Escolar Documentos
Profissional Documentos
Cultura Documentos
= 1, 2, 3, ..., n
In matrix form,
=
+
( ) ( )
( )
Objective:
i.
i.
Do (statistical test) individual test
ii.
Joint (overall) test
iii.
Goodness of fit
/0 : & = 0 34 / : & (<, >, )0
j= 1, 2, 3, ... . k
Alternative hypothesis = operational statement of our research agenda
Basis of decision: p-value true probability of incorrect rejection
9 < 0.05 <=>=?@ /0
ii.
iii.
iv.
/0 : =
= = = 0 A 4 C3= DE@DF @E GE H@
/0 = )I-&J = & = -&K L = 1 O
j= 1, 2, ..., k
* : proportion of variation collectively explained Y by Xs
*P adjusted * : remedy for loss of degrees of freedom
o Tobit model
o Heckit model
Discrete Dependent Variable Models
o Count models
Poisson model
Negative binomial model
o Duration models
= + % & & +
&'
) = Prb = 1| W , W
, , W
l mn
(ol mn )p
^(@) =
r
i
) = gjk
mnp
p ; |t| <
l mn
(ol mn )p
Let u=1 + = jt
du= = jt G@
) =
=1
ol mui
l ui
= ) =
l ui vw
i
= -gjk
Kp
) =
sK
l uivw
xi
jxi
= = hi
= ln y
xi
*Link function is a probability distribution function which will link the index from ~ to
[0,1]
Logistic pdf for the logit model
hi
) = ^(@)G@
jk
^(@) =
l mn
(ol mn )p
) =
(Logit Model)
ol ui
l ui
)
_D
= - + % & W&
1 )
&'
The logit of the individual is the linear function of the individuals characteristics
Solved via MLE (theres already an estimation)
Used in forecasting:
^ ) 0.05; success -> , = 1
^ ) 0.05; failure-> , = 1
xi
We will be able to determine the role of success factor of the ith individuals
x
y i z = - + &' & W& (Implicit Function of Differentiation)
jxi
1 )
)
(1 ))
)
)
) (1)
W&
W&
= &
(1 ) )
)
)
)
= & ) (1 ) )
W&
W&
)
= & ) (1 ) ); > = 2,3, , 1,2,3, D C<FDC_ =^^=?@4
W&
Probit Model- Standard Normal Cumulative Distribution Function
Link Function:
^(@) =
r
mnp
p
; < @ <
xi
Marginal Effects =
)
~
1 jhi p
=
=
W& W&
2
),
= ^(e- )
W&
D = D + D where:
D with observations of both Y and xs
D with observations only on the xs (Censored sample)
E.g. amount spent on a car by ith family is linearly dependent on the socio economic status
of the family (W& =jth socio economic variable)
-> OLS will result in Biased and inconsistent estimates
Bias= sample selectivity bias- solution is the procedure developed by James tobin via
sampling from censored normal distribution.
*wage of workers= censored variable
-major substitute for OLS
Given: V , , X; i=1,2n
O = ; marginal density function of the Y and xs with respect to Y.
FOC:
# =0 {=1,2,..k}
SOC:
# p
< 0 ; Hessian
(,i )
(,i )
Ve- X
= D3=<4= __4 <C@E
Ve- X
xi ()
xi ()
= = (w op
-The probability of being in the jth category with respect to the RRR reference/ base
category
Model 4: all parameters are time invariant, the slopes and intercepts vary
across individual microunits.
2. lagged ys
k-> predetermined variables
1
for i=1
t + t = Qt
W
t W1
W
t W1
Qt W1
, = 4@<?@<C_ 9C<C=@=<4
t
1
+
t
t
Qt
=
t
Qt
problem of identification
per equation:
= j ; t = j t
& =
C@ @ = 1,2, . . D; > = 1,2, . .
&
Nationally,
M- # of endogenous var in SEM
K- # of predetermined in SEM
m- # of endogenous variables in the equation
k- # or predetermined variables in the equation
(M+K) (m-k) 1
K-km-1
> over
= exact
Rank condition
an equation is identifiable if and only if there exsist atleast one subdeterminant (M-1) (m-1)
formed by the coefficients of the variable exclude from the equation evaluated.
L. Klein Close Economy Basic Model
equations: Structural model
1.
2.
3.
4.
5.
t = C + t + ? t + G *t + t (consumption function)
Tt = C + t + ? *t + t (investment function)
t = ?t + Tt + t (Identity)
t = C
+
t +
*t + G
)t +
t (Liquidity preference)
t = O + t + t (production function)
Standard form:
t + ) t = t t = t + 3t
Identification issue: how are we going to retrieve the s and s from the s?
