Journal of Economics and Sustainable Development www.iiste.

org
ISSN 2222-1700 (Paper) ISSN 2222-2855 (Online)
Vol.9, No.16, 2018

Analysis of Economic Development Disparities Between Districts

in North Sumatra
Sri Rosliana Lubis
Lecture, Akademi Pariwisata Medan, North Sumatera, Indonesia

Abstract
This study aims to determine and analyze the effect of GDP, population, development expenditure and
unemployment rate on the imbalance of economic development between districts / cities in North Sumatra. Data
were taken from 2010 - 2017 in 33 districts throughout northern Sumatra. The data collected were analyzed
using panel data regression. From the test data using Chow test and Haussmann test then the panel data
regression used is fixed effect method. Collectively the GDP variables, population, development expenditure and
unemployment rate affect the variables of economic inequality with R-square of 0.9949. GDP has a negative and
significant effect; the number of influential population and development expenditure has a positive and
significant effect on the disparity of economic development while the unemployment rate has no significant
effect on economic development in North Sumatra. To reduce the inequality or disparity of economic
development, the government should continue to improve the performance of the economy by finding new
sources of revenue and budget efficiency that can be used for development in every sector.
Keywords: Economic Disparity, Fixed Effect Method, Panel Data Regression

Introduction
In general, national development in developing countries is focused on economic development through
economic growth efforts. Where the goal of multidimensional economic development is to create growth and
changes in economic structure, social change, reducing or eliminating poverty, reducing disparity, and
unemployment (Todaro, 2000). The main cause of inequality is the difference of socio-economic structure and
other factors.
The results of the Easterly (1999) study reveal that high levels of disparities are a barrier to prosperity, the
growth of quality institutions, and the development of high quality education.
In North Sumatra if economic growth is seen between districts / cities, it will be seen how much inequality
that occurs from the economic side. Following in Figure 1 economic growth between districts / cities in North
Sumatra.

Source: BPS, North Sumatera In Figures Year 2017

Figure 1. Economic Growth of Regency / City of North Sumatra Year 2012 (percent)

1
Journal of Economics and Sustainable Development www.iiste.org
ISSN 2222-1700 (Paper) ISSN 2222-2855 (Online)
Vol.9, No.16, 2018

From Figure 1 above can be seen that economic growth in North Sumatra is relatively high, but the growth
is accompanied by an increasingly large regional inequality. High economic growth has often led to a widening
inequality between the poor (the rich and the poor) and the inequality or disparity between regions (advanced
and left behind). The increasing inequality between groups and between these areas can lead to problems of
social jealousy, the vulnerability of regional disintegration and the wider and sharper economic disparities.
In terms of demography, large population is an asset for regional development, but large population will
also be a burden for the government, especially local governments when the human resources are low. The
uneven distribution of the population will also be a dilemma for the government when the population is
concentrated in one region that will cause inequality. The centralized population will tend to be a burden for
local governments to provide employment opportunities that can absorb the labor force in the region. In contrast,
a relatively small number of people will also be an obstacle for regional development.

