Government Expenditure and Poverty in East Java Province

: This study aimed to examine the effect of government spending on poverty in East Java Province. Government spending was examined in terms of spending on education, health, and social protection. The method used was a quantitative approach. Multiple regression analysis was used to test three hypotheses in the study. The population of this study is the Expenditure Realization Report by Function in the APBD of 38 districts/cities in East Java Province in 2017-2021. This study used a saturated sample so that the entire population was used as a sample, amounting to 190 data. The results showed that education expenditure affects poverty with a positive and significant direction of influence. Health expenditure affects poverty, although with a negative direction of influence. Social protection expenditure affects poverty, although with a negative direction of influence. This study is expected to contribute to the government in allocating government spending so that it is following priorities and is right on target for people who are more in need.


INTRODUCTION
Poverty in Indonesia is a serious problem and a burden on people's lives, with a lack of food, shelter, and access to education, health, and other public services (Sianturi, 2021). The problem of poverty is important to be studied further because it is feared to be a challenge for Indonesia in achieving the first goal (Goal-1) of the Sustainable Development Goals (SDGs). Indonesia's development goals are to increase economic growth and reduce poverty, unemployment, and income inequality.
Economic development in Indonesia is centred in Java, but the problem of poverty is also centred in Java due to the dense population living in Java. Java is the island with the largest population, with 14 million people (52%) of Indonesia's 27.5 million poor people in 2021 (Datadoks, 2022). East Java has Indonesia's second-highest economic growth rate (Ardiansyah, 2017). However, this fact has not reduced the poverty rate in East Java Province. Poverty in East Java Province is still relatively high, ranking third in Java Island, as shown in Figure 1 below. The cause of high poverty in East Java Province is a large number of residents, but not balanced with an increase in employment. The Covid-19 pandemic has caused poverty to increase, so some companies have to reduce some of their employees (Edris, 2020), which impacts increasing unemployment. According to BPS East Java, poverty in East Java Province makes people   Table 1 above shows that the proportion of social protection expenditure is lower than the proportion of education and health expenditure. This means the government is more focused on education and health. In 2020 and 2021, the proportion of education and health expenditures increased from the previous years. The increase in that year was caused by the Covid-19 pandemic, which paralyzed all sectors, so the government had to spend more funds to overcome poverty.
Research by Ariwuni (2019) and Misdawita (2018) states that education spending significantly negatively affects poverty. Sirait (2022) states that health spending has a significant negative effect on poverty, while social protection spending has a significant positive effect on poverty. In contrast, Khairunnisa (2021) states that education and health expenditures do not significantly affect poverty.
Government spending is one of the instruments of pro-poor budgeting. Not many studies relate pro-poor budgeting in the realm of public sector accounting. Referring to the results of bibliometric analysis conducted by researchers, pro-poor budgeting can be associated with public or government spending. Therefore, government spending needs to be studied further to determine the extent of the government's commitment through the realization of spending on education, health, and social protection in alleviating poverty in East Java Province. H1: Education spending affects poverty H2: Health expenditure affects poverty H3: Social protection expenditure affects poverty

LITERATURE REVIEW Government Spending Theory
According to Mangkosoebroto, the theory of government spending consists of macro theory and micro theory (Aini, 2020). Macro theory explains that expenditures made by the government to purchase goods or services are intended as subsidies that benefit national economic growth (Tjodi, 2018). Government spending in macro theory was developed by several experts such as Rostow pandemic in 2019-2020. Surabaya City is one of the areas that has succeeded in reducing the poverty rate compared to other districts/cities. This is because the proportion of spending in Surabaya City is quite high compared to other regions. Education Expenditure Based on Figure 3, it can be seen that the realization of education expenditure in each district/city has stagnated. Only a few regions experienced fluctuations, namely Surabaya City, Sidoarjo Regency, Malang Regency, and Jember Regency. In the year of the Covid-19 pandemic, namely 2020-2021, education spending has increased quite high. Surabaya City has the highest absorption of expenditure, whereas, in 2021, the absorption of education function expenditure in Surabaya City has doubled compared to other districts/cities. The high absorption of education expenditure is used to finance facilities and infrastructure to support learning during the Covid-19 pandemic.   Figure 4 shows that in 2017-2019, the absorption of health function expenditure in each district/city did not experience significant changes or stagnation. However, in 2020 and 2021, the absorption of health function expenditure increased significantly, even doubling the expenditure absorption from the previous year. This is because, in 2020 and 2021, the government disbursed a large enough budget to overcome the Covid-19 pandemic, such as spending on vaccine support, health worker incentives, and other health expenditures. Surabaya City and Sidoarjo Regency have the highest absorption of health expenditure during the Covid-19 pandemic.

