FROM THE LAST MILE TO THE NEXT MILE 2022 VIETNAM P OVERTY AND EQUITY A SSESSMENT © 2022 International Bank for Reconstruction and Development / The World Bank 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 of the data included in this work. 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 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 This work is available under the Creative Commons Attribution 3.0 IGO license (CC BY 3.0 IGO) http://creativecommons.org/licenses/by/3.0/igo. Under the Creative Commons Attribution license, you are free to copy, distribute, transmit, and adapt this work, including for commercial purposes, under the following conditions: Attribution—Please cite the work as follows: World Bank. 2022. From the Last Mile to the Next Mile – 2022 Vietnam Poverty & Equity Assessment. Washington, DC: World Bank. License: Creative Commons Attribution CC BY 3.0 IGO Translations—If you create a translation of this work, please add the following disclaimer along with the attribution: This translation was not created by The World Bank and should not be considered an official World Bank translation. The World Bank shall not be liable for any content or error in this translation. Adaptations—If you create an adaptation of this work, please add the following disclaimer along with the attribution: This is an adaptation of an original work by The World Bank. Views and opinions expressed in the adaptation are the sole responsibility of the author or authors of the adaptation and are not endorsed by The World Bank. Third-party content—The World Bank does not necessarily own each component of the content contained within the work. The World Bank therefore does not warrant that the use of any third-party-owned individual component or part contained in the work will not infringe on the rights of those third parties. The risk of claims resulting from such infringement rests solely with you. If you wish to re-use a component of the work, it is your responsibility to determine whether permission is needed for that re-use and to obtain permission from the copyright owner. Examples of components can include, but are not limited to, tables, figures, or images. All queries on rights and licenses should be addressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; e-mail: pubrights@worldbank.org. Cover design: Saengkeo Touttavong From the Last Mile to the Next Mile i   FROM THE LAST MILE TO THE NEXT MILE 2022 VIETNAM P OVERTY AND EQUITY A SSESSMENT iv 2022 Vietnam Poverty and Equity Assessment  Acknowlegements The World Bank’s Poverty and Equity team in the Equitable Growth, Finance and Institutions Practice Group (EFI) led the preparation of this report. It includes the following team members: Judy Yang (Senior Economist), Matthew Wai Poi (Lead Economist), Anh Thi Bao Tran (Extended Term Consultant), England Rhys Can (Consultant), and Aisha Mohammed Abubakar (Consultant). This work is supported by the Australia – World Bank Group Strategic Partnership in Vietnam – Phase 2+ (ABP2+). The report greatly benefited from important discussions, inputs, and review provided by the following World Bank staff, external collaborators, and government agencies: ‒ Agriculture: Hardwick Tchale (Senior Agriculture Economist) ‒ Education: Nguyet Thi Anh Tran (Education Specialist), Michael Drabble (Senior Economist), and Dung Kieu Vo (Senior Education Specialist) ‒ Environment: Jun Rentschler (Senior Economist) and Rui Su (Consultant) ‒ Health: Christophe Lemiere (HD Practice Leader), Huong Lan Dao (Senior Health Specialist), and Anh Thuy Nguyen (Senior Operations Officer) ‒ Macroeconomics, Trade, and Investment: Dorsati Madani (Senior Economist), Hoang The Nguyen (Research Analyst) ‒ Poverty & Equity: Minh Cong Nguyen (Senior Data Scientist), Michal Myck (Consultant), Kajetan Trzcinski (Consultant), Monika Oczkowska (Consultant), Jon Jellema (Consultant), Cuong Nguyen Viet (Consultant), and Huong Dang (Consultant) ‒ Social Protection and Jobs: Robert Palacios (Lead Social Protection Specialist), Abla Safir (Senior Economist), and Nga Thi Nguyen (Social Protection Specialist) ‒ Social, Sustainability and Inclusion: Jose Antonio Cuesta Leiva (Lead Economist) and Sean Bradley (Lead Social Development Specialist), Helle Buchhave (Senior Social Development Specialist), and Giang Tam Nguyen (Senior Social Development Specialist) ‒ General Statistics Office: Quan The Nguyen (Deputy Director of the Socioeconomics and Environmental Statistics Department), To Thuy Hanh (Statistician, Socioeconomics and Environmental Statistics Department), and Mai Thi Thanh Nguyen (Deputy Director of the Population and Labor Statistics Department) ‒ Vietnam Academy of Social Sciences: Dr. Xuan Thanh Dang (Vice- President of VASS), Dr. Thang Nguyen (First Vice-Chairman, Socio-Economic Policy Advisory Council, Vietnam Academy of Social Sciences), Dr. Phuong Nguyen (CAF), Tam Tran (Economist, CAF), and Vu Hoang Dat (Economist, CAF) ‒ ILSSA: Ms. Thu Thi Pham (Head of Planning and International Co-operation Division Institute of Labour Science and Social Affairs (ILSSA)) ‒ IPSARD: Dr. Tran Cong Thang (Director General, Institute of Policy and Strategy for Agriculture and Rural Development (IPSARD)) ‒ UNICEF: Vincenzo Vinci (Chief Social Policy and Governance) ‒ Mekong Development Research Institute: Tung Duc Phung (Director) Report guidance was provided by: ‒ Hassan Zaman (Regional Director, East Asia and Pacific, EFI) ‒ Carolyn Turk (Country Director, Vietnam) ‒ Rinku Murgai (Practice Manager, East Asia and Pacific, Poverty and Equity Global Practice) The report is edited by Leslie Hunter and designed by Saengkeo Touttavong. We are grateful to Ngan Hong Nguyen (Senior External Affairs Officer), Anh Thi Quynh Le (External Affairs Officer), Ha Thanh Doan (Consultant) and Hoa Chau Nguyen (Program Assistant) for their contributions. From the Last Mile to the Next Mile v In addition to the large team that contributed to the report, the Poverty Assessment builds upon a large foundation of World Bank knowledge products produced over the last decade since the 2012 Poverty Assessment (Well Begun, Not Yet Done). First, several influential poverty updates have been published documenting Vietnam’s poverty reduction story. Recent reports include Climbing the Ladder (Pimhidzai, 2018), Better Opportunities for All (Pimhidzai et al., 2020), and Shared Gains (Pimhidzai and Niu, 2020). Climbing the Ladder discusses trends in economic mobility, constraints, and profiles those left behind. Better Opportunities for All zooms into the development story of rural households and discusses ways to improve their economic opportunities through better market integration. The most recent poverty update report, Shared Gains, updates our understanding of rural poverty using data from 2018. It emphasizes the need to increase off-farm opportunities by reducing distances through investments in physical, digital and human capital, lowering costs of migration, making better use of agricultural land, and increasing women’s empowerment. The report also addresses the issue of targeting the poor and assesses the strengths and weaknesses of the National Targeting Programs in Vietnam. Moreover, the conditions and development challenges of ethnic minorities are examined in Drivers of Socio-economic Development among Ethnic Minority Groups report. A series of reports also examine constraints and challenges in relation to labor, skills, and impacts from COVID-19. Skilling Up Vietnam discusses the education and skills challenges to meet labor demands of a modern market economy (Bodewig et al., 2014). The jobs challenge is updated and well discussed in Vietnam’s Future of Jobs report (Cunningham et al., 2018). While the share of wage jobs has increased dramatically, many of these off-farm jobs tend to be low in productivity, wages, and chances of advancement and will not be sufficient to achieve middle-class aspirations. More recently, a World Bank Vietnam COVID-19 report (A Year Deferred – Early Lessons and Experiences from COVID-19 in Vietnam) uses data from June 2020–March 2021 to illustrate the changing economic conditions of households and firms since the beginning of the pandemic (World Bank, 2021a). However, in Vietnam and the Southeast Asia region, COVID-19 quickly escalated in severity in April 2021. These new developments in 2021 will lead to worse outcomes than could have been predicted. Vietnam 2035: Toward Prosperity, Creativity, Equity, and Democracy (World Bank and MPI, 2016) is a sweeping examination of 2035 development goals and targets. A Systematic Country Diagnostic and an update were also completed within the last decade (World Bank, 2016, 2021b). The latest macroeconomic conditions and developments in Vietnam are discussed regularly in semi-annual Taking Stock reports as well as monthly Vietnam Macro Monitors. vi 2022 Vietnam Poverty and Equity Assessment CONTENTS Contents Acknowlegements v Currency Equivalents xvi Abbreviations and Acronyms xvi Overview 1 Introduction 1 Part I. A Decade of Significant Progress, but Last-Mile Challenges Remain 3 Part II. The Next Mile is the Road Ahead 10 Policies for the way forward 17 References 20 Notes 20 Part 1. A Decade of Significant Progress, but Last-Mile Challenges Remain 22 Chapter 1. Taking stock of welfare trends over a dynamic decade 24 1.1. Introduction: From the Last Mile to the Next Mile 25 1.2. Monetary poverty rates illustrate a decade of significant progress in poverty reduction 26 1.3. Subnational poverty trends 30 1.4. Grouping households into economic classes to monitor changes across the entire welfare distribution 35 1.5. Upward economic mobility within a lifetime for most 37 1.6. Trends across the welfare distribution 40 1.7. How new developments from COVID-19 have impacted poverty projections and inequality 43 1.8. Summary 45 1.9. References 46 1.10. Notes 47 Annex 1.1. Expenditure in the Vietnam Household Living Standards Survey 48 Annex 1.2. Global poverty measurement and Vietnam 49 Annex 1.3. Review of poverty lines 50 Annex 1.4. Distribution-sensitive poverty projections technical information 53 Chapter 2. A review of the drivers of poverty reduction 54 2.1. Changing household profiles over the last decade (2010–2020) 55 2.2. Understanding household income dynamics: the key to poverty reduction 61 2.3. Forces behind poverty reduction–decompositions 71 2.4. Summary 74 2.5. References 78 2.6. Notes 78 Annex 2.1. Chapter 2 Figures and Tables 79 From the Last Mile to the Next Mile vii Chapter 3. Challenges to reducing poverty among the remaining poor 80 3.1. Introduction – household characteristics chronically related to higher poverty rates 81 3.2. Poverty across regions 87 3.3. Ethnic minority poverty reduction 91 3.4. Agriculture incomes and livelihoods 94 3.5. Chronic Poverty: yesterday and tomorrow 103 3.6. References 104 3.7. Notes 104 Annex 3.1. Chapter 3 figures and tables 105 Annex 3.2. NTP Summary 106 Part 2. The Next Mile is the Road Ahead 108 Chapter 4. Early human captial formation and poverty reduction 110 4.1. Human Capital– a key driver of sustained growth 111 4.2. Vietnam’s performance in human capital vs regional and global peers 112 4.3. Variations in human capital formation exist within in Vietnam 114 4.4. Quantifying the sources of variation in child outcomes within Vietnam 119 4.5. Summary 126 4.6. References 129 4.7. Notes 129 Annex 4.1. Chapter 4 figures 130 Annex 4.2. Computing Dissimilarity index – Circumstances and opportunities 132 Chapter 5. Sustaining upward economic mobility 135 5.1. What will it take to build a middle-class? 136 5.2. Economic mobility for tomorrow – potential challenges adhead 138 5.3. Is the youth labor force transformating? 144 5.4. Conclusion 148 5.5. References 149 5.6. Notes 149 Annex 5.1. Chapter 5 Tables 150 Annex 5.2. Growth decomposition methology 151 Chapter 6. Building resilience and safety nets to guard against shocks 153 6.1. Introduction - Shocks, Coping, and the Need to Build Resilience 154 6.2. Households and shocks 155 6.3. What types of shocks are households more at risk from? 161 6.4. Household coping strategies 163 6.5. Social Protection programs in Vietnam today 167 6.6. Conclusion 173 6.7. References 174 6.8. Notes 174 viii 2022 Vietnam Poverty and Equity Assessment CONTENTS Chapter 7. Fiscal policy, financing future investments and building the middle class 176 7.1. Introduction – measuring fiscal impacts 177 7.2. Vietnam’s pre-COVID-19 fiscal policies: how is money raised and how is it spent? 178 7.3. Vietnam’s pre-COVID-19 fiscal incidence: who pays and who benefits? 182 7.4. Conclusion - Insights from international experience 187 7.5. References 191 7.6. Notes 192 Annex 7.1. Key assumptions in modelling tax and transfer incidence in Vietnam 193 Chapter 8. Policies for the way forward 195 8.1. Addressing a new context and Last Mile Chronic Challenges 196 8.2. Reaching Next Mile Aspirations 201 8.3. References 209 8.4. Notes 208 List of Figures Overview Figure O.1. Economic growth and poverty reduction, 1993–2020 2 Figure O.2. Issues of greatest concern, 2015–2020 2 Figure O.3. Poverty rates across different thresholds 3 Figure O.4. The number of poor, 2010–2020 3 Figure O.5. Household consumption inequality, 2010–2020 4 Figure O.6. Growth at the bottomof the distribution 4 Figure O.7. District level estimates of poverty 5 Figure O.8. Change in number of wage jobs, 2010–2020 6 Figure O.9. Number of jobs by skill level, 2010–2020 6 Figure O.10. Distribution-sensitive poverty projections for Vietnam, 2018–2023 7 Figure O.11. Poverty rates by dimensions revealing chronic poverty 9 Figure O.12. The distribution of poor varies at different thresholds of poverty 10 Figure O.13. Household expenditure on extra education courses at compulsory grade levels in public schools, by household quintile 12 Figure O.14. Variation in continuity of education across Vietnam’s regions 12 Figure O.15. Projected growth under different productivity scenarios, 1991–2045 12 Figure O.16. Projected income under different productivity scenarios, 2018–2045 12 Figure O.17. Rates of chronic insecurity and risk-induced insecurity by region, 2020 14 Figure O.18. Relative importance of idiosyncratic and covariate risk by region, 2020 14 Figure O.19. Impact of fiscal policy on inequality, 2018 16 Figure O.20. Reduction in inequality due to fiscal policy in international context 16 From the Last Mile to the Next Mile ix Chapter 1 Figure 1.1. Economic growth and poverty reduction, 1993–2020 25 Figure 1.2. Issues of greatest concern, 2015–2020 26 Figure 1.3. Poverty rates declined consistently 27 Figure 1.4. Illustration of poverty and poverty gap (LMIC poverty line) 29 Figure 1.5. Poverty rates ($3.20/day 2011PPP) , by region 31 Figure 1.6. Number of poor ($3.20/day 2011PPP) , by region 31 Figure 1.7. District level estimates of poverty 32 Figure 1.8. District level poverty comparison, 2009 vs. 2019 33 Figure 1.9. Share of districts with poverty rates above the national average, 2019 34 Figure 1.10. Distribution of economic classes (share of the population) 35 Figure 1.11. Distribution of economic classes (population in millions) 35 Figure 1.12. Population by economic class and by region, 2010–2020 36 Figure 1.13. Increases in consumption in panel households observed over two-year and four-year periods 37 Figure 1.14. Share of panel households moving up or sliding down an economic group 38 Figure 1.15. Economic class transitions over a two-year period 38 Figure 1.16. Household consumption growth across the welfare distribution 40 Figure 1.17. Comparing growth at the bottom and top of the distribution 41 Figure 1.18. Vietnam’s shared prosperity vs other East Asia countries 41 Figure 1.19. Relative inequality, Gini Index 42 Figure 1.21. Expenditure levels, 2020 42 Figure 1.20. Widening absolute inequality 42 Figure 1.22. Distribution-sensitive poverty projections for Vietnam, 2018–2023 43 Figure 1.23. Impact of changes in growth and inequality on poverty projections 44 Figure A.1.3.1. Comparison of povertyrates, by different lines 51 Chapter 2 Figure 2.1. Ageing Vietnam 55 Figure 2.2. Distribution of householdsby ethnicity, 2020 56 Figure 2.3. Education level in Vietnamese households, 2010–2020 56 Figure 2.4. Shifts in sector of main employment of Vietnamese households, 2010–2020 56 Figure 2.5. Employment levels, by sector (15+) 57 Figure 2.6. A substantial increase in wage jobs 58 Figure 2.7. More jobs and a shift from low- to medium-skilled jobs 58 Figure 2.8. Share of jobs by formality 59 Figure 2.9. Underemployment and unemployment rates rose in 2020 59 Figure 2.10. Average share of household income by source, 2020 61 Figure 2.11. Share of Vietnamese households receiving income from multiple sources, 2020 62 Figure 2.12. Trends in household income components, 2010 and 2020 64 Figure 2.13. Changes in household engagement in major sectors 65 Figure 2.14. Wage trends for selected sectors (2010–2020) 66 Figure B.2.3.1. Share of workers by selected sectors and economy 67 Figure B.2.3.2. Wages by economy and selected sectors 68 Figure 2.15. Evolution of household income from wages, 2020 vs. 2010 68 Figure 2.16. Wage trends in 2020, by household quintile 69 Figure 2.17. Family business income trends in 2020, by household quintile 70 x 2022 Vietnam Poverty and Equity Assessment CONTENTS Figure 2.18. Household-level cash transfers by household quintile 70 Figure 2.19. Percent of beneficiaries in the population, 2018 71 Figure 2.20. Growth-driven poverty reduction over the last decade 72 Figure 2.21. Inequality decompositions – urban/rural 72 Figure 2.22. Total poverty reduction 73 Figure 2.23. Poverty reduction, by urban and rural areas 73 Figure 2.24. Poverty reduction, by ethnicity 74 Figure B.2.4.1 Household consumption per capita—a decomposed identity 76 Figure A.2.1.1. Poverty reduction from2010 to 2020, decomposition results 79 Figure A.2.1.3. Household income decomposition, 2010-2020 79 Figure A.2.1.2. Poverty reduction from 2010 to 2020, decomposition results including wages by sector 79 Chapter 3 Figure 3.1. Poverty rates by dimensions revealing chronic poverty 82 Figure 3.2. Average annualized household consumption growth, by region 82 Figure 3.3. Populations by region, 2010-2020 83 Figure 3.4. Population distribution by region, 2020 83 Figure 3.5. Economic class distribution, by region and ethnicity, 2020 84 Figure 3.6. Poverty rates amongethnic minority households 84 Figure 3.7. Poverty rates among Kinh households 84 Figure 3.8. Population profiles by groups 86 Figure 3.9. Small area poverty estimates (2019), regional zoom-ins 87 Figure 3.10. Share of households receiving labor market income (2010 and 2020), by source and region 89 Figure 3.11. Average household labor market income (2010 and 2020), by sources and region 89 Figure 3.12. Poverty rates by geography 90 Figure 3.13. Share of provincial population that is ethnic minority 92 Figure 3.14. Share of workers employed by ethnicity and in the manufacturing sector, 2010 to 2020 93 Figure 3.15. Kinh-ethnic minority absolute income gaps across broad categories 93 Figure 3.16. The number of poor (in millions), by main household agricultural activity, 2020 97 Figure 3.17. Farmers are aging, particularly among Kinh 97 Figure 3.18. Participation in agricultural activities over time (2010–2020), national level 98 Figure 3.19. Participation in agricultural activities over time (2010–2020), by region 98 Figure 3.20. Household farm income by activity (2010–2020), national level 99 Figure 3.21. Household farm income by activity (2010–2020), by region 99 Figure 3.22. A comparison of household agricultural participation - Kinh vs. ethnic minorities 100 Figure 3.23. Crop productivity - Kinh vs. ethnic minorities 101 Figure 3.24. A comparison of household agricultural income - Kinh vs. ethnic minorities 102 Figure A.3.1.1. Wage trends by job type, Kinh vs. ethnic minorities 105 Chapter 4 Figure 4.1. Poverty rates by household education group 111 Figure 4.2. Learning-adjusted years of schooling by GDP per capita, Vietnam, and comparators, 2020 113 Figure 4.3. Fraction of children not stunted by GDP per capita, Vietnam, and comparators, 2020 114 Figure 4.4. Survival rates by GDP per capita,Vietnam, and comparators, 2020 114 Figure 4.5. Gap in socioeconomic-disaggregated HCI, 2013 115 Figure B.4.1.1. Framework for calculating the Human Capital Index 115 Figure 4.6. Enrolment rates by householdwealth quintile and age, 2020 116 Figure 4.7. Access to improved drinking-water sources and sanitation facilities, by income quintiles, 2014 116 Figure 4.8. School Attendance by age and gender, 2020 117 From the Last Mile to the Next Mile xi Figure 4.9. Enrolment rates by ethnicityand age, 2020 118 Figure 4.10. Access to improved drinking-water sources and sanitation facilities, by ethnicity, 2014 118 Figure 4.11. Dissimilarity index in education and health 119 Figure 4.12. Shapley decomposition of the dissimilarity index in education and health outcomes 121 Figure 4.13. Dissimilarity index in bundled opportunities 122 Figure 4.14. Equity in access is improving (dissimilarity in 2010 and 2020) 123 Figure 4.15. Continuity of education across regions and socioeconomic groups 124 Figure 4.16. Expenditure of wealthier families on compulsory education (primary and lower secondary) 125 Figure 4.17. Participation rates in extraclasses by household wealth and age 125 Figure 4.18. Persistence and relative mobility by cohort 126 Figure A.4.1.1. HCI benchmarking, Vietnam vs. peers 130 Chapter 5 Figure 5.1. Average labor productivity growth (GDP per worker), by sector 136 Figure 5.2. Employment shares and labor productivity by sector, 2002-2030 137 Figure 5.3. Projected growth (baseline and target scenario), 1991–2045 137 Figure 5.4. Projected income, 2018–2045 137 Figure 5.5. Household consumption per capita by household head occupation skill level 138 Figure 5.6. At its level of development, Vietnam is older than its peers 139 Figure 5.7. Potential workers, 1991–2050 139 Figure 5.8. Growth decomposition (varying demographic scenarios), 1991–2018 139 Figure 5.9. Reasons for remaining out of the labor force, by gender and cohort 140 Figure 5.10. Sectors having jobs with contracts remainconcentrated in the public sector and manufacturing (2010–2020) 141 Figure 5.11. Skill level of jobs, 2010 and 2020 141 Figure 5.12. Employment across digital occupation level, by country 142 Figure 5.13. Employment byoccupation skill category (share) 143 Figure 5.14. Share of workers by skill category and education in 2010-2020 143 Figure 5.15. Distribution of occupations among tertiary education workers, by skill and age groups 145 Figure 5.16. Number of workers, by occupation and birth cohort 145 Figure 5.17. Occupation skill intensities, by age cohort 146 Figure 5.18. Concentration of manufacturing jobs compared to othercountries in the region (15–24-year-old cohort) 147 Figure 5.19. Comparison of HCI and fullutilization-adjusted HCI, 2020 147 Figure 5.20. Decline in HCI due to ull utilization adjustment 148 Figure 5.21. Returns to education and skills over time 148 Chapter 6 Figure 6.1. Percent of households reporting worsening living conditions compared to five years ago 154 Figure 6.2. Top three reasons why conditions are same or worse than five years ago 155 Figure 6.3. Out-of-pocket healthexpenditure remains high 157 Figure 6.4. Catastrophic health expenditurehas been declining slightly (% of households) 157 Figure 6.5. Environmental conditions and risks by district 159 Figure 6.6. Share of districts with a high environmental risk factor, by district-level poverty rates 160 Figure 6.7. Poverty and environmental risk maps 162 Figure 6.8. Rates of chronic insecurity andrisk-induced insecurity by region, 2018 163 Figure 6.9. Relative importance of idiosyncraticand covariate risk by region, 2018 163 Figure B.6.1.1. Stylized examples of chronic insecurity and risk-induced insecurity 165 Figure 6.10. Use of financial services has improvedon some dimensions and less in others 167 xii 2022 Vietnam Poverty and Equity Assessment CONTENTS Figure 6.11. Borrowing from family andfriends is the most common source 167 Figure B.6.2.1. Vietnam had the second-largest relative gap between household losses and support, East Asia and Pacific 168 Figure 6.12. Social assistance benefit levels in Vietnam compared to other countries 170 Figure B.6.3.1. Key Elements of Vietnam’s Social Protection System 170 Figure 6.13. Vietnam lags other countries shifting to digital payments 172 Chapter 7 Figure B.7.1.1. Definition of income concepts and the role of fiscal instruments 177 Figure 7.1. Demographic composition of households by per capita consumption decile 180 Figure 7.2. Impact of fiscal policies by per capita consumption decile (excluding in-kind spending) 182 Figure 7.3. Impact of fiscal policies by per capita consumption decile (including in-kind spending) 183 Figure 7.4. The impact of fiscal policy on poverty by fiscal instrument 183 Figure 7.5. The impact of fiscal policy on inequality by fiscal instrument 184 Figure 7.6. Magnitude of different fiscal instruments and their cost-effectiveness at reducing inequality 185 Figure 7.7. Socioeconomic-disaggregated Human Capital Index (S-HCI) by component 186 Figure 7.8. The impact of fiscal policy on inequality after adjusting value ofin-kind benefits for S-HCI differences 186 Figure 7.9. Impact of fiscal policy on poverty by country income level 187 Figure 7.10. Impact of fiscal policy on inequality by country income level 189 Figure 7.11. Tax mix by country income level, direct and indirect taxes 190 Figure 7.12. Reduction in inequality from cash and non-cash fiscal instruments 191 Chapter 8 Figure B.8.2.1. Statistical Performance Index scores, 2019 199 Figure 8.1. Addressing the social protection coverage gap 205 List of Tables Chapter 1 Table 1.1. Poverty rates (2010–2020) 28 Table 1.2. Number of poor (2010–2020) 28 Table 1.3. Average poverty gap 29 Table 1.4. Summary of a decade of small area poverty estimates 32 Table B.1.3.1. Evolution of small area poverty estimation techniques 34 Table B.1.4.1. Summary of economic class definitions 36 Table 1.5. Economic mobility transitions from 2016 to 2018, by household characteristics 39 Table B.1.5.1. Number of panel households (over two-year periods) 39 Table A.1.2.1. Conversion rates 49 Table A.1.3.1. Median value of national poverty lines, by economy income classification (circa 2011 in PPP terms) 51 Table A.1.3.2. Conversions of the global poverty lines to Vietnamese dong 51 Table A.1.3.3. A comparison of poverty lines and monetary thresholds (VND per month per capita, ‘000) 52 From the Last Mile to the Next Mile xiii Chapter 2 Table 2.1. Population, 2009–2019 55 Table B.2.2.1. Income categories and definitions 63 Table 2.2. Social assistance beneficiaries 70 Table B.2.4.1. Variables and definitions 77 Chapter 3 Table B.3.2.1. Description of agricultural activities 96 Table 3.1. Agricultural activities - Kinh vs. ethnic minorities 100 Table 3.2. Crop yield and total factor productivity 101 Table A.3.1.1. Key household characteristics across main activity 105 Table A.3.1.2. A Summary of National Targeting Programs 106 Chapter 4 Table 4.1. Human Capital Index - Vietnam vs. peers (2010 vs. 2020) 112 Table 4.2. Human Capital Index disaggregated by gender 117 Table B.4.3.1. Decompositions of math test scores 128 Table A.4.1.1. Shapley decomposition of dissimilarity index 131 Table A.4.2.1. List of circumstances (potential sources of inequality) 132 Table A.4.2.2. Vietnam’s General Education System 133 Chapter 5 Table A.4.2.3. List of opportunities 134 Table A.5.1.1. Skill categorization of occupations and expected education level of workers 150 Chapter 6 Table 6.1. Environment indicator, definitions 160 Chapter 7 Table 7.1. Vietnam government revenues, 2018 179 Table 7.2. Vietnam central government expenditures, 2018 180 Chapter 8 Table 8.1. A new context and Last Mile policy summary 196 Table 8.2. Next Mile policy summary 201 xiv 2022 Vietnam Poverty and Equity Assessment CONTENTS List of Boxes Chapter 1 Box 1.2. Foster, Greer, and Thorbecke (FGT, 1984) class of indicators 30 Box 1.3. A history of small area poverty estimates 34 Box 1.4. World Bank East Asia and Pacific regional economic class groups 36 Box 1.5. VHLSS Panel sub-sample 39 Chapter 2 Box 2.1. Setting the growth foundation in earlier decades 60 Box 2.2. Measuring household income using the Vietnam householdliving standards survey 63 Box 2.3. Labor and wages, Vietnam vs. other EAP economies 67 Box 2.4. Group and income decompositions 75 Chapter 3 Box 3.1. Gender, ethnicity and the job market in Vietnam 95 Box 3.2. Categories of household agricultural activities in the VHLSS 96 Chapter 4 Box 4.1. World Bank’s human capital index 115 Box 4.2. The Association between socioeconomic characteristics and school quality 127 Box 4.3. Decomposing variation in education quality 128 Chapter 6 Box 6.1. Modelling risk and vulnerability among households 164 Box 6.2. Vietnam’s early COVID-19 response 168 Box 6.3. Vietnam’s Social Protection System 170 Chapter 7 Box 7.1. The Commitment to Equity methodology 177 Box 7.2. Limitations to CEQ methodology in Vietnam 181 Chapter 8 Box 8.1. Gender Dimensions in poverty reduction and inclusion efforts 199 Box 8.2. Data for better lives in Vietnam 199 From the Last Mile to the Next Mile xv Currency Equivalents Exchange rate, 2020 average Currency unit = Vietnamese dong (VND) US$1 = 23,208.37 dong Abbreviations and Acronyms ADB Asian Development Bank Information System ALMP Active Labor Market Program LUR Land use right ASEAN Association of Southeast Asian Nations MOET Ministry of Education and Training ASPIRE Atlas of Social Protection Indicators of MOF Ministry of Finance Resilience and Equity MOH Ministry of Health CEQ Commitment to Equity MOLISA Ministry of Labor, Invalids and CIT Corporate Income Tax Social Affairs EAP East Asia and Pacific MPI Multidimensional Poverty Index EET Exempt-Exempt-Taxed MPI Ministry of Planning and Investment EM Ethnic Minority MPS Ministry of Public Security EPT Environmental Protection Tax MPSARD Master Plan for Social Assistance ESC Employment Service Center Reform and Development ETS Emissions Trading System NDC Nationally Determined Contribution FDI Foreign direct investment NLSA National Large-Scale Assessment FGT Foster-Greer-Thorbecke NRD New Rural Development FLFP Female Labor Force Participation NTP National Targeting Program G2P Government-to-person NTPRP National Target Poverty GDP Gross domestic product Reduction Program GIC Growth incidence curve OECD Organisation for Economic Co- GNI Gross national income operation and Development GSO General Statistics Office O-NET Occupational Information Network HCI Human Capital Index OOP Out-of-pocket HCMC Ho Chi Minh City OPHI Oxford Poverty and Human HIV Human immunodeficiency virus Development Initiative HOI Human Opportunity Index PAPI Public Administration ICT Information and communication Performance Index technologies PISA Programme for International Student ILO International Labor Organization Assessment IO Input-output PIT Personal income tax IPL International poverty line PL Poverty line ISAS Integrated Social Assistance System PPP Purchasing power parity ISCED International Standard Classification PREM Poverty Reduction and Economic of Education Management LCU Local currency unit R&D Research and development LFS Labor Force Survey RISE Research on Improving Systems LIC Low-income country of Education LMIC Lower-middle-income country SA Social assistance LMIS Logistics Management SCOLI Spatial Cost of Living Index xvi 2022 Vietnam Poverty and Equity Assessment  SDRSP Sustainable Poverty Reduction Support Program SEDEMA Socio-economic Development in Ethnic Minority Areas SEDP Socio-Economic Development Plan SEDS Social Economic Development Strategy SES Socioeconomic status SME Small and medium enterprise SNG Subnational government SPR Sustainable Poverty Reduction STC Special Tax Code STEP Skills Toward Employment and Productivity TFP Total factor productivity TIMSS Trends in International Mathematics and Science Study TTE Taxed-Taxed-Exempt TVET Technical and vocational education and training U-HCI Underutilization-Adjusted Human Capital Index UMIC Upper-middle-income country UNDP United Nations Development Programme UNESCO United Nations Education, Scientific and Cultural Organization USD United States dollars VAT Value-added tax VHLSS Vietnam Household Living Standards Survey VND Vietnamese dong VSS Vietnam Social Security WB World Bank WDI World Development Indicators WHO World Health Organization WTO World Trade Organization From the Last Mile to the Next Mile xvii xviii 2022 Vietnam Poverty and Equity Assessment Overview Introduction Vietnam is a country on the move and in transition. Concerns over poverty amid high economic growth are Indicators are pointing in the right direction, with many not inconsistent; together they illustrate an absolute and positive economic and social developments. The amount of inclusive rise in living standards, but also a population progress that Vietnam has achieved in less than half a century that seeks economic security and aspires for more. Rapid since the war ended in 1975 has been nearly without parallel. developmental changes left some behind who did not have At the same time, Vietnam is a lower-middle-income country the opportunity to join the most dynamic sectors of the facing a challenging and uncharted road ahead to reaching economy, and also created a large segment of the population upper-middle and high-income country levels in a shifting that are no longer poor but not yet rich. About 85 percent global economic and climatic landscape. of households report that their living conditions got better in 2020 compared to 2016 (VHLSS). On the other hand, In less than half a century since the end of the Vietnam 63 percent of households felt their economic conditions were War and 35 years since the Doi Moi reforms, Vietnam has better in 2018 than what they were five years ago, according become a vibrant economy and a sought-after market to to UNDP PAPI data. Thus, household perceptions on changes the outside world. GDP per capita (2015 US$) grew from in economic conditions are slightly more pessimistic than $481 in 1986 to $2,655 in 2020. Rapid economic growth perceptions on changes in living conditions; in both cases the was broadly inclusive, and livelihoods in Vietnam improved degree of perceived improvement was more minor than major.2 dramatically. Based on the World Bank’s Lower-Middle Income Class ($3.20/day 2011PPP) poverty line, poverty rates declined from 16.8 percent in 2010 to 5.0 percent in 2020 (Figure O.1). T h ese co n cern s an d According to preliminary estimates for 2019, 5.7 percent of a sp iratio n s refl e c t th e Vietnamese are considered multidimensionally poor (GSO). n e e d to sim u ltan e o u sly ad d ress Last Mile ch ro n ic At the same time, despite remarkable progress, poverty p overty ch allen ges, remains a key concern among the population. In a survey w h ile en su rin g th ere are of citizens, from 2015 to 2020 poverty/hunger was selected as su stain ab le e co n o m ic the main issue that the government must tackle (Figure O.2). m ob ility p ath w ays to fu lfi ll When asked why poverty was the main concern, while many th e co u n try ’s Next Mile people worried about falling back into poverty, even more felt u p p er-m id d le an d h igh - that poverty is an overall drag on the economy and reduces in co m e a sp iratio n s national prestige (based on data from the 2018 UNDP PAPI). Half of those who said that hunger and poverty were their top concern earned more than 8 million VND per month,1 highlighting concerns over economic security, even among higher-earning individuals. From the Last Mile to the Next Mile 1 The Vietnam Poverty and Equity Assessment is organized into standards and provided better economic opportunities for large two parts motivated by addressing both Last Mile and Next numbers of youth entering the labor force. But poverty rates are Mile issues: still much higher among rural and ethnic minority families and households primarily engaged in agriculture (Chapter 3). These Part I reviews poverty and inequality trends over the groups are faced with persistent challenges of lower human last decade, 2010–2020. The decade saw high growth, great capital, lower-quality local public services, longer distances success in poverty reduction, and household profiles shifting to economic opportunities, and less access to financing and toward higher levels of education and more people working training. Policy is well aligned to tackle Last Mile chronic outside of agriculture (Chapters 1 and 2). Higher wages and poverty; three NTPs will be implemented during the 2021– the creation of off-farm jobs largely drove improved living 2025 Socioeconomic Development Phase. Figure O.1. Economic growth and poverty reduction, 1993–2020 100 3,000 90 2,500 GDP per capita (constant 2015 US$) 80 70 2,000 Poverty rate (%) 60 50 1,500 40 1,000 30 20 500 10 0 0 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 Extreme poverty ($1.90/day 2011PPP) LMIC PL ($3.20/day 2011PPP) UMIC PL ($5.50/day 2011PPP) GDP per capita (constant 2015 US$) Note: Poverty measurement methodology changed in 2010, rendering a break in series comparability. Source: World Bank WDI, PovcalNet. Figure O.2. Issues of greatest concern, 2015–2020 30 25 Percent of respondents 20 15 10 5 0 2015 2016 2017 2018 2019 2020 Poverty/Hunger Economic Growth/GDP Environment Health/Insurance Source: CECODES, VFF-CRT, RTA & UNDP (2021). 2 2022 Vietnam Poverty and Equity Assessment Overview Part II assesses opportunities for and challenges to adequately protect all households against all risks and should Vietnam’s path to achieving its Next Mile aspirations be modernized (Chapter 6). At the same time, investments in and creating greater prosperity for households and better skills and higher-quality education, as well as a stronger workers. During the 13th National Congress in 2019, the safety net which protects all, will require public investment. government announced its goal to become a high-income How this investment can be financed and how fiscal policy country by 2045. This is an ambitious target befitting a in general can drive prosperity more broadly is the subject country with a proven track record of inclusive growth, but of Chapter 7. Lastly, policy recommendations covering both it requires new approaches and even greater achievements. Last Mile and Next Mile agendas are presented in Chapter 8. The Next Mile is the road to upper-middle and high-income country standards. What will be needed to sustain the upward economic mobility of the millions who have left poverty and Part I. A Decade of now seek greater economic security in the face of new risks and challenges? Equitable human capital formation is a key Significant Progress, ingredient to Asia’s economic success. While completion but Last-Mile Challenges gaps are narrowing, some still remain for the most vulnerable Remain groups (Chapter 4). To achieve high-income status by 2045, real GDP will have to grow by nearly 7 percent per annum. A decade of continuous poverty reduction but Of those entering the work force, today’s youth are better inequality is rising educated, but are they ready to lead a high-skill and high- productivity transition and grow the middle class? Some Over the last decade, poverty declined impressively. indicators point to challenges to such a transition without On average, household consumption growth was strong at continuing reforms and improvements in education and skills about 5 percent per year. Based on the World Bank’s lower- (Chapter 5). Moreover, in a world with unpredictable risks middle income country (LMIC) poverty line ($3.20/day and challenges, promotion policies to boost upward economic 2011PPP), poverty declined from 16.8 in 2010 to 5 percent mobility must be complemented by protection strategies to in 2020 (Figure O.3).3 This equates to ten million people preserve gains. The current social protection system does not having moved out of poverty, reducing the number of poor to Figure O.3. Poverty rates across Figure O.4. The number of poor, 2010–2020 different thresholds 25 15 20 12 Number of poor (millions) Poverty rate (%) 15 9 10 6 5 3 0 0 2010 2011 2012 2013 2015 2016 2017 2018 2019 2020 2014 2010 2012 2016 2018 2020 2014 General poverty rate Multidimensional poverty rate World Bank LMIC poor ($3.20/day 2011PPP) GSO-WB poverty rate World Bank LMIC poverty rate Note: See Chapter 1 for definitions of the various lines. Source: GSO and World Bank. Source: World Bank staff calculations. From the Last Mile to the Next Mile 3 5 million in 2020 (Figure O.4). Growth over the last decade in A decade of district-level poverty trends Vietnam was broadly inclusive, as welfare improved across the illustrates more progress near economic centers entire economic distribution of households. Pathways out of than in remote areas poverty were widespread across most groups, as many families moved out of poverty within a single generation. Some groups District-level poverty maps produced over a decade shows had greater chances of climbing up to a higher economic class overall progress, but pockets of chronic poverty remain. group as well as lower rates of sliding down; these groups left The 2009 poverty map reveals high concentrations of poverty agriculture completely or gained formal employment over a throughout the high mountainous areas in the north and two-year period. central regions of the country (Figure O.7). Over the decade, poverty has declined impressively but has remained in lagging Inequality, on the other hand, increased marginally in the areas in the northwest and central highlands. The region that has latter half of the decade. The Gini index was at its highest achieved the largest absolute poverty reduction is the northeast, at the beginning of the decade, coinciding with the global which has benefited from expanding industrial activity. financial crisis, but quickly reverted (Figure O.5). Declining inequality was consistent with households at the lower ends Job creation and rising wage income were the of the welfare distribution experiencing higher-than-average main drivers of poverty reduction, but these growth rates from 2010 to 2014 (Figure O.6). In the latter half channels were interrupted by COVID-19 of the decade, however, consumption among richer households grew at faster rates than among poorer ones, resulting in a The patterns of poverty reduction over the last decade negative shared prosperity premium4 and increasing inequality. are strongly linked to economic growth. Rising wages, Over the decade, average household consumption Since the an increasing share of formal employment, and movement poorest households are increasingly concentrated in low-income out of low-productivity agriculture raised labor incomes. agricultural activities, these groups risk falling further behind and Due to a demographic dividend—the large young generation being less connected to the more vibrant sectors of the economy. entering working age—the labor force expanded by 4.5 million net workers from 2010 to 2020, even accounting for large exits from agriculture. Overall, the manufacturing and services sectors increased by 5.8 and 4.8 million net workers, respectively. Agricultural employment declined only in the Figure O.5. Household consumption Figure O.6. Growth at the bottom inequality, 2010–2020 of the distribution 8 45 7 40 6 5 Annualized growth rate (%) 35 4 Gini index 30 3 2 25 1 0 20 2010–12 2012–14 2014–16 2016–18 2018–20 2010 2012 2014 2016 2018 2020 Growth B40 Growth T60 Growth Mean Source: World Bank, PovcalNet. Source: World Bank staff calculations using VHLSS 2010–2020. 4 2022 Vietnam Poverty and Equity Assessment OvErviEw Figure O.7. District level estimates of poverty 2009 2015 2019 Poverty rate (%) 0.0 100.00 Note: Poverty rates based on the poverty line. Source: World Bank staff and GSO calculations using Census and VHLSS. latter half of the decade, with its workforce declining from sustained growth to date. However, with a much greater share 24.5 million to 17.7 million workers from 2015 to 2020. of care responsibilities due to social norms, women have less freedom to pursue more labor-intensive occupations, while Importantly, the new jobs created were mostly better family business income for women is lower than for men than those for generations before (Figure O.8). In the first because of fewer hours worked. The types of jobs available half of the decade, employment grew rapidly in the services to women differ from those for men, and even where men sector, but this was largely in unskilled occupations (Figure and women are eligible from both legal and sociocultural O.9). Patterns shifted halfway through the decade when standpoints, employers tend to prioritize male candidates in more medium-skilled jobs in services and manufacturing job advertisement. were created. Foreign direct investment was important in transforming jobs since most jobs in foreign-owned enterprises The increase in household wage income was substantial, are formal with higher wages compared to domestic firms. increasing by nearly 100 million VND in nominal terms, The services sector was also invigorated, as the number of or nearly tripling over the decade. Household wages international visitors increased from 5 million in 2010 to 18 also tripled in the lowest quintile, but the absolute increase million in 2019, most of them tourists. Wage growth was also amounted to just over 50 million VND. Incomes from family faster in the first half of the decade than the second, coinciding businesses increased by similar levels, but participation in with slower poverty reduction in the latter period. family businesses was less common among poorer households. For example, 16 percent of households in the bottom quintile From a gender perspective, job opportunities have been earn income from family businesses compared to 73 percent mostly inclusive. The rate of women working in Vietnam earning wage incomes in 2020. The share of households is 73 percent, much higher than the regional average and receiving wage income increased from 64 to 70 percent over averages of lower-middle, upper-middle, and high-income the decade. In particular, participation in medium-skilled and countries, which has contributed significantly to the non-farm wage employment increased, providing households From the Last Mile to the Next Mile 5 with more predictable and higher incomes. In the poorest were more likely to experience reduced working hours or to decile, the share of households engaged in manufacturing quit their jobs altogether. increased from 26.7 percent in 2010 to 35.8 percent in 2020. There was a comparable increase in households working in COVID-19 will lead to setbacks in poverty services overall (9.4 to 15.2 percent), but less of an increase reduction and widen inequality across both among poor households. monetary and non-monetary dimensions The sudden emergence of COVID-19 at the end of the The COVID-19 pandemic has set progress in poverty decade halted wage growth and improvements in job reduction back. In Vietnam, COVID-19 halted a decade quality. By the end of 2020, the spread of COVID-19 was of rapid wage growth, although progress in reducing poverty still manageable, but the labor market nonetheless experienced was sustained in 2020 over a two-year period compared to a rise in informality and underemployment.5 By the fourth 2018. This was a better outcome than for most developing quarter of 2020, about 830,000 people were estimated to countries in the region and globally (World Bank, 2020a). be underemployed, with the highest rates among those in Some impacts were still hard felt, as poverty rates among the the agriculture sector, their ranks swollen with returning Kinh majority and in urban areas increased slightly in 2020 manufacturing and services workers whose jobs in the cities had compared to 2018. Furthermore, the emergence of the Delta disappeared or been put on hold. The informality rate at the variant in mid-2021 led to further setbacks. Based on updated end of 2020 was 56.2 percent, an uptick following a continuous 2021 growth forecasts after the arrival of Delta, poverty rates decline in informality from 2016 to 2019 (GSO, 2021). Official are expected to be about 0.5 percentage points higher, and statistics also reported that 9.1 million workers (12.8 percent poverty reduction will likely slow in 2021 under a scenario of all workers) had either lost their jobs or had reduced wages with a 1 percent increase in inequality (Figure O.10). in the first quarter of 2021, and average labor incomes were 2.3 percent lower compared to the previous year (Ha and Minh, COVID-19 highlighted pre-existing inequalities and 2021). Unemployment rates also rose, more so for women. differences in coping and adaptation. Even before the COVID-19 has taken an outsize toll on female participants onset of the COVID-19 pandemic, new signs of widening in the labor force, many of whom are in the informal services inequality were emerging. The absolute difference in annual sector, which was hardest hit by the COVID-related lockdowns. per capita consumption between the poorest and richest 10 Due to school closures and caregiving responsibilities, women percent of the populations increased from VND 48.5 million Figure O.8. Change in number of Figure O.9. Number of jobs by wage jobs, 2010–2020 skill level, 2010–2020 5 4 Net change in employment (millions) Net change in employment (millions) 4 3 3 2 2 1 1 0 0 −1 −1 −2 −2 −3 −3 −4 2010–14 2015–20 2010–14 2015–20 Wage worker Own account worker High-skill Medium-skill Low-skill Employer Unpaid family worker Source: GSO, Labor Force Survey. Source: GSO, Labor Force Survey. 