GOVERNANCE GOVERNANCE EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT Refining Indonesia’s Intergovernmental Transfers Mechanism Fiscal Capacity Estimation to Incentivize Subnational Tax Effort Muhammad Khudadad Chattha Jürgen René Blum Assyifa Szami Ilman EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 1 © 2024 International Bank for Reconstruction and Development / The World Bank 1818 H Street NW Washington DC 20433 Telephone: 202-473-1000 Internet: www.worldbank.org This work is a product of the staff of The World Bank with external contributions. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy, completeness, or currency of the data included in this work and does not assume responsibility for any errors, omissions, or discrepancies in the information, or liability with respect to the use of or failure to use the information, methods, processes, or conclusions set forth. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Nothing herein shall constitute or be construed or considered to be a limitation upon or waiver of the privileges and immunities of The World Bank, all of which are specifically reserved. Rights and Permissions The material in this work is subject to copyright. Because The World Bank encourages dissemination of its knowledge, this work may be reproduced, in whole or in part, for noncommercial purposes as long as full attribution to this work is given. Any queries on rights and licenses, including subsidiary rights, should be addressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2625; e-mail: pubrights@worldbank.org. Attribution Please cite the work as follows: “Chattha, Muhammad Khudadad, Jürgen René Blum, and Assyifa Szami Ilman. 2024. “Refining Indonesia’s Intergovernmental Transfers Mechanism” © World Bank. Contents Acknowledgments iv Abbreviations v Executive Summary 1 1. Subnational Government’s Reliance on Transfers to Finance Spending 4 2. Case for Reforming General Allocation Fund (DAU) Formula to Incentivize Subnational Government Tax Effort 8 3. Limited Empirical Evidence on Estimating Potential Revenues 10 4. Conceptual Framework for Estimating Potential Revenues in Indonesia 12 5. Adding Relevant Controls and District Fixed Effects Reduces Distributional Implications 17 6. Conclusion 24 References 26 Annex A: Regression Results of Conceptual Framework 28 Annex B: Regression Results of Models with Added Controls 32 Figures Figure 1: Comprehensive Framework for Potential Revenue Estimation 1 Figure 2: Districts and Provinces’ Revenue Compositions, 2001–2020 5 Figure 3: The DAU Formula Under Law 33/2004 7 Figure 4: Comprehensive Framework for Potential Revenue Estimation 12 Figure 5: Changes in DAU from Model 1 to Model 2, by Percent Changes and Number of Gaining/Losing Districts 17 Figure 6: Changes in DAU from Model 1 to Model 2, by Geographical Distribution of Gaining/Losing Districts 18 Figure 7: Distributional Implications of Running Models with and without Fixed Effects 20 Tables Table 1: Evolution of Fiscal Capacity Formula 8 Table 2: Proposed Conceptual Predictors for Each District Groups 13 Table 3: Aggregated Potential Revenue Model 14 Table 4: Empirical Results from Including GRDP in the Model 15 Table 5: Top 10 Winning and Losing Districts After Modifications in Models 1 and 2 19 Acknowledgments This paper was prepared by the Governance Global Practice of the World Bank, led by Muhammad Khudadad Chattha with core authors comprising Jürgen René Blum and Assyifa Szami Ilman. It was written with the guidance of Alma Kanani (Practice Manager, Governance Global Practice, World Bank) and Habib Rab (Lead Economist, World Bank). The paper has been enriched through the peer review comments from Francisco Javier Arias Vazquez (Senior Economist, World Bank), Ruth Nikijuluw (Economist, World Bank), Ahmad Zaki Fahmi (Economist, World Bank), Khoirunurrofik (Lecturer, University of Indonesia), Adriyanto (Director, Ministry of Finance, Indonesia), Aditya Nur Yuslam (Coordinator, Ministry of Finance, Indonesia), Dian Putra (Coordinator, Ministry of Finance, Indonesia), Heri Sudarmanto (Coordinator, Ministry of Finance, Indonesia), and other colleagues within Indonesia’s Ministry of Finance. Rosalia Marcha Violeta (Consultant, World Bank) provided excellent data support. Gisella Elvir Lokopessy (Program Assistant, World Bank) provided administrative support. Jana Mirjam Silberring (Consultant, EEAM2) helped with proof reading. The work was financed by the governments of Canada and Switzerland and the European Union, through the World Bank’s Public Financial Management Multi-Donor Trust Fund for Indonesia. EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< iv Abbreviations Bea Balik Nama Kendaraan Bermotor BBNKB (Transfer of Motor Vehicle Title Fee) Bea Pengalihan Hak atas Tanah dan Bangunan BPHTB (Duty on the Acquisition of Land and Building Rights) CIT Corporate Income Tax DAK Dana Alokasi Khusus (Special Allocation Fund) DAU Dana Alokasi Umum (General Allocation Fund) DBH Dana Bagi Hasil (Revenue Sharing Fund) Dana Bagi Hasil Penerimaan Sumber Daya Alam DBHSDA (Revenue Sharing Fund from Natural Resources) DID Dana Insentif Daerah (Regional Incentive Fund) GRDP Gross Regional Domestic Product IMB Izin Mendirikan Bangunan (Building Permit) OSR Own Source Revenue PAD Penerimaan Hasil Daerah (Own Source Revenue) PBB Pajak Bumi dan Bangunan (Land and Property Tax) PBJT Pajak Barang dan Jasa Tertentu (Certain Goods and Services Tax) PIT Personal Income Tax RRS Representative Revenue System RTS Representative Tax System SNG Subnational Government VAT Value Added Tax EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< v Executive Summary Dana Alokasi Umum (DAU), or general allocation fund, is revenue in Indonesia. This paper fills the gap in the literature Indonesia’s most significant central government transfer to by developing a conceptual framework for estimating potential subnational governments (SNGs). It is a block grant that aims revenues for districts in Indonesia and applying it using to resolve horizontal imbalances between regions. Most SNGs existing data. rely heavily on this transfer and they have a low share of own source revenues (OSRs). This can partly be attributed to the DAU formula provided by Law 33/2004 that disincentivized A. Establishing a Conceptual SNGs to collect OSRs in favor of higher DAU transfers. Framework for Potential The Indonesian government has recently stipulated a new law Revenue Estimation on intergovernmental fiscal relations (Law No. 1/2022). The law revised the DAU formulation