THE ROLE OF COALITION PARTIES AND LOCAL GOVERNMENT EXPENDITURE: EVIDENCE FROM THE DEVELOPING COUNTRY

How to cite this paper: Rizky, H. P., Setiawan, D., & Winarna, J. (2021). The role of coalition parties and local government expenditure: Evidence from the developing country. Corporate Governance and Organizational Behavior Review, 5(2), 82–89. https://doi.org/10.22495/cgobrv5i2p8 Copyright © 2021 The Authors This work is licensed under a Creative Commons Attribution 4.0 International License (CC BY 4.0). https://creativecommons.org/licenses/by/


INTRODUCTION
This study aims at examining the role of coalition parties on the local government's decision on expenditures. Coalition parties are the coalition of several political parties that support the regional heads. This coalition is mostly established during the election of the regional head of the local government. This study focus on the coalition parties in the local government context in the emerging country. The previous study shows the importance of coalition parties on the local government decision in the public affair (Hendrawan, Berenschot, & Aspinall, 2021; Lewis & Hendrawan, 2019). Political parties established coalition parties to maximize the payoffs (Debus & Gross, 2016). This study examines the effect of coalition parties in the local government in the Republic of Indonesia on the decision to spend the money through mandatory expenditure, education expenditure, and health expenditure.
This study focuses on the role of coalition parties on local government expenditure, specifically education and health expenditure using the Indonesian context. Education and health expenditure are mandatory for the government. This is based on Law Number 36 of 2009 concerning health and the 1945 Constitution concerning education by the state. Furthermore, this expenditure is required by the government in line with Law Number 17 of 2003 concerning state finances. The two budgets are included in the mandatory state expenditure regulated by law. Mandatory expenditure comprises the budget for education, health, infrastructure, and village funds.
Each regional head election candidate is endorsed by a coalition of parties in legislative council seats. The legislature or Regional House of Representatives (DPRD) is directly elected by the people to represent their views in politics (Martin & Vanberg, 2014). Coalitions comprise several parties that jointly support candidates for regional heads. They are formed based on similar views of a good government system (Sitepu, 2016). Moreover, member parties make policy agreements with supported candidates to assist in goal realization (Kellam, 2017). Lewis (2020) provides evidence that coalition party in the pre-election periods have a significant effect on the outcome of the direct election in Indonesia. Candidates backed up by the larger coalition party in the pre-election period have a higher possibility to win the election.
The number of parties making up the government influences the system by forming the rewards associated with council seats (Ziegfeld, 2012). According to Lewis and Hendrawan (2019), political party coalitions influence public policies through regional heads when their candidates are elected. These policies include local government expenditure on public services through regional budgets.
Sitepu (2016) stated that national coalitions significantly affect the approval of the revenue and expenditure budget. According to Tans (2012), the most effective local government in Indonesia has a majority in the legislature, with most political parties supporting regional heads. Beland and Oloomi (2017) showed that party affiliation affected fund allocation, but not expenditure. Greer (2011) stated that policymaking with negligible coalition supporting the government is beneficial but ineffective. Additionally, the role of DPRD members in preparing the Regional Revenue and Expenditure Budget (APBD) raises problems, including corruption. According to detik.com, most DPRD members in Malang are involved in corruption due to the misuse of the APBD. Following this, Lewis (2017) stated that local governments with low levels of corruption spend resources more efficiently and provide better access to public services while improving performance. Therefore, the APBD should be implemented in line with applicable laws and regulations, because it supports economic growth (Afonso & Furceri, 2010).