6 2022 Vietnam Poverty and Equity Assessment Overview Figure O.10. Distribution-sensitive poverty projections for Vietnam, 2018–2023 No change in Gini +1% change in Gini +2% change in Gini 6.6 6.6 6.6 Poverty rate (%, $3.20/day 2011PPP) 6 5.2 4.9 4.9 4.7 4.6 5.0 5.0 4.3 5.0 4.0 4.7 4 4.4 4.4 4.2 3.7 4.1 3.4 3.9 3.5 3.5 3.1 2 0 22.4 22.4 22.4 Poverty rate (%, $5.50/day 2011PPP) 20 19.2 18.1 18.5 18.2 18.8 18.8 18.8 17.0 16.2 17.8 17.0 17.2 16.9 16.9 15.1 15.9 15 15.5 14.9 14.1 14.3 13.2 10 2018 2020 2022 2024 2018 2020 2022 2024 2018 2020 2022 2024 Growth estimates (version) March 2021 growth estimates October 2021 growth estimates Note: Poverty rates in 2018 and 2020 are based on survey data. Poverty rates in 2021 and onwards are projections. Simulations of changes in inequality, if any, are assumed to start in 2021. Distributionally sensitive simulations methods are based on Lakner et al. (2020). See Annex 1.3 for details. Source: World Bank staff estimations using VHLSS, 2018, 2020. in 2010 to VND 123.8 million in 2020. Over the last half continuity of education was uneven during COVID-19, and decade, the household consumption growth of the bottom 40 the pandemic has potentially widened gaps in human capital percent of the population was lower than the national average. formation because of the unequal capacity of schools across Education outcomes in Vietnam vary by socioeconomic the country. Jobs of the future will require greater digital status,6 and progress in reducing stunting had also stagnated. skills, but there are gaps in digital use and inclusion. Wealthier Women, those in the informal sector, and households in households are better able to participate in the digital economy the bottom 20 percent experienced the slowest recovery in both as sellers and as buyers on digital platforms. In the longer household income between June 2020 and March 2021 term, COVID-19 has widened inequality and disparities in (World Bank, 2021a). In terms of coping during COVID-19, opportunities, potentially impacting future economic growth poor households were more reliant on external sources such while limiting upward mobility for many. as borrowing, while rich households were better able to cope through their own means such as tapping into savings. Chronically higher poverty rates among certain groups is a Last Mile challenge –but some Inequality is expected to increase during COVID-19 positive trends have emerged for a range of reasons. Women bear a larger share of care responsibilities, and their labor market activities were more There is a persistent interplay between ethnicity, adversely impacted than men’s. Informal workers have agriculture, geography, and high rates of poverty in the least access to safety nets and experienced the greatest Vietnam. Remote mountainous areas where the poor are challenges when seeking government cash support. The concentrated are heavily populated by ethnic minorities, who From the Last Mile to the Next Mile 7 are also disproportionately more engaged in agriculture than NTPs are a long-standing policy instrument that the Kinh majority. These remote areas are located far from still play a role in poverty reduction economic centers and, partly due to unfavorable topology, agriculture is also less productive there. Among the remaining Chronic dimensions of poverty are well known, and poor in 2020, these groups are overrepresented; 79 percent assistance to groups with high poverty is well reflected are ethnic minorities (just 15 percent in the population), and in policy priorities. In the 2021–2025 SEDP phase, three 66 percent are engaged solely in agriculture (16 percent in National Targeting Programs (NTPs) are aimed at individuals, the population). The share of the national population in the households, and geographic areas across the dimensions Central Highlands and Midlands and Northern Mountains highlighted in this chapter (rural, agricultural, and ethnic regions is 6 and 13 percent, respectively, but these regions are minorities). NTPs provided high levels of investments to home to 21 and 42 percent of the poor. communes, with nearly VND 560 trillion (approximately $25 billion) of this for commune level programs under NTPs At the end of the decade, the poverty rate of ethnic from 2010 to 2019. minorities was still higher than those of the Kinh majority at the start of the decade, although the gaps are The design, targeting, and financing allocations of NTPs closing (Figure O.11). Kinh poverty rates were low in 2010 could be further refined to improve impact and results. at 9 percent and had fallen to near zero by 2020, while poverty Based on a survey of NTP- National Rural Development (2016- among ethnic minorities was still 27 percent. However, the 2020) projects and implementation, NTP-NRD provided high absolute gap in poverty had narrowed considerably, from 47.4 commune-level investments, but a smaller share went to the percentage points in 2010 to 26 percentage points in 2020, as poorest communes because these were more reliant on direct minority poverty rates have declined from 57 percent in 2010. central government transfers and had more limited external financing options (Pimhidzai and Niu, 2020). Much of the Despite persistently high poverty rates among some NTP spending was devoted to socio-economic infrastructure groups, there are new developments and progress for investments rather than spending on other basic social services. some. Ethnic minority workers are shifting into manufacturing However, lifting the remaining poor to the poverty line can be work, with their share in manufacturing increasing to 23 efficient under well-targeted household-level programs. For percent by 2020, similar to the rate of Kinh in manufacturing in instance, in 2020, the monetary gap to lift all poor7 above the 2010. However, geographic mobility remains limited, with the poverty line was 15 trillion VND. Thus, the cost of sufficient distribution of the ethnic minority population across regional cash to the poor to raise them to the poverty line can be smaller and urban divides largely remaining constant over the decade. compared to the overall NTP budget. Some regions exhibited lower increase in income growth, Beyond the poor, a more diverse share of the particularly in the Central Highlands region, where population is economically vulnerable wage employment with contracts is the least common. Moreover, this was the only region where household economic Rapid economic growth has lifted many Vietnamese out participation declined in wage jobs, family business, and family of poverty, but a large group of the population remains farms 2010 to 2020. The Midlands and Northern Mountains economically vulnerable. The rapid speed of developmental region, on the other hand, saw strong growth in the share of change nonetheless left behind those with less opportunity to the households entering wage employment. These income join the most vibrant sectors of the economy and created a large dynamics are consistent with recent developments in regional class of people that are not poor but also not yet middle class. poverty rates in 2020. Slow progress in the Central Highlands Thus, the poverty and equity agenda going forward is not just region has resulted in re-ranking, where it is now poorer than about raising minimum living standards and tackling chronic the Midlands and Northern Mountains, if by only a small poor; it is also about creating new and sustainable economic margin. The Mekong Delta region saw higher poverty rates in pathways for a more aspirational population and guarding these 2020 than 2018, a result of severe droughts and disruption to hard-won economic gains from shocks or crises. Furthermore, the agriculture sector. sustaining upward economic mobility at higher income levels has been challenging. While the risk of falling into extreme 8 2022 Vietnam Poverty and Equity Assessment Overview poverty is now low, achieving economic stability at higher aspiring to higher standards of living. For the 2021–2025 levels is still a relevant concern. In 2016, nearly 40 percent of SEDP period, MOLISA significantly raised the monetary the middle class slid to a lower economic group by 2018. threshold line for the classification of poor and near-poor groups (1.5 and 2 million VND per month, respectively), in As the country aspires to greater prosperity, a higher recognition of higher aspirations and better living standards. minimum standard of living is warranted. The current These higher monetary thresholds are close to the UMIC national monetary poverty lines represent the concept of PL ($5.50/day 2011PPP, or 1.8 million VND per month). minimum basic needs—a bare subsistence level—and not a At this poverty line, poverty is higher at 18.8 percent, or higher minimum level of consumption befitting a country 18.3 million people. Figure O.11. Poverty rates by dimensions revealing chronic poverty Region Household engagement in agriculture 60 60 50 50 39.9 Poverty rate (%) Poverty rate (%) 40 40 34.0 35.2 32.8 29.6 29.1 29.3 30 25.8 30 23.4 23.5 22.7 27.6 22.0 21.3 18.6 17.1 20 20 13.6 20.4 19.7 14.0 14.3 12.1 16.0 9.0 10.1 7.8 10 11.5 10 8.6 4.8 6.8 6.2 4.7 4.9 4.3 4.7 3.6 4.5 2.3 5.3 1.4 2 3.7 1.1 6.9 3.9 0.7 1.0 0 3.0 2.3 0 0.4 0.4 1.2 0.6 2020 2020 2010 2012 2016 2018 2010 2012 2016 2018 2014 2014 Central Highlands Northern & Coastal Central No agriculture Mekong Delta Red River Delta Some agriculture Midlands & Northern Southeast Only agriculture Mountains Urban vs rural Ethnicity 60 60 56.8 50 50 47.7 46.8 Poverty rate (%) Poverty rate (%) 40 40 35.9 34.6 30 30 22.1 27.2 20 20 15.3 16.9 10.8 9.4 9.4 10 7.0 10 4.4 3.7 4.6 2.8 2.1 1.1 1.2 1.3 1.0 1.5 6.4 0 0 2020 2010 2012 2014 2016 2018 2020 2010 2012 2016 2018 2014 Rural Urban Ethnic minority Kinh Note: Poverty rate according to the $3.20/day 2011PPP poverty line. Source: World Bank staff calculations using VHLSS 2010–2020. From the Last Mile to the Next Mile 9 Figure O.12. The distribution of poor varies at different thresholds of poverty Region Education 100% 100% 80% 80% Share of the population Share of the population 60% 60% 40% 40% 20% 20% 0% 0% Extreme & Economically Population Extreme & Economically Population moderate poor vulnerable moderate poor vulnerable Mekong Delta Central Highlands No education Upper secondary Southeastern Area Red River Delta Primary Vocational Northern & Coastal Central Midlands & Northern Lower Secodary Tertiary Region Mountainous Areas Note: The extreme and moderate poor are those living below the LMIC poverty line ($3.20/day 2011PPP). The economically vulnerable are those living between the LMIC and UMIC poverty lines. Household education categorization is based on the maximum level of education in the household. See World Bank 2018 for more details. Source: World Bank staff calculation using VHLSS. The population of the economically vulnerable8 is more Their populations had the education and skills ready to take than double the size of the population of poor. Moreover, on more complex jobs with higher productivity as well as this group has a different profile, requiring different policies the ability to effectively manage risks; in addition to growing to sustain their well-being at higher levels. The population’s the economy, this allowed for inclusive upward household geographic distribution means that the economically mobility. Today, this transition is as challenging as ever. Part II vulnerable are more prevalent in regions outside of the of the report reviews forward-looking themes related to these traditionally poor mountainous areas, with the difference in aspects for the Vietnam context. the share of the economically vulnerable compared to the poor is greatest in the Mekong Delta (Figure O.12). The share of the Human capital formation for inclusive poverty economically vulnerable with higher levels of education is also reduction and breaking intergenerational higher than it is for the poor. A more diverse economically poverty patterns vulnerable group also poses additional challenges to attaining Next Mile aspirations. Human capital—a combination of the education, skills, and health factors that largely determine labor productivity—has been a major driver of sustained Part II. The Next Mile is economic growth and is also a key ingredient in the Road Ahead breaking intergenerational poverty traps. For children, the education they receive and how healthy they grow up The Next Mile is the road to upper-middle and affects their future earnings, life expectancy, and human high-income country living standards capital as adults (World Bank, 2019). Falling short on human capital formation can limit their economic mobility in Only a handful of developing economies in the last half- adulthood (Narayan et al., 2018). While indicators of early century have successfully made the leap to high-income human capital formation in Vietnam at the national level status. Their success was partly based on a continuous structural are mostly in line with regional peers, there are gaps within transformation into ever more modern and productive sectors. Vietnam between different groups. For example, performance 10 2022 Vietnam Poverty and Equity Assessment Overview gaps between the bottom and top 20 percent of children are Education is a pathway out of poverty, but larger than average compared to other countries. Gaps in the COVID-19 has created large learning losses Vietnam socioeconomic status (SES) disaggregated Human Capital Index, harmonized test scores, and stunting rates are COVID-19 has added further challenges to equitable all larger than the average gap in a comparative study across 50 human capital development, affecting nutritional developing countries (D’Souza, Gatti, and Kraay, 2019). attainment, health, and educational continuity and outcomes (World Bank, 2021a). Between September Education completion varies with household 2020 and March 2021, 72 percent of households with a child characteristics, especially ethnicity and economic status between 6 and 18 years of age experienced school closures. (as measured by household per capita consumption School closures hit poor, ethnic minority, and low-achieving quintiles). Private expenditures in education vary students particularly hard, many of whom lacked access to substantially by household background, even at compulsory digital technologies. Over 60 percent of households in the grade levels in public schools. In 2020, households in the top lowest welfare quintile and ethnic minorities, and close to quintile spent 5.6 times more than households in the bottom 59 percent of those living in the Midlands and Northern quintile on extra classes for children attending public primary Mountainous Areas, Central Highlands, and the Mekong and lower secondary grades (Figure O.13). The Kinh spend Delta regions, did not have online learning offered during over seven times more than ethnic minorities. At the upper- school closures (Figure O.14). Public investment in human secondary level, the gap widens with the top quintile spending capital is now more important than ever to prevent a reversal as much as 10 times more on extra education courses, while as of the progress that has been made, as well as to prepare future many as 80 percent of their children are enrolled in education generations for a rapidly evolving economic landscape. up to their early 20s, compared to almost none in the bottom 20, and only 20 percent in the second and third wealth Graduating to high-income status requires quintiles. higher labor productivity Differences in household finances consistently explain Continuing on the path of a growth-driven welfare- the largest proportion of the differences in on-time enhancing process will require higher productivity education completion across children of different per worker to reach high-income country status. With backgrounds. By level of education, the importance participation and employment rates unable to increase much of household finances as a driver is greatest for on-time beyond that of 2012 to 2018 and if the population continues completion at higher levels. The association between student to age as projected, then the number of workers will fall, and learning and household characteristics is well established both the productivity of those who do work will need to increase in developed and developing countries, as it is for gaps in to sustain growth. Average real per capita GDP growth from educational outcomes between children from high and low 1991 to 2018 was 5.6 percent per annum; 6.7 percent would socioeconomic backgrounds.9 In Vietnam, a child’s education be required until 2045 to reach high-income status. To achieve outcomes relative to their parents’ tend to persist, indicating this, growth in productivity per worker will need to increase less intergenerational economic mobility. For example, across from the 2012–2018 annual rate of 5.3 percent—already the the East Asia and Pacific region, Vietnam has the smallest highest of the last three decades—to 6.6 percent per annum, share of the 1980s generation that ascended to the top quartile an acceleration of around 20 percent every year (Figure O.15). of education completion if their parents had low levels of At the rate observed from 2012 to 2018, Vietnam would be education10 (Narayan and Yang, 2019). Thus, narrowing gaps comfortably upper-middle income in 2045, but around $4,000 in education is important to realize human capital potential per capita short of high-income status (Figure O.16). for disadvantaged students. From the Last Mile to the Next Mile 11 Figure O.13. Household expenditure on extra Figure O.14. Variation in continuity of education courses at compulsory grade levels in education across Vietnam’s regions public schools, by household quintile 100 8,000 80 Share of households (%) 7,000 Annual VND (thousand) 6,000 60 5,000 4,000 40 3,000 20 2,000 1,000 0 No distance Online live SMS Paper-based Other 0 learning classes self-learning Extra courses Red River Delta Central Highlands 1(=poorest) 2 3 4 5 (=richest) Midlands & Northern Southeastern Area Mountainous Areas Mekong Delta Note: Categories by household quintiles. Average expenditures among households Northern & Coastal Central Region with children in public primary or lower secondary education. Source: World Bank staff calculations using VHLSS 2020. Note: SMS = Short Message Service. Source: World Bank Vietnam COVID-19 household monitoring surveys (round 5). Figure O.15. Projected growth under different Figure O.16. Projected income under different productivity scenarios, 1991–2045 productivity scenarios, 2018–2045 7 16,000 6.4% 5.9% 6 14,000 5.5% 0.3% HIC 1.0% 5.1% 5.3% 5 1.0% 12,000 GDP per capita (2010 USD) 0.1% 4 10,000 Growth rate 3 8,000 5.0% 4.4% 5.3% 5.3% 6.6% 2 6,000 UMIC 1 4,000 0 -0.1% 2,000 −0.1% -0.2% -0.2% LMIC -0.2% −1 0 1991–2002 2002–2012 2012–2018 2018–2045 2018–2045 2018 2045 Baseline HIC Productivity Employment Participation GDP - Baseline GDP - HIC Demographics GDP per capita Note: Both projected periods use UN projected total and working age populations and hold employment and labor force participation constant. Baseline uses 2012–2018 average worker productivity growth; HIC uses the growth required to achieve high-income status by 2045. See Annex of Chapter 5 for methodology. Income levels are converted from GNI per capita to GDP per capita based on the 2018 ratio of the two in Vietnam. Source: WDI and World Bank calculations. 12 2022 Vietnam Poverty and Equity Assessment Overview Sustaining upward economic mobility – ever more productive sectors. Their populations had the challenges ahead education and skills to take on more complex jobs with higher productivity; in addition to growing the economy, this allowed High rates of economic growth have transformed for inclusive upward household mobility. Productivity gains households, but future labor market challenges have can be achieved with better management and agglomeration been looming on the horizon that may impact upward in urban areas, shifts toward higher value-added economic economic mobility at higher income levels11 (Bodewig activity, and more widespread use of sophisticated et al., 2014; Cunningham et al., 2018; Pimhidzai and Niu, technologies, all which are well acknowledged but often face 2020; Mason and Shetty, 2019). These reports illustrate how challenges in implementation (World Bank, 20201b). changing domestic and global conditions will require a more skilled and productive labor force to sustain growth. The Promotion policies to sustain economic mobility many challenges include low completion at higher education must be complemented by protection strategies levels, high labor informality, low labor productivity, low to preserve gains wages, low skills, and an aging society. Informality has declined but remains widespread: in 2020, only 23.5 percent Vietnam is facing new vulnerabilities that require more of households had a household member employed with a sophisticated social protection systems. As the economy formal wage contract. Low-cost labor was one of the principal becomes increasingly globalized and in the face of changing reasons behind increasing foreign direct investment, primarily global and regional trade patterns, households are becoming into manufacturing. While wages in Vietnam doubled from more vulnerable to economic volatility and shocks. At the 2010 to 2020, they remain low at just under $300 per month. same time, environmental risks exist in different forms Projections estimate that the number of people aged 65 and throughout the country; for poor households, natural disasters over will increase from 10 percent of the population today to or accidents can become poverty traps, as they have the least 20 percent in 2045 (World Bank, 2021b). savings and personal safety nets. In addition, COVID-19 cast light those who are economically insecure, even if they are The challenges of reaching higher-productivity growth not poor in absolute terms. Even the relatively small shock are reflected in the characteristics of the current youth of COVID-19 in the early phases in Vietnam led to lingering workforce. Today’s youth in Vietnam are better educated longer-term impacts. Before the arrival of the Delta variant in than previous generations, but some indicators point to March 2021, about 30 percent of households still reported challenges in transitioning to higher-skilled jobs without lower household incomes than a year before (World Bank, continuing reforms in education, skill development, and 2021a). COVID-19 has also been a “gender-biased” crisis, as a transformation of the labor market. Despite higher increased childcare burdens due to prolonged school closures education completion, the youth cohort is concentrated in disproportionately affect women’s work. All these issues are medium-skilled occupations, more so than youth cohorts in likely contributing to expanding inequalities. They also trap other countries in the region. Employers in Vietnam report some groups in near-poverty situations. difficulty finding skilled labor. Based on the World Bank’s Enterprise Survey on Innovation and Skills in Vietnam While some households face chronic economic insecurity conducted in 2019, 22 percent of managers reported that the because of lack of livelihood-generating capacity, others biggest obstacle faced by the firm was an adequately educated are insecure because of the risks they are exposed to. In workforce (World Bank, 2021d). However, challenges are not addition to understanding why some households still live in purely from the labor supply side; the domestic private sector poverty, as Part I of the report does, it is important to also faces challenges in innovation and competition. examine those living below the higher and more aspirational “economic security” line ($5.50 per day, 2011PPP). This Addressing these challenges is necessary to sustain much larger group of households can be divided into those continuous economic transformation toward higher whose economic insecurity is “chronic” and those whose income. Only a handful of developing economies in the last insecurity is “risk-induced.” The chronically insecure live half century have successfully made the leap to high-income below the economic security line most of the time because status, requiring a continuous structural transformation into they lack the human capital and physical assets to earn a From the Last Mile to the Next Mile 13 sufficient living even in good years. These households most experience risk-induced insecurity. Conversely, in the Red likely need better access to economic opportunities, support River Delta, where chronic insecurity is low, risk-induced with cash transfers, and better delivery of basic services to insecurity is twice as prevalent as chronic insecurity. Other facilitate investments in physical and human capital. Those regions where risk-induced insecurity is higher than the who face risk-induced economic insecurity consume enough national average are the Southeast and the Mekong Delta. to live above the economic security line, but this consumption Nationally, idiosyncratic risk contributes to risk-induced can be highly variable because of idiosyncratic or covariate insecurity 1.2 times more frequently than covariate risk shocks (the former affecting individuals or specific households (Figure O.18). The Red River and Mekong Deltas are regions and the latter affecting whole communities, regions, or the where idiosyncratic risk matters a lot more (1.7 and 1.5 times, country), meaning that sometimes they fall below the line respectively), while in the Midlands and North and Central and into insecurity. These households do not necessarily need Coast, covariate risk is nearly as important. These differences the same investments in assets but do need protection from in chronic insecurity and risk profiles have important shocks, which might mean insurance programs to increase implications for operational and policy design. resilience. The social protection system does not adequately Around one in five Vietnamese usually live below the protect all households against all risks. Social assistance is economic security line, and another one in ten are fragmented and underfunded, and has a delivery system with vulnerable to occasionally falling below it due to shocks. implementation issues (Nguyen and O’Keefe, 2019). Social In Vietnam, the rate of chronic insecurity is twice that of risk- assistance programs are concentrated on particular categories induced insecurity, but this varies significantly by region. In of households, such as the very old, ethnic minorities, and addition, idiosyncratic risks contribute more to insecurity those with disabilities. Consequently, many poor households than covariate risks. The rate of chronic insecurity is 1.9 times not meeting these criteria are excluded. Instead of a flagship greater than that of risk-induced insecurity (Figure O.17). social program such as those in China, Indonesia, and the Regional outcomes vary widely. With so many chronically Philippines, there are many smaller individual programs, insecure in the Midlands, there are fewer people left to which leads to different systems for implementation and Figure O.17. Rates of chronic insecurity and Figure O.18. Relative importance of idiosyncratic and risk-induced insecurity by region, 2020 covariate risk by region, 2020 70 4.5 1.8 Ratio of Idiosyncratic/Covariate Risk 4 1.6 60 3.5 Ratio of Chronic/Risk 1.4 50 3 1.2 40 2.5 Percent 1 30 2 0.8 1.5 0.6 20 1 0.4 10 0.5 0.2 0 0 0 All Midlands & Northern Mountains North & Central Coast Central Highlands Southeast Red River Delta Mekong Midlands & Northern Mountains North & Central Coast Central Highlands Southeast All Red River Delta Mekong Chronic Economic Insecurity Risk-induced Chronic Economic Insecurity Chronic: Risk Ratio Chronic: Risk Ratio Economic Insecurity Source: VHLSS 2020 and World Bank’s Vulnerability Tool. 14 2022 Vietnam Poverty and Equity Assessment Overview inefficiencies. Relatively low spending compared to other many countries, COVID-19 highlighted the role that fiscal middle-income countries compounds these inefficiencies. policy can play in mitigating shocks. It is also an important Low spending is largely the result of low benefit levels, part of financing public investment in physical and digital which are spread thin and consequently have relatively little infrastructure and human capital which countries need to impact on poverty. In addition, the response to the pandemic make the transition to upper-middle and high-income status. highlighted weaknesses in delivery systems, specifically their At the same time, fiscal policy is one of the few instruments inability to respond quickly to large shocks. The first relief which governments can use to reduce inequality in the short package, which targeted 5 million informal sector workers, term. Households pay various taxes and benefit from public only ended up reaching 1 million. A significant problem spending in different ways. The net effect determines the was the inability to leverage administrative data and the extent to which fiscal policy directly reduces poverty and national ID database (currently being digitalized) as many inequality. The choice of public spending can also affect the other countries did. Meanwhile, social insurance, which is extent to which poverty and inequality are reduced in the designed to cushion shocks, is constrained by inadequate longer term. coverage of households, particularly those with informal workers. Coverage extends almost exclusively to workers in This report uses the Commitment to Equity fiscal the formal sector—a serious constraint given the country’s incidence analysis to trace how household income high informality rate. A voluntary contributory scheme was changes when taxes are paid and benefits received. Many introduced in 2006 but has failed to attract many informal poorer households benefit from direct cash transfers, which workers. Most workers have little recourse when subjected by themselves would reduce poverty by 1.05 percentage to shocks such as unemployment or disability. This gap was points, and electricity subsidies by another 0.15 points. exposed by COVID-19, as millions of non-poor and mostly However, taxes and payroll contributions, particularly VAT, urban informal sector workers in sectors such as transport excises and social security and health insurance contributions, and tourism suddenly lost their incomes and were without mean that the benefits received by the poor are outweighed insurance to soften the impact on their incomes. Women are by what they pay into the fiscal system.12 The impact on also not as well protected against shocks, with women-headed inequality is more positive (Figure O.19); taxes are either households tending to be more vulnerable than those headed largely neutral or progressive (in the case of personal income by men. Agriculture and services make up about 73 percent taxes and insurance contributions), while spending on health of female employment (ILO, 2021), the latter of which is also and especially education is also quite progressive. Therefore, concentrated in informal services, limiting their access to the the Gini coefficient of inequality drops by 5 points from its formal social insurance system. pre-fiscal to post-fiscal measure. Fiscal policy is a key government instrument that This reduction in inequality is around average for can enable inclusive prosperity in both the short lower-middle income countries, but many countries and the long term are more progressive. A five-point reduction in inequality (as measured by the Gini Index) puts Vietnam in the middle Fiscal policy can support the development of a prosperous range of lower-middle income countries with respect to fiscal and inclusive middle-class society. It can be used to provide progressivity (Figure O.20), but this would place the country public goods and services, for macroeconomic stabilization, in the bottom third of upper-middle income countries. to stimulate economic growth, and to reduce poverty. In From the Last Mile to the Next Mile 15 Figure O.19. Impact of fiscal policy on inequality, 2018 6.0 5.0 Percentage Point Reduction 4.0 3.0 2.0 1.0 0 Health Disposable income Cash transfers Education Other direct taxes SSC/HI Excises Environmental tax Consumable income Final income PIT VAT Electricity subsidy Note: The blue bars are direct taxes and transfers. The green bars are indirect transfers and taxes. The orange bars are in-kind. Poverty is measured at the $3.20 per person per day in 2011PPP terms used for LMICs. Figure O.20. Reduction in inequality due to fiscal policy in international context 0 −5 Point change in the Gini Index −10 −15 −20 −25 Namibia Argentina South Africa Costa Rica Brazil Mexico Iran Mexico Mexico Ecuador Belarus Colombia Colombia Venezuela Russia Albania Dominican Republic Armenia Peru Peru Indonesia Guatemala Jordan Ukraine Tunisia Kenya El Salvador Bolivia Vietnam Tanzania Nicaragua Egypt Ghana Comoros Honduras Sri Lanka Burkina Faso Uganda Ethiopia Poland Croatia Panama Uruguay Chile Romania Paraguay Lower High income Upper middle income Lower middle income income Note: Pensions treated as deferred income. Source: CEQ and World Bank databases and World Bank calculations (see Rodriguez and Wai-Poi, 2020). 16 2022 Vietnam Poverty and Equity Assessment Overview There is significant potential for Vietnam’s fiscal policy Policies for the way to do more in terms of investing for inclusive growth by adopting lessons from other countries. International forward experience suggests that countries become richer in part Policies are discussed following the Last Mile to Next Mile through greater public investment in drivers of economic framework of this report. Over the last decade, rapid economic growth and inclusive prosperity, increasingly financed by growth was broadly inclusive and livelihoods in Vietnam relying more on progressive direct taxation such as personal improved dramatically. The rapid speed of developmental income taxes rather than indirect taxation such as goods change, however, left some behind who did not have the and services taxes. In lower-middle income countries, where opportunity to join the most vibrant sectors of the economy indirect taxation underpins the majority of tax revenue, and has also created a large class of the population that are some choose to offset the tax burden on poorer households not poor but also not yet middle class. Thus, the poverty by applying lower rates or exemptions on staples consumed and equity agenda is no longer only about raising minimum by the poor such as food and clothing, as Vietnam does. living standards and tackling chronic poverty; it is also about However, this means sacrificing significant revenue, the creating new and sustainable economic pathways for a more benefits of which go mainly to richer households who also aspirational population. The emergence of COVID-19 added purchase these goods, and in greater amounts. Other LMICs to challenges looming in regard to skills, productivity, climate raise more revenue by closing such exemptions and are able to change, and an aging society. 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 Last Mile protecting key spending on human capital. With the fourth Tackling poverty for groups with persistently wave of COVID-19 beginning in April 2021 and never being high poverty rates is the Last Mile challenge. fully brought under control and the emergence of the Omicron variant in late 2021, the country (and the world) are still very much in the middle of the pandemic at the time of writing, with great uncertainty over future health, economic, and social trajectories. Nonetheless, key questions can begin to be asked. Is the current fiscal support package for households adequate? When and how should it be tapered, so that vulnerable households who may have exhausted their resilience do not resort to bad coping mechanisms, but emergency spending does Next Mile not continue for longer than is needed? How should fiscal gaps (deficits and debt) be closed as the recovery accelerates? How The Next Mile is the road to upper-middle can revenues be increased? What spending can be consolidated and high-income country standards. and what spending needs to be protected? What should the long-term fiscal framework look like in Vietnam to facilitate public investments in growth for tomorrow while reducing poverty and inequality today? In particular, the limited role cash transfers have traditionally played in Vietnam is evident in the lack of fiscal poverty reduction before COVID and in the scale and nature of its response. From the Last Mile to the Next Mile 17 Addressing Last Mile Chronic Poverty Reduction equipped with the necessary human capital and skills to Challenges engage in more productive and sophisticated jobs. Broad policy areas include investing in skills for the future, investing The concentration of poverty among geographically in higher-quality education, modernizing social protection disadvantaged regions calls for strengthening of area- systems for idiosyncratic shocks, and leveraging fiscal policy based anti-poverty intervention, modernization of to fund inclusive investments. the agricultural sector, and improving education for disadvantaged students. Analysis suggests that past NTPs Continuing improvements in higher education access helped improve access to services and achieved some positive and quality will be needed. Improving the relevance and impacts on improving welfare, but large gaps were also quality of tertiary education curriculum and staffing can identified (Pimhidzai and Niu, 2020). Strengthening of NTPs help reduce skill gaps and improve perceptions by businesses can include (i) ensuring that additional resources reach lagging who report difficulties recruiting for certain skills. In communes; (ii) ensuring that resource allocation is based on particular, businesses find it difficult to recruit for leadership, deprivations at the commune level, so that communes that managerial, socio-emotional, or job-specific technical skills. are more deprived receive more investment; (iii) earmarking To enhance curriculum quality, policies can be pursued resources across subsectors to ensure that adequate resources to improve funding of university research, improve staff are devoted to improving quality of human development quality and qualifications, design programs to recruit and services and livelihood interventions; and (iv) strengthening retain high quality staff, and to convert higher education commune-level monitoring by deploying online data teaching status from administrative staff to public servants. collection and aggregation tools to construct a centralized The quality of curriculum can also be improved through NTP database. Supporting agricultural productivity growth centrally coordinated investments to increase international is key to maintaining livelihoods for those remaining in the accreditation of programs, increase international exchange of rural economic system in the face of significant structural students and staff, and the internationalization of curriculum change. Access to new knowledge and innovations, including (World Bank, 2020b). application of appropriate digital technologies, would support productivity growth by substituting for labor intensity. A modern social protection system is needed. A decade Moreover, the social protection system can play a larger role ago, Vietnam’s social assistance coverage was in line or higher (discussed as part of the Next Mile agenda). than many East Asia and Pacific neighbors, but it now lags. Vietnam needs to provide more effective social assistance for To improve the participation of ethnic minorities in poorer households, with increased coverage and benefit levels the labor market, laws protecting the rights of ethnic to achieve greater poverty and inequality reduction and greater minorities can be further strengthened. The existing overall spending while consolidating the currently fragmented Ethnic Minorities Law, which was designed to advance the mix of programs to achieve greater efficiency (Nguyen policy agenda by recognizing ethnic minorities, improving and O’Keefe, 2019). Greater coverage of social insurance conservation of language and culture has yet to be enacted to protect all households from all risks can be achieved by (World Bank, 2021c). The Labor Code and related expanding coverage to non-poor informal workers by blurring legislation could be strengthened to prevent exploitation and the line between social assistance and insurance; workers make discrimination of ethnic minorities (World Bank, 2021c).  the contributions they can afford while the state subsidizes the remainder. Finally, a more flexible and adaptable delivery Achieving Next Mile Aspirations chain for both social assistance and insurance will improve effectiveness, including better data collection and use to The millions who have escaped poverty over the last determine who is in need and how this changes over time, and decade now need to continue climbing up to higher adoption of digital payment systems so that payments occur economic classes. The economically insecure require quickly and safely, reaching those who need them when they different policy support than the poor such as being provided need them. safety nets to prevent falling back into poverty, and becoming 18 2022 Vietnam Poverty and Equity Assessment Overview Fiscal policy can play a critical role in both driving which encourage wasteful energy use while mostly benefitting Vietnam toward high-income status and doing so in an richer households, could be redirected for more productive inclusive manner to assist the movement of people into a and inclusive purposes. For example, Vietnam needs to bring prosperous middle class. To finance the public investments social protection spending in line with international norms to needed to eliminate poverty and grow the economically secure develop a modern system which helps households manage the and middle classes, Vietnam can broaden its tax revenue many risks they face. base (personal income tax, property tax), explore the use of new taxes which both raise revenue and address negative Policies should have the dual objectives of tackling externalities (such as health taxes on alcohol, tobacco, and remaining Last Mile chronic poverty challenges, as well sugar-sweetened beverages, as well as environmental taxes such as setting the foundation to reach Next Mile aspirations. as on carbon), or encompass the growing digital economy, Whether or not these challenges end up being short-term and eliminate regressive tax expenditures. In addition, public growing pains or long-term bottlenecks to the welfare expenditures need to be directed in the right way. Inefficient trajectory of Vietnam’s households will depend on policy and inequitable expenditures such as electricity subsidies, action and prioritization. From the Last Mile to the Next Mile 19 References UNDP (United Nations Development Programme). 2018. Multidimensional Poverty in Vietnam. Reducing Poverty in Bodewig, Christian, Reena Badiani-Magnusson, Kevin McDonald, all its Dimensions to Ensure a Good Quality Life for All. New David Newhouse, and Jan Rutkowski. 2014. Skilling Up Vietnam. York, NY: UNDP. Preparing the Workforce for a Modern Market Economy. World Bank. 2018. Riding the Wave: An East Asian Miracle for the CECODES, VFF-CRT, RTA & UNDP. 2021. The 2020 Viet Nam 21st Century. World Bank East Asia and Pacific Regional Report;. Governance and Public Administration Performance Index (PAPI Washington, DC: World Bank. 2020): Measuring Citizens’ Experiences. --------------. 2019. World Development Report 2019: The Changing Cunningham, Wendy, and Obert Pimhidzai. 2018. Vietnam’s Future Nature of Work. Washington, DC: World Bank. Jobs – Leveraging Mega-Trends for greater prosperity. Washington, --------------. 2020a. Poverty and Shared Prosperity Report: Reversals of DC: World Bank. Fortune. Washington, DC: World Bank. Dsouza, R., Gatti, R.V., Kraay, A.C. 2019. A Socioeconomic --------------. 2020b. Improving the Performance of Higher Education in Disaggregation of the World Bank Human Capital Index. Policy Vietnam: Strategic Priorities and Policy Options. Washington, DC: Research Working Paper; no. WPS 9020 Washington, DC: World World Bank. Bank Group. --------------. 2021a. A Year Deferred – Early Experiences and Lessons from Lakner, Christoph, Daniel Gerszon Mahler, Mario Negre, and COVID-19 in Vietnam. Washington, DC: World Bank. Espen Beer Prydz. 2020. “How Much Does Reducing Inequality --------------. 2021b. How Will Vietnam Blossom?: Reforming Institutions Matter for Global Poverty?” Global Poverty Monitoring Technical for Effective Implementation. Washington, DC: World Bank. Note 13, Washington, DC: World Bank. --------------. 2021c. Reducing Poverty Among Ethnic Minorities in GSO. 2021. Major Findings. The 01/4/2020 Time-Point Population Vietnam: Priority Action Areas and Lessons. Washington, DC: Change and Family Planning Survey. World Bank. Ha, Thi and Anh Minh. 2021. Citizens, Businesses Hurt as Rising Prices --------------. 2012d. Vietnam - Science, Technology, and Innovation Report. Raise Inflation Concerns. VNExpress, May 19, 2021. https://e. Washington, DC: World Bank Group. vnexpress.net/news/business/economy/citizens-businesses-hurt-as- rising-prices-raise-inflation-concerns-4279567.html. ILO Online Database (https://www.ilo.org/global/statistics-and- databases/lang--en/index.htm) Notes Mason, Andrew D., and Sudhir Shetty. 2019. A Resurgent East Asia: Navigating a Changing World. World Bank East Asia and Pacific 1  This is equivalent to about $350 per month, and average monthly Regional Report. Washington, DC: World Bank. wages are 6 million VND per month. Narayan, Ambar, Roy Van der Weide, Alexandru Cojocaru, Christoph 2  On living conditions, 32.7 percent responded significantly better, 51.7 Lakner, Silvia Redaelli, Daniel Gerszon Mahler, Rakesh Gupta N. Ramasubbaiah, and Stefan Thewissen. 2018. Fair Progress? Economic percent slightly better, 8.5 percent the same, and 6 percent worse, based Mobility across Generations around the World. Washington, DC: on the VHLSS. On economic conditions, most respondents indicated World Bank. that conditions were “a little better” versus “much better,” based on Narayan, Ambar, and Judy Yang. 2019. Economic mobility across UNDP PAPI surveys. 3  generations in the developing East Asia and Pacific region. Poverty This report describes the poverty and inequality story of Vietnam and Equity Notes, September 2019, Number 16. primarily using the World Bank’s global absolute poverty lines. Based Nguyen, Nga Nguyet, and Philip B. O’Keefe. 2019. A Vision for the on Vietnam’s level of development, the natural choice for the poverty 2030 Social Protection System in Vietnam (English). Washington, threshold is the LMIC Poverty Line ($3.20/day 2011PPP). This line D.C.: World Bank. also makes sense when converted to Vietnamese dong, which converts to Pimhidzai, Obert; and Chiyu Niu. 2020. Vietnam Poverty and Shared about 1 million VND per person per month in January 2020 prices. At Prosperity Update: April 2020. – Shared Gains: How high growth this level, the threshold is similar to the MOLISA (general poverty rate) and anti-poverty programs reduced poverty in Vietnam. Washington, monetary near-poor threshold over the 2016-2020 Socio-Economic D.C.: World Bank Group. Development Phase (SEDP) period. Alongside this line, the report also Rodriguez, L and Wai-Poi, M. 2020. Fiscal Policy, Poverty and Inequality examines poverty rates at the Upper-Middle Income Class (UMIC) in Jordan: The Role of Taxes and Public Spending. Washington, DC: Poverty Line ($5.50/day 2011PPP), which is equivalent to about 1.8 World Bank. million VND per person per month in January 2020 prices. The UMIC PL is in between the urban and rural monetary poverty thresholds for the 2021-2025 SEDP and is also relevant for higher aspirations. 20 2022 Vietnam Poverty and Equity Assessment Overview 4  The shared prosperity premium is negative if growth of the richest 60 percent (“Top 60”) is higher than that of the poorest 40 percent (“Bottom 40”). 5  An individual is underemployed if they work less than 35 hours a week and would like to work more. 6  The Human Capital Index is 0.85 for children in the richest 20 percent of households, compared to 0.58 for children in the poorest 20 percent. Unsurprisingly, children in the top 20 percent have better nutrition, health, and education outcomes. For some outcomes, the gap between the top and bottom is larger in Vietnam than the average gap among other countries. For example, the gap in the Human Capital Index between the top 20 and bottom 20 in Vietnam is 0.27 points, higher than the average gap among 50 countries (0.15 points). See D’Souza, Gatti, and Kraay (2019). 7  As defined by the World Bank LMIC poverty line ($3.20/ day 2011PPP). 8  By World Bank regional definitions, the “economically vulnerable” are those that subsist above the LMIC PL but below the UMPC PL (World Bank, 2018). 9  In terms of gender, girls in Vietnam are better educated than their male counterparts on average. Girls outperform boys across multiple metrics: the Human Capital Index, scores on standardized tests, and expected years of schooling. Girls also exceed boys in childhood health indicators, such as survival and stunting indicators. While this was already the case in 2010, the gains made by girls in the decade since have outpaced those made by boys, widening the gap even further. Scrutinizing variations in access to opportunities across socioeconomic and demographic factors, gender plays a negligible role, and where it has a more significant impact is to the benefit of women. 10  The Poverty-to-Privilege indicator measures the share of the 1980s generation that attained education completion in the top quartile when their parents had only attained education in the bottom half. 11  This report primarily reviews labor supply-side constraints. 12  However, it is important to note that the analysis covers only 57 percent of all tax revenues and around a third of all central government expenditures; other spending, such as investments in infrastructure, may indirectly reduce poverty by boosting economic growth and creating more economic opportunities for poorer households. From the Last Mile to the Next Mile 21 PART 1. Part 1. A Decade of Significant Progress, but Last-Mile Challenges Remain 22 2022 Vietnam Poverty and Equity Assessment Part I of this report reviews poverty and inequality trends over the last decade, 2010–2020, and discusses their underlying drivers and remaining challenges. The decade saw great success in poverty reduction, but also widening inequality (Chapter 1). As the world was hit with the COVID-19 pandemic in 2020, poverty still fell compared to 2018. However, with the emergence of the Delta variant in Vietnam in early 2021, there will likely be additional setbacks to poverty reduction and increases in inequality in both monetary and non-monetary dimensions. Improving standards of living were driven to a large extent by the creation of off-farm jobs, filled by a large youth population entering the labor force and enjoying increasing wages (Chapter 2). While structural shifts out of agricultural activities is occurring, the poor are still concentrated in low-productivity and low-income agriculture. Poverty rates remain higher among rural and ethnic minority families and households primarily engaged in agriculture (Chapter 3). These groups face persistent challenges of lower levels of human capital, lower-quality local public services, and farther distance to economic opportunities. CHAPTER 1. Taking stock of welfare trends over a dynamic decade CHAPTER 2. A review of the drivers of poverty reduction CHAPTER 3. Challeneges to reducing poverty among the remaining poor From the Last Mile to the Next Mile 23 PART 1. Chapter 1. Taking stock of welfare trends over a dynamic decade Chapter 1. Taking stock of welfare trends over a dynamic decade Key Messages • Over 10 million people in Vietnam moved out of poverty over the last decade (2010–2020), based on the World Bank’s Lower-Middle Income Country (LMIC) poverty line ($3.20/day 2011PPP), and LMIC poverty rates dropped from 16.8 to 5 percent. • However, inequality is also rising, especially in the latter half of the decade. The shared prosperity premium has also been negative, meaning that the growth of household consumption among the bottom 40 percent is lower than the average. • In 2020, COVID-19 did not reverse the progress made on poverty reduction compared to 2018, but a more severe emergence of the Delta variant in April 2021 will lead to larger setbacks and possibly longer-term consequences from widening inequality. 24 2022 Vietnam Poverty and Equity Assessment Chapter 1. Taking stock of welfare trends over a dynamic decade 1.1. Introduction: From the While poverty has declined amid rapid economic Last Mile to the Next Mile growth, there is now also a large share of the population that is no longer poor but is not yet middle class. Poverty Chapter 1 documents poverty and inequality trends in and hunger was selected as a top issue that the government Vietnam over the last decade. The chapter’s sub-sections must tackle (Figure 1.2). The 2018 PAPI probed why poverty describe broad trends and patterns in monetary poverty, was the main concern and found that many worried about economic mobility, subnational poverty, distributional falling back into poverty, but even more felt that poverty is indicators of inequality, and the impacts of COVID-19 on an overall drag on the economy and reduces national prestige. poverty projections. A review of key principles of poverty Half of all respondents selecting hunger and poverty as measurement, in particular the poverty line and welfare their top concern in 2018 earned more than 8 million VND aggregate are discussed in Annexes. Deeper analyses of the per month,13 highlighting apprehensions over economic drivers of poverty reduction and poverty profiling across security. Statistics of high economic progress but also public dimensions of chronically high poverty rates are described in concern on poverty can be consistent; together they describe subsequent chapters. an absolute and inclusive rise in living standards, but also a population that seeks economic security and is still aspiring Vietnam is at a pivotal stage where it has not completely for more. These worries and aspirations reflect the need to outgrown last-mile challenges but also needs to lay a simultaneously solve last-mile chronic poverty challenges foundation for next-mile aspirations. Progress in the last and to create reliable upward economic mobility pathways to decade was substantial, and the Vietnamese economy averaged fulfill next-mile upper-middle and high-income aspirations. 6–7 percent growth rates per year. Household consumption per capita, the basis of poverty measurement, also increased by about 5 percent annually. Across different thresholds, poverty has declined considerably (Figure 1.1). Figure 1.1. Economic growth and poverty reduction, 1993–2020 100 3,000 90 2,500 GDP per capita (constant 2015 US$) 80 70 2,000 Poverty rate (%) 60 50 1,500 40 1,000 30 20 500 10 0 0 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 Extreme poverty ($1.90/day 2011PPP) LMIC PL ($3.20/day 2011PPP) UMIC PL ($5.50/day 2011PPP) GDP per capita (constant 2015 US$) Note: Poverty measurement methodology changed in 2010, rendering a break in series comparability. Source: World Bank WDI, PovcalNet. From the Last Mile to the Next Mile 25 PART 1. A DECADE OF SIGNIFICANT PROGRESS, BUT LAST-MILE CHALLENGES REMAIN Figure 1.2. Issues of greatest concern, 2015–2020 30 25 Percent of respondents 20 15 10 5 0 2015 2016 2017 2018 2019 2020 Poverty/Hunger Economic Growth/GDP Environment Health/Insurance Source: CECODES, VFF-CRT, RTA & UNDP (2021). 1.2. Monetary poverty standards, and the second is that the threshold should be rates illustrate a decade comparable across time to provide a consistent time series. The value of the World Bank global poverty lines when converted of significant progress in to VNDs is meaningful and relatable to the government’s own poverty reduction income thresholds. Thus, the trends and drivers of poverty reduction that will be described using World Bank global The determination of who is poor affects an array of poverty lines in this report are relevant for monitoring and policy decisions. Public investment, cash support, and social policy. Annex 1.3 reviews and compares poverty lines based assistance are some examples of direct government policies on Vietnam’s national and World Bank definitions. and actions that inform resource allocation and budgets. From a welfare monitoring perspective, poverty trends The World Bank’s working poverty measurement track the progress of improvement in livelihoods and living methodology is the Cost of Basic Minimum Needs standards among the most vulnerable population. Differences Approach (Ravallion, 2015). In the case of household in poverty reduction trends across population groups inform consumption as a welfare measure, welfare is measured the design and targeting strategies for poverty reduction and practically by valuing a basket of commodities or goods social assistance programs (see Chapter 6 for a discussion of that a household consumes based on expenditure data from the social protection system). household surveys.14 Households are assumed to gain value from consuming these goods and services, which increases This report describes poverty and inequality their welfare or their utility function.15 See Annexes 1.1 and trends using the World Bank global poverty lines 1.2 for additional information on household expenditure data and household consumption and global poverty measurement. What should the poverty threshold be? The threshold of Poverty reduction was impressive throughout poverty is not dogmatic. Household incomes, consumption, the decade and livelihoods exist on a spectrum. For the objective of describing trends in this report, there are two primary By 2020, 1 percent of Vietnamese lived in extreme considerations. The first is that the threshold of an appropriate poverty as defined by the International Poverty Line standard should reflect Vietnam’s minimum necessary living (IPL) ($1.90/day 2011PPP). In Vietnam, extreme poverty 26 2022 Vietnam Poverty and Equity Assessment Chapter 1. Taking stock of welfare trends over a dynamic decade is now almost eradicated (Figure 1.3). Most of the success in Poverty reduction remained resilient even amidst the extreme poverty reduction occurred before 2010, when it fell initial outbreaks of COVID-19 in 2020. While poverty the most, from 51.9 percent in 1993 to 14.1 percent in 2008 impacts may appear mild in 2020, with the onset of the (Table 1.116). In equivalent monetary terms, the value of the fourth wave in April 2021, the pandemic is expected to last IPL is closer to MOLISA monitoring standards from earlier for much longer (see Section 1.7 for poverty projections SEDP phases, and the remaining households living at these under COVID-19 scenarios). Another potential reason for levels are of high concern. lower adverse impacts measured in the 2020 VHLSS is the reference period for household expenditure in the surveys. By the end of 2020, the LMIC ($3.20/day 2011PPP) For some components of consumption, households provide poverty rate was 5 percent, a 1.6 percentage information over a 12-month recall period. For households point reduction from 2018. This was a remarkable interviewed in Q1-Q3 of 2020, their consumption patterns accomplishment, as poverty rose in many other countries would include periods in 2019 that are pre-COVID. amid the COVID-19 pandemic (World Bank, 2020). Nearly 10 million people escaped LMIC poverty over the last decade; The number of UMIC (US$5.50/day 2011PPP) poor the LMIC ($3.20/day 2011PPP) poor numbered 14.5 million is much larger than the LMIC poor (Table 1.2). Due in 2010 and 4.9 million in 2020. The remaining population to the distribution of household consumption, an increase living below the LMIC global poverty standard in 2020 in the poverty line from $3.20 to $5.50/day 2011PPP more are perceived to be at risk of meeting basic necessities at a than triples the poverty rate and the number of poor. Small standard that is relevant for the Vietnam context, bearing in increases in the poverty line will lead to sizeable increases mind that the equivalent value of the global LMIC poverty in the populations being classified into poverty, because of line converted into local currency is about 1 million VND per the large concentration of the population living near these person per month, which is the upper limit of the MOLISA standards. In 2020, the UMIC poverty rate was 18.9 percent, near-poor threshold from the 2015–2020 SEDP period. and the number of poor was estimated to be 18.3 million people. The UMIC poor population is slightly higher than MOLISA’s estimate of poor based on the 2021–2025 SEDP standard, which was estimated at 17.4 million.17 Thus, the UMIC threshold can also be used as an aspirational target. Figure 1.3. Poverty rates declined consistently 50 40 Poverty rate (%) 30 20 10 0 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 Extreme poverty ($1.90/day 2011PPP) LMIC PL ($3.20/day 2011PPP) UMIC PL ($5.50/day 2011PPP) Source: World Bank staff calculations using VHLSS 2010–2020. From the Last Mile to the Next Mile 27 PART 1. A DECADE OF SIGNIFICANT PROGRESS, BUT LAST-MILE CHALLENGES REMAIN Table 1.1. Poverty rates (2010–2020) POVERTY RATES (%) World Bank Global Poverty Lines Units: $1.90/day $3.20/day 2011PPP $5.50/day 2011PPP 2011PPP per capita, per day 2011PPP - IPL - LMIC - UMIC Approximate equivalent value in VND 0.6 1.0 1.8 Units: January 2020 VND in millions, monthly and per person 1993 52.3 79.7 94.3 1998 31.4 68.7 90.0 2002 37.0 70.1 89.0 2004 25.8 59.3 84.2 2006 18.8 50.6 79.9 2008 14.1 45.9 77.8 Significant change in consumption-based welfare measurement, new poverty data series is not comparable 2010 4.0 16.8 46.8 2012 2.7 13.0 40.8 2014 2.6 11.0 35.0 2016 1.8 7.8 27.2 2018 1.8 6.6 23.4 2020 1.0 5.0 18.8 Source: PovcalNet and World Bank staff calculations. Table 1.2. Number of poor (2010–2020) NUMBER OF POOR World Bank Global Poverty Lines $1.90/day $3.20/day 2011PPP $5.50/day 2011PPP Units: 2011PPP per capita, per day 2011PPP - IPL - LMIC - UMIC Approximate equivalent value in VND 0.6 1.0 1.8 Units: January 2020 VND in millions, monthly and per person 1993 37.8 57.6 68.2 1998 24.6 53.6 70.3 2002 30.2 57.1 72.5 2004 21.4 49.3 69.9 2006 15.9 42.8 67.6 2008 12.2 39.6 67.1 Significant change in consumption-based welfare measurement; new poverty data series is not comparable 2010 3.5 14.8 41.2 2012 2.5 11.6 36.6 2014 2.4 10.1 32.1 2016 1.7 7.3 25.5 2018 1.7 6.3 22.3 2020 1.0 4.9 18.3 Source: PovcalNet and World Bank staff calculations. 28 2022 Vietnam Poverty and Equity Assessment Chapter 1. Taking stock of welfare trends over a dynamic decade The average distance to the poverty line is also population, it would be enough to raise the poor population shrinking up to the poverty line (each poor person would need to receive a different amount depending on how far they are from the The average poverty gap illustrates the amount of line). In the case of Vietnam, the poverty gap in 2020 is 1.2 additional consumption necessary to raise the poor to percent of the LMIC poverty line (panel A, Table 1.3). The the poverty line (Figure 1.4). The gap is measured in terms total monetary gap is calculated as the value of the poverty gap of the percentage of the poverty line on a per person basis that over the entire year per person on average (panel B, Table 1.3). would raise everyone in poverty to the minimum threshold. In 2020, the monetary gap to lift everyone above the LMIC For example, an average poverty gap of 1 percent means that poverty line equals 15 trillion VND, and the total monetary if an amount of 1 percent of the poverty line is contributed gap for the UMIC poor is 116.9 trillion VND. from the entire population and transferred to only the poor Figure 1.4. Illustration of poverty and poverty gap (LMIC poverty line) A. Illustration of poverty B. Illustration of the poverty gap 80 80 Household consumption per capita 70 Household consumption per capita 70 60 60 (2011PPP, per day) (2011PPP, per day) 50 50 40 40 30 30 20 20 10 10 0 0 Percent of the population (%) Percent of the population (%) 2010 2020 LMIC PL ($3.20/day 2011PPP) 2020 LMIC PL ($3.20/day 2011PPP) Notes: In Panel B, the red shaded areas depict the poverty gap. Source: World Bank staff calculations using VHLSS. Table 1.3. Average poverty gap A. AVERAGE POVERTY GAP (SHARE OF THE POVERTY LINE, FGT(1)) 2010 2012 2014 2016 2018 2020 IPL ($1.90/day 2011PPP) 0.008 0.005 0.005 0.004 0.003 0.002 LMIC – PL ($3.20/day 2011PPP) 0.046 0.032 0.029 0.021 0.018 0.012 UMIC – PL ($5.50/day 2011PPP) 0.161 0.131 0.111 0.084 0.069 0.055 B. ANNUAL TOTAL GAP (2020 VND, TRILLION) 2010 2012 2014 2016 2018 2020 IPL ($1.90/day 2011PPP) 5.38 3.02 3.64 2.48 2.43 1.44 LMIC – PL ($3.20/day 2011PPP) 49.98 36.13 33.88 24.59 22.38 15.02 UMIC – PL ($5.50/day 2011PPP) 302.13 254.08 221.65 169.84 146.29 116.91 Notes: Currency conversions use 2011PPPs, not exchange rates. Source: World Bank staff calculations using VHLSS. From the Last Mile to the Next Mile 29 part 1. a deCade of signifiCant progress, BUt last-mile Challenges remain BOX 1.2. Foster, Greer, and Thorbecke (FGT, 1984) class of indicators The FGT family of indexes with parameter α. As the parameter increases, more importance is given to those below the poverty line. The poverty line is denoted as z and a household ’s consumption, or income is . The index measures the proportion of people in poverty (FGT0), the average distance (in monetary terms) to the poverty line (FGT1), and the dispersion in standards of living among the poor (FGT2, that is, the square of the average distance, in order to give more importance to those far from the line and underline inequality among the poor). Compared to national poverty reduction budgets, ranking. In 2010, the regional poverty rate in the Midlands was the poverty gap is small. For example, the New Rural the highest in the country. In 2020, the Central Highlands had Development National Targeting Program (NTP) the highest regional poverty rates, if even by a small margin. implemented by the Ministry of Agriculture and Rural Development has a budget of 2,600 trillion VND over a There are relevant concerns that economic vulnerability five-year period from 2021 to 2025 and does not cover all in the Mekong Delta will continue to worsen. While regions of the country. This is just one of three concurrent the increase in poverty in the Central Highlands in 2014 NTPs that will be implemented over the 2021–2025 SEDP appeared to be a one-time occurrence, the Mekong Delta phase (See Chapter 3 for more discussion on NTPs). Poverty is facing continuous stresses and challenges. In 2020, the reduction budgets include a much broader set of programs region suffered a dual crisis, from droughts and COVID-19. than simply cash support, including public investments, Concurrently, both consumption and income based poverty infrastructure, and other indirect expenditures. With well- rates increased in this region. The region has also received defined targeting, the government could afford the amount attention in the past years as a source of climate migrants. In of direct cash support needed to raise immediate cash flows 2020, migrants from the Mekong Delta accounted for about to poor households to a minimum level, and also continue to 37 percent of all migrants in the country (GSO, 2021). See invest in medium- and long-term development projects. Chapter 6 for a discussion on resilience and the intersection between poverty and natural hazards. 