Nov.

25, 2003, Revised February 23, 2010

Hayashi Econometrics

Solution to Chapter 2 Analytical Exercises

1. For any > 0,

1
0 as n .
n

Prob(|zn | > ) =
So, plim zn = 0. On the other hand,
E(zn ) =

1
n1
0 + n2 = n,
n
n

which means that limn E(zn ) = .

2. As shown in the hint,
(z n )2 = (z n E(z n ))2 + 2(z n E(z n ))(E(z n ) ) + (E(z n ) )2 .
Take the expectation of both sides to obtain
E[(z n )2 ] = E[(z n E(z n ))2 ] + 2 E[z n E(z n )](E(z n ) ) + (E(z n ) )2
= Var(z n ) + (E(z n ) )2

(because E[z n E(z n )] = E(z n ) E(z n ) = 0).

Take the limit as n of both sides to obtain

lim E[(z n )2 ] = lim Var(z n ) + lim (E(z n ) )2

=0

(because lim E(z n ) = , lim Var(z n ) = 0).

n

Therefore, zn m.s. . By Lemma 2.2(a), this implies zn p .

3. (a) Since an i.i.d. process is ergodic stationary, Assumption 2.2 is implied by Assumption 2.20 .
Assumptions 2.1 and 2.20 imply that gi xi i is i.i.d. Since an i.i.d. process with mean
zero is mds (martingale differences), Assumption 2.5 is implied by Assumptions 2.20 and
2.50 .
(b) Rewrite the OLS estimator as
b = (X0 X)1 X0 = S1
xx g.

(A)

Since by Assumption 2.20 {xi } is i.i.d., {xi x0i } is i.i.d. So by Kolmogorovs Second Strong
LLN, we obtain
Sxx xx
p

The convergence is actually almost surely, but almost sure convergence implies convergence
in probability. Since xx is invertible by Assumption 2.4, by Lemma 2.3(a) we get
1
S1
xx xx .
p

Similarly, under Assumption 2.1 and 2.20 {gi } is i.i.d. By Kolmogorovs Second Strong
LLN, we obtain
g E(gi ),
p

which is zero by Assumption 2.3. So by Lemma 2.3(a),

1
S1
xx g xx 0 = 0.
p

Therefore, plimn (b ) = 0 which implies that the OLS estimator b is consistent.

Next, we prove that the OLS estimator b is asymptotically normal. Rewrite equation(A)
above as

n(b ) = S1
ng.
xx
0

As already observed, {gi } is i.i.d. with E(gi ) = 0. The variance of gi equals E(gi gi ) = S
since E(gi ) = 0 by Assumption 2.3. So by the Lindeberg-Levy CLT,

ng N (0, S).
d

1
Furthermore, as already noted, S1
xx p xx . Thus by Lemma 2.4(c),

1
n(b ) N (0, 1
xx S xx ).
d

4. The hint is as good as the answer.

5. As shown in the solution to Chapter 1 Analytical Exercise 5, SSRR SSRU can be written as
SSRR SSRU = (Rb r)0 [R(X0 X)1 R0 ]1 (Rb r).
Using the restrictions of the null hypothesis,
Rb r = R(b )
= R(X0 X)1 X0

(since b = (X0 X)1 X0 )

n
1X
(where g
xi i .).
n i=1

= RS1
xx g

1
Also [R(X0 X)1 R]1 = n [RS1
. So
xx R]

0
1
0 1
R S1
SSRR SSRU = ( n g)0 S1
xx ( n g).
xx R (R Sxx R )

Thus

SSRR SSRU
0 2
1
0 1
= ( n g)0 S1
R S1
xx R (s R Sxx R )
xx ( n g)
2
s
= z0n A1
n zn ,
where

2
1
0
zn R S1
xx ( n g), An s R Sxx R .

By Assumption 2.2, plim Sxx = xx . By Assumption 2.5,

Lemma 2.4(c), we have:
1
0
zn N (0, R1
xx Sxx R ).
d

ng d N (0, S). So by

But, as shown in (2.6.4), S = 2 xx under conditional homoekedasticity (Assumption 2.7).

So the expression for the variance of the limiting distribution above becomes
1
1 0
0
2
R1
xx Sxx R = Rxx R A.

Thus we have shown:

zn z, z N (0, A).
d

As already observed, Sxx p xx . By Assumption 2.7, 2 = E(2i ). So by Proposition

2.2, s2 p 2 . Thus by Lemma 2.3(a) (the Continuous Mapping Theorem), An p A.
Therefore, by Lemma 2.4(d),
0 1
z0n A1
z.
n zn z A
d

But since Var(z) = A, the distribution of z A

z is chi-squared with #z degrees of freedom.