In Indonesia, decentralization allows the DPRD to oversee the administration of a clean, democratic, and pro-people local government. Moreover, this legislature performs a significant role in preparing and processing local government budgets. However, DPRD also abuses power by allocating the APBD for a certain period (Abdullah & Asmara, 2006). This is easier when the regional head is supported by most party coalition members of the council. According to Arends (2017), the effects of decentralization are not always positive. For instance, the health sector's performance is worse when decentralization is implemented. Artés and Jurado (2018) found that when the government has a single-party majority, the budget is better. This contradicts Nakanishi (2019), who did not prove budget effectiveness in a single-party government. Similarly, Persson, Roland, and Tabellini (2007) found that government expenditure would be greater with coalitions than with single parties. Furthermore, Jurado (2014) explained that the party system affects social government expenditure. These studies show that legislative single and multi-party systems affect the government's budget for education and health. Lewis and Hendrawan (2019) found that positive changes in health services only occurred in the two years following an election. Budget abuse becomes a problem in the final two years of most coalitions' terms in the legislature because of diversion to infrastructure expenditure. Coincidentally, Kis-Katos and Sjahrir (2017)  . Therefore, the current study investigates the effect of party coalitions supporting regional heads in legislating education and health expenditures in Indonesia's local governments. The control variables were population size, poverty level, economic growth, and local revenue. According to Cavalieri and Ferrante (2016), decentralization reduces infant mortality, but this depends on regional factors, such as wealth.
The structure of the paper is as follows. Further section discusses literature review and hypothesis development. Section 3 analyses the methodology that has been used to conduct the empirical study. Section 4 provides the result of the statistical testing. Section 5 presents a discussion of the statistical result, followed by Section 6 that provides conclusions, limitations of the study, and suggestions for future research.

LITERATURE REVIEW AND HYPOTHESIS DEVELOPMENT
This study refers to public finance theory. Musgrave and Musgrave (1989) stated that some unprofitable economic activities cannot be provided by the market or the private sector. Therefore, activities such as health and education provisions are the government's obligation.
Public finance is an economic field that studies the activities and methods used to fund government expenditure (Hyman, 2011). The government provides public goods that are not accessible in the market. According to Hyman (2011), the government is an organization that exercises authority and action for people living together in a society, to provide and finance public interest services.
One goal of the public sector is to provide community services, which means that the budget is drawn up for the people. However, this is influenced by various factors, such as using the budget for political interests. Hyman (2011) stated that, from an economic perspective, the goal of politics is to provide goods and services that benefit citizens. This shows that the political process determines the number of goods and services provided by the government. The political aspect also influences the way to manage public sector organizations (Alsaid & Mutiganda, 2018).
This study used the public choice theory developed by Buchanan and Tollison (1984). The theory is useful in collective and government social decision-making, and in explaining social and political phenomena. Accordingly, the DPRD, comprising various political institutions and regional heads, could make APBD policies, including mandatory budget allocations. A recent paper uses the idea of the public choice theory to explain public affairs such as the price of rice in Thailand (Ricks, 2018), rental housing (Owusu-Ansah, Ohemeng-Mensah, Abdulai, & Obeng-Odoom, 2018), and government expenditure (Bhandary & Nagesha, 2018).
Education expenditure is a mandatory government affair related to basic services, as stated in Law Number 23 of 2014, Article 12. According to Law Number 18 of 2016, the budget is allocated for education through the state ministry or institutions, regional and village funds, and financing. Its percentage compares education to the total state budget.
Government Law Number 20 of 2003 stipulates national education regulations. It ensures equal distribution of educational opportunities and improved quality, to face challenges based on the local, national, and international demands of the future. Therefore, planned, directed, and sustainable regulations are needed.
Law Number 36 of 2009, Article 171, stipulates that the health sector is allocated at least 5% of the state budget (APBN), excluding salaries. In comparison, local governments in provinces, cities, and regencies receive at least 10% of the APBD, excluding salaries. Furthermore, the budget is prioritized for public services, which is at least twothirds of the total health budget in APBN and APBD. This demonstrates the Indonesian government's commitment to combat health problems.
The government should focus on the health budget to improve people's lives. Law Number 36 of 2009 states that health development increases awareness, willingness, and the ability to live a healthy life, as an investment in productive human resources. Moreover, Gupta, Verhoeven, and Tiongson (2002) suggested that an increase in the health budget reduces child mortality, although this influence depends on government quality (Baldacci, Clements, Gupta, & Cui, 2008).
