Policy Research Working Paper 10538 Fiscal Policy and Equity Vietnam 2018 Fiscal Incidence Analysis Laura Rodriguez Kajetan Trzcinski Matthew Wai-Poi Poverty and Equity Global Practice August 2023 Policy Research Working Paper 10538 Abstract This paper examines the distributive and poverty reducing co-payments than what they receive in cash benefits, and effects of Vietnam’s fiscal system in 2018. The paper looks the fiscal system results in a small increase in poverty. The at the incidence across the distribution and the effect of progressivity of the fiscal system and its inequality-reduction (direct and indirect) taxes, subsidies, and social spending impact mostly comes from in-kind health and education (in cash and in-kind) on inequality and poverty in Vietnam spending. This reduction in inequality is about average for using the Commitment to Equity methodology. The overall lower-middle-income countries, but Vietnam could do pattern of taxes and transfers in Vietnam is moderately more to increase the progressivity of its fiscal system. progressive, but most households pay more in taxes and This paper is a product of the Poverty and Equity Global Practice. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://www.worldbank.org/prwp. The authors may be contacted at lrodrigueztak@worldbank.org, mwaipoi@worldbank.org, ktrzcinski@worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Fiscal Policy and Equity: Vietnam 2018 Fiscal Incidence Analysis1 Laura Rodriguez; Kajetan Trzcinski and Matthew Wai-Poi Keywords: fiscal incidence, poverty, inequality, taxation, social spending, transfers, Vietnam. JEL: D31, H22, I38 1 Acknowledgments: This paper was written under the overall guidance of Rinku Murgai. We are grateful to Jon Jellema, Monika Oczkowska and Michał Myckfor their contributions. We are also thankful to Arden Finn and Viet Anh Nguyen who provided peer review comments, to Judy Yang and other many World Bank colleagues and to staff from the Vietnam’s National Institute of Finance at the Ministry of Finance for discussion, comments and the information provided. Also, to the Vietnamese General Statistics Office who generously provided the household survey data needed for this work. This work was supported by the Australia - World Bank Group strategic partnership in Vietnam-Phase 2+ (ABP 2+). 1. Introduction This paper examines the distributive and poverty effects of the Vietnam fiscal system in 2018, serving as a pre-COVID-19 baseline. It is a technical background paper that details the methodology and analysis presented in Chapter 7 of the Vietnam Poverty and Equity Assessment report (World Bank 2022a). It will be followed up by an analysis of the fiscal system in 2020 and 2022 in subsequent papers. Different households pay various taxes and benefit from public spending in different ways. The net effect determines the extent to which fiscal policy directly reduces poverty and inequality. The choice of public spending can also affect how much poverty and inequality are reduced in the longer term. It is not just that fiscal policy can affect household income distribution today, but also it can finance public investments which can promote growth as well as reduce poverty and inequality in the long term. Thus, fiscal policy can play a critical role in both driving Vietnam toward high-income status and doing so in an inclusive manner which helps grow the size of the middle class. First, it can help finance the required investments needed for the country and its workers to become more productive and higher earners, such as modernization of agriculture, improved skills and higher-quality education, a more robust digital backbone, and accompanying services. Second, it can finance policies which can address constraints for poverty elimination for groups with persistently high poverty (Last Mile) and place Vietnam on the road to upper-middle and high-income country standards (Next Mile) (see World Bank 2022a) today, such as a modern social protection system and strengthening of National Target Programs (NTPs). Using the Commitment to Equity (CEQ) framework,2 the analysis in this paper seeks to answer three key questions: • Who pays and benefits from particular taxes and transfers in Vietnam? How progressive or regressive is each component of the fiscal system? • What is the net impact of all taxes and transfers on different households? • How does the combination of taxes and transfers affect poverty and inequality in Vietnam? The fiscal incidence assessment finds that most households in Vietnam pay more by way of taxes and co- payments than they receive in cash benefits. When in-kind health and education spending is included, the poorest four deciles are net beneficiaries of the fiscal system. Poverty at the lower-middle income poverty line (US$3.65 PPP) falls slightly when direct taxes and transfers are included, to then rise one point above the poverty rate at market income once indirect taxes and subsidies are added. While the fiscal system results in a small increase in poverty, it helps redistribution because of the progressive nature of direct taxation and in-kind spending, particularly on education. The Gini Index is reduced by nearly 3 points after accounting for all cash-based policies and by 5 points after including non-cash health and education spending. Compared to other LMICs, Vietnam’s fiscal system is one of the better performers with respect to poverty and around average with respect to inequality reduction. Vietnam raises more revenue from 2 The CEQ approach was developed by the Commitment to Equity Institute (CEQ Institute) at Tulane University. The methodology, implementation guidelines, applications, and software of the CEQ approach can be found in Nora Lustig (Ed.), Commitment to Equity Handbook. Estimating The Impact of Fiscal Policy on Inequality and Poverty (pp. 3-55). Brookings Institution Press. 2018. 2 the least progressive taxes and spends the most on the least progressive expenditures, indicating the potential for fiscally neutral reforms which would reduce inequality further. The paper consists of five sections: after the introduction, section 2 provides an overview of the fiscal system in Vietnam. Section 3 presents the data and the Commitment to Equity (CEQ) framework and how it is applied to the Vietnam case. Section 4 shows the results of the analysis; first it presents the overall poverty and inequality impacts of the different fiscal instruments and the system as a whole; then it shows the distributional analysis of the different fiscal interventions and their cost-effectiveness; finally, the results are put in international perspective. The last section concludes and discusses policy recommendations. 2. Overview of the Fiscal System This section provides an overview of Vietnam’s fiscal system in 2018. Taxes Tax revenues account for 18.4 percent of Vietnam’s GDP in 2018 (Table 1). Indirect taxes—value added tax (VAT), excises and environmental protection tax (EPT)—are by far the largest revenue components, representing 9.2 percent of GDP and almost half of tax revenues. Personal income tax (PIT) accounts for 9 percent of tax revenues and 1.7 percent of GDP. Alongside EPT, Vietnam also has a natural resources tax on the exploitation the country’s natural resources such as petroleum, minerals, natural gas, forestry products, and natural water. These are excluded from the analysis in this study, alongside taxes on the commercial sector, land and registration taxes. Corporate income tax, a large although declining revenue component,3 is not included in this analysis because its incidence cannot be easily or accurately allocated to households. As a result, the analysis here captures just under 60 percent of total tax revenues. PIT is levied on gross income from employment less exemptions and deductions. The income tax rate has seven tiers and it is progressive on the income level: the first 5 million Vietnamese dong (VND) of an individual’s monthly income are taxed at a 5 percent rate, income between 5 million and 10 million VND is taxed at 10 percent, income between 10 million and 18 million VND is taxed at 15 percent, and so forth (see Appendix 1, Table 5) up to 80 million VND, above which income is taxed at 35 percent. Vietnam’s PIT framework does not have a tax-free threshold, but there are significant personal and dependent deductions and allowances, in addition to the relatively low rates for lower tiers. There is a personal allowance of 9 million VND, a deduction for dependents4 of an additional 3.6 million VND per dependent and some deductions for social security contributions (social, health and unemployment insurance) and payments to charities. PIT is entirely borne by the worker. A rental tax of 5 percent applies to incomes from rents in excess of 100 million VND per year. There is also a 5 percent withholding tax that applies to interest income, royalties, income from services, with the exception of bank savings. A real estate tax for sales proceeds is 2 percent of the value of the land or 3 The CIT rate was cut by over 10 percentage points in the last two decades: the standard rate was 32 percent before 2004 and it is now 20 percent, with many reductions available which can make the rate as low as 10 percent. 