Panel Seemingly Unrelated Regression with Dummy Variables For
Economic Modeling Of Developed And Developing Country
MUHAMAD LISWANSYAH PRATAMA, RAHMA FITRIANI, SUCI ASTUTIK
Department of Statistics,
Brawijaya University,
St. Veteran No.1, Malang, East Java,
INDONESIA
Abstract: - This study aims to identify the relationship between population density, inflation, and
unemployment on the human development index, GNP, export-import, and urbanization in the category of
developed and developing countries using the Panel Seemingly Unrelated Regression (Panel SUR) with a
dummy variable as a slope component. This study uses economic data from 145 countries in the world obtained
through the official websites of the World Bank and the International Monetary Fund. The results showed that
the Fix Effect SUR model (AIC=84915.74) was better than the pooled SUR model (AIC=114936) and Random
Effect SUR (AIC=1148415). The results of the analysis using the Fix Effect SUR model show that population
density has a significant positive relationship with GNP, imports, and exports. There is a significant negative
relationship between the unemployment rate and GNP. In addition, the results obtained show that the effect of
population density on GNP in developed countries is positive and greater. The effect of the unemployment rate
on GNP in developed countries is negative and greater than that of developing countries. The results of the
analysis using the pooled SUR model and the Random Effect SUR are the same conclusion where population
density has a significant positive relationship to GNP, imports, and exports. There is a significant negative
relationship between inflation and GNP. The effect of population density and inflation on GNP, imports, and
exports in developed countries is positive and greater than that of developing countries. The effect of the
unemployment rate on GNP in developed countries is negative and greater than that of developing countries.
Key-Words: - Panel SUR, Pooled SUR, Fix Effect SUR, Random Effect SUR, World Economics, Developed,
and Developing Country.
Received: February 5, 2023. Revised: May 21, 2023. Accepted: June 19, 2023. Published: July 19, 2023.
1 Introduction
Seemingly Unrelated Regression (SUR) is a form of
multivariate regression that can accommodate the
residual correlation between equations (variety
structure between equations) so that each equation
in the model seems to stand alone or there is no
relationship between equations, but actually
between equations has linkage, [1], [2]. However,
there is a problem related to multivariate regression
analysis on economic variables, namely the dynamic
nature of the data so that the data at one point in
time is not sufficient to model the relationship
between each predictor variable to each response
variable. Therefore, data on the variables related to
several time periods for each research object (panel
data) are needed so that the regression model
obtained is better at describing the relationship
between the predictor variables and the response
variables so that a method that can accommodate
these problems is developed, namely multivariate
regression analysis on panel data.
Therefore, the Seemingly Unrelated Regression
Panel model was developed which is able to model
the relationship of one or more predictor variables to
more than one observed response variable from a
research object during a certain period of time on
panel data, [3]. The Seemingly Unrelated
Regression Panel Model is one of the developments
of multivariate regression analysis and panel data
regression analysis that allows statistical users to
analyze cause-and-effect relationships in data
resulting from combining cross-section and time
series data not only on one response variable.
Economic success is something that all
countries in the world want to achieve, especially
developing countries. A country is said to be
developed if it has a high and evenly distributed
economic level, a high standard of living, and
sophisticated technology. Developing countries
have a middle level of social welfare and there is no
economic equality, [4]. The benchmark for
economic success is economic development. Every
WSEAS TRANSACTIONS on ENVIRONMENT and DEVELOPMENT
DOI: 10.37394/232015.2023.19.67
Muhamad Liswansyah Pratama,
Rahma Fitriani, Suci Astutik
E-ISSN: 2224-3496
692
Volume 19, 2023
country must strive for economic development from
year to year so that the country does not have
problems in the economic field and has a prosperous
society. This is still a problem in developing
countries where developing countries continue to
strive for a better country's economy through
various economic actions and policies.
Economic success is a process because it is a
stage that must be lived by every country so it
requires hard work and cooperation between the
community, government, and other elements
involved in the long term so that economic success
can be achieved. Economic success can generally be
seen through 2 aspects, namely the economic aspect
which includes international trade and GNP, and the
social aspect which includes the human
development index and urbanization, [5].