Political party coalitions nominate one pair of regional head candidates (Law Number 10 of 2016). To be registered, the candidates must win at least 20% of the seats in the DPRD. Alternatively, they must obtain 25% of the accumulated valid votes in the general election for DPRD members in the stipulated area.
Law Number 17 of 2014 states that the DPR, which is referred to as DPRD at the regional level, comprises political party members elected through general elections. Each region has its DPR at the provincial, city, and regency levels with legislative, budgetary, and supervisory functions.
The coalition supporting the elected regional heads in the legislative seats approves the APBD and APBD-P revenue and expenditure budgets (Sitepu, 2016; Widyaningrum, Setiawan, & Brahmana, 2019). These budgets are tools for economic policy to realize economic growth and stability (Riharjo & Isnadi, 2010).

Hypothesis development
The theory of public finance states that public goods, including unprofitable services, are provided by the state. This is in line with the government's exercise of authority and action for people living together in a society to provide and finance public interest services (Hyman, 2011). In conducting these obligations and authorities, the APBD is allocated by the regional head and the DPRD under political influence. Furthermore, public choice theory explains how society's decisions are made by the government. In this case, the DPRD, consisting of various political institutions and regional heads, makes APBD policies, including mandatory budget allocations. Some members support the regional head during elections, while others do not have a supporting coalition. However, action is required for supported and elected regional heads to achieve these goals. Therefore, the formation of the APBD is suspected to involve political factors (Widyaningrum et al., 2019). The previous studies by Setiawan and Rizkiah (2017) and Garmann (2018) provide evidence that the incumbent head of local government uses his discretion to spend more budget in the election year, to get a more positive impact on the voters.
Education services are a focal point of the government as a method of increasing people's prosperity and the nation's competitiveness. Kis-Katos and Sjahrir (2013) demonstrated that a democratically elected government affects low education expenditure. This supports Lloyd (2015), who indicated that education was ineffective under a coalition government. However, education is an obligation of the government, as stated in the 1945 Constitution. According to Arifin, Trinugroho, Prabowo, Sutaryo, and Muhtar (2015), the multiparty government system increases corruption involving regional executive and legislative politicians, because of ineffective supervision by the DPRD.
Health expenditure is also a form of public service intended to bring about prosperity. Government decentralization has affected the originally centralized health expenditure to become regional. According to Arends (2017), decentralization leads to poorer performance in the health sector. This is in line with Kyriacou and Roca-Sagalés (2019), who stated that decentralization reduces local health service quality. However, Lewis and Hendrawan (2019) revealed a positive effect of decentralization on health expenditure in the two years after elections. Further, Apriliani and Khoirunurrofik (2020) provide evidence that Special Allocation Fund in Indonesia has no significant effect on health development performance in Indonesia, but it has a positive effect on the second year.
Education or health expenditure is a long-term investment. However, according to Lewis, Nguyen, and Hendrawan (2020), the delayed results in these sectors make local governments focus on visible services, such as infrastructure or grants. Furthermore, although basic health and education services should be provided by the government, they do not significantly benefit local officials, including the legislature. Therefore, these expenditures are reduced and diverted to other sectors (Mauro, 1998). In comparison, DPRD, as one of the APBD compilers, behaved opportunistically. This support Abdullah and Asmara (2006) who stated that APBD is used by DPRD members to fulfill their interests by increasing budget misuse. The reason is the significant role of the legislature in budget approval, supervision, and policy formulation.
From this study, the proposed hypotheses are as follows: H1: Coalition parties negatively affect education expenditure in Indonesian local governments.
H2: Coalition parties negatively affect health expenditure in Indonesian local governments.

METHODOLOGY
The current study uses quantitative methods to test the hypotheses. This study examines the impact of a political party coalition, that supports regional heads in the local legislative council, on government expenditure. Purposive sampling was used to select local governments based on various criteria. First, it needed to be Indonesia's regional government from 2016 to 2018. Second, it had to have a complete regional government financial report, including data according to the variables used. Third, the local government had to hold regional head elections in 2015. The observation was carried out on local government expenditure in education and health between 2016 and 2018, with reference to the 2014 legislative election results.
