INTERNATIONAL JOURNAL OF COMPUTING SCIENCE AND APPLIED MATHEMATICS.
VOL.
NO.
AUGUST 2024
Analysis of Human Development Index in West Nusa Tenggara Province with Spatial Panel Model Alfira Mulya Astuti.
Afifurrahman, and Habibi Ratu Perwira Negara AbstractAiThe purpose of this article is to examine the factors that influence the human development index (HDI) in West Nusa Tenggara using a spatial panel model.
This research is crucial because it can analyze correlations between regions and is more efficient, informative, and effective in HDI modeling.
The data structure is panel data, where observation units are the cities and regencies in West Nusa Tenggara Province for 2010 A human development index serves as the dependent The independent variables were per capita expenditure, average length of school, length of school expectations, and life The Rook contiguity and the customized matrix .
ransportation route.
are used to examine geographical impacts.
The results of the analysis indicate: .
there are spatial linkages between districts and cities in West Nusa Tenggara.
the SAR Fixed Effect model is the most appropriate spatial model to model the human development index.
the human development index can be improved simultaneously by factors such as life expectancy, expected length of schooling, average length of schooling, and per capita expenditure.
life expectancy is the main factor affecting the human development index.
Keywords: Spatial Analysis.
Human Development Index.
Panel Data.
I NTRODUCTION
UMAN Human development can be defined as a series of phases or actions taken to elevate the quality of human existence.
The United Nations Development Programme (UNDP) establishes four main elements in human development, namely equity, productivity, empowerment, and sustainability .
High economic growth cannot always alleviate social issues like poverty and the general level of Thus, one indicator that requires attention is human The measurement of human development progress refers to the methods developed or popularized by UNDP, namely: Human Development Index or HDI .
Badan Pusat Statistik (BPS) defines that HDI is a statistical measure used to measure the progress and quality of life of people in a country or region based on three indicators, namely, health, education, and decent standard of living .
By considering these indicators.
HDI can provide a more comprehensive picture of a countryAos progress in improving the quality of life of its people.
HDI provides guidance for governments in designing more effective development policies, identifying social gaps, and directing efforts to improve overall quality of life.
Astuti.
Afifurrahman.
Negara are with the Universitas Islam Negeri Mataram.
Gajah Mada Street.
Mataram 83116.
Indonesia e-mail:
alfiramulyastuti@uinmataram.
Manuscript received July 18, 2023.
accepted September 14, 2023.
One of the provinces aiming to raise the HDI value is West Nusa Tenggara (NTB).
According to BPS of West Nusa Tenggara province.
HDI of West Nusa Tenggara tends to increase .
Despite the increase every year, the HDI is still in the 29th position out of 34 provinces, which shows that West Nusa Tenggara province is in the category of the five lowest provinces in Indonesia in 2022.
There are several researchers who have studied the human development index of West Nusa Tenggara Province.
Rayes .
examined the human development index of West Nusa Tenggara using panel data for 2013Ae2017.
Sapurah et al .
studied the HDI of West Nusa Tenggara province using the fixed effect panel method for 2010Ae2017.
Pramuja et al .
studied the HDI of West Nusa Tenggara with the First Dif-ference Generalized Method of Moment (FDGMM) method for 2016 Ae 2020.
Rayes .
Sapurah et al .
, and Pramuja .
used panel data structure.
The benefits of using panel data are that the data is more informative, diverse, efficient, and able to measure effects that cannot be observed with pure cross-section data and pure time series .
They have not discussed spatial effects in their studies.
The spatial effect of a model is characterized by the presence of a weighted matrix (W ) in the model and is first applied using cross section data .
The advantage of spatial models is that they provide information about the direct, indirect, and total effects of independent variables .
Therefore, to complement the previous study, spatial effect studies need to be added to the study of human growth index in West Nusa Tenggara.
This is because West Nusa Tenggara is a province consisting of 8 regencies and 2 cities that are connected to each other.
The purpose of this article is to examine the factors that influence the human development index in West Nusa Tenggara using a spatial panel model.
There are 2 main differences from this study compared to previous studies.
First, the analysis model used is the spatial panel model.
Second, the observation time interval is 2010-2022.