Literature Review
If viewed historically, the concept of the division of the region itself was first proposed by Tiebout (1956) in an
article entitled "A Pure Theory of Local Expenditure". It is argued that regional expansion is analogous to a
model of perfectly competitive economy where local government has the power to maintain low tax rates,
provide efficient services, and permit each individual community to express its preference for each type of
service from different levels of government with "vote with their feet ". Then Swianiewicz (2002) reveals that
small local communities are more homogeneous, and easier to implement policies that suit the preferences of
most societies. The opportunity of the community to participate in small communities has a greater chance.
Furthermore, small local governments have low bureaucracies, such as administrative functions. In this case
Hofman et al (2005) said the division of the region is intended to bring public services closer to its constituents.
Therefore, among districts / cities there are different needs of public services with different characteristics, then
urban area should be expanded from the parent regency so that each region can specialize in the provision of
public services in accordance with the characteristics of the needs of the community.
The positive impact of regional expansion was also presented in the study conducted by Percik (2007). That
in New Autonomous Region (DOB) Bengkayang, Bombana and Wakatobi districts in the early years of their
administration, they did not prioritize to develop PAD, but prioritized for infrastructure development. Although
the division of the region is considered to bring the government closer to the community, Kerlin (2002) suggests
that the goal of improving the equally important administrative efficiency is not achieved. This is where there is
a failure to achieve the goal of regional expansion policy. Studies conducted by do'Carmo and Martinez-Vazquez
(2001) in the Czech Republic have revealed the failure of the division.
In addition to economic growth, there are also several factors affecting regional inequality. Research
conducted by Akai-Sakata (2005) and Lessman (2006) looks for factors that affect the level of regional
imbalances, including economic growth, agglomeration, and the number of people employed. Agglomeration is a
grouping of economic activities, generally homogeneous, the place. Agglomeration of economic activity
somewhere will cause the region to experience higher economic growth. Agglomeration of economic activities
for an industry will also be able to spur development in a region through the mechanisms of job creation and
improvement of people's welfare.
Previous research conducted by Jaime Bonet (2006), in which Bonet analyzed the effect of agglomeration
of production variables on regional inequality. The result of Bonet research shows that between the
agglomeration of production and the regional income imbalance there is a positive and significant relation at α =
1%. It means that every level of production agglomeration will increase the inequality of regional income.
One of the causes of regional imbalances is the difference in geographic conditions between regions.
Demographic conditions of a region include differences in growth rates and population structure, different levels
of education and health, differences in labor conditions and differences in work ethics held by the local
community concerned. Demographic conditions affect the magnitude of productivity of a region. Areas that have
good demographic capabilities will be able to have high productivity. Levels of people working affect the
productivity of a region, the higher the level of people who work in an area will cause the productivity of the
area is higher than the area with the number of people who work less. According to Lessman (2006), high
unemployment rates have an effect on the higher regional imbalances.

Research Method
The location of research conducted by North Sumatera Province from 2010 until 2017. Data type in this research
is quantitative data and data source used is secondary data. The data source used in this research is sourced from
the Central Board of Stastistic Province of North Sumatra in several publications. This research used panel data
analysis to know the influence of PDRB variable, population variable (POP), development expenditure (GE),
and workforce variable (AK) to economic development inequality between regency / city (VW) in Sumatera
Province North. The panel data model is:

2
Journal of Economics and Sustainable Development www.iiste.org
ISSN 2222-1700 (Paper) ISSN 2222-2855 (Online)
Vol.9, No.16, 2018

VWit = βo + β1 GDPit + β2 POPit + β3 GEit + β4 AKit + ɛ

The function models that will be used to know the imbalance of economic development among regencies in
North Sumatera Province are:
0 1 2 3 4
Where :
VW = Economic Inequality Index (percent)
GDP = Revenue per capita (million rupiah)
POP = Population (person)
GE = Development Expenditure (IDR)
AK = Labor Force (Person)
i = Cross Section: i = 1; , 2,3, ......., 28 District
t = Time series t = 2010 - 2017
β = Coefficient of Free Variable
µ_i = Fixed effect or random effect
µ_it = residual or error
Given the data used in this study is panel data, then to test the hypothesis used the model of Fixed Securities
and Random Effects (Greene, 2000). Secondary data processing and application of the three methods above
using the statistic (software) program Eviews version 6.0.