The dependent variables were education and health expenditures by the local government. Data were obtained from the local government financial reports and the DJPK website (http://www.djpk .kemenkeu.go.id/) by dividing the realization of education and health expenditures by the total budget in a period. In this case, education and health expenditures were local government services to the community (Lewis & Hendrawan, 2019).
The independent variable is the political party coalition, measured by the number of council seats with regional heads, during the election. Data were obtained from the central General Election Commissions (KPU) website (www.kpu.go.id) and KPU websites for each regency and city on Java Island. The control variables included population, economic growth, poverty level, and local own revenue (PAD). Population refers to the number of inhabitants in the regional area (Lewis & Hendrawan, 2019; Winarna, Widagdo, & Setiawan, 2017), economic growth uses PDRB as a proxy (Castañeda-Angarita, 2013), poverty level refers to the percentage of the poverty level in the regional area (Lewis & Hendrawan, 2019) and local own revenue refers to local own sources revenue of the local government.
The model used is as follows: where, PRE_PEND = education expenditure, the ratio between the realization of education expenditure divided by total expenditure; PRE_KES = health expenditure, the ratio between the realization of health expenditure divided by total expenditure; CP = coalition party, number of seats held by coalition party that support regional heads; Pop = ln number of the regional population; PDRB = product domestic regional bruto; MSKN = poverty level, percentage of the poverty level in the regional; PAD = local own revenue, ratio local own revenue to total local government revenue.

ANALYSIS
This section discusses the study's descriptive statistics, the statistical test result, and the discussion of the statistical result. Table 1 provides information on the descriptive statistics of the study. The descriptive study includes information on education expenditure, health expenditure, coalition party, and control variables, such as population, PDRB, level of poverty, and local own revenue. 0842 Notes: PRE_PEND = education expenditure, the ratio between the realization of education expenditure divided by total expenditure; PRE_KES = health expenditure, the ratio between the realization of health expenditure divided by total expenditure; CP = coalition party, number of seats held by coalition party that support regional heads; Pop = ln number of the regional population; PDRB = product domestic regional bruto; MSKN = poverty level, percentage of the poverty level in the regional; PAD = local own revenue, ratio local own revenue to total local government revenue.
The population comprised of level II local governments, including Indonesia's regencies and cities, between 2016 and 2018. This gave a total of 34 provinces, 415 regencies, 93 cities, 1 administrative regency, and 5 administrative cities. The samples were obtained by the purposive sampling method based on predetermined conditions. The local government had to hold a regional head election in 2015, and not a regional expansion area between 2012 and 2014 because the new region did not conduct legislative elections in 2014. Additionally, the local government was not an administrative regency or city, and the regional head was not from an independent channel. Furthermore, the criteria required data on the websites of the Directorate General of Fiscal Balance, or the Regional Government Financial Statements obtained from the Supreme Audit Agency (BPK), Central Statistics Agency (BPS), and the General Elections Commission (KPU). Table 3 shows that the total sample used is 632 (N), representing regencies and cities in 2016, 2017, and 2018, that fulfilled the sampling criteria. The maximum percentage for education expenditure was 41%, based on data from the Directorate General of Fiscal Balance website (www.djpk.kemenkeu.go.id). This value is quite substantial compared to the required minimum of 20%, while the lowest value was 10%. Furthermore, the average education expenditure of 26% shows that most regency and city local governments have complied with these regulations. However, others have not fulfilled the minimum of 20%.
The maximum value for health expenditure in the sample of regencies and cities was 32%. This percentage was higher than the required minimum of 10%. The minimum percentage was 0.05%, which was low. However, based on the average and median value of 13%, which exceeds 10%, most regencies and cities have complied with health expenditure regulations.