In this study, we used 2 spatial weighted matrices.
It aims to compare the performance of each spatial weighted matrix in modeling the human development index in West Nusa Tenggara and find out the factors that influence it based on the best model.
The results of this analysis are expected to assist local governments in determin-ing policies related to the human development index.
INTERNATIONAL JOURNAL OF COMPUTING SCIENCE AND APPLIED MATHEMATICS.
VOL.
NO.
AUGUST 2024
The discussion flow begins by presenting the reasons for modeling the human development index with the spatial panel model, describing the spatial panel model in Section 2, discussing the factors affecting the human development index in West Nusa Tenggara for 2010-2022 based on the model selected in Section 3, concluding and presenting recommendations for the next article in Section 4.
II.
M ATERIAL AND M ETHODS
Human Development Index The Human Development Index (HDI) measures the degree to which human development has had a positive influence on both peopleAos physical .
ealth and well-bein.
and nonphysical .
The UNDP defines human development as the process of giving people more choices in terms of their income, health, education, physical environment, and other factors.
Longevity and healthy living .
ong life and healt.
, knowledge, and living standards are the three funda-mental qualities that BPS .
claims should be used as the benchmark for assessing the Human Development Index.
According to Stanton .
, the HDI combines the human proxy for three fundamental abilities: health, education, and a respectable standard of life.
It is a measure of human growth.
A measurement of health is life expectancy (LE).
School enrolment (ENR) and literacy (LIT) are used as education The education (E) index is created by adding together the literacy and school enrolment indices to get a weighted A measure of living standards is GDP per capita (Y ).
The fundamental idea of human evolution, in HarrisonAos view .
, is measuring three dimensions.
The first is for people to live long and healthy lives.
the second is for people to learn.
and the third is for people to have access to the resources required for a respectable level of living.
The unweighted average of the components of the life expectancy index, the index of educational achievement component, and the real per capita GDP index component .
$) is the HDI of a nation.
HDI is divided into three categories by UNDP: low (HDI less than .
, low-medium (HCI between 50 and 65.
, uppermedium (HMI between 66 and 79.
, and high (HDI 80 and Lower-middle class .
ower-mediu.
to medium-top .
pper-mediu.
socioeconomic status are included in IndonesiaAos regional HDI.
The populationAos capacity to absorb and manage the sources of economic growth will be determined by the level of human development.
A key component of achieving economic growth is having strong links with or against institutional technology .
Panel Data Model Panel data combines cross-sectional and time-series data.
Crosectional data are gathered for multiple sample units simultaneously, whereas time-series data are gathered over time .
When using panel data, there is a difference in the coefficients for each crosectional unitAos slope and intercept .
The panel data regression model is generally written as shown in eq1 .
yit = AA xit uit i = 1, 2.
A A A , n and t = 1, 2.
A A A .
T , where subscript i represents the cross-sectional unit, t represents the time-series dimension, yit is the dependent variable for the iOeth cross-sectional unit and the tOeth period.
AA is the scalar or constant .
, xit is the independent variable for the iOeth crosectional units and the jOeth period, is the K y 1 parameter vector.
K is the number of independent variables, and uit is the error component .
The error component in panel data regression consists of a general component and a specific component.
The general error component (Ait ) is the error for the iOeth individual and the tOeth period.
The specific error component consists of an unobserved individual-specific effect .
i ) and an unobserved time-specific effect or t .
The error component in the panel-data regression model is expressed as follows:
uit = vi t Ait , .
Based on the error component, the model for the panel data consisted of one-way and two-way error component models.
A model in which there are only an individual-specific effects or a time-specific effects is called a one-way error component The panel data model consisting of these two specific effects is called a two-way error component model.
Based on the parameter estimation method, there are three model approaches that are commonly used for parameter estimation in panel data regression: pooled effect, fixed effect, and random effect models .
The selection of the research object can be used as a determinant to determine the effect of the panel on the model.
The fixed-effect model is the right model if the researcher chooses the object.
Meanwhile, if the object is chosen randomly from the population, the random-effects model is the right model to use .
This study focused on a fixed-effect model with a one-way error component.
This was because the object was chosen by the researcher.
This can be expressed as follows .
yit = AA vi xit Ait , .
