THE EFFECT OF GOVERNMENT EDUCATION SPENDING ON ECONOMIC GROWTH THROUGH SCHOOL ENROLLMENT IN INDONESIA

This paper examines the effect of government spending on education on growth as an indirect process through enrollment rate, specifically in Indonesia. Using a panel data set of 26 provinces from 2000 to 2010 and applying fixed effect approach of panel regression, we find that government spending on education has a positive relationship with enrollment with one-year lag; also, the data show that enrollment affect growth significantly. By enacting the chain rule, government spending on education has a positive effect in economic growth indirectly. Together, these findings suggest that raise government spending on education can increase enrollment rate and enhance economic growth.


INTRODUCTION
countries' data, they cover many nations or long Maintaining government spending on time span.No study has focused on one education has been a major goal in Indonesia, developing country's case, especially Indonesia's.where the government has been struggling since As education spending will have different effect in 2003 to keep the allocation for education at or different circumstances, it is important to study above 20% of the national and regionalbudget, country specific cases.Our paper will attempt to based on Law no.20 of 2003.Indonesia has this fill this gap by focusing on Indonesia alone.budget requirement for education because The paper will analyze the effect of government believes that it will enhance the government education spending on growth as an country's economic growth.A large body of indirect process through enrollment rate.Thus, it research supports the belief that government aims to find the effect of that spending on spending on education influences economic enrollment rate and economic growth, and the growth significantly (Baldacci, Clements, Gupta, direct effect of school enrollment rate on growth, and Cui, 2008; Barro & Sala-I-Martin, 1995; specifically in Indonesia, using a panel data set of Dauda, 2010;Musila & Belassi, 2004;and 26 provinces from 2000 to 2010.Prasetya & Pangestuty, 2012).
This paper is organized as follows:section 2 While most researchers agree there is provides the literature review, section 3 discusses relationship between the two, some do not the model and the dataset, section 4 presents the consider that government spending must first results, and section 5 offers the policy enhance human capital, to improve education, as implication.measured by enrollment rate (Baldacci et al., 2008;Gupta, Verhoeven, & Tiongson, 2002;and 2. LITERATURE REVIEW Rajkumar & Swaroop, 2007).Increased The relationship between government enrollment rates are likely to expand economic education spending and growth has been widely growth (Jalil & Idrees, 2013;McMahon, 1998; discussed among scholars with various results.and Wolff, 2000).Some researchers assert it has a positive and Although some studies look at the effect of significant relationship, such as Barro & Sala-Ihuman capital on growth and use developing Martin (1995), Musila & Belassi (2004), Baldacci et al., (2008), Dauda (2010), and