Therefore, this study aims to form a
mathematical model of the seemingly unrelated
regression model and develop it on panel data with
dummy variables and apply it in economic cases,
namely the relationship between HDI, GNP,
Imports, Exports, and Urbanization such as
Population Density, Inflation and Unemployment
Rates in developed and developing countries using
the Pooled SUR, Fix Effect SUR and Random
Effect SUR. Therefore, this study aims to show the
differences in the effect of Population Density,
Inflation, and Unemployment Rate on the Human
Development Index, GNP, Imports, Exports, and
Urbanization in developed and developing countries
so that this research can be beneficial for developing
countries in maximizing economic activities in their
countries as well as aspects what needs to be
improved to have a high economic level like in
developed countries, apart from that this research
can be useful for developed countries to maintain
the economic level in their country.
2 Literature Review
2.1 Seemingly Unrelated Regression (SUR)
Models
The SUR model uses m response variables as a
function of p predictor variables which can be seen
in equation (1), [6].
⁝
(1)
Equation (1) can be simplified into equation (2).
(2)
The form of the SUR equation in the matrix can
be seen in equation (3), [7].
(3)
2.2 Panel Seemingly Unrelated Regression
Models (Panel SUR)
The Panel Seemingly Unrelated Regression model is
an extension of the Seemingly Unrelated Regression
model that can be used in panel data. The Panel
Seemingly Unrelated Regression model was
developed to accommodate the dynamic nature of
data (data patterns always change over time), [3].
The Panel Seemingly Unrelated Regression model is
the same as the Seemingly Unrelated Regression
model (equation (1)) except that there is an element
of time presented in equation (4).
⁝
(4)
According to [3], there are 3 Panel Seemingly
Unrelated Regression models as follows.
a. Pooled Seemingly Unrelated Regression
Models
This model has the same structure as the general
model of Seemingly Unrelated Regression with the
addition of an element of time. The Pooled
Seemingly Unrelated Regression model is presented
in equation (5).
⁝
(5)
WSEAS TRANSACTIONS on ENVIRONMENT and DEVELOPMENT
DOI: 10.37394/232015.2023.19.67
Muhamad Liswansyah Pratama,
Rahma Fitriani, Suci Astutik
E-ISSN: 2224-3496
693
Volume 19, 2023
b. Fix Effect Seemingly Unrelated Regression
Models
In the fixed effects model, it is assumed that there is
a relationship between the characteristics of the
object and the predictor variables for each
response variable. The Fix Effect Seemingly
Unrelated Regression model is presented in equation
(6).
(6)
c. Random Effect Seemingly Unrelated
Regression Models
In the random effects model, it is assumed that there
is no relationship between the object characteristics
and the predictor variables for each response
variable. The object characteristics are assumed to
be random variables combined with random error
terms. The Random Effect Seemingly Unrelated
Regression model is presented in equation (7).
with,
(7)
2.3 Slope Dummy Seemingly Unrelated
Regression (SUR) Models
The Slope Dummy Panel Seemingly Unrelated
Regression model is a development of the Panel
Seemingly Unrelated Regression model which aims
to determine the relationship of one or more
predictor variables to more than one response
variable and can accommodate differences between
categories of dummy variables as follows.
a. Slope Dummy Pooled Seemingly Unrelated
Regression Models
This model has the same structure as the general
Seemingly Unrelated Regression model with the
addition of time elements and dummy variables as
slope components. The Slope Dummy Pooled
Seemingly Unrelated Regression model is presented
in equation (8).
⁝
(8)
b. Slope Dummy Fix Effect Seemingly
Unrelated Regression Models
In the fixed effects model, it is assumed that there is
a relationship between the characteristics of the
object and the predictor variables for each
response variable. The model of Slope Dummy Fix
Effect Seemingly Unrelated Regression is presented
in equation (9).
⁝
WSEAS TRANSACTIONS on ENVIRONMENT and DEVELOPMENT
DOI: 10.37394/232015.2023.19.67
Muhamad Liswansyah Pratama,
Rahma Fitriani, Suci Astutik
E-ISSN: 2224-3496
694
Volume 19, 2023
(9)
c. Slope Dummy Random Effect Seemingly
Unrelated Regression Models
In the random effects model, it is assumed that there
is no relationship between the object characteristics
and the predictor variables for each response
variable. The object characteristics are assumed to
be random variables combined with random error
terms. The Random Effect Seemingly Unrelated
Regression model is presented in equation (10).
⁝
with,
(10)
2.4 Parameter Estimation of Seemingly
Unrelated Regression Model with Ordinary
Least Squares Method
The ordinary least squares method (OLS) is a
method for estimating the parameters of the
Seemingly Unrelated Regression model in equation
(1) by minimizing the sum of the squared errors as
in equation (11).
