CAUSAL RELATIONSHIP BETWEEN GOVERNMENT EXPENDITURE, ECONOMIC GROWTH, AND POVERTY IN SOUTH KALIMANTAN HUBUNGAN KAUSAL ANTARA PENGELUARAN PEMERINTAH, PERTUMBUHAN EKONOMI, DAN KEMISKINAN DI KALIMANTAN SELATAN

The poverty rate in South Kalimantan is the 3rd lowest nationally in 2020 and the lowest in Kalimantan Region. However, the downward trend in the prices of several commodities that have been the backbone of the economy, such as rubber, palm oil, and coal, greatly affects the population's income, especially in rural areas. The study aims to analyze the effect of government expenditure on economic growth and poverty in South Kalimantan. The study used two analytical approaches: Descriptive and Inferential Analysis. A Scatter plot is used in Descriptive analysis to describe the relationship between government expenditure on economic growth and poverty. Cross-Sectional Time-Series FGLS and Generalized Ridge Regression modeling were used in the inferential analysis. Three out of five government expenditures statistically significantly affect economic growth: indirect personnel, direct personnel, and capital expenditure. Two of the five functions of government spending positively affect poverty reduction: direct and indirect spending.


INTRODUCTION
Government expenditure is one of the instruments used by the government to foster economic growth. The higher the spending, the more positive impact it will have on the economic growth, while small government spending will be reducing the economic activity of a region, thus potentially slowing down the economy. Based on data from the Statistical Bureau of South Kalimantan Province in 2020, the poverty rate in Kalimantan Selatan is the lowest in the Kalimantan region and the third lowest across provinces in Indonesia after DKI Jakarta and Bali. However, the downward trend in commodity prices, such as rubber, palm oil, and coal, dramatically affects the population's income, especially in rural areas. It is also affecting the poverty rate through the increase in the unemployment rate. The downward trend in commodity prices further results in stagnant economic growth in Kalimantan Selatan, which is still very dependent on the real sector as their backbone of economies.
The abovementioned conditions require serious government attention in addressing poverty and economic growth in Kalimantan Selatan. The local Government in Kalimantan Selatan has made efforts to reduce poverty and increase economic growth through various strategies, programs, and activities through government expenditure. To date, there has been no comprehensive study that examines the relationship between local government expenditure, economic growth, and poverty levels in Kalimantan Selatan across regencies and cities and focuses explicitly on poverty alleviation programs.