4 Children under 18 years old, those over 18 who are students and have low income (under 1 million VND per month), spouses and parents that are unable to work or have low income or are out of working age. 3 building. A tax of 10 percent applies to gifts5 in excess of 10 million VND per year. Finally, a tax rate of 0.03, 0.07 or 0.15 percent, depending on the land area used applies to residential property owned by the household. Table 1. Vietnam government revenues, 2018 2018 Percent of Included Macro- VND trillion GDP in CEQ validation‡ Total revenues & grants 1,364 24.6 Tax revenues 1,022 18.4 Oil revenues 66 1.2 CIT 49 0.9 No Natural resources tax 17 0.3 No Non-oil taxes 956 17.2 Taxes on goods and services 509 9.2 VAT*** 344 6.2 Yes 61 Excise tax 96 1.7 Yes 67 Natural resources tax 38 0.7 No Environmental protection tax 47 0.8 Yes 44 Taxes on international trade 87 1.6 No Direct taxes 296 5.3 Corporate Income Tax 251 4.5 No Personal income and other taxes* 94 1.7 Yes 72 Other taxes Agriculture land use tax 0 0.0 Yes Residence land and housing tax 2 0.0 No Registration tax 32 0.6 No Lottery 29 0.5 No** Non-taxes 334 6.0 No Fees and charges 38 0.7 No Land rents 28 0.5 Yes 85 LUR assignment 148 2.7 No Other non-taxes 121 2.2 No Grants 8 0.1 No Note: Lines underlined are included in the analysis. ‡ Macro validation is the ratio of each simulated tax item to the administrative revenue total. * Includes PIT, capital gains, property rental, real estate, inheritance/gifts, and business income taxes. ** Excise on lotteries is included. *** VAT from budget includes all VAT collected from fiscal accounts, not just form households. Source: MOF, GSO, IMF and World Bank staff calculations The standard VAT rate in Vietnam is 10 percent. There is also a reduced rate of 5 percent for essential goods and services, which include clean water, teaching aids, books, unprocessed foodstuffs, medicine and medical equipment, husbandry feed, various agricultural products and services, technical/scientific 5 Excluding those from direct family members. 4 services, rubber latex, sugar and its by-products, social housing, and certain cultural, artistic, and sport services/products. There are also some exempted items.6 Excises apply to selected goods and services such as alcohol, automobiles of less than 24 seats, motorcycles, aeroplanes, boats, petroleum, air-conditioners up to 90,000 British thermal units (BTU), cigarettes, playing cards, discotheques, massages, karaoke, casinos, gambling, golf clubs, and entertainment with betting and lotteries (Table 7 in the Appendix). Among the highest rates is an excise of 70 percent of the pre-tax factory price, which is applied to domestic tobacco products, with VAT then levied on the retail price. The EPT was introduced in 20127 and it is an indirect tax that applies to 7 types of goods that are deemed to be detrimental to the environment such as coal, refined fuels (gasoline, diesel kerosene, mazut, etc.), hydrochlorofluorocarbons (HCFCs), plastic bags and some chemical substances used in agriculture and forestry. The tax is calculated as an absolute amount on the quantity of the goods (Table 8, Appendix 1). Expenditures In contrast to revenues, which are largely centralized in Vietnam, spending is quite decentralized. Data limitations mean important sub-national spending on social assistance is excluded from this analysis, which is material in a highly decentralized fiscal system such as Vietnam’s, in particular social assistance. Nonetheless, both central and local health and education expenditures have been included in this paper, and they account for the majority of social spending analysed. Data limitations prevent a more comprehensive review. Vietnam’s total central public expenditures in 2018 were around 19 percent of GDP; 36 of this, 6.8 percent of GDP, was spent on social and non-social expenditures included in the analysis (Table 2). Education was the largest expenditure, followed by health. Direct cash transfers comprise a small part of total expenditures, 11 percent of those analysed. Administrative data on electricity subsidies were not available but have been estimated at about 27 trillion dong in the current work. Vietnam’s social protection system is based broadly on three mechanisms: (i) social insurance; (ii) geographically targeted development programs and budget equalization mechanisms to channel resources to poor provinces and poor districts (currently, National Target Programs for New Rural Development and Sustainable Poverty Reduction (NTP-NRD and SPR))8 and (iii) household-targeted social 6 Certain agricultural products; goods/services provided by individuals having annual revenue of 100 million VND or below; imported or leased drilling rigs, airplanes, and ships of a type that cannot be produced in Vietnam; transfer of land use rights (LUR) (detailed guidance is provided to specific cases); various financial services; various securities activities including fund management; capital assignments; foreign currency trading; debt factoring; certain types of insurance; medical services and elderly/disabled people care service; education, printing/publishing, public transportation, export of unprocessed natural resources, among others. 7 Environmental Protection Tax Law was passed in 2010 but came into effect in 2012. 8 Their objectives are: (a) For NTP-NRD: “To introduce New Rural Development to enhance the spiritual and material life of people; to acquire relevant social and economic infrastructure, to achieve appropriate economic structure and production organizational arrangements, to link agriculture with the industry and service sectors; to link rural development with urbanization; to achieve democracy, equality, stability, rich cultural and national identify in rural society; to protect the ecological environment; and to maintain national defence and security, social order and safety”; and (b) For NTP-SPR: “To achieve sustainable poverty reduction and to restrain the resurgence of poverty, in contribution to the achievement of economic growth, social security, improved living 5 assistance programs, primarily social assistance transfers by MOLISA. It also includes programs at the district and commune levels not covered in national policies. Here we only cover central social assistance transfers. But social assistance is fragmented and underfunded, and has a delivery system with implementation issues (Nguyen and O’Keefe, 2019). Social assistance programs are concentrated on particular categories of households, such as the very old, ethnic minorities, and those with disabilities. Consequently, many poor households not meeting these criteria are excluded. Instead of a flagship social program, Vietnam has many smaller individual programs, which leads to different systems for implementation and inefficiencies. Table 2. Vietnam central government expenditures, 2018 2018 VND Percent of Included in Macro- Trillion GDP CEQ validation Total Expenditure 1,435 25.9 Expenses 1,051 19.0 Total spending analysed 376 6.8 Direct Transfers 40 0.7 Yes 100 Education 220 4.0 Yes 93 Pre-school 34 0.6 Primary 63 1.1 Lower secondary 49 0.9 Upper secondary 22 0.4 TVET 31 0.6 Tertiary 21 0.4 Health 89 1.6 Yes 164 In-patient 43 0.8 Out-patient 46 0.8 Subsidies 27 0.5 Electricity 27 0.5 Yes NA Notes: Education total expenditures are based on the 2018 administrative data. Breakdowns by education level are based on 2013 UNESCO data, pro-rated to the 2018 total expenditures. Modelled estimates come close to the total level (modelled is 229 trillion compared to 220 trillion admin) and pro-rated levels except for TVET (overestimated) and tertiary (underestimated). Health total expenditures are based on the 2018 administrative data; WHO 2018 data is 154 trillion. The table uses modelled levels for in-patient and out-patient and pro-rates to the admin total for this table. The analysis uses the modelled absolute levels. Total modelled expenditure accounts for 92 percent of administrative expenditure analysed. ‡ Macro validation is the ratio of each simulated tax or spending item to the administrative budget total Source: Ministry of Finance, Ministry of Education, WHO Global Health Expenditure Database, GSO, IMF, and World Bank staff calculations Residential electricity consumption is subsidised in Vietnam. The subsidy received depends on the level of electricity consumption (kilowatts per hour) according to a tariff structure (Table 6 in Appendix 1). Anything below the cost recovery price per kwh of 2.493 VND is subsidized. The lowest 3 consumption slabs receive a subsidy while there is a tax imposed on the 3 higher consumption slabs. Households who conditions and income generation for people, especially in poor areas; to facilitate the poor and poor households to access basic social services (health, education, housing, clean and hygienic water, and information), in contribution to the achievement of household poverty reduction targets in period 2016 -2020 as set out in the National Assembly Resolution.” 6 use pre-paid card meters pay a flat tariff close to the cost recovery price.9 There was also an electricity transfer of 51,000 VND per month that poor households received.10 Vietnam provides universal access to education. The country has high primary school completion rates, strong gender parity, low student/teacher ratios, and a low out of school rate. Nonetheless, some challenges remain. Educational completion is much lower for children in the poorest households than the richest ones. By age 19, only a fifth of students from the poorest 20 percent remain in school, compared with 80 percent of those in the wealthiest 20 percent. Inequities persist across ethnicities and geographies, with ethnic minorities often falling behind in educational attainment. Meanwhile, enrolment in secondary school in rural areas is nearly 15 percentage points lower than in urban areas (76 percent vs. 90 percent). Geographic disparities also exist: children in the Mekong Delta and Central Highlands have been persistently falling behind in learning outcomes over the last decade, although regional gaps are slowly narrowing (World Bank 2022a). In 2020, health insurance covered 91 percent of the population (MOH 2021), following the introduction of reforms to the Health Insurance Law in 2015 to improve progress towards universal health coverage. Support was shifted from supply-side to demand-side subsidies on health insurance premiums for various groups such as poor and near poor, among others. These reforms were aimed at expanding participation in health insurance, promoting enrolment in health insurance through household subscription, and increasing entitlements to the insured (Thu et al., 2020). As a consequence of the reforms, most households are covered and coverage does not vary significantly across the income distribution. Despite the high level of health insurance coverage, out-of-pocket costs remained high at 45 percent of current health expenditure in 2018 (World Bank 2022a), reducing the net value of public health services. 