(11)
Then the sum of the squared errors is derived
from the parameter estimator and equated to zero as
in equation (12) to produce an estimator that is close
to the parameter value.
(12)
So that the regression parameter estimator is
obtained using the ordinary least squares method as
follows, [8].
(13)
2.5 Parameter Estimation of Seemingly
Unrelated Regression Model with
Generalized Least Squares Method
The generalized least squares method (GLS) is one
of the methods for estimating the parameters of the
Seemingly Unrelated Regression model in equation
(1) with the assumption that matrix V (matrix of
variance-covariance residual OLS) is known to be
presented in equation (14), [9].
(14)
with the matrix V as follows.
(15)
In general, the matrix V is not known, so it is
necessary to estimate the matrix V using a two-stage
aiken, such as equation (16).
(16)
WSEAS TRANSACTIONS on ENVIRONMENT and DEVELOPMENT
DOI: 10.37394/232015.2023.19.67
Muhamad Liswansyah Pratama,
Rahma Fitriani, Suci Astutik
E-ISSN: 2224-3496
695
Volume 19, 2023
with,
(17)
and
estimated using the ordinary least
squares method for each equation so that the matrix
V estimator is formed through a single equation
with OLS, [10]. Then a matrix V is formed as in
equation (18)
(18)
The estimator of the SUR model using GLS is
obtained as follows, [1].
(19)
3 Methodology
This study aims to model seemingly unrelated
regression on panel data with dummy variables and
its application in modeling economic success in
developed and developing counties. This study uses
the Seemingly Unrelated Regression Panel (SUR
Panel) model because the five aspects of economic
success (IPM, GNP, Import, Export, and
Urbanization) are related and panel data is used so
that the information obtained is better where it is not
only at one time, rather it can cover multiple time
periods at once. This study uses data on aspects of
economic success (HDI, GNP, Import, Export, and
Urbanization) and the factors that influence it
(population density, inflation, and unemployment
rate) in the world. This study uses balanced panel
data so that it only uses samples that have complete
data where there are 35 developing countries and
110 developed countries obtained from the official
websites of the World Bank and IMF. The data
source properties are presented in Table 1.
Table 1. Data Source
Variables
Period
Data source
Predictor
Population
Density
2010-
2019
World Bank
(https://data.wo
rldbank.org/)
→ Population,
total
Inflation
2010-
2019
World Bank
(https://data.wo
rldbank.org/)
→ Inflation,
consumer
prices (annual
%)
Open
Unemploym
ent Rate
2010-
2019
World Bank
(https://data.wo
rldbank.org/)
→
Unemployment
, total (% of the
total labor
force)
(modeled ILO
estimate)
Response
HDI
2010-
2019
Kaggle
(https://www.k
aggle.com/data
sets/tjysdsg/hu
man-
development-
index)
Import
2010-
2019
World Bank
(https://data.wo
rldbank.org/)
→ Imports of
goods and
services
(current US$)
Eksport
2010-
2019
World Bank
(https://data.wo
rldbank.org/)
→ Exports of
goods and
services
(current US$)
Urbanizatio
n
2010-
2019
World Bank
(https://data.wo
rldbank.org/)
→ Urban
Population (%
of total
population)
WSEAS TRANSACTIONS on ENVIRONMENT and DEVELOPMENT
DOI: 10.37394/232015.2023.19.67
Muhamad Liswansyah Pratama,
Rahma Fitriani, Suci Astutik
E-ISSN: 2224-3496
696
Volume 19, 2023
Dummy
Country
Category
2010-
2019
International
Monetary Fund
(https://www.i
mf.org/external
/pubs/ft/weo/20
22/01/weodata/
groups.htm)
The procedure of this research is as follows.
1. Checking the relationship between predictor
variables (assuming non-multicollinearity) using
VIF statistics
2. Testing pooled SUR model parameters using the
ordinary least squares (OLS) method, fix effects
SUR, and random effects SUR model parameters
using the general least squares (GLS) method
3. Testing the assumption of error correlation
between equations in pooled SUR, fix effects
SUR, and random effects SUR model using the
Lagrange Multiplier test statistic
4. Testing the relationship of the predictor variables
to the response variables of pooled SUR, fix
effects SUR, and random effects SUR model
simultaneously and individually
5. Selecting the best panel SUR model using the
smallest AIC value and the largest adjusted R2
6. Interpretation of the best panel SUR model
4 Result and Discussion
This study uses the pooled SUR model which is
estimated using OLS, Fix Effect SUR which is
estimated using GLS, and Random Effect SUR
which is estimated using GLS as follows.