The independent variable is the coalition, as measured by the number of DPRD members supporting the head of the party coalition in the legislature, compared to the total number of members. The descriptive statistics show that the highest and lowest percentages were 92% and 11%, respectively. Moreover, the average value of this variable was 37%, while the median value was 33%. Of the 632 samples, 478 data show that the coalition is below 50%, while 141 data show that it is above 50%. 0.0000 0.0000 N 632 632 Notes: *, **, *** significance at 1%, 5%, and 10% levels. CP = coalition party, number of seats held by coalition party that support regional heads; MP = main political parties that carrying the regional head; Pop = ln number of the regional population; PDRB = product domestic regional bruto; MSKN = poverty level, percentage of the poverty level in the regional; PAD = local own revenue, ratio local own revenue to total local government revenue.
The results of the regression test with the full sample show a significance value of 0.0854, with a beta coefficient of -0.0318. Therefore, the first hypothesis (H1) that the coalition negatively affects education expenditure in regencies and cities is accepted. Our study also checks the effect of main political parties that carry the regional election on the education expenditure. The result is consistent with the coalition party. Thus, the higher percentage of the seat held by the coalition party and the main political party that carried the regional election decreases the education expenditure. 0.0000 0.0000 N 652 652 Notes: *, **, *** significance at 1%, 5%, and 10% levels. CP = coalition party, number of seats held by coalition party that support regional heads; MP = main political parties that carrying the regional head; Pop = ln number of the regional population; PDRB = product domestic regional bruto; MSKN = poverty level, percentage of the poverty level in the regional; PAD = local own revenue, ratio local own revenue to total local government revenue.
These results show that an increased coalition percentage cannot guarantee high education expenditure. This is due to the high voting rights of coalition members that discourage the government's mandatory budget expenditures for education. Bursztyn (2016) stated that low education expenditure occurs because voters in developing countries choose governments that prioritize other activities, such as cash transfers. This is inseparable from voters who prefer something with quicker benefits than education. However, this contradicts Busemeyer (2008), who found a positive relationship between regional decentralization and education expenditure.
The results of the regression test of the health expenditure variable depicted a significance value of 0.0184, with a beta coefficient of -0.0154. Therefore, the second hypothesis (H2) that the coalition negatively affects health expenditure in regencies and cities that conducted regional head elections in 2015 is accepted. Furthermore, the two equations show that the larger the coalition, the smaller the expenditure on education or health. Most often, decentralization does not have a positive effect. In line with this, Arends (2017) found that health performance deteriorates when decentralization is implemented, due to weak local administrative capacity or the need to increase fiscal transparency. Similarly, Kyriacou and Roca-Sagalés (2019) found that the quality of health services decreased when decentralization was implemented. The lower standard occurred because the municipal and regional sample from the decentralization level was not considered. This study also tests the effect of the ratio of the seat held by the main political party that carries regional head to the health expenditure. The result is also consistent with the effect of coalition parties on health expenditure. Therefore, the higher percentage of the seat held by the main political parties and coalition leads to lower health expenditure.
The tendency of a majority DPRD coalition is high, meaning that the legislature has the drive and power to influence regional heads in their policies. The supporting coalition is formed because of a common goal, vision, and mission. However, to realize their goals, they need actions that influence the policies of regional heads in the ABD (Jurado, 2014), because the legislature can act opportunistically by utilizing its power when preparing the APBD. Both the legislature and the executive have absolute power to exploit budget allocations for personal and group interests (Arifin & Purnomowati, 2017).

DISCUSSION
This study investigated the effect of coalition parties that support regional heads on the local government expenditure, specifically into two mandatory expenditures: education and health expenditure. It is expected that the level of coalition parties has negatively increased the education and health expenditure. The result of the study confirms the expectations, that coalition parties have decreased the level of mandatory expenditures both education expenditure and health expenditure. The higher percentage of coalition parties holding a seat at the DPRD has pushed the local government to spend less in the education and health sectors.