6. For simplicity, we assumed in Section 2.8 that {yi , xi } is i.i.d. Collecting all the assumptions
made in Section 2.8,
(i) (linearity) yi = x0i + i .
(ii) (random sample) {yi , xi } is i.i.d.
(iii) (rank condition) E(xi x0i ) is non-singular.
(iv) E(2i xi x0i ) is non-singular.
(v) (stronger version of orthogonality) E(i |xi ) = 0 (see (2.8.5)).
(vi) (parameterized conditional heteroskedasticity) E(2i |xi ) = z0i .
These conditions together are stronger than Assumptions 2.1-2.5.
(a) We wish to verify Assumptions 2.1-2.3 for the regression equation (2.8.8). Clearly, Assumption 2.1 about the regression equation (2.8.8) is satisfied by (i) about the original
regression. Assumption 2.2 about (2.8.8) (that {2i , xi } is ergodic stationary) is satisfied
by (i) and (ii). To see that Assumption 2.3 about (2.8.8) (that E(zi i ) = 0) is satisfied, note first that E(i |xi ) = 0 by construction. Since zi is a function of xi , we have
E(i |zi ) = 0 by the Law of Iterated Expectation. Therefore, Assumption 2.3 is satisfied.
The additional assumption needed for (2.8.8) is Assumption 2.4 that E(zi z0i ) be none is consistent
singular. With Assumptions 2.1-2.4 satisfied for (2.8.8), the OLS estimator
by Proposition 2.1(a) applied to (2.8.8).
b
e = (
b ) (
e ) and use the hint.
(b) Note that
(c) Regarding the first term of (), by Kolmogorovs LLN, the sample mean in that term
converges in probability to E(xi i zi ) provided this population mean exists. But
E(xi i zi ) = E[zi xi E(i |zi )].
By (v) (that E(i |xi ) = 0) and the Law of Iterated Expectations, E(i |zi ) = 0. Thus
E(xi i zi ) = 0. Furthermore, plim(b ) = 0 since b is consistent when Assumptions
2.1-2.4 (which are implied by Assumptions (i)-(vi) above) are satisfied for the original
regression. Therefore, the first term of () converges in probability to zero.
Regarding the second term of (), the sample mean in that term converges in probability to E(x2i zi ) provided this population mean exists. Then the second term converges
in probability to zero because plim(b ) = 0.

(d) Multiplying both sides of () by

b )
e =
n(
=

n
1 X

i=1

n
1 X

zi z0i

i=1

zi z0i
"

n,

n
1 1 X

zi vi
n i=1

#
n
n

1X
1X 2
2 n(b )
xi i zi + n(b ) (b )
x zi .
n i=1
n i=1 i

Under Assumptions
2.1-2.5 for the original regression (which are implied by Assumptions
(i)-(vi)P
above), n(b ) converges in distribution to a random variable. As shown in
n
(c), n1 i=1 xi i zi p 0. So by Lemma 2.4(b) the first term
Pn in the brackets vanishes
(converges to zero in probability). As shown in (c), (b ) n1 i=1 x2i zi vanishes provided
E(x2i zi ) exists and is finite. So by Lemma 2.4(b) the second term, too, vanishes. Therefore,

b )
e vanishes, provided that E(zi z0i ) is non-singular.
n(
7. This exercise is about the model in Section 2.8, so we continue to maintain Assumptions (i)(vi) listed in the solution to the previous exercise. Given the hint, the only thing to show is
1
1 0
that the LHS of () equals 1
xx S xx , or more specifically, that plim n X VX = S. Write S
as
S = E(2i xi x0i )
= E[E(2i |xi )xi x0i ]
= E(z0i xi x0i )

(since E(2i |xi ) = z0i by (vi)).

Since xi is i.i.d. by (ii) and since zi is a function of xi , z0i xi x0i is i.i.d. So its sample mean
converges in probability to its population mean E(z0i xi x0i ), which equals S. The sample
mean can be written as
n

1X 0
z xi x0i
n i=1 i
n

1X
vi xi x0i
n i=1

1 0
X VX.
n

(by the definition of vi , where vi is the i-th diagonal element of V)

8. See the hint.

9. (a)
E(gt |gt1 , gt2 , . . . , g2 )
= E[E(gt |t1 , t2 , . . . , 1 )|gt1 , gt2 , . . . , g2 ]
(by the Law of Iterated Expectations)
= E[E(t t1 |t1 , t2 , . . . , 1 )|gt1 , gt2 , . . . , g2 ]
= E[t1 E(t |t1 , t2 , . . . , 1 )|gt1 , gt2 , . . . , g2 ]
=0
(since E(t |t1 , t2 , . . . , 1 ) = 0).

