1. ECONOMETRICS
( Case : Indonesia’s Economy)
S3 Public Policy of Trisakti University
Indra Yuspiar, SE, MAk, MPM, CIPM, CFRM, AMA
2. The Content of Presentation
1. What is a good thesis/dissertation?
2. Econometrics and economic research
3. Brief history of econometrics
4. Statistic cycle and other issues in regression
5. Model building in econometric analysis
6. The econometric model evolution
7. Implications in economic analysis
8. Concluding remarks
6. Novelty
1. It is something new or modified from
previous findings.
2. It may create, develop, add, complete, or
give new alternatives of theory, method,
formula, model, or other forms in a
scientific matter.
3. Novelty must be original.
7. Important Aspects Relates to Novelty
1. Literature review
2. State of the art
3. Methods
4. Discussion of the research findings.
5. General scientific statement
10. TheChallenges in Islamic Economic Research
Islamic Finance
Company
Islamic Trust Fund
Islamic Bond
Islamic Capital
Market
Islamic Social Funds
(ZISWaf)
Islamic Banking
Islamic Insurance
Economic
Growth
Poverty
Alleviation
Reducing
Unemployment
Financial System
Stability
Real Economic
Activities
15. 4. Statistic Cycle and Other Issues in Regression
1. Statistical cycle in the hypothesis testing
2. Some factors cause “significant”
3. Regression: tool vs method of analysis
4. The economic model vs econometric models
5. Econometrics as a methodology
16. An example of the theoretical model:
Gt = F (Yt , Nt)
Gt = Gross Investment
Yt= Gross National product
Nt= Nominal cost of capital
Gt* = o + 1 Yt + 2 Nt + e t
17. Statistic cycle:
Gt = F (Yt , Nt)
Gt* = o + 1 Yt + 2 Nt + et
Based on our data, we can estimate o, 1 , . The results of
statistical testing, we conclude that all independent variable
are not statistically significant even at 0,10 level. Why?
He answer is:
It depends on the value of sum of residual (RSS). That is:
e2
n – k
18. The consequences of high RSS in the regression:
1. The value of SE is “too high”
2. Value of t statistic is low
3. Value of F statistic is also low
4. The value of (R2) is also low
5. High probability of autocorrelation and
heteroscedasticity in the regression.
19. th 1 = 1 / SE 1
SE 1 = (Var 1 )1/2
Var 1 =
(Yi2)/(nNi2) x (e2)/n-k)
If t statistic is low
High value of SE,
Low value of t Statistic
High value of variance,
high value of SE
e2 value is high
The results of statistical
testing is insignificant
20. Fh = (b1 gi y1i + b2 gi y2I)/(k-1)
ei2 / (n-k)
R2 = 1 - e2 / Gi
2
Probablys exist due to value of e2
1. F staistic:
2. Value of R2:
3. Classical assumption:
21. An Example: Liniear model
Variable Coefficient Std. Error t-Statistic Prob.
GDP -0.220686 0.158328 -1.393853 0.1773
IR -0.555570 0.390491 -1.422745 0.1688
CPI -4.781396 2.696932 -1.772902 0.0901
C 129.5088 24.78379 5.225546 0.0000
R-squared 0.354875 Mean dependent var 44.88462
Adjusted R-squared 0.494176 S.D. dependent var 12.79790
S.E. of regression 9.102027 Akaike info criterion 7.395509
Sum squared resid 1822.632 Schwarz criterion 7.589063
Log likelihood -92.14162 F-statistic 2.141451
Durbin-Watson stat 2.631298 Prob(F-statistic) 0.897040
Dependent Variable: Investment
Method: Least Squares
Date: 06/10/20 Time: 09:59
Sample: 1969 2005
Included observations: 37
22. Estimate results in Log model
Variable Coefficient Std. Error t-Statistic Prob.
