Analysis of Poverty Determinants in the Ex-Kedu Residency 2015-2020

This study aims to analyze and determine the effect of the number of BPJS participants, especially PBI, education, UMK, and per capita income, on poverty in six Cities/Regencies of the Former Kedu Residency in 2015 – 2020. The method used is quantitative, where data acquisition is secondary data through various sources in time series and cross-section data. The cross-section data consists of six cities/districts, while the time series data is from 2015 to 2020. Technically, the data is analyzed using panel data regression with the Fixed Effect Model approach as an excellent model to use. The results show that the variable number of BPJS participants, especially PBI, positively affects poverty in six Cities/Regencies of the Former Kedu Residency in 2015-2020; the UMK and income per capita variables harm poverty. Meanwhile, the education variable measured by the school participation rate parameter does not affect poverty.


INTRODUCTION
The former Kedu Residency has the second-highest poverty rate in Central Java after the former Banyumas Residency, with an average poverty rate of 12.59%. The former Kedu Residency consists of 5 regencies and one city, where each district contributes significantly to the number of poor people in Central Java. Therefore, it is necessary to prioritize and pay special attention to the red zone areas to overcome poverty. To break the chain of poverty, the government can increase the Human Development Index through education, health, and community welfare. Health is the crucial investment in Human Resources and the most significant capital to build and improve human welfare. Poor health and nutrition indicators will impact other activities, such as reduced rights to life, income, and not optimal activities in carrying out work. The right to education is not implemented (Asrol & Ahmad, 2018). If a person's health level is in prime condition, then that person can access education optimally so that his level of welfare will increase in the long term. The higher the education taken, the higher the quality of resources and human life so that poverty can be reduced.
Health, education, income per capita, and minimum wage influence poverty (Asrol & Ahmad, 2018;Marmujiono, 2014). One of the essential indicators to see the quality of human resources in a country is health parameters. Therefore, the steps of the Indonesian government to improve the quality of health and create welfare for the poor through social insurance in the health sector to gain access to health so that health capacity increases through the JKN Program or National Health Insurance. (Ardinata, 2020). The program was implemented in 2014 and aims to protect health by facilitating access to health services and reducing finances. The scope of JKN program participation has increased until 2021 to reach 82.3% of the total population.
One of the JKN programs for the poor is to include them in the PBI (Recipients of Contribution Assistance). The central and regional governments bear the participants' monthly fees. The more people participate in the JKN program, the higher the quality of a person's health to encourage the population's welfare. Participation for low-income and underprivileged people in meeting the needs of the health sector is getting bigger so that more and more poverty is occurring, and there are many government responsibilities for developing public health. (Fahroez et al., 2018;Fithri & Kaluge, 2017).
Education is also closely related to poverty. One of the long-run investment factors, according to Amartya Sen, is that education will directly affect a person's level of welfare and will indirectly affect social and economic changes in the long term. (Tilak, 2002). Factors that influence the poverty of a country or region through the human resources approach, education becomes an aspect in alleviating poverty. Humans become a capital formed through training to teach knowledge and skills so that the quality and productivity of a person will increase, and income will also increase so that community welfare is achieved. (Sotomayor, 2021). A study that explains the occurrence of a negative and significant inverse relationship between education and poverty (Mardiyana, 2020), the higher a person's education, the lower the poverty that occurs.
The minimum wage is also closely related to fluctuations in poverty levels. The cause of changes in the poor population due to high and low MSEs affects public acceptance, which impacts meeting the needs of a decent life (Lutfi et al., 2016). The lower the minimum wage, the lower the income the community receives, causing a decrease in purchasing power to meet the needs of a decent life and the population's welfare so that the poverty level will increase. (Maipita, 2018;Sotomayor, 2021). There is an inverse relationship between the minimum wage and poverty; it is shown that the higher the minimum wage, the lower the poverty rate (Ramdhan et al., 2018).
Another factor that affects the poor population is income per capita. There is an inverse correlation between income per capita and the poor; the higher the income per capita or GRDP per capita, the greater the welfare of the people and vice versa. The increased income per capita is also driven by the increasing income of a region or GRDP. In his book, Sukirno (2016) states that when economic growth rates decrease and do not exceed the additional total population, there will be a decrease in income/capita if people's incomes increase, the ability to meet future needs and save. Conversely, if there is a decrease in people's income, it will result in the inability of the population to fulfill their needs.
Based on the description of the background, this study uses the dependent variable in the form of the number of poor people and the independent variable in the form of the number of BPJS participants, mainly PBI, education with APS parameters, UMK, and per capita income adopted from Dartanto's research in Djamhari et al. (2020), Safri (2021), Lutfi et al. (2016), Sotomayor (2021) dan Ramdani (2017). This study uses a differentiating variable from previous studies, namely health parameters with the National Health Insurance participation program, especially PBI (Recipient of Contribution Assistance). It aims to improve the quality and capacity of health by protecting the health of the poor as a result of health, including the human capital component. This study aims to analyze the determinants of poverty in six Cities/Regencies of the Former Kedu Residency in 2015 -2020.