Spatial Econometrics Model A spatial model is a statistical model related to the geographical conditions of the observation location .
The spatial econometrics model is aimed at data in the economic field that contains regional .
In this case, it is related to spatial dependency and regional heterogeneity.
Spatial dependency or spatial autocorrelation describes the dependencies between regions.
Spatial heterogeneity or spatial structure describes the diversity/variation of the model for each region .
Spatial dependency testing is the early detection of spatial dependencies in each model.
Two popular tests are used to assess spatial effects: the MoranAos I test and the Lagrange multiplier (LM) test .
The Lagrange multiplier test presents more detailed results than MoranAos index test.
This is because the LM test can detect the location of the spatial dependence that occurs, spatial lag or spatial error.
Both methods were applied to test the spatial effects of the model.
Spatial Autoregressive Model.
A model that incorporates a simple regression model with a spatial lag on the dependent INTERNATIONAL JOURNAL OF COMPUTING SCIENCE AND APPLIED MATHEMATICS.
VOL.
NO.
AUGUST 2024
variable using cross-sectional data is called the spatial autoregressive (SAR) or spatial lag model .
The SAR model requires the spatial effect to be on the dependent variable.
The spatial effect is represented by the matrix W.
eq4 is the general form of SAR.
yi = A Oc wi j y j Oc k xki Ai .
i = 1, 2.
A A A , n .
with A is the spatial autocorrelation coefficient.
Spatial Error Model (SEM).
A model in which there is a spatial correlation in the error.
The general form of SEM for cross-sectional data can be seen in eq5.
Fig.
1: The connection of the variables yi = Oc k xki ii , where ii = A Oc wi j i j Ai i = 1, 2.
A A A , n, with A is the spatial autocorrelation coefficient.
Spatial Panel Model Panel data modeling, which also examines the spatial effect of the dependent variable, is called the spatial autoregressive panel model.
The spatial error panel model examines the spatial effect of error.
The individual-specific effect was chosen as a characteristic of the panel data.
eq6 and eq7 are forms of the spatial auto-regressive panel model and spatial error panel model with a fixed effect, respectively.
yit = A Oc wi j y jt Oc k xkit Ait yit = Oc k xkit iit , where iit = A Oc wi j i jt Ait .
Methodology The panel data utilized in this study was obtained from the Badan Pusat Statistik of West Nusa Tenggara Province and was released on the BPS West Nusa TenggaraAos official The observation units are the following cities and regencies in West Nusa Tenggara Province: Mataram City.
Bima City.
North Lombok Regency.
East Lombok Regency.
Central Lombok Regency.
West Lombok Regency.
Sumbawa Regency.
West Sumbawa Regency.
Dompu Regency, and Bima Regency.
There are 13 years of observation, namely 2010 A human development index (HDI) serves as the dependent variable.
The independent variables were per capita expendi-ture (PCE), average length of school (ALS), old school expectations (OSE), and life expectancy (LE).
Fig.
presents the connection between variables used in this study.
The data in this study was analyzed using the R software.
Fig.
2: Human development index growth of west nusa tenggara from 2017 to 2022 that HDI in-creased in West Lombok.
East Lombok, and West Sumbawa.
figure2 demonstrates the spatial dependence of the human development index between districts and cities in West Nusa Tenggara.
On the distribution map, the close color resemblance of neighboring cities or districts serves as a clue.
Cities or districts that are close to each other contain attributes that are i.
R ESULT AND D ISCUSSION
Exploration Inferential A thematic map depicting the evolution of the human development index in West Nusa Tenggara over the previous six years is shown in figure2.
In NTB.
HDI typically rises Thematic maps with color differences make it obvious The panel regression modelsAo parameters were first estimated in this work.
The common effect model (CEM), the fixed effect model (FEM), and the random effect model (REM) are the three models that can be produced by modeling using INTERNATIONAL JOURNAL OF COMPUTING SCIENCE AND APPLIED MATHEMATICS.
VOL.
NO.
AUGUST 2024
the panel regression approach.
table1 shows the outcomes of parameter estimation.
weighted matrix.
figure3 illustrates the spatial weighted for side intersections and transport flows.