Review of Literature
Government expenditure has a theoretical basis seen from the balance of national income: Y = C + I + G + (X-M). From this equation, an increase or decrease in government expenditure will affect the aggregate national income (Mutia, 2021). Some considerations underlie government decisions in regulating their spending; it is not enough for the government to only achieve the ultimate goal of their expenditure policies and consider the intermediate targets who will benefit from the policy (Jermsittiparsert, 2019). Increasing expenditure for the sole purpose of increasing national income or expanding employment opportunities is not sufficient. Still, it also must take into account the employment aspect and how it will benefit them financially. The government also needs to avoid increasing its role in the economy from undermining the activities of the private sector (Dilliana, 2019).
Regional spending is used as part of government affairs that fall under the authority of the provincial government. According to the Minister of Home Affairs Regulation No.13, the year 2006 on Regional Financial Management Guidelines, the government expenditure structure consists of indirect and direct expenditures. Indirect expenditure includes personnel expenditure, interest expenditure, subsidy expenditure, grants expenditure, social assistance expenditure, revenue sharing fund expenditure, regional financing allocation, and unexpected expenditure (Dudzevičiūtė, 2018). Direct expenditure includes personnel expenditure, goods and services expenditure, and capital expenditure. Thus, based on the categorization, Government expenditure can be categorized as "exhaustive and transfers." Exhaustive means the purchase of goods and services directly consumed or used to produce other goods. In addition, government expenditure can also be money transfers to individuals for social purposes, companies as subsidies, or grants (Rini, 2018).
Government expenditure can also be categorized as cyclical and structural (Rahman, 2019). Cyclical expenditure usually occurred during an economic crisis or recession when there is a rise in the unemployment rates (Faisol, 2018). The role of government spending is to support and ease the unemployed while also stimulating economic growth by increasing overall expenditures (individual, household, business, and government). Structural expenditure is a necessary expenditure that occurs regardless of the condition of the economy (Kotlebova, 2020). Some examples of the expenditure are education, health, infrastructure, social security, defence, and more.
Based on economic growth theory, government expenditure directly relates to economic growth by allocating public resources and spending, which ultimately contributes to increased aggregate demand in the economy (Rini, 2018). Government spending aims to increase production capacity in projects that refer to economic growth, equal distribution of income, increase in welfare, and programs that directly touch underdeveloped areas. An increase in expenditure, in general, can lead to a more considerable increase in economic output because spending by one individual, household, business, or the government is income for another (Maneerat, 2020). Therefore, there is an urgency for the government to control and manage the economy through the central bank.
As mentioned earlier, high economic growth comes from increased government spending, which is the best part of the regional autonomy, which can empower the potential of each region (Nguyen, 2018). The granting of regional autonomy affects economic growth. It gives local governments the freedom to make their financial plans and policies to carry out economic development by optimally and sufficiently managing their resources. Local governments are expected to manage and develop the public sector to increase regional economic growth (Sepehrdoust, 2019).
Although the economic growth rate does not automatically answer various welfare questions and problems, it remains an essential element in its development programs to eradicate poverty (Amar, 2020). Multiple studies conducted by economists have noted that economic growth plays a vital role in reducing long-term poverty levels. Rapid economic growth and equitable distribution of income must be separated as development goals (Ding, 2019). Sometimes, these things cannot grow together since high economic growth does not necessarily guarantee a better income distribution.
The focus on economic growth stems from a focus on broad-based growth (Zafar, 2019). Broad-based growth means that the growth is felt and enjoyed by all people at all economic strata. However, rapid economic growth and more equitable income distribution can be achieved all at once, and few countries have proved it. The policy should not be emphasized on a development strategy that maximizes economic growth only enjoyed by a handful of people or an approach that focuses on more equitable income distribution (Chugunov, 2021). Still, these two things are essential to be achieved together.
Economic growth may generate and open opportunities and chances to improve the welfare of the people (Trabelsi, 2018). Strong growth and availability of job opportunities will result in stable household income, leading to an increase in spending and a strong economy. This will result in poverty reduction. The greatest challenge for the government is to design and implement policies that enable the poor to participate in the economy and so contribute to that growth. This includes policies to make labour markets work better, remove gender inequalities and increase financial inclusion.
The problem of poverty, which is synonymous with inadequate community income, must always be a priority in development (Syam, 2018). The agenda of the poverty eradication program is related to many factors caused by poverty itself since the effect of poverty will always be related to the fundamental conditions of society. Poverty eradication efforts are also inseparable from the amount of government expenditure. Even if the amount of government spending continues to increase yearly in the hope that it will improve people's welfare and reduce poverty, it should also see the accuracy of targeting in government spending. This is where the role of the government, through its expenditure, provides easier access for the poor to obtain basic services.

METHODS
This research was conducted at 9 (nine) Regencies and 2 (two) Cities in South Kalimantan from 2010-2019. This study uses two analytical approaches; Descriptive and Inferential Analysis. The descriptive analysis includes two-way graphs and tables and shows the relationship between government spending on economic growth and poverty. A Scatter Plot diagram is used to see a pattern of the relationship between 2 (two) variables with a numerical scale (interval or ratio).
The inferential analysis uses the Cross-Sectional Time-Series FGLS and the Generalized Ridge Regression for the model. Determination of the best model is based on several statistically relevant considerations; the number of observations, variable conditions, differences in the fulfilment of modelling assumptions, and the nature of data collection (panels and time series). Thus, the model used is also different for each group data set. At the regency and city level, the model used is the Cross-Sectional Time-Series FGLS Regression. While for provincial level, the Generalized Ridge Regression Model is used.
The Cross-Sectional Time-Series FGLS Regression Model is used for panel data where heteroscedasticity is detected in the data condition. This regression uses the Generalized Least Square to estimate model parameters. By using this method, the error variant will be homoscedasticity.
The bootstrap method was used to solve the limited number of observations (10 observations in 10 years from 2010 to 2019). The bootstrap method itself is based on resampling sample data with the condition that the data is returned in completing the sample size statistics hoping that the sample represents the actual population data. Bootstrapping allows the use of parametric statistical estimation, i.e., the Ridge Regression can be applied.
Both types of models use a data structure with a lag t-1 condition on the independent variable. This is based on the assumption that spending does not necessarily impact economic growth and poverty in the current year or equal economic growth and poverty. However, the impact was only seen in the following year.
The functional form can overcome the weakness of the intercept interpretation of ordinary linear functions, which sometimes do not match the substance of the problem. To avoid this, this study uses the Double Log functional form or the log-log model. The loglog model, often called the double log model or constant elasticity model transforms from a nonlinear model to a linear model by making a model in logarithmic form. More specifically, the independent variables and the dependent variable are changed into natural logarithms (ln). Consequently, the interpretation of the beta coefficient/parameter model values becomes the elasticity value.