1.3. Subnational poverty Small area poverty estimates at the district level trends18 reveal more heterogeneous variation in the success of poverty reduction The traditionally poor regions in Vietnam are the Central Highlands and the Midlands and Northern Mountains, Measuring poverty at the regional level is insufficient. where poverty rates were the highest throughout the Prominent inter-regional differences call for more granular decade (Figure 1.5). The Central Highlands region stands out statistics of poverty and welfare. For example, the Southeast as a region where poverty is decreasing but the number of poor region comprises only six provinces and municipalities, are not declining as quickly. Poverty also increased in 2014 in including Ho Chi Minh City (HCMC), and is the smallest the Central Highlands, but this was temporary, and poverty by land size. Moreover, HCMC accounts for more than half continued to decline consistently into 2020. However, the of the population of the region. However, the Midlands and number of poor in the Central Highlands declined by about Northern Mountains region is much larger by area (even only half a million people over the last decade, the smallest if it is small by population) and has the largest variation in amount across the six regions (Figure 1.6). Slower poverty provincial poverty based on the multidimensional index, reduction progress in this region also affected its relative ranging from 2.8 percent in Bac Giang to 36.7 percent in Dien 30 2022 Vietnam Poverty and Equity Assessment Chapter 1. Taking stock of welfare trends over a dynamic decade Bien in 2020. A similar observation is made for the Northern decisions are made at the local level, which requires a local- and Coastal Center region, which includes Da Nang city. Da level understanding of poverty trends. Over the decade, as Nang city on its own is a vibrant economic area but is part of the populations and settlements grew, some districts have the larger Northern and Central Coast area, characterized by been split. There were 27 more districts in 2019 than in 2009 varying levels of economic activity. (Table 1.4). (See Box 1.3 for additional history on the World Bank-developed small area poverty estimation tools.). For over a decade, the World Bank has worked with the General Statistics Office (GSO) on producing small area A comparison of small area poverty maps over a decade estimates of poverty at the district level. This level of shows overall progress, but pockets of chronic poverty disaggregation is extremely powerful, greatly enhancing the remain. Figure 1.7 illustrates district-level poverty maps from understanding of spatial poverty and dynamics since direct 2009, 2015, and 2019. The 2009 map shows initially high estimates of poverty from the household survey are only concentrations of poverty throughout the majority of the representative across six broad regions; the country had 63 high northern mountains and central regions of the country. provinces and over 700 districts in 2019. Many policies and Over the decade, poverty has declined impressively but has not Figure 1.5. Poverty rates ($3.20/day 2011PPP) , by region 50 39.9 40 34.0 32.8 Poverty rate (%) 29.6 30 25.8 23.4 27.6 22.0 18.6 17.1 20 13.6 20.4 19.7 14.3 12.1 16.0 11.5 9.0 10 6.8 6.2 8.6 4.8 4.7 3.6 4.5 5.3 1.4 2 3.7 0 3.0 2.3 0.6 0.4 0.4 1.2 2010 2012 2014 2016 2018 2020 Central Highlands Mekong Delta Midlands & Northern Mountains Northern & Coastal Center Red River Delta Southeast Source: World Bank staff calculations using VHLSS. Figure 1.6. Number of poor ($3.20/day 2011PPP) , by region 6 Number of poor (in millions) 5 4 3 2 1 0 Central Highlands Mekong Delta Midlands & Northern Northern & Red River Delta Southeast Mountains Coastal Center 2010 2012 2014 2016 2018 2020 Source: World Bank staff calculations using VHLSS 2010–2020. From the Last Mile to the Next Mile 31 part 1. a deCade of signifiCant progress, BUt last-mile Challenges remain disappeared in some lagging regions such as the Northwest Across all districts, wealthier districts with the lowest initial and Central Highlands. The region that has achieved the poverty rates in 2009 exhibited more poverty reduction when largest absolute poverty reduction is the Northeast, which has measured in percent terms (these districts had lower absolute benefited from expanding industrial activity (see Chapter 6 changes in poverty rates due to their low starting point). Thus, for bivariate maps of poverty and environmental factors). poverty reduction among chronically poor areas continues to be a challenge. Regions with districts exhibiting the slowest A scatter plot of district-level poverty rates highlights poverty reduction in precent terms include the Central areas with the slowest poverty reduction progress. Highlands, Northwest, and Mekong Delta (Figure 1.8). Table 1.4. Summary of a decade of small area poverty estimates YEAR OF ESTIMATE VHLSS CENSUS POVERTY LINES NUMBER OF DISTRICTS IN MAP 2009 2010 2009 • GSO-WB Poverty Line 685 Census • $1.25/day 2005PPP 2015 2014 2015 • GSO-WB Poverty Line 699 Inter-census • $1.90/day 2011PPP • $3.10/day 2011PPP 2019 2020 2019 • GSO-WB Poverty Line 712 Census • $3.20/day 2011PPP Note: An interim LMIC poverty line valued at $3.10/day was used before a finalized line was valued at $3.20/day (Joliffe and Prydz, 2016). Source: World Bank (2012), World Bank (2016), and Nguyen and Yang (2022) Figure 1.7. District level estimates of poverty 2009 2015 2019 Poverty rate (%) 0.0 100.00 Note: Poverty rates based on the GSO-WB poverty line. The consumption aggregates are comparable across all years. Source: World Bank staff and GSO calculations using Census and VHLSS. 32 2022 Vietnam Poverty and Equity Assessment Chapter 1. taking stoCk of welfare trends over a dynamiC deCade Figure 1.8. District level poverty comparison, 2009 vs. 2019 Absolute change in poverty rate (percentage point) Percent change in poverty rate (%) 0 Absolute reduction in poverty (percentage point, 2009–19) Percentage reduction in poverty (%, 2009–19 -10 0% -20 -30 -50% -40 -50 -60 -100% 0 20 40 60 80 100 0 20 40 60 80 100 Poverty rate (2009, %) Poverty rate (2009, %) Change in population of poor Central 100K Mekong River Delta North Central Coast North East 80K North West Population of poor (2019) Red River Delta 60K South Central Coast Southeast South 40K 20K 0K 0K 20K 40K 60K 80K 100K Population of poor (2009) Source: World Bank staff and GSO calculations using Census and VHLSS. Progress can also be visualized by the distribution of the national average has increased from 2015 to 2019. This districts that have poverty rates below or above the is consistent with rising regional poverty rates in the Mekong national average (Figure 1.9). The Northwest and Central Delta in 2020. In contrast, the share of districts with higher- Highlands remain the primary areas where progress is lagging. than-average poverty rates in the Northeast Mountains region Nearly 90 percent of their districts in these two regions in declined considerably in the latter half of the decade. This 2019 had poverty rates higher than the national average. The area is part of the Midlands and Northern Mountains region Mekong Delta has a low level of poverty compared to other and is the source of its regional poverty reduction owing to an regions, but the share of districts with poverty rates above influx of manufacturing jobs. From the Last Mile to the Next Mile 33 PART 1. A DECADE OF SIGNIFICANT PROGRESS, BUT LAST-MILE CHALLENGES REMAIN Figure 1.9. Share of districts with poverty rates above the national average, 2019 100 80 Share of districts (%) 60 40 20 0 North West North East Red River North Central Central South Central Southeast South Mekong River Delta Coast Highlands Coast Delta 2009 2014 2019 Note: Poverty rates based on the GSO-WB poverty line. The consumption aggregates are comparable across all years. Source: World Bank staff and GSO calculations using Census and VHLSS. BOX 1.3. A history of small area poverty estimates Poverty and welfare indicators are measured from household surveys that only sample selected enumeration areas. These samples, while carefully selected, can only be used to estimate poverty statistics at geographic levels that are representative in the surveys. Since household survey samples are small and do not have the coverage of a Census, poverty rates cannot be estimated in small areas. The methodology to produce small area estimates of poverty has a long and fruitful history with several notable updates. To solve the small and missing sample problem, Elbers et al. (2003) originally proposed a so-called “small area estimation” method to estimate poverty measures at small areas by combining a household survey and a census. This method is widely applied to estimate disaggregated poverty and welfare measures, especially in developing countries (see Bedi et al., 2007, and Bigman and Fofack, 2000, for review of applications). ELL’s estimates can be computed using the World Bank’s PovMap software. An intermediate theoretical update involving sample conditionality and sampling weights was applied in 2014 and accompanied an update to the PovMap software. New improvements have been made to the method to increase precision and reduce bias. Most recently, Corral, Molina, and Nguyen (2021) introduced a new method that improves the precision and reduces bias in small area estimates using updated Monte Carlo and imputation techniques. These estimates are referred to as the Census-EB estimates and can be produced using the World Bank-developed sae Stata package. Table B.1.3.1. Evolution of small area poverty estimation techniques ESTIMATES ELL (2003) H3- CBEB (2014) CENSUS EB (2021) Notes Original theoretical paper with Conditioned on survey sample available, Assumes survey data is not a estimates based on multiple and results are more precise than ELL. subsample of the census; estimates are imputation Added survey weights following Molina/ based on Monte Carlo and multiple Rao (2010) imputation techniques World Bank- PovMap software PovMap software update v2.5 / sae Stata sae Stata package developed tools package 34 2022 Vietnam Poverty and Equity Assessment Chapter 1. Taking stock of welfare trends over a dynamic decade 1.4. Grouping households More people became economically secure from 2010 to into economic classes 2020 than between 1993 and 2008. The size of the middle class also tripled from 2010 to 2020, increasing from 7.3 to monitor changes million people in 2010 to 22.6 million in 2020 (Figures 1.10 across the entire welfare and 1.11). Growth occurring at higher and higher levels of income illustrates a population whose consumption patterns, distribution aspirations, and living conditions are rapidly changing. A more complete view of household economic transitions A disaggregated view by region reveals that most of the and dynamics can be obtained by examining population expansion of the middle class occurred in the Red River groups defined by intervals of consumption, or economic Delta and Southeast regions (Figure 1.12). These regions classes. The intervals of consumption pertaining to economic are where the country’s largest cities are located (Hanoi and classes are purely based on household consumption per HCMC). The Southeast region is the only region where the capita, without regard to social class. Examining progress by middle class is growing and the absolute number of people different groups is also useful since groups in different welfare in other economic classes is declining, illustrating high rates brackets benefit from different types of support and policies. of absolute upward economic mobility. While the Red River Managing prosperity and stability requires supporting not Delta has an increasing share of the middle class, the absolute only the poor, but also the aspirations of the middle class. To population in other economic groups is still similar in number categorize households at higher ends of the distribution, this to a decade ago. This may be due to increasing migration, report uses definitions of economic classes from the World which is also quickly increasing the region’s population. Bank’s East Asia and Pacific regional report (see Box 1.4 for The poorer regions exhibited smaller expansions of higher definitions of economic classes). economic classes. The populations of extreme poor, moderate poor, and economically vulnerable have remained relatively Poverty reduction was not only remarkably successful constant in the two poorest regions. over the last decade, but the middle class grew quickly as well. Over the last decade, upward economic mobility occurred even faster for households at higher income levels. Figure 1.10. Distribution of economic classes Figure 1.11. Distribution of economic classes (share of the population) (population in millions) 100 100 80 80 Population (in millions) Share of the population 60 60 40 40 20 20 0 0 2010 2012 2014 2016 2018 2020 2010 2012 2014 2016 2018 2020 Extreme poor Moderate poor Extreme poor Moderate poor Economically vulnerable Economically secure Economically vulnerable Economically secure Middle-class Middle-class Source: World Bank staff calculations using VHLSS 2010–2020. From the Last Mile to the Next Mile 35 PART 1. A DECADE OF SIGNIFICANT PROGRESS, BUT LAST-MILE CHALLENGES REMAIN Figure 1.12. Population by economic class and by region, 2010–2020 25 20 Population (in millions) 15 10 5 0 2010 2012 2016 2018 2020 2010 2012 2016 2018 2020 2010 2012 2016 2018 2020 2010 2012 2016 2018 2020 2010 2012 2016 2018 2020 2010 2012 2016 2018 2020 2014 2014 2014 2014 2014 2014 Midlands & Red River Delta Central Highlands Northern & Southeastern area Mekong River Northern Mountains Central Region Delta Extreme poor Moderate poor Economically vulnerable Economically secure Middle-class Source: World Bank staff calculations using VHLSS 2010–2020. BOX 1.4. World Bank East Asia and Pacific regional economic class groups To highlight the evolution of households along the entire welfare distribution, this paper uses the economic class definitions defined in the World Bank’s Riding the Wave report (World Bank, 2018b). These economic class thresholds were developed for economies in the East Asia and Pacific region based on global poverty lines. Lower economic classes are bounded by the World Bank’s $1.90, $3.20, and $5.50/day 2011PPP poverty lines. The middle-class threshold of $15.00/day 2011PPP is similar in value to other thresholds for large regional analyses (Ferreira et al., 2013; World Bank, 2018b). In high-income countries, middle-class lines can be even higher (e.g., $50.00/day), but upper ends of the welfare distribution do not translate well in household surveys, where the tail distributions are the most problematic. The World Bank’s regional and global poverty thresholds are intended to be used for cross-country comparisons, but these thresholds are also suitable for examining the welfare story of Vietnam. While some of the lower-level World Bank poverty lines are comparable in monetary value to the MOLISA monetary dimension of poverty thresholds, these definitions do not reflect government definitions of economic classes. Rather, they are created to help illustrate the discussion on welfare distribution dynamics. Table B.1.4.1. Summary of economic class definitions DEFINITION VALUE IN 2011 APPROXIMATE OF ECONOMIC PPP PER DAY NOTES VALUE VND [1] CLASSES AND PER CAPITA Extreme poor < $1.90 < 626.5 This threshold is the World Bank’s international poverty line. Moderate poor ($1.9$-3.2) 626.5 - 1,055 The upper threshold is the moderate poverty line traditionally used by the World Bank in analyzing trends in developing East Asia and Pacific. Economically vulnerable ($3.2-$5.5) 1,055 – 1,813 Between the World Bank’s LMIC and UMIC poverty lines Economically secure ($5.5-$15) 1,813 – 4,946 Above the World Bank UMIC poverty line, but not yet middle class (see below). Middle class $15+ 4,946+ Those living on more than US$15.00 a day. This threshold is broadly consistent with the values used by other studies (World Bank, 2018b). [1] Units: Jan 2020 VND in thousands, monthly, and per capita. Summarized from World Bank (2018b). 36 2022 Vietnam Poverty and Equity Assessment Chapter 1. Taking stock of welfare trends over a dynamic decade 1.5. Upward economic The majority of households, 65–70 percent, stayed within mobility within a lifetime the same economic class over a two-year period. For some, changes in consumption were large enough to cross economic for most class groups. On average over the various selected periods, about 20 percent of households moved up into a higher This section utilizes the panel sub-sample of the VHLSS economic class, but about 15 percent slid down (Figure over available two- and four-year periods to better describe 1.14). Between 2016 and 2018, 38 percent of the population economic transitions across time and their drivers (see Box 1.5 changed economic classes, with 20 percent moving to a higher for information on the VHLSS panel). economic class. Upward economic mobility was stronger over a four-year period, but increases that moved a household up Absolute changes in household consumption two economic classes were rare. over the decade Stronger upward economic transitions at the Upward economic mobility was widespread, as most bottom, but less stability at the top households experienced increasing consumption. Across various two-year periods, the share of panel households By the end of the decade, falling back into poverty was experiencing increasing real consumption ranged from 57 less common. Between 2016 and 2018, less than 2 percent percent (2012–2014) to as high as 70 percent (2014–2016) of the nonpoor fell into poverty, half the likelihood (Figure 1.13). Increases in consumption were stronger over a from 2010 to 2012 (Figure 1.15.A). Even though fewer longer four-year period, as might be expected. From 2014 to households at the bottom of the distribution experienced 2018, almost three-quarters of the panel sample had greater absolute increases in household consumption toward the end absolute consumption. These transitions of households of the decade, absolute minimum living standards had already illustrate a more mobile society in the short run, but more improved, and the risk of falling into extreme poverty was low. stable absolute gains over longer horizons. However, despite the low rates of people sliding into extreme or moderate poverty, this still amounted to about 640,000 Changes in household consumption were large enough households in 2018. Therefore, managing the risk of people for about one-third of households to move into another falling back into poverty cannot be neglected. economic class, with most but not all moving upward. Figure 1.13. Increases in consumption in panel households observed over two-year and four-year periods Two-year periods Four-year periods 80 80 70 70 Share of panel households (%) Share of panel households (%) 60 60 50 50 40 40 30 30 20 20 10 10 0 0 2010–2012 2012–2014 2014–2016 2016–2018 2010–2014 2012–2016 2014–2018 Notes: Share of panel households experiencing an increase in real consumption per capita. Source: WB calculations using VHLSS panel From the Last Mile to the Next Mile 37 part 1. a deCade of signifiCant progress, BUt last-mile Challenges remain Churning at the higher end of the distribution was Better jobs were key pathways to upward more evident at the end of the decade. Later in the decade economic mobility over the last decade (2016–2018), while the likelihood of falling into poverty had substantially diminished, most of those who slid were at the Overall, location, economic activity, and education top. Maintaining middle-income status remains challenging, are all strong determinants of increasing welfare and at least over short time horizons. Close to a third of the middle- upwards economic mobility. Based on time-invariant class slid to a lower class during 2016–2018, with most sliding characteristics, households in urban areas, Kinh majority, and down just one class (Figure 1.15.B). Southeast or Northern and Central coast regions were most likely to experience increasing welfare between 2016 and 2018 that was large enough to move them to a higher economic Figure 1.14. Share of panel households moving up or sliding down an economic group 100% Share of panel households (%) 80% 60% 40% 20% 0 2010–2012 2012–2014 2014–2016 2016–2018 2010–2014 2012–2016 2014–2018 Two-year period Four-year period Slid down Same Moved up Source: World Bank staff calculations using VHLSS panel. Figure 1.15. Economic class transitions over a two-year period A. 2010–2012 Transitions B. 2016–2018 Transitions 1-Middle class 1-Middle class 1-Middle class 1-Middle class 2-Economically 2-Economically 2-Economically 2-Economically secure secure secure secure 3-Economically 3-Economically vulnerable vulnerable 3-Economically 3-Economically vulnerable vulnerable 4-Moderate poor 4-Moderate poor 4-Moderate poor 4-Moderate poor 5-Extreme poor 5-Extreme poor 5-Extreme poor 5-Extreme poor Notes: The extreme poor, living on less than the international poverty line ($1.90/day, 2011 PPP); The moderate poor, living on $1.90 to $3.20 / day (2011 PPP); The economically vulnerable, living on $3.20 to $5.50 a day (2011 PPP); The economically secure, living on $5.50 to $15 a day (2011 PPP). The middle class, living on more than $15 a day (2011 PPP). Source: World Bank staff calculations using VHLSS panel. 38 2022 Vietnam Poverty and Equity Assessment Chapter 1. Taking stock of welfare trends over a dynamic decade class (Table 1.5). The location of households is a primary Challenges exist for sustainable upward determinant of the likelihood of experiencing improving economic mobility economic conditions related to access to better economic opportunities. Stable and sustainable upward economic Among some groups, poverty reduction still faces mobility requires not just access to jobs but also to good jobs challenges of sustainability and resilience. Traditionally that can provide economic security. Groups that experienced poor regions experience high rates of churning—that is, high high rates of climbing and low rates of sliding are those rates of both upward (climbers) and downward (sliders) which exited agriculture completely or gained employment transitions—tempering net overall poverty reduction. with a formal contract. Households that have lower than Characteristically poor groups such as households in the average rates of sliding include those that were not engaged Midlands and Northern Mountains and ethnic minorities in agriculture over the entire period, exited agriculture, or experienced rates of climbing and sliding that were higher worked in manufacturing (see Chapter 2 for more discussion than the national average. This complicates persistent poverty on poverty reduction drivers). alleviation since not necessarily the same households are chronically poor from year to year (see Chapter 3 for a review of groups with chronically high poverty rates). Table 1.5. Economic mobility transitions from 2016 to 2018, by household characteristics % BOX 1.5. VHLSS Panel sub-sample STAY CLIMB SLIDE National 66.0 21.4 12.6 Across the biennial household expenditure modules, Geographic area (time invariant) about half of the sample are re-interviewed and half are Central Highlands 59.5 24.2 16.3 rotated. Furthermore, a quarter of the samples are re- Mekong Delta 67.7 20.1 12.2 interviewed over three consecutive rounds spanning four Midlands and Northern 57.4 26.3 16.4 years. The VHLSS 2010–2018 samples were drawn from Mountains a master sample frame based on the 2009 Census. Since Northern and Coastal Center 64.8 23.6 11.7 the 2020 VHLSS was sampled from the 2019 Census, a panel does not exist for this year. Red River Delta 71.4 17.7 10.9 Southeast 68.6 19.3 12.1 Table B.1.5.1. Number of panel households Rural 64.2 23.2 12.7 (over two-year periods) Urban 70.3 17.2 12.5 2012 2014 2016 2018 Ethnicity (time invariant) 2010 4,231 1,951 Kinh majority 69.3 19.4 11.4 2012 4,226 1,895 Ethnic minority 49.9 31.2 18.9 2014 4,275 1,936 Employment characteristics and dynamics 2016 4,281 No participation in agriculture 72.5 17.2 10.3 Exited agriculture in 2018 62.9 26.9 10.3 The selection of the panel sub-sample appears well Entered manufacturing in 2018 61.8 20.7 17.5 randomized. Since the panel does not follow migration or split households, there may be attrition bias. However, Pimhidzai Has formal contract in 2016 73.3 15.4 11.3 et al (2019) compared outcomes between the rolling panel Gained formal contract in 2018 61.3 29.3 9.4 sub-sample and the full sample and did not find evidence of Note: Climbers are households that are in a higher economic class in 2018 such a bias. than in 2016. Sliders are households in a lower economic class in 2018 than in 2016. Stayers are in the same economic class in 2018 as in 2016. Economic class definitions are based on World Bank global definitions as defined in Box 1.4. Source: World Bank staff calculation using VHLSS panel sub-samples. From the Last Mile to the Next Mile 39 PART 1. A DECADE OF SIGNIFICANT PROGRESS, BUT LAST-MILE CHALLENGES REMAIN 1.6. Trends across the relative to the rest of the East Asia and Pacific (EAP) region, welfare distribution based on international price comparisons (PPPs), also stretched purchasing power. With low taxation and relatively inexpensive living costs such as rent or mortgages, these rising incomes The concepts of poverty reduction, inequality, and were channeled into increasing household discretionary inclusive growth are connected. The literature notes consumption. At the beginning of the decade, households the importance of the impact of inequality on poverty at the lower ends of the welfare distribution enjoyed higher- reduction, highlighting a double-dividend effect (Alvaredo than-average growth rates (Figure 1.16). Higher growth rates and Gasparini, 2015; Bourguignon, 2004). At a macro level, at the bottom of the distribution were evident between 2010 changes in poverty can be decomposed into a combination and 2014. In later periods (2014–2016, 2016–2018, and of growth and redistribution effects (Bourguignon, 2003, 2018–2020), the highest rates of growth were observed among 2004; Datt and Ravallion, 1992; Ferreira, 2012). Through a households at the higher ends of the distribution. growth effect, growth in mean income or in consumption can drive poverty reduction. The second channel is redistribution The cost of maintaining decent living standards is or through changes in inequality. Reducing inequality has a increasing. Rising expenditures also reflect an increase in the double-dividend effect because it promotes poverty reduction cost of living, changes in what is considered to be necessary today and accelerates poverty reduction in the future by to sustain a quality life in a richer country, and redefined breaking cycles of intergenerational poverty traps. Lower concepts of minimum standards. As is to be expected in a levels of inequality have been empirically associated with rapidly growing economy, the costs of basic necessities begin to higher growth elasticity of poverty reduction; however, the accumulate with higher utilities, transportation needs, health empirical evidence is not conclusive (Bergstrom, 2020). expenditures, and housing costs. Households are spending a larger share of their income on non-food items, reflecting Consumption growth was consistently high over higher amounts of discretionary income. Expenditures on the decade, but in recent years it is higher among housing,19 durables, education, health, and other non-food the rich items are all increasing across the entire welfare distribution. Ownership in durables and assets is also increasing, Higher rates of consumption growth has shifted from consistent with higher expenditures. However, ownership the lower to the upper ends of the distribution. On of some higher-end items such as cars and computers is still average, household consumption growth was strong at about uncommon or limited to the wealthiest households. This 5 percent per year over the entire decade. A low cost of living is also seen through changes in household assets, where Figure 1.16. Household consumption growth across the welfare distribution 8 6 Annual growth (%) 4 2 0 2010–2020 2010–2012 2012–2014 2014–2016 2016–2018 2018–2020 Note: Growth incidence curves illustrate the trends in annualized household consumption growth per capita across different points of the welfare distribution. Due to the small sample size of the household survey, and also less precision in the tails, these curves are constructed from 10 points, each representing a welfare decile. Source: World Bank staff calculations using VHLSS. 40 2022 Vietnam Poverty and Equity Assessment Chapter 1. Taking stock of welfare trends over a dynamic decade ownership of landline phones and bicycles are decreasing, but distribution (bottom 40 percent) rather than on average or cell phone and motorcycle ownership is increasing. Diversity among those at the top. In recent periods, the growth of the in food consumption has also increased, with more purchases bottom 40 percent of the population has been lower than the notably in dairy and eating out. There is still some variation national average (Figure 1.17). As with the growth incidence across geographic areas, with the poorest regions spending curves, average consumption growth for the poorest 40 the largest part of their food consumption on grains, while percent grew more quickly than for the richest 60 percent wealthier regions have higher shares of food consumed as from 2010 to 2014 before falling behind. In the 7th edition meat or eating out. of the Global database of shared prosperity, Vietnam was the only country in the East Asia and Pacific region sample Measures of dispersion illustrate widening where the shared prosperity premium is negative, of where the inequality growth of the bottom 40 is lower than average growth across all households (Figure 1.18). Inequality in household consumption can be measured in a variety of absolute and relative ways. These measures Relative inequality has remained remarkably stable include ratios, standard deviations, Gini coefficients, and over the past decades, but began to increase again in absolute gaps. Across these measures, there is a similar story to the latter half of the decade. The Gini Index was 35.6 in what was illustrated using the GICs; household consumption 1993 and 36.8 in 2020 (Figure 1.19). Between 1993 and 2018, between the least and most well-off groups are widening in the only instance when inequality increased substantially recent years. in Vietnam was in 2010 near the global financial crisis, but it quickly reverted in 2012. In the first half of the decade, Promoting growth of the bottom 40 percent of the inequality declined, as growth was strongest at the bottom population is one of the World Bank’s twin goals. of the welfare distribution. But inequality slowly increased in The Shared Prosperity indicator measures the extent to the second half of the decade, and more noticeably in 2020. which economic growth is inclusive by focusing on income The absolute difference in annual per capita consumption growth rates of the population at the bottom of the income between the poorest and richest 10 percent has also widened Figure 1.17. Comparing growth at the Figure 1.18. Vietnam’s shared prosperity bottom and top of the distribution vs other East Asia countries 8 10 7 8 Percent or percentage point Annualized growth rate (%) 6 6 5 4 4 3 2 2 0 1 0 −2 Philippines Indonesia Mongolia Thailand 2013–16 2014–18 2011–18 2012–16 2012–15 2014–18 2012–18 Malaysia Vietnam 2010–2012 2012–2014 2014–2016 2016–2018 2018–2020 China Growth B40 Growth T60 Growth Mean Shared Prosperity (%) Shared Prosperity Premium (percentage point) Note: Measures based on real household consumption per capita. Source: World Bank staff calculations using VHLSS 2010–2020. Source: Global database of shared prosperity, seventh edition, World Bank 2020. From the Last Mile to the Next Mile 41 PART 1. A DECADE OF SIGNIFICANT PROGRESS, BUT LAST-MILE CHALLENGES REMAIN considerably (Figure 1.20). From 2010 to 2020, average The much higher consumption of the top 10 percent household consumption per capita among the bottom 10 is also seen across consumption sub-components percent of population increased from 7.4 to 12.1 million (Figure 1.21). While household expenditures are increasing VND per month. However, for the richest 10 percent of the incrementally across the first nine welfare deciles, the top 10 population, the increase was from 52.9 to 136 million VND percent of households have about double the expenditure in per month over the same period. some categories as the 9th decile, such as durables and rent. This illustrates a large gap among the fortunes of the top 10 Differences in consumption along the percent compared to the rest of the population. These gaps distribution in the same year also reveal much existed in 2010 as well but are wider in 2020. Thus, absolute faster growth at the top of the distribution gaps and inequality are also increasing. Figure 1.19. Relative inequality, Gini Index Figure 1.20. Widening absolute inequality 40 150 per capita (monthly VND, millions) Average household consumption 35 120 Gini index 90 30 60 25 30 20 0 2006 2008 2012 2014 2016 2018 2020 1993 1998 2002 2004 2010 2010 2012 2014 2016 2018 2020 Bottom 10 Bottom 40 Top 60 Top 10 Note: Measures based on real household consumption per capita. Poverty Note: Measures based on real household consumption per capita. measurement before and after 2010 is not comparable. Source: PovcalNet. Source: PovcalNet. Figure 1.21. Expenditure levels, 2020 60 Annual household consumption per capita 50 40 (VND, millions) 30 20 10 0 Food Other Water Electricity Waste Health Education Durables Rent 1 (=Poorest) 2 3 4 5 6 7 8 9 10 (=Richest) Source: World Bank staff calculations using VHLSS. 42 2022 Vietnam Poverty and Equity Assessment Chapter 1. Taking stock of welfare trends over a dynamic decade 1.7. How new across the country. Residents in these urban areas experienced developments from lockdowns that lasted several months. Industrial zones also reduced output, and the economic vulnerabilities of migrant COVID-19 have impacted laborers were also highlighted. poverty projections and Poverty projections using economic growth expectations inequality after the arrival of the Delta variant show stagnation in poverty reduction in 2021. Figure 1.22 compares changes Before the emergence of the Delta variant in April in poverty forecasts based on growth forecasts from March 2021, Vietnam was expected to enjoy a “V-shaped” and October 2021, and including scenarios with different recovery, with growth back to pre-COVID levels by distributional assumptions. Projections made in March 2021, Q3 2021. However, the COVID-19 context dramatically before the arrival of more transmissible variants, suggested that and unexpectedly worsened in April 2021, when cases poverty reduction would stay on track, but higher inequality began to accumulate quickly. Within a month the number could slow the pace of poverty reduction. Downgraded of COVID-19 cases in Vietnam surpassed the total number growth projections made in October 2021 estimated are that from the entire previous year. Cases were concentrated in poverty will be about a half percentage point higher—4.7 vs. the HCMC area and surrounding industrial zones at first 4.2 percent—in 2021 under the $3.20/day 2011PPP poverty and remained largely contained until November, when line (and under the scenario without increasing inequality) movement restrictions eased, and then spread more widely (Figure 1.22). However, this is most likely a lower-bound Figure 1.22. Distribution-sensitive poverty projections for Vietnam, 2018–2023 No change in Gini +1% change in Gini +2% change in Gini 6.6 6.6 6.6 Poverty rate (%, $3.20/day 2011PPP) 6 5.2 4.9 4.9 4.7 4.6 5.0 5.0 4.3 5.0 4.0 4.7 4 4.4 4.4 4.2 3.7 4.1 3.4 3.9 3.5 3.5 3.1 2 0 22.4 22.4 22.4 Poverty rate (%, $5.50/day 2011PPP) 20 19.2 18.1 18.5 18.2 18.8 18.8 18.8 17.0 16.2 17.8 17.0 17.2 16.9 16.9 15.1 15.9 15 15.5 14.9 14.1 14.3 13.2 10 2018 2020 2022 2024 2018 2020 2022 2024 2018 2020 2022 2024 Growth estimates (version) March 2021 growth estimates October 2021 growth estimates Note: Poverty rates in 2018 and 2020 are based on survey data. Poverty rates in 2021 and onwards are projections. Simulations of changes in inequality, if any, are assumed to start in 2021. Distributionally sensitive simulations methods are based on Lakner et al. (2020). See Annex 1.4 for details. Source: World Bank staff estimations using VHLSS, 2018, 2020. From the Last Mile to the Next Mile 43 PART 1. A DECADE OF SIGNIFICANT PROGRESS, BUT LAST-MILE CHALLENGES REMAIN estimate of the increase in poverty since this estimate assumes Inequality in both monetary and non-monetary no change in inequality. Inequality is expected to increase dimensions can be expected to increase during the because households that are least able to cope and have COVID-19 pandemic for a range of reasons. In terms intermittent sources of income are the most impacted during of coping during COVID-19, poor households were more lockdowns and a period of reduced retail spending (World reliant on external sources such as borrowing, while rich Bank, 2021b). Under assumptions of a 1 percent increase households were better able to cope using their own resources in inequality, poverty rates in 2021 are expected to stagnate. such as tapping into savings (World Bank, 2021b). Women Moreover, an unprecedented economic slowdown in 2021 bear a larger share of caregiving responsibilities, and their labor coupled with a 2 percent rise in inequality could potentially market activities were more adversely impacted than men’s. lead to an increase in poverty. Informal workers are the least inclined to have safety nets and experienced the most challenges when seeking government The adverse impact of inequality on poverty reduction cash support. The continuity of education was uneven during can be just as large as the impact of slower growth (Figure COVID-19, and the pandemic has potentially widened gaps 1.23). Small increases in inequality can slow poverty reduction, in human capital formation because of the uneven capacity especially when accumulated over multiple years. A downgrade of schools across the country. The future is digital, but there in the GDP growth forecast increases poverty estimates 0.5 are gaps in digital use and inclusion. Wealthier households are percentage points in 2021. However, a 1 percent increase in the more able to participate in the digital economy as both sellers Gini coefficient would increase poverty about 0.4 percentage and buyers on digital platforms. points, and a 2 percent increase in the Gini coefficient would increase poverty in 2021 by 0.5 percentage points. Globally, Increasing inequality can have long-term implications. it has been found that a 1-point decrease in the Gini index in Inequities can have long-term consequences: lost education is every country would lower global poverty by more than a 1 unlikely to be recovered, with consequences for lifetime wages; percentage point increase in GDP per capita (Lakner et al., sold assets cannot produce future income; and employment 2020). Bergstrom (2020) also finds that a 1-point reduction scarring is also associated with lower lifetime earnings. Larger in inequality leads to a more than 1 percent increase in mean businesses and wealthier households were also able to make incomes, based on global analysis. While recovering growth investments to reap larger sales from digital orders, which can narrow the differences in poverty projections by 2023 may lead to widening inequality down the road. Minimizing under scenarios of low changes in inequality, larger changes in future disparities will require forward-looking policies and inequality would lead to longer-term increases in poverty. improving existing support systems. Figure 1.23. Impact of changes in growth and inequality on poverty projections 1.8 Difference in $3.20/day 2011PPP 1.5 poverty rate (p.p) 1.2 0.9 0.6 0.3 0.0 Growth downgrade 1% increase in 2% increase in Growth downgrade Growth downgrade inequality inequality + 1% increase in Gini + 2% increase in Gini 2021 2022 2023 Source: World Bank staff estimations using VHLSS, 2020. 44 2022 Vietnam Poverty and Equity Assessment Chapter 1. Taking stock of welfare trends over a dynamic decade 1.8. Summary differences in child outcomes in education and health. The trends in human capital development are explored across The objective of Chapter 1 is to document broad trends ethnic, geographic, and socioeconomic groups to determine in poverty and inequality over the last decade (2010– existing inequalities. The dissimilarity in outcomes of these 2020), as well as provide poverty projections under new groups is decomposed and quantified to be attributable to and evolving COVID-19 developments. This introduction characteristics such as geography, household characteristics, provided a high-level assessment and intentionally deferred and expenditure quintile. some questions, which will be addressed in later chapters. Where to find answers to key questions is described below. Pathways for sustaining upward economic mobility (Chapter 5) What were the drivers of poverty reduction over the last decade? (Chapter 2) Over the past decade, growth in Vietnam has been inclusive, leading to a net reduction in poverty rates, The drivers of poverty reduction are overwhelmingly with many moving out of poverty in their lifetimes. attributable to growth factors throughout most of the This chapter discusses what the upward economic mobility decade. Chapter 2 describes broad labor market changes channels of the future may look like, and it highlights potential over the last decade and the evolution of household profiles challenges for Vietnam’s economic transition and upward and incomes in more detail. With respect to the drivers of economic mobility going forward. These challenges include poverty reduction, group and income decompositions are high labor informality, low job quality, low labor productivity, used to quantify and explain broad contributing factors from the need for upskilling, and an ageing population. household profiles or income sources. What are the risks to households falling into Why is poverty reduction so challenging for poverty? (Chapter 6) certain groups? (Chapter 3) COVID-19 has revealed a wider population of vulnerable There is an interplay between ethnicity, agriculture, groups. These groups risk falling into poverty traps when geography, and chronic poverty in Vietnam. Remote exposed to shocks, without the provision of adequate safety mountainous areas where the poor are concentrated are nets. Chapter 6 discusses the broader range of vulnerable heavily populated by ethnic minorities, who are also heavily groups by the different types of shocks they face, including engaged in agriculture. Chapter 3 further distills the observable covariate and idiosyncratic shocks. Covariate shocks are differences in poverty reduction across these chronic caused by large events that are felt by many households (e.g., dimensions and some reasons behind lagging progress. For climate change), while idiosyncratic shocks are those that may the poorest households, agriculture is more often the largest be unique to the household (e.g., accident). The chapter also source of household income, but agricultural incomes have explores the relationship between poverty, vulnerability, and also shown little growth over the decade. risk and its policy implications. It concludes with a discussion of Vietnam’s social protection system and preliminary insights Human capital and poverty reduction (Chapter 4) on how it can be improved to better guard against shocks. Human capital is a significant driver of sustained How can public financing be used effectively for economic growth and poverty reduction. Consisting of the poor? (Chapter 7) a combination of education, skills, and health factors, human capital formation in early life is a significant determinant All households pay taxes (whether personal income of social and economic outcomes. High-quality education taxes or indirect taxes such as value-added tax (VAT) and helps break intergenerational cycles of poverty by increasing excise taxes), and all benefit from some public spending labor productivity and growth. While education has (whether social assistance transfers, electricity subsidies, improved significantly in Vietnam, parental and household or public health and education). For some households the characteristics are still highly explanatory of remaining taxes are greater than the benefits, meaning they are net payers From the Last Mile to the Next Mile 45 PART 1. A DECADE OF SIGNIFICANT PROGRESS, BUT LAST-MILE CHALLENGES REMAIN Elbers, C., Lanjouw, J. O. [Jean O], & Lanjouw, P. (2003). Micro–level into the fiscal system, while others are net beneficiaries, as estimation of poverty and inequality. Econometrica, 71(1), 355–364. they receive more in benefits than they pay in. When the net Ferreira, Francisco H. G. 2012. Distributions in Motion: Economic beneficiaries are concentrated in the lower half of the income Growth, Inequality, and Poverty Dynamics. In The Oxford distribution and the net payers are concentrated in the upper Handbook of the Economics of Poverty, edited by Philip N. half, the fiscal system is progressive and can reduce both poverty Jefferson, 427–62. New York: Oxford University Press. and inequality. At the same time, the revenues generated from Ferreira, Francisco H. G., Julian Messina, Jamele Rigolini, Luis-Felipe taxes fund investments in physical and digital infrastructure as López-Calva, Maria Ana Lugo, and Renos Vakis. 2013. Economic well as human capital accumulation, which can both increase Mobility and the Rise of the Latin American Middle Class. economic mobility for the children of poorer and vulnerable Washington, DC: World Bank. households and drive economic growth. Chapter 7 addresses Foster, James, J. Greer, and E. Thorbecke. 1984. A Class of how well fiscal policy in Vietnam achieves these aims. Decomposable Poverty Measures. Econometrica, 761–765. Gibson, John. 2009. Regional Price Deflators for VHLSS 2008–2010 and GSO Capacity Building for Spatial Cost of Living Index. Mimeo. 1.9. References GSO. 2021. Major Findings. The 01/4/2020 Time-Point Population Change and Family Planning Survey. Alvaredo, Facundo, and Leonardo Gasparini. 2015. Recent Trends in Jolliffe, Dean, and Espen Beer Prydz. “Estimating international poverty Inequality and Poverty in Developing Countries. In Handbook lines from comparable national thresholds.” The Journal of Economic of Income Distribution, vol. 2A, edited by Anthony B. Atkinson Inequality 14, no. 2 (2016): 185-198. and François J. Bourguignon, 697–805. Handbooks in Economics Lakner, Christoph, Daniel Gerszon Mahler, Mario Negre, and Espen Series. Amsterdam: North-Holland. Beer Prydz. 2020. “How Much Does Reducing Inequality Matter Bedi, Tara, Aline Coudouel, and Kenneth Simler. 2007. More than a for Global Poverty?” Global Poverty Monitoring Technical Note 13, Pretty Picture: Using Poverty Maps to Design Better Policies and World Bank, Washington, DC. Interventions. Washington, DC: World Bank. Lakner, Christoph, Mario Negre, and Espen Beer Prydz. 2014. Bergstrom, Katy. 2020. The Role of Inequality for Poverty Reduction. Twinning the Goals: How Can Shared Prosperity Help to Reduce Policy Research Working Paper No. 9409. Washington, DC: Global Poverty? Policy Research Working Paper 7106. Washington, World Bank. DC: World Bank. Bigman, David, and Hippolyte Fofack. 2000. Geographical Targeting Mancini, Giulia, and Giovanni Vecchi. 2020. On the Construction of a for Poverty Alleviation: An Introduction to the Special Issue. World Welfare Indicator for Inequality and Poverty Analysis. WB DRAFT. Bank Economic Review, 14(1), 129–45. Marhuenda, Y., Molina, I., Morales, D., & Rao, J. (2017). Poverty Bourguignon, Alain Jean-Francois. 2003. The Growth Elasticity of mapping in small areas under a twofold nested error regression Poverty Reduction: Explaining Heterogeneity across Countries and model. Journal of the Royal Statistical Society, Series A (Statistics in Time Periods (English). Washington, DC: World Bank. Society), 180 (4), 1111–1136. Bourguignon, François J. 2004. The Poverty-Growth-Inequality Molina, I., & Rao, J. 2010. Small area estimation of poverty indicators. Triangle. Working Paper 125. New Delhi: Indian Council for Canadian Journal of Statistics, 38 (3), 369–385. Research on International Economic Relations. Mookherjee, D. and Shorrocks, A.F. (1982), “A decomposition analysis CECODES, VFF-CRT, RTA & UNDP. 2021. The 2020 Viet Nam of the trend in UK income inequality”, Economic Journal, Vol. 92, Governance and Public Administration Performance Index (PAPI No 328, pp. 886-902. 2020): Measuring Citizens’ Experiences. Nussbaum, Martha. 2011. Creating Capabilities. Cambridge, MA, Corral, P., Molina, I., & Nguyen, M. C. (2020). Pull your Small Area London: Belknap Press of Harvard University Press. Estimates up by the Bootstraps. World Bank Policy Research Nguyen, Minh Cong and Judy Yang. 2022. Technical Background Working Paper No. 9256. Washington, DC: World Bank. Paper on the Measurement of Small Area Estimates of Poverty in Datt, Guarav, and Martin Ravallion. 1992. Growth and Redistribution Vietnam. Washington, DC: World Bank. Components of Changes in Poverty Measures: A Decomposition Pimhidzai, Obert, Chiyu Niu, Diji Chandrasekharan Behr, Thang Cong with Applications to Brazil and India in the 1980s. Journal of Tran, Hardwick Tchale, Hardwick; Peter Kjaer Milne, Salman Zaidi, Development Economics, 38(2): 275–95. and Lan Vu Thi. 2019. Vietnam Poverty and Shared Prosperity Deaton, Angus, and Salman Zaidi.2002. Guidelines for Constructing Update: Better Opportunities for All (English). Washington, DC: Consumption Aggregates for Welfare Analysis. LSMS Working World Bank Group. Paper No. 135. Washington, DC: World Bank. Ravallion, Martin and Guarav Datt. 1992. Growth and Redistribution Demombynes, Gabriel and Linh Hoang Vu. 2015. Demystifying Poverty Components of Changes in Poverty Measures: A Decomposition Measurement in Vietnam. Vietnam Development Economics with Applications to Brazil and India in the 1980s. Journal of Discussion Paper 1. World Bank Group. Development Economics, 38, 275–295. 46 2022 Vietnam Poverty and Equity Assessment Chapter 1. Taking stock of welfare trends over a dynamic decade Ravallion, Martin. 2015. The Economics of Poverty: History, and given any amount of money, households may choose to spend it Measurement and Policy. New York, NY: Oxford University Press. differently. Theoretical underpinnings of consumer choice and welfare Ravallion, Martin, Shaohua Chen and P. Sangraula. 2009. Dollar a Day assume a utility-based approach. For a given level of income, households Revisited. The World Bank Economic Review, 23(2): 163–184. will consume items based on a normative utility function (Sen 1980, Rawls, John. 1967. A Theory of Justice. Cambridge. Nussbaum 2011, Ravallion 2015). 16  Sen, Amartya. 1980. Equality of What? Tanner Lectures on Human Poverty rates before 2010 are based on a different methodology. The Values, CUP. construction of the welfare aggregate was significantly updated in 2010. Thuong, Nguyen Thi Thu, Yme Van Den Berg, Tran Quang Huy, Do Trends before and after 2010 are different and not comparable. Anh Tai, Bui Nu Hoang Anh. Determinants of catastrophic health 17  MOLISA has forecasted that by January 2021, the country will expenditure in Vietnam. International Journal of Health Planning have about 16.6 percent of the total households, or about 4.473 and Management, 36(2), 316–333. DOI: 10.1002/hpm.3076 million households, with about 17.447 million people, having incomes UNDP (United Nations Development Programme). 2018. under those poverty levels (2.32 million household more than the Multidimensional Poverty in Vietnam. Reducing Poverty in figure of 2020). For more information, see: https://en.vietnamplus. all its Dimensions to Ensure a Good Quality Life for All. New vn/20212025-poverty-standards-to-help-better-identify-poor- York, NY: UNDP. households/189264.vnp World Bank. 1999. Vietnam Development Report 2000: Attacking 18  Regional poverty trends described in this section are part of a high- Poverty. Washington DC: World Bank. level discussion on poverty dynamics over the last decade. Poverty ---------------. 2012. Well Begun, Not Yet Done: Vietnam’s Remarkable profiling across other household characteristics, in particular dimensions Progress on Poverty Reduction and the Emerging Challenges, 2012 that reveal large variations in poverty rates, are further discussed in Vietnam Poverty Assessment. Washington, DC: World Bank. Chapter 3. ---------------. 2016. Vietnam Systematic Country Diagnostic. Sustaining 19  Success – Priorities for Inclusive and Sustainable Growth. The cost of housing is estimated for most households. Most households do not pay monthly rent, so housing value is imputed using ---------------. 2018a. Poverty and Shared Prosperity 2018: Piecing Together the Poverty Puzzle. Washington, DC: World Bank. hedonic models based on size, location, and housing quality. 20  ---------------. 2018b. Riding the Wave: An East Asian Miracle for the See Gibson (2009) for information on the Spatial Cost of Living 21st Century. World Bank East Asia and Pacific Regional Report;. Index (SCOLI). Washington, DC: World Bank. © World Bank. 21  This is often chosen because for many poor households informal ---------------. 2020. Poverty and Shared Prosperity Report: Reversals of incomes can be lumpy and look quite different when measured in one Fortune. Washington, DC: World Bank. week compared to another, varying with the agricultural season, the ---------------. 2021a. East Asia And Pacific Economic Update October amount of business a home enterprise does, or the number of hours 2021: Long Covid. Washington, DC: World Bank. of casual labor they can sell. Consumption is often more regular, as ---------------. 2021b. A Year Deferred – Early Lessons and Experiences savings or borrowing is used to smooth out lumpy incomes. Thus, from COVID-19 in Vietnam. Washington, DC: World Bank. consumption is a close reflection of their average welfare or living WHO. 2021. Global Health Expenditure Dataset. Accessed November 7, standards. In higher-income countries, national aggregates are often 2021. https://apps.who.int/nha/database/ViewData/Indicators/en based on household current income. As populations live farther from the concept of minimum needs, households are more likely to have stable formal incomes, and thresholds reflect what is regarded to be 1.10. Notes needed to maintain middle-class living standards. 22  For the 2006–2010 and 2011–2015 phases, the monetary poverty 13  Minimum monthly wages are about 4 million VND per worker. threshold for monitoring was fixed at the beginning of each five-year Average wages in 2020 were about 6 million VND per month. phase, which resulted in welfare trends not being comparable over Households usually have multiple sources of income and working adults. longer time horizons (Demombynes and Vu, 2015). 14  First principles and guidelines on the construction of the welfare 23  The ten non-monetary indicators include: adult education, child aggregate are illustrated using data from several countries including education, access to health care services, health insurance, housing quality, Vietnam in Deaton and Zaidi (2002). Almost two decades later, Mancini per capita housing area, drinking water supply, type of toilet/latrine, use and Vecchi (2020) conducted an updated review using a wider set of of telecommunication services, and assets for information access. countries and methodologies. 