for each SNG by substituting the sum of Penerimaan Hasil Daerah (PAD or own source There is limited knowledge on how potential government revenue) and Dana Bagi Hasil (DBH or revenue sharing revenue should be estimated, especially in the case of funds) with “potential revenue.” The exact estimation method Indonesia’s SNGs. This paper builds a comprehensive is left to the implementing regulations. framework to help estimate potential revenues in Indonesia. The framework is illustrated in Figure 1 which shows the three Existing literature provides little guidance and consensus on criteria that we use to identify predictors of revenue bases. an ideal estimation framework for local government potential > > > F I G U R E 1 - Comprehensive Framework for Potential Revenue Estimation 1. Identifying conceptual predictors using existing theory Conceptual predictors are indicators that would conceptually and/or theoretically predict SNGs’ 1. Conceptual 2. Empirical revenue stream (e.g., taxes). Predictors Evidence 2. Empirical evidence Indicators should be tested statistically or proven by existing literature before they get selected to estimate revenue base. 3. Applicability 3. Applicability in Indonesia in Indonesia The tested and selected indicators should be applicable in Indonesia. Thus, these indicators need to be supported by available data in Indonesia. EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 1 B. Building and Optimizing both approaches help improve the equity implications as well as the model’s fit. Potential Revenue Estimation Model However, adding district fixed effects is a much better, fairer, and systematic way to account for district characteristics. While mining GRDP aims to capture districts’ distinct economic This paper discusses the application of the conceptual landscape, applying district fixed effects would control for framework for potential revenue estimation using Indonesia’s district characteristics that mining GRDP could not fully existing macroeconomic indicators. We find that Gross capture. We find that including the district fixed effect leads to Regional Domestic Product (GRDP), population, and lower distributional implications. urbanization rate are good predictors for most tax sources. These three indicators also predicted the total district’s OSRs In sum, the paper provides a viable, theoretical, and an well, providing an empirical foundation for an aggregated empirical approach to estimate potential revenues for macro-based model to estimate all OSRs using the the DAU formula in Indonesia – a macroeconomic model abovementioned variables. using three main macroeconomic indicators: GRDP, population, and urbanization rate, along with district fixed effects. C. Distributional Implications of The rest of the paper is structured as follows: Section 1 provides Potential Revenue Estimation a description of Indonesia’s system of intergovernmental transfers and subnational taxation. It shows subnational fiscal reliance on transfers rather than OSRs. Section 2 makes This paper also analyzes the distributional implications of the case for reforming the DAU formula and explains recent moving to estimated potential revenue in the DAU formula. efforts by the Government of Indonesia on that front. Section 3 We find that the equity implications are limited with most discusses the limited existing empirical literature on estimating districts gaining, or losing, less than 20% of DAU. However, potential revenues for transfers formula. Section 4 explains there can be outliers in the estimation. and applies our conceptual framework for potential revenue estimation. It also provides the empirical justification to use We explored two approaches to increase the fairness of DAU an aggregated approach to estimation rather than estimating allocations. First, we add mining GRDP as a control variable each individual tax base. Section 5 applies the aggregated to incorporate SNGs’ economic landscape, because the approach to estimating potential revenues. Section 6 mining sector is generally not subject to property tax at the discusses the equity implications and makes the case for using subnational level and instead is taxed at the central level. district fixed effects. Finally, Section 7 provides a conclusion of Second, we control for district fixed effects to account for time- this paper. invariant district characteristics. Our simulations show that EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 2 1.Subnational Government’s Reliance on Transfers to Finance Spending Indonesia decentralized significant spending authority from the central government to subnational governments (SNGs) during the early 2000s.1 SNGs are now responsible for delivering essential public services, such as education, health, and infrastructure. Several studies have shown that Indonesia’s decentralization was among the largest and most ambitious programs ever to take place (Negara and Hutchinson 2021; Fritzen 2009). With significant shifts in spending authorities, SNGs are also responsible for collecting revenues to support their operations. Decentralization enables SNGs to collect several types of tax and retributions independently, as stated under Law 28/2009 on Regional Tax and Retributions before eventually being replaced most recently by Law 1/2022. Under this structure, SNGs can generate revenues from the following resources: • Own source revenues (OSRs) or Penerimaan Asli Daerah (PAD)2 are revenues that are collected by SNGs through regional tax and retributions. Province- and district-level governments have a different set of taxes that they can collect. Meanwhile, retributions are fees applied to each SNG’s public facilities and services. PAD include the following taxes and retributions: • District-level taxes: Land and property tax, duty on the acquisition of land and building rights (BPHTB), certain goods and services tax (PBJT),3 advertising tax, groundwater tax, swallow’s nest tax. 1. The central government decentralized most authorities to SNGs, except for constitutional, foreign, defense, religious, and monetary affairs. 2. In this paper, we will use OSR and PAD interchangeably. 