4.1 Fix Effect SUR
The results of testing the Fix Effect SUR model
with GLS give parameter estimators as shown in
Table 2.
Table 2. Parameter Estimator for Fix Effect SUR
Beta
Estimate
t value
Pr(>|t|)
X1.1
0.003
0.0165
0.9868
X2.1
0
0.00017
0.9999
X3.1
-0.002
-0.00072
0.9994
X1D.1
0.003
0.00018
0.9999
X2D.1
-0.003
-0.00045
0.9996
X3D.1
-0.001
-0.00019
0.9998
X1.2
4.945
30.59201
<0.0001
X2.2
-0.659
-1.39751
0.1623
X3.2
-11.476
-4.06096
<0.0001
X1D.2
35.213
2.02937
0.0425
X2D.2
-10.281
-1.66973
0.095
X3D.2
-17.207
-3.99049
0.0001
X1.3
0.93
5.75196
<0.0001
X2.3
-0.06
-0.12767
0.8984
X3.3
-2.29
-0.81038
0.4178
X1D.3
13.618
0.78483
0.4326
X2D.3
3.732
0.60613
0.5445
X3D.3
-4.575
-1.06105
0.2887
X1.4
0.922
5.70326
<0.0001
X2.4
-0.042
-0.08848
0.9295
X3.4
-2.1
-0.74324
0.4574
X1D.4
21.095
1.21572
0.2241
X2D.4
0.434
0.07042
0.9439
X3D.4
-4.804
-1.1142
0.2652
X1.5
0.203
1.25671
0.2089
X2.5
-0.007
-0.01464
0.9883
X3.5
-0.067
-0.02369
0.9811
X1D.5
0.075
0.0043
0.9966
X2D.5
-0.079
-0.01286
0.9897
X3D.5
-0.01
-0.00225
0.9982
: Significant at α = 5%
Through the table, it can be seen that Fix Effect
Seemingly Unrelated Regression (SUR) models are
as in equation (20).
WSEAS TRANSACTIONS on ENVIRONMENT and DEVELOPMENT
DOI: 10.37394/232015.2023.19.67
Muhamad Liswansyah Pratama,
Rahma Fitriani, Suci Astutik
E-ISSN: 2224-3496
697
Volume 19, 2023
(20)
The intercept for each equation (, , ,
and ) is different for each country because it
uses the Fix Effect SUR model which assumes that
there are characteristics for each country. The
intercept for each country in each equation is
presented in Table 3.
Table 3. The Intercept for Each Country
Country
Albania
0.808
80.418
29.392
25.406
72.283
Algeria
0.712
-838.82
-35.711
-29.187
21.175
Angola
0.544
-614.763
-33.577
-14.171
86.023
Armenia
0.798
113.39
36.13
31.945
-86.489
Australia
0.823
569.63
-23.988
-195.421
75.376
⁝
⁝
⁝
⁝
⁝
⁝
Zambia
0.565
-314.262
-13.743
-10.973
40.904
Zimbabwe
0.534
-321.892
-22.386
-22.08
31.758
Through this model, it can be seen that in
developing countries, every additional population
will increase the human development index, GNP,
imports, exports, and urbanization. Any increase in
inflation will increase the human development
index, but reduce GNP, imports, exports, and
urbanization. Any increase in the unemployment
rate will decrease the human development index,
GNP, imports, exports, and urbanization.
In addition, it can be seen that in developed
countries, each additional population will increase
the human development index, GNP, imports,
exports, and urbanization. Any increase in inflation
will reduce the human development index, GNP,
and urbanization, but increase imports and exports.
Any increase in the unemployment rate will
decrease the human development index, GNP,
imports, exports, and urbanization.
Table 2 shows that population density has a
significant positive relationship with GNP, imports,
and exports. There is a significant negative
relationship between inflation and GNP. The effect
of population density and inflation on GNP,
imports, and exports in developed countries is
positive and greater than in developing countries.
The effect of the unemployment rate on GNP in
developed countries is negative and greater than in
developing countries.
4.2 Random Effect SUR
The results of testing the Random Effect SUR
model with GLS give parameter estimators as
shown in Table 4.