Government Expenditure Model on Economic Growth in Regencies and Cities in South Kalimantan
The model used is calculated based on 2 (two) data sets: data from regencies and cities in South Kalimantan and (aggregate) data from South Kalimantan Province. Each data set has 2 (two) equations, the equation for economic growth and poverty. Thus, there are four operational models for this research.
The model of government spending on economic growth in Regencies and Cities in South Kalimantan is compiled by an independent variable as in economic growth and independent variables that are thought to affect economic growth, namely indirect personnel expenditure, grants, and social assistance expenditures, direct personnel expenditure, capital expenditure, and regional financing. The following is a summary table of the estimated model parameters.

Table 1. Estimation of Model Parameters of the Cross-Sectional Time Series FGLS Regression of Government Expenditures on Economic Growth
Variables In general, the Cross-Sectional Time-Series FGLS Regression model of government spending on economic growth is statistically significant. The P-value indicates this in the model, which is smaller than the α at 0.05. Thus, there is at least one type of government expenditure that affects economic growth. The test is then followed up with the partial test. Three out of five types of expenditure are statistically significant through a partial test of each independent variable (types of government spending). Still, the remaining 2 (two) are not substantial. Indirect personnel expenditure and direct personnel expenditure are significant at α = 5%, while capital expenditure is significant at α = 10%. The different levels of significance are influenced by the level of confidence taken by the researcher. The smaller the level of significance used, the greater the data needed by the researcher and vice versa.

Cross-Sectional Time-Series FGLS Regression
Although significant, the three types of expenditure have an inverse relationship. A negative sign indicates this in the regression coefficient value for each of the three types of expenditure. The regression coefficient value on indirect personnel expenditure is 0.7981, suggesting that the indirect personnel expenditure grows by one per cent. At the same time, the other independent variables remained constant (ceteris paribus), reducing economic growth by 0.79 per cent. Meanwhile, if direct personnel expenditure grows by one per cent, while the other independent variable remains constant (ceteris paribus), it will reduce economic growth by 0.33 per cent. And if the capital expenditure grows by one per cent, while the other independent variables are remained constant (ceteris paribus), it will reduce economic growth by 0.46 per cent. The scatter plot also shows this opposite relationship. There is a tendency for economic growth to decline when indirect personnel expenditure, direct personnel expenditure, and capital expenditure continue to increase. 2.000e+08 4.000e+08 6.000e+08 8.000e+08 pt 2.00 4.00 6.00 8.00 10.00 lpe 0 1.000e+08 2.000e+08 3.000e+08 4.000e+08 pl 2.00 4.00 6.00 8.00 10.00 lpe 0 2.000e+08 4.000e+08 6.000e+08 8.000e+08 m spending still indirectly impacted economic growth. Regional economic growth has slowed down in the last ten years in South Kalimantan due to the weakening financial situation. Personnel expenditure and capital expenditure have continued to increase due to increased demand in the government sector.

Government Expenditure Model on Economic Growth in South Kalimantan Province
The model for government spending and economic growth in South Kalimantan is constructed by economic growth as the dependent variable and regional financing and capital expenditure as an independent variable. The choice of independent variables was different from the model for Regencies and Cities due to the multicollinearity, which is very high. Some other government expenditures cannot be included in the model. The following is a summary table of the estimated model parameters. High multicollinearity occurs because there is a strong correlation between the two independent variables. Thus, it will cause a change in the value of the partial regression coefficient, a greater standard error value, and the value of the confidence interval is very wide, so it is difficult to reject the null hypothesis. Based on the statistical output, it is concluded that the relationship between all of the government expenditures on economic growth using the Generalized Ridge Regression model is statistically significant. This conclusion is drawn because the P-value of the model is less than α = 5%. Thus, testing can be continued partially for each independent variable.
The partial test results show that only the capital expenditure variable is statistically significant. The regional financing variable is not significant because the p-value is far above 5%. The magnitude of the effect of capital expenditure is indicated by the regression coefficient value of -1.8565. This means that when the capital expenditure of the Provincial Government of South Kalimantan grew by one per cent, then the economic growth contracted by 1.86 per cent. The results of this modelling are also in line with the scatter graph between capital expenditure and economic growth, which shows a pattern in opposite directions. Economic growth tends to decline at a time when capital spending continues to increase.
These findings are not in line with the various literature. This is because an efficient capital expenditure will generally be able to boost economic growth. However, in this study, the opposite happened. Conditions can occur if the use of capital does not directly have a practical impact on efforts to increase economic growth. Most of the capital expenditures were capitalized in non-productive businesses. The situation is the same as in the regencies and cities model; South Kalimantan's economic growth has tended to experience a slowdown in the last ten years due to a weakening economic situation, while capital expenditure has continued to increase due to increased demand in the government sector.