RESEARCH METHODS
This research utilizes a quantitative descriptive method. A study was conducted using secondary data. Data sources were obtained from the Central Statistics Agency (BPS) in annuals, BPJS Health, and agencies related to the research being conducted. Time series data or time-series from 2015 -2020. The cross/individual data are six Cities/Regencies in the former Kedu Residency covering Kebumen Regency, Wonosobo Regency, Purworejo Regency, Temanggung Regency, Magelang Regency, and Magelang City, with a total observation of 36.
Data collection techniques in research are conducted through library research. The variables used in the study, namely the dependent variable in the number of poor people, are the total population with ownership of average per capita expenditure below the poverty line. At the same time, the independent variable is the number of BPJS participants, mainly PBI (X 1 ), namely the number of BPJS participants for the poor (the poor and needy) by the criteria who are entitled to receive them where the central and regional governments bear the monthly contributions of these participants. Education (X 2 ) is in the form of School Participation Rate, the percentage of the number of school students at a specific age classification compared to the age group according to the level of education. The UMK (X 3 ) is the minimum average wage set by the City/Regency Government and given by employers every month to workers to achieve community welfare for each City/Regency. The income per capita (X 4 ) is the community's average income, calculated by comparing the Gross Regional Domestic Product of ADHK with the number of residents in a given year.
The data analysis technique in this study used panel data regression analysis. The following is the panel data regression equation model as follows: Log POV it =α+ β 1 log JKN it +β 2 APS it +β 3 logUMK it +β 4 log INCOME it + e it …. (1) With, LogPOV = Poverty (persons); α = Constanta; β 1 -β 4 = Coefisien; regression; Log JKN = Number of BPJS participants, mainly PBI (persons); APS = School participation rate (%); Log UMK = City/district minimum wage (Rp); Log INCOME = Income per capita (Rp); ɛ = Error term; i = County town (cross section); t = 2015-2020 a. Chow test Based on the regression results, the p-value of the Chi-square Cross-section in the Likelihood test above is lower than the level of =5%, namely 0.000 < 0.05. That is, rejecting hypothesis 0 and accepting hypothesis 1. It can be concluded that the model that is received and applied in data regression is the FEM (Fixed Effect Model) model. Therefore, further testing is needed to choose a good model, whether FEM or REM using the Hausmann test.

b. Hausman test
Based on the regression results, the p-value of the random cross-section in the Hausman test above is lower than the level of =5%, 0.0070 < 0.05. That is, rejecting hypothesis 0 and accepting hypothesis 1. It can be concluded that the model that is received and applied in data regression is the FEM (Fixed Effect Model) model.