The negative effect of the coalition parties on the education expenditures confirmed the previous study that shows the coalition government has ineffective to manage the education expenditures (Lloyd, 2015). Lloyd (2015) found that the coalition government has several weaknesses in the early childhood education policy, such as the relation between early education policy and another welfare policy was disconnected and too dominant the political belief in the market operations. This might be a coalition government that prefers to spend the money on other expenditures that provide fast results. The education expenditure has a strategic aspect on the long-term policy, therefore, the impact of education expenditure also has a long-term impact. The regional heads in the year around direct election prefer to spend the money on other activities, such as donations and social assistance expenditure (Setiawan & Rizkiah, 2017), because it will give a good impression to the incumbent regional heads (Setiawan & Rizkiah, 2017;Wiguna & Khoirunurrofik, 2021). The result of the study was also confirmed by Kis-Katos and Sjahrir (2013) that find the democratically elected regional head spends less on the education expenditure. The poor political institution will lead to poor public institution performance (El-Khishin & Zaky, 2021).
The current study also finds a negative relationship between coalition parties that carry the regional heads and the health expenditure. The result of the study confirms the expectation that coalition parties spend less on health expenditure. Previous studies, such as Arends (2017) and Kyriacou and Roca-Sagalés (2019), find that decentralization has a negative effect on health expenditure. The local government expects that the central government still have a significant contribution to the health sector. Therefore, local governments prefer to spend the money on other expenditures. As mentioned by a previous study (Wiguna & Khoirunurrofik, 2021) that the incumbent regional heads in the year of direct election prefer to spend the money on donation, social assistance, or financial assistance. Thus, coalition parties push the regional heads to use their discretion to build a good image. It will provide a higher probability to re-elect during the election (Setiawan & Rizkiah, 2017). Lewis et al. (2020) argue that health expenditures have long-term results, thus local governments focus on visible services, such as infrastructure or grants.

CONCLUSION
This study empirically shows the effect of coalitions supporting regional heads, on government education and health expenditure. The study's hypothesis expected that coalition parties have a negative effect on both education expenditure and health expenditure in the Indonesian local government. The statistical test result confirms the expectation. The coalition parties that supported regional heads had a negative effect on education expenditure. The higher percentage of the coalition party has reduced the level of the education expenditure in the local government. Therefore, the coalition party prefers to spend the money on other expenditures rather than education expenditure. Further, the result of the statistical test shows that coalition parties have negatively affected health expenditure. The higher percentage of seats held by the coalition parties in the DPRD have significantly affected local government decisions on health expenditure. The level of health expenditure decreases when the level of the coalition party increases. The result of the study indicates that coalition parties prefer to get the payoff from their decision to support the regional heads. This result is inline with the previous studies by Lewis and Hendrawan (2019) and Hendrawan et al. (2021).
The test results highlighted that coalitions negatively affected education and health expenditure. This implies that the high prevalence of supporting coalitions does not improve community education, health services, and expenditure. Therefore, the government may focus on their interests at the expense of education and health, driven by the fact that the two sectors are long-term investments whose results take time. Furthermore, local governments may prefer to allocate their budgets to sectors with faster rewards; however, the priority needs of local governments could be determined by the community or voters who prefer policies with direct and immediate benefits, rather than those that take years to yield positive results. The coalition party supports the regional head's decision to spend the money on activities that have a faster impact. This result is in line with the previous study that shows regional heads choose to use discretion on the spending for their best interest (Kis-Katos & Sjahrir, 2017; Setiawan & Rizkiah, 2017; Wiguna & Khoirunurrofik, 2021).
This study has several limitations, including the independent variable proxy, which is measured by the coalition supporting the regional head compared to the number of DPRD members. Furthermore, certain data on education and health expenditures have not been separated. However, Indonesia adheres to a presidential form of government with unique characteristics (Lewis, 2016). Therefore, these results cannot be generalized and may vary between countries and governments.
Further research should use education and health expenditure variables with more detailed proxies, such as the Regional Government Financial Reports (LKPD). This study implies that the coalition supporting regional heads is one factor that negatively influences education and health expenditure. This should be examined because one of the goals of the public sector is community service. Therefore, although the coalition supporting regional heads has a significant percentage, it should not neglect its primary purpose of community service.