15  The consumption of commodities is constrained by a household’s budget and income. Of course, household preferences are different, From the Last Mile to the Next Mile 47 PART 1. A DECADE OF SIGNIFICANT PROGRESS, BUT LAST-MILE CHALLENGES REMAIN ANNEX 1.1. Expenditure in the Vietnam Household Living Standards Survey The first consumption module was collected in the 1993 The measurement of an expenditure-based welfare Vietnam Household Living Standards Survey (VHLSS). concept for Vietnam was updated in 2012 using the The VHLSS collects household expenditures on food, non- 2010 VHLSS and is described in detail in the World Bank food, housing, durables, education, and other categories. Food 2012 Poverty Assessment Report (World Bank, 2012). captures essential calories and nutrients. Non-food items Important updates included the measurement of imputed encompass everyday essential goods, services such as health rent and durables, developing the spatial and temporal price and education, utilities, water, and value from using housing deflator index,20 and updating the reference basket. These or other durable items for extended periods. Consumption changes and improvements were quite substantial, and welfare categories can be either own-produced, gifted, or purchased, measurements before and after 2010 are not comparable. The an important distinction that separates the consumption large sample of the VHLSS includes approximately 45,000 aggregate from current income concepts. Values from households, and since 2015, it has been conducted every year. housing are estimated since many Vietnamese do not pay The collection of household expenditures used for the World rent or mortgage, yet there is still a welfare utility present that Bank’s consumption-based poverty measurement is conducted needs to be measured. The time reference of consumption biannually and on a smaller sample of approximately 9,400 varies by item. Expenditures on regular items such as food are households that is representative at the regional level. The asked in relation to the last 30 days, but other items are asked 2010–2018 VHLSS were sampled from the 2009 Census. The in reference to the last six months or year. 2020 VHLSS is sampled from an updated sampling frame based on the 2019 Census. The welfare aggregate is a monetized value of a bundle of goods and services based on household consumption and expenditures. For the purposes of measuring welfare, the welfare aggregate based on household consumption includes adjustments, imputations, and valuations. First, household consumption is measured in per capita terms to account for differences in consumption due to household size. Second, household consumption is spatially adjusted following a geographic price deflator, the Spatial Cost of Living Index, or SCOLI (World Bank, 2012). 48 2022 Vietnam Poverty and Equity Assessment Chapter 1. Taking stock of welfare trends over a dynamic decade ANNEX 1.2. Global poverty measurement and Vietnam To measure poverty using the World Bank’s global poverty Table A.1.2.1. Conversion rates thresholds, the monetary value of the consumption VIETNAMESE aggregate collected from the VHLSS must be converted DONG (TO US$1) from Vietnamese dong to international dollars. Specifically, Purchasing power parity (2011, revised) 7,528.385 the units of measurement for global poverty are 2011PPP Households and NPISHS final consumption international dollars and in per capita per day terms. expenditures Exchange rate (2011, period average) 20,509.75 An important distinction is that conversion to Note: The total value of actual and imputed final consumption expenditures international dollars utilizes purchasing power parity incurred by households and NPISHs on individual goods and services. It also (PPP) conversions rather than exchange rates. PPP allows includes expenditures on individual goods and services sold at prices that are not economically significant. a country’s income and consumption data to be converted Source: World Bank International Comparison Program, World Development into globally comparable terms taking into account cost-of- Indicators. living comparability across countries. The PPP of currency of an economy is defined as the number of currency units required to purchase a basket of goods and services with one unit of the currency of a reference or base country. The PPP is computed based on price data collected by countries across the world for a common basket of goods and services. For a fixed basket of international goods valued at US$1, those items can be purchased in Vietnam for 7,528.385 Vietnamese dong. The observation that the PPP conversion factor is much lower than the exchange rate means that cost of living is low in Vietnam. From the Last Mile to the Next Mile 49 PART 1. A DECADE OF SIGNIFICANT PROGRESS, BUT LAST-MILE CHALLENGES REMAIN ANNEX 1.3. identifies the poor and near-poor based on a combination of non-monetary and monetary principles. In 2015, the government approved a plan to update the country’s poverty monitoring basis to include a multidimensional measure. Review of poverty lines The new multidimensional index includes ten non-monetary indicators and is reported annually by the GSO at the national There are many concepts of poverty, each with its own and provincial levels (UNDP, 2018).23 From 2016 to 2020, merits. The World Bank’s bi-annual Poverty and Shared those who subsist below a low monetary threshold based Prosperity Report, Piecing Together the Poverty Puzzle, on household income are considered poor, while the near- expanded the menu of indicators that it uses to measure poor are those living within a higher monetary range and globally comparable poverty rates, including higher absolute were also deprived in 3 out of 10 non-monetary indicators. poverty lines, relative poverty lines, and a multidimensional These definitions changed again for the 2012–2025 period poverty measure (World Bank, 2018a). Countries also set where the poor and near-poor are now defined by the same their own national poverty lines, determined by national monetary threshold but different non-monetary deprivations. governments that are best suited to a country’s specific Comparing the 2006–2010 and 2021–2025 planning periods, conditions. Some countries explicitly measure happiness, the urban monetary threshold for the poor has increased 5-fold, such as in Bhutan, or poverty based on a consensus measure, and the rural monetary threshold has increased about 5.8-fold. as in Tonga. In many developing countries, monetary poverty is measured based on household consumption rather than GSO-WB Poverty Line household income.21 The World Bank and the General Statistics Office (GSO) Vietnam’s National Poverty Measurement jointly developed a monetary consumption-based poverty line following the Cost of Basic Needs methodology. The National poverty reduction budgeting and goal setting first monetary poverty line was developed using the 1993 is based on standards set by the Ministry of Labour, VHLSS (World Bank, 1999). The original GSO-WB poverty Invalids and Social Affairs (MOLISA). MOLISA’s line was thus based on economic conditions of households just poverty lines are also referred to as budgeting lines since shortly after the initiation of Doi Moi reforms. Due to rapid national estimates of poor people influences the poverty economic progress and changes in household consumption reduction budget. Standards and targets are set for a five- patterns, the GSO-WB poverty line was substantially revised year Socioeconomic Development Plan (SEDP) cycle. For using newer consumption patterns and data from the 2010 example, different standards were set for the 2006–2010, VHLSS. Cost of Basic Needs poverty lines are constructed by 2011–2015, 2016–2020, and 2021–2025 periods (Table first defining a food poverty line based on minimum calorie A.1.3.3.). However, the identification of poor and near-poor requirements, and then a non-food poverty line is calculated to inform the implementation of poverty reduction programs based on average non-food expenditures of a baseline group. at local levels may include further conditions defined by This update is explained in detail in the World Bank’s local governments. The government in Vietnam is highly 2012 Poverty Assessment report (World Bank, 2012). The decentralized, and while national standards are developed for purpose of the GSO-WB poverty lines is primarily to ensure monitoring targets, household targeting may rely on different consistency in poverty monitoring over time and across space. criteria determined at local levels. The values of the GSO-WB and World Bank lower middle- income country (LMIC) poverty lines are very similar. The definitions of poor and near-poor have changed across SEDPs. For the 2006–2010 and 2011–2015 phases, World Bank Global Poverty Lines the poor and near-poor were determined by only household income.22 The general poverty rate reflected the share of The construction of the World Bank global poverty households earning income less than the income threshold for lines was informed by national poverty lines across the the poor, and was last published in 2016. Today, for policy, developing world. Extreme poverty informing SDG #1 planning, and budgeting at the national level, MOLISA is defined by the World Bank’s International Poverty Line 50 2022 Vietnam Poverty and Equity Assessment Chapter 1. Taking stock of welfare trends over a dynamic decade ($1.90/day 2011PPP) that represents the average of national • Finally, in the upcoming 2021–2025 monitoring period, poverty lines from the poorest countries (Ravallion, Chen, income thresholds to identify the poor and near-poor are and Sangraula, 2009). The global poverty line most suitable the same, and near equivalent value to UMIC PL ($5.50/ for the Vietnam context is the LMIC poverty line, derived day 2011PPP). The UMIC PL converts to about 1.8 million as the median of a set of national poverty lines from LMIC VND per capita per month, while the 2021–2025 thresholds countries and valued at $3.20/day in 2011 PPP US dollars are 1.5 and 2.0 million VND for rural and urban areas (Joliffe and Prydz, 2016). At this value, the LMIC poverty line respectively (Table A.1.3.2). is representative of local minimum basic needs that LMIC countries themselves define as necessary for day-to-day living Table A.1.3.2. Conversions of the global (Table A.1.3.1.). poverty lines to Vietnamese dong USD VND 2020 (JAN) Table A.1.3.1. Median value of national poverty 2011PPP MONTHLY PER lines, by economy income classification PER DAY CAPITA (circa 2011 in PPP terms) International Poverty Line 1.90 626,512 ECONOMY, INCOME CLASSIFICATION MEDIAN Lower-middle-income 3.20 1,055,178 country poverty line Low-income economy 1.90 Upper-middle-income 5.50 1,813,588 Lower-middle-income economy 3.20 country poverty line Upper-middle-income economy 5.50 High income 21.70 7,155,429 High-income economy 21.70 Note: Using PPP conversion factors to LCU. Note: Values are rounded to the nearest 0.10. Economies are classified on the basis of official World Bank income classifications, which rely on measures of Poverty rates across different thresholds illustrate similar per capita gross national income. Estimates are based on national poverty lines trends in impressive poverty reduction over the last in 126 economies. The selected poverty line for each economy is the one that is closest in time to 2011. decade (Figure A.1.3.1). From 2010 to 2016, the LMIC Source: Jolliffe and Prydz (2016). poverty rate trended between the GSO-WB poverty rate and MOLISA’s general poverty rates. Since 2016, poverty rates across different metrics have converged more closely and This report primarily uses the World Bank global describe similar levels of poverty for the remainder of the decade. poverty lines Figure A.1.3.1. Comparison of poverty It is instructive to understand the value of the World rates, by different lines Bank global poverty lines in Vietnamese dong. The value of Vietnam’s monetary poverty thresholds has slowly 25 increased, reflecting increasing average standards, aspirations, 20 and costs of living. Poverty rate (%) 15 • The World Bank’s International Poverty Line (IPL) of $1.90/ 10 day 2011PPP is equivalent to about 626,512 VND per person per month. At the beginning of the decade during the 5 2011–2015 SEDP phase, MOLISA’s poverty threshold was 0 below the equivalent value of the IPL. 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 • The upper threshold for the identification of the near poor General poverty rate Multidimensional poverty rate during 2016–2020 SEDP phase used thresholds in similar GSO-WB poverty rate World Bank LMIC poverty rate equivalent values to the LMIC poverty line of $3.20/day Note: Multidimensional poverty rate is preliminary in 2020. 2011PPP. Source: GSO and World Bank. From the Last Mile to the Next Mile 51 Table A.1.3.3. A comparison of poverty lines and monetary thresholds (VND per month per capita, ‘000) PART 1. MOLISA GSO WORLD BANK WB-GSO   MULTIDIMENSIONAL POVERTY MULTIDIMENSIONAL POVERTY GLOBAL POVERTY LINES POVERTY LINE Welfare concept An index based on income per capita and An index based on income per capita and Consumption per capita Consumption per non-monetary dimensions non-monetary dimensions capita   Rural Urban Rural Urban National National Rural monetary Urban monetary Rural monetary Urban monetary International Lower-middle Upper-middle   thresholds (poor thresholds (poor thresholds (poor, thresholds (poor Poverty Line income country income country and near-poor) and near-poor) and near-poor), and near-poor) ($1.90/day poverty line ($3.20/ poverty line ($5.50/ 2011PPP) day 2011PPP) day 2011PPP) 2010 2006–10 SEDP 2006–10 SEDP     352 592 1,018 653 52 2022 Vietnam Poverty and Equity Assessment Poor < 200 Poor < 260 2011             2012 2011–15 SEDP 2011–15 SEDP     463 779 1,339 873 Poor: 400 Poor: 500 2013             Near-poor: Near poor: 2014 400–520 500–650     522 880 1,512 964 2015             2016 700 900 531 895 1,539 969 or 700-1,000 or 900-1,300 A DECADE OF SIGNIFICANT PROGRESS, BUT LAST-MILE CHALLENGES REMAIN 2017 2016–20 SEDP 725 or 725-1,035 935         2016–20 SEDP or 935-1,350 Poor < 700 Poor < 900 2018 Near-poor: 700- 755 or 755-1,080 975 574 967 1,662 974 Near-poor: 900 - 1,000 or 975-1,400 1,300 2019             2020     628 1,058 1,819 1,033 2021 2021–25 SEDP 2021–25 SEDP             Poor < 1,500 Poor < 2,000 Near-poor < 1,500 Near-poor < 2,000 Note: This table only documents the monetary thresholds under various poverty lines and indices. MOLISA and GSO concepts also include non-monetary components that are not listed in this table. Values rounded to nearest thousand. MOLISA standards are only monetary in the 2006–2010 and 2011–2015 periods. Non-monetary dimensions under the MOLISA concepts include basic social services in healthcare, education, housing, clean water and hygiene, and information. The non-monetary components under the GSO Multidimensional Poverty Index includes adult education, child education, access to health care services, health insurance, housing quality, per capita housing area, drinking water supply, type of toilet/latrine, use of telecommunication services, and assets for information access. Chapter 1. taking stoCk of welfare trends over a dynamiC deCade ANNEX 1.4. Distribution-sensitive poverty projections technical information At a macro level, changes in poverty can be decomposed Assumptions on the shape of the growth incidence into a combination of growth and redistribution effects curve and changes in inequality affect the degree of how (Bourguignon 2003, 2004; Datt and Ravallion 1992; Ferreira growth is distributed across households. Inequality is 2012). Poverty projections can be made taking into account modeled through parameterized assumptions regarding how these two channels. In a simplified scenario, poverty can welfare grows at different rates along the welfare distribution. be projected on the basis of growth alone, when growth is The shape of this growth curve along the welfare distribution assumed to be even across the entire distribution under a is referred to as the growth incidence curve. On the basis neutral-distribution assumption. This assumption allows for of empirically observed growth incidence curves using the simplified calculations but may not be accurate in all cases. Vietnam Household Living Standards Survey, a linear growth A distribution-sensitive poverty projection assumes that incidence curve is assumed. A linear assumption also yields growth is uneven across the welfare distribution. Calculations more conservative projections of poverty rates: of distributional sensitive poverty projections follow the methods described in Lakner et al. (2020), and are calculated using the povsim STATA command documented in Lakner, where Negre, and Prydz (2014). = the growth rate for percentile group i. = a percentile group . The poorest households are The initial poverty rate is the 2018 $3.20/day 2011PPP percentile group 1, and richest are percentile group 100. poverty rate in Vietnam, based on the most recently available and = growth parameters that can be considered to reflect survey data. Poverty projections begin in 2019. a transfer and tax. Growth of household consumption is assumed to follow The final Gini obtained will depend on the values of and similar patterns of gross domestic product (GDP) per capita . The parameters and are values that solve the following growth from national accounts. equation that yields the desired change in inequality. ‒ Actual GDP growth rates are used for poverty projections in 2019 and 2020. For 2021–2023, growth projections are used for poverty projection calculations. ‒ A pass-through rate is fixed at 1. This assumes that Poverty in 2018 and 2020 is based on the initial welfare values the welfare aggregate grows at the same rate as GDP . The final welfare ( ) is determined by the formula or private consumption per capita. This is the most below, assuming everyone is taxed in proportion to their optimistic scenario. initial welfare ( ) and their rank ( ): where = the initial value of welfare for a percentile group . From the Last Mile to the Next Mile 53 PART 1. Chapter 2. A review of the drivers of poverty reduction Chapter 2. A review of the drivers of poverty reduction Key Messages • Livelihoods continued to improve over the last decade (2010–2020), as a large expansion of the economy in off-farm sectors created a net 10.6 million new jobs in the manufacturing and services sectors. Importantly, these jobs became available as a youth bulge in the population came of age, and these jobs were better than for the generation before. • Wage income was the most dynamic source of household income, but some other sources, in particular household farming income, remained relatively stagnant over the decade. • Household profiles are also shifting, exhibiting higher education levels and also becoming older. • Widening inequality is driven by increasing inequality between urban and rural areas. 54 2022 Vietnam Poverty and Equity Assessment Chapter 2. A review of the drivers of poverty reduction 2.1. Changing household have at least one child. Single-adult female-headed households profiles over the last with a child are much more common than single-adult male- headed households with a child. Skip-generation households, decade (2010–202024) those with only seniors and children, are less common. About 10 percent of households are multi-generational with at least The decade was marked by broad changes not just in the one child, working age adult, and senior citizen. Yet, as the economy but also in the profile and characteristics of population is aging, the number of households with older households. This section compares the profile of households members is also increasing (Figure 2.1). at the beginning and end of the decade. Profiles and indicators are presented at the household level rather than individual Figure 2.1. Ageing Vietnam because poverty is a household-level indicator.25 8 The population grew by 11 million people from 7 6.7 6.4 2009 to 2019 6.2 # of households (in millions) 6 5.1 5.1 The Vietnamese population grew by 11 million in the 5 4.7 4.6 decade between 2009 and 2019 according to Census data, 4 with wealthier regions experiencing higher population 3 2.7 2.5 growth (Table 2.1). Vibrant economic centers gained the 2.2 2 most in population; the Southeast region grew in both 1.5 1.2 absolute and percentage terms, followed by the Red River 1 0.1 0.0 Delta. The population in the Southeast region increased by 3.8 0 million people. The Red River Delta region also added about 10–19 20–29 30–39 40–49 50–59 60–69 70+ 3 million people. The population grew the least in absolute 2010 2020 numbers in the Mekong Delta, which is consistent with recent outmigration trends related to worsening climate conditions. Note: Categories are by the oldest member of the household. Source: World Bank staff calculations using the VHLSS, 2010-2020. Table 2.1. Population, 2009–2019 2009 2019 Most ethnic minorities still live in the Midlands Central Highlands 5,115,135 5,842,605 and Northern Mountain region Mekong Delta 17,191,470 17,273,621 Households with an ethnic minority head represent less Midlands and Northern 11,013,590 12,532,801 Mountains than 13 percent of the population, or about 3.4 million households. The regional distribution of ethnic minorities Northern and Coastal 18,835,154 20,187,228 did not change considerably over the decade. Across regions, Central they are still most heavily concentrated in the Midlands and Red River Delta 19,584,287 22,543,562 Northern Mountain region, where they even outnumber the Southeast 14,067,361 17,828,818 Kinh majority (Figure 2.2). They are next most concentrated National 85,801,996 96,208,635 in the Central Highlands region, where ethnic minorities Source: GSO, Census yearbooks. comprise 30 percent of the regional population and the population of Kinh is the smallest. These two regions also In 2020, there were about 26.7 million households in have the highest poverty rates. The linkages between ethnic Vietnam, with an average of 3.7 members and 2.2 adults minorities and their locations are discussed in Chapter 3. Ethnic per household. The most common household composition, minorities are less represented particularly in regions with large representing over half of all households, are multi-adult cities and vibrant economies. These households are the fewest households with at least one child. Sixty percent of households in number in the Red River Delta (about 120,000 households). From the Last Mile to the Next Mile 55 PART 1. A DECADE OF SIGNIFICANT PROGRESS, BUT LAST-MILE CHALLENGES REMAIN Figure 2.2. Distribution of households Household profiles of education and sector of by ethnicity, 2020 economic activity have changed the most 8 7 Education completion has increased considerably. The # of households (in millions) 6 maximum level of education in a household has improved 5 over the past decade. In 2010, in half of all households, the maximum level of education of any member was lower- 4 secondary level or less (Figure 2.3). By 2020, this share fell 3 to about 44 percent of all households. There are now an 2 additional 1.7 million households with a member who 1 has completed tertiary education. However, tertiary-level 0 educated households are still the minority. Chapters 4 and 5 examine education completion in more detail, as well as Central Highlnads Midlands & Northern Mountains Northern & Coastal Central Southeast Red River Delta Mekong Delta differences in completion across certain groups, which will affect future labor market profiles. Households are shifting out of agriculture as a main Kinh Ethnic minority activity.26 Over the last decade, the number of households Source: World Bank staff calculations using the VHLSS, 2020 with someone engaging in the manufacturing or services sector increased by nearly 4 million each. In 2010, agriculture was the most common sector in which at least one person in a household was employed as their main job. In 2020, households with primary engagement in agriculture became the least common, a reversal over the decade. In 2020, about 9.7 million households, or 35.6 percent of all households, Figure 2.3. Education level in Vietnamese Figure 2.4. Shifts in sector of main employment of households, 2010–2020 Vietnamese households, 2010–2020 8 15 6.9 13.1 7 12.2 11.6 12 # of households (in millions) # of households (in millions) 6 5.7 5.5 9.5 9.6 5 4.7 4.7 8.3 9 4.0 4.1 4 3.7 3.5 3.0 3 6 2 1.5 1.4 3 1 0 0 No Primary Lower Upper Vocational Tertiary Agriculture Manufacturing Services education Secondary Secondary 2010 2020 2010 2020 Note: Maximum level of completed education of any household member. Note: Not exclusive, households with multiple working adults can be engaged in Source: World Bank staff calculations using the VHLSS, 2010 and 2020. multiple sectors as main jobs. Source: World Bank staff calculations using the VHLSS, 2010 and 2020. 56 2022 Vietnam Poverty and Equity Assessment Chapter 2. A review of the drivers of poverty reduction have some engagement in the agricultural sector (Figure 2.4). of unskilled occupations declined dramatically, while the This does not necessarily mean these households are reliant on number of medium-skilled jobs increased in net by almost agricultural income (agriculture is one of the lowest-paying 3 million (Figure 2.7). These new medium-skilled jobs were income streams); it is more likely that agriculture is an available primarily plant and machine operators and assemblers. While activity for diversification and income supplementation. the number of medium-skilled occupations grew, growth in higher-skilled occupations was slower. The size of the high- Due to a demographic dividend, the labor force skilled workforce remained steady at about 10 percent of the expanded by a net 4.5 million workers from 2010 to total labor force from 2010 to 2020. In absolute numbers, 2020. Over the entire decade, the manufacturing and services the high-skilled workforce grew by about 1.8 million workers, sector increased in net by about 5.8 and 4.8 million workers, while the medium-skilled workforce grew by almost 4 million. respectively. Larger changes were more pronounced in the Despite a more educated youth population, younger workers latter half of the decade, when growth in manufacturing was are moving into manufacturing and manual labor. See Chapter more pronounced and agriculture jobs declined dramatically 5 for a more in-depth discussion on skills and education. (Figure 2.5). Agricultural employment began to decline significantly between 2015 to 2020, from 24.5 million to Informality is declining, but most jobs are still without 17.7 million workers at about an equal pace annually. The contracts (Figure 2.8). The labor market in Vietnam is services sector transformed in the latter half of the decade still highly informal, with only 27 percent of wage workers through tourism. Between 2010 and 2015, international having a contract through which employers contribute to visitors increased by 3 million annually (from 5 to 8 million). social insurance and health care. At the household level, over By 2019, the number of international visitors had reached 18 40 percent of households in 2020 now have a member whose million, an increase of nearly 10 million. primary wage job has a contract, compared to 31.4 percent in 2010. Having a job contract aligns with being entitled to social Consistent with the structural transformation out insurance, as well as having unemployment insurance. But 21 of agriculture, jobs shifted from low to medium million households, or 81 percent of all households, still have skilled. In the first half of the decade, new jobs were just as at least one household member who either has a wage job likely to be in wage employment as in unpaid family work without a contract, is engaged in self-employed agriculture, or (Figure 2.6). In the second half of the decade, the number is engaged in a self-employed business. Figure 2.5. Employment levels, by sector (15+) 30 Increasing services Increasing manufacturing and declining agriculture 25 20 Workers (millions) 15 10 5 0 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 Prel. 2020 Agriculture Manufacturing Services Source: GSO. From the Last Mile to the Next Mile 57 PART 1. A DECADE OF SIGNIFICANT PROGRESS, BUT LAST-MILE CHALLENGES REMAIN Figure 2.6. A substantial increase in wage jobs Figure 2.7. More jobs and a shift from low- to medium-skilled jobs 6,000 4,000 Net change in employment (thousands) 5,000 3,000 Net change in employment (thousands) 4,000 2,000 3,000 2,000 1,00 1,000 0 0 −1,000 −1,000 −2,000 −2,000 −3,000 −3,000 2010–2014 2015–2020 −4,000 2010–2014 2015–2020 Wage worker Employer High-skill Medium-skill Low-skill Own account worker Unpaid family worker Source: GSO, LFS. Source: GSO, LFS. Foreign direct investment (FDI) was important in a strong correlation between injections of FDI, the increasing transforming jobs, since most jobs in foreign-owned number of manufacturing jobs, and higher wages. However, enterprises are formal with higher wages compared to there are also potential challenges to manufacturing jobs, domestic firms. At the beginning of the decade, Vietnam since manual labor jobs prefer younger workers, and these are was still a new entrant to the World Trade Organization not necessarily jobs that one works into old age. (WTO), having entered in 2007, but timing also coincided with lingering effects from the 2008–2009 global financial A challenging ending to the decade in 2020 with crisis. Immediately after entry into the WTO, growth in the emergence of COVID-19 the share of labor in manufacturing was strong from 2007 to 2010 but stagnated at the beginning of the 2010s. The COVID-19 halted a period of rapid income and wage absolute number of manufacturing workers remained growth for workers and households in Vietnam. Real relatively constant, coinciding with a period of stagnant FDI household income per capita measured using the VHLSS growth. As the foreign ownership cap was raised in 2015, FDI in 2020 declined by 5 percent compared to 2019.27 In also increased, with the number of net new manufacturing comparison, real median household income grew at an jobs quickly following. average of 7.2 percent per year from 2010 to 2018.28 Before the full onset of COVID-19, wages were 9 percent higher in Across different categories of wage employment, the first quarter of 2020 than in the same quarter the year there are linkages between the manufacturing sector, before. However, for the remainder of 2020, wages were formality, and foreign-owned firms. Virtually all jobs lower than in corresponding quarters in the previous year. in foreign-owned firms are formal in nature, or have a work Unemployment and underemployment rates also ticked up contract and offer social insurance. In 2020, about half of in 2020 (Figure 2.9). all jobs with contracts were in the manufacturing sector. The public sector is the only other sector also that has high Over a year after the onset of COVID-19, households rates of formal jobs. In net, there were over 2 million more and firms are still on the path to recovery. The incidence manufacturing jobs in 2020 than 2018, and 1.9 million of of income loss lightened considerably over time, though it them were formal. About half of all domestic private sector did not disappear completely. According to responses from firms offer formal work contracts. Foreign companies also the World Bank monitoring surveys, about 30 percent of pay double the wages that domestic firms do. Thus, there is households self-reported still having lower incomes in March 58 2022 Vietnam Poverty and Equity Assessment Chapter 2. A review of the drivers of poverty reduction 2021 than the year before. Official statistics reported that 9.1 recover, and almost all businesses were open by January 2021. million workers (12.8 percent of all workers) had either lost Although sales remained lower than in previous years, the their jobs or had reduced wages in the first quarter of 2021, reductions in employment were smaller than at the start of and average labor incomes were 2.3 percent lower compared the pandemic. However, a full recovery will be delayed due to to the previous year (Ha and Minh, 2021). Firms started to more severe outbreaks that emerged in 2021. Figure 2.8. Share of jobs by formality 100 14.8 9.3 17.0 80 23.1 Share of workers (%) 14.9 19.1 12.8 60 15.9 40 46.6 56.2 20 32.8 40.4 0 2010 2020 2010 2020 Has contract No contract Agriculture and fishing Mining and quarrying Manufacturing Electricity, gas and water Construction Commerce Education, health and social work Transport, storage, and comms Public admin Financial, insurance and real estate Other Source: World Bank staff calculations using VHLSS 2010–2020. Figure 2.9. Underemployment and unemployment rates rose in 2020 A. Underemployment rate B. Unemployment rate 6 8 7 5 6 Underemployment (%) Underemployment (%) 4 5 3 4 3 2 2 1 1 0 0 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 General 15–24 25–49 General 15–24 25–49 50+ Male Female 50+ Male Female Source: GSO. From the Last Mile to the Next Mile 59 PART 1. A DECADE OF SIGNIFICANT PROGRESS, BUT LAST-MILE CHALLENGES REMAIN BOX 2.1. Setting the growth foundation in earlier decades This box reviews the broad policy decisions and events that UNESCO held in 1991 that provided a policy framework and have led to lasting changes and created the foundations for strategic direction for improvement in the national education successful poverty reduction during the most recent decade. system. The gradual actions included universalization of primary education by 2000, progress toward universal lower In the immediate post-war period, from 1975 to 1985, secondary education by 2005 with nationwide coverage by Vietnam experienced a sharp economic deterioration. 2010, and improved linkages between secondary, technical, During the reconstruction period, Vietnam continued and vocational education. Consequently, the government to adopt a centrally planned economy by expropriating increased its budget dedicated to education and with that, individual properties and nationalizing assets such as land and made gradual improvements in the provision of education. capital to form cooperatives. However, these initial policies Net enrollment rates have increased for all levels. were not successful, and Vietnam plunged into an economic crisis at the end of 1970s and the beginning of 1980s as the The country’s intensified efforts to integrate deeper country endured soaring inflation, food shortages, damaged into the global economy contributed to further infrastructure, and worsening living standards. During this poverty reduction. Beginning in the late 1980s, tariffs were period, about 70 percent of the population was living below unilaterally reduced and numerous quantitative restrictions the poverty line. on trade abolished. Adoption of the Foreign Investment Law in 1987 opened the country to FDI inflows. Membership To end a decade of stagnant growth and high levels of in the Association of Southeast Asian Nations (ASEAN) extreme poverty, the government implemented policy in 1995 and its associated Asian Free Trade Area, and the reforms that greatly improved living standards. A series U.S.-Vietnam Bilateral Trade Agreement in 2001 were of market-oriented reforms was launched in the late 1980s and important steps in the integration process. Following trade early 1990s under the Doi Moi, which ignited a remarkable liberalization, the government prioritized exports of labor- period of economic growth and poverty reduction that still intensive industries such as garment, textile, footwear, and continues to this day. The 1993 Land Law was among the most sea product industries, creating a favorable environment for important milestones in terms of rural reform and is credited off-farm job creation and thereby poverty reduction. Thanks with successful poverty reduction among farmers during this to the government’s adjustments to the industrial policy and period. The law granted households five basic rights, including trade liberalization, structural changes in the labor market to transfer, exchange, inherit, rent, and mortgage their land, became the key pillar to poverty alleviation between 1993 and and extended the lease term to 20 years for annual cropland 2002, with increasing shifts of workers out of the agricultural and 50 years for perennial cropland. The change in land sector to the industry and construction sectors. ownership created strong incentives for farming households to make long-term investments and expand agricultural With the introduction of the new Enterprise Law, production. Farmers’ incomes grew as their crops could now citizens could establish and operate private businesses be sold at market rather than be collected by the state. Farm with limited intervention from government officials. income increased by 60 percent between 1993 and 1998. The most important innovation introduced by the Enterprise Law was the simplification of registration procedures and Since 1991, education has been considered a national the associated elimination of many business licenses, which priority for both social and economic development and sharply reduced transaction costs for businesses and helped as a vehicle for supporting Doi Moi. The education reform install greater business confidence. The promulgation of the process reached practical ground when a joint study was new legislation was considered one of the most important generated by the Ministry of Education of Training (MOET), factors facilitating the establishment of new business and United Nations Development Program (UNDP), and generating many jobs in small and medium private firms. Continued > 60 2022 Vietnam Poverty and Equity Assessment Chapter 2. A review of the drivers of poverty reduction BOX 2.1. Setting the growth foundation in earlier decades (Continued) Vietnam accelerated its negotiations for WTO somewhat with the lifting of the foreign ownership cap, FDI membership and officially ascended to the WTO resurged, corresponding to a further increase in the share in January 2007. Becoming a WTO member has had of employment in manufacturing. Entry into the global important implications for Vietnam’s development, economy coincided with a growing demographic dividend. catalyzing the creation of off-farm services and export- This synchronization helped to create jobs for millions of oriented manufacturing jobs. The export value of goods young people entering the labor force and drove a strong and services as a percentage of GDP rose from less than 10 phase of upward economic mobility. percent in 1986 to 106 percent in 2019. In 2015, aligning 2.2. Understanding Households rely on different types of income household income sources depending on their background dynamics: the key to The composition of household income varies across the poverty reduction welfare distribution, and in expected ways. Across the distribution, the average contribution from different income The East Asia region and Vietnam stand out for achieving sources is illustrated as a share of total household income an extremely successful growth-driven poverty reduction in Figure 2.10. The role and importance of labor market process, where high levels of economic growth provided a income is clear. For the majority of households, labor market decade of prosperity through job creation and rising wages, income from family farms, non-farm family businesses, or thus lifting households out of poverty. Rising wages lifted wage employment averages to 90 percent of total household all boats and created better jobs. This section provides more income. At most, for households in the top decile, non-labor detail to enable a deeper understanding of the dynamics of market income comprises over 15 percent of average total specific income sources and how income patterns look across household income. By labor income sources, the shares evolve the entire household distribution, not just among the poor. across the distribution in expected ways: larger sources of See Box 2.2 for a description of household incomes. total income from richer households are from family business Figure 2.10. Average share of household income by source, 2020 100 Share of total household 80 60 income (%) 40 20 0 1 2 3 4 5 6 7 8 9 10 (=poorest decile) (= richest decile) Family Farm Family Business Wage SA Remittances Pension Other Note: Includes zero incomes for households who do not receive a particular income source. Source: World Bank calculations using VHLSS 2020. From the Last Mile to the Next Mile 61 PART 1. A DECADE OF SIGNIFICANT PROGRESS, BUT LAST-MILE CHALLENGES REMAIN income, whereas family farm income contributes more to Broad income trends over the last decade reveal poor households. The role of cash from social assistance the importance of wages programs is small as a share of total household income. Wage income is the most important source of household Poorer households are more likely to be income because it provides a high amount of income diversified across broad income sources. for a majority of households. Total household wage income includes earnings on the first, second, and third jobs A high share of Vietnamese households receives income that individuals were engaged in over the last 12 months. from multiple sources (Figure 2.11). Since income sources At the household level, virtually all households have at least are recorded over the last 12 months, diversification29 can be one individual that is employed; only about 5 percent of due to either simultaneous or seasonal income streams from households in Vietnam do not have any employed adult. multiple employed adults. About 30 percent of households In Vietnam, only a subset of those who are working report in Vietnam had at least three income streams over the last year earning individual wages or salaries. Due to a mix of factors since the date of interview across broad categories, reflecting including a large agricultural workforce, informality, and adequate diversification of activities and sources. Poorer classification of informal activities such as family businesses, households tend to have slightly higher income diversification earnings are not always reported as wage income. Lottery across broad categories. sellers are a particularly unique occupation, where workers in this sector (sector 92) often report they are employed but not Income diversification can be viewed as a coping earning wages because they earn income from commissions mechanism or a way to secure additional income. Poorer on tickets sold and thus report business income. Even with households tend to have more diversification. Diversification these distinctions, there is large variation in the type of work of income can be beneficial when earnings are through and quality of work associated with wage employment. passive incomes, but multiple labor-intensive activities can be Wage income can be derived from any sector (agriculture, inefficient. For poor households, diversifying within agriculture manufacturing, and services) and obtained through informal may be a way to raise their agricultural incomes or provide more work or skilled contract work. Reporting wages from the food to the household. In contrast, many households at the agriculture sector is uncommon, as the majority of households upper ends of the distribution are primarily wage earners. report it as part of self-employed family farming. Figure 2.11. Share of Vietnamese households receiving income from multiple sources, 2020 A. Average number of income sources, by decile B. Distribution of the number of income sources 3.5 40 # of households (in millions) Average number of income sources 3.0 35 2.5 30 25 2.0 20 1.5 15 1.0 10 0.5 5 0.0 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 (=poorest) (=richest) Source: World Bank staff calculations using the Vietnam Household Living Standards Survey, 2020. 62 2022 Vietnam Poverty and Equity Assessment Chapter 2. A review of the drivers of poverty reduction Measuring household income using the Vietnam household BOX 2.2. living standards survey For the purposes of welfare analysis, income sources are examined at the household unit. The examination of trends at the household level and considering all sources of income is one of the main differences between welfare analysis conducted using the household survey and other labor and income analyses using labor force surveys (LFS). Another difference is that the VHLSS records information on income for the last 12 months, while the LFS records labor market activity over the last month and does not include income sources from non-labor market activity, or detailed sector or business costs. Thus, labor diversification across households is more complete with the VHLSS than the LFS. Sources of household income are collected from the VHLSS. All income sources are collected at the household level with the exception of wages, which are collected at the individual level (Table B.2.2.1). These concepts are aggregated to construct current household income. This construction is World Bank developed and may not follow official GSO income aggregate construction methods. World Bank concepts are in net, and costs from farming or business operations are deducted. Taken together, income information from the VHLSS is broader and more complete than what is available from other data sources. Table B.2.2.1. Income categories and definitions INCOME CATEGORY NOTES 1 Family farms Net farm income. See Chapter 3 for agriculture sub-categories. 2 Non-farm family business Net household non-farm income. 3 Wages, bonus, and severance Wages are accumulated across the first, second, and third jobs over the last 12 months. Not all employed activities pay wages and salaries. Where individuals do not provide wage information, they are likely to report income sources under family businesses or agriculture modules. 4 Pensions Asked to individuals currently not working 5 Social Assistance Direct cash transfers from Module 4 and unemployment benefits. Does not include indirect benefits such as subsidies or loans. 6 Remittances Domestic and international remittances 7 Financial 8 Rent Rental or leasing incomes. 9 Other Net hunting, interest revenue, gifts from wedding or funeral, receipts from insurance payments. Some differences may appear between individual and household-level analysis of incomes and wages. For example, one main difference is that reliance on wages in manufacturing appears higher in the VHLSS, but in the LFS, employment in services is higher than in manufacturing. This is because the sector of activity may be related to wages, farm, or non- farm business incomes. These differences arise from taking into consideration multiple jobs from all working members in households and across the entire year. In the VHLSS, half of all workers in the primary job stated that they were working but did not receive a salary. These earnings may be reflected in family business (non-ag), or agriculture modules. For example, a common case are lottery workers who do not report any wage income. Inconsistency remains in how individuals report their activities, especially when these activities are informal. Labor market indicators are most accurately represented by the LFS, which details individual-level employment outcomes. Trends found in the household survey are consistent with broad labor trends illustrated earlier in this chapter. From the Last Mile to the Next Mile 63 PART 1. A DECADE OF SIGNIFICANT PROGRESS, BUT LAST-MILE CHALLENGES REMAIN From 2010 to 2020, total household wage incomes is marginal. Remittances are not a primary transfer as in increased in importance as the primary source of income some countries, such as Thailand (domestic) or Nepal and for households (Figure 2.12). The share of households the Philippines (international), where migration is much receiving wage income increased from 64 to 70 percent. more important. Lacking the large-scale worker migration Changes in the amount of household wage income from 2010 programs such as in the Philippines and Nepal, international to 2020 was also substantial. Annual household wage income migration and remittances are limited to richer segments of increased by nearly 100 million VND in nominal terms, or the population in Vietnam. Remittances from international nearly tripling in one decade. Income from non-farm family sources can be high but are also seldom reported. Over time, businesses also increased at a similar level as wages, but many a marginally higher share of households in 2020 received fewer households are engaged in this activity. remittance and social protection incomes than in 2010. Pension income increased over the last decade, but only about Both public and private transfers comprise a small share 10 percent of households received any form of pension in of average household income. Remittances and social 2020. The increase in the share of households with pension protection provide smaller amounts of household income, income is reflective of an aging population. Pension incomes and contributions remained relatively static over the decade. are primarily received by retirees from public sector positions Domestic remittances are the most cited income source (85 and wage jobs with contracts. percent of all households), but the average amount received Figure 2.12. Trends in household income components, 2010 and 2020 Annual household income (VND, million) 150 Annual household income 120 (VND, millions) 90 60 30 0 Wage, bonus, Family business Family farms Remittance Social assistance Pension Other income severance (cash transfer only) 2010 2020 Share of household with income source (%) 100 Share of households (%) 80 60 40 20 0 Wage, bonus, Family business Family farms Remittance Social assistance Pension Other income severance (cash transfer only) 2010 2020 Notes: Nominal values. World Bank-constructed income categories. See Box 2.2 for definitions. Average incomes are only for households who report having the income source. Social assistance income includes only selected direct cash transfers. Source: World Bank staff calculations using VHLSS 2010 and 2020. 64 2022 Vietnam Poverty and Equity Assessment Chapter 2. A review of the drivers of poverty reduction The poorest households remain reliant on Households are exiting agriculture on average, but the agriculture poor still remain significantly engaged. In 2010, 67.2 percent of households received agriculture income, compared Agricultural income is still common among poor to 53.8 percent in 2020. Across the various sub-components households and provides these households with only a of agricultural, participation has fallen across all categories, small amount of income compared to wage income. The particularly in low-revenue activities such as by-products gap between household-level agriculture, wage, and business or rice (more details in Chapter 3). However, engagement incomes has widened considerably over the last decade. In in agriculture remains high for extreme and moderate poor 2010, total agricultural household income was still lower than households (Figure 2.13). Agriculture is one of the lowest total wages or family business incomes, but by no more than revenue-generating economic activities that households can 20 million VND. But in 2020, the gap had widened to over be engaged in. When examining dynamics in nominal total 100 million VND per year. earnings, agriculture incomes remain similar in 2010 and 2020. Figure 2.13. Changes in household engagement in major sectors Share of household engaged in agriculture 100 Share of households (%) 80 60 40 20 0 1 2 3 4 5 6 7 8 9 10 (poorest decile) (richest decile) 2010 2020 Share of households engaged in manufacturing 60 Share of households (%) 40 20 0 1 2 3 4 5 6 7 8 9 10 (poorest decile) (richest decile) 2010 2020 Share of households engaged in services 80 Share of households (%) 60 40 20 0 1 2 3 4 5 6 7 8 9 10 (poorest decile) (richest decile) 2010 2020 Note: Households can be engaged in more than one major economic sector. Source: World Bank staff calculations using VHLSS 2010 and 2020. From the Last Mile to the Next Mile 65 PART 1. A DECADE OF SIGNIFICANT PROGRESS, BUT LAST-MILE CHALLENGES REMAIN Unpacking determinants of agricultural incomes is a key skill occupations, contract employment, manufacturing, and aspect to better understanding the chronic nature of poverty. services sectors. Even in the early half of the decade, while See Chapter 3 for a detailed discussion on agricultural the number of jobs in manufacturing was not increasing in income sources. net, jobs were still transforming and becoming more likely to be formal and were getting better paid. Labor in the services Variation in wage growth by job type sector expanded more than manufacturing, but services jobs were more likely to be informal. Informality is still high in Wage growth was high across all sectors over the last Vietnam but has declined significantly over time. Thirty-one decade but highest in manufacturing (Figure 2.14). percent of households had income from formal wage work in Statistics are shown at the household level but align with 2010, and this increased to 41 percent in 2020. national trends in wages from labor force surveys. Nominal wages increased across all sectors, doubling on average over the Growth in occupations by skill level has been primarily decade. Wage growth among poorer households was higher in in medium-skilled occupations over the last decade. The the first half of the decade than the second, coinciding with share of households with a worker in a medium-skilled wage faster poverty reduction in the early 2010s. By sector, wages job increased from 31 to 42 percent over the decade. Wages in manufacturing grew the fastest, followed by public sector are highest for high-skill jobs (leaders, high-level professionals, wages. Nominal wages in manufacturing nearly doubled and mid-level professionals), but only about 16 percent of between 2010 and 2014, and by 2019 they had nearly tripled. households have a working member at this skill level, and Between 2015 and 2020, wage growth slowed down slightly only 3 percent of households in the bottom quintile (Figure overall, with wages seeing declines in 2017 and then again 2.16). Moreover, the absolute number of high-skill jobs is not in 2020. increasing as quickly. The stunted growth of high-skill jobs could become a constraint to future growth (see Chapter 5 Participation in wage employment has increased for a discussion of this issue). Across different categories of dramatically over the last decade. From 2010 to 2020, the wage jobs, the poorest quintile earn only half or a third of the share of households that have at least one member receiving wages as the top quintile. Jobs in the poorest quintile are also wages or salaries has increased from 53.3 percent to 64.7 more likely to be informal; 18 percent of households in the percent (Figure 2.15). Participation in wage employment lowest quintile have income from formal wage employment increased substantially in some categories, including medium- compared to 52 percent of households in the top quintile. Figure 2.14. Wage trends for selected sectors (2010–2020) 350 8 Average monthly wages (VND, millions) 300 7 6 Wage index (2010=100) 250 5 200 4 150 3 100 2 50 1 0 0 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 Agriculture, forestry and fishing Manufacturing Construction Wholesale and retail trade; repair of motor Public administration and defence; All sectors vehicles and motorcycles compulsory social security Source: ILO database. 66 2022 Vietnam Poverty and Equity Assessment Chapter 2. A review of the drivers of poverty reduction BOX 2.3. Labor and wages, Vietnam vs. other EAP economies Compared to neighboring economies, Vietnam now has a larger share of the workforce in manufacturing. This dominance in manufacturing revealed itself in the last few years. Compared to regional neighbors, it is also the only country where the share of the labor force in manufacturing has been increasing in the last few years. In the short run, this is not necessarily a concerning trend since a large share of the services sector in Vietnam is informal and would pay lower wages and provide less stability for its workers. However, relying on a manual and physically demanding sector to provide jobs will become challenging as the population ages, and is also limiting growth in potentially more value-added services sectors. Figure B.2.3.1. Share of workers by selected sectors and economy Trade, transport, hotel, Agriculture Manufacturing restaurant Public services 70% 60% 50% Share of workers (%) 40% 30% 20% 10% 0% 2010 2015 2020 2010 2015 2020 2010 2015 2020 2010 2015 2020 Year Year Year Year Indonesia Malaysia Philippines Vietnam Lao PDR Myanmar Thailand Source: ILO database. Vietnamese wages on an international level are still low, at a few hundred US dollars per month. Low-cost labor was one of the main reasons behind increasing FDI and primarily in manufacturing. Wages in Vietnam doubled from 2010 to 2020, from $400 per month to over $800 per month (2017PPP). Average monthly earnings are higher than in Philippines but still trails the Upper-Middle Income regional neighbors of Thailand and Malaysia. Wages are still low at an international level and remain uncompetitive for economic activity relying on higher skills of higher value-added. Across sub-sectors, the wage premium for high-skill jobs is much larger than for manufacturing jobs. While Malaysia and Thailand have the highest wages across most categories, the premium for professional jobs is much greater than for lower-skill jobs. Vietnam will face challenges retaining high-skill labor force without increasing salaries. Continued > From the Last Mile to the Next Mile 67 PART 1. A DECADE OF SIGNIFICANT PROGRESS, BUT LAST-MILE CHALLENGES REMAIN BOX 2.3. Labor and wages, Vietnam vs. other EAP economies (Continued) Figure B.2.3.2. Wages by economy and selected sectors All Manufacturing Professional 3,000 3,000 3,000 2,500 2,500 2,500 Average monthly earnings (2017PPP) Average monthly earnings (2017PPP) Average monthly earnings (2017PPP) 2,000 2,000 2,000 1,500 1,500 1,500 1,000 1,000 1,000 500 500 500 0 0 0 2010 2015 2020 2010 2015 2020 2010 2015 2020 Malaysia Thailand Philippines Vietnam Cambodia Indonesia Myanmar Source: ILO Database. Figure 2.15. Evolution of household income from wages, 2020 vs. 2010 A. Share of households with wage income, by type of wage job B. Average household wage income, by type of wage job 80 200 Annual income (VND, millions) 70 Share of household (%) 60 150 50 40 100 30 20 50 10 0 0 Services Services Low-skill Low-skill Primary job With contract Agriculture Medium-skill High-skill Primary job With contract Agriculture Medium-skill High-skill Manufacturing Manufacturing 2010 2020 2010 2020 Note: Annual nominal household wage income. Source: World Bank staff calculations using VHLSS 2010 and 2020. 