Result and Discussions

From the calculation result using williamson index method, it can be seen that disparity in North Sumatera
province during 2010 until 2017 tends to fluctuate. In 2005 the index of williamson North Sumatra of 0.0523
decreased relative small in 2012 by 0.0503. Table 1 below presented williamson index of districts in North
Sumatra province during 2010-2017.
Table 1. Index of Williamson District / City of North Sumatra Year 2010 – 2017
No District 2010 2011 2012 2013 2014 2015 20116 2017
1 Nias 0.0957 0.0936 0.0918 0.0953 0.1062 0.0578 0.0578 0.0573
2 Madina 0.0811 0.0871 0.0867 0.083 0.0815 0.0797 0.0799 0.0793
3 Tapsel 0.0951 0.0918 0.0945 0.0377 0.037 0.0371 0.0385 0.0392
4 Tapteng 0.0846 0.0879 0.0893 0.0889 0.0897 0.0896 0.0898 0.0905
5 Taput 0.0469 0.0447 0.047 0.0519 0.0519 0.0539 0.0549 0.055
6 Tobasa 0.0222 0.0162 0.0165 0.0119 0.0132 0.0134 0.0114 0.0115
7 Labuhan Batu 0.0092 0.0037 0.0017 0.0025 0.0221 0.0251 0.0265 0.0275
8 Asahan 0.1167 0.113 0.0257 0.0247 0.0249 0.0266 0.0291 0.0296
9 Simalungun 0.0667 0.0677 0.0686 0.0673 0.0649 0.0639 0.0654 0.0653
10 Dairi 0.0179 0.02 0.0208 0.0247 0.0239 0.0244 0.026 0.0261
11 Karo 0.0246 0.013 0.0083 0.0111 0.01 0.0082 0.0068 0.0075
12 Deli Serdang 0.0062 0.0139 0.0234 0.0304 0.035 0.0419 0.044 0.0489
13 Langkat 0.0485 0.0604 0.0641 0.054 0.0524 0.0504 0.0523 0.0515
14 Nias Selatan 0.0785 0.0703 0.0705 0.0755 0.0756 0.0799 0.0814 0.0813
15 Humbang Has 0.0334 0.0312 0.0311 0.0384 0.0384 0.0412 0.0417 0.0419
16 Pakpak Barat 0.0265 0.0259 0.0298 0.0309 0.0314 0.031 0.0311 0.0314
17 Samosir 0.0106 0.0101 0.0111 0.0058 0.0041 0.0031 0.0037 0.0037
18 Sergei 0.0424 0.0431 0.0455 0.0407 0.0373 0.0347 0.0359 0.036
19 Sibolga 0.0095 0.011 0.0119 0.0055 0.0041 0.0033 0.0129 0.005
20 Tanjung Balai 0.0053 0.0026 0.0013 0.0016 0.0005 0.0011 0.003 0.0042
21 Pematang Siantar 0.0076 0.0073 0.0082 0.0082 0.0069 0.0066 0.0074 0.007
22 Tebing Tinggi 0.0099 0.0098 0.0101 0.013 0.0123 0.0129 0.0129 0.0124
23 Medan 0.301 0.3172 0.3272 0.3265 0.3389 0.3492 0.3562 0.3733
24 Binjai 0.0135 0.0146 0.0162 0.0148 0.0138 0.014 0.0145 0.0142
25 Padang Sidimpuan 0.0533 0.0537 0.0544 0.0049 0.0541 0.0565 0.057 0.0588
Estimation using fixed effect method based on chow test result and haussman test result, where chow test
result is shown in Table 2. below:

3
Journal of Economics and Sustainable Development www.iiste.org
ISSN 2222-1700 (Paper) ISSN 2222-2855 (Online)
Vol.9, No.16, 2018

Table 2. Chow Test Results

Redundant Fixed Effects Tests
Pool: Untitled
Test cross-section fixed effects
Effects Test Statistic d.f. Prob.
Cross-section F 503.940360 (24,146) 0.0000
Determination of use between method of common effect and fixed effect by using chow test, where if
probability> 0.05 Conversely, if the probability is < 0.05, then the use of fixed effect method is better used.From
result of estimation with chow test obtained probability <0.05 so that method of fixed effect better to use.The
next stage is to test the haussman to determine whether a better fixed effect or random effect method is used.
Haussman test results can be seen in table 3 below:
Table 3. Haussman Test Results
Correlated Random Effects - Hausman Test
Pool: Untitled
Test cross-section random effects
Test Summary Chi-Sq. Statistic Chi-Sq. d.f. Prob.
Cross-section random 37.562195 4 0.0000
Determination of use between method of fixed effect and random effect by using haussman test, where if
probability> 0.05 then random effect method better use. Conversely if probablity <0.05, then the use of fixed
effect method is better used. From result of estimation with haussman test obtained probability <0,10,0,0,0,0 so
that method of fixed effect better to use.