(by the linearity of conditional expectations)

(b)
E(gt2 ) = E(2t 2t1 )
= E[E(2t 2t1 |t1 , t2 , . . . , 1 )]
=
=

(by the Law of Total Expectations)

E[E(2t |t1 , t2 , . . . , 1 )2t1 ]

(by the linearity
2 2
2
E( t1 )
(since E(t |t1 , t2 , . . . , 1 ) = 2 )

of conditional expectations)

= 2 E(2t1 ).
But
E(2t1 ) = E[E(2t1 |t2 , t3 , . . . , 1 )] = E( 2 ) = 2 .
(c) If {t } is ergodic stationary, then {t t1 } is ergodic stationary (see, e.g., Remark 5.3 on p.
488 of S. Karlin and H. Taylor, A First Course in Stochastic Processes, 2nd. ed., Academic
Press, 1975, which states that For any function , the sequence Yn = (Xn , Xn+1 , . . . )
generates an ergodic stationary process whenever {Xn }
is ergodic
Thus the
Pn
stationary.)
Billingsley CLT (see p. 106 of the text) is applicable to nb
1 = n n1 t=j+1 gt .
b0 converges in probability to E(2t ) = 2 . As shown in (c),
(d) Since 2t is ergodic stationary,

4
nb
1 d N (0, ). So by Lemma 2.4(c) n bb10 d N (0, 1).
10. (a) Clearly, E(yt ) = 0 for all t = 1, 2, . . . .

(1 + 12 + 22 )2

( + ) 2
1
1 2
Cov(yt , ytj ) =
2

for
for
for
for

j
j
j
j

=0
= 1,
= 2,
> 2,

So neither E(yt ) nor Cov(yt , ytj ) depends on t.

(b)
E(yt |ytj , ytj1 , . . . , y0 , y1 )
= E(yt |tj , tj1 , . . . , 0 , 1 )
(as noted in the hint)
= E(t + 1 t1 + 2 t2 |tj , tj1 , . . . , 0 , 1 )

t + 1 t1 + 2 t2 for j = 0,

for j = 1,
1 t1 + 2 t2
=

for j = 2,

2 t2

0
for j > 2,
which gives the desired result.

(c)

1
Var( n y) = [Cov(y1 , y1 + + yn ) + + Cov(yn , y1 + + yn )]
n
=

1
[(0 + 1 + + n2 + n1 ) + (1 + 0 + 1 + + n2 )
n
+ + (n1 + n2 + + 1 + 0 )]

1
[n0 + 2(n 1)1 + + 2(n j)j + + 2n1 ]
n
n1
X
j
= 0 + 2
1
j .
n
j=1
=

(This is just reproducing (6.5.2) of the book.) Since j = 0 for j > 2, one obtains the
desired result.

(d) To use Lemma 2.1, one sets zn = ny. However, Lemma 2.1, as stated in the book, inadvertently misses the required condition that there exist an M > 0 such that E(|zn |s+ ) < M
for all n for some > 0.Provided this technicalcondition is satisfied, the variance of the
limiting distribution of ny is the limit of Var( ny), which is 0 + 2(1 + 2 ).
11. (a) In the auxiliary regression, the vector of the dependent variable is e and the matrix of
.
regressors is [X .. E]. Using the OLS formula,
" 1 0 #
nX e
1
b
b =B

.
1 0
nE e
X0 e = 0 by the normal equations for the original regression. The j-th element of

1 0
nE e

is

n
1
1 X
(ej+1 e1 + + en enj ) =
et etj .
n
n t=j+1

which equals
bj defined in (2.10.9).
Pn
(b) The j-th column of n1 X0 E is n1 t=j+1 xt etj (which, incidentally, equals j defined on
p. 147 of the book). Rewrite it as follows.
n
1 X
xt etj
n t=j+1
n
1 X
=
xt (tj x0tj (b ))
n t=j+1

n
n
X
X
1
1
=
xt tj
xt x0tj (b )
n t=j+1
n t=j+1

Pn
The last term vanishes because b is consistent for . Thus n1 t=j+1 xt etj converges in
probability to E(xt tj ).
The (i, j) element of the symmetric matrix n1 E0 E is, for i j,
nj
1 X
1
(e1+ij e1 + + enj eni ) =
et et(ij) .
n
n t=1+ij

Using the relation et = t x0t (b ), this can be rewritten as

nj
nj
1 X
1 X
t t(ij)
(xt t(ij) + xt(ij) t )0 (b )
n t=1+ij
n t=1+ij

(b )0

nj
X

n t=1+ij

xt x0t(ij) (b ).