3. Data and Methodology Data This study uses data from the 2018 Vietnam Household Living Standards Survey (VHLSS) conducted by the General Statistics Office (GSO) with the purpose of monitoring living standards in the country and used for official poverty and inequality statistics. Crucially, the survey contains information on household incomes, and additional consumption expenditure information is also collected for 25 percent of the sample. The sample size is close to 47,000 households and it is representative at the national, regional, urban/rural and provincial levels. VHLS 2018 was collected in four periods (one in each quarter) through 2018, so we matched it to the 2018 fiscal data. The work follows the CEQ methodology, an internationally recognised fiscal incidence diagnostic method developed by the Commitment to Equity Institute. This approach uses standard incidence analysis for each tax and transfer, which are allocated to households based on the information in the household survey. The innovation of the CEQ approach is to combine the sectoral incidence analysis to model the net impact of taxes and transfers on households and determine their welfare and distributional impacts. A CEQ assessment uses eight different income concepts starting from ‘pre-fiscal’ or ‘market income’; that is, the income before all fiscal interventions (see Lustig 2018 and Figure 1). This includes all income from work (salaries and self-employed income), capital, self-provision of goods and services, remittances and other private transfers, private pensions income and the value of imputed rent. Under the CEQ approach, 9 We do not have enough information on the use of pre-paid card meters to allocate this. 10 This was suspended in 2019. 7 there are two options for treating pensions. Here we treat Pensions as Deferred Income (PDI), which uses market income plus pensions11 (hereafter simply referred to as ‘market income’) as the starting point of the assessment. This treats contributions to pensions made during working-years as mandatory savings that would be enjoyed later in life; during retirement the income from contributory pensions is considered part of the pre-fiscal income. Alternatively, contributions can be treated as a tax and payments as a transfer.12 The movement from one income concept to the next is done by iteratively assigning taxes and transfers as seen in Figure 1. ‘Disposable income’ is that resulting from adding direct government cash and near- cash transfers and subtracting direct taxes and contributions to market income. ‘Consumable income’ then subtracts indirect taxes and adds indirect subsidies. ‘Final Income’ adds the public cost of providing in-kind transfers (services which are not received as cash benefits, namely health and education). Income concepts and the values of taxes and transfers are reported in Vietnamese dong per month. Figure 1. Definition of CEQ Income Concepts and Fiscal Interventions in Vietnam MARKET INCOME (PGT scenario) Contributory social insurance old-age pensions (net of contributions) MARKET INCOME PLUS PENSIONS (PRE-FISCAL INCOME) Direct taxes: PIT, non- agricultural land use tax, property rental income tax, NET MARKET capital gains tax and taxes on INCOME gifts/prizes Direct cash transfers (social allowance for war veterans, social assistance, and natural disasters allowance). DISPOSABLE INCOME Indirect taxes: VAT; excises Indirect subsidies: electricity CONSUMABLE (tobacco, alcohol, cars, INCOME gasoline, lotteries, etc.); EPT In-kind transfers (health; education) net of user fees FINAL INCOME Source: Adapted from Lustig (2018) 11 Where income from pensions is added to factor income and contributions to old-age contributory pensions are subtracted from factor income. 12 Since the status as pensioner status is not directly recorded in VHLS we use the self-reported income and expenditure information to determine the amount of pension income received and the contributions made to the pension system. 8 In Vietnam, the WB-GSO poverty measurement uses consumption as the welfare aggregate. Consequently, the calculation of the income concepts begins by equating household consumption to disposable income and then working backward (subtract direct transfers and add direct taxes) to construct market income. To assess the impact of the fiscal system on poverty we use the $3.65 and $6.85 (2017 PPP) international poverty lines used for lower-middle income countries (LMICs) and upper-middle income countries (UMICs) respectively. Impacts across deciles of the consumption distribution and on the Gini Index are used to capture the distributional impacts. The Vietnam CEQ accounts for the following fiscal instruments: Personal income tax (PIT), capital gains tax, non-agricultural land use tax, property rental income tax and taxes on gifts/prizes, cash transfers, VAT, excises and environmental taxes, electricity subsidies and health and education in-kind benefits. The allocation of the fiscal instruments to households is done primarily based on the reported information in the household survey. This means that analysis already incorporates inclusion and exclusion errors in the allocation of different interventions; for example, some intended social assistance beneficiaries do not receive support for various reasons. The details about the allocation of the different fiscal instruments are presented next. Allocating direct taxes To allocate PIT paid by households we look at the self-reported employment incomes and impute the expected PIT burden that corresponds to those incomes minus any allowances and deductions.13 Households that have income from formal non-agricultural business activities are assumed to be taxed according to the PIT rates rather than the business income tax rates because we do not have enough information to allocate these taxes.14 We assume that only formal incomes are taxed. To account for informality, we follow the following procedure with the aim of matching as closely as possible the national informality rates in ILO.15 Formal employment for employees is assigned to those who: i) live in Red River Delta, Northern and Coastal Central Regions, Southeastern Area or the Mekong Delta regions, received a salary for the job and were entitled to paid leave and social insurance and worked in the public sector or had a work contract; ii) live in either Midlands and Northern Mountainous Areas or Central Highlands regions, received a salary for the job and were entitled to paid leave and social insurance or worked in the public sector and had a work contract; iii) live in Ho Chi Min or Hanoi and receive a salary for the job, worked for at least 20 days in the last month in that job and had a contract, was entitled to paid leave and social insurance; iv) work as leaders in agencies or as high and mid-level professionals; v) have compulsory health insurance; or vi) work in foreign companies. For any secondary employment, since the data is not as rich, we simply assign formality if the person receives salaries or wages for this job. For self-employed people, we assume that those who paper receiving business revenues from a registered business are formal incomes. PIT in the survey is likely underestimated because of missing top incomes from household surveys and because PIT revenue may also include foreigners. However, we do not attempt to allocate or scale up PIT or other 13 That is, self-reported labour incomes are assumed to be after tax, and PIT withheld is backed out from the reported after-tax income and the PIT schedule. 14 Household businesses are subject to a different regime, in which they pay a percentage tax on the turnover, including a VAT component and a PIT component. The rates are different depending on the type of business activities: trading (1% VAT + 0.5% PIT), Services and construction exclusive of building materials (5% VAT + 2% PIT), Production, transport, service associated with goods; construction inclusive of building materials (3% VAT + 1.5% PIT), and other activities (2% VAT + 1% PIT). 