The McGraw-Hill Series Economics essentials of economics brue, McConnell, and Flynn Essentials of Economics

Coefficient
Std. Error
t
-Statistic
Prob.
C
−
0.447481
0.362943
1.232924
0.2188
X
1.171128
0.075386
15.53500
0.0000
R
-squared
0.503480
Mean dependent var.
0.499826
Adjusted 
R
-squared
0.501394
S.D. dependent var.
7.849594
S.E. of regression
5.542759
Durbin-Watson stat.
1.984746
Sum squared resid.
7311.877
Prob. (
F
-statistic)
0.000000
F
-statistic
241.3363
From these results we see that the intercept is not statistically different from zero, although
the slope coefficient (the Beta coefficient) is highly statistically significant. This suggests
that the regression-through-the-origin model fits the data well. Besides, statistically there is
no difference in the value of the slope coefficient in the two models. Note that the standard
error of the slope coefficient in the regression-through-the-origin model is slightly lower
than the one in the intercept-present model, thus supporting Theil’s argument given in
footnote 4. Even then, the slope coefficient is statistically greater than 1, once again con-
firming that returns on the stocks in the cyclical consumer goods sector are volatile.
By the way, note that the
r
2
value given for the regression-through-the-origin model
should be taken with a grain of salt, for the traditional formula of
r
2
is not applicable for such
models.
EViews,
however, routinely presents the standard
r
2
value even for such models.
EXAMPLE 6.1
(
Continued
)
guj75772_ch06.qxd 11/08/2008 03:50 PM Page 153

---