Maddala System
Included = 1
Excluded = 0
Eqn#
1
2
3
4
5
6
7
C
1
0
1
0
0
0
0
I
0
1
1
0
0
0
0
N
0
0
0
0
1
1
1
P
0
0
0
1
0
1
1
R
1
1
0
1
0
0
0
Y
1
1
1
1
1
0
0
W
0
0
0
0
0
1
1
G
0
0
1
0
0
0
0
T
1
0
0
0
0
0
0
M
0
0
0
1
0
0
0
K-k
2
3
2
2
3
3
3
m-1
2
2
2
2
1
2
2
Status
Exact
Over
Exact
Exact
Over
Over
Over
1
1
0
0
0
0
0
0
1
1
1
0 0
0 0
1 0
0 0
1 1
1 1
0
1
0
0
0
0
0
0
0
0
0
Simultaneity
Exogeneity
t = + t +
t + t
t = O + O t + 3t
3t = *+ <=4GC_ E^
If - is significant
t = + t +
t + 3t + t
Reject /0 : = 0
Step 1: Get the fitted value (,) of the endogenous variable at right hand side
1
Matrix of endogenous variable coefficients
Eg. t = C _(hl| llt)_ + = t + ^ t + Qt
t = C + t + = t + ^ t + Qt
t = C
+
t +
t +
t + ^
t + Q
t
{t } {Qt }
{ h , t } {t } {Qt }
{
t } {Q
t }
t = O + % tj +
'0
t = + % O& tj& + t
&'0
&'
What you are today is influenced by what you were before plus how the authorities molded you
to be how you are today
Focus: DL(p) model
t = O + % tj + t *(9)
'0
Alt-Tinbergen Method
o Sequential OLS
o Bottom-up
o Will start in simple regression up to complicated model
Hendry Top-down Method
o Will start in big model
o AIC (choose the model p* with smallest AIC
P finite
Almon Model
Koyck model
Adaptive Expectation
Rational Expectation
Infinite DL Models
t = O + % tj + t {(9)
'0
t = O + 0 t + tj + + tj + + t
= 0 = 0, 1,2, 3,
<C@= E^ G=?C (0 1)
1. t = O + 0 t + 0 tj + 0 tj + + 0 tj
+ + t
D^D@= G4@<@3= _CF EG=_
Lag (1) by 1 period multiply the result by , subtract the outcome from (1)
t tj = O (1
)
+ 0 t + t tj
lls
s&Ktlt
ARDL (1, 0)
Three issues:
Autocorrelation
Simultaneity bias
Non-linear in parameter
SEM static/contemporaneous
t = O + 0 t + tj + 3t
Estimation of O , 0 & W: {. 4 E@!
1 3C<V-X
; = 1
G
G = 4@C@4@?
2
Liviathan Method
tj has to be proxied by tj
FOC:
- = -0 -
*.. = %( t O
-0 t - tj )
t'
= 0 normal equation
= % -&
&'0
= 4 E^ 0 + + (?_C@3=)
p
-0
= % -& =
= lim % -& D^D@= 4=<=4 ^D?@ED
k
1
&'0
&'0
Median Lag amount of time lag (lapse of time) within 50% of total effect would be
manifested
Mean Lag amount of time on the average, the total effect would be perceptible
Koyck Model
With = 0 ; = 0, 1, 2, 3,
ARDL (0,1)
Problems:
t = O + % tj + t
'0
t = O + 0 t + tj + 3t
1. O = O(1 )
2. 3t = t tj *(1)
3. tj =DGEF=DE4
t = O + % tj + t
'0
= O0 + % C& t
&'
Basic Concepts
o Application of econometrics to TS data
o Data on variables captured and recorded in regular intervals of time (e.g. annual,
quarterly, monthly, etc.)
o Historical data-frequently/massively available
Autocorrelation
Spurious regression (non-sensical)
Random walk phenomenon
o Tomorrows stock market price, the best prediction is closing price today
random walk forecasting
ARCH effect (Autoregressive Conditional Heterescedasticity) conditional volatility in
stock market
Forecasting
Non-stationarity of most variables
e.g. {t } @ = 1, 2, 3, ,
{, , , }
{t } {t }
If {t } is weakly stationary process, its first 2 moments are said to be time-invariant.
i.e. $(t ) =
3C<(t ) = $(t ) = S
White noise most basic of all stationary SPs building block of all TS models, t
Random walk most basic of all non-stationary stochastic processes, t = tj + t
t
tj
t||ls |
t
|s
= t
= t
tl
l
= G^^=<=D?DF E9=<C@E<;
= 1 {;
= (t tj )
t ~T(0) 4@C@EDC<
t = )Tt
t = TD^t
t = )Tt )Ttj
TD^t = TD^
TD^
t
tj
(0)
G = 0,1,2
t = tj + t
t = t
t = tj + t ; = 1
+ tj + t ;
s|t
|ltl|
Kt||lt
llt
=1
RWM:
t = tj + t
t = (1 )tj + t
t = Otj + t
@=
Thats why they developed
O
DE@ 3C_G
4=(O)
Shortcoming: t = Otj + t
t is highly correlated but Dickey and fuller assumed that it is white noise.
t = tj + % tj + t
'
RWMD:
t = + tj + % tj + t
'
RWMDT:
t = + tj + t + % tj + t
'
Cointegation Analysis
It is a property of 2 or more non-stationary variables to be linked together by a LR equilibrium
relationship
Robert engle and Clive Granger (1982)
Work for cointegration analysis to check if theres spurious regression
Augmented Engle Granger Test (AEG)
If t , t ~T(0) if their SRF
, = - + - t ; t = , ~T(0)
We perform unit root test on both variables and see to it that they are non-stationary.
Stage 2: Run OLS on t CDG t to get SRF and obtain t , ADF t and if t ~T(0) X & Y are
cointegrated
Short-run dynamics
ECM: Error Correction model
Granger Representation Theorem
- If the variable used in regression are cointegrated, an ECM representation
of the LR model is possible.
In k=2:
t = O0 + O t + O tj + t
As are matrices.
j
t = t + tj + % tj + t
'
Johansen
s&Ktlt tl|t
| l|||
lt|
tl|
Johansen more versatile and can be used for more than two variables