15 https://ilostat.ilo.org/topics/informality/ 9 direct taxes because we only have available information on the government revenues from all income taxes taken together, not distinguishing between income tax, capital income tax, non-agricultural land use tax, property rental income tax and taxes on gifts/prizes. Voluntary health insurance contributions for the self-employed are randomly assigned to those with a voluntary health insurance card and calibrated to the number of voluntary contributors in the administrative data; pensioners pay the full 4.5 percent contributions (employer plus employee rates). Property rental income tax is allocated using self-reported information on household income from leasing of residential land or houses. Capital gains tax are allocated by applying the corresponding rate to the self- reported household interests from savings incomes. Similarly, gifts/inheritances taxes are allocated based on the self-reported gifts incomes from non-family members (from abroad and in the country).16 We do not have information in the survey to allocate the real estate tax on property sales. For the non-agricultural land tax, we use the estimated value of the land17 (if the property was bought today) to assign the corresponding tax rate. This tax is allocated to households who own their residence. There is no information on the survey about the value or size of any second homes or properties owned by the household to apply the tax in these cases. Allocating indirect taxes For VAT and excises, we match the rates to the various expenditure items in the household survey. The direct burden is simply the household net expenditure by the corresponding tax rate.18 For VAT-exempt items we calculate the indirect effect using an Input-Output (IO) table and a cost-push model.19 This assumes that higher producer prices will be pushed to consumers in the final sale price of the product. The current model does not account for informality in indirect taxes even though it is well-established in the literature (e.g., Bachas et al. 2021, see discussion in World Bank 2022b) that a greater share of poor household consumption is informal, and thus has a lower effective tax rate for the same goods. The VHLSS does not record the place of purchase in the expenditure module (e.g., street vendor versus large supermarket) to be able to use this to account for informality on VAT. This means that at the moment, the regressivity of VAT in Vietnam is likely overestimated. A future extension of the work may apply informality rates and the informality Engel curve for Thailand20 to Vietnam. Environmental protection taxes are treated as an indirect effect of the items taxed and thus, we follow the standard approach to allocate their incidence. However, the IO table has highly aggregated categories, requiring blunt assumptions about the weight of the items taxed within the input sector. We use various 16 For rental income savings interests and gifts, it is assumed that the household survey information is net of taxes. Thus, we first backtrack it to obtain the gross income and calculate the tax paid. 17 There is no information in the survey about the size of the land plot. The use of value creates a distortion in tax estimates: land in more urbanised areas will generally have higher value (we estimate a higher tax paid) but be smaller in area (should have a lower tax paid). 18 For excises, this will likely overestimate the household incidence because excises in Vietnam are applied at factory instead of retail prices. For example, the effective excise rate is estimated to be less than 40 percent for tobacco products and less than 50 percent for alcoholic products, at retail prices. 19 VAT-exempt items do not have VAT charged to their final price but sellers are not able to claim back VAT paid on inputs; consequently there is embedded VAT in the final price of VAT-exempt goods. This does not apply to goods and services with a zero VAT rate (for which VAT on inputs can be claimed). 20 One of the few comparable countries in EAP for which this data is available. 10 sources of information available about the proportion of the output of a given sector that can be attributed to the production of the taxed product. Allocating cash transfers (social assistance) The program fragmentation, including that between the central and local governments, makes it difficult to allocate social assistance in Vietnam. There are 19 specific programs listed in the VHLSS alone, including cash transfers, subsidies, and in-kind provisions. For this section, we allocate three sources of direct cash transfers—social allowance for war veterans (meritorious service to the revolution), social assistance for social protection beneficiaries, and natural disasters and fires allowance—for which information is collected in the survey but since we do not have information on how these are assigned or targeted, we present them as an aggregate.21 None of the direct transfers are scaled, as we do not have enough disaggregated information (in the survey nor in the budget) to do so. Allocating electricity subsidies Administrative data on electricity subsidy spending is not available. The current analysis applies the difference between the return for one kWh of electricity and the subsidized price for one kWh. To estimate the subsidy received per household we use information on the households’ expenditure in the electricity bill to estimate the quantity of electricity consumed. This is then matched to the tariff(s) corresponding to this quantity to calculate the subsidy received. The electricity tariff structure is in Appendix 1. Allocating in-kind benefits: Education The composition of a household has a strong influence over how much it is likely to benefit from certain types of spending. For example, households with school-aged children will benefit from education spending when kids go to school, and young children and older people are more likely to use health care services relative to older children and younger adults. Poorer households in Vietnam tend to be younger, with more children. As households get richer, they have fewer children, with half or more of the richest two deciles having none at all (and those that do are more likely to attend non-subsidised private schools). At the same time, there are almost no elderly in the poorer half of the distribution (and relatively few in any household, although this will change as the country ages) (see figure in Appendix 2). The monetary value of public education is allocated as the cost per student per level of education (pre- primary, primary, lower and upper secondary, TVET and higher education) calculated from administrative budget and enrollment data. This is a government cost approach, in which the use-value to in-kind benefits is given by the average cost of service provision per beneficiary. This represents what the household would have to pay to use the service at the government’s cost. Data from past years on costs of each level of education or each kind of treatment are used, then uprated while maintaining their proportion to the total government expenditures in 2018.22 We deduct costs paid by households (tuition fees and contributions to schools/class) from the benefits. These are reported in the VHLSS but it is not possible to disaggregate them to allocate them to specific 21 In addition, there is an electricity benefit transfer for poor households. We do not allocate this specifically and treat it as part of the cash transfers. 22 Scaling of in-kind benefits was standard in earlier CEQs but there has been a movement away from this practice. In Vietnam, any scaling would be small as the gap between national accounts private consumption and VHLSS consumption is small. 11 children or by level of education. Thus, only total net and gross education benefits are calculated and these are not split by level of education. In a sensitivity analysis, we use a value-to-household adjustment which accounts for differences in health and education outcomes across the household per capita consumption distribution; households with better outcomes from human capital development are considered to receive more value from those public services.23 We use information from the Socioeconomically Disaggregated Human Capital Index (S- HCI) (D’Souza, Gatti and Kraay, 2019), which captures the expected future human capital of children born today for children of different parts of the income distribution (see Box 1). Households with better outcomes from human capital development are considered to receive a higher-value public service than those with lower outcomes. We adjust the education benefits to account for the income quintile gradient in learning-adjusted years of school, which combines information the quality and quantity of education that a child can be expected to obtain. Allocating in-kind benefits: Health We follow a similar approach to allocate health benefits. That is, a government cost approach allocating benefits to people who use public health services.24 Six types of visits, depending on the hospital type and whether they are inpatient or outpatient. We deduct out-of-pocket expenditures (hospital fees, thank you cash for doctors, VIP service fees, purchase additional medication, equipment, transportation, care giving, etc. related to that examination / treatment) incurred by households. Budget data is from WHO but this only provides an aggregate expenditure. To allocate to the different types of health care visits, we use ratios to distribute the total unit cost by type of service. These come from Flessa and Nghiem (2004), who estimate the costs of each kind of health intervention in Vietnam (inpatient vs. outpatient at various types of facility); these values are uprated to match the total public spending on health care in 2018. Out-of-pocket costs for health care use reported by households in VHLSS are deducted from the health benefits. As with in-kind education benefits, we also present quality-adjusted health benefits using information form the S-HDI (D’Souza, Gatti and Kraay 2019). We adjust health benefits to account for the income quintile gradient in under-five stunting rates. 4. Results Vietnam’s pre-COVID-19 fiscal incidence: Who pays and who benefits? Net impact of fiscal policy on households by income decile Most households in Vietnam pay more by way of taxes and co-payments than they receive in cash benefits. Figure 2 and Figure 3 summarise the results across the income distribution. Households are ranked according to market income and allocated to deciles. The poorest 10 percent are decile 1, the next 23 Following Rodriguez and Wai-Poi (2020). 24 Alternatively, instead of assigning health in-kind benefits based on service utilization, under the insurance approach benefits are allocated to those that have public health insurance. The rationale is that if they have insurance, they would be able to access the health care if they are to need it, which can arguably be what matters to both households and for fiscal incidence analysis. The insurance approach is more relevant in countries where health care access barriers are low and public insurance programmes are widespread. This method has not been applied in this paper but can be used in further updates to this analysis. 