The type of argument that is by now routine (similar to the one used on p. 145 for
(2.10.10)) shows that this expression converges in probability to ij , which is 2 for i = j
and zero for i 6= j.
b = B. Since xx is non-singular, B is non-singular. So B
b 1
(c) As shown in (b), plim B
converges in probability to B1 . Also, using an argument similar to the one used in (b)
b = 0. Thus the formula in (a)
for showing that plim n1 E0 E = Ip , we can show that plim
b converges in probability to zero.
shows that

0 0
1 0
SSR
1 0
b . Show that n = n e e
b
(d) (The hint should have been: n E e =
. The SSR from
b

the auxiliary regression can be written as

.
.
1
1
b 0 (e [X .. E])
b
SSR = (e [X .. E])
n
n
.
1
b 0e
= (e [X .. E])
(by the normal equation for the auxiliary regression)
n
.
1
1 0
b [X .. E]0 e
= e0 e
n
n
" 1 0 #
1 0
nX e
0
b
= ee
1 0
n
Ee
n

1
b0
= e0 e
n

"

0
b

#
(since X0 e = 0 and

1 0
b ).
Ee=
n

b = 0 and plim
b = 0. By Proposition 2.2, we have plim n1 e0 e = 2 .
As shown in (c), plim
Hence SSR/n (and therefore SSR/(n K p)) converges to 2 in probability.
(e) Let

(pK)

..
.

Ip

.
, V [X .. E].

The F -ratio is for the hypothesis that R = 0. The F -ratio can be written as
F =

1
b 0 R(V0 V)1 R0
b
(R)
(R)/p
.
SSR/(n K p)

()

b in (a) above, R
b can be written as
Using the expression for

0
..
(K1)
b 1

b=
R
0
. Ip B
b

(pK)
(p1)

(pK)

..
.

b 11
B

(KK)
b 21
B

Ip

(pK)

b 12
B

(Kp)
b 22

(K1)

(p1)

(pp)

b 22
b.
=B
0

Also, R(V V)

()

R in the expression for F can be written as

1
1 b 1 0
b
RB R
(since V0 V = B)
n
n
11
"
b 12
b

B
B
..
1
(KK) (Kp)
=
0
. Ip b 21
b 22
n (pK)
B
B

R(V0 V)1 R0 =

(pK)

(pp)

(Kp)

Ip

1 b 22
B .
( )
n
Substitution of ( ) and () into () produces the desired result.
(f) Just apply the formula for partitioned inverses.

b p , it should be clear that the modified Box-Pierce

(g) Since nb
nb
/ 2 p 0 and
0
1
b
b (Ip )
b) is asymptotically equivalent to nb
b / 4 . Regarding the
Q (= n
0 (Ip )1
22
b given in (f) above. Since
pF statistic given in (e) above, consider the expression for B
the j-th element of n1 X0 E is j defined right below (2.10.19) on p. 147, we have

b = 1 E0 X S1 1 X0 E ,
s2
xx
n
n
so
i1
h
b
b 22 = 1 E0 E s2
.
B
n
b 22 p 12 (Ip )1 , and pF is asymptotiAs shown in (b), 1 E0 E p 2 Ip . Therefore, B
=

b / 4 .
cally equivalent to nb
0 (Ip )1

12. The hints are almost as good as the answer. Here, we give solutions to (b) and (c) only.
(b) We only prove the first convergence result.
r
!
r
!
r
X
X
1X
r
1
1
xt x0t =
xt x0t =
xt x0t .
n t=1
n r t=1
r t=1
The term in parentheses converges in probability to xx as n (and hence r) goes to infinity.
(c) We only prove the first convergence result.
!

!
r
r
r
r

1 X
1 X
r
1 X

xt t =
xt t =
xt t .
n
n t=1
r t=1
r t=1
The term in parentheses converges in distribution to N (0, 2 xx ) as n (and hence r) goes
to infinity. So the whole expression converges in distribution to N (0, 2 xx ).