Spatial Analysis of The Human Development Index in Indonesia Before and During The Covid-19 Pandemic

In 2020 Indonesia’s HDI reached 71.94 or in a very high category, it is in 107th place out of 189. The HDI in every province in Indonesia shows varying numbers and has increased every year, but there has been a slowdown increase from 2019 to 2020, which caused by the Covid-19 pandemic. This study aims to analyze the variables which affect HDI in Indonesia before and during the Covid-19 pandemic, it also to analyze Covid-19 pandemic affect HDI in Indonesia. This research uses a quantitative method with descriptive analysis and inferential analysis by spatial regression using geoda software. The HDI before the Covid-19 pandemic is significantly affected by the percentage of poor residents and the university gross enrollment rate, while the during Covid-19 pandemic, HDI is also significantly affected by Gross Regional Domestic Product (GRDP) per capita and ratio of the number of hospital beds to 1.000 residents. Based on the spatial dependency test, it is known that there is a spatial dependency for HDI in Indoncovid-10esia during Covid-19 pandemic period. From these results, it is necessary to have a policy that is oriented to reducing poverty, increasing the scope of education services, improving macroeconomic conditions in a region and also ensure a sufficient number of hospital beds, especially during the Covid-19 pandemic.


Introduction
The Human Development Index (HDI) is a measure of human development success in a region, which was originally an annual publication from UNDP since 1990 (Noorbakhsh, 1998).According to Arriani and Chotib (2021), human development is a measurement to measure the fulfillment of basic human needs such as health, education, and the economy in order to live a more decent life.The calculation of the index consists of three main factors, namely Gross Domestic Product (GDP) per capita, literacy rate and life expectancy at birth, although many critics say that besides these three main components there should be other components to consider (Haq, 1995).In 2010, UNDP changed the indicators for the knowledge dimension from the original literacy rate to years of schooling (Arriani and Chotib, 2021) The determination of HDI in Indonesia is based on 3 dimensions, namely longevity and healthy living, knowledge and a decent standard of living, so that HDI can explain how people access development results in obtaining income, health and education.In 2020, Indonesia's HDI score reached 71.94 or maintained its high-performing status, but it still ranked 107th out of 189 countries and is below other neighboring countries such as Singapore, Brunei Darussalam, Malaysia, and Thailand.Based on the HDI in 2020 which was published by The Central Agency of Statistics, it is known that there's an increase in HDI every year since 2010, but the increase has slowed from 2019 to 2020, including due to the Covid-19 pandemic.
According to Brodeur et al (2021), Covid-19 pandemic has had a significant impact on the slowdown in the economy, as predicted by the World Bank that Global Gross Domestic Product (GDP) in 2022 will decrease by compared to 2019 by 5.2%.In addition to the decrease in GDP, the impact can have broad effects such as on the supply of labor and production supply chains.This phenomenon is also in line with what is happening in Indonesia, according to data from the Central Agency of Statistics, there has been a contraction in economic growth from 2019 to 2020 of 2.07%.With this phenomenon, further analysis is needed regarding the impact of Covid-19 pandemic on HDI in Indonesia, considering that the economic/income dimension is one of the dimensions that used in determining HDI.
Based on the HDI data of 34 Indonesian provinces in 2017 (before the Covid-19 pandemic period) and 2020 (during the covid-19 pandemic period), shows that the HDI achievement has a very varied value from moderate to very high category.Considering the varying HDI values, need to analyze the variables that affect the HDI in Indonesia.This is very important to determine the policy recommendations that are needed to improve the HDI in Indonesia because Indonesia has an HDI rank which lows at the global level.This analysis is also expected can be used as an assessment of the success of government policies, particularly those related to human quality development strategies.Human development-oriented policies are sorely essential, considering people are the primary capital of development.
Based on previous research, there are several variables that affect HDI in several cities in Indonesia.According to Halid and Yapanto (2021), HDI in Gorontalo is affected by several factors including income and years of schooling.It was also explained that Gorontalo City's HDI is the highest compared to other cities in Gorontalo Province.One of them was influenced by the presence of many good colleges and high schools in the city.According to Setiawan and Hakim (2013), an increase in GDP has a significant impact on an increase in HDI in Indonesia, as it affects the improvement of people's wealth.According to Prasetyoningrum and Sukmawati (2018), there is an association between HDI and poverty in Indonesia, where an increase in HDI will have a significant impact on poverty reduction.This result is consistent with Arriani and Chotib (2021) that the HDI in Central Java is significantly influenced by the poverty line and GDP per Capita with a positive correlation, and that HDI is also affected by the unemployment rate and poverty rate with a negative correlation.According to Farida et al (2022), HDI in East Java is significantly influenced by high school enrollment rates and the number of health facilities.
According to Pratowo (2012), regional spending will significantly affect the HDI values in Central Java.Further stated by Sofilda et al, 2015 and Fattah & Muji, (2012), the budget allocations for education, health and housing and public facilities or infrastructure have a significant impact on HDI.Therefore, in order to improve human development as one of the fundamental assets of development, government programs should focus more on the aspects of HDI formation, namely education, welfare and health, by allocating budgets for these aspects.
This study will analyze the influence of several independent variables such as Gross Regional Domestic Product (GRDP) per capita, the percentage of poor residents, the University Gross Enrollment Rate and the ratio of the number of hospital beds to 10.000 residencts affecting HDI in Indonesia in 2020 which represents the period during the Covid-19 pandemic and also in 2017 which represents the period before the Covid-19 pandemic.The selection of this variable is based on the dimensions used to determine the HDI, namely the knowledge dimension, the decent standard of living dimension and the health dimension, which are the main dimensions used in calculating the HDI in Indonesia.This study also determines whether or not there is a spatial dependency of HDI in Indonesia.By knowing the spatial dependency on HDI in Indonesia and the variables affecting it, it is hoped that it will provide leading policy recommendations needed to increase the HDI in all provinces in Indonesia and raise the HDI ranking at the global level.