12 poorest 10 percent decile 2, up to the richest 10 percent. Figure 2 shows in (trillion) Vietnamese dong (VND) terms how much is paid and received across the income distribution, while Figure 3 shows how much is paid and received as a percentage of market income. The dotted line considers the aggregate impact of fiscal policy, excluding non-cash in-kind health and education spending. The first feature of note is that all but the poorest decile of households are net contributors in cash terms into the fiscal system (and even the first decile sees net benefits of less than 2 percent of their already low incomes). This contribution represents 10 percent or less of household market income for deciles 2-4 and then rises above 10 percent deciles 5-8 and is above 20 percent only for the top two deciles. Indirect taxation— largely VAT – is modestly regressive in relative terms, ranging from 5.1 percent of the poorest decile’s average income to 4.4 percent for the richest. The progressive nature of the aggregate fiscal cash impact is driven by direct taxation, part of which is personal income tax but most of which are contributions to social insurance, particularly social security contributions. Also of note is that direct transfers are relatively small albeit concentrated in the poorest decile and representing just over 10 percent of their income on average. When in-kind health and education spending is included (solid line in the figures), the poorest four deciles are net beneficiaries of taxes and public spending, with the poorest decile receiving net benefits of 34 percent of market income. The majority of the in-kind benefits come from education, which is worth around 13–22 percent of market income for the poorest two deciles. However, even when health and education spending is included, the net payment into the fiscal system for richer households remains relatively similar. In education, this reflects the number of richer households without children (or with children enrolled in private schooling). In health, it reflects overall low value of health benefits.25 Figure 2 Payments of Taxes and Benefits of Public Spending by Household Market Income Decile (Trillion VND) 10 Trillions 5 Fiscal Revenue / Expenditure (billion VND) 0 -5 -10 -15 -20 -25 -30 Decile 1 2 3 4 5 6 7 8 9 Decile 10 Household per capita market income decile Direct Taxes SSC Indirect Taxes (VAT+SST) Indirect Taxes (EPT) Direct Transfers Indirect Subsidies In-kind Spending (education) In-kind Spending (health) Total Impact Total Cash Impact 25 These can reflect high out-of-pocket (OOP) costs to access public health. Although we include an estimation of OOP costs (as reported by households in the survey), these appear small compared to the evidence indicating high OOP costs in Vietnam. See Appendix 2 for disaggregation of in-kind benefits. 13 Notes: Households are grouped into per capita market income deciles. Direct taxes include personal income tax, capital gains tax, property rental tax, non-agricultural land tax and gifts tax. Indirect taxes include VAT, SST and EPT. Direct transfers include social protection transfers, social allowance for war veterans and natural disasters and fires allowance. Indirect subsidies are residential electricity subsidies. In-kind spending includes health and education, net of costs or user fees. In- kind spending on health based on use of inpatient and outpatient healthcare. Contributory pension contributions and receipts are treated as deferred savings and income. Total cash impact excludes in-kind benefits. Source: World Bank calculations from VHLSS 2018 Figure 3 Payments of Taxes and Benefits of Public Spending by Household Market Income Decile (Percentage of Market Income) 50 Fiscal Revenue / Expenditure (percentage of market 40 30 20 income) 10 0 -10 -20 -30 Decile 1 2 3 4 5 6 7 8 9 Decile 10 Household per capita market income decile Direct Taxes SSC Indirect Taxes (VAT+SST) Indirect Taxes (EPT) Direct Transfers Indirect Subsidies In-kind Spending (education) In-kind Spending (health) Total Impact Total Cash Impact Notes: See Figure 2 Source: World Bank calculations from VHLSS 2018 Given the pattern of benefit and burden across the household income distribution, what is the impact of fiscal policy on poverty and inequality? At market incomes, the LMIC ($3.65/day 2017 PPP) poverty rate in Vietnam in 2018 was 5.5 percent and it was 20.8 percent at the UMIC poverty line ($6.85/day 2017 PPP).26 Figure 4 shows that poverty at the LMIC line falls slightly to 5.3 percent when direct taxes and transfers are included (disposable income), to then rise to 6.4 percent once indirect taxes and subsidies are added (consumable income). Using the UMIC line, poverty increases steadily from market to disposable (22.2 percent) and to consumable income (25.3 percent). The difference is that close to this higher UMIC poverty line (in decile 3 of market income), households start to pay more in direct taxes and make sizeable social security and health insurance contributions which are not matched by what they receive in direct transfers. Figure 5 (panel a) shows how each fiscal instrument contributes to poverty reduction or increases poverty (Individual subcomponents are scaled to the marginal effect of each fiscal category). This discussion is 26 Note that this differs from the results in WB (2022a) which uses the international poverty lines in 2011 PPP. 14 focused on the LMIC poverty line, as it renders a similar poverty rate to that obtained when using the Vietnamese national poverty line. Direct cash transfers contribute 0.8 points of poverty reduction (that is, they reduce poverty by 0.8 points). Direct taxes reduce this contribution by 1 point, adding mostly to social security and health insurance contributions. While electricity subsidies make a minor contribution to poverty reduction, VAT, environmental taxes, and particularly excises more than offset the impact of direct transfers and result in overall poverty being higher after all cash fiscal policies are considered (at consumable income). Spending on health and education is not included in the poverty impacts as these benefits are non-cash.27 However, they not only benefit poorer households more today, as reflected in the inequality discussion next, but also benefit them in the long-term. The returns to education in terms of better adult health outcomes and higher earnings are well-documented and poorer children disproportionately benefit (for example, see Holla et al. 2021 and World Bank 2022b). Figure 4. (a) Poverty and (b) inequality change between income concepts (a) (b) 30 39 38.2 38 25 25.3 37 Poverty headcount 35.7 35.5 20 22.2 36 20.8 15 Gini index 35 34 33.3 10 5.5 6.4 33 5.3 32 5 31 0 30 Market Disposable Consumable Final Income Market Disposable Consumable Final Income Income Income Income Income Income Income (m+p) (m+p) Poverty (US$ 3.65 PPP) Inequality (Gini) Poverty (US$ 6.85 PPP) Note: Poverty is measured at the lower-middle income (US$ 3.65) and upper-middle income (US$ 6.85) international poverty lines in 2017 PPP. Inequality is measured by the Gini Index. Market income includes all income from wages and earnings, capital incomes and rents, and private remittances. Disposable income subtracts direct taxes and adds direct transfers. Consumable income subtracts indirect taxes and adds indirect subsidies. Final income adds in-kind health and education spending. Source: World Bank analysis from VHLSS 2018 27 These non-cash benefits cannot be traded or used to pay expenses such as rent or food, have an opportunity cost to use, and are not necessarily valued by households at the same price as the cost of the government to deliver them (as they are valued here). 15 Figure 5. The impact of fiscal policy on (a) poverty ($3.65 2017 PPP) and (b) inequality by fiscal instrument (a) 1.0 Percentage Point Reduction 0.5 0.0 -0.5 -1.0 -1.5 b) 6.0 Percentage Point Reduction 5.0 4.0 3.0 2.0 1.0 0.0 Source: World Bank analysis from VHLSS 2018 Note: The figure presents marginal effects on poverty (inequality) reduction. Individual subcomponents are scaled to the marginal effect of each fiscal category. For example, the marginal effects of VAT, excises and environmental taxes are scaled to sum to the marginal effect of indirect taxes as a category. A positive bar (green)—falling down from the previous category—represents a contribution to poverty (inequality) reduction. A negative bar (red)—going up from the previous category—represents an increase in poverty (inequality). 16 While the fiscal system results in a small increase in poverty, it does help redistribution because of the progressive nature of direct taxation and in-kind spending, particularly on education.28 The Gini Index measure of inequality is reduced by nearly 3 points after accounting for all cash-based policies and by 5 points after including non-cash health and education spending (Figure 4 and Figure 5 panel b). The key progressive instruments are direct taxes and transfers, which reduce inequality by nearly 3 points, driven mostly by social security and health insurance contributions, and education spending, which contributed nearly another 2 points of inequality reduction. Nonetheless, when adjusting in-kind benefits to account for the lower value of the services received by the poorest households, the contribution of health and education benefits to inequality reduction falls by around fifteen percent, and the overall redistributive effect of Vietnam’s fiscal system declines by 7 percent, to around 4.5 points in total (see Box 1). Indirect transfers (electricity subsidies) and indirect taxes were largely neutral and had little effect on inequality. In the case of electricity subsidies, this results from expenditures on electricity are very low overall, and because there is a steep gradient in electricity consumption, meaning that poorer people in Vietnam use less electricity than richer ones, and an existing block tariff structure which subsidizes smaller consumers. Even though poverty is increased slightly by the fiscal system in Vietnam, this