68 2022 Vietnam Poverty and Equity Assessment Chapter 2. A review of the drivers of poverty reduction Figure 2.16. Wage trends in 2020, by household quintile 60 250 50 200 Annual income (millions) Share of households (%) 40 150 30 100 20 50 10 0 0 Q1 Q2 Q3 Q4 Q5 Q1 Q2 Q3 Q4 Q5 (Bottom 20) (Top 20) (Bottom 20) (Top 20) Agriculture Manufacturing Services Low-skill Medium-skill High-skill Contract (1st) Note: Annual nominal household wage income. Source: World Bank staff calculations using VHLSS 2020. Non-farm family business income is less registered businesses (Figure 2.17). There are clear differences common, but large businesses can yield high in business income between types of family businesses, in earnings particular if they are registered and by size. Overall, nearly a third of household businesses tend to be retail, but registered Nationwide, non-farm family business income was the businesses are more likely to be in wholesale. second-most important contributor to upward economic mobility after wage growth. Household business income Direct public cash transfers are a small overall is less common than wage employment. Only about 30 contribution to average household income even percent of households receive income from this source. The for poor households prevalence of household businesses declined slightly over the last decade. The share of households with nonfarm business Direct cash transfers from social assistance programs income dropped slightly to 31 percent in 2020 compared to make up a small share of household incomes. The 34 percent in 2010, but the average income from household poorest households are more likely to receive direct social businesses rose substantially across all forms of businesses. assistance income, but annual amounts are still small relative The richest households did not reduce their participation in to other income sources (Figure 2.18). The poverty reduction family businesses; thus, changes in family business income interventions in Vietnam can be divided into direct cash, had a more modest effect on poverty rates, as most households indirect monetary support, and broader geographically that received income from businesses were already above the targeted public investments. Administratively, about 7 million poverty line. people are receiving direct cash support by being identified as a poor or near-poor household (Table 2.2). Benefit levels are Registered businesses are extremely uncommon but low, and coverage is limited. Benefit levels are also declining are highly profitable. Only 1.2 percent of households in value over time because they are not indexed to inflation. reported income from formally registered businesses. Small About 15 percent of households reported receiving a social household businesses and unregistered businesses are more benefit of some kind in the 2020 VHLSS (social assistance, common, but they earn only about half as much as large- merit, or disaster relief). From the Last Mile to the Next Mile 69 PART 1. A DECADE OF SIGNIFICANT PROGRESS, BUT LAST-MILE CHALLENGES REMAIN Figure 2.17. Family business income trends in 2020, by household quintile 25 600 Annual income (VND, millions) 500 Share of households (%) 20 400 15 300 10 200 5 100 0 0 Q1 Q2 Q3 Q4 Q5 Q1 Q2 Q3 Q4 Q5 (Bottom 20) (Top 20) (Bottom 20) (Top 20) Registered business Small household business Registered business Small household business No registration No registration Note: Household-level income. Source: World Bank staff calculations using VHLSS 2020. Figure 2.18. Household-level cash transfers by household quintile 25 7 Annual income (VND, millions) 6 Share of households (%) 20 5 15 4 10 3 2 5 1 0 0 Q1 Q2 Q3 Q4 Q5 Q1 Q2 Q3 Q4 Q5 (Bottom 20) (Top 20) (Bottom 20) (Top 20) 2010 2020 2010 2020 Note: Annual household nominal income among recipients. Source: World Bank staff calculations using VHLSS 2010 and 2020. Table 2.2. Social assistance beneficiaries ESTIMATED NUMBER OF AVERAGE BENEFIT AMOUNT CATEGORIES OF EXISTING VULNERABLE GROUPS BENEFICIARIES (MONTHLY) Whole country People in poor and near-poor households 7 million people VND 750,000 per person Social assistance beneficiaries 2.84 million people VND 1.5 million per person Merit people (National Devotees) 1 million people VND 1.5 million per person Note: Merit people (or National Devotees) are people who have contributed during revolution and war times. Dutta (2019) estimates for 2018 are used. 70 2022 Vietnam Poverty and Equity Assessment Chapter 2. A review of the drivers of poverty reduction Social assistance through direct cash transfers are small, 2.3. Forces behind but indirect benefits can be larger. There is a much wider set of programs spanning relevant ministries that provide poverty reduction– support to poor households in different forms. At the decompositions population level, support with health insurance premiums is the most commonly received support (24 percent), followed The previous sections discussed household profiles and by subsidies for electricity (6 percent) (Figure 2.19). Benefits income trends over the decade and alluded to potentially from these programs are not include in measures of household important drivers. This section quantifies the channels and current income. However, remaining types of benefits are drivers of poverty reduction by decomposing poverty rates by rarely cited to be received. Targeting performance of these broad groups and income sources, as well as across different additional programs varies. Social assistance schemes include periods over the decade.30 Nationally, poverty reduction was programs not just targeted to the poor but also based on higher during the first half of the decade, was concentrated categorical targeting—that is, people with disabilities, people in initially poorer areas, and was almost exclusively a rural with HIV, the elderly, and National Devotees (war veterans). phenomenon. From 2010 to 2016, LMIC $3.20/day Some programs are more received by poor households than 2011PPP poverty rates declined from 16.8 to 7.8 percent, and others. Food allowances and hardship allowances are almost from 2016 to 2020, poverty further declined from 7.8 to 5 exclusively received by households in the lowest decile of the percent. Progress in the latter half of the decade was slower, welfare distribution. Other programs are less well targeted, but the absolute percentage point change in poverty is not such as support for payment of health insurance premiums, just a function of progress, but also of the level of poverty policy scholarships, agricultural extension, and tuition in the initial year. The drivers of poverty reduction are exemptions. overwhelmingly attributable to growth factors throughout most of the decade, with a smaller role played by redistribution as a channel of poverty reduction. From 2010 to 2020, over 90 percent of poverty reduction at the LMIC poverty line was Figure 2.19. Percent of beneficiaries in the population, 2018 Other Food-emergency allowance Food allowances Financial hardship allowances for poor households Financial hardship allowance for people with merit Financial hardship allowance for low-income public employees Electricity bill support Support to clean/improve water sources Housing support for the poor Vocational training for the poor Policy scholarship Tuition exemption and discount for the poor Discount exemptions of medical examinations Support with health insurance premium Support with machines and materials for production Preferential credit for the poor Petrol support for fishing boats/vessels Support to access overseas employment Agriculture, forestry, and fishery extension Provision of arable land to poor ethnic minority households 0 5 10 15 20 25 Share of population (%) Non-poor Poor Note: Poor defined based on the $3.20/day 2011PPP poverty line. Source: World Bank staff calculations using VHLSS 2018. From the Last Mile to the Next Mile 71 PART 1. A DECADE OF SIGNIFICANT PROGRESS, BUT LAST-MILE CHALLENGES REMAIN attributed to growth of the mean of household consumption day 2011PPP) poor are positioned in distribution in the first (Figure 2.20). A growth-driven poverty reduction process decile. Observing changes in the bottom deciles, many poor occurred thanks to a decade of prosperity primarily due to households increased their economic participation in both an expansion in the non-farm labor force and rising wages. the manufacturing and services sectors, but as a whole more Redistribution contributed to poverty reduction in the earlier are engaged in manufacturing. Thus, manufacturing wages half of the decade, but still had a relatively small effect overall. played a stronger role in reducing LMIC poverty. Drivers of income inequality Figure 2.20. Growth-driven poverty reduction over the last decade Decomposition of inequality (as measured by GE(0)) 2 between population groups in Vietnam illustrate channels Percentage point change in poverty rate 1 of the changes in inequality based on the method developed 0 by Mookherjee and Shorrocks (1982). The rise in inequality −1 is largely related to income effects and widening urban and rural disparities. Figure 2.21 illustrates that inequality was −2 increasing between urban and rural areas, highlighting that −3 economic vibrancy and jobs in urban areas was key for poverty −4 reduction but the vibrancy did not always reach rural areas. −5 −6 Figure 2.21. Inequality decompositions – urban/rural 2010–2012 2012–2014 2014–2016 2016–2018 2018–2020 0.15 Growth Distribution 0.12 Note: World Bank LMIC Poverty Line ($3.20/day 2011PPP) 0.09 Point change Source: World Bank staff calculations using the VHLSS. 0.06 0.03 Labor market incomes are almost exclusively explanatory 0.00 of poverty reduction and are consistent with a growth- −0.03 driven poverty reduction process. Among sources of household income, labor income was the primary driver of −0.06 poverty reduction from 2010 to 2020, and these trends are 2010–2016 2016–2020 2010–2020 broadly consistent across regions and also ethnicity (Figure Pure inequality With effect A.2.1.1). Job creation and rising wages as a promotive strategy Between effect Income effect worked well throughout the decade, with high economic Note: Measure of change in GE(0) is based on real household consumption growth. However, workers are not as well protected by per capita. safety nets during downturns. The economic crisis caused Source: World Bank staff calculations using VHLSS 2010–2020. by COVID-19 revealed economic vulnerabilities of informal and labor migrant workers. Low levels of private and public transfers proved to be a source of vulnerability. Changes in sources of household income also affect inequality (Figure A.2.1.3). The decline in farm income By sector, poverty reduction over the last decade was is increasing income inequality in most regions. Since the attributable to growth in both manufacturing and poorest households are more engaged and dependent on services jobs (Figure A.2.1.2). Across the three broad agriculture income, a decline in this income source worsens primary sectors, the manufacturing sector welcomed the most inequality. This is compounded by strong growth trends households from the lower end of the welfare distribution, observed for other income sources that are well received by closely followed by the services sector. The LMIC ($3.20/ households at the higher end of the distribution. 72 2022 Vietnam Poverty and Equity Assessment Chapter 2. A review of the drivers of poverty reduction In the Central Highlands region, labor market conditions these areas. The small contribution to poverty reduction are worsening and inequality is increasing. Increasing from urban areas is explained by the fact that most people in inequality in the Central Highlands region is attributable to a these areas already live above the poverty line as defined by lower share of employed adults as well as less labor income per the LMIC PL. However, the COVID-19 pandemic has raised employed adult. This is consistent with previous descriptive new awareness to issues of livability and economic security statistics on the declining household-level economic activity in urban areas. Urban areas were by far more affected by in that region in Chapter 1 and will be further discussed in lockdown policies and mitigation measures than rural areas. Chapter 3. The Central Highlands region has experienced Schools in urban areas remained closed for much longer than the slowest poverty reduction over the last decade and, unlike in rural areas. Furthermore, business activity in urban areas any other region, the share of households engaged in wage experienced reduced hours or limited operations for a longer employment declined. time. COVID-19 revealed one of the main disadvantages of urban areas: crowded, small, and low-quality housing. Decomposition of poverty rates by groups Ethnic minorities have seen relatively slow poverty Using classifications of urban and rural areas in each reduction compared to the Kinh majority, but are survey round, poverty reduction appears to have making progress (Figure 2.24). The Kinh majority occurred almost exclusively in rural areas. The decline experienced higher absolute poverty reduction even after in poverty is attributed to rural areas (Figure 2.23).31 The taking into account that they started from lower initial poverty urban-rural poverty gap narrowed considerably over the last rates than ethnic minorities. In the latter half of the decade, decade. In 2010, rural and urban poverty rates were 22 and 4.5 a larger portion of overall poverty reduction is associated percent respectively, compared to 7 and 1.5 percent in 2020. with improvements among ethnic minorities, since poverty Geographic variation in poverty is still much larger by regions rates among the Kinh majority is now very low, at about than by the urban-rural classification. See Chapter 3 for more 1 percent. In the latter half of the decade, ethnic minority discussion on regional poverty reduction patterns. economic outcomes improved, as larger shares started to exit agriculture. Poverty rates among ethnic minorities remain While urban areas contributed little to overall LMIC high even though the absolute gap between the Kinh and poverty reduction, economic vulnerabilities exist in ethnic minorities is narrowing. Figure 2.22. Total poverty reduction Figure 2.23. Poverty reduction, by urban and rural areas 0.0 3.0 Percentage point change in poverty rate −2.0 0.0 -3.0 Point change −4.0 −6.0 −6.0 −8.0 −9.0 −10.0 −12.0 −12.0 −15.0 2010–2014 2016–2020 2010–2020 2010–2016 2016–2020 2010–2020 Rural Urban Population shift effect Interaction effect Note: World Bank LMIC Poverty rates ($3.20/day 2011PPP). Note: World Bank LMIC Poverty rates ($3.20/day 2011PPP). Source: World Bank staff calculations using the VHLSS 2010–2020. Source: World Bank staff calculations using the VHLSS 2010–2020. From the Last Mile to the Next Mile 73 PART 1. A DECADE OF SIGNIFICANT PROGRESS, BUT LAST-MILE CHALLENGES REMAIN Figure 2.24. Poverty reduction, by ethnicity Effects from ageing are starting to show 2.0 Change in poverty rate (percentage point) Poverty reduction decompositions with age-specific 0.0 factors are showing some early impacts of ageing. −3.0 Household employment shares and labor income can be decomposed to be sourced by either young or old household −6.0 members, where younger cohorts are those under 35 years of −9.0 age. There are some small indications suggesting lower labor force participation among the youth cohort as well as lower −12.0 earnings. These factors can be driven by multiple factors, −15.0 such as youth cohorts staying in school for longer, but it is 2010–2014 2016–2020 2010–2020 not clear if a smaller poverty reduction impact from youth income is associated with lower productivity or lower-quality Kinh Ethnic Minority Population shift effect Interaction effect work. The profile of the youth cohort in the labor market is discussed further in Chapter 5. Note: World Bank LMIC Poverty rates ($3.20/day 2011PPP). Source: World Bank staff calculations using VHLSS 2010–2020. 2.4. Summary By education levels, households with fewer members The role of economic growth in supporting poverty having attended secondary education experienced reduction continued in importance throughout the the largest absolute decline in poverty rates, which last decade. The role of certain income sources and sectors is consistent with labor expansion based on low-skill for poverty reduction was clearly illustrated, both in terms manual labor. Due to dynamic household profile changes, of those that matter most (e.g., wage growth), and those population shift effects are the largest for decompositions that played a lesser role (e.g., social assistance). However, by education and sector groups. A large (and negative) challenges are looming on the horizon related to the factors population shift effect is consistent with descriptive statistics that have sustained growth and poverty reduction in the past. of large shifts in the composition of household education These include high informality, low skills, low wages, and low levels and sectors of activities toward groups that had initially growth in high-skill jobs (Chapter 5). Furthermore, to better lower poverty rates. This is consistent with households having guard against risks and provide broader safety nets, the role improved profiles of education and shifting out of agriculture. of social assistance and protection must increase (Chapter 6). 74 2022 Vietnam Poverty and Equity Assessment Chapter 2. a review of the drivers of poverty reduCtion BOX 2.4. Group and income decompositions Huppi-Ravallion decomposition of poverty rates by population groups Profiling of the population is already suggestive of the drivers and determinants of poverty reduction since some characteristics were more dynamic than others. To understand the scale of poverty reduction, the contribution changes in poverty rates can be decomposed and attributed to progress across different groups based on household-level characteristics. Given that the national poverty rate is an average of poverty rates of specific population groups, weighted by the population share of each group, changes in poverty rates can also be thought of as the result of changes in poverty incidence in some groups, the relative size of groups, or a mixture of both. A decomposition (Ravallion and Huppi, 1991) that separates the total poverty change into these components can be formalized as follows. Given poverty rates for final and initial year ( and , respectively): where stands for the poverty incidence and represents the population share in group k for initial (final) period. 1. The first right-hand term represents the intra-sectoral effect, that is, the change in poverty that can be attributed to changes in the incidence of poverty in each k group of the population, assuming the relative size of the population groups remains as in the initial year of the comparison. 2. The second term stands for the population shift effects, namely, the change in poverty that is due to changes in population shares, assuming poverty incidence in each group stays at the level of the initial period. 3. The third term is an interaction effect, which indicates if there is correlation between changes in poverty incidence and population movements. A positive sign of the interaction effect would indicate—on average—a growing population share among groups where poverty is growing, which in turn would imply a smaller decline (or a larger increase) in the poverty rate. A negative sign, on the other hand, would imply that groups with declining (increasing) poverty also have a growing (declining) population share, which accelerates poverty reduction. Shapley income decomposition Early strategies to decompose poverty primarily decomposed changes targeting summary statistics. Changes in poverty were decomposed into effects from growth and redistribution (Datt and Ravallion, 1992), growth, redistribution, and prices (Kolenikov and Shorrocks, 2005), or changes in groups (Ravallion and Huppi, 1991). These decompositions targeted summary statistics, or the mean of the distribution. Yet, measuring changes in the aggregate may be misleading since economic growth may disproportionally impact growth across the distribution. It is most informative to analyze how the distribution changes over time from various factors. More recent decompositions utilizing the full distribution of consumption enable the measurement of the contributions to poverty reduction from various components. Changes in the distribution of household consumption per capita across two periods can be decomposed into contributions by calculating a Shapley-Shorrocks estimate for each component based on a modified methodology from Barros et al. (2006). The adaptations to the methodology contribute to the Continued > From the Last Mile to the Next Mile 75 part 1. a deCade of siGnifiCant proGress, But Last-MiLe ChaLLenGes reMain BOX 2.4. Group and income decompositions (Continued) literature in several ways (Inchauste et al., 2014; Azevedo et al., 2013). First, the updated methodology focuses on consumption rather than income. Second, the decomposition computes a cumulative counterfactual distribution by adding one variable at a time. Third, cumulative decompositions are calculated along every possible path and averaged to address path dependence. There are two caveats with respect to the described method. First, it is not possible to infer whether the changes are in returns or in endowments. For example, did wages increase because the working population became more educated, or did the minimum wage increase? Second, the method is not a general equilibrium method, meaning that the counterfactual distributions are not the result of an economic equilibrium (Azevedo et al., 2013). This implies that the contributions to poverty changes do not have a causal effect, but rather are an exercise used to detect distributional changes. It is assumed that by keeping all other components constant, each income component is allowed to change, and the effect on the change in the distributional statistics of the welfare aggregate is detected. The richness of household surveys allows changes in poverty across two cross-sections to be decomposed into multiple explanatory factors. Poverty is a function of household consumption, and changes in poverty can be decomposed into the factors, as illustrated in Figure B.2.4.1. The decomposition is an identity that can be expanded or reduced based on data availability. For example, labor and non-labor income can be further decomposed if desired. Figure B.2.4.1 Household consumption per capita—a decomposed identity Propensity to consume Share of occupied Household consumption adults per capita Household Share of adults Labor income income per capita per adult Labor income per Income per adult occupied adult Non-labor income per adult Note: An illustration of Equation 5. Household per capita income ( ) is total household income from all members ( ) divided by household size ( ). (1) Assuming that only adults aged 15+ contribute to household income, the identity can be rewritten as below. Working-age adults can receive wage income, while adults past retirement can receive pension income. Continued > 76 2022 Vietnam Poverty and Equity Assessment Chapter 2. a review of the drivers of poverty reduCtion BOX 2.4. Group and income decompositions (Continued) (2) Income per adult can be separated into labor and non-labor sources. In the case of Vietnam, labor income includes wage, net business, and net farm components that are described in Annex 2.2. Non-labor income includes public transfers, remittances, and other sources. (3) Finally, in the case of labor income, we need to only account for employed or occupied adults. (4) To create a mapping between household consumption and income, a variable, propensity to consume, is introduced. Household consumption per capita is expressed as below and is also identical to the illustration shown in Figure B.2.4.1. (5) To measure the contribution of the change in poverty from different components, the identity in Equation (5) is used. The contribution of each component to the total change in poverty is the average Shapley-Shorrocks estimate across all possible paths. Table B.2.4.1. Variables and definitions VARIABLE DEFINITION Household consumption per capita Propensity to consume Household income per capita Total household income (household h) Total income from household member i Total labor income from household member i. Assuming labor income only from occupied or employed adults. Total non-labor income from household member i. Number of household members Number of adult household members aged 15+ Number of adult household members aged 15+ and are also employed or occupied From the Last Mile to the Next Mile 77 PART 1. A DECADE OF SIGNIFICANT PROGRESS, BUT LAST-MILE CHALLENGES REMAIN 2.5. References World Bank. 2012. Well Begun, Not Yet Done: Vietnam’s Remarkable Progress on Poverty Reduction and the Emerging Challenges, 2012 Azevedo, Joao Pedro, Gabriela Inchauste, Sergio Olivieri, Jaime Saavedra, Vietnam Poverty Assessment. Washington, DC: World Bank. and Hernan Winkler. 2013. Is Labor Income Responsible for ---------------. 2018a. Poverty and Shared Prosperity 2018: Piecing Poverty Reduction? A Decomposition Approach. Policy Research Together the Poverty Puzzle. Washington, DC: World Bank. Working Paper No. WPS 6414. Washington, DC: World Bank. ---------------. 2018b. Vietnam Future of Jobs. Washington, DC: Barros, Ricardo Paes de. Carvalho, Mirela de. Franco, Samuel. World Bank. Mendoça, Rosane (2006). “Uma Análise das Principais Causas da ---------------. 2021. A Year Deferred – Early Lessons and Experiences Queda Recente na Desigualdade de Renda Brasileira.” In: Revista from COVID-19 in Vietnam. Washington, DC: World Bank. Econômica. Volume 8, número 1, p.117-147. Universidade Federal Fluminense. Rio de Janeiro. Datt, Gaurav, and Martin Ravallion. 1992. Growth and Redistribution Components of Changes in Poverty Measures: A Decomposition with Applications to Brazil and India in the 1980s. Journal of 2.6. Notes Development Economics, 38: 275–296. 24  This chapter limits analysis to the past decade. Poverty trends in Dutta, Puja. 2019. Access to Social Assistance among Ethnic Minorities the past decade are difficult to compare to the previous decade due to in Vietnam. Note. changes in data in 2010. Historical discussions of poverty trends in Gill, Indermit, Ana L. Revenga, and Christian Zeballos. 2016. Grow, Vietnam can be found in these reports: Glewwe, Gragnolati, and Zaman Invest, Insure: A Game Plan to End Extreme Poverty by 2030. (2002) and World Bank (2012). Policy Research Working Paper Series, No. 7892. Washington, DC: 25  Studies on intra-household poverty exist, but most of them cannot World Bank. disaggregate consumption by household members, and the only Glewwe, Paul, Zaman Hassan, and Michele Gragnolati. 2002. Who individual-level disaggregation outcome is education, which is more Gained from Vietnam’s Boom in the 1990s? An Analysis of Poverty easily measured (World Bank, 2018a). and Inequality Trends. Economic Development and Cultural Change 26  50(4): 773–792. This categorization is based on the primary activity of each working Ha, Thi and Anh Minh. 2021. Citizens, Businesses Hurt as Rising Prices household member. 27  Raise Inflation Concerns. VNExpress, May 19, 2021. https://e. Vietnam’s General Statistics Office (GSO) reported a 2 percent vnexpress.net/news/business/economy/citizens-businesses-hurt-as- decline in household income in nominal terms (https://e.vnexpress.net/ rising-prices-raise-inflation-concerns-4279567.html. news/business/data-speaks/binh-duong-overtakes-major-cities-tops- Inchauste Comboni, Maria Gabriela, João Pedro Azevedo, B. Essama- per-capita-income-4282618.html). Nssah, Sergio Olivieri, Trang Van Nguyen, Jaime Saavedra- 28  Consumption based poverty rates are measured bi-annually between Chanduvi, and Hernan Winkler. 2014. Understanding Changes in 2018 and 2020. Poverty. Washington, DC: World Bank Group. 29  For the sake of diversification, income sources are first categorized Kolenikov, Stanislav, and Anthony Shorrocks. 2005. “A Decomposition only by size groupings in Box 2.2. Any agriculture income is considered Analysis of Regional Poverty in Russia.” Review of Development Economics 9 (1):25–46. to be the same category, whether it is from crops or fishing. For example, if a household grows multiple types of crops, it is still considered to have Ravallion, Martin, and Monika Huppi. 1991. Measuring Changes in Poverty: A Methodological Case Study of Indonesia during an one type of income source rather than separate income streams for the Adjustment Period. The World Bank Economic Review 5(5): 57–82. purposes of this analysis. Wage income is also considered to be from one group initially, and not disaggregated by sectors or occupations. ILO Online Database (https://www.ilo.org/global/statistics-and- 30  databases/lang--en/index.htm) See Box 2.4 for technical details on decomposition methods. 31  Shapley, L. 1953. A Value for n-person Games. In H. W. Kuhn and A. W. All areas from 2010 to 2018 maintained the same urban and rural Tucker (eds.). Contributions to the Theory of Games. Princeton, classifications as they were sampled from the 2009 Census. The urban NJ: Princeton University Press. and rural area classifications were changed only for the VHLSS 2020 Shorrocks A.F. 1999. Decomposition Procedures for Distributional data set, since it was the first to be sampled from the 2019 Census. Analysis: A Unified Framework Based on Shapley Value. University of Essex and Institute for Fiscal Studies. Mimeo. 78 2022 Vietnam Poverty and Equity Assessment Chapter 2. A review of the drivers of poverty reduction ANNEX 2.1. Chapter 2 Figures and Tables Figure A.2.1.1. Poverty reduction from Figure A.2.1.2. Poverty reduction from 2010 to 2020, 2010 to 2020, decomposition results decomposition results including wages by sector LMIC ($3.20/day 2011PPP) National 15 LMIC ($3.20/day 2011PPP) 10 5 15 Decomposed change in 0 Decomposed change in poverty rate (p.p) poverty rate (p.p) −5 10 −10 −15 5 −20 −25 0 −30 −35 −5 Southeast National Midlands & Northern Mountains Northern & Coastal Center Central Highlands Red River Delta Mekong Delta Kinh majority Ethnic minority −10 −15 −20 2010–2020 2014–2020 Adult share Employed share Farm income Adult share Employed share Labor income (p.e) Non-farm income Non-labor income Farm income Labor income (Ag, p.e) Propensity score Labor income (Manu, p.e.) Labor income (Serv, p.e) Non-farm income Propensity score Note: Consumption-based poverty. Note: Consumption-based poverty. Source: World Bank staff calculations. Source: World Bank staff calculations. Figure A.2.1.3. Household income decomposition, 2010-2020 0.09 0.06 Decomposed changed in 0.03 Gini index (p.p.) 0.00 −0.03 −0.06 −0.09 −0.12 −0.15 National Midlands & Red River Northern & Central Southeast Mekong Ethnic Kinh Northern Delta Coastal Center Highlands Delta minority majority Mountains Adult share Employed share Labor income (p.e) Farm income Non-farm income Non-labor income Note: Household income Gini. Source: World Bank staff calculations. From the Last Mile to the Next Mile 79 PART 1. Chapter 3. Challenges to reducing poverty among the remaining poor Chapter 3. Challenges to reducing poverty among the remaining poor Key Messages • Chronic poverty continues to be concentrated in households primarily engaged in agriculture, ethnic minorities, and those living in rural areas. • Despite chronic challenges for these groups, some trends are shifting: ‒ Ethnic minorities are now experiencing a similar pace of poverty reduction as the Kinh, though absolute gaps remain due to delayed progress. ‒ Regionally, the Central Highlands is now the poorest region. Poverty also increased in the Mekong Delta in 2020. 80 2022 Vietnam Poverty and Equity Assessment Chapter 3. Challenges to reducing poverty among the remaining poor Poverty rates are still much higher among rural and rates depending on household location and circumstances. ethnic minority families and households primarily Over the last decade, nearly 10 million people escaped LMIC engaged in agriculture. These groups face persistent poverty ($3.20/day 2011PPP), but gaps across sub-groups did challenges of lower human capital, lower-quality local public not always narrow. For example, when examining households services, greater distance to economic opportunities, and less by their economic engagement in agriculture, the absolute access to financing or training. This chapter describes the difference in poverty rates remain similar in 2010 and 2020. patterns of poverty among these groups in further detail and reviews prominent constraints and challenges. Economic centers with the highest economic growth also saw the highest household consumption growth among the poor 3.1. Introduction – household characteristics The Southeastern region had the highest average growth in household expenditure per capita. This region grew chronically related to the fastest and by almost one percentage point higher than higher poverty rates the Northern and Coastal Center or Red River Delta regions (Figure 3.2). High-growth regions have high-performing Poverty rates are known to be chronically higher urban areas (HCMC, Da Nang, and Hanoi). The Mekong by some household characteristics Delta is the slowest growing region on average. Only in two instances over the last decade did a region experience average There is an interplay between ethnicity, agriculture, negative growth. The first case was from 2012 to 2014 in the geography, and poverty in Vietnam. Remote Central Highlands region, and the second case was from 2018 mountainous areas where the poor are concentrated are to 2020 in the Mekong Delta. The contraction in the Mekong heavily populated by ethnic minorities. Ethnic minorities Delta from 2018 to 2020 was more severe, with household are also disproportionately more engaged in agriculture than consumption growth being negative across all deciles of the the Kinh majority. These remote areas are located far from distribution. Over the entire decade, the poorest households economic centers and - partly due to unfavorable topology, experienced the highest growth rates in the regions with the agriculture is also less productive there. Group-wise poverty largest cities, the Red River Delta and Southeast regions. rates illustrate the sometimes-large differences in poverty rates (Chapter 1 discusses regional and small area poverty trends). by household characteristics (Figure 3.1). When households are divided by education levels,32 geographic region, primary The poorest regions are the smallest in population and economic sector, or ethnicity, large variations in within-group experiencing the slowest population growth. Population poverty rates emerge, and the differences between the groups size across the six regions ranges from about 5 million to are more apparent. over 20 million people (Figure 3.3). About 20 percent of all households live in the poorer Central Highlands and Characterizations of groups with the highest poverty Midlands and Northern Mountains regions (Figure 3.4). rates have remained largely the same over the last decade Over the shorter-term horizon between 2018 and 2020, the (World Bank, 2012). Households that were already nearest populations are estimated to have declined in the two poorest to markets and better educated in 2010 were most able to take and least populated regions: the Central Highlands and the advantage of the rapidly changing economy and structural Midlands and Northern Mountains regions. It is likely that transformations. However, the least dynamic economic the slowdown in population growth is also partly due to sectors remained weak, and regions without strong economic outmigration since these two regions are the most heavily growth poles experienced the lowest population growth while populated with ethnic minority households, which tend to cities and industrial zones expanded quickly. Livelihoods have more children than the Kinh majority group. have broadly improved in absolute terms but at different From the Last Mile to the Next Mile 81 PART 1. A DECADE OF SIGNIFICANT PROGRESS, BUT LAST-MILE CHALLENGES REMAIN Figure 3.1. Poverty rates by dimensions revealing chronic poverty Region Household engagement in agriculture 50 50 39.9 Poverty rate (%) Poverty rate (%) 40 40 34.0 35.2 32.8 29.6 29.1 29.3 30 25.8 30 23.4 23.5 22.7 27.6 22.0 21.3 18.6 17.1 20 20 13.6 20.4 19.7 14.0 14.3 12.1 16.0 9.0 10.1 7.8 10 11.5 10 8.6 4.8 6.8 6.2 4.7 4.9 4.3 4.7 3.6 4.5 2.3 5.3 1.4 2 1.1 3.7 6.9 3.9 0.7 1.0 0 3.0 2.3 0 0.4 0.4 1.2 0.6 2020 2020 2010 2012 2016 2018 2010 2012 2016 2018 2014 2014 Central Highlands Northern & Coastal Center No agriculture Mekong Delta Red River Delta Some agriculture Midlands & Northern Southeast Only agriculture Mountains Urban vs rural Ethnicity 60 60 56.8 50 50 47.7 46.8 Poverty rate (%) 40 40 35.9 30 30 34.6 Poverty rate (%) 22.1 27.2 20 20 15.3 16.9 10.8 9.4 9.4 10 7.0 10 4.4 4.6 3.7 2.8 2.1 1.2 1.3 1.0 1.5 1.1 6.4 0 0 2020 2010 2012 2014 2016 2018 2020 2010 2012 2016 2018 2014 Rural Urban Ethnic minority Kinh Note: Poverty rate according to the $3.20/day 2011PPP poverty line. Source: World Bank staff calculations using VHLSS 2010–2020. Figure 3.2. Average annualized household consumption growth, by region Midlands and North and National Urban Rural Northern Red River Central Coastal Central Southeast Mekong Mountains Delta Highlands Region Delta 10 Annual growth (%) 5 0 −5 2010–20 2010–12 2012–14 2014–16 2016–18 2018–20 2010–20 2010–12 2012–14 2014–16 2016–18 2018–20 2010–20 2010–12 2012–14 2014–16 2016–18 2018–20 2010–20 2010–12 2012–14 2014–16 2016–18 2018–20 2010–20 2010–12 2012–14 2014–16 2016–18 2018–20 2010–20 2010–12 2012–14 2014–16 2016–18 2018–20 2010–20 2010–12 2012–14 2014–16 2016–18 2018–20 2010–20 2010–12 2012–14 2014–16 2016–18 2018–20 2010–20 2010–12 2012–14 2014–16 2016–18 2018–20 Notes: Real household consumption per capita per day (2011PPP). Source: World Bank staff calculations using VHLSS. 82 2022 Vietnam Poverty and Equity Assessment Chapter 3. Challenges to reducing poverty among the remaining poor Figure 3.3. Populations by region, 2010-2020 Figure 3.4. Population distribution by region, 2020 Number of households (in millions) 30 25 1.5m Population (in millions) 20 5.3m 4.6m 15 10 3.3m 6.6m 5 5.5m 0 2010 2012 2014 2016 2018 2020 Central Highlands Mekong Delta Midlands & Northern & Coastal Central Northern Mountains Southeast Central Highlands Mekong Delta Red River Delta Midlands & Northern Mountains Northern & Coastal Central Red River Delta Southeast Source: World Bank staff calculations using the VHLSS, 2020. The richer regions are growing faster. The Red River In 2020, ethnic minorities constituted about 4 out of Delta remains the most populous region, but the Southeast the remaining 5 million poor. In contrast, only 29 percent region is growing faster. The primary cities in Vietnam have of the poor were ethnic minorities in 1998, highlighting the different degrees of influence on the regional statistics. The slower pace of poverty reduction realized by this group. Over population of HCMC is more than half the population of the entire decade, the annualized growth rate of household the entire Southeast region, and thus is a primary driver consumption per capita from 2010 to 2020 was about 10 of statistics in the region. Hanoi has less than one-fifth the percent per year for both the Kinh and ethnic minorities. population of the Red River Delta. These are the two largest From 1998 to 2010, per capita consumption grew at an annual cities and economic poles in Vietnam; the Southeast and Red rate of 7.4 for ethnic minorities and 9.4 for Kinh (World River Delta are home to 40.5 percent of all households and are Bank, 2012). Despite improving growth rates, since ethnic where industrial zones are most concentrated. minorities started at a lower consumption level, absolute gaps in household consumption are still high. Poverty rates among ethnic minorities have halved over the decade Ethnic minorities are most populous in the Midlands and Northern Mountains regions (Figure 3.5). While In 2020, ethnic minorities comprise about 78 percent poverty rates among ethnic minorities are highest in both the of the LMIC ($3.20/day 2011PPP) poor, despite being Midlands and Northern Mountains and Central Highlands only 15 percent of the total population. A bit less than regions, a large share of ethnic minorities in the North are one-third of ethnic minorities are poor based on this standard economically secure and even middle class. Poverty rates compared to just 1.3 percent of Kinh. This is not to say among ethnic minorities are also declining much faster in poverty reduction did not occur among ethnic minorities. In the Midlands and Northern Mountains, and in 2020, are at fact, the number of ethnic minorities who are poor has halved, similar rates to the ethnic minority poverty in the Northern from about 7.8 million in 2010 to 3.9 million in 2020, even and Coastal Central region (Figure 3.6). Thus, across 53 with high fertility rates and population growth. However, ethnic minority groups, there is still high variation in living the progress of exiting poverty among ethnic minorities is conditions. While the Kinh majority is the largest and most much slower than among the Kinh, who managed to reduce geographically spread out ethnic group across the country, their poor population from 7.1 million in 2010 to about the remaining 53 ethnic minority groups are scattered and 1.1 million in 2020. tend to be concentrated in their local areas. In 2016, the poorest EM groups were the La Hu, Mang, and Mong groups From the Last Mile to the Next Mile 83 PART 1. A DECADE OF SIGNIFICANT PROGRESS, BUT LAST-MILE CHALLENGES REMAIN (Chi et al., 2018). The Muong, San Diu and Khmer were household assets among different ethnic minority groups. identified among the group of ethnic minorities with higher Even among the same ethnic minority group, gender-related levels of socio-economic development (World Bank, 2019). factors such as the sex of household head are associated with Meanwhile, Kho Mu, Mong and Xo Dang were among lower further variation in socioeconomic outcomes (UNWOMEN, performing groups. Significant variations exist in per capita 2021). Due to limited scope, an in-depth analysis on income, poverty headcount, and non-monetary aspects such heterogeneity in socioeconomic development between and as housing conditions, access to water and adequate hygiene, within ethnic groups are deferred for future research. Figure 3.5. Economic class distribution, by region and ethnicity, 2020 25 20 Population (millions) 15 10 5 0 Ethnic Kinh Ethnic Kinh Ethnic Kinh Ethnic Kinh Ethnic Kinh Ethnic Kinh minority minority minority minority minority minority Midlands & Northern Red River Delta Northern & Coastal Central Highlands Southeast Mekong Delta Mountains Central Region Extreme poor Moderate poor Economically vulnerable Economically secure Middle class Note: Based on ethnicity of the household head. Source: World Bank staff calculations using VHLSS 2020. Figure 3.6. Poverty rates among Figure 3.7. Poverty rates among Kinh households ethnic minority households 80 15 70 12 60 Poverty rate (%) Poverty rate (%) 50 9 40 30 6 20 3 10 0 0 2010 2012 2014 2016 2018 2020 2010 2012 2014 2016 2018 2020 Central Highlands Mekong Delta Central Highlands Mekong Delta Midlands & Northern & Coastal Midlands & Northern & Coastal Northern Mountains Central Region Northern Mountains Central Region Red River Delta Southeast Red River Delta Southeast Note: Poverty rate according to the $3.20/day 2011PPP poverty line. Poverty rates Note: Poverty rate according to the $3.20/day 2011PPP poverty line. among the EM sample in Red River Delta is more variable due to small sample size. Source: World Bank staff calculations using VHLSS 2010-2020. Source: World Bank staff calculations using VHLSS 2010-2020. 84 2022 Vietnam Poverty and Equity Assessment Chapter 3. Challenges to reducing poverty among the remaining poor Households relying on agriculture remain the The profile of the poor vs. the population as a poorest whole The poorest households remain the most engaged Comparing the composition of the poor to the in agriculture. Households who receive their income composition of the population is also informative. from only family farming activities comprise 16 percent These comparisons tell a similar story about which of the population, while they represent 66 percent of the groups may be overrepresented in poor populations. For LMIC poor. About 38 percent of households in the lowest example, agricultural, ethnic minorities and households consumption decile rely on agriculture as their primary in the mountainous regions are overrepresented in poor source of income. Hence, agriculture is particularly important populations. Seventy-seven percent of the $3.20/day poor in for the livelihoods of the poor, and these households also 2020 are households with agriculture as a main activity. engage in subsistence agriculture. Outside of family farming, about 10 percent of wage workers are engaged in either the Generally, groups with higher poverty rates also agricultural sector or in food and beverage processing in account for most of the poor population, but it is not the manufacturing sector. For the sake of analysis on family always the case. In 2020, ethnic minorities were 15 percent farming, these wage workers are not considered. of the population, but 79 percent of the LMIC poor. Also, 42 percent of the poor live in the Midlands and Northern Poverty among agricultural households dropped in the Mountains, even though they are 13 percent of the national first half of the decade, but then stagnated in the second population (Figure 3.8). However, not all groups are the half of the decade. Among households engaged only in same size, and the share of the poor population can be high agriculture, LMIC-PL ($3.20/day 2011PPP) poverty rates in groups with lower poverty rates. For example, 20 percent declined from 35 to 23.4 percent during the first half of the of the LMIC poor population reside in the Northern and decade, from 2010 to 2016. However, from 2016 to 2020, Central Coast region, even though the poverty rate in the poverty among households only engaged in agriculture reduced region is only 4.7 percent. Thus, regional averages in poverty only from 23.4 to 21 percent. The size of the agricultural labor may be too ambiguous, and more granular statistics are force remained steady in the first half of the decade, and then needed at least at the provincial level. declined significantly over the second half. Households with partial or full engagement in manufacturing or services tend to Chronic dimensions of poverty are well known and have lower poverty rates than households who are solely reliant well reflected in policy priorities. In the 2021–2025 in agriculture. For example, poverty rates for households in 2020 Socioeconomic Development planning phase, three National solely engaged in agriculture was about 21 percent, compared Targeting Programs (NTPs) are aimed at assisting individuals, to 1 percent for households not engaged in agriculture at all. households, and areas based on the dimensions of high The significant of sectors of economic engagement in poverty poverty (rural, mountainous, engaged in agriculture, and reduction is discussed in Chapter 2. ethnic minorities) that are discussed in this chapter. NTPs are a broad-based policy instrument focused on alleviating The profile of agricultural households in Vietnam is poverty for the extreme poor in Vietnam. The mechanisms of largely unsurprising. In line with common perception, NTPs are broad, ranging from large-scale public investment, households with smaller land sizes, that have older household loans, and credits to extension services. members, are ethnic minorities, less educated and headed by women, and living in larger homes with dependent children The following sections describe these groups individually, are more likely to be poor. Remoteness and the extent of but as stated at the outset, there is an interplay between reliance on agricultural income increase the risk of poverty. these chronic characteristics. Section 3.2 discusses The relationship between the age of the household head constraints to poverty reduction in remote areas, Section 3.3 and poverty likelihood is convex (U-shaped), implying that discusses challenges faced by ethnic minorities which are more as people grow older, the risk of poverty increases. The risk numerous, and Section 3.4 describes agricultural incomes and of poverty is higher for those mainly dependent on farm challenges to low levels of income generation in this sector. activities other than non-farm activities. From the Last Mile to the Next Mile 85 PART 1. A DECADE OF SIGNIFICANT PROGRESS, BUT LAST-MILE CHALLENGES REMAIN Figure 3.8. Population profiles by groups Region Household engagement in agriculture 2010 2020 2010 2020 100% 100% 18% 13% 21% 19% 17% 16% Share of the population (%) Share of the population (%) 45% 21% 22% 62% 31% 23% 21% 50% 42% 50% 27% 66% 25% 25% 59% 24% 23% 20% 28% 15% 18% 16% 0% 0% Poor Population ($3.2) Population Poor Population ($3.2) Population Poor Population ($3.2) Population Poor Population ($3.2) Population Central Highlands Mekong Delta Not engaged in family farming Midlands & Northern Northern & Coastal Some engagement in family farming Mountains Central Only engaged in family farming Red River Delta Southeast Urban vs Rural Ethnicity 2010 2020 2010 2020 100% 100% 8% 11% 36% 21% 30% Share of the population (%) 47% 85% Share of the population (%) 84% 50% 50% 92% 70% 89% 64% 79% 53% 0% 0% 16% 15% Poor Population ($3.2) Population Poor Population ($3.2) Population Poor Population ($3.2) Population Poor Population ($3.2) Population Urban Rural Kinh majority Ethnic minorities Source: World Bank staff calculations using VHLSS 2010, 2020. 