4
Journal of Economics and Sustainable Development www.iiste.org
ISSN 2222-1700 (Paper) ISSN 2222-2855 (Online)
Vol.9, No.16, 2018

Table 4. Estimation Result of Fixed Effect Method

Dependent Variable: LOG (VW?)
Method: Pooled EGLS (Cross-section weights)
Variable Coefficient Std. Error t-Statistic Prob.
C -9.782287 1.919361 -5.096637 0
LOG(GDP?) -0.177096 0.050583 -3.501086 0.0006
LOG(POP?) 0.57342 0.140704 4.075369 0.0001
LOG(GE?) 0.036701 0.019571 1.875256 0.0628
LOG(TPT?) 0.014832 0.016386 0.905152 0.3669
Fixed Effects (Cross)
_NIAS—C 1.028671
_MADINA—C 0.947268
_TAPSEL—C 0.553278
_TAPTENG—C 1.103665
_TAPUT—C 0.687944
_TOBASA—C -0.384188
_LBATU—C -1.292764
_ASAHAN—C -0.107045
_SIMALUNGUN--C 0.483473
_DAIRI--C -0.057431
_KARO--C -1.07417
_DSERDANG--C -0.52706
_LANGKAT--C 0.231886
_NISEL--C 1.031321
_HUMBAHAS--C 0.602904
_PAKPAKB--C 0.876519
_SAMOSIR--C -1.205839
_SERGAI--C 0.111041
_SIBOLGA--C -0.857669
_TBALAI--C -2.453909
_PSIANTAR--C -1.148807
_TTINGGI--C -0.466664
_MEDAN--C 1.881713
_BINJAI--C -0.49247
_PSIDEMPUAN--C 0.528333
Effects Specification
Weighted Statistics
R-squared 0.994995 Mean dependent var -15.30609
Adjusted R-squared 0.994035 S.D. dependent var 16.64737
S.E. of regression 0.389273 Sum squared resid 22.12385
F-statistic 1036.513 Durbin-Watson stat 1.221867
Prob(F-statistic) 0
The F-value value is equal to 1036,513 with F-statistical probability of 0.000 which means together
independent variables (GDP, POP, GE, and TPT) affect the dependent variable (DV). The estimation results have
met the suitability test model for the simultaneous test, so the estimation results can be used for the analysis.
The model is said to be better if R2 gets closer to 1. The model estimate yields R2 of 0.9949. That is, the
existence of independent variables (GDP, POP, GE and TPT) is able to explain the dependent variable (DV) of
99.49 percent, the rest of which is 0.51 percent explained by other variables outside the model. With R2 = 0,9949
then result of estimation fulfill conformity test from coefficient aspect of determination. The estimation result is
feasible to be analyzed.
Partial test is also called the test of significance. The GDP t-value is equal to -3.501086 with the probability
of 0.0006 smaller than α = 0.05 which means the GDP variable significantly affects the DV negatively, the POP
t-statistic value is equal to 4.075369 with the probability of 0, 0001 smaller α = 0,05 meaning POP variable
significantly influence DV positively, GE variable with t-statistic value equal to 1,875256 with probabilita equal
to 0,0628 smaller α = 0,05 meaning that GE variable significantly influence DV positive. While the variable TPT
t-statistical value of 0.905152 with probabilita of 0.3669 bigger than α = 0.05 which means TPT variable does
not significantly affect the DV. Thus, the model estimation results have met the suitability test of the partial test

5
Journal of Economics and Sustainable Development www.iiste.org
ISSN 2222-1700 (Paper) ISSN 2222-2855 (Online)
Vol.9, No.16, 2018

aspect. Model estimation results can be analyzed.