Classical Assumption Test
This research is free from the problem of classical assumptions. The following explains the output of classical assumption testing to see the data distribution in the study conducted.
Based on the normality test, the probability of JB is 0.359142, which shows a value greater than the significance level of =5% or 0.05 (0.359142 > 0.05), so the residual value is normally or distributed.
The test results to determine whether or not there is a strong relationship between the independent variables used in the study, namely utilizing the multicollinearity test. The following table presents the output of the multicollinearity test as follows: Based on table 3 above, it is known that the correlation coefficient between the variables used in the study is not more than 0.8, so it can be concluded that there is no multicollinearity problem in panel data regression.
The condition of the emergence of variable inequality in a regression model of the residual variable from one observation to the following observation requires heteroscedasticity testing. The following are the results of the heteroscedasticity test obtained as follows. Based on the results of table 4 regression, the overall p-value generated by each independent variable is more significant than 0.05; it can be concluded that there is no heteroscedasticity problem in panel data regression.
As for the autocorrelation test, it was concluded that H0 was accepted because dU < d < 4-dU; which was 1.236 < 2.047956 < 2.7642; meaning that there was no autocorrelation problem, either positive or negative; So the panel data regression model is free from autocorrelation problems.

T-test
The partial effect of the independent variables on the dependent variable was carried out by t-test by comparing t-count and t- Based on the output of Table 5 above, the results showed that the variables of the number of BPJS participants, mainly PBI, UMK, and per capita income, significantly affected the number of poor people in six Cities/Regencies as Ex-Kedu Residency in 2015-2020. At the same time, the education variable has no significant impact on the number of poor people. It is proven that the p-value of the variable number of BPJS participants, especially PBI is 0.0887 < 0.10, the UMK variable has a p-value of 0.0005 < 0.05, and the income per capita variable has a p-value of 0.0003 < 0.05. The education variable has a p-value of 0.8913 > 0.05, which has no effect.

F -Test
The test criteria used a significance test of 0.05 with df1 of 4 and df2 of 31, and the F-table value shows 2.679. Based on testing the independent variable on the dependent variable from the calculation of F-count of 3389,766 > F-table of 2,679 or a probability of 0.000000 < 0.05; This means this model can explain the relationship between variables; namely the number of BPJS participants, mainly PBI, education (School Participation Rate), UMK, and income per capita jointly or simultaneously influence the number of poor people.

R Square (R 2 )
The results of R 2 from the regression model of this study obtained the Adjusted r square value of 0.998854 or 99.88%. The result shows the percentage contribution of the influence of the independent variables, namely the number of BPJS participants, especially PBI, education (School Participation Rate), UMK, and income per capita on the dependent variable, namely the number of poor people is 99.88%. At the same time, the remaining 0.12% is explained by other independent variables not observed in this study.

Regression Model Output
Based on The number of BPJS participants, especially PBI (β1), has a significant effect and a positive coefficient value of 0.107061. If the number of JKN program participants, especially PBI, increases by 1%, then the number of poor people or the number in poverty will increase by 0.1071%; This happens with a note that the other independent variables have fixed-valued assumptions. This study's results align with research (Ratisyah, 2019) explained that BPJS membership has a positive and significant effect on the number of poor people. The increase in BPJS membership, especially Contribution Assistance Recipients (PBI), was driven by low incomes and unemployment. The government covers and includes someone in the health sector social security by the number of poor people. If the health sector social security is more, it indicates high poverty.
Conversely, if the health sector social security borne by the government is lower, it illustrates the low poverty level; This is by the condition of poverty in the Kedu Ex Residency area, classified as high, namely the second poorest in Central Java in 2020; This encourages the government to help and provide more social security for the health sector, thereby expanding PBI participation. If someone becomes a member of JKN membership where the government pays the insurance premium, the probability of accessing health facilities such as treatment, outpatient, or inpatient is much less or limited. The lack of access to health services and the lack of fulfillment of health needs, even though utilizing free health services, encourage someone to be lazy to seek treatment if they suffer from an illness or illness so that their health status decreases due to low productivity. (Soewondo et al., 2021). When one's productivity is low, one's welfare will fall, increasing the poverty rate.
The BPJS participation program, especially PBI, has not reduced the poverty rate. As a result of mistargeting, a place to reduce poverty through the health insurance provided pushes the large budget or government spending, causing ineffectiveness and inefficiency in government spending on health. (Rolindrawan, 2015). BPJS PBI premiums are one of the government's cost dependents where health sector costs positively influence poverty. This is in line with research (Kinanti, 2019) which states that the positive influence of government spending on the health sector on poverty is that expenditures made by the government to improve the quality of health are not accompanied by an increase in employment opportunities resulting in higher unemployment, birth rates, and poverty rates.