3. This includes but not limited to food and beverage/restaurant tax, electric power/street lighting tax, hotel services tax, parking services tax, and art and entertainment tax (karaoke, night clubs, traditional, sports events, etc.). EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 3 • Province-level taxes: Vehicle tax, transfer of motor transfers (49 percent of total transfers for district-level vehicle title fee (BBNKB), heavy equipment tax, and 29 percent for province-level government in 2020). cigarette tax, vehicle fuel tax, surface water tax, non- This transfer aims to equalize SNGs’ capacity to fund their metal minerals and rock tax. operational needs. • Regional retributions: General services,4 business • Special allocation fund (Dana Alokasi Khusus services,5 and certain licensing.6 or DAK) is another grant that aims to fund specific responsibilities that align with national priorities. This • Intergovernmental transfers are revenues transferred special purpose grant comprises physical DAK (DAK from the central government to SNGs. Fisik) to finance capital expenditures, and non-physical ones (DAK Non-Fisik) to finance public service delivery • Other eligible revenues are revenues generated through operational costs. DAK comprised about 15 percent of other channels of revenue, such as revenue sharing from total transfers in 2020 for district-level and 37 percent for provinces and village funds. province-level government. Despite having various taxes and retribution, SNGs in • SNGs also receive revenue sharing funds (Dana Bagi Indonesia remain largely dependent on intergovernmental Hasil or DBH), a central government redistribution of tax transfers to finance their spending. Intergovernmental revenues to SNGs.8 DBH comprised about 9 percent of transfers comprised around 60 percent of total district revenues total transfers in 2020 for district-level and 20 percent for and 55 percent of total province revenues in 2020 (Figure 2). province-level government. SNGs’ reliance on transfers has decreased substantially since 2001,7 when the decentralization policy was first implemented • Finally, there are other transfers that constitute 27 percent under Law 22/1999 on Regional Governments and Law of the total transfers pool for district-level and 14 percent 25/1999 on Intergovernmental Fiscal Transfers. These laws for province-level governments. These include autonomy were replaced by Law 32/2004 and 33/2004, and recently by funds for specially treated regions, regional incentive Law 1/2022. The different types of intergovernmental transfers grants (Dana Insentif Daerah or DID) for SNGs with better in Indonesia are briefly described below: governmental performance, village funds for funding development projects in villages, and de-concentration • The general allocation fund (Dana Alokasi Umum or funds for funding national programs under specific DAU) is a non-earmarked, general-purpose block grant line ministries. that constitutes the largest portion of central government 4. Retribution fees applicable for general services, such as for medical, cleaning, civil administration, funeral, parallel parking, traditional market, vehicle emission testing, fire hazard maintenance, official map printing, lavatory waste management (septic tank), weight remeasurement service for traditional market transactions (tera ulang), educational retribution fees, telecommunication tower maintenance services, and traffic light and pedestrian maintenance. 5. Retribution fees applicable for commercial purposes. An example is the utilization of regional assets (land and property) for commercial affairs: wholesale market, auction house, bus terminal, government-managed parking buildings, government-owned hotel and dormitories, slaughtering house, government-owned port, recreation and sports venues, water passage facilities, and government retailer products. 6. Retribution fees applicable for specific permits, including construction permit (IMB), alcoholic drink sales permit, crowd permit, transportation route permit, aquaculture and fisheries permit, and expatriate working permit. 7. Transfers to districts declined from 90 percent of district revenue in 2001 to 62 percent of district revenue in 2020. Transfers to provinces declined from around 61 percent of provincial revenue in 2001 to 55 percent of provincial revenue in 2020. 8. Central government taxes eligible for DBH include property tax on mining, forestry and plantation, personal income tax, value-added tax, as well as other income from tobacco excises and natural resource revenues. EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 4 > > > F I G U R E 2 - Districts and Provinces’ Revenue Compositions, 2001–2020 DISTRICT REVENUE COMPOSITION, 2001-2020 900 100% Transfers as a % of total district revenue 800 90% 700 80% 70% 600 60% Rp trillion 500 50% 400 40% 300 30% 200 20% 100 10% 0 0% 01 1 02 003 004 005 006 007 008 009 010 01 012 013 014 015 016 017 018 019 020 20 20 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 PAD DAU DAK DBH Other Transfers PROVINCE REVENUE COMPOSITION, 2001-2020 450 100% Transfers as a % of total province revenue 400 90% 350 80% 70% 300 60% Rp trillion 250 50% 200 40% 150 30% 100 20% 50 10% 0 0% 1 7 1 7 00 00 002 003 004 005 006 00 008 009 010 01 012 013 014 015 016 01 018 019 020 20 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 PAD DAU DAK DBH Other Transfers Source: Ministry of Finance; World Bank staff calculations. Note: DAK = Dana Alokasi Khusus (Special Allocation Fund), DAU = Dana Alokasi Umum (General Allocation Fund), DBH = Dana Bagi Hasil (Revenue Sharing Fund), PAD = Pendapatan Asli Daerah (own source revenue), Rp = Indonesian rupiah. EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 5 DAU’s significant contribution to district revenues is partly due to the regulatory framework, which mandates the central a result of the regulatory structure of intergovernmental government to allocate at least 26 percent of net domestic transfers. District governments primarily rely on the DAU to revenue for DAU9 and eventually distributes 90 percent of all finance their spending, which contributes about 41 percent of DAU to district-level governments. total district revenues. This heavy reliance on the DAU is partly 9. This is calculated by subtracting revenue sharing funds (DBH) from the national tax revenue. EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 6 2. Case for Reforming the General Allocation Fund (DAU) Formula to Incentivize Subnational Government Tax Efforts The DAU formula under Law 33/2004 became a disincentive for districts to collect revenue bases assigned to them. The DAU formula equalizes the difference between fiscal needs and fiscal capacity. Prior to Law 1/2022, fiscal capacity was proxied using actual district’s own source revenues (PAD) and revenue sharing funds. Under this formulation, district DAU allocation decreases when districts