Professional Documents
Culture Documents
Stock Watson 4E Exercisesolutions Chapter5 Instructors
Stock Watson 4E Exercisesolutions Chapter5 Instructors
5.1 (a) The 95% confidence interval for b1 is {-5.82 ± 1.96 ´ 2.21}, that is
-10.152 £ b1 £ -1.4884.
bˆ 1 - 0 -5.82
t act = = = -2.6335.
SE( bˆ 1) 2.21
The p-value is less than 0.01, so we can reject the null hypothesis at the 5%
significance level, and also at the 1% significance level.
bˆ 1 - (-5.6) 0.22
t act = = = 0.10
SE ( bˆ 1) 2.21
The p-value is larger than 0.10, so we cannot reject the null hypothesis at the
10%, 5% or 1% significance level. Because b1 = -5.6 is not rejected at the 5%
(d) The 99% confidence interval for b0 is {520.4 ± 2.58 ´ 20.4}, that is,
467.7 £ b0 £ 573.0.
(b) The hypothesis testing for the gender gap is H 0 : b1 = 0 vs. H1 : b1 ¹ 0. With a t-
bˆ 1 - 0 2.12
statistic t act = = = 5.89, the p-value for the test is
SE ( bˆ 1) 0.36
p-value = 2Φ(−|t act |) = 2Φ (−5.89) = 2 × 0.0000 = 0.000 (to four decimal places).
The p-value is less than 0.01, so we can reject the null hypothesis that there is no
gender gap at a 1% significance level.
(c) The 95% confidence interval for the gender gap b1 is {2.12 ± 1.96 ´ 0.36}, that
(d) The sample average wage of women is bˆ0 = $12.52/hour. The sample average
(e) The binary variable regression model relating wages to gender can be written as
either Wage = b0 + b1Male + ui , or Wage = g 0 + g 1Female + vi . In the first
regression equation, Male equals 1 for men and 0 for women; b 0 is the
5.2 (continued)
g 0 = b 0 + b1 ,
g 0 + g 1 = b0 .
gˆ 0 = bˆ 0 + bˆ 1 = 14.64,
gˆ1 = bˆ 0 - gˆ 0 = -bˆ 1 = -2.12.
uˆ i = vˆi. Thus the sum of squared residuals, SSR = å i =1 uˆ i , is the same under the
2 n
unchanged.
! = 14.64 − 2.12Female,
Wages R2 = 0.06, SER = 4.2.
(c) The increase in wages for college education is b1 ´ 4. Thus, the counselor’s
assertion is that b1 = 10/4 = 2.50. The t-statistic for this null hypothesis is
t= 2.37−2.50
0.10
= −1.3, which has a p-value of 0.19 Thus, the counselor’s assertion
5. 5 (a) The estimated gain from being in a small class is 13.9 points. This is equal to
approximately 1/5 of the standard deviation in test scores, a moderate increase.
(d) Yes. Students were randomly assigned to small or regular classes, so that SmallClass is
independent of characteristics of the student, including those affecting testscores, that is
u. Thus E(ui | ClassSizei) = 0.
5.6. (a) The question asks whether the variability in test scores in large classes is the
same as the variability in small classes. It is hard to say. On the one hand, teachers in
small classes might able so spend more time bringing all of the students along, reducing
the poor performance of particularly unprepared students. On the other hand, most of
the variability in test scores might be beyond the control of the teacher.
(c) Yes. If Y and X are independent, then b1 = 0; but the p-value in (a) was 0.03. This
means that only in 3% of all samples, the absolute value of t-statistic would be
2.13 (the value actually observed in this sample) or larger.
5.8. (a) 43.2 ± 2.05 ´ 10.2 or 43.2 ± 20.91, where 2.05 is the 5% two-sided critical value
from the t28 distribution.
(c) The one sided 5% critical value is 1.70; tact is less than this critical value, so that
the null hypothesis is not rejected at the 5% level.
1
(Y1 + Y2 +!+ Yn )
5.9. (a) β = n
so that it is linear function of Y1, Y2, …, Yn.
X
1 1
E( β |X 1 ,…, X n ) = E (Y + Y +!+ Yn )|X 1 ,…, X n )
X n 1 2
1 1
= β ( X +!+ X n ) = β1
X n 1 1
5.10. Let n0 denote the number of observation with X = 0 and n1 denote the number of
å å
n n
X = n1; X = n1| n; n11 i =1 X iYi = Y1;
i =1 i
observations with X = 1; note that
å
n
( )
( X i - X ) 2 =å i =1 X i2 - nX 2 = n1 - nn1 = n1 1 - nn1 =
n
; n1Y1 + n0Y0 = å i =1Yi ,
n
2
n1n0
i =1 n
so
Y = nn1 Y1 + nn0 Y0
that
∑ i=1
n
( X i − X )(Yi − Y ) ∑ i=1
n
X i (Yi − Y ) ∑ i=1
n
X iYi − Yn1
β̂1 = = =
∑ i=1 ( X i − X )
n 2
∑ i=1 ( X i − X )
n 2
n1n0 |n
n n⎛ n n ⎞
= (Y1 − Y ) = ⎜ Y − 1 Y1 − 0 Y0 ⎟ = Y1 − Y0 ,
n0 n0 ⎝ n n ⎠
æn n ö n n +n0
and bˆ0 = Y - bˆ1 X = ç 0 Y0 + 1 Y1 ÷ - (Y1 - Y0 ) 1 = 1 Y0 = Y0
èn n ø n n
5.11. Using the results from 5.10, bˆ0 = Ym and bˆ1 = Yw - Ym . From Chapter 3,
bˆ0 = 523.1
s2 s2
SE (Ym ) = Snm SE (Yw - Ym ) = nmm + nww .