Table 4. Parameter Estimator for Random Effect
SUR
WSEAS TRANSACTIONS on ENVIRONMENT and DEVELOPMENT
DOI: 10.37394/232015.2023.19.67
Muhamad Liswansyah Pratama,
Rahma Fitriani, Suci Astutik
E-ISSN: 2224-3496
698
Volume 19, 2023
Beta
Estimate
t value
Pr(>|t|)
X1.1
0.00050
0.00464
0.9963
X2.1
0.01001
0.00496
0.9960
X3.1
0.04678
0.02141
0.9829
X1D.1
0.00433
0.00356
0.9972
X2D.1
0.15974
0.0072
0.9943
X3D.1
0.00804
0.00143
0.9989
X1.2
6.18306
57.41236
<0.0001
X2.2
-5.77721
-2.86327
0.0042
X3.2
-0.59539
-0.27251
0.7852
X1D.2
27.48431
22.58424
<0.0001
X2D.2
244.05
11.00268
<0.0001
X3D.2
15.39835
2.73385
0.0063
X1.3
1.19832
11.12692
<0.0001
X2.3
-0.85747
-0.42497
0.6709
X3.3
0.73194
0.33501
0.7376
X1D.3
7.79384
6.40431
<0.0001
X2D.3
83.81546
3.77871
0.0002
X3D.3
3.77439
0.67011
0.5028
X1.4
1.20177
11.15896
<0.0001
X2.4
-0.88664
-0.43943
0.6604
X3.4
0.746
0.34144
0.7328
X1D.4
8.51871
6.99994
<0.0001
X2D.4
78.88334
3.55635
0.0004
X3D.4
3.33352
0.59184
0.5540
X1.5
0.03325
0.3087
0.7576
X2.5
0.71171
0.35273
0.7243
X3.5
3.98798
1.82529
0.068
X1D.5
0.44179
0.36302
0.7166
X2D.5
15.70975
0.70825
0.4788
X3D.5
0.37756
0.06703
0.9466
: Significant at α = 5%
Through the table, it can be seen that Random
Effect Seemingly Unrelated Regression (SUR)
models as in equation (21).
(21)
Through this model, it can be seen that in
developing countries, each additional population
will increase the human development index, GNP,
imports, and exports, but reduce urbanization. Any
increase in inflation will reduce the human
development index, GNP, imports, exports, and
urbanization. Any increase in the unemployment
rate will reduce GNP, imports, and exports, but
increase the human development index and
urbanization.
In addition, it can be seen that in developed
countries, each additional population will increase
the human development index, GNP, imports,
exports, and urbanization. Any increase in inflation
will increase the human development index, GNP,
imports, exports, and urbanization. Any increase in
the unemployment rate will increase the human
development index, GNP, imports, exports, and
urbanization.
Table 4 shows that population density has a
significant positive relationship with GNP, imports,
and exports. There is a significant negative
relationship between unemployment and GNP. In
addition, the results show that the effect of
population density on GNP in developed countries
WSEAS TRANSACTIONS on ENVIRONMENT and DEVELOPMENT
DOI: 10.37394/232015.2023.19.67
Muhamad Liswansyah Pratama,
Rahma Fitriani, Suci Astutik
E-ISSN: 2224-3496
699
Volume 19, 2023
is positive and greater than in developing countries
and the effect of the unemployment rate on GNP in
developed countries is negative and greater than in
developing countries.
4.3 Pooled SUR
The results of testing the pooled SUR model with
OLS give parameter estimators as shown in Table 5.
Table 5. Parameter Estimator for Pooled SUR
Beta
Estimate
t value
Pr(>|t|)
X0.1
0.6794
0.021
0.983161
X1.1
-1.7E-06
0
0.999987
X2.1
-0.002
-0.001
0.999237
X3.1
0.000734
0
0.99981
X1D.1
0.001328
0.001
0.999134
X2D.1
0.04745
0.002
0.99834
X3D.1
0.01169
0.002
0.998344
X0.2
67.34
2.092
0.036471
X1.2
6.133
55.645
<0.0001
X2.2
-6.968
-3.325
0.000887
X3.2
-5.159
-1.671
0.094678
X1D.2
27.19
22.201
<0.0001
X2D.2
232.9
10.217
<0.0001
X3D.2
15.76
2.798
0.005152
X0.3
51.14
1.589
0.112155
X1.3
1.161
10.53
<0.0001
X2.3
-1.762
-0.841
0.400458
X3.3
-2.734
-0.886
0.37579
X1D.3
7.568
6.18
<0.0001
X2D.3
75.36
3.306
0.000952
X3D.3
4.049
0.719
0.472242
X0.4
58.67
1.823
0.068397
X1.4
1.158
10.511
<0.0001
X2.4
-1.924
-0.918
0.358485
X3.4
-3.23
-1.046
0.295377
X1D.4
8.259
6.745
<0.0001
X2D.4
69.19
3.035
0.002416
X3D.4
3.648
0.648
0.517159
X0.5
54.21
1.684
0.092155
X1.5
-0.00674
-0.061
0.951218
X2.5
-0.2471
-0.118
0.906112
X3.5
0.3136
0.102
0.919063
X1D.5
0.2021
0.165
0.868902
X2D.5
6.749
0.296
0.767208
X3D.5
0.6684
0.119
0.905539
: Significant at α = 5%
Through the table, it can be seen that Pooled
Seemingly Unrelated Regression (SUR) models are
as in equation (22).