C -2.514534 0.669854 -3.753856 0.0007
LGDP 0.698566 0.099704 7.006393 0.0000
IR 0.100918 0.052175 1.934218 0.0617
LCPI 0.411257 0.121255 3.391671 0.0018
R-squared 0.995533 Mean dependent var 9.585885
Adjusted R-squared 0.995127 S.D. dependent var 1.234014
S.E. of regression 0.086141 Akaike info criterion -1.963846
Sum squared resid 0.244872 Schwarz criterion -1.789693
Log likelihood 40.33115 F-statistic 2451.617
Durbin-Watson stat 0.798455 Prob(F-statistic) 0.000000
Dependent Variable: Log Investment
Method: Least Squares
Date: 06/10/20 Time: 20:56
Sample: 1969 2005
Included observations: 37
23. Estimate in Partial Adjustment model
Variable Coefficient Std. Error t-Statistic Prob.
C -1.484513 0.692522 -2.143634 0.0400
LGDP 0.379049 0.136011 2.786902 0.0090
IR 0.090872 0.046814 1.941151 0.0614
LCPI 0.262595 0.118919 2.208178 0.0348
LINV(-1) 0.395620 0.126874 3.118217 0.0039
R-squared 0.996329 Mean dependent var 9.644796
Adjusted R-squared 0.995856 S.D. dependent var 1.197589
S.E. of regression 0.077095 Akaike info criterion -2.159315
Sum squared resid 0.184252 Schwarz criterion -1.939382
Log likelihood 43.86767 F-statistic 2103.664
Durbin-Watson stat 1.359673 Prob(F-statistic) 0.000000
Dependent Variable: LINV
Method: Least Squares
Date: 06/10/20 Time: 21:10
Sample(adjusted): 1969 2005
Included observations: 36 after adjusting endpoints
24. Note: Partial Adjustment
Model (PAM)
Gt = F (Y t, N t)
Gt* = o + 1 Yt + 2 Nt + e t
Gt = o + 1 Yt + 2 Nt + 3 Gt-1 + v t
Where :
Gt = Actual data
= b1/(b1+b2); 3 = (1- );
o = o / ; 1 = 1 / ; 2 = 2 /
25. 1. As a universal law, economic theory explains long-run phenomena
(long-run equilibrium)
2. Disequilibrium and adjustment of economic agents
3. Anticipated and unanticipated variable
4. Short and long run empirical model (dynamic model)
5. Data stationary and spurious regression
6. The best model criteria
7. The general to specific approach
5. Model Building in Econometric Analysis
26. Econometrics as a Method helps us to reveal:
• Adjustment process
• Adaptive expectation
• Efficiency
• Effectivity
• Elasticity
• Market competition
• Shock variables
• Disequilibrium situation
• Asymmetric effects
• etc
27. 6. Econometric Models Evolution
Linier
Regression
The model
is not valid
to apply in
the
economic
research
now days.
Simple
Dynammic
Regression
Partial
Adjustment,
Adaptive
Model, Shock
Absrober,
autoregressive,
simple dynamic
model. Etc
Advanced
Dynamic
Econometrics
Autoregresive
Model, Moving
Average, Box
Jenkins,
GARCH,Cointeg
ration, ECM,
VAR, VECM,
autoregresive
model.
Simple Panel
Data
Analysis,
Modified
VECM and
other
dynamic
models
Dynamic Panel
Data Analysis
Cointegration
and VECM for
Panel Data
1926 1960 1990 2000 2010 2017
1. Bootstraping
Regression
2. Fuzzy
Econometrics
28. Regression
Cross Section
Time Series Panel Data
A Theoritic
ADL
Cost/loss
Functon
Dynamic Model
Dynamic Panel Data:
Cointegration-Vector Error Correction Model
(VECM)
PAM (1967)
SAM (1980)
OE-ECM (2006)
OESAM (1982)
ECM (1987)
N-ECM (1990)
Theoritic
Special Case for
Cross Section Data
(SEM.......)