TABLE I: Parameter estimation results for panel regression Variable Intercept lnOSE lnALS lnPCE Common effect model Estimate p-value 000*** 000*** 000*** 000*** Fixed effect model Estimate p-value 000*** 000*** 000*** 000*** Random effect model Estimate p-value 000*** 000*** 000*** 000*** *** Significant at = 1%.
table1 demonstrates that for all three panel regression models, the p-values for the variableAos life expectancy, length of schooling, average length of schooling, and per capita expenditure were 0.
All independent variablesAo P-values are lower This indicates that for all three panel models at = 0.
01, the human development index is positively and significantly influenced by life expectancy, predicted length of schooling, average length of schooling, and per capita The Chow test and Hausman test were used to determine which panel model was the best.
Test of Chow to decide whether to use FEM or CEM.
Test of Hausman to determine whether to use FEM or REM.
table2 displays the outcomes of both assessments.
TABLE II: Parameter estimation results for panel regression Chow test Hausman test F statistic p-value Chi-square statistic p-value 000*** Significant at: *** = 1%, * = 15%.
Fig.
3: Spatial weighted matrix plot: .
rook contiguty, .
transportaion route.
The Moran I test and the Lagrange Multiplier (LM) can be used to detect whether there are spatial dependencies .
in a regression model .
The Moran I test demonstrates global or thorough spatial entanglement.
The spatial reliance on endogenous factors or error of model is revealed by the LM test.
table3 displays the outcomes of the Moran I and LM tests.
According to table3, there TABLE i: Results of spatial dependence test Statistic The Chow testAos p-value was 0.
000, which indicates that it was a statistically significant result based on the analysis shown in Table 2Aos findings.
Thus, comparing CEM and FEM.
FEM is the better model.
The p-value was 0.
126 for the Hausman test.
FEM is the model of choice if utilizing = 15%.
This indicates that at least one fixed district .
is influencing the model.
Consequently, the fixed effect model was selected as the panel model to continue in the spatial The Rook contiguity matrix and the customized matrix .
ocioeconomic relations approac.
are the weighted matrices (W ) used to examine geographical impacts.
The intersection of the regional sides connecting sites led to the selection of the rook contiguity matrix to portray the interrelationships.
Regionally intersecting areas are seen as having similar characteristics, such as the North and West Lombok Regencies.
Because the two districts physically border one other or intersect geographically, they are seen as having spatial links.
The customized weighted matrix was selected because places without side junctions may be connected to other regions due to economic ties or the proximity of other socioeconomic factors .
Routes for land, sea, and air transportation are used as a guide when creating a customized LM1 (SEM) LM2 (SAR) CLM1 (SAR) CLM2 (SEM) LMH (SAR and SEM) Rook Contiguity Estimate p-value 000*** 000*** 000*** 000*** 000*** Customized Estimate p-value Moran I 000*** 000*** 000*** 000*** Significant at: *** = 1%, * = 15%.
are global spatial dependencies at = 1%, as indicated by the Moran I values for the Rook contiguity and customize weighted matrices.
According to the LM value, there is a spatial dependency on both lag (SAR) and error (SEM) for both weighted matrices employed at = 5%.
The weighted matrix that will be used for additional modeling will therefore be the Rook contiguity and customized weighted matrix.
A fixed effect model and a spatial autoregressive model are coupled to form the SARFE model (SARFEM).
The fixed effect model and the spatial error model are combined to form the SEFE model (SEFEM).
table4 displays the estimation outcomes of the SARFEM and SEFEM for the rook contiguity and customized weighted matrix.
In both the customized matrix and the rook contiguity weighted matrix, table4 shows that the value of the spatial effect coefficient on the dependent variable (HDI) is clearly A significant influence on = 1% was seen for INTERNATIONAL JOURNAL OF COMPUTING SCIENCE AND APPLIED MATHEMATICS.
VOL.
NO.
AUGUST 2024
TABLE IV: Parameter estimation results for spatial panel
Estimate WlnHDI
lnOSE
lnALS
lnPCE
Loglikelihood
Weighed Matrix
Rook Contiguity Customized SARFEM
SEFEM
SARFEM
SEFEM
084***
536***
452***
487***
459***
214***
217***
209***
209***
133***
129***
126***
125***
172***
150***
161***
155***
647***
249***
ln HDIBR,2022 = Oe0.
ln HDIDR,2022 ln HDIBC,2022 536 ln LEBR,2022 0.