(22)
Through this model, it can be seen that in
developing countries, any increase in population
will reduce the human development index and
urbanization, but increase GNP, imports, and
exports. Any increase in inflation will reduce the
human development index, GNP, imports, exports,
and urbanization. Any increase in the
unemployment rate will reduce GNP, imports, and
exports, but increase the human development index
and urbanization.
WSEAS TRANSACTIONS on ENVIRONMENT and DEVELOPMENT
DOI: 10.37394/232015.2023.19.67
Muhamad Liswansyah Pratama,
Rahma Fitriani, Suci Astutik
E-ISSN: 2224-3496
700
Volume 19, 2023
In addition, it can be seen that in developed
countries, each additional population will increase
the human development index, GNP, imports,
exports, and urbanization. Any increase in inflation
will increase the human development index, GNP,
imports, exports, and urbanization. Any increase in
the unemployment rate will increase the human
development index, GNP, imports, exports, and
urbanization.
Table 5 shows that population density has a
significant positive relationship with GNP, imports,
and exports. There is a significant negative
relationship between unemployment and GNP. In
addition, the results show that the effect of
population density on GNP in developed countries
is positive and greater than in developing countries
and the effect of the unemployment rate on GNP in
developed countries is negative and greater than in
developing countries.
4.4 Compare Pooled, Fix Effect, and Random
Effect SUR
The best model used to model the relationship
between population density, inflation, and the
unemployment rate to the human development
index, GNP, imports, exports, and urbanization is
the model that has the smallest AIC value among
the Pooled SUR, Fix Effect SUR and Random
Effect SUR models.
Table 6. Compare between Pooled SUR, Fix Effect
SUR, and Random Effect SUR
Pooled
SUR
Fix Effect
SUR
Random
Effect SUR
AIC
114936
84915.74
114845.7
Table 6 shows the best model for modeling the
relationship between population density, inflation
and unemployment on the human development
index, GNP, export-import, and urbanization in the
category of developed and developing countries is
the Fix Effect SUR model because Fix Effect SUR
has the smallest AIC value compared to the Pooled
SUR model and Random Effect SUR. In the Fix
Effect SUR model, the parameter estimators are not
as significant as in the Pooled SUR model and the
Random Effect SUR model. However, the Fix
Effect SUR model can accommodate individual
characteristics (countries) as well as the average
difference in the value of the response variable
between countries that are not obtained through the
Pooled SUR and Random Effect SUR models.
4.5 Interpretation of The Best Panel SUR
Model
Table 6 shows that the best model for modeling the
relationship between population density, inflation,
and unemployment to the human development
index, GNP, export-import, and urbanization in
developed and developing countries is the Fix Effect
SUR as in equation (23).
(23)
Through this model, the SUR fix effect model
for developing countries (=0) is obtained as in
equation (24).
WSEAS TRANSACTIONS on ENVIRONMENT and DEVELOPMENT
DOI: 10.37394/232015.2023.19.67
Muhamad Liswansyah Pratama,
Rahma Fitriani, Suci Astutik
E-ISSN: 2224-3496
701
Volume 19, 2023
(24)
Meanwhile, the SUR fixed effect model for
developed countries (=1) is as in equation (25).
