Beruflich Dokumente
Kultur Dokumente
Econometrics I, Lecture 3
Yasushi Kondo
1000
0 , 1000 1
0 , 100 1
Chapter 3
Multiple Regression Analysis: Estimation
Primary drawback of simple regression analysis
SLR.4: E
=0
The key assumption SLR.4 often fails.
Omitted variables (other factors included in the error term, u)
are likely correlated with x.
Difficult to draw ceteris paribus conclusions.
,
, unobserved factors in u
Primary interest = effect of
on
+2
if
=0
2016/4/11
1 1
+ +
2 2
log
log
Fitted values
| 1,
2, ,
0, 1, ,
to get OLSE
=1
CEO tenure
Minimize (3.12)
Residuals
=0
=
=
+ +
= 1, ,
= 1, ,
=0
(3.16)
(3.17)
=
=
=0
(3.18)
+
1
+ 2 2 + +
1 + 2 2 + +
1 1
=1
1
=1
if
==
=0
1 1
=0
=1
9
10
Eq. (3.19)
log
= 0.284 + 0.092
+ 0.0041
+ 0.022
11
2016/4/11
(3.22)
1
13
MLR
1 1
SLR
1 1
2 2
and
= 80.12 + 5.52
1 1
2 1
corr
if
= 0 or
=0
15
= 83.08 + 5.86
,
= 0.12
Homework assignment
SST = =1
SSE = =1
SSR = =1
2
2
+ 0.243
16
Goodness-of-fit
on
2, ,
Part of
that is uncorrelated with the other s
after the effects of the other s are partialled out, or
netted out.
on
14
Residual,
=1
1 : residual by reg.
21
1
=1
Sum of residuals = 0
18