does not necessarily imply net negative transfers for all poor households. The fiscal impoverishment indicator quantifies how many people were pre-fiscal poor and made poorer by the fiscal system or were non-poor and made poor. Excluding in-kind transfers in health and education and using the US$3.65 poverty line as benchmark, 5.5 percent of individuals have been impoverished by fiscal policy (Table 3). In other words, they were poor to begin with and were made poorer because they paid more in taxes than they have received in direct or indirect transfers, or, their market income was above the poverty line but their consumable income fell below. On the other hand, 1.4 percent of individuals were made richer by fiscal policy, that is they were lifted above the poverty line because received more in cash or near cash benefits than they paid in taxes and contributions, or, even if they remained poor, they were better off after taxes and transfers. Table 3. Fiscal Loss or Gain at Consumable Income by Poverty Line (international poverty lines, in 2017 purchasing power parity (PPP) U.S. dollars. Fiscal Impoverishment Fiscal Gains Post-fiscal losers Post-fiscal gainers (percent total (percent total population) population) US$ 2.15 1.3 0.7 US$ 3.65 5.5 1.4 US$ 6.85 22.6 3.3 Note: The table shows loss and gains based on households’ consumable income, which is defined as market income − direct and indirect tax payments + direct transfers. It excludes in-kind transfers (for education or health care). a. “Fiscal impoverishment” is the extent to which fiscal system (of taxes and transfers) either (a) make a substantial portion of the poor even poorer, of (b) make the nonpoor poor. In other words, if the poor pay more in taxes than they receive in transfers, they are fiscally impoverished (Higgins and Lustig 2016). b. “Fiscal gains to the poor” is the opposite of fiscal impoverishment, representing more in gains from transfers than they pay in taxes (Higgins and Lustig 2016). 28 Although in-kind health and education spending is excluded from the poverty impact as they are not cash benefits. 17 Box 1. The value of in-kind health and education spending can also take into account quality Just over two-fifths of the total reduction in inequality through Vietnam’s fiscal system comes from health and education spending. However, poorer children do not benefit from education and health spending to the same extent as richer children. The value to households may not be the same as the cost of providing them, as is assumed in the main results. Quality may vary from province to province or between urban and rural locations, for example. Some classrooms may have better equipment than others, smaller teacher-pupil ratios or more textbooks. To test how sensitive this key contributor to fiscal inequality reduction is to the assumption that the benefit to a household is the cost of service provision, an alternative benefit level can be constructed. For this purpose, the Human Capital Index (World Bank 2020) can be disaggregated across household welfare quintiles (D’Souza, Gatti, and Kraay 2019). Figure X shows how well each quintile scores on the different HCI components and overall relative to the richest quintile. For example, the poorest quintile stays in school for only 80 percent of the time that the richest quintile does. Moreover, the average test scores for the poorest quintile are only 87 percent those of the richest quintile, so even when they do stay in school, they do not learn as much. Therefore, when their years of schooling are adjusted for how much they learn, the poorest quintile has only 72 percent of the educational attainment as the richest quintile. Stunting outcomes are even more unequal. Adjusting the value to a household in a particular quintile of health and education spending by their human capital outcomes means inequality is reduced by 7 percent less. The value of health and education spending to each household can then be adjusted according to that household’s S -HCI quintile result relative to that of the richest quintile, reflecting the lower human capital outcomes that household achieves. After making this adjustment, the contribution of health and education benefits to inequality reduction falls by around 15 percent, from 2.4 points to 1.9 points, and the overall redistributive effect of Vietnam’s fiscal system declines to around 4.5 points in total. 100 Quintile score relative to Quintile 5 80 60 40 20 0 Q1 2 3 4 Q5 Q1 2 3 4 Q5 Q1 2 3 4 Q5 Q1 2 3 4 Q5 Q1 2 3 4 Q5 Q1 2 3 4 Q5 Survival Years Test scores Adjusted years Stunting S-HCI Note: For each component (and the aggregate index), the Q5 score is normalized at 100 and the other quintile scores normalized relative to this. That is the normalized score for Quintile X is Qx / Q5 Source: D’Souza, Gatti and Kraay (2019) and World Bank calculations The issues with quality and value of such spending to households notwithstanding, in most countries spending on human capital accounts for much of the inequality-reducing effect of fiscal policy. Vietnam is relatively unusual in that less than half of the fiscal impact on inequality comes from health and education spending. In part this reflects relatively low spending levels in these sectors. For example, while Vietnam’s education spending (4.0 percent of GDP) is broadly in line with that of upper middle income (3.9 percent) and middle income (3.7 percent) averages, it is lower in health (2.3 percent compared to 3.3 percent for upper middle-income countries and 2.8 percent for middle income ones). 18 Impact and cost-effectiveness of fiscal policy instruments on inequality Vietnam raises more revenue from the least progressive taxes and spends the most on the least progressive expenditures, indicating the potential for fiscally neutral reforms which would reduce inequality further. The previous section examined how much each fiscal instrument increased or reduced inequality. The progressivity of the different fiscal interventions with respect to the distribution of market income is summarised in the Kakwani coefficients29 (Table 4). These confirm that direct taxes in Vietnam are progressive while indirect taxes are close to neutral. All direct transfers all progressive, as are electricity subsidies. In-kind education and health spending also have positive Kakwani coefficients, a reflection of their progressivity. Table 4 Kakwani Coefficient and Marginal Contribution Figure 6 Budget (VND billion) and Effectiveness of Main Taxes of Main Taxes and Transfers and Transfers 6.0 16 Kakwani Marginal 14 Coefficienta Contributionb 5.0 12 Budget (% of GDP) Main taxes 4.0 10 Effectiveness PIT 0.5 0.6 8 VAT and SST 0.0 -0.2 3.0 6 EPT 0.0 0.0 2.0 4 Direct transfers 2 1.0 Cash transfers 0.7 0.6 0 Indirect transfers - -2 Electricity subsidy 0.2 0.2 In-kind transfers Education benefits 0.6 0 Health net benefits 0.3 0.5 Budget (% of GDP) (LHS) Effectiveness (RHS) Notes: a. Kakwani coefficients measure whether a fiscal intervention exercises an equalizing or unequalising force. It is measured as the difference in the concentration curve of the tax and the Gini for market income (or the inverse in the case of transfers). Progressive interventions have positive Kakwani coefficients, and regressive ones have negative coefficients. b. Marginal contribution is the points which the Gini Index is reduced between market income and consumable income, or between Consumable and Final income (for in-kind transfers); a positive contribution is a reduction in Gini, negative contribution is an increase in Gini. Budget Is the budget identified in the survey as a percentage of GDP. Effectiveness is marginal contribution over budget. Source: VHLSS, 2018 Government Budget, administrative data and World Bank analysis. Figure 6 considers the progressive instruments and asks which is the most cost-effective in reducing inequality. That is, given the money spent on each instrument (or the revenue raised), how much is inequality reduced? A cost-effectiveness index is constructed (on the right-hand axis), which is the decrease in inequality per dong raised (taxes) or spent (spending). Second, this is compared to the size of spending (or revenue in the case of a tax) on that instrument (left-hand axis). Personal income taxes reduce inequality the most per dong of revenue raised. Other direct taxes are half as effective at reducing inequality and indirect taxes even less so. If Vietnam were looking to increase tax revenues, 29 It is measured as the difference in the concentration curve of the tax/transfer and the Gini for market income. It ranges between -1 and 1; progressive interventions have positive Kakwani, and regressive ones have negative coefficients. 