collect more tax and non-tax revenues which is a clear disincentive for district revenue mobilization (Figure 3 shows the formulas wherein an increase in PAD leads to a reduction in DAU). > > > F I G U R E 3 - The DAU Formula Under Law 33/2004 DAUi = ( Fiscal Gapi Total Fiscal Gap ) * Total DAU Pool Fiscal Gapi = Fiscal Needsi - Fiscal Capacityi Fiscal Capacity = δ1 PADi + δ2 DBHi, 0 <δ ≤ 1 Source: Government of Indonesia, Law 33/2004. EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 7 The DAU formula aims to resolve horizontal imbalances capabilities. Martinez-Vasquez and Boex (1997a) argued that among local governments by covering SNG deficits. This using the amount of revenues collected as a measure of fiscal is reflected by how the DAU formula considers the fiscal gap, capacity is associated with several problems. For example, which measures the difference between fiscal needs (i.e., two regions with similar “fiscal capacity” may eventually necessary expenses) and fiscal capacity (i.e., SNG’s revenue, generate different amount of revenues due to differences in such as own source revenue and revenue sharing funds) compliance or tax rate. Furthermore, the study also argued (Takahata, Khoirunurrofik, and Dartanto 2021). DAU’s ability that using actual data may be harmful if the quality of such to cover the SNG deficit remains limited, given that DAU is data is questionable. Thus, using actual collected revenues, predetermined in the national budget (APBN). Given the such as PAD and DBH in the case of Indonesia, can be strong focus on equalization, the DAU formula treated PAD misleading when determining a district’s true fiscal capacity. as a deduction factor to accommodate more funding for SNGs As an alternative, Martinez-Vasquez and Boex (1997b) and with less capability to generate their own revenue. Shah and others (1994) proposed some proxies to measure fiscal capacity such as macro indicators (i.e., personal income/ Existing literature provides limited evidence to support disposable income, regional GDP) and the representative tax the use of PAD as a fiscal capacity proxy. This is because system (RTS).10 estimating fiscal capacity has been conceptually and empirically challenging. Yilmaz and Zahir (2020) explained In the case of Indonesia, changes to fiscal capacity that some countries concentrate only on assessing revenue- measurement have evolved numerous times and can be raising ability, while others focus more on ensuring that districts seen (Table 1). However, PAD has always been a primary have the capacity to provide similar public service packages to indicator for the fiscal capacity formula. their citizens but are paying less attention to revenue collection > > > T A B L E 1 - Evolution of Fiscal Capacity Formula 2001 2002-2003 2004-2005 2006-Now Avg (PADi + PBBi PADi + DBH^i 0.5 * PADi + DBH^i PADi + DBHi + BPHTBi ) + 0.75 * DBHSDAi + DBHSDAi * Avg (GRDPsdaIndex + GRDPnonsdaIndex + Working Age Index) Source: Hofman et al. (2006). Note: PAD = own source revenue (Penerimaan Asli Daerah), PBB = property tax, BPHTB = duty on the acquisition of land and building rights, GRDPsda Index = Index of SNGs with dominant revenue from natural resources, DBHSDA = revenue sharing fund from natural resources, DBH^ = revenue sharing funds from tax collections, DBH = revenue sharing funds. The Indonesian government has recently reformed the revenue estimates instead of the historical sum of own source DAU formula to incentivize SNG revenue effort through revenue (PAD) and revenue sharing funds (DBH). However, Law 1/2022.11 This law made several important changes the law leaves it up to the implementing regulations to define to the architecture of intergovernmental transfer and SNG the method for estimating the potential revenue, which can be revenues. The new law has changed the DAU formula for both a challenge and an opportunity. measuring fiscal capacity, that is, determining the potential 10. The Representative Tax System (RTS) is an approach where regional governments measure their fiscal capacity by counting the revenue that they could raise if they employed all standard sources at the nationwide average intensity of use (Shah et al. 1994). Several pieces of information are needed, such as information on the tax bases and tax rates for each SNG. 11. Law 1/2022 replaces Law 33/2004 on intergovernmental transfers and Law 28/2009 on subnational revenues. EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 8 3. Limited Empirical Evidence on Estimating Potential Revenues The previous DAU formula theoretically disincentivizes SNGs to collect more OSRs, but empirical evidence remains mixed. With the existing DAU formula, SNGs have the choice to make less effort in collecting OSR to gain transfers in the form of DAU. However, available evidence shows mixed results. Lewis and Smoke (2017) found that due to limited OSR contribution to Indonesia’s SNGs’ revenue streams, SNGs paid little attention to the negative incentives in the current DAU formula. This contradicts Fadliya and McLeod (2010), who argued that the formula would disincentivize SNGs in generating their revenues. Results were also mixed in other developing economies (Mogues and Benin 2012; Huang, Lo, and She 2012). Nevertheless, there are strong reasons to review the existing DAU formula and incentivize SNGs to collect more OSRs. Theoretically, Martinez-Vasquez (1997b) shared a concern discussed in the previous section regarding how OSRs could provide misleading representation of SNGs’ fiscal capacity. A local government’s leaders may also find it politically challenging to increase their region’s tax effort. Raising local tax may be seen as an unpopular policy that brings some political costs to local leaders. Hence a stronger incentive for OSR collection by the central government could be useful. Existing literature provides little to no consensus on how potential revenue should be estimated, particularly those incorporating the characteristics of Indonesia’s SNGs. Mawejje and Sebudde (2019) compare various approaches to revenue forecasting methods (Rubin, Mantell, and Pagano 1999; Williams and Kavanagh 2016) and the challenges of utilizing advanced revenue forecasting methods in local governments (Reddick 2004; Batóg and Batóg 2021). However, none of these bodies of literature has shed light on which approach would suit best for Indonesia. EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 9 Existing literature identifies several ways of estimating However, some other countries that are willing to go potential revenues. Differences between various approaches through more complex and rigorous estimation methods are usually related closely to mathematical complexity, data adopt macroeconomic modeling, representative revenue requirements, overall sophistication, and accuracy (Reddick systems (RRS), and microeconomic approaches. 