Scribd
Upload
58 views

Problem Set 3 - Solution: T 1 T T T 1 T 1 T 1 T 1 T T T 1 T

This document contains the solutions to problems in an introductory econometrics theory course. It provides detailed working through seven problems involving concepts like OLS regression, projections, and the properties of estimators. Key steps and results are shown for regressing variables on each other, partitioning variance, and deriving the distributions of estimators under different assumptions.

Uploaded by

damnedchild
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
58 views

Problem Set 3 - Solution: T 1 T T T 1 T 1 T 1 T 1 T T T 1 T

This document contains the solutions to problems in an introductory econometrics theory course. It provides detailed working through seven problems involving concepts like OLS regression, projections, and the properties of estimators. Key steps and results are shown for regressing variables on each other, partitioning variance, and deriving the distributions of estimators under different assumptions.

Uploaded by

damnedchild
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

Intro.

Econometric Theory (Fall 06/07)

1-1

Nese Yildiz

Problem Set 3 - Solution


1. (3.2) Let and e denote the OLS coefficients and residuals from regressing Y on X.
Let and e denote the OLS regression coefficients and residuals from regressing
Y on Z.
Z = Z(Z T Z)1 Z T Y = XC(C T X T XC)1 C T XY
= XCC 1 (X T X)1 (C T )1 C T X T Y

= X(X T X)1 X T Y = X .
In addition, e = Y Z = Y X = e.
2. (3.6) OLS coefficient from regressing y on X equals
(X T X)1 X T y = (X T X)1 X T X(X T X)1 X T y

= (X T X)1 X T y = ,
where is the OLS coefficient from regression of Y on X.
3. (3.7) Note
P X = P [X1 X2 ] = X(X T X)1 X T X = X = [X1 X2 ].
Since the equality must hold for each entry this means that P X1 = X1 . Similarly,
M X1 = (In P )X1 = X1 P X1 , which by the first part equals X1 X1 = 0.
4. (3.10) Note that the question assumes that P1 and P2 are well
 Tdefined, which

X1 X 1
0
T
T
T
means that X1 X1 and X2 X2 are both invertible. X X =
,
0
X2T X2
since X1T X2 = 0. Then



 (X1T X1 )1
0
X1
P = X 1 X2
0
(X2T X2 )1
X2


 (X1T X1 )1 X1T
= X 1 X2
= X1 (X1T X1 )1 X1T + X2 (X2T X2 )1 X2T = P1 + P2 .
(X2T X2 )1 X2T
5. (3.12) The first regression cannot be run. This is because 1 = d1 + d2 . As a result,
the data matrix X does not have full rank, i.e. X T X is not invertible. The other
two regression can be run as they do not suffer from this problem.

Intro. Econometric Theory (Fall 06/07)

1-2

Nese Yildiz

a) Note d1i + d2i = 1, so that yi = d1i 1 + d2i 2 + ei = 1 + d2i (2 1 ) + ei .


Thus, = 1 and = 2 1 .
b) 1T d1 = n1 and 1T d2 = n2 .
c) The answer depends on the definition of e. If e is the population residual from
linear projection of Y on d1 and d2 (or on d1 and the constant, or on d2 and
the constant), then this assumption is automatically satisfied. If e denotes
the structural unobservable, then this assumption has content. For example,
if y denotes wages this assumption would mean that the unobservables in the
wage equation are uncorrelated with gender.
6. (3.13)
a) By construction

Pn

i d1i
i=1 e

= 0. Moreover, d1i d2i = 0 for each i. Therefore,


Pn
Pn
[
1 d21i + 2 d1i d1i + ei d1i ]
i=1 yi d1i
= i=1
n1
n1
Pn

d
1 1i
= 1 .
= Pi=1
n
d
i=1 1i

The argument for 2 is similar.


b) y is the residual vector from regressing y on d1 and d2 , and X denotes the
residuals from regressing X on d1 and d2 .
c) They are the same.
7. (4.4)
a) E(y|X) = (X T 1 X)1 X T 1 X + (X T 1 X)1 X T 1 E(e|X) = .
b) Let A = (X T 1 X)1 X T 1 .

V ar(|X)
= AV ar(y|X)AT
= (X T 1 X)1 X T 1 2 (X T 1 X)1
= 2 (X T 1 X)1
c) y X = y XAy = (I XA)y. SoM1 = I XA, and the claim follows if we
show that M1 X = 0. But, M1 X = X X(X T 1 X)1 X T 1 X = X X =
0.

Intro. Econometric Theory (Fall 06/07)

1-3

Nese Yildiz

d)
M1 1 M1 = M1 [1 1 X(X T 1 X)1 X T 1 ]
= [I 1 X(X T 1 X)1 X T ][1 1 X(X T 1 X)1 X T 1 ]
= 1 21 X(X T 1 X)1 X T 1
+ 1 X(X T 1 X)1 X T 1 X(X T 1 X)1 X T 1
= 1 1 X(X T 1 X)1 X T 1 .
e) By part (c), E(s2 |X) =
using part (d) we have

1
E[
eT 1 e|X]
nK

1
E[eT M1 1 M1 e|X].
nK

Then

E[eT M1 1 M1 e|X] = tr(E[eT M1 1 M1 e|X]) = E[tr(eT M1 1 M1 e)|X]


= E[tr(eeT M1 1 M1 )|X] = tr(E[eeT M1 1 M1 |X])
= 2 tr(M1 1 M1 )
= 2 tr(In X(X T 1 X)1 X T 1 )
= 2 [n tr(X(X T 1 X)1 X T 1 )]
= 2 [n tr(X T 1 X(X T 1 X)1 )]
= 2 [n tr(IK )] = 2 (n K).
Thus, E(s2 |X) = 2 .
f) s2 is an unbiased estimator for 2 .

You might also like