Prof. Kiseok Lee ([email protected])
TA Junyong Kim ([email protected])
Lecture Note 1
1
Econometrics
Econometrics is Economic Measurement, measurement in terms of statistics and mathematics.
Why should we learn this subject?
1
Economic theories may b
Prof. Kiseok Lee ([email protected])
TA Junyong Kim ([email protected])
Lecture Note 3
1
Random Property of
OLS estimators of regression model is
=
=
We can transform the regression model (2.1) through differencing.
=
+
=
+
+
=
=
+
(
=
+
)+
+
+
Note
Prof. Kiseok Lee ([email protected])
TA Junyong Kim ([email protected])
Lecture Note 2
1
Regression Model
1.1
Two Variable Regression Model
Definition
Simple Regression Model; suppose we have a regression model (2.1)
=
( )
=
( | )
=
+
(2.1)
+
Where
:
Prof. Kiseok Lee ([email protected])
TA Junyong Kim ([email protected])
Lecture Note 4
1
Standardization of
Assumption (Homoscedasticity)
mean 0 and variance
= cfw_1,2,3, , ,
are distributed as normal with
, i.e.
(0,
~
)
To reflect above property, Pr
Prof. Kiseok Lee ([email protected])
TA Junyong Kim ([email protected])
Lecture Note 6
1
Coefficient of Determination
1.1
R-square
=
+
=
+
+
(6.1)
(6.2)
Through difference (6.2) from (6.1) below equation can be constructed.
( )
With
=
=
(
)+
and
=
=
Prof. Kiseok Lee ([email protected])
TA Junyong Kim ([email protected])
Lecture Note 5
1
Confidence Interval
The
statistic
=
~
Where
that is
is the null hypothesis and
= , the above random variable
freedom. Note that the variable
=
under the null hypo
Prof. Kiseok Lee ([email protected])
TA Junyong Kim ([email protected])
Lecture Note 8
1
Partial Autocorrelation
In general, the AR(p) model can be represented as
=
+
+ +
+
All AR models have the autocorrelation functions that are hard to identify the
Prof. Kiseok Lee ([email protected])
TA Junyong Kim ([email protected])
Lecture Note 7
1
Heteroskedasticity
1.1
Heteroskedasticity Problem
Suppose the following model
=
Where
+
+
has mean 0 and variance
(Figure 3.1.3)
Classical Assumption (3) is
(
(3.
Prof. Kiseok Lee ([email protected])
TA Junyong Kim ([email protected])
Lecture Note 9
1
Unit Root
MA(1) process and MA(2) process are
=
+
=
+
+
AR(1) process is
=
+
=
+
=
+
This AR(1) process can be transformed into MA() as
=
+
+
+
In the case of MA(
Prof. Kiseok Lee ([email protected])
TA Junyong Kim ([email protected])
Lecture Note 13
1 Structure Change
Consider the model Vt = 1,2,- .71
Yt = a + ,BXt + ut
If the model experiences an abrupt change in parameter values at a certain point in time 15
Prof. Kiseok Lee ([email protected])
TA Junyong Kim ([email protected])
Lecture Note 11
1
Time-series Analysis
Time-series analysis is consist of 3 steps as
(1) Identification (of the model)
(2) Estimation (of the model parameters)
(3) Diagnostic chec
Prof. Kiseok Lee ([email protected])
TA Junyong Kim ([email protected])
Lecture Note 12
1 Conditional Heteroskedasticity
1.1 ARCH Model
Consider the following model
Yt = a+Xt+ut
ut ~i.i.d. N(00.2)
ht = (utfl-Qt 1)
= yo + huh (12.1)
[2t_1 : The
Prof. Kiseok Lee ([email protected])
TA Junyong Kim ([email protected])
Lecture Note 10
1
Durbin-Watson d Test
Suppose the model
=
+
+
Assumption (Durbin-Watson d Test)
DW d test assumes
(1) Dependent variables are fixed non-random. Cf. Classical Assu