2004). First, qualitative approaches rely on subjective opinion Estimation using macroeconomic approaches rely on of experts to forecast revenues. The approach is often used to aggregated macroeconomic data for variables that may enhance the results produced using other approaches. Second, provide information behind changes in revenues. RRS extrapolative or trend-based approaches rely on time series approaches estimate potential revenues using microeconomic analysis of historical revenue data. These approaches provide indicators by assessing both tax base and tax rate. This a more rigorous process for potential revenue estimation approach utilizes disaggregated economic census data to and incorporate simple and accessible data. Finally, there measure the tax base based on each individual SNG tax. The are causal approaches, where deterministic or econometric estimated tax base is then multiplied by the tax rate. However, techniques use variables that may play an important role in it is essential to interpret the result with caution, given that this influencing revenues in the future. While these approaches approach does not consider the behavioral response of tax yield useful information for policymakers, they require more implementation (i.e., once the tax is implemented, that would time and resources compared with other approaches. lead to a reduction in the tax base as taxpayers would respond to the change). United States,13 Canada,14 Italy, and Poland In practice, some countries use actual revenue data use this approach when estimating potential revenues. The as a proxy for potential revenue. Indonesia,12 Greece, microeconomic approach usually utilizes individual tax base Portugal, and the United Kingdom are some countries who (administrative data) and gathers new data from surveys. The use this approach. Taking this approach is practical and sophistication of such models depends on data availability. provides transparency, which are two desirable features of fiscal capacity measurements given that many policymakers are unfamiliar with complex and technically demanding measurements (Martinez-Vazquez and Boex 1997a). 12. This is the practice conducted by the Indonesian government prior to the ratification of Intergovernmental Fiscal Relations Law (Law 1/2022). 13. The study found that the RRS approach poorly predicts fiscal capacity in US states, as reflected by weak correlation between the fiscal gap (revenue capacity – expenditure need) and distributed grants (Gordon, Auxier, and Iselin 2016). 14. Taylor, Keenan, and Carbonneau (2002). EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 10 4. Conceptual Framework for Estimating Potential Revenues in Indonesia We build on the existing literature by developing a comprehensive conceptual framework to estimate potential revenues for Indonesia, taking into account the data constraints in the country. This framework aims to identify predictor(s) of potential revenue of the main tax revenue sources at the district level in Indonesia. We use three criteria to identify the predictors, given below: 1. Conceptual/theoretical link with revenue source: These are variables that we expect in theory to predict the revenue stream. 2. Backed by empirical evidence where available: Empirical evidence is needed to provide understanding of whether selected indicators in practice predict their respective revenue streams in a robust manner. We search for available empirical literature for indicators identified via the first criteria. 3. The indicator accurately predicts respective revenue sources using historic data from Indonesia: The selected indicators should be applicable to Indonesia. This means that (1) the data for the relevant indicator should be available for Indonesia, and (2) the indicator should be able to predict respective revenue sources using Indonesia’s data. EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 11 > > > F I G U R E 4 - Comprehensive Framework for Potential Revenue Estimation 1. Conceptual 2. Empirical Predictors Evidence 3. Applicability in Indonesia We apply this framework to four categories of district-level were tested separately using Indonesia’s subnational fiscal taxes in Indonesia, which constitute about 80 percent of data from 2010 to 2018 for districts with available data. We also the district’s OSRs. These groups are (1) recurrent property reviewed the indicators’ ability to predict own source revenues tax; (2) property transfer tax; (3) street lighting tax; and (4) (see Annex A and B for a detailed regression results). We only hotel, restaurant, and entertainment taxes. Table 2 provides a applied the conceptual framework in this study to districts in summary of results. For each group, the conceptual predictors Indonesia and not to provinces.15 15. We exclude provinces from this model because of their different nature and sample size compared to districts. EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 12 > > > T A B L E 2 - Proposed Conceptual Predictors for Each District Groups Group of revenue sources Proposed conceptual predictors Results summary at district level 1. Recurrent property Gross regional domestic product (GRDP) Recurrent property tax tax and property Higher economic activity might mean higher • GRDP and total district population transfer tax number of properties, higher valuations, and have a strong association with property higher number of property transfers. tax revenue. • Using the fixed effect model, all proposed Total district population conceptual predictors have strong Higher number of people residing in a district association with property tax revenue. might proxy for more properties and hence higher property and property transfer taxes. Property transfer tax • GRDP and urbanization rate have Urbanization rate a strong association with property The urban parts of the country are more tax revenue. likely to have higher property valuations and • Population has a predictive value once more transfer of properties. we control for urbanization. See Table A.1, Table A.2, and Table A.3 for the regression results. 