Figure 2. Two-Way Relationship between Capital Expenditures and Economic Growth in South Kalimantan, 2010-2019
Source: Statistical Bureau, 2020

Government Expenditure Model on Poverty in Regencies and Cities in South Kalimantan
The percentage of poor people compiles the model of government spending on poverty in regencies and cities in South Kalimantan as a dependent variable, indirect personnel expenditure, grants, social assistance spending, direct personnel expenditure, capital expenditure, and regional financing as the independent variable. The following is a summary table of the estimated model parameters. The model of government expenditure on poverty in regencies and cities in South Kalimantan is statistically significant. It can be further interpreted according to the independent variable. The partial test results show that only direct and indirect personnel expenditures are significant in the model. Meanwhile, the other three variables did not meet the significance. The two variables have a negative relationship at -0.77 and -0.29, respectively. The result is in line with various theories; if indirect personnel expenditure grows by one per cent while other influencing factors remain constant (ceteris paribus), it will reduce poverty by 0.77 per cent. The same is true for direct personnel expenditure; If direct personnel expenditure grows by one per cent, while other influencing factors remain constant (ceteris paribus), it will reduce poverty by 0.29 per cent. This result confirms the variables' scatter plot graph: poverty with indirect personnel and direct personnel expenditures. The percentage of poverty continues to decline when there is an increase in indirect personnel and direct personnel expenditures.

Government Expenditure Model on Poverty Rate in South Kalimantan
The model of government spending on economic growth in South Kalimantan is compiled by a dependent variable, namely economic growth, and an independent variable that is thought to affect economic growth, namely grants and social assistance expenditure, fund sharing expenditure, and capital expenditure. The choice of independent variables was different from regencies and cities due to the very high multicollinearity, so some expenditure variables cannot be included in the model. The following is a summary table of the estimated model parameters. The significance of the government expenditure model on poverty in South Kalimantan Province is statistically significant, which is indicated by the p-value of 0.037, which is less than 0.05. Meanwhile, partially, only grants and social assistance expenditure significantly affect poverty in South Kalimantan. The regression coefficient for grants and social assistance expenditure is -0.4452. This means that if the grants and social assistance expenditure grow by one per cent, then poverty in South Kalimantan can reduce by 0.44 per cent. These results are also explained in more detail with a scatter plot graph between capital expenditure and economic growth. When capital expenditure continues to increase, the poverty rate in South Kalimantan tends to decline.

CONCLUSION
Based on the research findings, it can be concluded that there are 3 out of 5 types of government expenditure that statistically have a significant effect on economic growth in South Kalimantan; indirect personnel expenditure, direct personnel expenditure, and capital expenditure. Although significant, the three types of expenditures have opposite relationships. This means that even though indirect personnel expenditure, direct personnel expenditure, and capital expenditure each grew 0.79-1 per cent, they have not increased economic growth significantly. Economic growth declined at a time when those aforementioned variables continued to increase. Two out of five types of government expenditure positively affect reducing poverty; direct spending and indirect spending. However, these two expenditures each grew only 1 per cent.

RECOMMENDATION
Several things that can be recommended in accordance with the results of this study are: the government needs to increase the budget for the five functions of government spending by preparing an appropriate and proportional budget for each instrument. The increase in government spending also needs to be done because it has a direct impact on the poor as beneficiaries. In addition, other cross-sectoral support is needed that can affect the reduction or reduction of poverty. So that the policy direction issued by the government is right on target. Further research needs to be done, especially after the Covid-19 pandemic which has a direct impact on the community, especially on economic aspects such as poverty, reduced income, unemployment, and employment.