86 2022 Vietnam Poverty and Equity Assessment Chapter 3. Challenges to reduCing poverty among the remaining poor 3.2. Poverty across regions Coastal Central region. Industrial zones are primarily dotted around the large cities of HCMC and Hanoi, providing an Economic development and poverty reduction in remote abundance of factory jobs. The poorest two regions are the and rural areas are by nature more challenging. The Central Highlands and Midlands and Northern Mountains, remaining poor in these areas are constrained by a lack of which also have higher shares of regional GDP associated with human, physical, and financial capital. Households in poor agriculture. One exception is the Mekong Delta region, where areas are disadvantaged by their distance to economic centers, the agricultural share of GDP is also high, but poverty is lower. unfavorable topography, or limited access to land. Given One explanation is that households in the Mekong Delta are these well-known determinants and regional poverty trends more engaged in the country’s higher-valued aquaculture, already described in Chapter 1, this section provides more and cultivation of its flat land rice is also more productive information on the important relationship between regional than elsewhere in the country. The Central Highlands and the income dynamics and poverty reduction. Mekong Delta region have the lowest composition of GDP derived from industry. Regions with less vibrant economic centers were less successful at reducing poverty Regions with high concentrations of industrials parks also have high shares of households relying on wage Much of the variation in geographic poverty trends income from the manufacturing sector. Unsurprisingly, is reasonably correlated with Vietnam’s terrain and at the higher end, the Southeast region leads with the location of economic growth poles. Regional differences largest proportion of households with income from wage in poverty trends are strongly associated with the presence of jobs characterized by having formal contracts or that are in economic centers and cities. Regions with the lowest poverty the manufacturing or services sectors. At the national level, rates contain the three largest and most prosperous cities. The growth in manufacturing wages was particularly high and Red River Delta is home to Hanoi city, HCMC is located in an important driver of poverty reduction. In 2020, about 20 the Southeast, and Da Nang is located in the Northern and percent of households in the poorer Central Highlands region Figure 3.9. Small area poverty estimates (2019), regional zoom-ins Midlands & Northern Mountains Central Highlands Poverty rate 0.0 100.0 Note: The color gradient that transitions between blue and orange depicts the national average poverty rate from the 2019 small area poverty map. Poverty rates are higher in orange districts than the national average, and lower in blue districts. Source: World Bank staff calculations. From the Last Mile to the Next Mile 87 PART 1. A DECADE OF SIGNIFICANT PROGRESS, BUT LAST-MILE CHALLENGES REMAIN received wages from manufacturing, while in other regions, Households in less economically vibrant regions the percentage of households receiving manufacturing earn lower incomes and experience lower income wages ranged from 35 to 47 percent. The importance of growth manufacturing labor is that it provides an easy entry for medium-skilled jobs and pays much higher salaries that can be Across the six broad geographic regions, dynamics of earned in agriculture alone. household incomes and poverty are related. For example, the two poorest regions have the highest level of household While large cities provide strong economic spillovers to participation in family farming (Central Highlands and neighboring areas, rural economies do not have the same Midlands and Northern Mountains, at 70 and 80 percent degree of positive spillover effects. Poverty rates are higher respectively) (Figure 3.10). However, the share of households in smaller urban areas. There are many locations outside of that receive wage income increased by over 10 percentage principal cities that are administratively classified as urban but points in the Midlands and Northern Mountains region remain at lower levels of development. In 2010, 27 percent but declined in the Central Highlands over the course of the of the urban population was in extra-small cities, and these decade. This contrast mirrors the faster poverty reduction areas accounted for 55 percent of the urban poor (Lanjouw success in the Midlands and Northern Mountains. The Central and Marra, 2018). Similarly, there are gaps in services between Highlands is the only region where economic participation in small urban versus large urban areas. Small area poverty maps all forms of labor market activities (wages, family businesses, that illustrate poverty rates at the district levels also reveal and agriculture) decreased at a household level from 2010 to the lack of positive economic spillover from smaller rural 2020. Consistent with having the slowest progress in poverty urban areas (Figure 3.9). Rural economies still have small reduction in terms of changes to the number of poor over widespread regional spillovers, which is illustrated by very low the last decade, the Central Highlands region exhibits the poverty rates in the immediate municipal districts (blue) but lowest average wage income across most types of wage jobs higher poverty in neighboring districts (orange). For example, (services, manufacturing, and different skill levels). Another the immediate areas of Dalat in the Central Highlands are less striking difference between the composition of income in the poor than surrounding areas despite the region’s proximity Central Highlands and other regions is the slower progress of to HCMC, Nha Trang, and the coast. In the case of Sapa in increasing jobs with a formal contract. The Central Highlands the Northern Mountains, it is not its own district, but the also exhibits the lowest share of households (22.2 percent) district encompassing Sapa in Lao Cai is still poorer than the with someone who receives wages with a contract, a sign of national average. lower economic stability and security. On the other hand, the share of households with income from jobs with contracts in The limitations of spillover effects can be partly the Midlands and Northern Mountains is now 32 percent. attributed to market connectivity. There is significant geographical variation in road density. Access to marketplaces, The levels of both wage and family business income unsurprisingly, follows a similar spatial pattern to road grew by the largest amount in the wealthier regions: the density, with a higher concentration—and therefore, greater Red River Delta and Southeast regions (Figure 3.11). accessibility. The Southern half of Vietnam has the greatest Average agricultural incomes in these two regions are higher density of roads, with the Northern Red River Delta region partly because households are engaged in multiple activities, and Da Nang also reporting high levels of road density (all key but income for some sub-sectors are higher in the wealthier urban centers). The rest of the country has a comparatively regions. The Southeast also has the largest share of households low concentration of roads. The Northwestern and Central exclusively receiving wage income as a source of labor market Highlands (mountainous) regions of the country report income. The two wealthiest regions lead in having the highest much less accessibility, but points of good connection have rates of wage growth among all job categories. In terms of lower poverty rates. participation in high-skill wage jobs, the Southeast ties with the Red River Delta; about 20 percent of households in each region receive income from a high-skill wage job. 88 2022 Vietnam Poverty and Equity Assessment Chapter 3. Challenges to reducing poverty among the remaining poor Figure 3.10. Share of households receiving labor market income (2010 and 2020), by source and region 100 Annual income (VND, millions) 80 60 40 20 0 2010 2020 2010 2020 2010 2020 2010 2020 2010 2020 2010 2020 Midlands & Northern Red River Delta Northern & Coastal Central Highlands Southeast Mekong Delta Mountains Central Region Wage, bonus, and severance Family business Agriculture Note: Household-level income. Source: World Bank staff calculations using VHLSS 2010, 2020. Figure 3.11. Average household labor market income (2010 and 2020), by sources and region 250 Annual income (VND, millions) 200 150 100 50 0 2010 2020 2010 2020 2010 2020 2010 2020 2010 2020 2010 2020 Midlands & Northern Red River Delta Northern & Coastal Central Highlands Southeast Mekong Delta Mountains Central Region Wage, bonus, and severance Family business Agriculture Note: Annual nominal household level income. Source: World Bank staff calculations using VHLSS 2010, 2020. Geography will remain a chronic challenge to appropriate for industrial zones, and the feasible placement poverty reduction of roads and connectivity. The natural geography is also very costly to tame or alter. Areas around Hanoi and HCMC By geography, households in communes located in high and the central coast have low poverty rates, illustrating the mountainous areas are dramatically poorer (Pimhidzai large spillover effects from these economic powerhouses. The et al, 2019). While there has been a longstanding association limits of the economic spillover from large cities also follow between geography and poverty, it may still be surprising geographic characteristics; poverty rates generally begin to rise just how significantly higher poverty is in high mountains as the plains begin to merge into mountains. Participation in compared to other geographic areas. The poverty rate in high non-farm economic activity is also directly related to access, mountains is almost 36 percent, compared to 7.6 percent in as youths living in high mountain areas are least likely to the low mountains. Natural geography largely determines participate in off-farm wage employment (Figure 3.12). the location of ports, plains suitable for agriculture, areas From the Last Mile to the Next Mile 89 PART 1. A DECADE OF SIGNIFICANT PROGRESS, BUT LAST-MILE CHALLENGES REMAIN Figure 3.12. Poverty rates by geography Area-based targeting programs have played an 80 important role in regional development, but 70 support needs to be better channeled to the 60 poorest communes33 Poverty rate (%) 50 40 Over the last decade, National Targeting Programs (NTPs) 30 provided high levels of investments to communes. The 20 government invested nearly VND 560 trillion (approximately 10 US$25 billion) in commune-level programs under NTPs from 0 2010 to 2019. Expenditures per commune averaged to about Inland Hills, Coastal Low High delta midlands Mountains mountains 85 billion VND per commune (US$3.5 million) over the whole period, or about US$350,000 per commune per year. NTP LMIC ($3.20/day) poverty line financing was a mix of central government, provincial, and UMIC ($5.50/day) poverty line commune-level funds, including credit and community-level contributions. On aggregate, the combination of state budget Source: World Bank staff calculations using VHLSS 2018. and local government financing made up close to 60 percent of commune-level investment financing. Increasing poverty can be related to rising risks Most poor households, identified using the World Bank’s monetary poverty lines, live in communes that Geography does not fully determine poverty. The poor are also designated as poor under the P135 program. In tend to reside in hilly and mountainous areas, where land 2018, about 65 percent of $3.20/day 2011PPP poor lived in is less productive and more difficult to cultivate. However, P135 communes, which are targeted communes under the there are also poor households located in relatively fertile Sustainable Poverty Reduction National Targeting Program. coastal and inland delta areas. In both the inland delta and Most communes in the high mountains also benefited the coastal areas, the median amount of land cultivated by the from more government programs, with over 90 percent of poor is less than 20 percent of the median amount cultivated communes receiving some form of government support from by the nonpoor. Poor households are also less likely to have a 2015 to 2018. Mountainous communes were also more likely to formal land title to use as financial collateral. With increasing report programs being related to hunger and poverty reduction, climatic changes, households in extremely low-lying flat lands investment in economic development, and infrastructures. are also economically vulnerable. Other types of geographies are prone to natural shocks, and these regions may experience However, due to financing options, the poorest communes temporary increases in poverty during natural disasters but ultimately received less overall investment. Communes are also under threat of more permanent reversals in growth classified by the government as extremely poor in 2010 are if natural environmental conditions continue to worsen. In more reliant on state contribution, which made up close to half 2020, severe droughts coincided with an increase in regional of the investment spending, with local revenues contributing poverty rates in the Mekong Delta, which rose back to near another 29 percent. Extremely poor communes received 2016 levels. The Mekong Delta and other low-lying areas, 14 percent more in state budget investment than nonpoor while not constrained by geography today, will need to take communes, but nonpoor communes spent 80 percent more decisive action now to prevent worsening environmental than extremely poor communes from their own-generated trends from deteriorating living conditions (see Chapter 6). revenues and four times as much from private sector and community participation. Thus, the poorest communes invested less because they mostly relied on central government financing without incremental resources. Overall, extremely poor communes drew fewer resources from private sector participation and community contributions than other types of communes. 90 2022 Vietnam Poverty and Equity Assessment Chapter 3. Challenges to reducing poverty among the remaining poor Lower investments can be linked to less success in that have been identified, including physical and economic achieving NTP-NRD status, a recognition given to connectivity, basic household livelihood assets, market communes that reached targets in 19 monitoring linkages, labor market mobility, access to education, and indicators during NTP-NRD (2011–2015). This access to external support initiatives (World Bank, 2019). is apparent from a comparison of the level and sources of financing of communes that achieved Natural Rural Geographic remoteness and poor linkages are Development (NRD) status and those that did not. associated with higher poverty Communes that had achieved NRD status by 2016 had an average investment of more than VND 60 billion during the Poor linkages to markets hinder ethnic minorities in first phase of the NTPs. This was more than twice the average improving incomes and livelihoods. Physical connectivity, amount spent by communes that had not achieved NRD production of high-value crops, prejudice from the Kinh status. At the end of 2019, communes achieving NRD status majority, and the degree of market power exercised by local had on average spent VND 120 billion since 2010, compared traders, retailers, and shop owners are factors that affect to VND 62 billion average spending in communes not market linkages for ethnic minorities. Physical connectivity achieving NRD status. Failure to achieve NRD status is thus determines ethnic minorities’ access to markets and other mostly a financing gap issue. economic opportunities available in the wider economy. Deprived ethnic minority communities are more likely to live NTP investments improved connectivity across all in areas with poor physical connectivity, which in turn keeps communes, but gaps remain in poor communes. them trapped in poverty. Poor physical connectivity negatively Resource limitations imposed sharper trade-offs in multi- impacts ethnic minorities’ ability to participate in market deprived poor areas. Faced with a trade-off between many value chains, engage in waged employment and access to needs, the poorest communes favored roads and water above other services (World Bank, 2019). The level of connectivity is human development investments. As infrastructure needs not only determined by distance or level of remoteness alone, are being met, investments now need to shift to improving but also factors such as weather and current conditions of quality in upper-secondary school and health facilities. Due infrastructure. Some groups are heavily subjected to problems to broad implementation of NTPs, the welfare impacts are related to seasonality, such as landslides and monsoons, which hard to quantify. Widespread investment, development, and worsen connectivity during certain times of the year. resources were broadly available to most communes. But the poorest communes in some instances received less overall Migration provides a pathway for further support. The analysis using commune-household linked employment opportunities for some ethnic VHLSS is consistent with findings from a detailed study of minorities NTP effectiveness (Pimhidzai and Niu, 2021). In recent years, the data suggest that there has been some migration by ethnic minorities to economic zones, 3.3. Ethnic minority but the majority continues to live in provinces with poverty reduction many ethnic minorities. Between 2009 and 2019, the share of ethnic minorities in the population increased in 14 out of 19 provinces identified as being part the Northern, Factors determining high poverty levels among ethnic Central, or Southern economic zones. Notably, Bac Ninh minorities are manifold and complex. A high level of province, a major industrial center, experienced the highest heterogeneity exists between different ethnic minority groups, growth in ethnic minorities as a share of the population, from and as such the extent to which these factors contributed to 0.3 percent in 2009 to over 5.5 percent in 2019. Similarly, poverty among specific groups also varies. Nonetheless, an the share of ethnic minority population also increased in examination of the drivers of socioeconomic development a number of Northern key economic zone provinces such facing all ethnic minorities as a group in Vietnam is important as Hai Duong, Hung Yen and Hai Phong (Figure 3.13). in understanding the chronic challenges for the remaining Research by CARE International also found an increase in the poor. The following section summarizes the key drivers number of migrant workers among different ethnic groups in From the Last Mile to the Next Mile 91 PART 1. A DECADE OF SIGNIFICANT PROGRESS, BUT LAST-MILE CHALLENGES REMAIN Share of provincial population Figure 3.13. Share of provincial population that is ethnic minority 15 that is ethnic minority (%) 12 9 6 3 0 Bac Ninh Hung Yen Hai Phong Hai Duong Vinh Phuc Hanoi Quang Ninh 2009 2019 Source: 2009 and 2019 Census. five provinces selected for the study, in 2019 compared to the Participation in the labor market outside of agriculture previous year (CARE, 2020).34 Overall, while there may be a is key to poverty reduction for ethnic minorities. Poorer degree of migration to economic zones, the largest share of ethnic minorities are heavily dependent on agriculture, while ethnic minorities still resides in the same provinces in 2019 better-off ethnic minorities earn the largest share of their as in 2009, such as Son La, Ha Giang, Gia Lai, Dak Lak. income outside of agriculture. Language barriers and low Policies to support ethnic minorities should maintain focus educational attainment among some ethnic minority groups on provinces with a large population of ethnic minorities. At limit their access to only unskilled, lower-paying jobs. Closer the same time, as migration continues to play an increasingly proximity to wage employment opportunities is associated important role in raising household income and reducing with better economic outcomes, and it enables more ethnic poverty, as well as providing a pathway to employment for minority women to participate in the labor force. Some ethnic minority youth and women, policies to support ethnic progress has been seen. In 2020, a strong shift was observed in minority migrant workers are also needed. the share of ethnic minorities working in the manufacturing sector, and impressively across all regions (Figure 3.14). In 2020, the share of ethnic minorities working in manufacturing was similar to the rate of the Kinh Income differentials between Kinh and ethnic majority working in manufacturing at the start of minorities remain the decade Across the broad categories of household income, Kinh Economic connectivity such as proximity to economic households receive more income across all sources except clusters or industrial parks is an important determinant for agriculture (Figure 3.15). The Kinh even receive higher of poverty amongst ethnic minorities. As public average social protection incomes, which includes benefits investment has improved road connectivity for ethnic based on categorical and non-poverty-related characteristics. minorities over the years, economic connectivity has become Among the same classification of wage jobs, wages and wage increasingly important for poverty reduction. Proximity to growth are higher among the Kinh than ethnic minorities industrial areas enables more ethnic minority women and (Figure A.3.1.1 in Annex). Even among high-skilled and youth to participate in the labor market. For some ethnic contract jobs, household incomes are lower on average for minority groups, distance to location of work can affect ethnic minority families than Kinh families. Of course, there perception on whether women should take on paid work are other factors at play, such as location. Demombynes and outside of the home (World Bank, 2019). Closer distances Testaverde (2018) find that ethnic minorities working in wage also enable women to commute to work given their limited jobs still earn about 6 percent less than the Kinh in comparable use of transportation. jobs (between 2011 and 2014). 92 2022 Vietnam Poverty and Equity Assessment Chapter 3. Challenges to reducing poverty among the remaining poor Figure 3.14. Share of workers employed by ethnicity and in the manufacturing sector, 2010 to 2020 Ethnic minority Kinh 50 50 40 40 Share of workers (%) Share of workers (%) 30 30 20 20 10 10 0 0 2010 2012 2014 2016 2018 2020 2010 2012 2014 2016 2018 2020 Red River Delta Midlands & Northern Mountain Central Highlands Northern & Coastal Central Region Southeast Mekong Delta Source: World Bank staff calculations using VHLSS 2010–2020. Figure 3.15. Kinh-ethnic minority absolute income gaps across broad categories 80 70 60 Annual VND (millions) 50 40 30 20 10 0 −10 −20 Agriculture Wage, bonus, Family business Pension Remittances Social Protection severance (cash transfers) 2010 2020 Source: World Bank staff calculations using VHLSS 2010 and 2020. Social assistance for ethnic minorities has transfers in cash and kind, as well as other social care services. improved over time, but under-coverage Examples of large regular social assistance programs include continues to be a challenge social transfers to vulnerable groups, subsidies in the form of cash transfers for electricity, and education-related cash Coverage of ethnic minorities in Vietnam’s social transfers. Ethnic minorities also benefit from specific poverty assistance system has improved over the years, but reduction programs such as the National Target Program challenges remain. Social assistance spending in Vietnam has for Sustainable Poverty Reduction. In 2016, 52 percent of been increasing annually, both in terms of nominal spending ethnic minority households were covered by at least one social and as a share of GDP (Dutta, 2019). The current social assistance program. assistance system consists of regular and irregular/emergency From the Last Mile to the Next Mile 93 PART 1. A DECADE OF SIGNIFICANT PROGRESS, BUT LAST-MILE CHALLENGES REMAIN While ethnic minorities are increasingly covered by compared to 16.3 percent for women nationwide (World social assistance programs, challenges remain with Bank, 2021c). Similarly, nearly a quarter of ethnic minority the complexity, fragmentation, and funding of these women are illiterate, compared with just 13.3 percent of programs as well as coverage. Multiple programs exist ethnic minority men (CARE, 2020). Women with lower without a comprehensive overview of how they can educational attainment and poor command of the national contribute to improving the well-being of ethnic minorities. language do not participate in community activities and Furthermore, considerations of the different challenges consequently are not part of household decision making. The faced by different ethnic minority groups have not been burden of unpaid care work also falls heavily on women (see incorporated in the design of these programs, making them Box 3.1). Child marriage, early childbirth and high fertility less effective in addressing the most pressing issues. Coverage rates are also contributing factors to high poverty among issues remain due to the arbitrary exclusion of some poor ethnic minority groups. Having a higher number of children ethnic minority households from the official poor list, due below working age within the household is correlated with to changes in local-level poverty reduction targets. In the economic hardship among some ethnic groups where birth categorically targeted social transfer, groups such as single control is not used or where a son preference exists (World parents with small children, people with disabilities, and Bank, 2019). children without support are excluded. Unfair competition from local traders and retailers can Finally, a wide range of other cultural and social also adversely impact market linkages for poor ethnic factors impact the pace of poverty reduction minorities. Ethnic minorities face prejudice from the Kinh majority, who control lowland markets (World Bank, 2019). Ethnic minorities are not represented in key leadership In addition, where local traders and retailers also act as money positions. While ethnic minorities are well represented in lenders, they are able to set output prices for agricultural the commune administration system, leadership positions output by ethnic minorities. Indebtedness of ethnic minorities in these organizations are often held by Kinh cadres. There to these lenders in turn reduces their income through are limited opportunities for ethnic minorities to break into lower, pre-determined output prices, perpetuating further these positions, given government policies of moving Kinh indebtedness. This can result in land and asset transfer from cadres to remote areas, as well low educational attainment ethnic minorities to repay high-interest loans. Furthermore, and language barriers. Lack of representation, even in non-key high costs and seasonal availability of transportation increase positions for smaller ethnic minority groups, reduces access to reliance on traders for the poorest. resources and opportunities for ethnic minorities. External and self-perception of ethnic minorities can 3.4. Agriculture incomes hinder their economic development. Prejudice exists between the ethnic majority Kinh and minority populations, and livelihoods which reduces ethnic minorities’ access to opportunities and The analysis focuses on household-level agricultural public services. This prejudice also exists based on socio- activities among smallholder farmers. Agricultural economic status (e.g., between better off and more deprived households are engaged in five activities, including the groups). Many ethnic minority groups internalize stereotypes cultivation of crops, livestock rearing, forestry, fisheries, about their inferiority and timidity. These prejudices and and agricultural services (see Box 3.2 for a description of negative self-perceptions can widen the social distance these activities). In the analysis that follows, a household’s between groups and make integration more difficult. main activity or main crop is defined as having the largest income source from a particular agricultural activity. The Ethnic minority women are disadvantaged because they section is divided into three parts. The first part focusses on face greater social and economic inequalities. Across profiling poverty, poverty reduction progress, and agricultural educational indicators, ethnic minority women are lagging outcomes among farm households, respectively. The final part behind ethnic minority men and women nationwide. The rate takes a closer look to unpack some sources of disparities in of graduation for ethnic minority women was 10.2 percent farm outcomes including productivity. 94 2022 Vietnam Poverty and Equity Assessment Chapter 3. Challenges to reducing poverty among the remaining poor BOX 3.1. Gender, ethnicity and the job market in Vietnam Ethnic minority women are disadvantaged because they face greater social and economic inequalities. Across educational indicators, ethnic minority women are lagging behind ethnic minority men and women nationwide. The rate of graduation for ethnic minority women was 10.2 percent compared to 16.3 percent for women nationwide (World Bank, 2021c). Similarly, nearly a quarter of ethnic minority women are illiterate, compared with just 13.3 percent of ethnic minority men (CARE, 2020). Women with lower educational attainment and poor command of the national language do not participate in community activities and consequently are not part of household decision making. Women are expected to carry out traditional housewife duties in addition to agricultural labor, leaving them unable to form expansive social networks. These factors limit ethnic minority women’s access to resources and livelihood opportunities. Child marriage, early childbirth, and high fertility rates are also contributing factors to high poverty amongst ethnic minority groups. Young girls and women once married are often asked to stay home to take care of children, causing many to drop out of school early. World Bank analysis found that early marriage and pregnancy are also associated with stunting and intimate partner violence (World Bank, 2018). Women from ethnic minority groups are disadvantaged in the job market, as they face both gender and ethnicity-related barriers. Ethnic minorities are less likely to have a wage-paying job than Kinh or Hoa people. Only 17 percent of non-Kinh or Hoa working women have a wage-paying job, compared to 22 percent of Kinh or Hoa women and 32 percent of men who belong to an ethnic minority group. In addition, women from ethnic minority groups that receive a wage make 17 percent less than ethnic minority males, 35 percent less than Kinh or Hoa women, and 50 percent less than non-ethnic minority men (World Bank, 2019). What is more, female ethnic minorities are disproportionately concentrated in low-wage, unstable agricultural self-employment jobs. In agriculture, for example, 59 percent of women from ethnic minority ethnic minority groups are self-employed, compared to 40 percent of women from ethnic majority groups (World Bank, 2019). The probability of being poor partly depends on Households most engaged in crop cultivation and that the type of agricultural activity a household is rely on staples have the highest rates of poverty. Poverty engaged in rates are generally inversely related to the value of outputs. For example, fruit products and industrial crops are more Poverty rates are high among households who receive expensive, and households that primarily depend on sales of the majority of their agriculture income from forestry. these crops are less likely to be poor. Those cultivating staples Households most engaged in forestry experienced a dramatic are the poorest partly because they are engaged in subsistence decline in poverty rates in 2016, but then saw little change farming. More households in the poorest decile consume afterwards. Households engaged in forestry have more home-produced foods compared to those in the highest land but less income. Although forestry households are the decile (91% vs. 47%). Families producing food crops are also largest landholders, production is often small-scale, and the more likely to consume a large portion of their crops at home type of trees planted may not be the popularly exported compared to people who cultivate fruit crops (i.e., the latter ones (such as wood and timber). Forestry output takes time group would more likely sell their output to domestic and/or to yield returns, and access to markets and services is often foreign consumers). Hence, if crop cultivators can efficiently poor. Moreover, these households may not own appropriate produce more and sell their output at more competitive forestry equipment for processing large-scale harvests to cover prices, the detrimental effects of outright subsistence farming domestic and foreign demand. They are for the most part can partially be overcome. ethnic minorities, with no formal education and living in the Midlands. From the Last Mile to the Next Mile 95 PART 1. A DECADE OF SIGNIFICANT PROGRESS, BUT LAST-MILE CHALLENGES REMAIN BOX 3.2. Categories of household agricultural activities in the VHLSS One important aspect to better understanding high poverty and low income is to describe the agriculture sector as households experience it. Analysis uses data from the Vietnam Household Living Standards Survey (VHLSS) to analyze households that depend on agriculture for their livelihood. The VHLSS captures agricultural activities primarily through household-level agricultural income modules. A small share of household members are engaged in the agriculture sector in the form of wage employment. A small share of family businesses are engaged in food processing, but that is considered to be manufacturing. Table B.3.2.1. Description of agricultural activities ACTIVITY DESCRIPTION IS THERE PER M2? IS THERE UNIT VALUES (Y/N) Income sources (net) Agricultural land Rental income Livestock, hunting, Animals, birds, apiculture and associated products No Yes, some in kgs some in heads trapping e.g., milk, eggs, honey, etc. Agricultural services Ploughing, irrigation, etc. No No Forestry Palm, bamboo, timber, cinnamon, firewood, Has land size No tending, forest renovation and improvement, etc. Fisheries Aquaculture and capture Has pond size Yes Crops Rice, staples, industrial, fruits, and by-products Has land size Yes Rice Ordinary, sticky, specialty Has land size Yes Staples (food crops) Corn, potato, beans, vegetables, etc. Has land size Yes Industrial crops Annual and perennials e.g., soybean, cotton, Has land size Yes coconut, rubber, coffee, etc. Fruits Fruit trees e.g., orange, pineapple, jackfruit etc. Has land size Yes Agricultural by- Leaves, stems, jute, stalks, straw, grass, etc. No No products Among agricultural households, poverty rates are lowest rising poverty also seen in the Mekong Delta in the same year, among those who receive the majority of their incomes since that is the region with the highest level of engagement from land leasing, fruit, or aquaculture. Land leasing is a in aquaculture. While not shown because of small sample passive activity, and while it yields low incomes, it is most likely sizes, households engaged in agricultural services also have the supplemental income for households engaged also in services lowest poverty rates and correspondingly high incomes. or manufacturing. Prices of fruit and aquaculture products are higher, which keeps poverty low. The improvement in For 84 percent of poor households in agriculture, the the livelihoods of aquaculture households over the decade main agricultural income sources are rice, staples, has been the most impressive. The relative decline of poverty livestock, and forestry (Figure 3.16). For the most part, for aquaculture households from 2010 to 2020 was the households who engage in these tend to be the poorest. An fastest, and comparable to the pace among non-agriculture exception is that poverty rates for households principally households. For households in aquaculture, poverty rates engaged in industrial crops are slightly higher than for rice- declined from 24 to 4.4 percent over the decade, or an average producing households in recent years, but the number of poor of 15.6 percent per year. However, a slight increase in poverty is lower because fewer households cultivate industrial crops. among aquaculture households in 2020 is related to the 96 2022 Vietnam Poverty and Equity Assessment Chapter 3. Challenges to reducing poverty among the remaining poor Figure 3.16. The number of poor (in millions), by main household agricultural activity, 2020 1.5 Number of Poor (millions) 1.2 0.9 0.6 0.3 0.0 Not engaged Land Rice Staples Industrial Fruit By-products Livestock Agricultural Forestry Aquaculture in agriculture crops services Note: World Bank LMIC ($3.20/day 2011PPP) poverty line. Main agricultural activity is the one that yields the highest income. Poverty rates for households with the largest income source from by-products or agricultural services is not shown because of small sample. Households may also earn incomes outside of agriculture. Source: World Bank staff calculations using VHLSS 2020. Characteristics of households in the forestry sector Employment in agriculture is in decline. In 2020, half of stand out as more unique (Table A.3.1.1. in Annex). all Kinh farmers were aged 50 and over, illustrating the main Compared to households mainly engaged in other activities, source of ageing, while ethnic minority farmers tend to be there is a high concentration of ethnic minorities in forestry much younger (Figure 3.17). Persistence in agriculture among (73 percent), who have the highest share of uneducated (18 ethnic minorities also foreshadows longer-term challenges to percent) and the youngest (47 years) household heads, and poverty reduction. Across all regions and all sub-sectors, fewer who manage the largest land areas, averaging 4.3 hectares. households were engaged in agriculture in 2020 than in 2010 These households mainly reside in hilly and remote regions (Figure 3.18). The exit from agriculture is a combination of of Vietnam, with over three-quarters having at least one factors including ageing, growing economic opportunities dependent child compared to others. Moreover, the highly elsewhere, and changing employment choices among educated, male-headed, Kinh group is the most likely to be younger generations. Across sub-sectors, the largest decline in engaged in profitable agricultural services, hence, to fetch participation was among some of the agricultural activities that the highest average household income among this group of generate the lowest income: by-products and rice production. households. The share of all households earning income from agricultural Figure 3.17. Farmers are aging, particularly among Kinh Ethnic minority Kinh 100 100 Share of agriculture workers (%) Share of agriculture workers (%) 80 80 60 60 40 40 20 20 0 0 2010 2012 2014 2016 2018 2020 2010 2012 2014 2016 2018 2020 0–9 10–19 20–29 30–39 40–49 50–59 60–69 70+ Source: World Bank staff calculations using VHLSS 2010–2020. From the Last Mile to the Next Mile 97 PART 1. A DECADE OF SIGNIFICANT PROGRESS, BUT LAST-MILE CHALLENGES REMAIN by-products was 41 percent in 2010, dropping to 21 percent crops (perennials). People living in the Southeast region, with in 2020. Participation in rice farming dropped from 44 to its small land size, and half of its population from Ho Chi 28 percent over the same period. However, households are Minh City, are the least likely to be engaged in agricultural diversified and often earn income from family farming in activities. Hence, the extent to which regions are engaged in addition to other labor activities. Households can also earn less profitable agricultural activities is likely to explain the income from more than one type of agricultural activity. variations in regional poverty rates among these households. Across regions, households are engaged in a variety of Agricultural incomes across sub-sectors remain types of agricultural activities. In 6 out of 10 sub-sectors, modest compared to other sources of non- the Midlands and Northern Mountains region has the highest agricultural income participation: by-products, forestry, fruit, livestock, rice, and staples (Figure 3.19). Mekong Delta households are the most Incomes from some agricultural activities are increasing, likely to be seafood producers (39 percent), while a large share but even these higher incomes are low compared to income of households in the Central Highlands cultivate industrial from wages or family businesses. Figure 3.20 disaggregates Figure 3.18. Participation in agricultural activities over time (2010–2020), national level Share of households with income source (%) Agricultural Agricultural Industrial Aquaculture By-products Forestry Fruit crops Livestock Rice Staples Total land services 60 Annual growth (%) 40 20 0 2012 2016 2020 2012 2016 2020 2012 2016 2020 2012 2016 2020 2012 2016 2020 2012 2016 2020 2012 2016 2020 2012 2016 2020 2012 2016 2020 2012 2016 2020 2012 2016 2020 Note: Agricultural land data missing in 2014. Source: World Bank staff calculations using VHLSS 2010–2020. Figure 3.19. Participation in agricultural activities over time (2010–2020), by region Agricultural Agricultural Any Industrial agricultural Aquaculture By-products Forestry Fruit Livestock Rice Staples land services crops activity Share of households (%) 80 60 40 20 0 2010 2012 2014 2016 2018 2020 2010 2012 2014 2016 2018 2020 2010 2012 2014 2016 2018 2020 2010 2012 2014 2016 2018 2020 2010 2012 2014 2016 2018 2020 2010 2012 2014 2016 2018 2020 2010 2012 2014 2016 2018 2020 2010 2012 2014 2016 2018 2020 2010 2012 2014 2016 2018 2020 2010 2012 2014 2016 2018 2020 2010 2012 2014 2016 2018 2020 Central Highlands Midlands & Northern Mountain Red River Delta Mekong Delta Northern & Coastal Central Region Southeast Source: World Bank staff calculations using VHLSS 2010–2020. 98 2022 Vietnam Poverty and Equity Assessment Chapter 3. Challenges to reducing poverty among the remaining poor farm income trends by activities and crops and shows that Changes in agricultural incomes also vary by region. For while crop income increased to a lesser extent, income from example, despite lower participation rates in some household agricultural services, aquaculture, and livestock substantially activities such as aquaculture, forestry, and livestock, the increased throughout the decade. Higher livestock income also Red River Delta and the Southeast regions still enjoy higher reflects increasing diversity in diets and higher consumption of growth in average incomes from these sub-sectors, perhaps milk products by the domestic population. However, even for benefitting from higher local prices (Figure 3.21). Other households engaged in sub-sectors that are enjoying increasing regional advantages include higher incomes from rice in the incomes, total incomes from even the most lucrative activities Mekong Delta and higher incomes from perennials in the are still well below wage incomes. Average annual household Central Highlands and Southeast regions. income from wages totaled about 138 million VND in 2020 compared to the highest-paying agriculture sub-sector: 51.5 million VND in agricultural services. Figure 3.20. Household farm income by activity (2010–2020), national level Annual VND (thousands) Agricultural Agricultural Industrial Aquaculture By-products Forestry Fruit Livestock Rice Staples Total land services crops Annual VND (thousands) 40,000 20,000 0 2012 2016 2020 2012 2016 2020 2012 2016 2020 2012 2016 2020 2012 2016 2020 2012 2016 2020 2012 2016 2020 2012 2016 2020 2012 2016 2020 2012 2016 2020 2012 2016 2020 Note: Nominal average annual net income among engaged households. Source: World Bank staff calculations using VHLSS 2010–2020. Figure 3.21. Household farm income by activity (2010–2020), by region Agricultural Agricultural Industrial Any Aquaculture By-products Forestry Fruit Livestock Rice Staples Agricultural land services crops activity 150K Annual VND (’000) 100K 50K 0K 2010 2012 2014 2016 2018 2020 2010 2012 2014 2016 2018 2020 2010 2012 2014 2016 2018 2020 2010 2012 2014 2016 2018 2020 2010 2012 2014 2016 2018 2020 2010 2012 2014 2016 2018 2020 2010 2012 2014 2016 2018 2020 2010 2012 2014 2016 2018 2020 2010 2012 2014 2016 2018 2020 2010 2012 2014 2016 2018 2020 2010 2012 2014 2016 2018 2020 Central Highlands Midlands & Northern Mountain Red River Delta Mekong Delta Northern & Coastal Central Region Southeast Source: World Bank staff calculations using VHLSS 2010–2020. From the Last Mile to the Next Mile 99 PART 1. A DECADE OF SIGNIFICANT PROGRESS, BUT LAST-MILE CHALLENGES REMAIN Sources of disparities between Kinh and ethnic Table 3.1. Agricultural activities - minorities Kinh vs. ethnic minorities Despite remarkable headway in overall poverty KINH EM DIFF. reduction, including a steady fall in ethnic minority Panel A: Activity poverty, there remains a considerable gap in agricultural Crops 0.90 0.97 -0.07*** outcomes between the Kinh majority and ethnic minorities. Overall, ethnic minorities are more likely to Livestock 0.64 0.82 -0.19*** engage in agriculture compared to the majority Kinh (88% vs. Ag. services 0.02 0.01 0.01*** 56%), as well as across all sub-sectors (Figure 3.22). Table 3.1 Forestry 0.11 0.69 -0.60*** panel A shows that the most noticeable difference between Fisheries 0.21 0.28 -0.05*** the Kinh and ethnic minorities is in engagement in livestock Observations 4488 1482 (82% vs. 64%) and forestry (69% vs. 11%). Ethnic minorities are marginally more likely to be engaged in crops and fishing Panel B: Crop by a 7 percentage point difference each. In panel B of Table Rice 0.62 0.80 -0.21*** 3.1, the absolute difference in engaging across crops between Staples 0.56 0.82 -0.30*** the two groups is largest in rice and staples compared to Industrial crops 0.34 0.37 -0.01 industrial crops and fruit trees. This implies that despite the vast presence of ethnic minorities in agriculture, they are more Fruits 0.48 0.58 -0.09*** likely to engage in subsistence agriculture and in profoundly By-products 0.53 0.68 -0.17*** less productive activities compared to the Kinh majority. Observations 4003 1430 Note: ‘Diff.’ is the difference in means between the two groups with significance Assessing crop yield and total factor productivity (TFP) * p < 0.10, ** p < 0.05, *** p < 0.01 based on t-tests. shows that poor and ethnic minority households are less Source: World Bank staff calculations using VHLSS 2018. productive (Table 3.2). In panel A, the difference in farm- level yield and TFP between poor and nonpoor households is statistically significant. Although there is no systematic difference in crop yields between ethnic minorities and the Kinh, the significantly lower TFP among ethnic minorities in panel B is potentially concerning. Figure 3.22. A comparison of household agricultural participation - Kinh vs. ethnic minorities 80 70 Share of households (%) 60 50 40 30 20 10 0 Ethnic Kinh Ethnic Kinh Ethnic Kinh Ethnic Kinh Ethnic Kinh Ethnic Kinh Ethnic Kinh Ethnic Kinh Ethnic Kinh minority minority minority minority minority minority minority minority minority Agricultural Aquaculture By-products Forestry Fruit Industrial Livestock Rice Staples services crops 2010 2020 Source: World Bank staff calculations using VHLSS 2010–2020. 100 2022 Vietnam Poverty and Equity Assessment Chapter 3. Challenges to reducing poverty among the remaining poor Table 3.2. Crop yield and total factor productivity the average compared to their ethnic minority counterparts. This points toward information asymmetries in prices or (1) (2) YIELD (LOG) TFP access to markets. Panel B shows that although productivity is generally low in rice and staples regardless of ethnicity (rice is Panel A: Poverty as equally low as ₫2,000 per kg for both groups), productivity Poor (=1) -0.349*** -0.195*** in fruits and cash crops per square meter is profoundly higher (0.050) (0.028) for the ethnic majority. R-sq 0.357 0.391 Observations 5602 5314 While ethnic minorities are more engaged than Kinh in Panel B: Ethnicity every agricultural activity, they earn less in every activity and have experienced less agricultural income growth. EM (=1) 0.037 -0.121*** Figure 3.24 shows that Kinh majority generally earn higher (0.071) (0.042) than the overall average, and farm income gaps between the R-sq 0.350 0.358 Kinh and ethnic minorities are significantly wide. In line with Observations 5602 5314 these findings, the most profitable activities are agricultural Note: TFP is expressed in hours of labor and is estimated based on yield, input services and fisheries, while forestry is the least profitable. expenditures, land size, and capital. District FE included. SE clustered at the Although it appears that households engaged in cultivating enumeration area level. *** p<0.01 ** p<0.05 * p<0.10. crops have a sizeable income, this is largely driven by fruit trees Source: World Bank staff calculations using VHLSS 2014. and industrial crops. Therefore, being productive is harder in rice and staples regardless of ethnic affiliation. Considering productivity and unit values across crops further explains welfare dynamics between Kinh and Disparities persist between the Kinh majority and ethnic ethnic minorities (Figure 3.23). Overall, average price minorities, and chronic challenges largely remain among per unit is lowest in rice (₫7,000) and highest in cash crops ethnic minorities. Ethnic minorities are more likely to be (₫22,000) – panel A. However, the major source of variation engaged in less profitable activities and to use less fertilizer; they between the two groups is in the value per kilogram for are less likely to monocrop, they have lower TFP, and they earn staples and fruits. For example, Kinh staples are worth more less farm and total household income. Income-maximizing than double those of ethnic minorities (₫24,000 vs ₫10,000). households might concentrate in the production of the most Therefore, the Kinh charge significantly higher prices than profitable crops. While some potential sources of disparities Figure 3.23. Crop productivity - Kinh vs. ethnic minorities A. Crop-Unit Value B. Crop-per square meter 25 350 300 20 250 15 1000/unit 200 1000/m2 10 150 100 5 50 0 0 Rice Staples Industrial Fruits Rice Staples Industrial Fruits Total Kinh Ethnic minority Total Kinh Ethnic minority Source: World Bank staff calculations using VHLSS 2018. From the Last Mile to the Next Mile 101 PART 1. A DECADE OF SIGNIFICANT PROGRESS, BUT LAST-MILE CHALLENGES REMAIN between Kinh and ethnic minorities have been highlighted, Putting agricultural land to better use to increase the additional explanations driving these outcomes may be incomes of those left behind in agriculture is imperative. linked to differences in knowledge/skills, quality of crops, Across lowland and highland areas, poor and nonpoor land productivity (upland vs. lowland), and risk strategy. households cultivate similar amounts of land. However, poor Availability of geo-coded data and qualitative work could households are less likely to grow industrial crops; they devote further explain these mechanisms. A suggestive implication more of their land to cultivating less profitable traditional for family farms in Vietnam is to enlighten and encourage cereal crops. Not only do poor households plant suboptimal households, especially those that are multicropping, crops; their yields are lower than those of nonpoor households, to specialize in the production of fewer crops to yield which cultivate the same crop type on the same type of land improved outcomes, boost household welfare, and ensure (Pimhidzai et al., 2019). This could be a result of lower sustainable livelihoods. This underscores the importance educational attainment, which results in lower professional of skill development in farming techniques and production and management skills. This is one of the underlying reasons decision making. for their poverty. Furthermore, poor households have limited access to credit to invest in perennial crops, as these crops The ability to engage in the agriculture value chain or often need expensive intermediate inputs and take longer to diversify into cash crops is limited for ethnic minority yield returns. The lack of access to credit generally stems from groups. The poorest ethnic minority groups have very little the lack of collateralized fixed assets and land user certificates. capacity to diversify into cash crops without the availability Hence, the remaining challenges regarding the lack of of indigenous crops (World Bank, 2019). This could be education and lack of financial capital should be addressed attributable to environmental, location-specific factors, to further exploit the agricultural potential from non-crop the lack of productive landholdings and capital, poor labor cultivation in reducing rural poverty. capacity, or potential risk aversion. Similarly, participation in agriculture value chains is very low for most ethnic minorities, despite the proximity of some of them to value chains for products such as coffee, rubber, aquaculture, and others. Figure 3.24. A comparison of household agricultural income - Kinh vs. ethnic minorities 70 Annual income (VND, millions) 60 50 40 30 20 10 0 Ethnic Kinh Ethnic Kinh Ethnic Kinh Ethnic Kinh Ethnic Kinh Ethnic Kinh Ethnic Kinh Ethnic Kinh Ethnic Kinh minority minority minority minority minority minority minority minority minority Agricultural Aquaculture By-products Forestry Fruit Industrial Livestock Rice Staples services crops 2010 2020 Source: World Bank staff calculations using VHLSS. 102 2022 Vietnam Poverty and Equity Assessment Chapter 3. Challenges to reducing poverty among the remaining poor 3.5. Chronic Poverty: Yet, the population of the economically vulnerable is more yesterday and tomorrow than double the size of the population of poor. Moreover, this group has a different profile, which necessitates different Chapter 3 examined the characteristics of regions and policies to sustain their well-being at higher levels. Due to the ethnicities to distill the main observable differences that distribution of the population, the economically vulnerable are associated with higher rates of poverty. Regional and are more prevalent in regions outside of the traditionally poor ethnic differences in the progress of poverty reduction have mountainous areas, with the difference in the share of the persisted. Some changes were seen in 2020. For example, the economically vulnerable compared to the poor greatest in the Mekong Delta experienced a severe drought as well as supply Mekong Delta. The share of the economically vulnerable with disruptions from COVID-19, which upset its decade-long higher levels of education is also higher than it is for the poor. trend of poverty reduction. The Midlands and Northern Addressing the needs of this group, those out of poverty but Mountains region has proven more successful than the Central still vulnerable, and creating pathways where there is a fair Highlands in reducing poverty from a high starting point at chance for all Vietnmaese to improve their livelihoods is the the beginning of the decade. This may be linked to a dynamic Next Mile challege described in Part 2 of this report. labor market process, where the region has experienced a high increase in manufacturing wage employment, while the Central Highlands has a more static labor market. For the poorest households, agriculture is frequently the largest source of household income. To conclude, the chapter reviewed the types of agricultural activities households are engaged in and their association with poverty reduction. An important contribution of this chapter is to describe and better understand trends in household farm income and productivity. Income from agricultural activity is the weakest source of labor market income, with much lower growth rates over the last decade compared to wage and household business incomes. From the Last Mile to the Next Mile 103 PART 1. A DECADE OF SIGNIFICANT PROGRESS, BUT LAST-MILE CHALLENGES REMAIN 3.6. References World Bank. 2012. Well Begun, Not Yet Done: Vietnam’s Remarkable Progress on Poverty Reduction and the Emerging Challenges, 2012 CARE International. 2020. Promoting Gender Mainstreaming in the Vietnam Poverty Assessment. Washington, DC: World Bank. National Target Programme on Socioeconomic Development in ---------------. 2017. Report No: 114017-VN. Washington, DC: Ethnic Minority and Mountainous Areas in 2021-20330. CARE: World Bank. Vietnam. https://www.care.org.vn/wp-content/uploads/2020/06/ ---------------. 2018. Vietnam Future of Jobs. Washington, DC: Gender-mainstreaming-in-National-Target-Programme- World Bank. June-2020-1.pdf ---------------. 2019. Drivers of Socioeconomic Development Among Chi, Pham Linh, Phung Duc Tung, Nguyen Quynh Chi, Dang Kim Ethnic Minority Groups in Vietnam. Washington, DC: World Bank. Chi, Bui Dang Minh, and Ta Tinh Khanh Van. 2018. 54 Ethnic https://openknowledge.worldbank.org/handle/10986/32307 Minority Groups: Why the Differences? License: CC BY 3.0 IGO. Demombynes, Gabriel and Mauro Testaverde. 2018. Employment ---------------. 2020. World Bank National Accounts Data, and Structure and Returns to Skill in Vietnam: Estimates Using the OECD National Accounts Data Files. https://data.worldbank. Labor Force Survey (March 12, 2018). World Bank Policy Research org/indicator/NV.AGR.TOTL.ZS?locations=VN [Retrieved Working Paper No. 8364. Washington, DC: World Bank. 27/10/2021]. Dutta, Puja. 2019. Access to Social Assistance among Ethnic Minorities ---------------. 2021a. A Year Deferred – Early Experiences and Lessons in Vietnam. Note. from COVID-19 in Vietnam. Washington, DC: World Bank. Lanjouw, P. and M.R. Marra. 2018. Urban poverty across the Spectrum ---------------. 2021b. A Trade Based Analysis of the Economic Impact of Vietnam’s Towns and Cities. World Development 110: 295–306. of Non-Compliance with IUU Fishing: The Case of Vietnam. Office of the Government of Vietnam. 2021. PM Decision 1719 on Washington, DC: World Bank. NTP-SEDEMA, National Assembly Resolution 24/2021/QH15 ---------------. 2021c. Reducing Poverty Among Ethnic Minorities in on NTP-SPR, National Assembly Resolution 25/2021/QH15 Vietnam: Priority Action Areas and Lessons. Washington, DC: on NTP-NRD. World Bank. Pimhidzai, Obert. 2018. Climbing the Ladder: Poverty Reduction and Shared Prosperity in Vietnam (English). Washington, DC: World Bank Group. Pimhidzai, Obert, Chiyu Niu, Diji Chandrasekharan Behr, Thang Cong Tran, Hardwick Tchale, Hardwick; Peter Kjaer Milne, Salman Zaidi, 3.7. Notes and Lan Vu Thi. 2019. Vietnam Poverty and Shared Prosperity 32  How education levels are tied to poverty will be discussed more in Update: Better Opportunities for All (English). Washington, DC: Chapter 4. World Bank Group. 33  This section summarizes findings from the Shared Gains report Pimhidzai, Obert and Chiyu Niu. 2021. Vietnam Poverty and Shared (Pimhidzai and Niu, 2021). See Annex 3.2 for a summary of objectives Prosperity Update: March 2021. – Shared Gains: How High and targets for the various NTPs and over different periods. Growth and Anti-Poverty Programs Reduced Poverty in Vietnam. 34  Washington, DC: World Bank Group. Internal. Research was conducted in five provinces where many ethnic Tùng, P.Đ., N. V. Cường, N. C. Thịnh, N. T. Nhung, Nand T. T. K. minorities reside. These include Dien Bien, Bac Kan, Quang Tri, Kon Vân. 2016. Ethnic Minorities and Sustainable Development Goals: Tum and Tra Vinh. From these 10 communes with high rate of migrant Who will be left behind? Results from Analyses of the Survey on workers were selected. The 10 communes include Muong Phang and the Socioeconomic Situation of 53 Ethnic Minorities in 2015. Ngoi Cay (Dien Bien province), Thuan Mang and Na Phac (Bac Kan Hanoi: Committee for Ethnic Minority Affairs (CEMA), UNDP province), Dak Xu and Bo Y (Kon Tum province), Dak Rong and Ba and IrishAid. Nang (Quang Tri province), and Thanh Son and Tra Cu (Tra Vinh province). 104 2022 Vietnam Poverty and Equity Assessment Chapter 3. Challenges to reducing poverty among the remaining poor ANNEX 3.1. Chapter 3 figures and tables Figure A.3.1.1. Wage trends by job type, Kinh vs. ethnic minorities 50 Share of households (%) 40 30 20 10 0 Ethnic Kinh Ethnic Kinh Ethnic Kinh Ethnic Kinh Ethnic Kinh Ethnic Kinh Ethnic Kinh minority minority minority minority minority minority minority Formal job Agriculture Manufacturing Services Low-skill Medium-skill High-skill 2010 2020 Source: World Bank staff calculations using VHLSS 2010 and 2020. Table A.3.1.1. Key household characteristics across main activity 2018 CROPS LIVESTOCK AG. SERVICES FORESTRY FISHERY Ethnic minority (=1) 0.35 0.25 0.14 0.73 0.07 Female headship (=1) 0.16 0.17 0.00 0.10 0.17 No education (=1) 0.09 0.11 0.00 0.18 0.07 Age in years 51.04 52.79 49.03 46.49 52.24 Household size (#) 3.90 3.48 4.08 4.47 3.97 Household w/ child (=1) 0.63 0.55 0.67 0.75 0.67 Net household income (log) 11.27 11.30 12.12 11.09 11.88 Land size (ha) 1.82 1.11 1.07 4.30 2.47 Observations 1155 250 22 118 225 Source: World Bank staff calculations using VHLSS 2018. From the Last Mile to the Next Mile 105 ANNEX 3.2. PART 1. NTP Summary Table A.3.1.2. A Summary of National Targeting Programs SOCIOECONOMIC DEVELOPMENT PLAN NATIONAL TARGETING PROGRAMS (NTPS) TARGETS AND COVERAGE (SEDP) PHASES 2011-2015 16 NTPs focused on sectors such as health, education, water, The objective of NTP-NRD was to develop rural communes in all 63 provinces, transport, agriculture, and rural development. The NTP for New involving 11 activity groups that were linked to 19 social and economic monitoring 106 2022 Vietnam Poverty and Equity Assessment Rural Development (NTP-NRD) was the largest, investing over indicators. Achievement of all 19 NTP criteria qualified communes, districts, and US$9 billion through the state budget, increasing to US$37 billion provinces for recognition as having attained “National Rural Development status,” when other financing sources such as bank credits, private sector a largely symbolic recognition. However, this approach that has led to some biases in investments and commune contributions are included. Meanwhile, funding to prioritize those districts and communes with the best chance of achieving Program 135 (P135) continued to finance investments in extremely “NRD status.” poor communes where ethnic minorities are concentrated. 