Social Protection Expenditure
Based on Figure 5, it can be seen that the realization of social protection expenditure did not experience much fluctuation, even in several districts/cities experiencing stagnation, such as in Jombang, Kediri, Lamongan, Lumajang, Madiun, and Magetan. In 2017-2021, the highest absorption of social protection expenditure was in Surabaya City, Sidoarjo Regency, and Malang City. During the Covid-19 pandemic years, namely 2020 and 2021, the realization of social protection spending in each Regency / City has decreased compared to previous years. According to the Regional Fiscal Study of East Java Province in 2020, social protection expenditure is affected by the confusing policy and budget reallocation through the centralization of implementing the Covid-19 pandemic countermeasures.

Normality Test
The normality test aims to determine whether each variable is normally distributed. The data normality test can be done with the Kolmogorov-Smirnov test. The data criteria are said to be normally distributed, namely if the significance value is more than 0,05. Conversely, if the significance value is less than 0,05, the data is not normally distributed.

Multicollinearity Test
The multicollinearity test was conducted to determine whether the regression model found a strong correlation between the independent variables. The model is considered free from multicollinearity symptoms if the tolerance value > 0,10 and the VIF value < 10.  Table 3, it can be seen that the tolerance value of all independent variables is greater than 0.10, and the VIF value is smaller than 10, so it can be said that the independent variables are free from multicollinearity symptoms.

Autocorrelation Test
The autocorrelation test is carried out to test whether there is a correlation between confounding errors in period t and period t-1 (previous) in the regression model. The Durbin-Watson test can be used to test whether autocorrelation occurs. There is no autocorrelation if the calculation results DW> dU and 4-DW> dU. Based on Table 4, it can be seen that the Durbin-Watson value is 1,064. With 5% significance, the number of n = 190 and 3 independent variables (k = 3), the dL value is 1,7306, and the dU is 1,7947. The D-W value (1,064) < dU value (1,7947) and the 4-DW value (2,936) > dU (1,7947), it can be concluded that autocorrelation occurs. For this reason, it is necessary to transform the data to overcome this problem by changing the regression model into the form of the Durbin two-step equation with the Durbin-Watson d method. Based on Table 5, it can be seen that the Durbin-Watson value is 2,151. With a significance of 5%, the number of n=189 (because it is transformed, the number of samples is eliminated 1) and 3 independent variables (k = 3), the dL value is 1,7298 and dU is 1,7942. DW value (2,151) > dU value (1,7942) and 4-DW value (1,849) > dU value (1,7942), it can be concluded that there are no autocorrelation symptoms.

Heteroscedasticity Test
The heteroscedasticity test is carried out to test whether there is an inequality of variance in the regression model from the residuals of one observation to another. The Glejser test can be carried out to determine whether heteroscedasticity symptoms occur. Based on Table 6, it can be seen that the significance value of all independent variables is more than 0,05, so it can be said that there are no symptoms of heteroscedasticity in the regression model. Based on Table 7, the equation results are as follows: Y = 9,459 + 7,870X1 -5,203X2 -5,920X3 + e The multiple linear regression test results show that a one-unit increase in the Education Expenditure variable can increase the Poverty variable by 7,870. Increasing one unit of the Health Expenditure variable can reduce the Poverty variable by 5,203. A one-unit increase in the Social Protection Expenditure variable can reduce the Poverty variable by 5,920.

Determination Coefficient Test (R 2 Test)
The R 2 test measures the regression model's ability to explain the dependent variable's variance. Based on Table 8, the coefficient of determination is 18.4%, which shows that the independent variables (Education, Health, and Social Protection Expenditure) can explain the dependent variable (Poverty). While other variables influence the rest.

Model Fit Test (F Test)
The F test was conducted to test the effect of all independent variables on the dependent variable. The hypothesis requirement is accepted if the significance level is less than 0,05. Based on Table 9, the test results show a significance value of 0.000 <0.05, so it can be said that all independent variables simultaneously affect the dependent variable.