THE EFFECT OF GOVERNMENT EDUCATION SPENDING ON ECONOMIC GROWTH THROUGH SCHOOL ENROLLMENT IN INDONESIA
The first process is the effect of government Prasetya & Pangestuty (2012).On the other hand, education spending on enrollment rate.Lotto (2011) found that, in Nigeria, the Researches claim it is a positive and significant relationship is negative or not significant in the one (Baldacci et al, 2008, and Gupta, Verhoeven, short run while possibly positive in the long run. and Tiongson, 2002); especially, if supported by Additionally, Blankenau and Simpson (2004) good governance (Rajkumar and Swaroop, concluded that the relationship between 2007).Using cross-section data, Rajkumar and education spending and economic growth exists Swaroop (2007) define educational attainment as because government provides financial support the proportion of school-aged children who finish for formal schooling in many countries; however, secondary school.However, Gupta, Verhoeven, there is no clear empirical validation of this and Tiongson (2002), define it as gross relationship because education spending may enrollment ratio in primary and secondary crowd out other growth factors such as tax education; similarly, Mankiw, Romer and Weil structure and government size.
(1992) use enrollment rate as a proxy of human Observing a strong relationship between capital accumulation.economic growth and public spending on In Baldacci et al. (2008), education is education, many scholars treat the latter as a measured by the summation of primary and direct factor of growth, such as Barro (1990), secondary enrollment rate and given a separate Blankenau and Simpson (2004) and Weil, 1992;and Wolff, 2000).All of them use capital as the determinant of economic growth.
international data; for instance, Mankiw, Romer, From this point of view, government education and Weil (1992) use 98 countries, Barro and Salaspending should be treated as an indirect factor I-Martin (1995) use 97, andWolff (2000) use 24 for economic growth through enrollment members of the OECD.A time series approach (Baldacci et al., 2008).Thus, we must first also proves that different levels of education examine the process from government education positively and significantly affect to economic spending to enrollment rate, and then from growth, for example Pakistan from 1960 to 2000 enrollment rate to economic growth.(Jalil & Idrees, 2013).
Using East Asian countries' data from 1965 to 1990 and applying the ordinary least square with correction for heteroscedasticity and autocorre-lation analysis for the data, McMahon (1998) finds that primary education spending from the government is very important in the early period.However, this spending matters less after primary education has generally been   (2002).They mainly use gross enrollment ratio enrollment rate, and economic growth in two (GER) as the education variable.They also use processes.First, we use the education equation to many variables to explain GER, such as initial show the effect of government education income, education spending, child mortality, spending on enrolment rate.Second, we use the urbanization, share of female students, and Solow growth model with enrollment rate as a repetition rate.However, due to limited time and measurement of human capital.Finally, using data availability, this paper only used initial both models and applying the chain rule, we income as a control variable.Therefore, the obtain the effect of education spending on general equation is as follows: growth.
To be consistent with previous research related to growth, we employ a panel data approach in both growth and education equations to discern not only the time effect, but also the This equation observes the impact of individual effect (Islam, 1995).Using panel data, education spending on GER in each level of we will see the individual effect in each province schooling where it is a proxy of human capital in in Indonesia will in this study.Ours is a regional, the growth equation.The use of gross enrollment not national, analysis.rate as a proxy is consistent with much of the literature examining the role of human capital on 3. MODEL AND DATA growth (Baldacci et al., 2008;Mankiw, Romer, & 3.1. Model Weil, 1992).To obtain the impact of government Indonesia has four levels of education (see spending on education on growth, this paper uses Appendix 1): pre-school, basic, secondary and general specifications for real per capita income higher education.In this paper, we used basic and growth and enrollment rate as used in the existing secondary education without accounting for nonliterature.Baldacci et al. (2008) build a separate formal education.Basic education is primary model for the education equation, using school, while secondary education is consisted of enrollment rate, as a determinant of economic lower and upper secondary school.GERs are growth.Therefore, we employ two models, a taken for each level of schooling.Therefore, we growth model and an education model.In both, had three education equations representing three we include lagged values of government levels of schooling.We used the GER of each spending and education variables.
schooling level as the education measurement.Baltagi (2001) argues that panel approach is Some variables that will influence GER, based on the most applicable method for growth model, the equation above are: since it does not require technology to be the • Income level (y).An increase in per capita same across individuals and provides a better income tends to increase demand for dynamic effect for each.Islam (1995) adds that it education, making a positive coefficient.lets us separate the effect of "capital deepening" This variable is the logarithm of real per and technological and institutional differences.capita gross regional domestic product (GRDP).

• Education Spending (EduSpending). This
? is real GRDP growth.variable expresses education spending of ?denotes the investment ratio, in terms of provincial and municipal governments as a gross fixed capital formation (GFCF) per percent of GRDP.A one period lag account GRDP.A higher investment ratio raises the for the attainment time of the spending's stock of physical capital.impact.
? refers to the stock of human capital, • h and m indicate the time-specific effect and represented by GER.The GER of each province-specific effect, respectively.
school's level will be estimated separately to • e is error term.(see Appendix 2 for the derivation): ? defines the lagged logarithm of per capita income.The negative direction of growth as initial per capita income increases shows that there is a conditional convergence to the steady state.
? h t and m i indicate the time-specific effect and province-specific effects, respectively.
? e is error term.The aim of this study is to investigate the cannot provide comparable and consistent data effect of education spending at a provincial level (see Table 1 for group of provinces, and Table 2 on economic growth indirectly through for more description of the data).
enrollment.Tables 2 and 3 show results for Data for GRDP, GFCF, and GER are from growth and education equations.Per previous Statistic Indonesia, while data for spending is discussions, we used the panel approach to from Ministry of Finance.obtain the best panel approach.The results of the coefficient of income level means that an Hausman test (see Appendix 3), for both models, additional 1% in income will increase GER by showed that applying fixed effect is more 0.09, 0.06, and 0.56 point in elementary, lower appropriate.In most cases, the coefficients are secondary, and upper secondary school, statistically significant and the directions areas respectively.expected (see Appendix 4 and 5 for more detail results).