(25)
4.6 Discussion
Through subbab 4.4, the best model for modeling
the effect of population density, inflation and the
unemployment rate on the human development
index, GNP, imports, exports, and urbanization is
the fixed effect SUR model because it has the
smallest AIC value compared to the pooled SUR
and random effect SUR models. The AIC values in
the three models are very large due to the large
number of estimators, sample sizes, and time
periods used. In the fixed effect SUR model, more
parameter estimators are used compared to the
pooled SUR and random effect SUR models
because the SUR fix effect model pays attention to
differences in the average response variables for
each country (country characteristics). In addition,
the fixed effect SUR model has the smallest AIC
value when compared to the pooled SUR and
random effect SUR models. This can happen
because the SUR fix effect model is able to
accommodate country characteristics (differences in
the average response variables for each country) and
the parameter estimators are significant as shown in
Table 4, Table 5, and Table 6, which means that
there is a real difference between the average
response variables of one country and the other. In
the SUR fix effect model, a corrected determination
coefficient value of 0.9846 is obtained, which
means that 98.46% of the diversity of response
variable values (human development index, GNP,
imports, exports, and urbanization) can be explained
by predictor variables (population density, imports,
and exports) and 1.54% the rest is explained by
other variables not included in the model. The SUR
fix effect model obtained is presented in equation
(22).
Through the SUR fix effect model in developed and
developing countries, it can be seen that:
For every increase of 1 million population,
the human development index in developing
countries increases by 0.001, while the human
development index in developed countries
increases by 0.006. For every 1 million
population increases, GNP in developing
countries increases by US$ 26.68 billion,
while GNP in developed countries increases
by US$ 41,611 billion. For every 1 million
increase in population, imports to developing
countries increased by US$ 2,952 billion,
while imports to developed countries
increased by US$ 14,633 billion. For every 1
million increase in population, exports to
developing countries increased by US$ 2,657
billion, while exports to developed countries
increased by US$ 12,116 billion. For every
increase of 1 million population, urbanization
WSEAS TRANSACTIONS on ENVIRONMENT and DEVELOPMENT
DOI: 10.37394/232015.2023.19.67
Muhamad Liswansyah Pratama,
Rahma Fitriani, Suci Astutik
E-ISSN: 2224-3496
702
Volume 19, 2023
in developing countries increases by 0.103%,
while urbanization in developed countries
increases by 0.272%.
With every 1% increase in the inflation rate,
the human development index in developing
countries does not change at all, while the
human development index in developed
countries decreases by 0.003. For every 1%
increase in the inflation rate, GNP in
developing countries decreased by US$ 0.055
billion, while GNP in developed countries
decreased by US$ 9.823 billion. For every 1%
increase in the inflation rate, imports to
developing countries decreased by US$ 0.009
billion, but imports to developed countries
increased by US$ 3.767 billion. For every 1%
increase in the inflation rate, exports to
developing countries increased by US$ 0.002
billion, while exports to developed countries
increased by US$ 0.473 billion. For every 1%
increase in the inflation rate, urbanization in
developing countries decreases by 0.009%,
while urbanization in developed countries
decreases by 0.09%.
If the unemployment rate increases by 1%,
the human development index in developing
countries will decrease by 0.002, while the
human development index in developed
countries will decrease by 0.003. If the
unemployment rate increases by 1%, GNP in
developing countries will decrease by US$
10,358 billion, while GNP in developed
countries will decrease by US$ 26,377
billion. If the unemployment rate increases by
1%, imports in developing countries will
decrease by US$ 2,195 billion, while imports
in developed countries will decrease by US$
6,669 billion. If the unemployment rate
increases by 1%, exports to developing
countries will decrease by US$ 2,019 billion,
while exports to developed countries will
decrease by US$ 6,737 billion. If the
unemployment rate increases by 1%,
urbanization in developing countries will
decrease by 0.072%, while urbanization in
developed countries will decrease by 0.087%.
5 Conclusion
The Fix Effect SUR model is the best model to
model the relationship between population density,
inflation, and the number of unemployed on the
human development index, GNP, export-import, and
urbanization in the category of developed and
developing countries (AIC=84915.74). Through the
Fix Effect SUR model, it is found that population
density has a significant positive relationship with
GNP, imports, and exports. There is a significant
negative relationship between unemployment and
GNP. In addition, the results show that the effect of
population density on GNP in developed countries
is positive and greater and the effect of the
unemployment rate on GNP in developed countries
is negative and greater than in developing countries.
Therefore, it is expected that all countries pay
attention to population density and the
unemployment rate because it can affect the
country's GNP, imports, and exports. It is hoped that
developing countries will be able to carry out
population efficiency so that with an increasing
population, the influence on GNP, imports, and
exports will be greater like developed countries
where most of the population of developed countries
has good quality human resources. On the other
hand, it is hoped that developing countries will be
able to add decent jobs to the population to reduce
the unemployment rate so that the influence on
GNP, imports, and exports will be greater like in
developed countries where most of the population of
developed countries are already working and have
decent jobs and salaries.