Regression model is to be exposed to multicolinearity when there is a perfect linear relationship between
some or all independent variables of a regression model. In this study multicollinearity test using partial
correlation method, partial correlation test results can be seen in table 5 below;
Table 5. Coefficient of Determination Among Dependent Variables
Dependent Variabel Independent Variabel R² Result
GDP POP, GE, TPT 0,988 No Multicollinearity
POP GDP, GE, TPT 0,984 No Multicollinearity
GE GDP, POP, TPT 0,857 No Multicollinearity
TPT PDRBP, POP, GE 0,705 No Multicollinearity
Table 5 shows that the regression result of the independent variables (partial correlation) yields the
determination coefficient (R2) as follows: R2PDRB (POP, GE, TPT) equal to 0.988; R2POP (PDRBP, GE, TPT) equal to 0.984;
R2GE (PDRB, POP, TPT) equal to 0.857; R2TPT (GRDP, POP, GE) equals 0.705 Compared with R2 (PDRBP, POP, GE, TPT) of 0.995
then the coefficient of determination of free variables PDRBP, POP, GE and TPT smaller. This means the model
is good and can be analyzed.
The estimation experiment was conducted on variables considered to have an effect on income disparities,
such as GRDP, population, development expenditure and unemployment rate. The result of significant estimation
given the GRDP variable, population and development expenditure.
Estimation results in all districts in North Sumatra province based on the amount of intercept disparities
income estimation results above also can be seen that the average increase in regional income disparity is
constantly unaffected by other factors is the largest compared with other regional income disparity is City of
Tanjung Balai is equal to -2.453909. While the increase of regional income disparity in the second rank is the
disparity of income of Medan that is equal to 1.881713. In the third order increase in income disparity is
Labuhan Batu district that is equal to -1.292764.
Increase in income disparity of the lowest area is in Dairi regency of 0.057431 followed by Asahan regency
of -0.107045 and Serdang Bedagai regency of 0.111041.
High regional development disparities are caused by large population where from estimation result,
coefficient of high is variable of population, variable of GRDP and variable of economic growth and variable of
unemployment rate. The city of Medan and the city of Tanjung Balai has a relatively high population and
unemployment rate. While Labuhan Batu district has GRDP and, economic growth and population are relatively
low. Meanwhile Dairi district with GRDP and population is relatively low, while Asahan and Serdang Bedagai
districts PDRB and the relatively moderate population with relatively low unemployment rate.
This result is in accordance with previous theories and hypotheses which suggest that the increase of GRDP
and government spending will decrease disparity, in contrast to increasing population and unemployment rate
will increase disparity.
The PDRB regression coefficient equals -0.177096. This means that if GRDP rises 100 percent, then the
income disparity will decrease by 17.71 percent. Conversely, if the GRDP drops 100 percent then the income
disparity will rise by 17.71 percent. The effect of this GRDP variable is relatively high and significant at 90
percent confidence level. This shows that GRDP has a negative and significant impact on income disparities in
North Sumatra.
The regression coefficient of the population variable is 0.573420 which means that when the population
increases 100 percent will increase the disparity of economic development of 57.34 percent. The influence of
population to the disparity of economic development is the highest relative of other independent variables and
positively and significantly influence at 90 percent confidence level to the disparity of economic development in
North Sumatra.
Inequality in this study is divided into two namely the imbalance in the province and the imbalance between
provinces. Regression analysis tools are also used to examine possible factors as determinants of income
inequality within provinces.
The estimation results show the regression coefficient of development expenditure variable of 0.036701.
Means that every increase in development expenditure of 100 percent then the disparity of economic
development will increase by 3.67 percent. The influence of development expenditure variable on the disparity
of economic development positively and significantly at 95 percent confidence level.
The results obtained from this study that investments negatively affect the disparity of economic
development, the number of labor force negative affects the disparity of economic development, the allocation of
development assistance funds from the central government is uneven and evenly distributed areas that receive
areas that get too much help can increase the level disparity between regions. This is due to the concentrated
development in the developed regions compared to the regions still lagging, as the developed regions compare to
the remaining areas, as the developed regions have better facilities than the less developed ones.