2. Street lighting tax Gross regional domestic product (GRDP) • Both predictors have strong association for electricity with street lighting tax in Indonesia. Higher economic activity might mean higher number of activities that would need See Table A.4 for the regression result. electricity power. Electricity access Electricity access is distributed widely via streetlight cables that connect power sources and buildings across the district. 3. Hotel, restaurant, and Gross regional domestic product (GRDP) • GRDP, accommodation GRDP and, entertainment taxes or GRDP for the accommodation sector percentage of urban population have a Higher economic activity might mean higher strong association with these taxes. consumption level. Measures of overall • Other direct measures of consumption consumption might be associated with might predict these taxes even better. higher consumption of hotels, restaurants, and entertainment. See Table A.5 for the regression result. Urbanization rate The urban parts of the country are more likely to have more hotels and restaurants. EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 13 Based on the results outlined in Table 2, we found that The results from the conceptual framework provide GRDP, population, and urbanization rate are good a justification for an aggregated, macro approach to predictors for most revenue sources. As a standalone estimating potential revenues. This is because the results predictor, GRDP is a relatively strong predictor of district showed that GRDP, population, and urbanization (in order revenue sources. Furthermore, sectoral GRDP and urban of importance) are good predictors of most district revenue population help further improve predictability and could sources when analyzed separately. This provides a strong be added to the model. Population and urbanization rates reason to use these variables to predict total district revenues. are essential variables that could also capture a district’s demographic nature. While other similar variables such as Table 3 shows that the aggregated model with GRDP, working age population and labor force could define a district’s urbanization, and population predicts PAD well. Model (1) productivity, they were not used in this study because of data shows that GRDP on its own is a good predictor of PAD as constraints.16 Considering how these indicators are easily shown by the high R-square and model fit. Models (2) and accessible, an establishment of simple potential revenue (3) show that adding additional covariates including year fixed estimation model is viable in the case of Indonesia. effects, population, and urbanization successively improve the model fit as well. > > > T A B L E 3 - Aggregated Potential Revenue Model We also show that nominal GRDP should be used in the model rather than real GRDP because price changes are reflected in PAD. Table 4 shows that the model fit decreases substantially if we use real GRDP instead of nominal GRDP. The R-square decreases when we compare Model (1) and (2), and again when we compared Model (3) and (4). We add province fixed effects to the model as added controls. 16. Furthermore, incorporating variables such as the labor force or working age population may not significantly enhance the accuracy of potential revenue estimation, especially in a context characterized by a high level of informality in Indonesia (Rothenberg et al. 2016). These variables might not offer optimal predictive value for our model since they provide only a partial representation of the economic landscape. EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 14 > > > T A B L E 4 - Empirical Results from Including GRDP in the Model EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 15 5. Adding Relevant Controls and District Fixed Effects Reduces Distributional Implications Improving the Potential Revenue Estimation Model by Introducing Selected Control Variables While an aggregated model does have good predictive power, it is also important to look at the distributional implications for districts of adding controls. This section aims to understand how adding more variables to the model would change the DAU allocation of districts. Specifically, we aim to answer the following questions: 1. What are the distributional implications of moving from actual PAD to potential revenues in the DAU formula by using a model that only uses GRDP? 2. How does DAU’s overall distribution change when we improve the model by adding population, urbanization, and sectoral GRDP17 to the model? 3. Would including district fixed effects lead to fairer DAU allocation as we consider time invariant district variant characteristics? 17. We add mining GRDP because it is not subject to district PAD and instead is subject to a central government property tax on mining, forestry, and plantations. EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 16 Moving from actual to potential revenue estimates using B.1 for the regression result). Furthermore, the geographical a GRDP only model would lead to change of less than distribution of DAU is varied with high tax effort districts, such 30% in DAU for most districts (Figure 5). Our simulations as those on the island of Java-Bali, gaining DAU (Figure 6, show a lower change in DAU with more districts gaining DAU Panel A). Districts that would lose DAU are on the islands of when we add controls in addition to GRDP. As we improve Sumatra, Kalimantan, and Papua. Shifting to the improved the model by adding population, urbanization rate, and mining model helps districts in Central and North Kalimantan, as well GRDP to predict potential revenue, we find that the number as those in the southern part of Sumatra, which gain more of districts that gained more DAU have increased (see Table DAU (Figure 6, Panel B). > > > F I G U R E 5 - Changes in DAU from Model 1 to Model 2, by Percent Changes and Number of Gaining/ Losing Districts PANEL A (MODEL 1) ( O N LY G R D P ) 100 Frequency 50 0 -100 -50 0 50 100 Change in DAU (%) Loss Win PANEL B (MODEL 2) ( G R D P, P O P U L A T I O N , U R B A N I Z A T I O N R A T E , A N D M I N I N G G R D P ) 150 100 Frequency 50 0 -100 -50 0 50 100 Change in DAU (%) Loss Win Note: GRDP = Gross Regional Domestic Product, DAU = Dana Alokasi Umum (General Allocation Fund). EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 17 > > > F I G U R E 6 - Changes in DAU from Model 1 to Model 2, by Geographical Distribution of Gaining/Losing Districts PANEL A (MODEL 1) ( O N LY G R D P ) Kalimantan Sumatra Districts gaining DAU Districts losing DAU Incomplete data PANEL B (MODEL 2) ( G R D P, P O P U L A T I O N , U R B A N I Z A T I O N R A