews D.W.K. (1992) Further Evidence on the Great Crash, the Oil
Price Shock and the Unit Root Hypothesis, Journal of Business and Economic Statistics, 10,
251-270.

Q
Q-, 42, 43
, 54
, 55
, 69, 86
, 69
, 69

,
83
, 16
N- , 230, 242
, 203
, 17
, 212
, 133
, 73
, 77
(ECM),
204, 242
, 258
, 77
, 78
, 204
, 74
, 241
, 242, 255
, 76
, 14
, 15
, 231
, 114
, 160
, 160
k, 116
, 127
, 127
, 117
, 15
, 66
, 231
- , 66
-, 66
, 203
, 63, 116,
133
, 14, 15
ARIMA(p, k, q), 117

, 195, 203
, 203, 212, 221
ARMA ,
204
, 206
,
221
, 221
, 203, 221
, 206
, 256
leads and lags, 232

, 226
,
211, 226, 230
, 221
, 223
, 195

, 83
, 83
, 81
, 82
, 82
, 82
, 82
, 81
, ,
81
, 160, 197,
227
Newey West, 160, 227
, 160
, 78

, 221
,
221
, 23, 73
, 74
, 47
, 47, 48
, 203, 213, 220, 236
, 212
, 213
, 215
, 213
, 214
, 220, 239
VAR, 222
(ECM), 222
ECM, 261
, 261
, 222

,
259
, 261

, 259
, 206, 213
, 206
, 258
, 16, 55
, 16
DF-GLS, 164
, 134, 140, 147
ARMA(p, q), 146
, 142, 148
, 145
, 150

, 44
, 45
, 44

(KPSS), 164
, 163
, 163
, 172
, 181
, 182
, 159
, 161
, 163
, 182

, 4, 8
, 119, 127
, 186, 193
,
47
ARX
, 66
(ADL), 67
, 67, 68
, 69
(ECM), 204
, 124
, 31, 47, 48, 49, 54

, 9
, 23

, 13
, 207
, 196
, 78
ARMA
, 35, 53
, 58
, 35
, 35, 47
, 47

c A, 9
c A, 11
c B, 11
c C, 12
c D, 63
c E, 64
F, 65
, 34
, 32
, 205

Jarque Bera, 58
, 18, 24, 25, 28,
37, 99, 139, 144
, 208
, 111
p, 23
, 84
TS DS , 128
TS-, 129
, 143
, 167
, 129
, 169
, 167
, 153
TS , 143
KPSS, 164
, 165
, 168
, 172
, 174
, 175

, 32
, 34
, 34
, 32
q
, 26, 48
backcasting, 49
, 48
, 111
, 114

(ARMA), 30
, 32
, 30
, 21, 67, 84, 94
, 42
, 42
, 166
, 84
, 81

, 16
, 83
, 113
, 128
, 152, 171
, 114
, 176
, 176
, 176
, 177
, 113, 220
, 187
, 16, 36, 38, 48,

127, 159
, 37
, 36
, 38, 46
, 77
, 77
, 25
, 48
, 26
ARMA,
36
. . .. www.iet.ru
14

Носко В.П. - Эконометрика. Введение в регрессионный анализ рядов - 2002

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Носко В.П. - Эконометрика. Введение в регрессионный анализ рядов - 2002

Uploaded by

Copyright:

Available Formats

..

S2/2 , S2 = RSS/(n p), RSS

F(v1, v2, , vn) = P{ X1 < v1, X2 < v2, ... , Xn < vn }.

Cov(Xt , Xt +) = E [(Xt E(Xt))(Xt + E(Xt+))] .

D(Xt) = D(aXt1 + t) = a2D(Xt1) + D(t), t = 1, , n.

100 150 200 250 300 350 400 450

100 150 200 250 300 350 400 450

a1 Xt1 + a2 Xt2 + + ap Xtp

a(L) = 1 (a1L + a2 L2 + + ap Lp).

= a(L) = (1 a1 a2 ap) = a(1).

(k) = a1 (k1) + a2 (k2) + + ap (kp) , k > 0 .

(k) = a1 (k1) + a2 (k2) + + ap (kp) , k > 0 .

= / (1 a1 a2) = 4.375/(1 0.25 + 0.125) = 5,

100 150 200 250 300 350 400 450

(k) = E [(Xt )(Xtk

100 150 200 250 300 350 400 450

100 150 200 250 300 350 400 450

c) MA(2) Xt = 5 + t 0.75t1 + 0.125 t2 :

100 150 200 250 300 350 400 450

, a(z) = 0 1 2/3z + 1/48

ACF, , 2/T = 0.0895

lg(P2) = lg(630) + 2lg(0.05) + 34lg(0.95) = 0.560.

P-, M = 14, 15, 17 22, 0.05,

100 150 200 250 300 350 400 450 500

2/T = 0.089 PACF,

T = XT bT 1 = (XT ) b(XT 1 ) + b2(XT 2 )

, [Bartlett (1946)] [Box,

QBP 2(M p q).

2(M p q), QBP ,

xt = (1, yt 1, yt 2, , yt p)T , = ( , a1 , a2 , , ap)T .

xt = (xt1, xt2, , xtp)T

[ : plim (n 1 XnT ) = 0];

[ : (n 1/2 XnT ) N(0, 2Q)].

[Mann, Wald (1943)] ( ).

xt = (xt1, xt2, , xtK)T

plim xt xtT = Q .. plim X nT X n = Q ,

E(xt) = = const , Cov(xt) = Q , Cov(xt k , xt +s, l) = kl (s)

zt = (zt1, zt2, , ztM)T , = (1 , 2 , , M)T.

t ~ i.i.d., E( t) = 0, D( t) = 2 > 0, E t m < m = 1, 2, ;

Mean dependent var

Mean dependent var

Mean dependent var

y2t = 2 + 21.1 y1, t 1 + 21.2 y1, t 2 + + 21.p y1, t p +

yt = (y1t , y2t , , ykt )T , = ( 1, 2, , k )T , t = (1t , 2t , , kt)T ,

1 0 0.7 L 0.2 L 1 0.7 L 0.2 L

C (1) = A1 (1) B(1) =

Mean dependent var

Mean dependent var

Mean dependent var

0.013941 0.020679 0.674149 0.5018

0.049508 0.019722 2.510238 0.0137

-0.001126 0.021384 -0.052674 0.9581

0.046032 0.018096 2.543741 0.0126

0.007088 0.015431 0.459354 0.6470

0.020438 0.013757 1.485648 0.1406

S.E. of regression 0.133909

Akaike info criterion -1.153476

0.532005 0.048276 11.02001 0.0000

S.E. of regression 0.134740

Akaike info criterion -1.150947

Mean dependent var

Akaike info criterion

Mean dependent var

Mean dependent var