Upload
andrew-kane
View
213
Download
0
Embed Size (px)
Citation preview
Reporting Results, and choosing a functional form.
Hill et al chapter 6.
Explained and Unexplained Variation
1 2 ˆ
ˆ ˆ
ˆ ˆ( )
t t t
t t t
t t t
y b b x e
y y e
y y y y e
TotalExplained Unexplained
Decomposing the variation2 2
2 2
2 2
ˆ ˆ( ) [( ) ]
ˆ ˆ ˆ ˆ( ) 2 ( )
ˆ ˆ( )
t t t
t t t t
t t
y y y y e
y y e y y e
y y e
1. 2( ) ty y = total sum of squares = SST
2. 2ˆ( ) ty y = explained sum of squares = SSR
3. 2ˆ te = error sum of squares = SSE
link to proof
Coefficient of determination
2 1SSR SSE
RSST SST
Closer to 1, the better job we have done in explaining variation in y If 2R=1, SSE=0, and the model fits the data “perfectly.”
If y and x are show no linear association, the fitted line is
“horizontal,” and identical to y, SSR=0 and 2R=0.
When 0 < 2R < 1, it is “the percentage of the variation in y about its
mean that is explained by the regression model.”
Example Sum of Mean Source DF Squares Square Explained 1 25221.2229 25221.2229 Unexplained 38 54311.3314 1429.2455 Total 39 79532.5544
SST = 2( ) ty y = 79532. SSR = 2ˆ( ) ty y = 25221. SSE = 2ˆ te = 54311.
2 1SSR SSE
RSST SST
= 0.317
SSE/(T 2) = 2̂ = 1429.2455
Reporting Results2ˆ =40.7676 0.1283 0.317
(s.e.) (22.1387)(0.0305) t ty x R
2ˆ 40.7676 0.1283 0.317
( ) (1.84) (4.20) t ty x R
t
Choosing a functional form
• Model has been assumed to be linear in the parameters.
• Linear in parameters: parameters are not multiplied together, raised to a power etc.
• variables, however, can be transformed in any convenient way, as long as the resulting model satisfies assumptions SR1-SR5 of the simple linear regression model.
A functional form for food expenditure
It is expected that food expenditure will rise at a decreasing rate with income
Some common functional formsType
Statistical Model
Slope
Elasticity
1. Linear
1 2t t ty x e
2 2
t
t
x
y
2. Reciprocal
1 2
1t t
t
y ex
2 2
1
tx 2
1
t tx y
3. Log-Log
1 2ln( ) ln( )t t ty x e 2t
t
y
x
2
4. Log-Linear (Exponential)
1 2ln( )t t ty x e
2 ty
2 tx
5. Linear-Log (Semi-log)
1 2 ln( )t t ty x e 2
1
tx 2
1
ty
6. Log-inverse
1 2
1ln( )t t
t
y ex
2 2t
t
y
x 2
1
tx
Empirical issues
2ˆ 0.638 0.210 0.649
(0.064) (0.0022) ( . .)t ty x R
s e
An alternative functional form
3 2ˆ 0.874 9.68 0.751
(0.036) (0.824) ( . .)t ty x R
s e
Are the residuals normal (Jarque-Bera)
2
2
3
13
4
14
2
1
3
6 4
ˆ
ˆ
ˆ
T
ii
T
ii
T
ii
kTJB S
eS
ek
e
T
2
2
0
1
22
2.874 3400.3969 1.077
6 4
H : Residuals are normally distributed
H : Residuals are not normally distributed
5.99
JB
Appendix: proof
y y e y e y et t t t t b b x e y e b e b x e y et t t t t t t1 2 1 2
C o n s i d e r t h e t e r m s e t a n d x et t
e y b b x y T b b xt t t t t 1 2 1 2 0
x e x y b b x x y b x b xt t t t t t t t t 1 2 1 22 0
T h e l a s t e x p r e s s i o n s i n e a c h o f t h e s e e q u a t i o n s b e c o m e z e r o f r o m t h e n o r m a l e q u a t i o n s t h a t a r e u s e d t o s o l v e f o r t h e l e a s t s q u a r e s e s t i m a t o r s .
S u b s t i t u t i n g e t = 0 a n d x et t = 0 b a c k i n t o t h e o r i g i n a l e q u a t i o n , w e o b t a i n y y et t 0 .
back