Growth Equation
The impact of different levels of schooling 4.1.Education Equation on economic growth may vary between nations Gov ernment spending on education positively (Jalil & Idrees, 2013); also, within regions of a affects GER after a one-year lag.Baldacci, et al nation.In the human capital augmented (2008) found the similar result.The coefficient production function (growth equation), we implies that an increase in education spending of substituted the human capital variable with GER 1% of GRDP would increase enrollment rates by of each level of schooling to measure the effect a factor of 0.13 for primary, 0.12 for lower from each of them on growth.We separated secondary, and 1.12 for upper secondary in the primary, lower secondary, and upper secondary following year (see Table 3).
school into three equations to avoid the risk of The income level also matters significantly multicollinearity, which would lead to incorrect to GER, especially last year income.The inferences.All levels of GER were found to capita have negative directions, which implies and significantly contribute to economic growth that each province in Indonesia will converge to a at a 1% statistical significance level, whether steady state.This result matches the convergence using 2SLS or IV regression, except for an concept in the neoclassical growth model (Solow, insignificant coefficient of primary GER using 1956) which says that a country's per capita IV regression.In the 2SLS regression, a 1% growth rate tends to be inversely related to its increase in the primary, lower secondary and starting level.upper secondary GER should raise the growth 29 rate by 0.025, 0.028, and 0.003 points, 4.3.The effect of education spending on respectively.In the IV regression, the instrument growth through enrollment rate variables are education spending ratio in terms of

Variable
We calculated the effect of government GRDP per capita, total value of education education spending on growth through spending, initial income, and log of population enrollment rate based on the marginal effect of growth.The coefficient of GER means that a 1% education spending on GER and the marginal increase in the lower and upper secondary GER effect of GER on growth.Based on the chain rule, should increase the growth rate by 0.056 and the multiplication of these effects will be the 30 0.019 percentage points , respectively.effect of education spending on growth.

These results are in line with McMahon
The table 5 shows that increasing 0.01 (1998) and Jalil and Idrees (2013).They could be percent of education spending in terms of GRDP due to investment in human capital by household will increase growth by 0.003 point using 2SLS or government, as well as physical capital, but regression; while using IV regression, 0.01 probably not the technological progress largely percent increase of education spending will responsible for the high per capita growth in East increase growth by around 0.002 point.Asia (McMahon, 1998).
Therefore, government spending has a positive Another interesting matter is the coefficients impact on economic growth. of initial GRDP per capita.In both 2SLS and IV regression, we found that initial GRDPs per    Where Y is output, H is human capital, K is physical capital, L is labor, and A is the level of technology.L and A are assumed to grow exogenously at rates n and g so that: The intensive form of the Solow production function by defining all variables per effective labor is: Assuming both physical and human capitals are accumulating factors, their equations of motion for are: Physical capital: Thus, the motion for physical capital is

Figure 1 .
Process from education spending to economic growth it better understand the progress in developing 3.1.2.Growth equation human capital for each province at different Drawing upon Mankiw, Romer and Weil stages.(1992), the growth equation is based on Solow's ? is summation of population growth, neoclassical growth framework.It adds human technological and depreciation rate.Following capital accumulation to the Solow growth model.Mankiw, Romer, and Weil (1992), we assume that The growth per capita output equation is below is 0.05.
(2002) chapter 2,p.45 on the summary of This study attempted to examine the effect of functional forms involving logarithm.For a level-log model, government education spending on growth the interpretation of So, a one percent through school enrollment in Indonesia during increase in x will raise y by 30 Loc.cit. the period 2000 to 2010.It has investigated using

Apendix 2 .
Derivation of growth equationFollowingMankiw, Romer, and Weil (1992) model, the equation started with Solow growth model by adding human capital accumulation and featuring Cobb-Douglas production function is:

Table 1 .
Group of provinces it

Table 2 .
At start, we had 286 estimate the education equation and growth data points, but only 260 were effective after model.The panel method has two kinds of effect, regression due to limited availability of data (e.g.fixed and random.We employed the Hausman GER of Aceh in 2000 and 2001) and generating test before running the regression in order to Summary descriptive statistics

Table 3 .
Education equation

Table 4 .
Growth equation

Table 5 .
Marginal Effect a Taken from Table3b Taken from Table4c Calculation based on chain rule where is education spending on growth, is education spending on GER, and is GER on growth.
S1, S2, S3) are equalivalent to Bachelor Master, Ph.D. Specialist programs are programs for academic and/or professional futher education Basic education is compulsory and free of change Pre-school is optional Appendix 1. Table of education system in Indonesia Note.TVETipedia, retrieved from www.unevoc.unesco.org