214 ln OSEBR,2022 133 ln ALSBR,2022 0.
172 ln PCEBR,2022 vBR = Oe0.
n HDIDR,2022 ln HDIBC,2022 ) 536 ln LEBR,2022 0.
214 ln OSEBR,2022 133 ln ALSBR,2022 0.
172 ln PCEBR,2022 0.
the spatial effect on the HDI variable for the rook contiguity weighted matrix.
Customized weighted matrices, however, do not follow this specific rule.
For the rook contiguity and customized weighted matrices, the value of the spatial effect coefficient on error exhibits a positive and significant sign.
Both for the rook contiguity weighted matrix and for the customized weighted matrix for SARFEM and SEFEM, life expectancy, length of schooling, average length of schooling, and per capita expenditure had a positive and substantial impact on the human development index at = 1%.
The SARFEM with the rook contiguity weighted matrix, denoted by eq8, has the highest R2 and loglikelihood value is the most effective model for simulating the human development index in West Nusa Tenggara.
ln HDIBR,2022 = Oe0.
536 y ln .
214 y ln .
133 y ln .
172 y ln .
) 0.
HDIBR,2022 = exp.
= 73.
ln HDIit = Oe0.
084 Oc wi j ln HDI jt 536 ln LEit 0.
214 ln OSEit 133 ln ALSit 0.
172 ln PCEit vi i = BR.
DR.
BC,W L,CL.
EL.
NL.
MC.
SR,W S.
t = 2010, 2011.
A A A , 2022.
and vi is a fixed effect .
istrict or cit.
, which value is shown in table5.
TABLE V: Regency or city effects.
Regency/city Bima regency Dompu regency Bima city West Lombok regency Central Lombok regency East Lombok regency North Lombok regency Mataram city Sumbawa regency West Sumbawa regency Furthermore, it is known that the life expectancy, school length expectancy, average length of schooling, and per capita expenditure of Bima district in 2022 are 66.
years, 13.
58 years, 8.
17 years, and Rp.
8,699,000 per year, respectively.
The HDI of Dompu regency and Bima city in 2022 is 69.
15 percent and 76.
84 percent.
Based on eq9, the following formula is used to determine the human development index for Bima regency in 2022:
ln HDIBR,2022 = 4.
Code boundaries of the Bima district.
eq9 can be used to get the human development index in the Bima regency.
Significant at: *** = 1%.
Effect The following describes the modelAos interpretation of eq8:
The Bima district will be used as an example area for the spatial autoregressive fixed effect model (SARFEM) to calculate the human development index.
Dompu regency and Bima city are regions and cities that cross the Using a spatial autoregressive fixed effect model, the human development index of the Bima Regency in 2022 results in 73.
27 percent.
Based on BPS statistics for 2022, there is a 5.
7 percent difference in the HDI of the Bima If other variables are constant, a spatial effect coefficient with a negative and significant sign indicates that the human development index in West Nusa Tenggara can be decreased by districts or cities that are bounderies to the reference district or city.
A value of 0.
536 indicates that, given all other factors remain constant, if life expectancy improves by 1 percent.
West Nusa TenggaraAos human development index will similarly rise by 0.
536 percent.
If all other factors remain constant, a value of 0.
indicates that if the length of school expectancy rises by 1 percent, the human development index in West Nusa Tenggara will similarly grow by 0.
214 percent.
If all other factors remain constant, a value of 0.
indicates that if the average length of education grows by 1 percent.
West Nusa TenggaraAos human develop-ment index will similarly rise by 0.
133 percent.
If all other factors remain constant, a value of 0.
172 indicates that if per capita expenditure grows by 1 percent.
West Nusa TenggaraAos human development in-dex will similarly rise by 0.
172 percent.
INTERNATIONAL JOURNAL OF COMPUTING SCIENCE AND APPLIED MATHEMATICS.
VOL.
NO.
AUGUST 2024
IV.