T E , A N D M I N I N G G R D P ) Kalimantan Sumatra Districts gaining DAU Districts losing DAU Incomplete data EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 18 Including added controls also significantly reduces the would help increase the DAU allocation to this region by 124 DAU change for outliers. Table 5 illustrates how adding percent relative to the status quo. A major reason for this is controls (Model 2) reduces the DAU change for the top 10 that there are GRDP components that are not subject to SNG losing and gaining districts. For example, district 3 of the top 10 taxation; for example, the mining portion of GRDP comes from winning districts would likely experience a loss of 70 percent of the central government’s property taxation. This means that their DAU in the same observed year if the potential revenue resource rich districts, such as the ones in Kalimantan, benefit estimation model relies only on GRDP, as reflected in Model when we control for factors such as mining GRDP. Hence, 1. However, applying Model 2 would benefit this district, as Model 2 has lower distributional implications. adding more control in the potential revenue estimation model > > > T A B L E 5 - Top 10 Winning and Losing Districts After Modifications in Models 1 and 2 Top 10 winning districts Top 10 losing districts % DAU change % DAU change Difference % DAU change % DAU change Difference District District (Model 1) (Model 2) (p.p.) (Model 1) (Model 2) (p.p.) 1 -209 186 394 1 1275 107 -1168 2 -132 131 264 2 186 75 -110 3 -70 124 195 3 85 28 -57 4 -71 105 176 4 218 178 -40 5 -95 64 159 5 63 24 -39 6 29 126 97 6 62 36 -27 7 -29 67 97 7 89 63 -26 8 -40 51 92 8 68 42 -26 9 -35 50 86 9 37 21 -16 10 -38 47 85 10 23 10 -13 Notes: “% DAU Change” is defined as the difference between actual and estimated PAD, as a share of DAU. This means that the % DAU change can be lower than -100% of DAU due to a substantial increase of estimated PAD. Figures below -100% are only there for illustrative purposes. DAU = General Allocation Fund, PAD = Penerimaan Hasil Daerah (Own Source Revenue). EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 19 Incorporating added controls to the model of potential district variant characteristics into account. By using a revenue estimation has therefore lower distributional fixed effect approach, the potential revenue estimation model implications. Adding control variables from only GRDP (Model will not be influenced by each district’s fixed characteristics 1) to GRDP, population, urbanization, and mining GRDP (Stock and Watson 2020). For example, if a district has a (Model 2) helps capture a district’s important characteristics consistent composition of GRDP (e.g., it earns a consistently relevant to tax potential. high share of GRDP from oil/gas) or a consistently high/low level of tax administration capacity, then the fixed effects model will account for that. Achieving Fairer DAU Allocations by Including district fixed effect leads to lower distributional Including District Fixed Effects in implications. As illustrated in Figure 7, we found that the the Model model with district fixed effects has a relatively lower change in PAD. This is because without district fixed effects, we are comparing across districts that may be very different from Adding district fixed effects takes the fairness of DAU each other. By including district fixed effects, we are using the allocations one step further as it takes time-invariant, variation within districts which leads to a fairer DAU allocation. > > > F I G U R E 7 - Distributional Implications of Running Models with and without Fixed Effects PANEL A .02 .015 Density .01 .005 0 -100 0 100 200 Change in PAD (%) Without district fixed effects With district fixed effects EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 20 PANEL B 400 % Change in PAD without district fixed effects 350 300 Y = X line 250 200 150 100 50 0 -50 -100 -100 -50 0 50 100 150 200 250 300 350 400 % Change in PAD with district fixed effects PA N E L C - G E O G R A P H I C A L D I S T R I B U T I O N O F DAU F O R D I S T R I C T F I X E D E F F E C T S M O D E L ( G R D P, Y E A R A N D D I S T R I C T F I X E D E F F E C T S ) Districts gaining DAU Districts losing DAU Incomplete data Note: DAU = Dana Alokasi Umum (General Allocation Fund), GRDP = Gross Regional Domestic Product, PAD = Penerimaan Hasil Daerah (Own Source Revenue). EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 21 Overall, the inclusion of district fixed effects is a better – in comparison with models that do not use district fixed and simpler way to improve the accuracy of the potential effects. This approach also does not require additional revenue estimation model. District fixed effects help specific variables, such as mining GRDP, to capture all the minimize differences between the estimated and actual PAD characteristics of districts in Indonesia. figures by generating lower mean and standard deviation EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 22 6. Conclusion Indonesia has taken the right step to reform the existing DAU formula by moving from actual SNG revenues to potential revenues as a proxy for fiscal capacity. Indonesia’s SNGs are highly reliant on central government transfers, particularly from DAU. As DAU aims to resolve horizontal imbalances among local governments by covering deficits in certain SNGs, its distribution formula provides a disincentive for districts to collect their revenues independently. The new Law 1/2022 changes this situation by replacing PAD and DBH with potential revenue as a proxy for fiscal capacity. The law, however, does not elaborate on the estimation method. This paper outlines a viable approach to estimating SNGs’ potential revenue in Indonesia. Given the limited understanding of how local governments should measure their potential revenues, this paper has developed a comprehensive conceptual framework to identify ideal indicators that could predict potential revenues by considering three criteria: 1. Linkage with conceptual/theoretical findings: Indicators, in theory, should be able to predict SNGs’ revenue stream. 2. Availability of empirical evidence: Indicators in practice should be able to predict SNGs’ revenue stream in a robust manner, as suggested by existing bodies of literature. 