The coefficient of determination, R^2 , is the percentage of the variance of the
dependent variable explained by the independent variable. The coefficient of
determination cannot exceed one nor be less than zero. In the case of R^2 = 0,
the regression line’s Y= Yand no variation in the dependent variable is ex-
plained. If the dependent variable pattern continues as in the past, the model
with time as the independent variable should be of good use in forecasting.
The firm can test whether the a^and b
^
coefficients are statistically dif-
ferent from zero, the generally accepted null hypothesis. A t-test is used to
test the two null hypotheses:
H 0 : a^= 0
HA: a^≠ 0
H 0 : b
^
= 0
HA: b
^
≠ 0
The H 0 represents the null hypothesis while HArepresents the alternative
hypothesis. To reject the null hypothesis, the calculated t-value must ex-
ceed the critical t-value given in the t-tables. The calculated t-values for a^
and b
^
are found by:
(5.13)
The critical t-value, tc, for the .05 level of significance with N– 2 degrees of
freedom can be found in a t-table in any statistical econometric text.
If ta> tc, then reject H 01.
If tb> tc, then reject H 02.
The null hypothesis is that =0 can be rejected and therefore is statistically
different from zero. The t-value of bleads to the rejection of =0, and is sta-
tistically different from zero. One has a statistically significant regression
model if one can reject H 02.
We can create 95 percent confidence intervals for aand b, where the
limits of aand bare:
t
a
S
NM
MNX
t
b
S
M
N
a
e
XX
XX
b
e
XX
=
−
+
=
−
ˆ ()
()
ˆ ()
α
β
2