2016-2020 The government consolidated the 16 NTPs into two NTPs for the 2016–2020 implementation period to increase efficiency and reduce overlapping activities. The two programs are the New Rural Development (NTP-NRD) and the Sustainable Poverty Reduction Program (NTP-SPR) with the P135 falling under the NTP-SPR. NTP-NRD is designed to upgrade services and infrastructure NTP-NRD has four ambitious objectives: (a) 50 percent of communes to meet NRD for rural communities across all 63 Provinces of Vietnam. It standards (achieve 15 of the 19 preset criteria), and each province, and each city under A DECADE OF SIGNIFICANT PROGRESS, BUT LAST-MILE CHALLENGES REMAIN encompasses 11 activity groups linked with 19 economic and social Central Authority should have at least one district that meets NRD standards (i.e., criteria relating to poverty, education, health, transportation, water meeting all 19 criteria); (b) Communes, on average, to meet 15 out of 19 NRD criteria, supply, irrigation, livelihoods, agricultural production, markets, and no commune to achieve less than 5 criteria; (c) Basic production and quality of life culture, energy, environmental issues, communication, and requirements to be achieved for rural citizens in areas such as transportation, power security. supply and domestic water, schools, and health stations; and (d) income levels to increase by at least 1.8 times compared with 2015. NTP-SPR supports infrastructure, livelihoods, basic services, and NTP-SPR has four ambitious targets: (a) lowering the poverty rate by an average of capacity building for the country’s 94 poorest districts and 310 1.5 percent per year; (b) improving the livelihoods and quality of life for the poor by communes in coastal areas through sub-programs: (i) Program increasing per capita income of poor households by 1.5 times from 2015 to 2020; (c) 30A; (ii) P135; (iii) Production support, livelihood diversification implementing poverty reduction mechanisms and policies in a consistent and effective and scaling up of poverty reduction models in non-P30A and manner to improve living conditions and enhance access to basic social services for the non-P135 communes; (iv) Communications and information poor; and (d) investing in the infrastructure of poor districts, communes, and villages poverty reduction. NTP-SPR falls under the overall purview of with special difficulties following NTP- NRD criteria. MOLISA, with CEMA playing a strong role as it has relatively independent management of Program 135 (SPR-P135). Program-135 (SPR-P135) is one of its five sub-programs, referred to as projects, which supports 2,240 poor communes and 33,723 poor villages in ethnic minority and mountainous areas. SOCIOECONOMIC DEVELOPMENT PLAN NATIONAL TARGETING PROGRAMS (NTPS) TARGETS AND COVERAGE (SEDP) PHASES 2021-2025 NTP-NRD supports continued development of the agriculture NTP-NRD has four key targets. These include: (a) 80% of communes reach NRD sector, promoting sustainable rural development and urbanization. standards, increase average income per capita of rural inhabitants by 1.5 times compared The program also aims to improve living standards of rural to 2020; (b) 50% of provinces, communes and cities to reach NRD standards; (c) At communities and promote gender equality. NTP-SRD is designed least 15 provinces and cities meeting NRD standards; (d) 60% of villages in special areas to improve infrastructure in rural communities, modernizing it including border area, mountainous, coastal, and island areas reach NRD standards. and ensuring its climate resilience. Total budget is 196,332 billion VND (39,632 billion VND from national budget and 156,700 billion from local budget) NTP-SPR implements reduction in multidimensional poverty, Main targets of NTP-SPR are as follows: (a) Reduce the share of multidimensional poor and supports poor households in improving livelihoods, by 1-1.5% each year; (b) Reduce the share of poor ethnic minority households by 3% access to services, and living standards. This program supports each year; (c) Alleviate poverty for 30% of poor districts and 30% of poor coastal and poverty alleviation in the poorest districts and coastal and island island communes. communes. Total budget is 75,000 billion VND (48,000 billion VND from national budget, 12,690 billion from local budget, 14,310 from other sources of funding) NTP-Socioeconomic development of ethnic mountainous areas Key targets of NTP-SEDEMA: (SEDEMA) will classify disadvantaged ethnic minority communes as those with high rates of poverty, high ratios of untrained ‒ Double average income per capita of ethnic minorities compared to 2020 workers, poor roads, high child mortality rates, and high illiteracy ‒ Reduce share of ethnic minority households in poverty by 3% each year, reducing rates. the share of poor communes and villages by half. The broad objective of NTP-SEDEMA is to utilize the ‒ 100% of communes with tarmac or cement roads for cars to the center, 70% of potential and comparative advantage of ethnic mountainous villages with solid roads for cars. 100% of schools and health centers have structure areas, promoting innovation, driving economic growth, and reinforced. 99% households have access to national electricity grids or other ensuring social security; reducing poverty and the gap between sources. 90% of ethnic minority people have access to water. 100% have access to living standards and income and the national average; develop television and media. Improving infrastructure in poor areas and villages. infrastructure, improving connectivity; develop education ‒ Complete resettlement of 90% of households. Resettle/develop 60% of and training, health, and culture; improving living standards; households living in remote or at-risk areas. improving the quality and quantity of ethnic minority cadres ‒ Increase share of 5 year-olds attending kindergarten to 98%, share of children and workers; promoting ethnic minority cultures while reducing attending primary school to 97%, share of children attending secondary school backwardness; improving gender equality, addressing issues with to 95%, 60% for high school. Reach 90% for share of people over 15 who can read women and children; build resilient local political system, ensuring and write in Vietnamese. security, and improving unity and trust in government. ‒ Improve access to health care, eradicating certain illnesses. Increase health insurance coverage to 98%. Access to maternal care to 80%, reducing malnutrition to lower than 15%. ‒ Enabling 50% share of working age population to receive appropriate vocational training ‒ Preserve cultural traditions ‒ Develop capacity of ethnic minority cadres Total budget for the 2021-2025 period is 137,664 billion VND (104,954 billion VND from national budget, 10,016 billion VND from local budget, 19,727 billion VND from loans, 2,976 billion VND from other sources of funding) Challenges to reducing poverty among the remaining poor Chapter 3. From the Last Mile to the Next Mile 107 Source: World Bank, 2017; Office of the Government of Vietnam, 2021 PART 2. Part 2. The Next Mile is the Road Ahead 108 2022 Vietnam Poverty and Equity Assessment The Next Mile is the road to upper-middle and high-income country living standards. Human capital—a combination of the education, skills, and health factors that largely determine labor productivity— has been a key driver of sustained economic growth and is also a key ingredient to break intergenerational poverty traps. An important aspect to ensure equitable economic pathways includes finishing the job of narrowing education gaps (Chapter 4). Economic growth has transformed the country, but preserving these gains and creating new, sustainable economic pathways towards higher levels of income is the challenge ahead (Chapter 5). Yet, Vietnam faces challenges in creating pathways based on a transition to high-skilled labor. Its labor market is still characterized by low wages, high informality, medium-skills, and a slow expansion of high-skilled occupations. In a world with unpredictable risks and challenges, promotive policies to sustain economic mobility must be complemented by protection strategies to preserve gains and guard against poverty traps. Households face individual and communal risks such as accident and illness, unemployment, and environmental—which will be exacerbated by climate change—as well as future pandemics and economic shocks. Yet the current social protection system does not adequately protect all households against all risks and needs to be modernized (Chapter 6). Addressing these challenges will require significant public investment. However, a rapidly aging population and rising middle-class demands will bring additional spending pressures. How can fiscal policy finance these investment needs while also helping reduce poverty and inequality? (Chapter 7). CHAPTER 4. early human capital formation and poverty reduction CHAPTER 5. Sustaining upward economic mobility CHAPTER 6. Building resilience and safety nets to guard against shocks CHAPTER 7. Fiscal policy, financing future investment and buidling the middle class From the Last Mile to the Next Mile 109 PART 2. Chapter 4. Early human captial formation and poverty reduction Chapter 4. Early human captial formation and poverty reduction Key Messages • High-quality education can break inter-generational cycles of poverty, promote human capital formation, and support poverty reduction, with positive implications for labor productivity, economic growth, and average life expectancy. • However, family backgrounds still explain variations in on-time education completion, and private expenditures in education vary substantially across household wealth quintiles and ethnicity, even for public education at compulsory levels. • The government and private sector have a role in promoting equitable access and consistent quality. 110 2022 Vietnam Poverty and Equity Assessment Chapter 4. Early human captial formation and poverty reduction 4.1. Human Capital– a key human capital formation can limit their economic mobility in driver of sustained growth adulthood (Narayan et al., 2018). In high-income countries, levels of educational attainment are closely linked to economic Human capital is a combination of the education, performance. Similarly, strong health indicators are essential skills, and health factors that largely determine labor to support economic growth and productivity. High-quality productivity. These factors have been a key driver of the education can break intergenerational cycles of poverty, promote sustained economic growth and poverty reduction observed human capital formation, and support poverty reduction, with across East Asia over the past several decades (Gill, Revenga, positive implications for labor productivity, economic growth, and Zeballos, 2016; Permani, 2009; World Bank, 2018). and average life expectancy. Educated populations also tend to Widespread basic literacy and numeracy gained through have higher levels of social tolerance, intercommunity trust, and education supported employment in manufacturing and demand for public services. product-assembly sectors in the early stages of economic development in the region. Following rapid economic growth, Education is a strong determinant of poverty. The some countries saw educational attainment accelerate across difference in poverty rates between households with the least generations, facilitating the development of knowledge- and and the most education can be as large as the variation in regional skill-intensive activities. With the region’s strong focus on poverty rates discussed in Chapter 3. In 2020, poverty rates education, educational attainment levels have converged among households with primary education or less was about with global averages. The level of schooling increased even 15 percent at the LMIC (US$3.20/day 2011PPP) poverty line, while the population of the region doubled. The rapid rise while virtually no households with tertiary education were in educational attainment has continued to this day, with in poverty (Figure 4.1). The population living in households higher numbers of students completing secondary school and with primary education or less comprise about 20 percent of progressing to tertiary education (World Bank, 2018). the total population, but they also constitute about half of the LMIC poor population. The gaps in poverty rates are even The development of human capital plays a vital role in larger at the UMIC (US$5.50/day 2011PPP) poverty line. A boosting growth and reducing poverty and inequality. cohort effect is partly at play since education completion rates For children, the education they receive and how healthy have improved considerably across generations and youths are they grow up affects their future wages, life expectancy, and less engaged in agriculture. However, differences in poverty by human capital as adults (World Bank, 2019a). Falling short on education level persist even accounting for these effects. Figure 4.1. Poverty rates by household education group Maximum education in the HH LMIC poverty ($3.20/day 2011PPP) UMIC poverty ($5.50/day 2011PPP) 80 80 70 70 60 60 Poverty rate (%) Poverty rate (%) 50 50 40 40 30 30 20 20 10 10 0 0 2010 2012 2014 2016 2018 2020 2010 2012 2014 2016 2018 2020 No education Primary Lower secondary Upper secondary Vocational Tertiary Note: Household education category determined by the maximum level of education of any household member. Source: World Bank staff calculations using VHLSS. From the Last Mile to the Next Mile 111 PART 2. THE NEXT MILE IS THE ROAD AHEAD 4.2. Vietnam’s averages on stunting, it scores below average in stunting performance in human globally and child survival rates, having higher stunting and lower survival rates than half of all countries in the world. capital vs regional and global peers Vietnam performs well in compulsory education completion. Vietnam’s youth possess high human capital potential. The World Bank’s Human Capital Index (HCI) measures Across the different dimensions of the HCI, Vietnam actual and potential productivity levels for the next ranks relatively high in terms of education quality. generation of workers (World Bank, 2019). Despite being a Vietnam’s strong performance on the HCI is partly lower-middle-income country, Vietnam is among the top attributable to its very high Harmonized Test Scores. Scoring (quartile) performers worldwide in terms of the HCI, not 519 out of 625, Vietnam’s result ranks 16th in the world, only outperforming other countries in the East Asia and surpassing that of many high-income countries, and second Pacific region, but also significantly surpassing other lower- only to Singapore in ASEAN. However, international middle-income countries (LMICs). According to the 2020 assessments in Vietnam are not without some sensitivity. The series of HCI data, a child born in Vietnam today will be 69 OECD did not rank Vietnam in the 2018 PISA round, stating percent as productive when she grows up as she could be if she that the international comparability of the results could enjoyed complete education and full health, compared to 59 not be ensured. The PISA score used in the World Bank’s percent for the average child in the EAP region, and 48 among 2020 calculations were from the 2015 assessment round. LMIC children (Table 4.1). International comparisons are not without challenges, and this chapter will compare differences in test scores between However, there is scope to further improve human students in Vietnam. capital formation across some dimensions. Among ASEAN nations, Vietnam’s performance on the overall On the expected number of school years, a measure of HCI ranks only second to Singapore, but the picture is more the quantity of education, Vietnam performs relatively nuanced when examining sub-indicators. Across the five well in the region. In Vietnam, a child who starts school indicators used to compute the World Bank’s HCI, Vietnam at age 4 can expect to complete 12.9 years of schooling by scores significantly above average compared to peer groups in her 18th birthday. Adjusting for the quality of schooling, as education but scores closer to the average in health indicators. proxied by harmonized test scores, the World Bank estimates While Vietnam performs better than the EAP and LMIC that children in Vietnam are expected to complete 10.7 years Table 4.1. Human Capital Index - Vietnam vs. peers (2010 vs. 2020) EAST ASIA AND LOWER MIDDLE VIETNAM MALE + FEMALE INDICATOR PACIFIC INCOME 2010 2020 2010 2020 2010 2020 HCI Component 1: Survival Probability of Survival to Age 5 0.977 0.979 0.972 0.978 0.944 0.959 HCI Component 2: Expected Years of School 11.8 12.9 12.0 11.9 9.6 10.4 Schooling Harmonized Test Scores 533 519 500 432 407 392 HCI Component 3: Survival Rate from Age 15-60 0.864 0.867 0.847 0.864 0.760 0.804 Health Fraction of Children Under 5 Not Stunted 0.773 0.762 0.691 0.759 0.690 0.750 Human Capital HCI Index (HCI) 0.66 0.69 0.68 0.59 0.45 0.48 Source: World Bank HCI (2020). 112 2022 Vietnam Poverty and Equity Assessment Chapter 4. Early human captial formation and poverty reduction of schooling (Figure 4.2). This measure of ‘learning-adjusted 2020, the average child survival rate has increased from 97.7 years of schooling’ is designed to incorporate differences in to 97.9 percent, which translates to around 2 fewer deaths per the quality of education provided as proxied by tests scores–a 1,000 live births. In 2020, nearly 21 in 1,000 children will not given number of years in school does not imply the same survive until age 5 in Vietnam. While adult survival rates in level of learning across different countries. While Vietnam’s Vietnam are higher than LMIC, UMIC, and EAP averages, learning-adjusted years of schooling are very high (high- there are further gains to be made to reach the levels seen in income country average is 10.3 years), it is subject to similar high-income countries. sensitivity issues since it is adjusted by harmonized test scores. Malnutrition among children is one of the dimensions where the challenges are greatest in Vietnam. In 2020, Figure 4.2. Learning-adjusted years of schooling by almost a quarter of children under 5 years old are stunted. GDP per capita, Vietnam, and comparators, 2020 The stunting rates in Vietnam are ranked third highest in the ASEAN region and are higher than the LMIC average. 14 Singapore Stunting, defined as a low height-to-age ratio, is indicative of impaired growth and development in children resulting 12 Vietnam from chronic malnutrition, repeated infection, or inadequate Learning-adjusted years of education 10 Philippines psychosocial stimulation. Early childhood stunting has long- lasting impacts on cognitive development, leading to poorer 8 Malaysia educational performance, lower adult wages and productivity, Cambodia Thailand and other health conditions in later life. Nutrition is especially 6 Indonesia concerning among ethnic minority children. A 2019 study 4 Myanmar Lao PDR finds that one in three ethnic minority children are stunted and one in five are underweight (Mbuya et al., 2019). 2 0 Improving access to maternal health care, early childhood 6 7 8 9 10 11 12 health care, and nutrition is needed to further reduce Log Real GDP Per Capita at PPP under-5 mortality rates. In 2020, the under-5 mortality Source: World Bank staff calculations using HCI data sheet, WDI. rate stood at 21 deaths per 1,000 live births, higher than that of all ASEAN countries except Philippines and Cambodia (Figure 4.3). This is reflective of the prenatal, infant, and early Vietnam’s strong performance on the education childhood environments in Vietnam. Infectious diseases, component of the HCI reflects the government’s focus birth complications, and congenital birth defects are leading and commitment to education, in particular, general causes of death for children globally. Chronic malnutrition education. A World Bank report has identified a number also makes children more susceptible to dying from common of success factors in Vietnam’s education system including childhood illnesses. To further improve child survival rates the government’s commitment to educational development, and reach UMIC and HIC levels, Vietnam needs to ensure investment in general education, basic inputs and equity, high equitable access to quality maternal and postnatal care and household expenditure on education, and development of the early childhood health services, as well as address the existing teaching workforce (World Bank, 2020). challenge of poor childhood malnutrition. Health indicators have scope for improvement. Adult survival rates in Vietnam are comparable to those of other ASEAN nations. The probability that Across health indicators, there is still progress to be made a 15-year-old person will reach their 60th birthday is 87 in ensuring that the youth population achieves good percent (Figure 4.4). There are several primary observable health that will enable them to productively contribute factors that can be said to have an impact on adult mortality, to the economy. Over the last decade, progress in improving including noncommunicable diseases, access to healthcare, health outcomes in Vietnam has been slow. From 2010 to income levels, labor laws, and lifestyle factors. As of 2017, From the Last Mile to the Next Mile 113 PART 2. THE NEXT MILE IS THE ROAD AHEAD public health expenditure was equivalent to 2.8 percent of 4.3. Variations in human GDP in Vietnam. The share of 30 to 70 year-olds dying from noncommunicable diseases such as cardiovascular disease, capital formation exist cancer, diabetes, or chronic respiratory diseases was 17 percent within in Vietnam in 2016. While Vietnam’s adult survival rates are comparable to those of other countries in the region, to achieve HIC levels While at the national level, indicators of human capital by 2045 will require continued investment in improving the formation appear mostly in line with regional peers, health of its economically productive population. there are gaps within Vietnam across different groups. This section illustrates the differences in human capital formation outcomes across gender, household socioeconomic Figure 4.3. Fraction of children not stunted by GDP groups, regions, and ethnicities. When structural gaps per capita, Vietnam, and comparators, 2020 exist that limit the opportunities of entire groups, the full 100% potential of human capital development will not be reached. Disparities in access to nutritional, medical, educational, 90% Thailand and social resources can impact children’s future economic productivity or contribute to the misallocation of human Fraction under 5 not stunted 80% Philippines Vietnam capital, inhibiting future economic growth. Malaysia Indonesia Gaps in human capital outcomes in Vietnam 70% Cambodia between children from bottom and the top 20 Lao PDR percent of households are larger than the average 60% Myanmar Socioeconomic status-disaggregated HCI (SES-HCI35) 50% scores can be calculated to illustrate variation in human 6 7 8 9 10 11 12 Log Real GDP Per Capita at PPP capital formation across different household wealth quintiles. Family characteristics and resources are strongly Source: World Bank staff calculations using HCI datasheet, WDI. linked to the outcomes of children. The HCI is 17 points lower than the average for children in the poorest 20 percent of households, compared to 16 percent above the average Figure 4.4. Survival rates by GDP per capita, observed in children in the richest 20 percent. Unsurprisingly, Vietnam, and comparators, 2020 children in the top 20 percent have higher nutrition, health, 100% Singapore and education outcomes (Figure 4.5). Vietnam Thailand 90% Philippines Performance gaps in Vietnam between the bottom Malaysia and the top 20 percent of households are larger than Survival Rate Age 15–60 Cambodia 80% Indonesia the average. Gaps in the Vietnam SES-disaggregated HCI, harmonized test scores, and stunting rates are all larger than 70% Lao PDR the typical gap in a comparative study across 50 developing Myanmar countries (D’Souza, Gatti, and Kraay, 2019). The gap in the SES-disaggregated HCI between the top 20 and bottom 20 is 60% 0.27 points, higher than the average gap in 50 countries (0.15 points). The percentage of children that are not stunted in the 50% richest 20 percent of families is 94 percent, compared to only 59 6 7 8 9 10 11 12 Log Real GDP Per Capita at PPP percent in the poorest 20 percent of households. This translates to a 35-percentage point gap in stunting rates between the top Source: World Bank staff calculations using HCI datasheet, WDI. and bottom compared to a 19-percentage point gap on average. 114 2022 Vietnam Poverty and Equity Assessment Chapter 4. Early human captial formation and poverty reduction Figure 4.5. Gap in socioeconomic-disaggregated HCI, 2013 2.0 1.5 Q5/Q1 ratio 1.0 0.5 0.0 Probability of survival Expected years Harmonized Learning-adjusted Fraction of children SES - disaggregated to age 5 of school test scores years of school under 5 not stunted HCI Note: Ratio of average outcomes between children in Q1 and Q5. SES-HCI (2013). Source: World Bank 2019b. BOX 4.1. World Bank’s human capital index The World Bank created the Human Capital Index (HCI) to assess the conditions that will shape the productivity of the next generation (Kraay, 2018). The HCI is part of a broader World Bank Group initiative that emphasizes and prioritizes investment in people. By anchoring the index in economic theory, it raises awareness of the opportunity cost of inaction and bolsters demand for effective interventions. To accelerate evidence-based investments in people for greater equity and growth, the HCI benchmarks the accumulation of human capital across the world. The index provides an indicator for how the current level of health and education will impact productivity in the next generation of workers. The HCI encompasses key health measures, including survival, education, and health. These three broad categories are measured through the under-5 mortality rate, the under-5 stunting rate, and the survival rate for adults aged 15–60. It also estimates the quantity and quality of education by measuring learning-adjusted years of schooling—the expected number of years of schooling received by age 18, adjusted to reflect educational quality as proxied by standardized test scores. The HCI benchmarks different aspects of human capital including education (expected years of schooling and harmonized test scores) and health (child survival, child stunting, and adult survival rates). These composite indicators are combined to estimate the expected productivity of a child born today as a future worker relative to a benchmark of complete education and full health. Figure B.4.1.1. Framework for calculating the Human Capital Index Contribution to Productivity from Survival Survival to age 5 Expected years of schooling Human Capital Index Contribution to Productivity from School Harmonized test scores Adult survival rate Contribution to Productivity from Health Not stunted rate From the Last Mile to the Next Mile 115 PART 2. THE NEXT MILE IS THE ROAD AHEAD School enrollment rates among children from households human contact. This not only prevents the spread of diseases, in the bottom income quintile lag behind the rates for reduces instances of malnutrition, it also improves the plight children from households in the richest quintile at of women and girls by increasing school attendance with higher grade levels. Starting with universal enrollment in separate sanitary facilities (WHO, 2022a). younger years, differences in enrollment widen toward the end of lower secondary school in the late teen years, at which Gender gaps in human capital formation point over a third of children from households in the bottom quintile have dropped out (Figure 4.6). By age 19, less than In early years, girls outperform boys in human capital one-fifth of 19-year-old children from households in the formation. Girls scored 0.73 on the HCI while boys bottom quintile remain in school. By contrast, 80 percent of scored 0.65 overall (Table 4.2). Across HCI components, students from households in the top quintile continue with girls outperformed across all indicators, scoring higher on their education up to college or university. standardized tests (10 points), attending schools for longer (0.7 years), having higher probability of survival until age 5 Disparities exist among socioeconomic groups in their (0.7 percentage points), having a lower share of stunting (2 access to improved drinking water sources. Measured as percentage points), having higher adult survival rates (12 access to drinking-water sources that are protected from outside percentage points). contamination, including fecal matter and harmful chemicals, nearly a hundred percent of the top two quintiles have access While in the early stages of life girls in Vietnam do to improved drinking water sources. While this remains fairly well, there are still strides to be made for both genders high, only two thirds of households in the poorest quintile have to reach their full potential. Figure 4.8 shows that while access to safely managed drinking water sources (Figure 4.7). schooling up to the lower secondary level could be said to be universal, the proportion of children who are in school The benefits of improved sanitation are well declines rapidly beyond 14 years of age, with small differences documented, yet just over a third of the poorest group in this pattern across genders. For individuals aged 20-24, a in Vietnam can access this. This is in stark contrast to the slightly smaller proportion of women are attending school accessibility of safely managed sanitation for the top 3 income than that of men, reflecting a lower proportion of individuals quintiles, which ranges from 95 to 100 percent. Improved actively pursuing advanced degrees among women than men. sanitation facilities hygienically keep human waste away from Differences in human capital accumulation between genders Figure 4.6. Enrolment rates by household Figure 4.7. Access to improved drinking-water sources wealth quintile and age, 2020 and sanitation facilities, by income quintiles, 2014 96.1 99.2 99.5 100 99.9 120 100 95.5 90.6 Share of cohort population (%) 100 74.3 80.7 80 Share of households (%) 80 60 60 38.2 40 40 20 20 0 0 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Poorest 2nd 3rd 4th Richest (quintile=1) (quintile=5) Poorest 2 3 4 Richest (quintile=1) (quintile=5) Improved water Improved sanitation Source: World Bank staff calculation using VHLSS2020. Source: MICS survey (UNICEF, 2014). 116 2022 Vietnam Poverty and Equity Assessment Chapter 4. Early human captial formation and poverty reduction Table 4.2. Human Capital Index disaggregated by gender     MALE FEMALE     2010 2020 2010 2020 HCI Component 1: Survival Probability of survival to age 5 0.973 0.976 0.981 0.983 HCI Component 2: School Expected years of schooling 11.7 12.5 11.9 13.2 Harmonized test scores 530 514 536 524 HCI Component 3: Health Survival rate from age 15-60 0.808 0.813 0.92 0.924 Fraction of children under 5 not stunted 0.763 0.752 0.784 0.773   Human Capital Index 0.63 0.65 0.68 0.73 Note: Maximum harmonized test score is 625. Source: World Bank HCI. reflect the different challenges that boys and girls face at Geographically, the Central Highlands and Mekong River different stages in the life cycle. Limitations of the HCI mean Delta regions have slightly lower primary enrollment rates that it captures human capital development during stages in than the rest of the country. However, there is a substantial which girls slightly outperform boys, and as such does not disparity between secondary school enrollment rates in rural necessarily translate to improved life outcomes. and urban areas, with urban enrollment rates being nearly 15 percentage points higher than in rural areas (90 vs. 76 percent, respectively). Geographically, the Red River Delta and the Figure 4.8. School Attendance by age and gender, 2020 North Central and Central Coast regions have comparatively 100 higher secondary enrollment rates than the rest of Vietnam. In terms of access to health services, the average access time to School attendance rate (%) 80 a health facility is 48 minutes in urban areas and 75 minutes 60 in rural areas. Even more striking is the maximum time taken to access a health facility: 3,000 minutes (50 hours) in urban 40 areas and 7,200 minutes (120 hours) in rural areas. 20 The majority of children aged 6 to 14 in Vietnam live in 0 rural areas, with only 31 percent residing in urban areas. 10–14 15–19 20–24 Vietnam’s industrialized and urbanized fertility rates declined, Female Male from around 6 births per woman in 1975 to around 2 births per Source: World Bank staff calculations using VHLSS 2020. woman in 2018. This decline was more pronounced in urban areas than rural ones, and is reflected by the mean number of children under 14 in each household. All regions have a lower Outcomes vary by geography, related to the mean number of children per household in urban areas, with provision of public services the notable exception of the Northern and Central Coastal area. The majority of urban children come from the Southeast Human capital formation is associated not with family and Red River Delta regions, accounting for just over half of all resources, but also with public resources in terms of the urban children. The Southeast region is also the only region in quality, access, and availability of schools and health care Vietnam with a larger population of children in urban areas, all services. Thus, a child’s location also has consequences for the other regions being majority rural. Ethnic disparities exist in their opportunities, outcomes, and human capital formation the urban-rural divide. While 36 percent of children belonging potential. Primary school enrollment is nearly universal to the Kinh majority reside in urban areas, only 9 percent of at 98.5 percent, with only small urban-rural differences. children not of Kinh ethnicity reside in urban areas. From the Last Mile to the Next Mile 117 PART 2. THE NEXT MILE IS THE ROAD AHEAD Ethnic minority children underperform in undernutrition among ethnic minority populations are education and health multisectoral. Ethnic minority children are fed less nutritious food, get sick more frequently, and have less access to health Ethnic minority children fall out of schooling earlier services as well as water and sanitation resources compared than other children (Figure 4.9). Poorer ethnic communities to children from the Kinh majority (Mbuya et al., 2019). in geographically remote areas experience higher dropout On average, women living in mountainous and remote areas rates. Remoteness also reduces exposure to the national get married younger, have lower educational attainment, language for ethnic minority children, which is a factor and receive less reproductive care compared to their peers in the educational gap between Kinh and ethnic minority in lowland and urban areas. Poverty, an important basic children. One of the reasons for the underlying variation in determinant of undernutrition, is also concentrated among enrollment rates is the opportunity cost of sending children ethnic minorities. Stunting rates matter for poverty reduction to school. For many poor families, the labor contribution because they are a key predictor of lower educational of a child is too valuable to justify keeping him or her in attainment, adult health, and future earnings. school. The long-term gains from education are simply not commensurate with the short-term sacrifice in income or Access to safely managed water and sanitation are major expenditure. Several nonfinancial factors are also constraints determinants of good health. The vital importance of to educational attainment among poor and ethnic minority this has been reflected in the United Nation’s Sustainable households, including poor infrastructure and accessibility, Development Goals number 6, which aims to ensure the limited quality of teachers, and the perceived irrelevance of the availability and sustainable management of water and school curriculum. (Pimhidzai, 2018). The lack of education sanitation for all. Unsafe and contaminated water leads in turn reduces access to better employment or productive to illnesses such as cholera and typhoid. Similarly, poorly opportunities among the poor, resulting in low earnings, and managed sanitation spreads diarrheal diseases which in turn perpetuates intergenerational trends. leads to stunting in children. The arrival of COVID-19 has further highlighted the significant role of safe water, sanitation Nutritional challenges remain, especially for ethnic and hygiene for human health and development. This section minorities. Stunting rates among children age 5 in certain analyses data from the Multiple Indicator Cluster Surveys underdeveloped regions have stagnated over the last decade, conducted by UNICEF for Vietnam for 2013-2014. More up particularly in remote mountainous regions with large to date survey results are yet to be published. ethnic minority populations. The key determinants of Figure 4.9. Enrolment rates by ethnicity Figure 4.10. Access to improved drinking-water and age, 2020 sources and sanitation facilities, by ethnicity, 2014 120 100 Share of cohort population (%) 100 80 Share of households (%) 80 60 60 40 40 20 20 0 0 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Improved water Improved sanitation Ethnic minority Kinh Ethnic minority Kinh Source: World Bank staff calculation using VHLSS2020.. Sources: MICS Survey (UNICEF, 2014). 118 2022 Vietnam Poverty and Equity Assessment Chapter 4. Early human captial formation and poverty reduction Access to improved sanitation and water-drinking The dissimilarity index – a measure of dispersion sources are higher for Kinh ethnic majority, compared of outcomes across groups to ethic minorities in Vietnam (Figure 4.10). While only 5 percent of Kinh households are still using unimproved The dissimilarity index is informative of how variation in drinking water sources, this figure stood at 25 percent of outcomes across groups (see Annex 4.2 for measurement ethnic minority households. These households are consuming details). It quantifies variations across groups by comparing water from unsafe sources such as unprotected wells, springs, overall and group-specific outcomes. The dissimilarity index surface water (WHO, 2022b). This is likely to result in poorer can be interpreted as the degree of reallocation between health, higher expenditure on medical costs and lower school groups to achieve equality of access or a constant coverage attendance for children. On sanitation, Kinh majority are rate across all groups of children aged 6–14. For example, in 1.7 times more likely to be able to access improved sanitation the case of on-time lower-secondary completion, 89 percent facilities such as flush or pour-flush to piped sewer system, or of children in the relevant age group have achieved on-time septic tank pit latrines among other safely managed options. completion. Depending on the group of interest, children With only half of ethnic minorities being able to use improved have different rates of on-time education completion. sanitation facilities, this will have serious consequences for the However, the difference between groups is small. To achieve socioeconomic development of this group. the same completion rate across all demographic and socio- economic groups, the difference is completion among only 5.6 percent (dissimilarity index) of children (Figure 4.11). 4.4. Quantifying the sources of variation in Among indicators of on-time education completion, the dissimilarity index increases with grade level. Among child outcomes within compulsory levels of education, there is higher dissimilarity Vietnam in on-time completion at the lower-secondary level than at the primary. The dissimilarity in on-time completion The previous section illustrated broad human capital of upper-secondary education (which is non-compulsory) trends in Vietnam and how these outcomes varied is significantly higher at 16 percent. The rate of on-time across ethnic, geographic, and socioeconomic groups. upper-secondary completion in general is lower, averaging This section quantifies these differences in outcomes using 68 percent. Studies on the determinants of school dropouts decompositions by specific characteristics and family in Vietnam provide evidence on the correlation between the backgrounds. probability of dropping out and such factors as age, working time, primary education, number of siblings, parental Figure 4.11. Dissimilarity index in education and health 0.20 0.155 0.15 Dissimilarity Index 0.10 0.058 0.056 0.044 0.050 0.05 0.014 0.009 0.001 0.00 Access to flush Access to safe Access to health Enrolled in Some Primary Lower Upper toilet water insurance school primary secondary secondary Health Education One time completion enrollment Source: World Bank staff calculations using VHLSS 2020. From the Last Mile to the Next Mile 119 PART 2. THE NEXT MILE IS THE ROAD AHEAD education, household’s per capita expenditure, the cost of of births by ethnic minority women were assisted by trained schooling (Vo and Trinh, 2005), and gender (Bui, 2011). staff. In contrast, 74 percent of births for the poorest quintile Baulch et al. (2012) consider two groups of dropouts: (1) nationwide and 95 percent for remaining quintiles were children who have completed lower secondary grades (and are assisted by trained staff. already enrolled in upper-secondary school), and (2) children who have not completed lower secondary school. They find Ethnic minority students, particularly those in that the strongest predictors were parental education for mountainous areas, tend to experience “double the first sub-sample and the amount of paid or unpaid work segregation” by ethnicity and concentrated poverty. The undertaken by the child in the previous survey round. For the network of boarding and semi-boarding secondary schools second sub-sample, however, only the mother’s education and which cater to above 85 percent of ethnic minority students ethnicity are essential factors behind the likelihood of dropout (i) have caps limiting non-Kinh students to 5-10 percent before lower secondary completion. This is followed by poor of total school enrollment and (ii) can only serve students socioeconomic status—poverty and parental education less coming from the most disadvantaged communes and villages than lower secondary education. in ethnic minority and mountainous areas. Substantial heterogeneity is masked under the distinction between Kinh Household wealth, location, and ethnicity are vs. “ethnic minorities” category. Most of the main data sources large predictors of child education outcomes. on learning outcomes in Vietnam, such as PISA and NLSA, are collected at the level of “ethnic minority” classification, The dissimilarity index can be decomposed along which merges together ethnic groups with vastly different different sets of circumstances that may affect a child’s learning experiences and outcomes. outcome. For comparison, four sets of circumstances are used, consisting of (1) geographical factors and gender, (2) Considering all the circumstances together (Set 4), household characteristics, (3) household characteristics household geographic and wealth factors are the most and expenditure quintile, and (4) a combination of all explanatory of a child’s education outcome. Household circumstances from sets 1-3 (Figure 4.12). Examining the wealth (as measured by household consumption per capita decomposition results among these four sets enables discerning quintiles) consistently makes up the largest proportion of the the relative importance of each of the circumstances in driving dissimilarity of on-time education completion. The association the dissimilarity of outcomes. between student learning and household characteristics is a well- established empirical finding both in developed and developing Ethnicity plays a larger role in the variations in certain countries, and gaps in education outcomes between children outcomes among children. When examining only Set from high and low socioeconomic backgrounds are widely 2 indicators, a parent’s ethnicity is more often the largest documented. For instance, Filmer and Pritchett (1999) use explanatory factor of dissimilarity, more so than parental data from 35 developing countries to show that most shortfalls education. When controlling for the full set of circumstances in basic education indicators (e.g., enrollment and dropout) (Set 4), ethnicity has a larger explanatory role than urban/ are due to children in the bottom 40 percent of the wealth rural or household education circumstances for some distribution. Likewise, Schuetz, Ursprung, and Woessmann indicators. Indicators where ethnicity is still a strong factor (2008), using TIMSS data of 54 countries, found that student are access to flush toilet and health insurance. Moreover, while and household characteristics have significant effects on household wealth absorbs many of the dissimilarity effects, by achievement. A recent meta-analysis of 49 empirical studies on all accounts variation in education enrollment and outcomes socioeconomic status and academic outcomes in 38 developing still vary considerably by ethnic group. On health indicators, countries similarly finds that parents’ socioeconomic status ethnic minorities and those living in remote provinces have has a strong impact on students’ academic attainment and lower access to health services and experience poorer health achievement, with effect sizes stronger in more economically outcomes. The child mortality rate in rural areas in 2016 was developed countries and moderated by grade level, gender, and 26 per 1,000 live births compared with 12.7 per 1,000 live racial and ethnic background (Kim et al., 2019). births in urban areas (World Bank, 2021b). Only 68 percent 120 2022 Vietnam Poverty and Equity Assessment Chapter 4. Early human captial formation and poverty reduction Figure 4.12. Shapley decomposition of the dissimilarity index in education and health outcomes 100 Set 1: Child 80 circumstances Share (%) (gender, region) 60 40 20 0 Enrolled in Some Primary Lower Upper Access to Access to Access to school primary secondary secondary flush toilet safe water health insurance Enrollment On-time education completion Health 100 Set 2: Parental 80 circumstances Share (%) 60 40 20 0 Enrolled in Some Primary Lower Upper Access to Access to Access to school primary secondary secondary flush toilet safe water health insurance Enrollment On-time education completion Health 100 Set 3: Parental 80 circumstances + Share (%) wealth 60 40 20 0 Enrolled in Some Primary Lower Upper Access to Access to Access to school primary secondary secondary flush toilet safe water health insurance Enrollment On-time education completion Health 100 Set 4: All 80 circumstances Share (%) 60 40 20 0 Enrolled in Some Primary Lower Upper Access to Access to Access to school primary secondary secondary flush toilet safe water health insurance Enrollment On-time education completion Health Child gender (female, male) Region Urban or rural Kinh or ethnic minority HH age HH gender HH education HH wealth quintile Note: See Table A.4.1.1. for values. HH = household head. Source: World Bank staff calculations using VHLSS 2020 From the Last Mile to the Next Mile 121 PART 2. THE NEXT MILE IS THE ROAD AHEAD Learning outcomes in Vietnam suffer from acute but others, such as access to the internet, are much less geographical inequality, with the Mekong Delta and common. Thus, having access to a lower number of indicators Central Highlands regions persistently falling behind is more easily attainable. other regions in the past 10 years. Although the regional gaps have gradually narrowed, these lagging regions continue Figure 4.13. Dissimilarity index in bundled opportunities to suffer from low net enrollment, high dropout rates at the upper-secondary level, and low progression (survival) 0.30 between education levels, between 2010–2020. Students 0.25 from the Mekong Delta region rank at the bottom on most Dissimilarity index assessed subjects at all levels from primary to upper-secondary 0.20 education. Moreover, there has been no national narrative 0.15 encouraging educational advancement for this group in the 0.10 way there has been for ethnic minorities, nor has there been one regionally. The regional development master plan for 0.05 Mekong Delta currently has little to say about education. 0.00 8 out of 11 9 out of 11 10 out of 11 Complementarity of opportunities - Children Note: Total of 11 indicators in a bundle, see Annex 4.2. In the bundle, on-time need joint access to multiple opportunities to be education completion is considered one indicator. able to succeed in society. Source: World Bank staff calculations using VHLSS 2020. Access to education and opportunities are not substitutes for one another but rather are complements. Children The main determinant of dissimilarity for bundled require access to multiple opportunities to achieve their full opportunities is household wealth (Table A.4.1.1). When potential. It is also useful to see how different factors influence accounting for all factors (Set 4), household consumption the dissimilarities in access to opportunities across the board. expenditures explain nearly half of the dissimilarity. While We construct an outcome that is a minimum bundle across a ethnicity and region are the second and third strongest range of 11 complementary indicators, including education, determinant for vulnerability defined by having eight or more health, housing quality, access to services, and assets (see Annex opportunities, their contributions decrease as the size of the 4.2). The importance of complementary opportunities was bundles increases further; the explanatory contributions from made especially salient since the onset of COVID-19 during urban status and the education of the household head increases school closures and prolonged lockdowns. Access to distance and eventually overtakes ethnicity and region. Gender is not learning and the quality of continuity varied considerably a major contributor of dissimilarity in bundled opportunities across regions, and access to the internet and home computers when decomposed using this set of circumstances. A similar became essential for education continuity. pattern with ethnicity emerges in that it matters less as the definition of bundled opportunities become increasingly Children can be said to be vulnerable if they have access stringent. to fewer than a minimum number of opportunities in a bundle (8, 9, or 10 out of 11). We determine a threshold Over time, variation has narrowed under which a child is said to be vulnerable by looking at the proportion of individuals deemed vulnerable under each Over a decade, dissimilarity of opportunities and access threshold. While upwards of 80 percent have access to at least has declined across a range of indicators, highlighting 8 opportunities, only around 60 percent have access to at least broad improvements (Figure 4.14). One of the most 9 opportunities, and around 30 percent have access to all 10 significant gains is access to the internet and health insurance. opportunities. As expected, dissimilarity is increasing when Improvements in access and opportunities are largely driven non-vulnerability is defined as having a larger number of by the scale effect, meaning a proportional increase in coverage opportunities (Figure 4.13). Some components in the bundle across all groups. For example, being fully insured sees the are common or nearly universal, such as access to electricity, largest gain of 47 percentage points, 30 percentage points 122 2022 Vietnam Poverty and Equity Assessment Chapter 4. Early human captial formation and poverty reduction of which are contributed by the scale effect. While equity in regions were most affected. As many as 50 out of 63 provinces educational outcomes is improving, there is larger variation closed schools early in January and February 2021 ahead of in complementary indicators such as access to improved Tet holidays because of the third outbreak. housing, internet, health insurance, safe water, and sanitation. Dissimilarity still remains high in the most advanced indicators Adverse learning effects from school closures remain such as access to computers, internet, and on-time education challenging because of limited access to distance learning completion of upper-secondary education, illustrating that and remedial learning. Among households with children there is room for improvement to reach upper-middle and having school disruptions, only 61 percent had access to high-income country levels. online classes and almost 20 percent did not have access to any distance learning opportunities between September 2020 Inequality in education continuity persisted and March 2021. In locations where online classes were not during COVID-1936 available or accessible, the Ministry of Education and Training and the provincial Departments of Education and Training Complementary access and opportunities were arranged to broadcast lessons on television and/or radio. especially important during COVID-19, which revealed However, these mediums account for only a negligible portion discontinuities in education and learning losses. During of distance learning. Short Message Service and paper-based the first outbreak in early 2020, Vietnam’s education sector self-study are the second most popular methods, but they are responded early and decisively to COVID-19 risks and closed generally considered ineffective by teachers and students. all schools for almost three months—among the longest closures in the world. Although containment measures School closures hit poor, ethnic minorities, and low- became more localized and targeted over time, school closures achieving students particularly hard (Figure 4.15). Over were still widespread. Between September 2020 and March 60 percent of households in the lowest welfare quintile and 2021, 72 percent of households with a child between 6 and ethnic minorities, and close to 59 percent of those living in 18 years of age experienced school closures. Households with the Midlands and Northern Mountainous Areas, Central children from urban areas, the Red River Delta, and Southeast Highlands, and the Mekong Delta regions did not have online Figure 4.14. Equity in access is improving (dissimilarity in 2010 and 2020) 80 70 60 Dissimilarity Index 50 40 30 20 10 0 Access to home computer Access to home internet Access to mobile Access to flush toilet Access to safe water Access to health insurance 8 out of 11 9 out of 11 10 out of 11 Enrolled in school Access to electricity Adequate housing Housing quality Some primary Primary Lower secondary Upper secondary Assets Health Housing Bundled opportunities Enrollment On-time education completion (total of 11) 2010 2020 Note: See Annex 4.2 for definition of wider set of indicators. Source: World Bank staff calculation using VHLSS 2010 and 2020 From the Last Mile to the Next Mile 123 PART 2. THE NEXT MILE IS THE ROAD AHEAD learning offered during school closures. By contrast, only 4 in the form of early summer breaks or complete closure percent of households in the Red River Delta region (Hanoi without remedies or access to distance learning. The latest wave area) did not receive distance learning. The Red River Delta of closures also coincides with a key assessment and testing region and the Southeastern region where Hanoi and Ho Chi period. It is still uncertain how these high-stakes assessments, Minh City are located, respectively, had high rates of online including secondary graduation exams, will be implemented. live classes (86 and 71 percent, respectively). Given that high-stakes exams are used as a key qualification to enter upper-secondary schools and higher education, school These regional differences reflect not only inequality interruption and exam cancellation can have potential long- across households but also inequality in public services term consequences for education completion. Because more across regions. Access rates and performance of the Midlands poorly resourced provinces are less able to provide education and Northern Mountainous Area, Central Highlands, and continuity, disparities in education completion may also widen the Mekong Delta regions have persistently lagged other across socioeconomic groups. regions at all levels of education. These regions also suffer from a technological infrastructure deficit, inhibiting their Education expenditures vary substantially across ability to implement distance learning. Poor information household backgrounds. and communication technology capacity is reflective not just of insufficient infrastructure but also of weaker governance, Investment in supplementary education services and in addition to poor and unsustainable financing, given that resources is tied to family resources. Richer households governance and financing for general education are highly devote substantial private expenditures to secure better decentralized to the provinces. outcomes for their children than poorer households do. At every level of education, wealthier families spend more than The situation remains fluid, and provinces continue poorer ones. Starting at compulsory levels of education reopening and reclosing schools in a relatively disruptive (primary and secondary), households in the top quintile manner in response to new outbreaks. The arrival of more spend 5.6 times more on extra courses than households in transmissible COVID-19 variants in mid 2021 resulted in the lowest quintile (Figure 4.16). Kinh households spend widespread school closures in 49 out of 63 provinces, mostly over seven times more than ethnic minority households on Figure 4.15. Continuity of education across regions and socioeconomic groups A. Access to distance learning, by income quintile B. Access to distance learning, by region 100 100 Share of affected households (%) Share of affected households (%) 80 80 60 60 40 40 20 20 0 0 No distance Online SMS Paper-based Other No distance Online SMS Paper-based Other learning live class self-learning learning live class self-learning Bottom 20 Q2 Q3 Q4 Top 20 Red River Delta Midlands & Northern Northern & Coastal Mountains Central Central Highlands Southeastern Area Mekong Delta Note: Household welfare quintiles (Q) are based on household consumption per capita in 2018. SMS = Short Message Service. Source: World Bank Vietnam COVID-19 household monitoring surveys (round 5, March 2021). 124 2022 Vietnam Poverty and Equity Assessment Chapter 4. Early human captial formation and poverty reduction extra classes during compulsory education. These differences When child outcomes are related to their parents, translate into disparities in educational attainment, slowing intergenerational outcomes persist. intergenerational mobility and exacerbating inequality. For children around age 8, the participation rate in extra classes The association between a child’s outcome and their is almost twice as high among top quintile households parents is referred to as intergenerational persistence compared to bottom quintile households (Figure 4.17). or relative mobility. The dissimilarity index decomposition Among children aged 12, the participation rates in top quintile for Vietnam shows that household education is a strong households are almost three times higher. Moreover, among determinant of child education when controlling only for wealthy households, participation in extra classes is increasing parental backgrounds (Set 2 variables in Figure 4.12), but its with age, while in the poorest households, participation is role diminishes when controlling for wealth (Set 3 variables). declining with age. These differences in education investment Across countries in the region, a mixed story of progress in create uneven opportunities and realities for future education relative educational mobility is evident. In three of the six attainment, skill development, and access to jobs. At the large countries with estimates for multiple cohorts—Lao upper-secondary levels, the disparity is even greater where the PDR, Timor-Leste, and Indonesia—children’s educational top quintile is spending nearly 10 times more on extra classes achievements are becoming less associated with their parents’ than households in the lowest quintile. education between the 1950s and the 1980s cohorts. In other words, persistence across generations is declining in these Poorer households have more children and even less to countries, and relative mobility is increasing (Figure 4.18). Yet spend on education per child. 