3. Applicability in Indonesia’s case: Data for the indicators should be available in Indonesia, which should be able to predict respective revenue sources. EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 23 The application of the conceptual framework shows that the latter approach minimized distributional implications three main predictors of potential revenue. Through – as reflected in changes in PAD. District fixed effects are a several regression tests, we found that GRDP, population, better way to improve the accuracy of the potential revenue and urbanization rates are good predictors for these estimation model, given its simplicity in capturing SNG revenue sources. characteristics without identifying specific sectoral GRDP predictors as control variables. Placing district fixed effects to the potential revenue estimation model resulted in a fairer DAU allocation. This Incorporating a temporal dimension into the future paper explored two approaches in improving the distributional application of the main predictor holds the potential for implications of the model: (1) adding control variables that enhanced effectiveness. While this study primarily focuses incorporate SNGs’ economic landscape and (2) placing on evaluating the predictive capacity of proposed variables district fixed effects. Both approaches aimed to ensure that for district-level potential revenue across Indonesia, it may be the potential revenue estimation model would capture SNGs’ useful to explore the effect of lagged predictors. This could be characteristics so that DAU distribution would incorporate each an important extension of this analysis in the future. SNG’s realistic capacity subject to their potential. We found EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 24 References Awasthi, R., T. L. Le, and C. You. 2020. “Determinants of Property Tax Revenue: Lessons from Empirical Analysis.” Policy Research Working Paper 9939, World Bank, Washington, DC. Batóg, B., and J. Batóg. 2021. “Regional Government Revenue Forecasting: Risk Factors of Investment Financing.” MDPI 9 (12): 210. Fadliya, and R. McLeod. 2010. “Fiscal Transfers to Regional Governments in Indonesia.” Working Paper 2010/14, Australian National University, Arndt-Corden Department of Economics. Fritzen, S. A. 2009. “Public Administration.” In Contemporary Southeast Asia, Regional Dynamics, National Differences, edited by M. Benson, 2nd edition, 74–90. Basingstoke: Palgrave Macmillan. Gordon, T., R. Auxier, and J. Iselin. 2016. Assessing Fiscal Capacities of States: A Representative Revenue System –Representative Expenditure System Approach, Fiscal Year 2012. Washington, DC: Urban Institute. Hofman, B., K. Kadjatmiko, K. Kaiser, and B. S. Sjahrir. 2006. Evaluating Fiscal Equalization in Indonesia. Washington, DC: World Bank. Huang, Jr-Tsung, Kuang-Ta Lo, and Po-Wen She. 2012. “The impact of fiscal decentralization on tax effort of China’s local governments after the tax sharing system.” The Singapore Economic Review 57: 1250005-1. Lewis, B. D., and P. Smoke. 2017. “Intergovernmental fiscal transfers and local incentives and responses: The case of Indonesia.” Fiscal Studies 38 (1): 111–139. Mawejje, J., and R. K. Sebudde. 2019. “Tax revenue potential and effort: Worldwide estimates using a new dataset.” Economic Analysis and Policy 63: 119–129. Martinez-Vazquez, J., and J. Boex. 1997a. “Fiscal Capacity: An Overview of Concepts and Measurement Issues and Their Applicability in the Russian Federation.” GSU Andrew Young School of Policy Studies Working Paper 97-3, SSRN, Georgia State University. Martinez-Vazquez, J., and J. Boex. 1997b. An Analysis of Alternative Measures of Fiscal Capacity for Regions of the Russian Federation. Mogues, T., and S. Benin. 2012. “Do external grants to local governments discourage own revenue generation? A look at local public finance dynamics in Ghana.” World Development 40: 1054–1067. Negara, S. D., and F. E. Hutchinson. 2021. “The Impact of Indonesia’s Decentralization Reforms Two Decades On: Introduction.” Journal of Southeast Asian Economies 38 (3): 289–295. EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 25 Reddick, C. 2004. “Assessing Local Government Revenue Forecasting Techniques.” International Journal of Public Administration 27 (8–9): 597–613. Rothenberg, A. D., A. Gaduh, N. E. Burger, N. E., C. Chazali, I. Tjandraningsih, R. Radikun, C. Sutera, and S. Weilant. 2016. “Rethinking Indonesia’s Informal Sector.” World Development 80: 96–113. Rubin, M., N. Mantell, and M. A. Pagano. 1999. “Approaches to Revenue Forecasting by State and Local Governments.” Proceedings: Annual Conference on Taxation and Minutes of the Annual Meeting of the National Tax Association, 92: 205–221. Shah, A., Z. Qureshi, A. Bagchi, B. Binder, and H. F. Zhou. 1994. Intergovernmental Fiscal Relations in Indonesia: Issues and Reform Options. Washington, DC: World Bank. Stock, J. H., and M. W. Watson. 2020. Introduction to Econometrics. Fourth Edition. Pearson. Takahata, J., K. Khoirunurrofik, and T. Dartanto. 2021. “Intergovernmental Transfers in Indonesia: The Risk-Sharing Effect of Dana Alokasi Umum.” Journal of Southeast Asian Economies 38 (1): 81–99. Taylor, M., S. Keenan, and J. F. Carbonneau. 2002. “The Canadian Equalization Program.” Journal of Official Statistics 18 (3): 393–407. Williams, D., and S. Kavanagh. 2016. “Local government revenue forecasting methods: Competition and comparison.” Journal of Public Budgeting, Accounting and Financial Management 28: 488–526. Yilmaz, S., and F. Zahir. 2020. “Issues in intergovernmental fiscal transfers: Public finance and political economy considerations.” In Intergovernmental Transfers in Federations, 30–40, Edward Elgar Publishing. EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 26 Annex A. Regression Results of Conceptual Framework > > > T A B L E A . 1 - Potential Revenue Estimation: Regression Results for Property Tax Revenue Source: Awashti et al. (2020). > > > T A B L E A . 2 - Potential Revenue Estimation: Regression Results for Recurrent Property Tax (Fixed Effect) EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 28 > > > T A B L E A . 3 - Potential Revenue Estimation: Regression Results for Property Transfer Tax > > > T A B L E A . 4 - Potential Revenue Estimation: Regression Results for Street Lighting Tax EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 29 > > > T A B L E A . 5 - Potential Revenue Estimation: Regression Results for Hotel, Restaurant, and Entertainment Tax EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 30 Annex B. Regression Results of Models with Added Controls > > > T A B L E B . 1 - Empirical Results from Including Urbanization Rate and Mining GRDP in the Model EQUITABLE GROWTH, FINANCE & INSTITUTIONS INSIGHT <<< 32