This study is limited to using balanced panel data
and dummy variables as slope components, so it is
hoped that future research can use unbalanced panel
data so that all countries can be sampled and use
dummy variables as intercept components to
distinguish the average value of the response
variable in the category developed and developing
countries.
References:
[1] Zellner, A, An Efficient Method of Estimating
Seemingly Unrelated Regressions and Tests for
Aggregation Bias, Journal of the American
Statistical Association, Vol.57, No.298, 1962,
pp.348–368,
https://doi.org/10.1080/01621459.1962.104806
64.
[2] Alaba, O. O., Adepoju, A. A., & Olaomi, O. E,
Seemingly Unrelated Regression with
Decomposed Variance-Covariance Matrix: A
Bayesian Approach, Journal of the Nigerian
Association of Mathematical Physics, Vol.51,
2019, pp.137-144.
[3] Xu, X., Šarić, Ž., Zhu, F., & Babić, D,
Accident severity levels and traffic signs
interactions in state roads: a seemingly
unrelated regression model in unbalanced panel
WSEAS TRANSACTIONS on ENVIRONMENT and DEVELOPMENT
DOI: 10.37394/232015.2023.19.67
Muhamad Liswansyah Pratama,
Rahma Fitriani, Suci Astutik
E-ISSN: 2224-3496
703
Volume 19, 2023
data approach, Accident Analysis & Prevention,
Vol.120, 2018, pp.122–129,
https://doi.org/10.1016/j.aap.2018.07.037.
[4] Heshmati, A., Kim, J., & Park, D, Fiscal Policy
and Inclusive Developing in Advanced
Countries: Their Experience and Implications
for Asia, Asian Development Bank Working
Paper Series, Vol.442, 2014.
[5] (text in Indonesian) Kuncoro, M, Ekonomi
Pembangunan. Jakarta: Penerbit Salemba
Empat, 2006.
[6] Moon, H. R. & Perron, B, Seemingly Unrelated
Regressions, The New Palgrave Dictionary of
Economics, 2006, pp.1-9.
[7] Pramoedyo, H., Ashari, A., & Fadliana, A,
Forecasting and Mapping Coffee Borer Beetle
Attacks Using GSTAR-SUR Kriging and
GSTARX-SUR Kriging Models, ComTech:
Computer, Mathematics and Engineering
Applications, Vol.11, No.2, 2020, pp.65-73,
https://doi.org/10.21512/comtech.v11i2.6389.
[8] Dwivedi, T.D., & Srivastava, V.K, Optimality
of least squares in the seemingly unrelated
regression equation model, Journal of
Econometrics, Vol.7, No.3, 1978, pp.391–395,
https://doi.org/10.1016/0304-4076(78)90062-3.
[9] Nisak, S. C, Seemingly unrelated regression
approach for GSTARIMA model to forecast
rainfall data in Malang southern region
districts, CAUCHY: Jurnal Matematika Murni
dan Aplikasi, Vol.4, No.2, 2016, pp.57-64,
https://doi.org/10.18860/ca.v4i2.3488.
[10] Biorn, E, Regression systems for unbalanced
panel data: a stepwise maximum likelihood
procedure, Journal of Econometrics, Vol.122,
No.2, 2004, pp.281–291,
https://doi.org/10.1016/j.jeconom.2003.10.023.
Contribution of Individual Authors to the
Creation of a Scientific Article (Ghostwriting
Policy)
-Muhamad Liswansyah Pratama made
conceptualization, data curation, formal analysis,
investigation, resources, software, supervision,
visualization and writing original draft.
-Rahma Fitriani and Suci Astutik provided funding
acquisition, methodology, project administration,
validation and review and editing.
Sources of Funding for Research Presented in a
Scientific Article or Scientific Article Itself
No funding was received for conducting this study.
Conflict of Interest
The authors have no conflict of interest to declare.
Creative Commons Attribution License 4.0
(Attribution 4.0 International, CC BY 4.0)
This article is published under the terms of the
Creative Commons Attribution License 4.0
https://creativecommons.org/licenses/by/4.0/deed.en
_US
WSEAS TRANSACTIONS on ENVIRONMENT and DEVELOPMENT
DOI: 10.37394/232015.2023.19.67
Muhamad Liswansyah Pratama,
Rahma Fitriani, Suci Astutik
E-ISSN: 2224-3496
704
Volume 19, 2023