19 broadening the income tax base would both bring in more money and reduce inequality at the same time. Similar results can be seen on the expenditure side. Direct cash transfers are nearly three times more effective at reducing inequality than less well-targeted electricity subsidies, yet almost as much is spent on the subsidies as the transfers. Education and health transfers, despite their progressivity, are not as cost-effective because of the large budgets attached to those transfers. Results in international context This section benchmarks Vietnam’s performance with other countries in the CEQ database (see World Bank 2022b), and then discusses how some middle-income countries are able to achieve more pro-poor outcomes with different combinations of fiscal instruments. The previous section showed that Vietnam’s fiscal policy in 2018 increased short-term poverty slightly by 0.9 points once all direct and indirect taxes and transfers are taken into account, excluding the many beneficial long-term poverty reduction effects of public investments in health and education. At the same time, inequality is reduced by 3 points when only cash instruments are considered and by 5 points when in-kind spending on health and education are also included. How do these results compare to other countries? Using the CEQ methodology across a selection of low to high-income countries, richer countries tend to have the most progressive mix of fiscal policies that help further reduce poverty. But particularly compared to other lower middle-income countries, Vietnam is one of the better performers with respect to poverty and around average with respect to inequality. The small increase in poverty from fiscal policy is a relatively good outcome for a lower-middle-income country, which average a 1.5 point increase and some of which see 5 or 7 point increases (Figure 7).30 Figure 7 Fiscal Impact on monetary poverty (points reduction in poverty headcount at international poverty lines in US$ 2011 PPP) 10.00 Change in Poverty Rate (percent) 8.00 6.00 4.00 2.00 0.00 -2.00 -4.00 -6.00 -8.00 -10.00 Spain Bolivia India Chile Venezuela Malaysia Guatemala Turkey Dominican Republic Armenia Vietnam Thailand Russia Russia Botswana El Salvador Lesotho Ethiopia Romania South Africa Peru Nicaragua Zambia Ecuador Guinea Togo United States Ivory Coast West Bank and Gaza High income Upper middle income Lower middle income Low income Direct taxes and transfers Indirect taxes and transfers Net impact Source: World Bank (2022b) and World Bank analysis from VHLSS 2018 30 For international comparison purposes, poverty reduction in Vietnam is estimated at the international LMIC poverty line of US$3.20 in 2011 PPP. 20 Figure 8. Fiscal Impact on Inequality (points reduction in Gini Index) 0 Change in Gini Index (points) -5 -10 -15 -20 -25 Mexico Botswana China Spain Uruguay Jordan Ivory Coast Panama Ukraine Georgia Venezuela Costa Rica Ecuador Russia Guatemala Indonesia El Salvador Egypt Cambodia South Africa Dominican Republic Peru Iraq Iran Malaysia Honduras United States Mauritius Thailand Albania India Moldova Sri Lanka Tanzania Comoros Togo Gambia Niger Croatia Namibia eSwatini Kenya Nicaragua Ghana Uganda Mali Romania Brazil Zambia Mongolia Bolivia Ethiopia Argentina Colombia Lesotho Turkey Belarus Paraguay West Bank and Gaza Guinea Tunisia Burkina Faso Tajikistan Vietnam High income Upper middle income Lower middle income Low income Cash taxes and transfers In-kind spending on H+E Net fiscal impact Source: World Bank (2022b) and World Bank analysis from VHLSS 2018 The impact of fiscal policy on reducing inequality is more mixed across economies at different levels of development (Figure 8). The reduction in inequality from Vietnam’s fiscal policy is around average for lower-middle-income countries. In all countries inequality declines after fiscal policy. The 3-point decrease in Vietnam ranks in the middle of lower-middle-income countries. Similarly, when in-kind spending is included, Vietnam’s ranking among these countries remains unchanged, although it gets closer to the best performers. Indirect taxation tends to be the main source of burden for the poor. Of the 33 low, lower and upper middle-income countries out of the 48 countries for which data are available in the CEQ cross-country database, the increase in poverty occurs when moving from disposable income to consumable income. That is, the burden of indirect taxes such as VAT and excises can cause households to become impoverished if there is not an offsetting benefit from direct cash transfers. However, in all high-income countries except for Romania and under half of upper middle-income countries, poverty is reduced at both the disposable and consumable income stage; direct transfers are well targeted enough with sufficient coverage generosity to more than offset the effect of indirect taxation. A reliance on indirect taxation initially presents a trade-off with respect to revenue collection and equity (Fuchs, Sosa, and Wai- Poi 2021). Some countries apply extensive VAT exemptions or preferential rates to staples such as food and clothing. These exemptions can be quite meaningful to poorer households for whom these items make up a greater share of their consumption baskets. At the same time, richer households also consume these goods and in greater amounts, meaning that richer households usually benefit more from the often- considerable tax expenditures. Countries with such exemptions have more neutral or even progressive indirect taxation incidence but lower revenue collection, as is discussed in Fuchs et al. (2021). The degree of informality in an economy should also be considered. Indirect taxes are usually not applied to purchases of goods and services at informal locations, such as street stalls. Some degree of indirect taxation is still included in the price, as inputs into production are often bought on the formal market (such as electricity), 21 but VAT is not applied to the final good, meaning the effective tax rate is lower. As Bachas, Gadenne, and Jensen (2021) show for 32 developing countries, the share of household consumption spent in the informal sector declines with income; the rich face an effective rate twice as high as that of the poor on average. This phenomenon reverses the revenue- equity trade-off of exemptions and preferential rates: greater informality means more progressive indirect taxation but lower overall collection rates. Figure 9. Tax mix by country income level, direct and Figure 10. Expenditure mix by country income level, indirect taxes 35 50 30 25 20 15 10 5 0 0 LIC LMIC Vietnam UMIC HIC LIC LMIC Vietnam UMIC HIC OECD Other Pensions Social protection (ex. Pens.) Health VAT Excises Trade* PIT + payroll PT CIT Education Source: World Bank (2022b) and World Bank analysis from Source: World Bank (2022b) and World Bank analysis from VHLSS 2018 VHLSS 2018 Note: Social protection in Vietnam includes direct transfers (0.7 percent) and electricity subsidies (0.5 percent). Pensions included under ‘Other’ for Vietnam. The mix of taxes used by a country has a strong influence on how progressive its fiscal policy will be. Direct taxes, such as corporate and personal income taxes, property taxes, and inheritance taxes, are powerful tools for reducing inequality. They are paid largely by richer households, directly reducing the inequality arising from market incomes. But part of the revenues can also be used to finance pro-poor investments, such as in human capital and physical infrastructure, thus indirectly reducing inequality in the future. Indirect taxes, such as VAT and excises, are much less progressive (as has been seen to be the case in Vietnam) and often regressive, representing a greater percentage of market incomes for poorer households. However, collecting personal income tax requires significant tax administrative capacity. As a consequence, most developing countries rely largely on increasing indirect taxation as they increase tax revenues as a percentage of GDP and move to a higher income level. World Bank (2022b) and Bachas, Sosa, and Wai-Poi (2023) show that as countries move from low income to lower-middle to upper-middle- income levels, they largely increase their tax revenues not through progressive personal income tax but instead through greater levels of less progressive indirect taxation (Figure 9). Even non-OECD high-income countries are more reliant on indirect taxation relative to direct taxation. It is only in OECD high-income countries that direct taxation significantly outweighs indirect taxation in the revenue mix. Consequently, for upper-middle-income countries looking to transition to high-income status, strengthening tax administrative capacity and broadening the personal income tax base is a sound way to both increase revenue collection and reduce inequality. For other developing countries, greater revenues are realistically going to be collected through indirect taxes. However, non-progressive taxation can be combined with progressive transfers to achieve a net progressive fiscal system. As this paper has made clear, the impact on poverty and inequality of taxes 22 should not be considered in isolation. All taxes will place a burden on at least some households. It is the net effect of these taxes and the spending they finance which determines the extent to which a fiscal system reduces poverty and inequality. The most effective instrument for reducing inequality is targeted direct cash transfers. In almost all countries, poorer households enjoy a greater share of cash transfer spending than richer households which represents an even greater percentage of their market income (World Bank 2022b). Thus, a combination of increased indirect taxes—perhaps through the removal of inefficient exemptions and preferential rates—can both raise revenues and reduce poverty and inequality if it is combined with an expansion of a targeted safety net. A country can only spend what it raises in revenues, so richer countries (which raise higher revenues) can spend more as a percent of their national income (Figure 10). But fiscal policy can also reduce inequality without necessarily spending more. Inefficient spending on energy and food subsidies can be better directed toward targeted direct transfers. A fraction of the subsidy spending can achieve greater reductions in poverty and inequality when spent as targeted cash transfers (Fuchs et al. 2021). 