The Effect of Education on the Number of Poor People
The education variable, or APS (β2), has no significant effect, and the coefficient is 0.000279. This study's results align with and support previous research, which states that education has no significant impact on poverty (Giovanni, 2018;Kurniawan, 2018;Putri et al., 2019).
In reality, what happened in six Cities/Regencies of the former Kedu Residency is that the higher the level of education taken by children, the lower the participation rate. The trend of school participation aged 16-18 years in six Cities/Regencies as Ex-Kedu Residency fluctuates yearly for 2015-2020 but is classified as low. A person's low education encourages people to accept an offered job. In contrast, a person's high education enables a person's tendency to choose a suitable job based on previous track records, so a highly educated person will ignore that job offers that are not suitable. Therefore, a highly educated person is even more challenging to find because of job competition in the labor market; This encourages the opportunity for an increasing number of poor people (BPS, 2020a).

The Influence of MSEs on the Number of Poor People
The UMK variable (β3) has a significant effect and has a negative coefficient value of -0.281487; This means that the City/Regency Minimum Wage increases by 1%, thus reducing the amount of poverty by 0.281487% by assuming the other independent variables have a fixed value. The research results align with and support several previous studies: (Lutfi et al., 2016;Oktaviana & Wahed, 2021;Ramdhan et al., 2018), which explain that the MSE has a negative and significant effect on poverty.
Every year there is a tendency to change the City/Regency Minimum Wage in the former Kedu Residency due to adjustments above a decent living (KHL) to meet living standards (efficiency, health, welfare, basic needs) by taking into account inflation rates and economic growth that occurs. As the government determines, employers provide minimum wages to their workers in the formal sector; This positively impacts the amount of UMK workers receive, meaning that a worker's income tends to increase. The increasing wages earned by someone in the former Kedu Residency encourage an increase in someone's efforts, especially the poor, in meeting the necessities of life such as purchasing power/consumption and high spending. Therefore, a person is more prosperous and will be free from the phenomenon of poverty so that poverty decreases.

The Effect of Per Capita Income on the Number of Poor People
The income per capita variable (β4) has a significant effect and has a negative coefficient value of -0.744131. This figure means that if the income per capita obtained by the community increases by 1%, it reduces the amount of poverty that occurs by 0.744131%. However, this is by assuming the other independent variables are fixed. The results of this study are by the hypothesis that has been made and according to the theory put forward by Sukirno (2006) that the high per capita income will encourage a high level of one's purchasing power so that one's welfare increases. The suitability of the poverty circle theory proposed by Nurkse in Sukirno (2006) is that the low level of capital formed when viewed from the supply side of the money affects the low level of community income so that the level of productivity and investment ability in the future is relatively low. The results of this study are also in line with and support several previous studies, namely: (Candra et al., 2012;Fadlillah, 2013;Sari, 2018), which explain that income per capita has a negative and significant effect on poverty.

CONCLUSION
Based on the analysis results, it can be concluded that the number of BPJS participants, especially PBI, has a positive and significant effect on population poverty in the Six Cities/Regencies of the Former Kedu Residency in 2015-2020. The education/school participation rate variable does not significantly affect population poverty. The MSE has a negative and significant impact on population poverty. The income per capita has a negative and significant effect on population poverty.