Partial Parameter Significance Test (t-test)
The t-test is conducted to determine the effect of each independent variable on the dependent variable. If the significance value is less than 0,05, H0 is rejected, and Ha is accepted. Based on Table 10, the significance value of the Education Expenditure variable is 0,000 < 0,05. Decision H0 is rejected, and H1 is accepted with the conclusion that Education Expenditure affects poverty with a positive and significant direction of influence. This means that education expenditure has a significant effect on increasing poverty.
The significance value of the Health Expenditure variable is 0,015 < 0,05, so decision H0 is rejected, and H2 is accepted with the conclusion that Health Expenditure affects poverty, although with a negative and significant direction of influence. This means that health expenditure has a significant effect on reducing poverty.
The significance value of the Social Protection Expenditure variable is 0,000 < 0,05, so decision H0 is rejected, and H3 is accepted with the conclusion that Social Protection Expenditure affects poverty, although with a negative and significant direction of influence. This means that Social Protection Expenditure has a significant effect on reducing poverty.

DISCUSSION The Effect of Education Expenditure on Poverty in East Java Province
Based on the t-test results, the significance of education expenditure in influencing poverty is 0,000 < 0,05, so H1 is accepted, which means that education expenditure significantly increases poverty. This result is not following the theory of government spending, which explains that government spending is used to improve the welfare of the community, one of which is so that people avoid poverty. Megawati & Sebayang (2018) said that education spending has not focused on improving the quality of education because the allocation is mostly used for educators' salaries. According to Iksan (2022) in the daily news Lingkarjatim.com, the education office of East Java Province has committed acts of fraud in the management of grant funds, school committee assistance funds, School Operational Assistance (BOS) funds, and Education Operational Support Costs (BPOPP). Jaka Jatim obtained the findings through the BPK RI LHP Notes for 2020 and 2021. This is what causes the allocation of education spending not to be absorbed effectively and on target, so it has not been able to reduce poverty.
This research is in line with the research of Utami & Rofiuddin (2022), Ningrum & Nuryadin (2021), and Wahyuni (2021), which say that education expenditure has a positive and significant effect on poverty. This study contradicts Demak (2020) and Febrianti (2022) research, which states that Education Expenditure significantly negatively affects poverty. The Effect of Health Expenditure on Poverty in East Java Province Based on the t-test results, the significance value of health expenditure in influencing poverty is 0,015 < 0,05, so H2 is accepted, which means that health expenditure significantly reduces poverty. By realizing large health function expenditures, local governments have adequate fiscal policies in carrying out programs to improve health status, such as free medication, improving the quality of nutrition, and providing maternal and child programs. Increasing the level of health is one of the factors to increase one's work productivity which impacts the quality and physical ability of a person to work (Thahir, 2021). A person who has a good level of health will be more productive in working to increase income. With a high income, a person will be free from the cycle of poverty.
This research aligns with Sari (2018) statement that Health Function Expenditure negatively and significantly affects poverty. In contrast to the research of Misdawita & Sari (2018), which says that Health Expenditure has a positive and significant effect on poverty.

The Effect of Social Protection Expenditure on Poverty in East Java Province
Based on the t-test results, the significance value of Social Protection Expenditure in influencing poverty is 0,000 < 0,05, so H3 is accepted, which means that Social Protection Expenditure significantly reduces poverty. Sinaga (2022) said that the PEN social protection cluster program with the greatest effect on poverty reduction is the pre-employment card program. The program is intended for people affected by Covid-19. With social protection from the state to the community, the community is guaranteed access to social protection and will gain access to opportunities both in the economic and social fields (Sihombing, 2022).
This research aligns with Aini (2020), which states that Social Protection Expenditure negatively and significantly affects poverty. In contrast, Sirait (2022) states that Social Protection Expenditure positively and significantly affects poverty.

CONCLUSION
The results of this study provide empirical evidence that education spending affects poverty, which means that higher education spending will impact increasing poverty. The increase in poverty is because education expenditure has not focused on improving the quality of education and has been misappropriated by irresponsible individuals. Health spending affects poverty, which means that health spending will reduce poverty. This is because a person with good health will be more productive in working to increase income. Social protection spending affects poverty, which means that social protection spending will reduce poverty. Through social protection funding, people are guaranteed access to social services the government provides.
The limitation of this study is that the research data is only within five years from 2017-2021, so it does not consider changes that have occurred in the past and the future. Future researchers are expected to expand the research time to compare research results. The variables in this study are still limited, and future researchers can add other variables that can affect poverty so that the results are more diverse.