C ONCLUSION
Based on the analysis and discussion, it is possible to draw the following conclusions: there are spatial linkages between districts and cities in West Nusa Tenggara, the SAR Fixed effect model is the most appropriate spatial model to use when modeling the human development index.
the human development index can be improved simultaneously by factors such as life expectancy, expected length of schooling, average length of schooling, and per capita expenditure.
and life expectancy is the main factor affecting the human development I MPLICATIONS AND R ECOMMENDATIONS The analysisAos findings indicate that life expectancy has the most influence on how the human development index is The level of health and wellbeing of the population is better in a place where life expectancy is higher.
The government ought to attempt to raise public awareness of the need to maintain a healthy lifestyle and make it simpler for individuals to access medical facilities.
The government is also expected to consider the state of the immediate neighborhood, as well as any areas connected by transit networks, when determining policy.
Examining economic variables outside that employed in this study will help researchers improve the studyAos findings further.
Researchers can also create studies of the spatial effects connected to the usage of weighted matrices.
Additionally, the next researchers can study economic elements not only as single equations but also as simultaneous equations.
R EFERENCES
Sarkoro and Z.
Zulfikar.
AuDana alokasi khusus dan pendapatan asli daerah terhadap indeks pembangunan manusia .
tudi empiris pada pemerintah provinsi se-indonesia tahun 2012-2.
,Ay Riset Akuntansi dan Keuangan Indonesia, vol.
1, pp.
54Ae63, 2016.
Yolanda.
AuAnalysis of factors affecting inflation and its impact on human development index and poverty in indonesia,Ay European Research Studies Journal, vol.
XX, pp.
38Ae56, 2017.
BPS.
AuIndeks pembangunan manusia .
uman development inde.
Ay .
Rayes.
AuHuman development analysis: The role of demographic factors in west nusa tenggara in the period of 2013-2017,Ay 2019.
Sapurah.
Gunartha, and N.
Fitriyani.
AuPanel data regression analysis of human development index in west nusa tenggara province with fixed effect model,Ay Eigen Mathematics Journal, 2021.
Pramuja.
Sari.
Arifin.
Azizurrohman, and S.
Supiandi.
AuLong- and short-term analysis on the human development index in west nusa tenggara,Ay Journal of Enterprise and Development, vol.
58Ae72, 2023.
Baltagi.
AuEconometric analysis of panel data,Ay John Wiley Sons, .
Astuti.
Setiawan.
Zain, and J.
Purnomo.
AuA review of panel data on spatial econometrics models,Ay J Phys Conf Ser.
1490, 2020.
Astuti.
Setiawan.
Zain, and J.
Purnomo.
AuThe extended algorithm for quasi maximum likelihood parameter estimation,Ay in In: AIP Conference Proceedings, 2021.
Anselin.
AuSpatial econometrics: Methods and models,Ay Springer Netherlands, 1988.
Astuti.
Setiawan.
Zain, and J.
Purnomo.
AuA modified generalized estimating equation approach for simultaneous spatial durbin panel model: Case study of economic growth in asean countries,Ay Decision Science Letters, vol.
12, pp.
369Ae388, 2023.
LeSage.
Applied Econometrics Using MATLAB.
Jaya and Y.
Andriyana.
AuAnalisis data spasial: Perspektif bayesian,Ay Alqaprint Jatinangor.
Elizabeth.
The Human Development Index: A History.
Amherst, 2007.
Makiko Ito.
The Human Development Index: a search for a measure of human values.
Ramirez.
Ranis, and F.
Stewart.
Economic Growth and Human Development.
Gujarati and D.
Porter.
Basic Econometrics.
New York: The McGrawHill Companies, 2009.
Baltagi.
Econometrics.
New York: Springer, 2021.
Anselin.
Spatial Panel Econometrics.
In: The Econometrics of Panel Data.
Berlin: Springer Netherlands, 2008.
Edi.
Quasi-maximum Likelihood untuk Regresi Panel Spasial (Studi Kasus: Laju Pertum-buhan Ekonom.
LeSage.
Introduction to Spatial Econometrics.
New York:
Taylor Francis Group, 2009.
Stakhovych and T.
Bijmolt.
AuSpecification of spatial models: A simulation study on weights matrices,Ay Papers in Regional Science, 88, pp.
389Ae408, 2008.