Method 2.1 Data sources and variables
This research will analyze the correlation between the independent variables such as the percentage of poor residents, GRDP per capita, level of education (Gross Enrollment Rate in University), the ratio of the number of hospital beds per 1.000 population on the dependent variable, namely HDI in the 2017 and 2020 periods in 34 provinces in Indonesia.All data used is secondary data obtained from the publications of the Central Agency of Statistics of Indonesia and Ministry of Health of Indonesia.The selection of variables is based on several indicators related to the basic dimensions that are used to measure HDI in Indonesia, as shown in Table 1.The ratio of the number of hospital beds per 1.000 residents Hospital

Data Analysis Procedures
In this research, classical regression, spatial regression, Moran's I and LISA will be used using Geoda software.

Classical Regression
Classical regression is used to determine the correlation between the independent and dependent variables.The linear regression equation used in this study can be explained by the equation ( 1): HDI = constant + β1GRDPcap + β2Univ.enrollmentrate+ β3Poverty + β4Hospital (1) Information: β1, β2….. βp: Regression coefficient

Moran's I
Moran's I is a technique in spatial analysis to see the spatial relationships that occur in an area (Gittleman and Kot, 1990).According to Arriani and Chotib (2021), Moran's I is used to see dependencies between locations.Morans'S I value uses the following hypothesis: H0 : I = 0 (no correlation between locations) H1 : I ≠ 0 (there is a correlation between location)

Local Indicator of Spatial Association (LISA) and Moran Scatter Plot
To analyze the spatial dependence in each location, LISA and Moran Scatter Plot test is conducted.In this test, there are four cluster values observed in neighboring areas.The four cluster values are the first quadrant (High-High), the second quadrant (Low-High), the third quadrant (Low-Low), and the fourth quadrant (High-Low).

Spatial Regression
In addition to looking at the relationship between the independent and dependent variables, the spatial regression modelling in this study will also look at how the spatial correlation or the influence of the independent variables on neighbouring areas.Based on Chotib, et al (2021) there are several steps in carrying out spatial regression.The first is to test the classical regression model to see the influence relationship between the dependent variable and the independent variable and also to determine the appropriate model to use in this study.Regression is then continued to determine the most appropriate regression model to use.The selection of the regression model is based on the results of a comparison of the significance level of the probability value in each LM-Error and LM-Lag, where significant is indicated by a probability ≤ 0.05.If both are significant then proceed with comparing Robust LM-Error and Robust LM-Lag if the Robbust LM Error value is more significance then proceed with spatial error regression, whereas if Robbust LM-Lag is more significant then proceed with spatial lag regression, as the plot shown shown in Figure 1.

