Download Econometrics

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Document related concepts
no text concepts found
Transcript
Econometrics
Jerzy Mycielski
2013
Jerzy Mycielski ()
Econometrics
2013
1 / 14
Topics
Statistical inference
Most important distributions of random variables
Assumptions of CNLRM
Distributions of b and e
t statistics
Simple and joint hypothesis, con…dence intervals
Joint hypothesis and F statistics
Constrained minimization and second form of F statistics
Jerzy Mycielski ()
Econometrics
2013
2 / 14
Normal and related distributions
Linear function (combination) of random variables with multivariate
normal distribution has multivariate normal distribution
z s N (µ, Σ)
a, B nonrandom
a +Bz s N a + Bµ, BΣB0
If z s N (0, Σ) and Σ is nonsingular matrix (r
z0 Σ
1
r ), then
z s χ2r
If z s N 0, σ 2 I and matrix A is nonrandom idempotent matrix and
Rank (A) = r , then
z0 Az
s χ2r
σ2
Jerzy Mycielski ()
Econometrics
2013
3 / 14
Normal and related distributions
if random variable Z s N (0, 1) and random variable W s χ2r and Z
and W are independent, then
Z
q
W
r
s tr
If random variable Z s χ2k and random variable W s χ2r and Z and
W are independent, then
Z
k
W
r
Jerzy Mycielski ()
s F (k, r )
Econometrics
2013
4 / 14
Hypothesis testing and con…dence intervals
α signi…cance level - probability of type I error (rejection of true H0 )
kα critical value for test statistics at signi…cance level α
F (kα ) = 1
α
k value of the test statistics in the sample
α signi…cance level calculated on the basis of the sample
α =1
F (k )
α is called p (p-value)
Value of k
Value of α
Decision
k > kα
k < kα
α <α
α >α
Rejection of H0
Cannot reject H0
Jerzy Mycielski ()
Econometrics
2013
5 / 14
Rejection of null hypothesis
k > kα , α < α =) H0 rejected
Jerzy Mycielski ()
Econometrics
2013
6 / 14
Non rejection of null hypothesis
k < kα , α > α =) H0 not rejected
Jerzy Mycielski ()
Econometrics
2013
7 / 14
Hypothesis testing and con…dence intervals
Simple hypothesis
Regression for Cobba-Douglas model
-----------------------------------------------------------------------------q j
Coef.
Std. Err.
t
P>jtj
[95% Conf. Interval]
-------------+---------------------------------------------------------------l j
1.236623
r j
.0080266
.017226
0.47
0.644
-.026944
.0429972
t j
.0106685
.0014495
7.36
0.000
.0077258
.0136112
_cons j
-5.31012
3.678503
-1.44
0.158
-12.77788
2.157639
.3801312
3.25
0.003
.4649154
2.00833
------------------------------------------------------------------------------
Hypothesis of constant returns to scale βl = 1
t=
Jerzy Mycielski ()
1.236623 1
= . 622
.3801312
Econometrics
2013
8 / 14
Con…dence intervals for forecast and for forecast error
Con…dence interval for forecast error based on Philips curve
2. 523
2.0226 = 12. 220
17. 323 + 2. 523
2.0226 = 22. 426
25
20
bezrobocie
10
15
5
0
0
5
bezrobocie
10
15
20
25
17. 323
0
10
20
30
40
0
10
inflacja
95% przedz. ufności
wart. dopas.
wart. zaob.
Jerzy Mycielski ()
20
30
40
inflacja
95% przedz. ufności
wart. dopas.
wart. zaob.
Econometrics
2013
9 / 14
Joint hypothesis testing
Example (Invalid formulation of joint hypothesis)
For β = [ β1 , β2 ]0 , Hβ = h
β1 + β2 = 1
β1 + β2 = 2
1 1
in this case H =
,h=
1 1
Jerzy Mycielski ()
1
2
Econometrics
2013
10 / 14
Formulating the joint hypothesis
Model for the log of wages
7
ln (wagei ) = β0 + β1 sexi + β2 agei +
∑ γs Ds ,i + εi
s =2
Hypothesis: join insigni…cance of all variables apart form constant
β1 = β2 = γ1 = γ2 = γ3 = γ4 = γ5 = γ6 = 0
Test result: R 2 = 0.2975, N = 6509, K = 9:
F =
6500 0.2975
= 344. 56
8 1 0.2975
p-value:
α =1
Jerzy Mycielski ()
1
F8,6500
(344. 56) t 0.0000 < α = 0.05
Econometrics
2013
11 / 14
Formulating the joint hypothesis
Hypothesis: education is not signi…cant
γ1 = γ2 = γ3 = γ4 = γ5 = γ6 = 0
Model with constraints
ln (wagei ) = β0 + β1 sexi + β2 agei + εi
Test result: SR = 1771.145. S = 1342.156.
F =
(1771.145 1342.156) 6
= 346. 26
1342.156/ 6500
p-value:
α =1
Jerzy Mycielski ()
1
F6,6500
(346. 26) t 0.0000 < α = 0.05
Econometrics
2013
12 / 14
Formulating the joint hypothesis
Hypothesis: only secondary and higher education is signi…cant
8
γ =0
>
>
< 6
γ5 = 0
γ = γ3
>
>
: 4
γ3 = γ2
Model with constraints
ln (wagei ) = β0 + β1 sexi + β2 agei + γ4 D4,i
+γ4 D5,i + γ4 D6,i + γ7 D7,i + εi
grouping the variables we obtain:
ln (wagei ) = β0 + β1 sexi + β2 agei +
γ4 (D4,i + D5,i + D6,i ) + γ7 D7,i + εi
Jerzy Mycielski ()
Econometrics
2013
13 / 14
Formulating the joint hypothesis
De…ne
secondaryi
higheri
= D4,i + D5,i + D6,i
= D7,i
resulting model
ln (wagei ) = β0 + β1 sexi + β2 agei + γ4 secondaryi + γ7 higheri + εi
SR = 1346.927
F =
(1346.927 1342.156) 4
= 5. 78
1342.156/ 6500
p-value:
α =1
Jerzy Mycielski ()
1
F4,6500
(5. 78) t 0.0001 < α = 0.05
Econometrics
2013
14 / 14
Related documents