m
and Plugging in the numbers and
ˆ ˆ
SE (b0 ) = 6.22; b1 = -38.0 ˆ
SE (b1 ) = 7.65.
and
var (H i ui ) µX
s b2ˆ = , where H i = 1 - Xi.
n éë E ( H i2 )ùû E ( X i2 )
0
2
Using the facts that E (ui | X i ) = 0 and var (ui | X i ) = s u2 (homoskedasticity), we have
æ ö
ç µx ÷ µx
E ( H i ui ) = E u -ç
X u = E (ui ) -
÷
E [ X i E (ui | X i )]
E(X ) E ( X i2 )
ç i 2 i i÷
çç i ÷÷
è ø
=0-
µx ´ 0 = 0,
E ( X i2 )
and
ìæ ö 2 üï
ïç
ïï ç µX ÷
÷ ïï
E [( H i ui ) 2 ] = E u -
íç i Xu i i ÷ ýï
ïç
ï çè E æçè X i2 ö÷ø ÷÷
îï ø ïþï
ì 2 ü
ï
µX é µ ù ï
ïï 2 ï
=E u -2 í i X i ui2 + ê æ X 2 ö ú X i2ui2 ïý
ï
ï E æçè X i2 ö÷ø êë E çè X i ÷ø úû ï
ï
îï þï
2
µ é ù
é µ ù é 2 æ 2 öù
= E u - 2 æ X 2 ö E ê X i E èæç ui2| X i øö÷ ú + ê
æ 2ö
çç i ÷÷
X
ú E ê X i E çè ui | X i ÷ø ú
êë æ 2ö úû êë úû
è ø
E çè X i ÷
ø
êë E çè X i ÷ø úû
2
µ é µ ù æ 2ö
= s u2 - 2 æ X 2 ö µ s u2 + ê X
ú E çç X i ÷÷s u2
X
æ 2ö
E çè X i ÷
ø
êë E çè X i ÷ø úû è ø
æ µ2 ö
= ç1 - æ X 2 ö ÷s u2 .
ç E ç Xi ÷ ÷
è è øø
æ µ X2 ö 2
var (H i ui ) = E[( H i ui ) ] = ç1 - ÷s . 2
ç E ( X i2 ) ÷ u
è ø
5.12 (continued)
ìæ ö 2 üï ì 2 ü
ïç
µX ï
µX é µ ù ï
E ( H i2 ) = E 1 -
÷
ïï ç ÷ ïï ïï 2 ïï
X = E 1- 2 X + ê X
ú X
E ( X i2 ) E ( X i2 ) êë E ( X i ) úû
íç i ÷ ýï í i i ýï
ïç ï 2
÷÷
ï çè ø ïïþ ï ï
îï îï þï
2
µ X2 + é µ X ù E X 2 = 1- µ X2 .
=1- 2 2
ê 2
ú
E ( Xi
i
) êë E ( X i ) úû
2 ( ) E ( Xi )
Thus
æ µX2 ö 2
ç1 - ÷s
var (H i ui ) ç E ( X i2 ) ÷ u s u2
s b2ˆ = = è ø =
0
é nE ( H 2 )2 ù æ ö
2
æ µX2 ö
ûú nç1 - µ X 2 ÷
2
ëê i
n ç1 - ÷
ç E ( Xi ) ÷ ç E ( X i2 ) ÷
è ø è ø
E ( X i2 )s u2 E ( X i2 ) s u2
= = .
n[ E ( X i2 - µ X2 )] ns X2
(b) Yes, this follows from the assumptions in KC 4.3 and conditional
homoskedasticity
(c) They would be unchanged for the reasons specified in the answers to those
questions.
(d) (a) is unchanged; (b) is no longer true as the errors are not conditionally
homoskesdastic.
Xi Yi b̂
only on but not on , is a linear function of Y.
∑ i=1
n
X i E(ui |X 1 ,…, X n )
(b) E( β̂ |X 1 ,…, X n ) = β + = β since E(ui |X 1 ,…, X n ) = 0
∑ i=1
n
X 2j
∑ i=1
n
X i 2Var (ui |X 1 ,…, X n ) σ2
(c) Var ( β̂ |X 1 ,…, X n ) = = n 2
⎡ ∑ i=1
n
X 2j ⎤⎦
2
∑ i=1 X j
⎣
5.15. Because the samples are independent, bˆm ,1 and bˆw,1 are independent. Thus
var ( bˆm,1 - bˆw,1 ) = var ( bˆm,1 ) + var( bˆw,1 ). Var ( bˆm,1 )
is consistently estimated as
[ SE ( bˆm,1 )]2
Var (bˆw,1 ) [ SE ( bˆw,1 )]2 ,
and is consistently estimated as so that
var( bˆm,1 - bˆw,1 ) [ SE ( bˆm,1 )] + [ SE ( bˆw,1 )]2 ,
2