Académique Documents
Professionnel Documents
Culture Documents
by
James H. Stock and Mark W. Watson
7.1 and 7.2
Regressor
College (X1)
Female (X2)
(1)
8.31**
(0.23)
3.85**
(0.23)
(2)
8.32**
(0.22)
3.81**
(0.22)
0.51**
(0.04)
17.02**
(0.17)
1.87
(1.18)
Age (X3)
Northeast (X4)
Midwest (X5)
South (X6)
Intercept
(3)
8.34**
(0.22)
3.80**
(0.22)
0.52**
(0.04)
0.18
(0.36)
1.23**
(0.31)
0.43
(0.30)
2.05*
(1.18)
(a) The t-statistic is 8.31/0.23 = 36.1 > 1.96, so the coefficient is statistically
significant at the 5% level. The 95% confidence interval is 8.31 (1.96 0.23).
(b) t-statistic is 3.85/0.23 = 16.7, and 16.7 > 1.96, so the coefficient is statistically
significant at the 5% level. The 95% confidence interval is 3.85 (1.96 0.23).
7.3. (a) Yes, age is an important determinant of earnings. Using a t-test, the t-statistic is
0.51/0.04 = 12.8, with a p-value less than .01, implying that the coefficient on age
is statistically significant at the 1% level. The 95% confidence interval is 0.51
(1.96 0.04).
(b) Age [0.51 1.96 0.04] = 5 [0.51 1.96 0.04] = 2.55 1.96 0.20 =
$2.16 to $2.94
7.4. (a) The F-statistic testing the coefficients on the regional regressors are zero is 7.38.
The 1% critical value (from the F3, distribution) is 3.78. Because 7.38 > 3.78,
the regional effects are significant at the 1% level.
(b) The expected difference between Juanita and Molly is (X6,Juanita X6,Molly) 6 = 6.
Thus a 95% confidence interval is 0.43 (1.96 0.30).
(c) The expected difference between Juanita and Jennifer is
(X5,Juanita X5,Jennifer) 5 + (X6,Juanita X6,Jennifer) 6 = 5 + 6.
A 95% confidence interval could be contructed using the general methods discussed
in Section 7.3. In this case, an easy way to do this is to omit Midwest from the
regression and replace it with X5 = West. In this new regression the coefficient on
South measures the difference in wages between the South and the Midwest, and a
95% confidence interval can be computed directly.
college,1992
college,2012
college,1992
Because college,2012 and college,1992 are computed from independent samples, they are
independent, which means that cov( college,2012 , college,1992 ) = 0 .
Thus, var( college,2012 college,1992 ) = var( college,2012 ) + var( college,1998 ).
1
This implies that SE( college,2012 college,1992 ) = (0.222 + 0.332 ) 2 = 0.40.
Thus, the t-statistic is (8.32 8.66)/0.40 = 0.85. The estimated change is not
statistically significant at the 5% significance level (0.85 < 1.96).
0.485
2.61
7.8. (a) Using the expressions for R2 and R 2, algebra shows that
R2 = 1
n 1
n k 1
(1 R 2 ), so R 2 = 1
(1 R 2 ).
n k 1
n 1
Column 1: R 2 = 1
420 1 1
(1 0.049) = 0.051
420 1
Column 2: R 2 = 1
420 2 1
(1 0.424) = 0.427
420 1
Column 3: R 2 = 1
420 3 1
(1 0.773) = 0.775
420 1
Column 4: R 2 = 1
420 3 1
(1 0.626) = 0.629
420 1
Column 5: R 2 = 1
420 4 1
(1 0.773) = 0.775
420 1
H 0 : 3 = 4 = 0
(b)
H1 : 3 , 4 0
Unrestricted regression (Column 5):
2
Y = 0 + 1 X 1 + 2 X 2 + 3 X 3 + 4 X 4 , Runrestricted
= 0.775
FHomoskedasticityOnly =
=
2
2
( Runrestricted
Rrestricted
)/q
, n = 420, kunrestricted = 4, q = 2
2
(1 Runrestricted )/(n kunrestricted 1)
(0.775 0.427)/2
0.348/2
0.174
=
=
= 322.22
(1 0.775)/(420 4 1) (0.225)/415 0.00054
5% Critical value form F2,00 = 4.61; FHomoskedasticityOnly > F2,00 so Ho is rejected at the
5% level.
(continued on the next page)
7.8 (continued)
(c) t3 = 13.921 and t4 = 0.814, q = 2; |t3| > c (Where c = 2.807, the 1% Benferroni
critical value from Table 7.3). Thus the null hypothesis is rejected at the 1%
level.
Yi = 0 + X1i + 2 ( X1i + X 2i ) + ui
and test whether = 0.
(b) Estimate
Yi = 0 + X1i + 2 ( X 2i aX1i ) + ui
and test whether = 0.
(c) Estimate
2
2
SSR
7.10. Because R2 = 1 TSS
, Runrestricted
Rrestricted
=
SSRrestricted SSRunrestricted
TSS
2
1 Runrestricted
= SSRunrestricted
. Thus
TSS
F=
=
=
2
2
( Runrestricted
Rrestricted
)/q
2
(1 Runrestricted )/(n kunrestricted 1)
SSRrestricted SSRunrestricted
TSS
SSRunrestricted
TSS
/q
/(n kunrestricted 1)
and
7.11. (a) Treatment (assignment to small classes) was not randomly assigned in the
population (the continuing and newly-enrolled students) because of the
difference in the proportion of treated continuing and newly-enrolled students.
Thus, the treatment indicator X1 is correlated with X2. If newly-enrolled
students perform systematicallydifferently on standardized tests than continuing
students (perhaps because of adjustment to a new school), then this becomes
part of the error term u in (a). This leads to correlation between X1 and u, so that
E(u|Xl) 0. Because E(u|Xl) 0, the 1 is biased and inconsistent.
(b) Because treatment was randomly assigned conditional on enrollment status
(continuing or newly-enrolled), E(u|X1, X2) will not depend on X1. This means
that the assumption of conditional mean independence is satisfied, and 1 is
unbiased and consistent. However, because X2 was not randomly assigned
(newly-enrolled students may, on average, have attributes other than being
newly enrolled that affect test scores), E(u|X1, X2) may depend of X2, so that 2
may be biased and inconsistent.