5. Conclusions This paper uses the CEQ fiscal incidence analysis to trace how household income changes when taxes are paid and benefits received. Fiscal policy can be used to both reduce inequality in Vietnam today and finance sustainable pro-growth and pro-poor investments for the future such as public investment in physical and digital infrastructure and human capital that countries need to make the transition to upper- middle- and high-income status. In many countries, COVID-19 highlighted the role that fiscal policy can also play in mitigating shocks (e.g., World Bank 2022b, Chapter 4). Overall, in Vietnam the benefits received by the poor are outweighed by what they pay into the fiscal system and poverty is slightly increased by the fiscal system. The impact on inequality is more positive, especially since spending on health and especially education is quite progressive. This reduction in inequality is about average for lower-middle-income countries, but many countries are more progressive. There is significant potential for Vietnam’s fiscal policy to do more in terms of investing for inclusive growth by adopting lessons from other countries. International experience suggests that countries become richer in part through greater public investment in drivers of economic growth and inclusive prosperity, increasingly financed by relying more on progressive direct taxation such as personal income taxes rather than indirect taxation such as goods and services taxes. While broadening the PIT tax base for higher equity and potential revenue gains is a long-term aim, in the short-term, using a combination of indirect taxes and targeted transfers can still a good approach for Vietnam at the current stage of development. In lower-middle-income countries, where indirect taxation underpins the majority of tax revenue, some choose to offset the tax burden on poorer households by applying lower rates or exemptions on staples consumed by the poor such as food and clothing, as Vietnam does. However, this means sacrificing significant revenue, the benefits of which go mainly to richer households who also purchase these goods, and in greater amounts. Other LMICs raise more revenue by closing such exemptions and are able to nonetheless reduce poverty and inequality to a greater degree and at lower cost by spending some of the extra revenue on targeted direct transfers to poorer households. Similarly, Vietnam’s approach of subsidizing energy is much less cost-effective at reducing poverty and inequality than broader social assistance coverage with more generous benefit levels. The path to fiscal and economic recovery from COVID-19 will need to address fiscal deficits while maintaining adequate support for vulnerable households and protecting key spending on human capital. 23 References Bachas, Pierre; Gadenne, Lucie; Jensen, Anders. 2020. Informality, Consumption Taxes and Redistribution. Policy Research Working Paper; No. 9267. World Bank, Washington, DC. https://openknowledge.worldbank.org/handle/10986/33851 Dsouza, R., Gatti, R.V., Kraay, A.C. 2019. A Socioeconomic Disaggregation of the World Bank Human Capital Index. Policy Research Working Paper; no. WPS 9020 Washington, D.C.: World Bank Group. http://documents.worldbank.org/curated/en/752571569852714031/A-Socioeconomic- Disaggregation-of-the-World-Bank-Human-Capital-Index Flessa, Steffen, and Nghiem Tran Dung. 2004. Costing of services of Vietnamese hospitals: identifying costs in one central, two provincial and two district hospitals using a standard methodology. The international journal of health planning and management, 19, no.1: 63-77 Fuchs, A., Sosa, M. and Wai-Poi, M. 2021. Progressive Domestic Resource Mobilization for a COVID-19 Recovery, Briefing Note, Washington, DC: World Bank Group Higgins, S. and Lustig, N., 2016, Can a poverty-reducing and progressive tax and transfer system hurt the poor?, Journal of Development Economics, 122, issue C, p. 63-75, Holla, Alaka; Magdelena Bendini, Lelys Dinarte and Iva Trako. 2021. Is Investment in Preprimary Education Too Low? Lessons from (Quasi) Experimental Evidence across Countries, World Bank Policy Research Working Paper 9723, Washington, DC: World Bank Lustig, N. (Ed.) 2018. Commitment to Equity Handbook. Estimating the Impact of Fiscal policy on Inequality and Poverty. New Orleans and Washington, D.C.: CEQ Institute at Tulane University and Brookings Institution Press. Ministry of Health, 2021. https://moh.gov.vn/hoat-dong-cua-lanhdao-bo/- /asset_publisher/TW6LTp1ZtwaN/content/het-nam2020-co-87-96-trieu-nguoi-tham-gia-bhyt- bao-phu-90-85-dan-so [Accessed 15th November 2021] Nguyen, Nga Nguyet, and Philip B. O’Keefe. 2019. A Vision for the 2030 Social Protection System in Vietnam. Washington, D.C.: World Bank Group Rodriguez, L and Wai-Poi, M. 2020. Fiscal Policy, Poverty and Inequality in Jordan: The Role of Taxes and Public Spending. World Bank, Washington, DC. Thu Thuong NT, Van Den Berg Y, Huy TQ, Tai DA, Anh BNH. 2020. Determinants of catastrophic health expenditure in Vietnam. International Journal Health Planning and Management;1–18. https://doi.org/10.1002/hpm.3076 World Bank. 2020. The Human Capital Index 2020 Update: Human Capital in the Time of COVID-19. World Bank, Washington, DC. © World Bank. https://openknowledge.worldbank.org/ handle/10986/34432 World Bank. 2022a. From the Last Mile to the Next Mile – 2022 Vietnam Poverty & Equity Assessment. Washington, DC. http://hdl.handle.net/10986/37952 24 World Bank. 2022b. Poverty and Shared Prosperity Report 2022: Correcting Course. Washington, DC: World Bank. http://hdl.handle.net/10986/37739 25 Appendix 1. Assumption Tables and Additional Information Table 5. Personal income tax rates Income Rate (percent) Up to 5 million per month 5 Above 5 to 10 million 10 Above 10 to 18 million 15 Above 18 to 32 million 20 Above 32 to 52 million 25 Above 52 to 80 million 30 Above 80 million per month 35 Note: rates applied to salary and wage income Table 6 Electricity tariffs for residential customers Customer group Rate (VND/kWh) For the kWh from 0 – 50 1,678 For the kWh from 51 – 100 1,734 For the kWh from 101 – 200 2,014 For the kWh from 201 – 300 2,536 For the kWh from 301 – 400 2,834 For the kWh from 401 kWh onwards 2,927 Retail price for household electricity via prepaid card meter 2,461 Cost-recovery 2.439 Table 7. Excise tax rates Item Tax rate (percent) Cigarettes 70 Alcoholic beverages with at least 20 degrees 65 Alcoholic beverages with less than 20 degrees 35 Beers 65 Motor vehicles with a capacity fewer than 24 people 10 to 150 Motorbikes with an engine displacement exceeding 20 125 cc Airplanes 30 Yachts 30 Gasoline 7 to 10 Air conditioners with a capacity not exceeding 10 90,000 BTU Playing cards 40 Votive gilt papers and votive objects 70 26 Dance studios 40 Massage and karaoke facilities 30 Casinos 35 Gambling dens 30 Golf courses 20 State lottery 15 Table 8. Environmental protection tax Item Tax Anthracite coal (hard coal) 30000 VND/kg Other types of coal 15000 VND/kg HCFCs, ozone-depleting substances and powerful 5000 VND/kg greenhouse gases Plastic bags 50000 VND/kg Gasoline 4000 VND/L Kerosene 1000 VND/L Lubricants and mazut 2000 VND/L 27 Appendix 2. Additional figures Figure 11. Direct taxes. Absolute impact (VND billion) Figure 12. Direct taxes. Relative impact (percent of market income) 0 0.0 Billions -500 -0.5 -1,000 -1.0 -1,500 -1.5 Fiscal Revenue (VND) Fiscal Revenue (Percentage of Market Income) -2,000 -2.0 -2,500 -2.5 -3,000 -3.0 -3,500 -3.5 -4,000 -4.0 -4,500 -4.5 -5,000 -5.0 Decile 2 3 4 5 6 7 8 9 Decile 1 2 3 4 5 6 7 8 9 10 1 Household per capita market income decile 10 Household per capita market income decile PIT Capital Gains Tax Property rental Tax Tax on gifts PIT Capital Gains Tax Non-agricultural land tax Total Impact Property rental Tax Tax on gifts Figure 13. Indirect taxes. Absolute impact (VND billion) Figure 14. Indirect taxes. Relative impact (percent of market income) 0 0 Billions Fiscal Revenue (Percentage of Market Income) -2 -1,000 -4 -2,000 -6 Fiscal Revenue (VND) -3,000 -8 -4,000 -10 -12 -5,000 -14 -6,000 -16 -7,000 -18 -8,000 -20 Decile 2 3 4 5 6 7 8 9 Decile Decile 2 3 4 5 6 7 8 9 Decile 1 Household per capita market income decile 10 1 Household per capita market income decile 10 Indirect Taxes (VAT+SST) Indirect Taxes (EPT) Indirect Taxes (VAT+SST) Indirect Taxes (EPT) Total Impact Total Impact 28 Figure 15. Direct transfers. Absolute impact (VND billion) Figure 16. Direct transfers. Relative impact (percent of market income) 1,000 12 Billions 900 Fiscal Expenditure (Percentage of Market Income) 10 800 700 8 600 Fiscal Expenditure (VND) 500 6 400 300 4 200 2 100 0 1 2 3 4 5 6 7 8 9 10 0 Household per capita market income decile Decile 2 3 4 5 6 7 8 9 Decile 1 Household per capita market income decile 10 Social assistance transfers Total Impact Social assistance transfers Total Impact Figure 17. Indirect subsidies. Absolute impact (VND billion) Figure 18. Indirect subsidies. Relative impact (percent of market income) 300 1.4 Billions Fiscal Expenditure (Percentage of Market Income) 250 1.2 1.0 200 0.8 Fiscal Expenditure (VND) 150 0.6 100 0.4 50 0.2 0 Decile 2 3 4 5 6 7 8 9 Decile 0.0 1 Household per capita market income decile 10 Decile 2 3 4 5 6 7 8 9 Decile 1 Household per capita market income decile 10 Electricity Subsidy Total impact Electricity Subsidy Indirect Subsidies 29 Figure 19. In-kind transfers. Absolute impact (VND billion) Figure 20. In-kind transfers. Relative impact (percent of market income) 3,500 35 Billions 3,000 30 25 2,500 20 2,000 15 1,500 10 5 1,000 0 500 1 2 3 4 5 6 7 8 9 10 -5 0 1 2 3 4 5 6 7 8 9 10 -500 Health OOP fees Education fees Education gross benefits Health gross benefits Health gross benefits Education gross benefits Education fees Health OOP fees Total In-kind benefits Total In-kind benefits Figure 21. Demographic composition of households by per capita consumption decile 30 Figure 22. The impact of fiscal policy on inequality after adjusting value of in-kind benefits for S-HCI differences 6 5 Percentage Point Reduction 4 3 2 1 0 Direct taxes and Indirect taxes and In-kind benefits Discounting Discounting Final impact transfers transfers education health Source: World Bank analysis from VHLSS 2018 and D’Souza, Gatti and Kraay 2019 Note: The figure presents marginal effects on poverty (inequality) reduction. Individual subcomponents are scaled to the marginal effect of each fiscal category. A positive bar (green)—falling down from the previous category—represents a contribution to poverty (inequality) reduction. A negative bar (red)—going up from the previous category—represents an increase in poverty (inequality). 31