Results and Discussion
Based on recapitulation data of the Indonesian HDI in 2017 and 2020 published by the Central Agency of Statistics, it is known that the HDI values in Indonesia vary between 59.09 -80.77 as shown in Table 2. Figure 2 and Figure 3 show that there has been no change in variation and composition of HDI values in Indonesia, so it means the Covid-19 pandemic has not had a significant effect on HDI values in Indonesia.Both before the Covid-19 pandemic period and during Covid-19 pandemic period, there is an increase in HDI values in each province, although in several provinces there is an increase in the percentage of poverty and a decrease in GDRP per capita.Based on the results of the classical regression test to find out the independent variables that affect the dependent variable, namely HDI before Covid-19 pandemic period and during the Covid-19 pandemic period, the following correlation results were obtained: The equation for the period during Covid-19 pandemic as the following Equation ( 3 Based on the classical regression results, shows that there is no change in the variables affecting the HDI in Indonesia before and during the Covid-19 pandemic period.During these two periods, the University Gross Enrollment Rate and the percentage of poor residents were very significant in affecting the HDI in Indonesia.Meanwhile, the ratio of the GRDP per capita and ratio of hospital beds to 1,000 population do not significantly affect the HDI in Indonesia. Before the Covid-19 pandemic period, IPM in Indonesia has a negative correlation with the percentage of poverty with a coefficient of 0.276, which means that every 1% decrease in poverty will increase the HDI value by 0.276.Meanwhile, during the Covid-19 pandemic Period, every 1% decrease in poverty will increase the HDI value by 0.4117 HDI in Indonesia.The relationship between HDI and poverty is in line with previous research as has been done by Arriani and Chotib (2021) and Prasetyoningrum and Sukmawati (2018).
Another dimension that affects HDI is knowledge or education.Based on the equation above, HDI in Indonesia has a positive correlation with the University Gross Enrollment Rate with a coefficient of 0.1266 when before the Covid-19 pandemic period, which means that every 1% increase in the University Gross Enrollment Rate will increase the HDI by 0.1266.Whereas for the period during Covid-19 pandemic, the HDI in Indonesia has a positive correlation with the University Gross Enrollment Rate with a coefficient of 0.158, which means that every 1% increase in the University Gross Enrollment Rate will increase the HDI by 0.158.This result is in line with research conducted by Halid and Yapanto (2022) that the existence of a university can affect HDI scores in an area.Based on the relationship between the HDI and the university gross enrollment rate, the higher the gross enrollment rate for university education in an area, the better the human development in that area.