AR, MA, ARMA,
ARIMA, ARCH-
GARCH, BOX-
JENKINS.......
Fuzzy, Boostraping Cointegration-VECM
for Panel Data
VAR
Econometric Models Based on Data Scheme
31. Independent Variables Dependent Variable
DLGEO DLGED DLGET
Constanta
DLGDP
DLPOP
LGDP (-1)
LPOP (-1)
DSHOCKDB
SHOCKDB (-1)
SHOCKDB (-2)
SHOCKDB (-3)
ECT
-2.832 (-2.36) b
0.949 (2.90) a
-0.124 (-0.06)
0.461 (1.99) b
-1.103 (-1.98) b
0.019 (3.97) a
0.027 (4.15) b
-
-
0.151 (2.61) b
-12.727 (-2.54) b
2.731 (2.95) b
-1.976 (-0.33)
2.032 (2.27) b
-4.732 (-2.24) b
0.070 (4.73) a
0.103 (3.76) a
0.012 (0.671)
0.028 (1.89) c
0.499 (3.23) a
-5.407 (-3.03) a
1.569 (3.94) a
-1.304 (-0.41)
0.798 (2.50) b
-1.832 (-2.45) b
0.034 (5.82) a
0.051 (5.34) a
0.004 (0.69)
0.013 (2.15) b
0.344 (4.14) a
F
R2
4.79308
0.56340
4.06227
0.64232
6.628474
0.73057
Autokorelasi: 2 = 3.46 (0.18) 2 = 0.52 (0.77) 2 = 1.16 (0.55)
(Breusch-Godfrey)
Heteroskedastik 2 = 16.04 (0.31) 2 = 21.7 (0.24) 2 = 14.3 (0.71)
(White test)
Test Spesifikasi LR = 1.96 (0.16) LR = 0.63 (0.42) LR = 1.41 (0.23)
(Ramsey test)
32. Example 2: Vector Error Correction Model (VECM)
The model is GE= f (REV)
VECM model:
REVt = 11 + 12 GEt-1 + … + 1i GEt-i + 11 REVt-1
+ … + 1i REVt-i + 1 ECTt + v t
GEt = 21 + 22 GEt-1 + … + 2i GEt-i + 21 REVt-1 + … +
2i REVt-i + 2 ECTt + v t
33. Model Chi Square restriction test
(lag=1):
Coefficient
ECT
∆ LTAX ∆ LGET
∆ LTAX - 0.958 -0.299
(0.327) (-3.568) a
-
∆ LGET 5.052 -0.097
(0.024) b (-1.184)
Keputusan pada model jangka pendek:
Keputusan pada model jangka panjang:
LTAX
LTAX
LGET
LGET
Bivariate Causality Test Tax -Expenditure
34. 0 .0 4
0 .0 6
0 .0 8
0 .1 0
0 .1 2
3 4 5 6 7 8 9 1 0
1 2
T a h u n k e
R e s p o n s e o f L G E T t o O n e S . D . L T A X I n n o v a t i o n
35. 0 . 3 0
0 . 2 5
0 . 2 0
0 . 1 5
0 . 1 0
0 . 0 5
0 . 0 0
3 4 5 6 7 8 9 1 0
1 2
T a h u n k e
R e s p o n s e o f L T A X t o O n e S . D . L G E T I n n o v a t i o n
36. 3
2
1
0
- 1
4
3 4 5 6 7 8 9 1 0
G G D P
L G E T
L M 3
L T A X
1 2
T a h u n k e
R e s p o n s e o f G G D P t o O n e S . D . I n n o v a t i o n s
37. - 0 .1 0
- 0 .0 8
0 . 0 4
0 . 0 2
0 . 0 0
- 0 .0 2
- 0 .0 4
- 0 .0 6
1 2 3 4 5 6 7 8 9 1 0
L T A X
L G E O
L G E D
L G D P
R e s p o n s e o f L P I t o O n e S . D . I n n o v a t i o n s
T a h u n k e