6
Journal of Economics and Sustainable Development www.iiste.org
ISSN 2222-1700 (Paper) ISSN 2222-2855 (Online)
Vol.9, No.16, 2018

Conclusions and Recommendations

The coefficient of determination on the estimation result of economic development disparity variable in North
Sumatera can be explained by the variables of GDP, population, and development expenditure and
unemployment rate with the model used. The variables used to explain the variables of economic development
disparity show the direction of influence in accordance with the hypothesis. GDP has a negative and significant
effect; the number of influential population and development expenditure has a positive and significant effect on
the disparity of economic development while the unemployment rate has no significant effect on economic
development in North Sumatra. The coefficient value of the variables describing the disparity variable of
economic development, the largest is the population variable, followed by the GDP variables, development
expenditure, and unemployment rate. To reduce the inequality or disparity of economic development, the
government should continue to improve the performance of the economy by finding new sources of revenue and
budget efficiency that can be used for development in every sector. It is better for the government to open new
jobs by making it easier for investors to invest in the regions by issuing regulations and policies that encourage
the growth of these investments. With the growth of investment in the area will absorb a lot of labor, which in
turn will reduce unemployment. The government should do more socialization about the importance of family
planning in regulating the birth rate by providing cheap and free contraceptives to new partners, providing
counseling and making health facilities convenient and accessible to the public. There should be more research
on inequality with wider and more complex coverage and methods and with varied support variables that will
add and enrich the knowledge treasury.

References
Akai, Nobuo and Masayo Sakata. 2005. Fiscal Decentralization, Commitment and Regional Inequality: Evidence
from State-level Cross-sectional Data for the United States. CIRJE Discussion Papers. http://www.cirje.e.u-
tokyo.ac.jp/research/dp/2005/2005cf315.pdf
Bonet, Jaime. 2006. Fiscal Decentralization and Regional Income Disparities: Evidence from The Colombian
Experience. Ann Reg Sci 40:661-676.
http://www.springerlink.com/content/36595348343m6660/fulltext.pdf
Do Carmo, O.J., and Martinez-Vazquez, J. (2001). Czech Republic intergovernmental fiscal relations in the
transition Europe and Central Asia. Poverty Reduction and Economic Management Series, World Bank
Technical Paper, (517).
Forbes, Kristin J. 2000. Reassessment of The Relationship Between Inequality amd Growth. The American
Economic Review, Vol. 90 No. 4, hal 869-887. Amerika Serikat.
http://web.mit.edu/kjforbes/www/Papers/Inequality-Growth-AER.pdf.
Hofman, B., Fitrania; F., and K. Kaisera. Unity in diversity? The creation of new local governments in a
decentralising Indonesia. Bulletin of Indonesian Economic Studies, 41(1): pp. 57-79.
Lessman, Christian. 2006. Fiscal Decentralization and Regional Disparity: A Panel Data Approach for OECD
Countries. Ifo Working Papers. http://papers.ssrn.com/sol3/papers.cfm?abstract_id=936874.
Percik. 2007. Proses dan Implikasi Sosial-PolitikPemekaran: Studi Kasus di Sambas dan Buton. USAID
Democratic Reform Support Program (DRSP) dan Decentralization Support Facility (DSF), Jakarta.
Swianiewicz, P. 2002. Consolidation or Fragmentation? The Size of Local Governments in Central and Eastern
Europe. Local Government and Public Service Reform Initiative, Open Society Institute Budapest,
Budapest.
Tiebout, C.M. 1956. A Pure Theory of Local Expenditures. The Journal of Political Economy. 64(5): pp. 416-
424.