Spatial Dependency Test
The results of the classical regression above, it is then followed by a spatial dependency test and the results are shown in Table 3.This test is conducted to find out which econometric model is appropriate to apply.The results from the spatial dependent test show that in the period during the Covid-19 pandemic Moran's I has a significant result with a significant Lagrange Multiplier (error), while Lagrange Multiplier (lag) is not significant.From the results of comparing the Lagrange Multiplier (error) and Lagrange Multiplier (lag) values, the econometric model uses the spatial error model.When Lagrange Multiplier (error) is significant, it means that there is a spatial dependency for HDI in Indonesia with variables other than those contained in the model affecting HDI values in neighboring areas.Whereas in the period before Covid-19 pandemic it was known that Moran's I, Lagrange Multiplier (lag) and Lagrange Multiplier (error) are not significant so there was no spatial dependency of HDI between locations or provinces in Indonesia.The variables used in this study and other variables do not affect HDI in neighboring areas in the period before the Covid-19 pandemic.Based on the results of the spatial error regression in the period during the Covid-19 pandemic, there are differences in the results with the classical regression test.The spatial error test has a bigger R 2 value than classical regression.The value of standard error in all of variables in spatial error regression are smaller than classical regression.It means the spatial error regression is more accurate than classical regression.To determine the variable that affect the HDI during the Covid-19 pandemic period, can use spatial error regression.From this regression, the results were obtained that during the Covid-19 pandemic, other variables such as GRDP per capita, the ratio of number of hospital beds to 1,000 residents also significantly affect the HDI in Indonesia.The equation for the period during the Covid-19 pandemic is as the following Equation ( 4 Based on the equation above, during the Covid 19 Pandemic, every 1% increase in GRDP per capita will increase HDI by 0.00000002795, every 1% increase in the College Gross Education Rate will increase HDI , every 1% decrease in the percentage of poverty will reduce the HDI by 0.194 and every 1% increase in the ratio of hospital beds per 1,000 population will increase HDI by 2.615.Meanwhile, the result before the Covid-19 pandemic can still refer to the equations in classical regression. According to Bhakti (2012), every year the government must continue to strive to increase the HDI value.Increasing the HDI needs to be done through policies oriented to the economy, education and health.Therefore, increasing the HDI can be done through increasing economic growth through inclusive development and followed by increasing and equalizing people's welfare, poverty alleviation, expanding access and improving the quality of education and expanding the reach of health services.Furthermore, Arriani (2021) stated that a policy program is needed that can integrate HDI and also achieve SDG 1 which focuses on poverty alleviation and SDG 8 which focuses on per capita economic growth.
By obtaining the variables that significantly affect the HDI value, there should be a policy which in line with the framework of reducing the poverty rate, bearing in mind that reducing poverty greatly affects the increase in the HDI.Another policy needed to increase HDI is through policies that are oriented toward improving education services in order to increase Education Participation Rates, especially up to Collage level.Meanwhile, from the results, it emphasized that in order to improve human development during the Covid-19 pandemic, policies that were oriented towards increasing GRDB are needed by increasing and equitable distribution of people's welfare in each province.Another policy is increase health services by ensuring sufficient availability of hospital beds.
In order to increase the value of HDI in Indonesia, the government needs to ensure that the programs and budgets are oriented toward human development programs in order to improve the economic, health and educational aspects which are the basic dimensions in achieving HDI.Through this policy, it is hoped that in the end all provinces in Indonesia can have HDI in a good category, so can increase Indonesia's HDI ranking at the world level.

LISA and Moran Scatter Plot
Based on the results of the LISA test and Moran's scatterplot as shown in Figure 4 and Figure 5, it is informed that no clustering occurs because the points are spread out and not too concentrated in the middle of the quadrant both before the Covid-19 pandemic and during the Covid-19 pandemic.In these two periods there were 2 Provinces that were in quadrant I (High-High),which means there were 2 Provinces that had high HDI and were among the provinces with high HDI, namely West Java and Banten and there were 2 Provinces that were in quadrant III (Low-Low), it means there are 2 provinces that have low HDI scores and are among the provinces with low HDI, namely Papua and West Papua.From theese figures it can be seen that cluster I is located on the island of Java or the western part of Indonesia, while cluster III is located in the eastern region of Indonesia, this may indicate that human development is still uneven in Indonesia.

Conclusion
Based on an analysis through spatial econometrics of HDI in Indonesia, shows that the reduction in poverty and also an increase in university enrollment rates are very significant in increasing the value of HDI in Indonesia, whereas during the Covid-19 pandemic period the variable such as the ratio of the number of hospital beds per 1.000 population and GRDP per capita also affects the HDI value in Indonesia.Based on the value of Moran's I, it is known that there is no spatial correlation of HDI in Indonesia before the Covid-19 pandemic period, whereas during the Covid-19 pandemic there is a spatial dependency, it means the HDI and also the variables that affect it in a province can affect the HDI of the province in the surrounding.
By knowing the variables that significantly affect the HDI, policies that are oriented towards improving people's welfare and increasing access to education, especially up to the university level, are needed in order to increase the HDI rate in Indonesia.During the Covid-19 pandemic, it is necessary to improve macroeconomic conditions and increase access to health services by increasing the number of hospital beds in order to increase the HDI in Indonesia.
For further research, it can be recommended to look at regional expenditure variables related to improving the economy, education and health to provide recommendations for fiscal policy priorities that are significant in increasing the HDI in Indonesia.In addition, it is also necessary to model the Gross Enrollment Rate variable for education under university level to see the minimum education required.

Figure 1 .
Figure 1.Spatial Lag Test Selection Flow and Spatial Error Test Source: Anselin and Rey (2014) in Chotib et al (2021)

Table 1 .
Variables used in the research

Table 2 .
Recapitulation Dependent Variable and Independet Variables