One Decade, 20 Percent Education Budget: How About Causality Between Education Success and Poverty?

In 2019, exactly one decade of the government expenditure on education in Indonesia (central and local government) was allocated 20 percent. The purpose of this study was to analyze the causality relationship between government budget on education, education success (proxied by the mean years of schooling), and poverty (proxied by the number of poor people) in Indonesia. The data analyzed is secondary data, to be precise panel data from 34 provinces in Indonesia over a period of five years (20152019). The analysis technique used is the Granger Causality Test. The results showed that the government budget on education had a significant effect on the mean years of schooling and had a causal relationship with poverty. Meanwhile, poverty has been shown to affect the mean years of schooling. Based on the results of this analysis, it is for the government to consistently prioritize the budget for the education sector because it is proven to have an impact on education success and poverty alleviation. In addition, the government also needs to pursue poverty alleviation programs such as subsidizing cash assistance for student in poverty.


INTRODUCTION
The Sustainable Development Goals (SDGs), also known as global goals, were adopted by all member states of the United Nations (UN) in 2015 as a universal call to act to end poverty, protect the planet and ensure that all people will have peace and prosperity by 2030 (www.id.undp.or).
Indonesia is a developing country that will soon become a developed country so it is very serious in realizing the SDGs. Indonesia is also one of the countries with the largest population in the world, and in 2045 it is predicted that there will be a demographic bonus in Indonesia (Wisnumurti et al., 2018). Education is a factor that really determines the quality of the nation's generation. The provision of education requires at least the cost of fulfilling the financing to provide service standards.
However, the education process cannot run without money, so there needs to be support from the government, such as the APBD for education (Nandani et al., 2018 both in terms of allocation and use.

Education
Education is an effort to guide children from birth to reach physical and spiritual maturity, in the interaction of nature and its environment (Nurkholis, 2013). The measurement of educational  increase, greater than the personal costs that must be incurred. To be able to maximize the difference between expected benefits and estimated costs, the optimal strategy for someone is to try to complete education as high as possible (Todaro, 2000). Basically, the definition of poverty can be seen from two sides, namely absolute poverty and relative poverty.
According to Todaro (2000)    The causality relationship is a shortterm relationship between certain groups using an econometric approach that includes a reciprocal relationship (Fauzi, 2007). With the optimal lag length, p, the Granger causality test working principle on panel data is based on the pooled regression model as follows.
2) The education budget does not affect poverty and poverty does not affect the education budget.
3) The success of education does not affect poverty and poverty does not affect the success of education.
The basis for rejection of the null hypothesis is by using a probability criterion smaller than 0.1 (<0.1).

Unit Root Test
Before carrying out the Granger Causality Test, a unit root test is carried out so that the data becomes stationary.
In addition, it can also be seen whether the data contains unit roots or not. If the variable contains unit roots, then the data is said to be non-stationary.

Lag Length Test
Granger causality test need to pay attention to determining the lag length.
The lag length must be correct because it is important to avoid misspecified models due to the lag length being too short, or reducing the degree of freedom too long. Lag testing using the Eviews application can make it easier to determine the right lag length. So that good and correct data results can be obtained. In Eviews, determining the optimal lag is by looking at the sign (*) at the most for each lag option. Table 2 shows the results of the lag length test in the research model. Based on the output shown in Table 2, the optimal lag according to the LR, FPE, AIC, and HQ criteria is the smallest and most (*) shown in lag 3.

Granger Causality Test
Granger causality test is used to see the direction of the relationship between education and poverty. Whether there is a relationship can be seen from the probability value of each causality test which is then compared with alpha 0.1. Based on the output of Table 3 shows that: 1 Thus, it can be concluded that the Granger causality test has a two-way relationship between the education budget and poverty.
3) The success of education statistically does not significantly affect poverty, which is indicated by a probability value of 0.1593> α = 0.1 at lag 3.