92 percent of households in on average, persistence at the bottom is also becoming more the bottom quintile have a child under 16 compared to 72.5 pronounced, as mobility of those in the bottom half into the percent of households in the top quintile. Households at the top quartile is declining in five out of the six countries. The bottom of the distribution also have more children on average share does not exceed 20 percent in any country in EAP for (2.3 in the bottom quintile, compared to 1.3 in the top the 1980s generation. quintile). Richer households not only have more resources, but have fewer children to invest in education, which can lead to even larger gaps to poorer households with more children. Figure 4.16. Expenditure of wealthier families on Figure 4.17. Participation rates in extra compulsory education (primary and lower secondary) classes by household wealth and age 8,000 100 Participation rate in extra classes 7,000 80 Annual VND (thousand) 6,000 5,000 60 4,000 3,000 40 2,000 20 1,000 0 0 Non-resident fee Contributions Parent/class fund Uniform Books Other study materials Extra courses Other Age 8 Age 12 Top quintile Bottom quintile Source: Le and Nguyen (2016). Young Lives Surveys. 1 (=poorest) 2 3 4 5 (=richest) Note: Other study materials include paper, pens, bags, and notebooks. Source: World Bank staff calculation using VHLSS2020. From the Last Mile to the Next Mile 125 PART 2. THE NEXT MILE IS THE ROAD AHEAD Figure 4.18. Persistence and relative mobility by cohort 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 1940 1950 1960 1970 1980 1940 1950 1960 1970 1980 Intergenerational Persistence (IGP) Privilege (Bottom Half to Top Quartile) China Indonesia Lao PDR Malaysia Mongolia Philippines Thailand Vietnam Note: See Annex 4.2 for definition of wider set of indicators. Source: World Bank staff calculation using VHLSS 2010 and 2020 Economic growth is a key driver of income and 4.5. Summary educational mobility but is not sufficient on its own. Examples from individual countries illustrate how relative The education system in Vietnam is on the right path: mobility in education is not an inevitable consequence of a new modern curriculum is being adopted, teachers economic growth (Narayan et al, 2018). In three of these are being retrained, large and targeted investments are countries, relative mobility increased with national income; being made to the poorest areas in the country, and but not in China, Mongolia, or Vietnam. Upward mobility new policies are under way to make lower-secondary from the bottom to the top has increased unambiguously education tuition-free. While general education is with economic growth only in Malaysia. That China continuously improving, household wealth remains a large and Vietnam showed increasing persistence despite rapid determinant of a child’s education outcomes and private economic progress may be in part related to the significant expenditures in education varies substantially between poor changes experienced by these countries in the last half century, and rich families. which led to the early and later cohorts growing up under very different circumstances, including the wars that affected Prior to the pandemic, a range of educational and those born from the 1950s to the 1970s. Promoting mobility socioeconomic inequalities have become more salient, in education is also likely to raise income mobility across especially between the top and the bottom of wealth generations in the region. The evidence that exists for a few quintiles as well as between ethnic minorities and Kinh countries does not indicate a significant disconnect between groups. The gaps in education access and learning outcomes educational and income mobility in the EAP region. Income begin very early in life and accentuate as children move up the mobility in the five EAP countries for which estimates are education ladder. Educational performance varies significantly available appears to line up well with educational mobility in based on socioeconomic backgrounds, and pathways into a cross-country comparison (Narayan and Yang, 2019). quality employment are much less clear for those not going on to tertiary education. This issue is perpetuated by the schooling system which benefits high-performing students while leaving behind those with learning difficulties, at least at the general education level. 126 2022 Vietnam Poverty and Equity Assessment Chapter 4. Early human captial formation and poverty reduction Students from disadvantaged backgrounds are more The COVID-19 pandemic has radically changed the likely to suffer from school failures and dropouts. The lives and education of children in Vietnam, highlighting cause of—and responsibility for—students’ low learning inequalities in education and skills formation. Although outcomes is due to (i) school failures: a deficient and/ in relative terms Vietnam was not hit as hard by COVID-19 or inadequate provision of quality education by schools as other countries by the pandemic thus far, in education, and the educational system, appropriately catering to the Vietnam has had one of the longest school closures in needs of students, especially those from the disadvantaged the world. The lack of face-to-face teaching overall and background, and being placed in low performing schools; and inequality in access to distance learning across socioeconomic (ii) correlated factors that schools or the education systems backgrounds have massively disrupted the education of all can have little or no influence, such as place of residence, child children, but especially those from disadvantaged families poverty, and family environment. (poor, migrants, those with disabilities, lagging ethnic groups). These worse impacts will likely also have longer-term effects on their educational progression and labor market performance, further exacerbating existing inequalities, and limit intergenerational mobility. BOX 4.2. The Association between socioeconomic characteristics and school quality While not explicitly discussed in this chapter, outside of family and child characteristics, the quality of schools are also important determinants to child outcomes. A large literature addresses why children from higher socioeconomic backgrounds have better educational outcomes, focusing predominantly on differences at home and in school. Home factors that seem to influence educational outcomes the most include income, parental education, and early childhood nutrition (Glewwe and Miguel, 2007). Children from lower- income backgrounds, who tend to attend lower quality schools, are less likely to participate in higher education (Yang and Qiu, 2016). Furthermore, high levels of income inequality and poverty can lead to great variability in the allocation of educational resources across households, which may deter children exposed to other inequalities (e.g., health, nutrition) from accessing schooling opportunities (Wongmonta and Glewwe, 2017). In addition to inequalities in household inputs, family socioeconomic status influences the kind of school and classroom environment to which the student has access. A large body of evidence from both high-income and developing countries indicates that access to quality schooling varies by children’s socioeconomic background. A review of the data in and studies on countries in the South Asia region show the same—in some cases achievement for a student in the top income quintile is three to four times that of a student in the bottom quintile (World Bank, 2014). From the Last Mile to the Next Mile 127 PART 2. THE NEXT MILE IS THE ROAD AHEAD BOX 4.3. Decomposing variation in education quality Similar to education completion, education quality can vary by family conditions or circumstances. Nguyen (2006) studied test scores and school quality, and observed that children from poor households in better-off villages (without government financial support as they are in disadvantaged areas) faced higher risk of school dropout. Following the Shapley decomposition techniques described in Chapter 4, the sources of variation in test scores can also be studied. Since only binary indicators can be decomposed, a threshold of high or low scores must be determined. As a starting point, children that score above the 75th percentile are considered to be high math test score achievers. The sources of variation in test scores across different levels of education illustrate how differences in educational resources can compound across the educational cycle for children. The decomposition of the dissimilarity index of test scores shows similar results to other decompositions on the variation of education completion, namely there are large influences from household wealth, ethnicity, parental education, and region, but lesser influence due to a child’s gender. Table B.4.3.1. Decompositions of math test scores PRIMARY (R4, 2019) LOWER SECONDARY (R2, 2018) LOWER SECONDARY (R5, 2020) Dissimilarity index 0.284 0.270 0.257 Shapley decomposition of the D-index HH wealth 23.95 16.744 29.00 Parents’ Education (max) 19.610 18.78 Ethnicity 32.14 26.700 26.31 Gender 4.80 0.278 0.38 Province 38.80 36.660 25.50 Source: World Bank staff calculations using the RISE data. 128 2022 Vietnam Poverty and Equity Assessment Chapter 4. Early human captial formation and poverty reduction 4.6. References Pimhidzai, Obert.2018. Climbing the ladder: poverty reduction and shared prosperity in Vietnam (English). Washington, D.C.: World Baulch, Bob; Vu Hoang Dat, Nguyen Thang. 2012. Do Vietnamese Bank Group. Schools Provide the Right Education for an Industrialising Schütz, Gabriela, Heinrich W. Ursprung, and Ludger Woessmann. Country? Working Paper No. 81. September 2012. 2008. Education Policy and Equality of Opportunity. Kyklos. Bui, Thai Quyen. 2011. “School dropout trends in Vietnam from 1998 Volume 61, Issue 2, page 279–308. to 2006.” Education in Vietnam: 152-170. Singapore: Institute of UNICEF. 2014. Retrieved from https://mics.unicef.org/surveys Southeast Asian Studies. Vo Tri Thanh, and Trinh Quang Long. 2005. Can Vietnam achieve one D’Souza, Ritika; Gatti, Roberta; Kraay, Aart. 2019. A Socioeconomic of its Millennium Development Goals? William Davidson Institute Disaggregation of the World Bank Human Capital Index. Policy Working Papers 776. Ann Arbor, MI: William Davidson Institute at Research Working Paper No. 9020. World Bank, Washington, DC. the University of Michigan. Filmer, Deon, and Lant Pritchett. “The effect of household wealth on Wongmonta S. and P. Glewwe. 2017. An analysis of gender differences in educational attainment: evidence from 35 countries.” Population household education expenditure: the case of Thailand. Education and development review 25, no. 1 (1999): 85-120. Economics. Gill, Indermit S., Revenga, Ana L., Zeballos, Christian. 2016. “Grow, World Bank. 2014. Student Learning in South Asia: Challenges, invest, insure: a game plan to end extreme poverty by 2030 (English)”. Opportunities, and Policy Policy Research working paper; no. WPS 7892. Washington, DC: ------------- 2018. Growing Smarter: Learning & Equitable Development World Bank Group. in East Asia and Pacific. World Bank East Asia and Pacific Regional Glewwe, Paul, Edward A. Miguel. 2007. Chapter 56 The Impact of Report. World Bank Group. Child Health and Nutrition on Education in Less Developed ------------- 2019a. World Development Report 2019: The Changing Countries, Editor(s): T. Paul Schultz, John A. Strauss, Handbook Nature of Work. Washington, DC: World Bank. of Development Economics, Volume 4, Pages 3561–3606. ------------- 2019b. Vietnam. Insights from Disaggregating the Human Kim, Sung Won, Hyunsun Cho, and Lois Y. Kim. «Socioeconomic status Capital Index. and academic outcomes in developing countries: A meta-analysis.» ------------- 2020. Vietnam’s Human Capital: Education Success & Future Review of Educational Research 89, no. 6 (2019): 875-916. Challenges Kraay, Aart. 2018. “Methodology for a World Bank Human Capital ------------- 2021a. A Year Deferred – Early Lessons and Experiences from Index”. World Bank Policy Research Working Paper 8593. COVID-19 in Vietnam. Washington, DC: World Bank. Washington, DC: World Bank. ------------- 2021b. Reducing Poverty Among Ethnic Minorities in Le D. Thuc, Nguyen Thi Thu. 2016. Inequality in educational Vietnam: Priority Action Areas and Lessons. opportunities and outcomes: Evidence from Young Lives data in Vietnam. WHO. 2022a. Retrieved from https://www.who.int/data/gho/data/ themes/topics/sdg-target-6-ensure-availability-and-sustainable- Mbuya, Nkosinathi Vusizihlobo; Atwood, Stephen J.; Phuong, management-of-water-and-sanitation-for-all Huynh Nam.2019. Persistent Malnutrition in Ethnic Minority Communities of Vietnam: Issues and Options for Policy and ------------- 2022b. Retrieved from https://www.who.int/data/nutrition/ Interventions (English). International Development in Focus nlis/info/improved-sanitation-facilities-and-drinking-water-sources Washington, D.C.: World Bank Group. Yang, Juan; Muyuan Qiu. 2016. “The impact of education on income Narayan, Ambar, Roy Van der Weide, Alexandru Cojocaru, Christoph inequality and intergenerational mobility”, China Economic Lakner, Silvia Redaelli, Daniel Gerszon Mahler, Rakesh Gupta N. Review 37:110-125. Ramasubbaiah, and Stefan Thewissen. 2018. Fair Progress? Economic Mobility across Generations around the World. Washington, DC: 4.7. Notes World Bank. Narayan, Ambar, and Judy Yang. 2019. Economic mobility across generations in the developing East Asia and Pacific region. Poverty 35  SES-disaggregated outcomes are computed using middle-income and Equity Notes, September 2019, Number 16. countries to allow for disaggregation. Computations are made using Nguyen Linh Phuong. 2006. “Effects of social class and school same general methodology as the global HCI but are not directly conditions on educational enrollment and achievement of boys and comparable due to differences in data sources. girls in rural Vietnam. International Journal of Educational research. 36  Vol 45, pp. 153-175. See A Year Deferred – Early Lessons and Experiences from COVID-19 in Vietnam (World Bank, 2021a). Permani, R. 2009. “The Role of Education in Economic Growth in East Asia: A Survey”, Asia-Pacific Economic Literature 2(1): 1–20. From the Last Mile to the Next Mile 129 PART 2. THE NEXT MILE IS THE ROAD AHEAD ANNEX 4.1. Chapter 4 figures Figure A.4.1.1. HCI benchmarking, Vietnam vs. peers Probability of survival to age 5 Singapore Malaysia Thailand Brunei Darussalam Survey Upper Middle Income ASEAN average Vietnam Cambodia Philippines Lower Middle Income 0.0 0.2 0.4 0.6 0.8 1.0 Expected years of schooling Harmonized test scores Singapore Singapore Burnei Darussalam Vietnam Philippines Cambodia Vietnam ASEAN average School Thailand Malaysia Malaysia Brunei Darussalam ASEAN average Thailand Upper Middle Income Upper Middle Income Lao PDR Lower Middle Income Lower Middle Income Lao PDR 0 3 6 9 12 15 0 100 200 300 400 500 600 Adult survival rate Children not stunted under 5 Singapore Singapore Brunei Darussalam Thailand Malaysia Upper Middle Income Health Thailand Brunei Darussalam Vietnam Malaysia ASEAN average ASEAN average Upper Middle Income Vietnam Cambodia Lower Middle Income Lao PDR Philippines Philippines Cambodia 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 Source: World Bank, 2020, HCI vintage 2020 130 2022 Vietnam Poverty and Equity Assessment Table A.4.1.1. Shapley decomposition of dissimilarity index BUNDLED     OPPORTUNITIES ENROLLMENT ON-TIME EDUCATION COMPLETION HEALTH (TOTAL OF 11) ACCESS ACCESS ACCESS TO 8 OUT 9 OUT 10 OUT ENROLLED IN SOME LOWER UPPER TO TO   PRIMARY HEALTH OF 11 OF 11 OF 11 SCHOOL PRIMARY SECONDARY SECONDARY FLUSH SAFE INSURANCE TOILET WATER Child gender Set 1: Child (female, male) 2.4 2.5 4.0 10.5 11.2 6.3 22.7 13.8 3.9 1.4 0.9 circumstances (gender, Region 62.9 60.2 42.3 82.0 87.4 93.7 74.5 52.0 61.9 54.6 69.9 region) Urban or rural 34.7 37.3 53.7 7.5 1.4 0.0 2.8 34.2 34.2 44.0 29.2 Kinh or ethnic minority 71.5 51.5 34.1 36.9 15.7 51.3 55.5 55.5 76.6 19.2 64.7 Set 2: Parental HH education 19.6 34.1 59.0 42.1 47.7 24.9 13.1 36.2 18.5 63.3 25.9 circumstances HH gender 4.0 9.2 5.1 7.6 5.6 15.1 11.8 6.9 2.9 13.6 1.0 HH age 5.0 5.2 1.8 13.4 31.0 8.7 19.6 1.4 2.0 3.9 8.5 Kinh or ethnic minority 23.8 19.0 13.7 12.8 8.1 14.1 20.6 21.7 33.2 9.2 52.1 HH education 8.1 13.2 23.2 13.9 28.7 7.9 6.1 17.5 9.8 29.0 13.8 Set 3: Parental circumstances HH gender 1.8 3.8 2.4 3.2 3.7 4.5 5.9 2.7 1.7 6.7 0.4 + wealth HH age 2.2 2.0 1.1 6.4 20.2 3.4 10.9 0.7 1.2 2.1 6.0 HH wealth quintile 64.1 62.1 59.6 63.8 39.4 70.1 56.4 57.4 54.1 53.1 27.8 Kinh or ethnic minority 17.2 13.5 9.9 8.2 5.3 9.0 12.2 15.3 22.5 5.5 21.8 Child gender (female, male) 0.6 0.6 0.9 3.7 2.7 1.5 8.7 5.0 1.2 0.5 0.3 HH education 4.9 8.1 15.6 6.9 18.8 5.0 2.8 10.0 5.7 9.8 7.9 Set 4: All HH gender 1.1 2.2 1.5 1.7 2.6 2.8 2.2 1.7 1.0 2.8 0.4 circumstances HH age 1.7 1.4 1.0 2.9 12.6 2.8 5.5 0.3 0.7 0.9 3.9 HH wealth quintile 43.5 42.9 46.7 36.2 27.0 43.0 31.0 40.5 34.9 18.2 18.2 Region 21.6 21.9 12.2 37.7 30.6 35.4 36.5 18.3 23.5 38.9 33.7 Urban or rural 9.5 9.6 12.2 2.7 0.4 0.6 1.2 9.0 10.5 23.4 13.9 Early human captial formation and poverty reduction Chapter 4. From the Last Mile to the Next Mile 131 Source: World Bank staff calculations using VHLSS2020. PART 2. THE NEXT MILE IS THE ROAD AHEAD ANNEX 4.2. and infrastructure. The list of opportunities we look at are provided in Table A.4.2.3. While the definitions for access to education, assets, and electricity are generally well defined, we refer to standards defined by the United Nations and the Computing Dissimilarity World Health Organization for access to improved drinking index – Circumstances water and sanitation. and opportunities Opportunities should be exogenous, not determined by the individual or their prior actions. For that reason, Circumstances – factors that can influence a we restrict the main analysis to children aged 6–14. This child’s opportunities is equivalent to the age range from the start of primary education to the completion of lower-secondary education. A set of circumstance variables are selected and defined While it is difficult to ascertain why an individual is in a certain as the factors that contribute to the disparity of access circumstance, the assumption that children are where they are to opportunities. Variation in children’s outcomes are largely due to the birth lottery is still plausible. attributable to their circumstances. In this analysis, three broad groupings of circumstances are considered: the characteristics Access to an opportunity is equated with the opportunity of the children themselves, the characteristic of their itself for the purposes of this analysis. While it can be households, and their geographic location (Table A.4.2.1). debated whether opportunity should be defined as the access to or utilization of the opportunity, due to the data we base A child’s opportunities this analysis upon the assumption of access-as-opportunity is made. We thus define the use of clean, piped water as A child’s opportunities are defined as access to basic the opportunity for clean water, while the availability of needs and infrastructure deemed necessary during clean, piped water elsewhere in the village is not considered formative years to the fulfillment of a child’s potential. as opportunity as long as the household still uses collected This generally covers three main realms: health, education, rainwater. Table A.4.2.1. List of circumstances (potential sources of inequality) CATEGORY CIRCUMSTANCE DESCRIPTION Geographic Region There are six geographic regions defined in the VHLSS: Red River Delta Midlands and Northern Mountains Northern and Coastal Central Central Highlands Southeast Mekong Delta Urban/Rural Binary indicator for being located in an urban or rural area Child Characteristics Gender Binary indicator for gender Ethnicity Binary indicator for being of the Kinh ethnic majority Household Gender of Head Binary indicator for gender of household head, defined as being male Characteristics Education of Head Binary indicator for education of household head, defined for having completed lower secondary Age of Head Binary indicator for household head age, defined as being below median Household consumption quintile Quintiles based on the household consumption aggregate in nominal LCU terms. 132 2022 Vietnam Poverty and Equity Assessment Chapter 4. Early human captial formation and poverty reduction The definition of opportunity considers only the Table A.4.2.2. Vietnam’s General Education System utilization of an opportunity and does not take into INTENDED AGE account the quality of the opportunity being utilized. LEVEL GRADES RANGE This is due to data constraints, as the data we rely on cannot Primary Grades 1–5  6–11 measure or distinguish high quality opportunities from low quality ones. Secondary Grades 6–9 (Lower Secondary) 11–15 Grades 10–12 (Upper Secondary) 15–18 For the purposes of this analysis, some assumptions need to be made in order to harmonize the general and vocational educational streams. In the education module A minimum bundle of goods of the VHLSS, there are two primary questions on level of education: highest qualification earned (question 2), and It is possible to define an opportunity as a minimum current grade attending (question 6). Question 2 is divided bundle of goods and services. From Table A.4.2.3, ten into two parts: 2a asking for the highest general education opportunities were chosen to construct the bundle. These certificate earned, and 2b asking for the highest vocational include school enrollment, on-time education completion, education certificate earned. As such, one can roughly discern access to safe drinking water, improved sanitation, electricity, from question 2 the trajectory of an individual’s education, mobile, internet, health insurance, and a permanent housing provided that there was only one shift between general structure and a non-overcrowded dwelling. Choosing education and vocational educational streams. However, overlapping deprivations may involve some subjective choice when thinking about the level of education that has been of what is included in the bundle. The indicators should be as achieved in itself, or that being attended by an individual, one different from each other as possible to create a combination must categorize the different vocational certificates into the as opposed to interrelated goods and services. This bundle buckets of lower-secondary/upper-secondary/post-secondary. represents the minimum complementary opportunities that This gets tricky because there is no specification on what type children need with consideration to data availability, but also of certificate that the individual is in the process of earning or aligns with dimensions of the Multidimensional Poverty Index has earned, adding to the complexity that is the incongruence for Vietnam and OPHI’s MPI. The weighting of different of the general and vocational education streams. As such, for indicators is also another consideration that can be somewhat the purposes of the analysis, we group vocational education arbitrary. All indicators are set to have equal weights to one into one with upper-secondary education. As such, for the another. Taking this approach avoids imposing a hierarchy of analysis of access to educational opportunities, we restrict the importance, which seems to be the most intuitive. sample to between 6 and 21 years of age instead of 6 and 14 as in the other parts of the analysis. Additionally, certain opportunities cannot be measured from the data, such as access to healthcare facilities. The education system in Vietnam consists of compulsory While access to healthcare facilities is covered in the survey five years of primary and four years of lower-secondary data used, it is only asked when individuals fall ill. As such, it education followed by further levels of education, upper- would be impossible to disentangle the fact that individuals secondary and beyond. Starting from lower-secondary, got sick (possibly indicating not having the opportunity of students have a choice of general or vocational education a healthy constitution) and the opportunity of having the streams and are free to switch between general and vocational access to a hospital if the individual fell ill. streams, as illustrated in Table A.4.2.2. From the Last Mile to the Next Mile 133 PART 2. THE NEXT MILE IS THE ROAD AHEAD Table A.4.2.3. List of opportunities CATEGORY OPPORTUNITY DESCRIPTION VHLSS data Education Attending School Children in age window 6-14 Completing education on time Some Primary Age window: 6-11 Primary Age window: 12-15 Lower-Secondary Age window: 16-18 Upper-Secondary Age window: 19–21 Access to Infrastructure Housing Adequate living Space Square-meters of living space per capita is more than 8m2 Housing quality Floor AND wall are permanent/durable materials Improved Drinking Water Sources of water that are protected from outside contamination, in particular from fecal matter. Improved Sanitation Sanitation facilities that are covered Electricity Household has access to electricity Internet Household has access to internet (at home or internet café) Healthcare Health Insurance All members of the household have health insurance Access to Assets Phone Household owns mobile or land line phone Computer Household owns computer Bundled Opportunities Not Vulnerable Opportunities can also be bundled. A child is not vulnerable if he or she has joint access to a minimum (8, 9, 10 out of 11) number opportunities: school attending, on-time education completion, improved drinking water, improved sanitation, electricity, internet, mobile, health, housing. See the section Complementarity of opportunities for analysis of bundled opportunities. RISE Data Education quality Math and reading test scores High above 75th percentile. Note: HOI analysis is calculated for children 6-14 years of age. Exceptions are the individual opportunities of on-time completion of lower-secondary education. 134 2022 Vietnam Poverty and Equity Assessment PART 2. Chapter 5. Sustaining upward economic mobility Chapter 5. Sustaining upward economic mobility Key Messages • Economic growth has transformed the country but preserving these gains and creating new sustainable economic pathways to build a secure middle class is the challenge ahead. • Higher labor productivity and high skilled led growth is necessary for HIC transition as well as generating upwards economic mobility • The labor market will have to recognize and address challenges of high informality, slow growth of high-skill and digital drops, and slow transition into higher skill occupations among the youth. From the Last Mile to the Next Mile 135 PART 2. THE NEXT MILE IS THE ROAD AHEAD 5.1. What will it take to 4 percent but increased in the latter half of the decade build a middle-class? (Figure 5.1). In the latter half of the decade, mirroring stronger labor market transitions out of agriculture and Vietnam needs a productivity-led growth path to into wage jobs as discussed in Chapter 2, labor productivity reach higher levels of prosperity. The growth strategies growth increased to nearly 6 percent. However, the sector that underpinned Vietnam’s poverty reduction success in with the highest productivity growth was agriculture, which the previous decade are no longer ones that can sustain the is now facing an aging workforce and will need to modernize country’s growth path to higher income levels and the build- to sustain productivity growth rates. Despite increasing labor up of a large and secure middle class. Even before the onset of productivity, the services sector remains highly informal and COVID-19, the efficacy of the developing East Asia and Pacific is susceptible to economic shocks, as witnessed by adverse region’s growth model based on low-skilled outward-oriented employment shocks from COVID-19. The labor productivity growth was expected to diminish as the world rapidly changes patterns of the manufacturing sector has varied inconsistently. (Mason and Shetty, 2019). Given these emerging challenges, Vietnam may not be as well positioned as it was a decade ago To increase labor productivity, productivity in the when it reaped advantages from a demographic dividend, low services and industry sectors must increase. In 2002, labor costs, and a structural transition of the labor force based industry was Vietnam’s smallest but most productive sector, on low-skill occupations and out of agriculture. making up just over 10 percent of employment but with productivity nearly three times the national total. It was Productivity growth had been a major driver of GDP followed by services with 20 percent employment share and growth in the early years of Vietnam’s economic around 1.75 times total productivity; agriculture formed two- transition following market liberalization. Labor thirds of employment but was less than half total productivity productivity growth between 1990 and 2000 was driven by (Figure 5.2). By 2030, 70 percent of employment will be in intra-industry improvements as production processes began the two more productive sectors but their advantage over total to modernize, new technologies were adopted, and less productivity no more than 1.4 times. Thus, total productivity productive firms were forced out due to competitive market over the last two decades has grown because low productivity forces. Post-2000, productivity growth emerged from inter- agriculture has shrunk. However, this means that, for total industry improvements, with workers moving from low- productivity to increase in the future, it is not enough for productivity agriculture to more productive manufacturing services and industry sectors to increase employment share; and services sectors (Cunningham et al., 2018). In the early they must also increase sectoral productivity. half of the decade, labor productivity growth was below Figure 5.1. Average labor productivity growth (GDP per worker), by sector 10 Annualized growth rate (%) 8 6 4 2 0 −2 2005–2009 2010–2014 2015–2020 Total Agriculture, forestry and fishing Industry and construction Service Source: World Bank staff calculations using GSO and WDI data. 136 2022 Vietnam Poverty and Equity Assessment Chapter 5. Sustaining upward economic mobility Figure 5.2. Employment shares and labor productivity by sector, 2002-2030 2002 2018 2030 (p) 300 300 300 Sectoral productivity % of total productivity Sectoral productivity % of total productivity Sectoral productivity % of total productivity 250 250 250 200 200 200 150 150 150 100 100 100 50 50 50 0 0 0 0 0 0 10 20 30 40 50 60 70 80 90 100 10 20 30 40 50 60 70 80 90 100 10 20 30 40 50 60 70 80 90 100 Share of total employment (%) Share of total employment (%) Share of total employment (%) Agriculture Industry Services Note: (p) = predicted. Source: World Bank, 2009. Figure 5.3. Projected growth (baseline and target Figure 5.4. Projected income, 2018–2045 scenario), 1991–2045 7 16,000 6.4% 5.9% 6 5.5% 0.3% 14,000 HIC 1.0% 5.1% 5.3% 5 1.0% 12,000 GDP per capita (2010 USD) 0.1% 4 10,000 Growth rate 3 8,000 5.0% 4.4% 5.3% 5.3% 6.6% 2 6,000 UMIC 1 4,000 0 -0.1% 2,000 −0.1% -0.2% -0.2% -0.2% LMIC −1 0 1991–2002 2002–2012 2012–2018 2018–2045 2018–2045 2018 2045 Baseline Reaching HIC target Productivity Employment Participation GDP - Baseline GDP - HIC Demographics GDP per capita Note: Both projected periods use UN projected total population and working-age population and hold employment and labor force participation constant. Baseline uses 2012–2018 average worker productivity growth; HIC uses the growth required to achieve high-income status by 2045. See Annex 5.2 for methodology. Income levels are converted from GNI per capita to GDP per capita based on the 2018 ratio of the two in Vietnam. Source: World Bank staff calculations. From the Last Mile to the Next Mile 137 PART 2. THE NEXT MILE IS THE ROAD AHEAD For Vietnam to achieve high-income status by 2045, 5.2. Economic mobility annual productivity growth of nearly 7 percent would be required. If participation and employment rates stay the for tomorrow – potential same as from 2012 to 2018 and the population continues to challenges ahead age as projected, then the number of workers will fall and the productivity of those who do work will need to increase to Since the postwar era, Vietnam has successfully used sustain or even increase growth. Average real per capita growth promotive strategies through job creation to improve from 1991 to 2018 was 5.6 percent per annum. A 6.7 percent living standards and reduce poverty. While medium- rate would be required until 2045 to reach high-income skilled manufacturing jobs created economic pathways off status. To achieve this, growth in productivity per worker the farm and out of poverty for many, incomes from these would need to increase from the 2012–2018 annual rate of 5.3 jobs are still too low to support the build up of a middle percent—already the highest of the last three decades—to 6.6 class. Future labor market challenges have also been looming percent per annum, an acceleration of around 20 percent each on the horizon in regard to the viability of a continuously year (Figure 5.3). At the 2012–2018 rate, Vietnam would be transforming the labor market towards more skilled and comfortably upper-middle income but US$4,000 per capita sophisticated activities (Cunningham et al., 2018, Pimhidzai short of high-income status (Figure 5.4). and Niu, 2020). Moreover, reports with a global perspective have described the need for developing economies to update A high-skill labor transition will help boost upwards their growth strategies involving the manufacturing sector economic mobility. Better jobs will not only boost economic as well as address informality (World Bank, 2017; Ohnsorge growth, it will further improve household living standards and and Yu, 2021; Mason and Shetty, 2019). This section reviews conditions. Household consumption varies by the occupation challenges for the Vietnam context including low education skill level of the household head (Figure 5.5). Households at higher grade levels, high labor informality, low labor where the head is in a high-skilled and non-manual occupation productivity, low skills, and an aging workforce, which this is nearly three times larger than households where the head section summarizes and reviews. has an unskilled occupation. An aging population means a smaller labor force in the future. Figure 5.5. Household consumption per capita by household head occupation skill level The demographic dividend is nearly spent. The working age share of the population is projected to have begun 100 declining in 2018, just as the share of the elderly population Annual household consumption is beginning to increase. While 6.7 percent of the population per capital (VND, millions) 80 was aged 65 years or above in 2017, this is expected to reach 60 21 percent by 2050, transforming Vietnam from a young to an aged society. Moreover, Vietnam is projected to become 40 old before it becomes rich, by aging at levels that are much 20 higher than other lower-middle-income developing countries (Eckardt et al, 2016). While Thailand and China are older in 0 the region, these two countries are also much richer and may 2010 2012 2014 2016 2018 2020 be better able to afford care for an aging society (Figure 5.6). High-skilled non-manual Low-skilled non-manual Moreover, social assistance programs are also larger in these Skilled manual Unskilled two countries. An aging society increases the dependency ratio, reduces the eligible working-age population, and Note: Household consumption in annual per capita nominal VND. increases the demands on social services. Source: World Bank staff calculations using VHLSS 2010-2020. 138 2022 Vietnam Poverty and Equity Assessment Chapter 5. Sustaining upward economic mobility Figure 5.6. At its level of development, Vietnam is older than its peers 14 14 THA Share of total population over 65 (%) Share of total population over 65 (%) THA CHN 2019 2020 12 12 10 10 VNM VNM 2020 8 8 MYS 2019 6 6 IDN 2019 MNG 4 4 2019 2 2 1960 1970 1980 1990 2000 2010 2020 0K 2K 4K 6K 8K 10K 12K Year GDP per capita constant (2010 US$) Cambodia Indonesia Malaysia Philippines Vietnam China Lao PDR Mongolia Thailand Source: WDI. The combination of an increasing share of elderly 2018 if the demographic dividend had not been lost (Figure and a declining share of working-age people—which 5.8 and Annex 6.2).37 This means that Vietnam will need to is projected to accelerate (Figure 5.7)—has several move its relatively declining workforce into more knowledge- implications for jobs and growth. The decline in the intensive activities that allow for high labor productivity working-age population means that a key driver of Vietnam’s to sustain robust growth and upward economic mobility rapid growth will diminish, making other sources even more (Cunningham et al., 2018; World Bank & MPI, 2016). vital; growth would have been 1.2 points higher from 2012 to Figure 5.7. Potential workers, 1991–2050 Figure 5.8. Growth decomposition (varying demographic scenarios), 1991–2018 100% 7 6 5 80% 4 3 2 60% 1 0 40% −1 1991 2002 2012 2018 2040 2050 1991– 2002– 2012– 1991– 2002– 2012– 2002 2012 2018 2002 2012 2018 Working age Participation Employment (actual) (alternate) Productivity Employment Participation Demographics GDP per capita Source: WDI (working age 1991–2018, participation, employment); UN Projections (working age 2018 until 2050). Note: 2012–2018 Alt. assumes all growth components the same as actuals for 2012–2018 except for Demographics (working age population) which are equal to the 2002–2012 value. See Annex 5.2 for methodology. Source: WDI and World Bank calculations. From the Last Mile to the Next Mile 139 PART 2. THE NEXT MILE IS THE ROAD AHEAD Most of Vietnam’s growth components are exogenous likely to offer formal work (Figure 5.10). Wage employment in or maximized. An aging population is given while the services sector is more likely to be informal and is comprised unemployment is near-zero and participation rates among the of more low-skilled activities than manufacturing. A significant working age population are some of the highest in the world. share of workers working in family farms or household Even in a country with strong family care responsibilities, enterprises, meaning they do not have access to social or health female labor force participation is 73 percent in Vietnam insurance through their employment, as small enterprises are compared to the 59.4 percent average in the East Asia and not required by law to provide social benefits. Pacific region. Among females aged 20–39, only 6 percent are out of the labor force due to family care responsibilities Older, rural, and ethnic minority demographics are often (Figure 5.9). With an aging population, fewer new workers are excluded from formal wage jobs due to educational coming into the labor force while existing workers are leaving requirements. Older workers and those from rural areas have at a greater rate. Among working adults, labor participation lower educational attainment, work primarily in household- cannot be significantly increased; of those participating, almost based production, and are trapped in low-quality jobs. Over all have already found employment. Thus, the only potential half of all workers aged 50 and over and three-quarters of ethnic source of growth that can increase is worker productivity and minorities are engaged in farming. The ILO categorizes 52 percent to have jobs in the labor market that allows every worker to of Vietnamese workers in 2019 as being engaged in “vulnerable work to their full potential and not be underutilized. employment,” that is, family or own-account workers. COVID-19 highlighted the large vulnerabilities of the Figure 5.9. Reasons for remaining out of the labor informal and labor migrant workforce (World Bank, force, by gender and cohort 2021a). Informal workers are more concentrated in lower- 8 skilled services sectors, which were also more affected by social 7 distancing restrictions. Informal workers had few coping Population (millions) 6 mechanisms, safety nets, or formal assistance with job search. 5 Most informal workers (64 percent) who experienced declines 4 in employment or income could only respond to labor shocks 3 by waiting until social distancing ended (GIZ, 2021). 2 1 Most jobs today are either low-skilled or manual. 0 0–9 10–19 20–29 30–39 40–49 50–59 60–69 70+ 0–9 10–19 20–29 30–39 40–49 50–59 60–69 70+ Manual but skilled jobs are concentrated in Female Male manufacturing, and these are also the types of jobs that Disabled/ill Homemaker Retired have increased the most over the last decade (Figure Students Other In the labor force 5.11). From 2010 to 2020, the composition of jobs has shifted Note: This variable is not available in the VHLSS 2020. away from agriculture to occupations in manufacturing, Source: World Bank staff calculations using 2018 VHLSS. non-trade services, or the trade and construction sectors. Ten million new jobs were created in the manufacturing and services sectors combined between 2010 and 2020, yet Job informality is still high with implications growth in the manufacturing and services sectors has yet to be towards income instability and access to social accompanied by complementary growth in high-skilled non- insurance. manual occupations. High-skilled non-manual jobs are most numerous in public-service sectors such as education and Informality remains widespread; in 2020, only 23.5 public administration. The financial sector is more private percent of households have an employed household sector and high-skill oriented but employs a smaller share of member with a formal wage contract. Outside of the public workers. The current analysis uses occupation as a proxy for skill; sector, the manufacturing sector provides the most contract jobs there is likely unobserved within-sector heterogeneity which due to the prevalence of foreign-owned firms, which are more further skills surveys could better capture and differentiate. 140 2022 Vietnam Poverty and Equity Assessment Chapter 5. Sustaining upward economic mobility Figure 5.10. Sectors having jobs with contracts remain concentrated in the public sector and manufacturing (2010–2020) 25 Number of workers (millions) 20 15 10 5 0 2010 2012 2014 2016 2018 2020 2010 2012 2014 2016 2018 2020 2010 2012 2014 2016 2018 2020 2010 2012 2014 2016 2018 2020 2010 2012 2014 2016 2018 2020 2010 2012 2014 2016 2018 2020 2010 2012 2014 2016 2018 2020 2010 2012 2014 2016 2018 2020 2010 2012 2014 2016 2018 2020 2010 2012 2014 2016 2018 2020 2010 2014 2016 2018 2020 Agriculture Mining and Manufacturing Electricity, Construction Commerce Education, Transport, Financial, Public admin Other and fishing quarrying gas and water health and storage, and insurance and social work social work real estate Has contract No contract Source: World Bank staff calculations using VHLSS. Figure 5.11. Skill level of jobs, 2010 and 2020 Number of workers by skill level of jobs in 2010 25 Number of workers (millions) 20 15 10 5 0 Agriculture Mining and Manufacturing Electricity, Construction Commerce Transport, Education, Financial, Public Other and fishing quarrying gas and storage and health, Insurance, admin water communication social work real estate Number of workers by skill level of jobs in 2020 25 Number of workers (millions) 20 15 10 5 0 Agriculture Mining and Manufacturing Electricity, Construction Commerce Transport, Education, Financial, Public Other and fishing quarrying gas and storage and health, Insurance, admin water communication social work real estate High-skilled non-manual Low-skilled non manual Skilled manual Unskilled manual Note: See Table A.5.1.1 for skill categorization. Source: World Bank calculations using LFS 2010 and 2020. From the Last Mile to the Next Mile 141 PART 2. THE NEXT MILE IS THE ROAD AHEAD Building broad economic security and a consumer class high-skill in 2009 compared to 11.9 percent in 2020. With is no longer just about moving people to wage jobs, but higher education, workers are for the most part shifting into increasingly more about improving the quality of those medium-skilled manufacturing occupations rather than low- jobs to sustain wage income growth. Gains in skilled or skill occupations like earlier generations. While the growth of non-manual employment are mostly seen among households the high-skill occupations may be expected to be slower, one in the top quintile. While the absolute number of skilled aspect that points to future challenges is that younger workers manual jobs nearly doubled in the lowest quintile, the size are entering manufacturing and medium-skill occupations at of the unskilled labor force did not significantly decrease. higher rates than in neighboring economies. It is too early to Modern jobs will need to shift toward higher-value added see if structural upgrading into higher-skilled service jobs will activities to improve job quality, drawing from the global occur at the same pace. shift toward knowledge-intensive production processes and complex value chains. This is needed if Vietnam wants to Despite higher average education in the population, move toward the service segments, and more sophisticated, those with tertiary education are more likely to be higher value chains. engaged in medium-skilled jobs in 2020 than in 2010 (Figure 5.14). 86 percent of workers with tertiary workers Vietnam has a high share of low digital occupations in 2010 worked in high-skill occupations compared to 74 compared to nearby economies. Figure 5.12 illustrates the percent in 2020. The slow growth of high-skilled jobs is distribution of jobs as classified by the intensity of digital skills due to several factors. Vietnam’s ability to move into higher across a set of Southeast Asia economies. About 74 percent value-added jobs is hindered by existing education and skills of occupations in Vietnam are classified as having low digital systems which does not address priority skills, along with a skill needs, more jobs than in neighboring countries where shortage among higher-skill occupations. Young workers are comparable data was available. better educated but, due to a mix of supply- and demand-side factors, are not finding or being hired for better jobs. Youths The growth of high-skilled jobs has been gradual with the right skill sets may be discouraged by low domestic despite a more educated workforce. wages. Existing older workers have less education and larger skill gaps. While Vietnam’s current education system has Based on official surveys, the size of the labor force in improved significantly, most workers do not benefit from high-skill occupations has been marginally increasing, these improvements, having long since left the education but not as quickly as the size of the medium-skilled system. At the same time, enrollment in tertiary education is workforce (Figure 5.13). 9.1 percent of occupations were stagnating (World Bank, 2020b). Figure 5.12. Employment across digital occupation level, by country 0.3% 3% 1.4% 1.4% 100 28.6% 33.4% 26.1% 24.7% Share of employment (%) 80 60 40 71.1% 63.6% 72.5% 73.9% 20 0 Cambodia Thailand Malaysia Vietnam Low digital occupations Medium digital occupations Highly digital occupations Source: World Bank 2021b using SEAD data. 142 2022 Vietnam Poverty and Equity Assessment Chapter 5. Sustaining upward economic mobility Figure 5.13. Employment by instance, behavioral skills appear to be more important in the occupation skill category (share) services sector rather than the manufacturing sector, which 60 values problem-solving skills more. 50 Share of employment (%) Employers in Vietnam report difficulty in finding skilled 40 labor, sometimes at rates higher than in the region. 30 Based on the World Bank’s Enterprise Survey on Innovation and Skills in Vietnam conducted in 2019, 22 percent of 20 managers reported that the biggest obstacle faced by the firm 10 was an adequately educated workforce (World Bank, 2021b). 0 This rate is nearly double the rate from a similar survey conducted in 201538. Nearly half of employers surveyed in 2009 2010 2011 2012 2013 Prel. 2020 2014 2015 2016 2017 2018 2019 the Skills Toward Employment and Productivity (STEP) High-skill Medium-skill Low-skill surveys in Vietnam indicated that graduates did not have the skills needed in their workplace (Bodewig et al., 2014). Labor Note: Occupations are grouped into broad skill categories defined by the ILO. force surveys suggest that firms often prefer workers with Source: GSO and World Bank calculations. technical and vocational education and training skills, which may undermine student interest in university education. Occupational mismatch can stem from both supply and Employers identify a lack of adequately skilled demand factors. About 11 percent of young workers in 2016 workers as a constraint were overqualified, but an even larger share, 33 percent, was under qualified (Nguyen and Kenichi, 2018). While a wage Employers and employees in East Asia particularly premium for tertiary-educated workers may reflect their emphasize job-specific technical skills and critical higher productivity, because employers are most concerned thinking and behavioral skills (World Bank, 2019). In about skills gaps in those occupations that university workers general, creativity and IT skills tend to have a higher relative hold (technical, professional, and managerial professions), importance within UMICs, whereas practical knowledge higher returns most likely reflect a shortage of those with is particularly crucial in LMICs (based on firm surveys in (even limited) university education (Cunningham and Indonesia, the Philippines, and Vietnam). Despite these Pimhidzai, 2018). Recent analysis also shows returns to non- overarching trends, the relative importance of generic skills routine analytical tasks are increasing over time in Mongolia, varies by sector, trade orientation, and foreign ownership. For the Philippines, and Vietnam (Macdonald, 2018). Figure 5.14. Share of workers by skill category and education in 2010-2020 100 80 Share of workers (%) 60 40 20 0 2010 2015 2020 2010 2015 2020 2010 2015 2020 2010 2015 2020 2010 2015 2020 Primary Lower secondary Upper-secondary Vocational Tertiary High-skill Medium-skill Low-skill Source: World Bank calculations using LFS. From the Last Mile to the Next Mile 143 PART 2. THE NEXT MILE IS THE ROAD AHEAD Increasing the demand for high-skilled jobs 5.3. Is the youth labor will require a more competitive and innovative private sector force transforming? Rapid technological change has tightened the link Many firms in Vietnam today are small and uncompetitive. between human capital and economic output. Previous Compared to other countries in the region, firms in Vietnam sections illustrated the need for higher labor productivity, are slower to shift to complex production, are less innovative, but also some challenges to increasing productivity to reach and have a lower number of patents (World Bank, 2021b). upper-middle and high-income targets and preserve patterns Private domestic firms are less productive than foreign-owned of upward economic mobility. Today’s youth in Vietnam firms. The labor productivity of foreign-owned enterprises is are better educated, but will they lead a high-skill and high- almost five times higher than that of private domestic firms productivity labor transition? To further examine the potential (World Bank, 2020a). Foreign-owned firms dominate the direction of labor market transitions in the near future, the export sector, with private domestic firms unable to compete. next section describes the characteristics, outcomes, and Only 17 percent of domestic firms are in the export sector circumstances of the youth today. A few stylized statistics raise (Eckhardt et al., 2016). This presents a lost opportunity for doubts that a high-skill transition will occur without policy domestic firms to gain from technology transfer and better interventions related to improving education and skills linkages through exposure to foreign buyers. Younger cohorts are primarily engaged in Domestic private firms are mostly small and less medium-skill occupations. productive as they lack the scale to access technology and invest in fixed capital. The non-farm business sector Is the more educated youth cohort working in better jobs? mostly consists of households and small enterprises (World Vietnam’s population is getting more educated. As discussed in Bank, 2020a). In 2017, there were over 5.7 million non- Chapter 1, the transformation of households with increasing farm businesses in the country. Nearly 98 percent of these education levels was one of the most dramatic changes in firms operate in the informal sector (World Bank, 2017). profiles over the last decade. The largest improvements in The median firm, including household businesses, has three education are among the younger cohorts. In 2010, 8.4 workers and most likely operates in traditional sectors such percent of 30–39 year-olds had completed tertiary education, as construction, repairs, and food preparation. Small firms in compared to 17.5 percent in 2020. Among 20–29 year-olds Vietnam underinvest in physical capital and are slow to adopt in 2010 and 2020, a large improvement in tertiary education technology. The quality of investment is also low, reflecting completion was also observed. However, a comparison across inefficiencies in capital allocation. As a result, the median cohorts shows that higher educated individuals in their 20s domestic firm is lagging in productivity compared to countries in 2010 were more concentrated in professional jobs than with similar income per capita. These firms face difficulties 20 year-olds in 2020 (Figure 5.15). Participation in high- in expanding due to lack of scale, access to technology, and skilled occupations among tertiary-educated workers is not competitive pressure to boost productivity (Eckhardt et al., increasing as rapidly among the younger cohort. Among the 2016). Barriers to increasing productivity for small firms also population that has completed tertiary education, the share include low innovation capacity. Investment in research and engaged in low-skilled and non-manual work is increasing. development (R&D) in Vietnam amounts to 0.4 percent of This is mainly driven by younger cohorts working in more GDP, compared with 2.1 percent in China and 1.3 percent medium-skill occupations despite achieving higher education in Malaysia. completion. The comparison illustrates that recent cohorts appear more engaged in medium-skill occupations and less so in high-skill occupations, despite being more educated. 144 2022 Vietnam Poverty and Equity Assessment Chapter 5. Sustaining upward economic mobility Figure 5.15. Distribution of occupations among tertiary education workers, by skill and age groups 100 Share of workers with tertiary 80 education (%) 60 40 20 0 2010 2012 2014 2016 2018 2020 2010 2012 2014 2016 2018 2020 2010 2012 2014 2016 2018 2020 2010 2012 2014 2016 2018 2020 Age 20–29 Age 30–39 Age 40–49 Age 50–59 High-skill non-manual Low-skilled non-manual Skilled manual Unskilled manual Note: See Table A.6.1.1. in Annex for occupation skill categorization. Share of active workforce with completed tertiary education. Source: World Bank calcualtions using LFS. Occupation choice has shifted across cohorts. For example, they entered the workforce, they entered jobs in the largest older cohorts are still primarily engaged in elementary numbers in manual and routine jobs. Due to a lack of detail occupations in agriculture reflective of much lower education in household and labor force surveys, there is ambiguity in the levels (Figure 5.16). The cohort born in the 1980s and 1990s types of activities being conducted across occupations and are most likely to be working as plant and machine operators the level of within-occupation heterogeneity. These details and assemblers, consistent with anecdotal evidence of a are important for better understanding the future of skills preference for younger factory workers. The 1990s cohort in Vietnam. was still in school at the start of the decade in 2010, and as Figure 5.16. Number of workers, by occupation and birth cohort 15 12 Workers (millions) 9 6 3 0 2010 2012 2014 2016 2018 2020 2010 2012 2014 2016 2018 2020 2010 2012 2014 2016 2018 2020 2010 2012 2014 2016 2018 2020 2010 2012 2014 2016 2018 2020 2010 2012 2014 2016 2018 2020 1940–49 1950–59 1960–69 1970–79 1980–89 1990–99 Elementary occupations Skilled agricultural and fishery workers Plant and machine operators Craft and related trade workers Service workers, shop market sales Clerks Technicians and associate Professionals Managers professionals Source: World Bank staff calculations using LFS. From the Last Mile to the Next Mile 145 PART 2. THE NEXT MILE IS THE ROAD AHEAD Compared to youths in other countries, analytical and interpersonal skills is lower than in neighboring Vietnamese youth are more engaged in routine countries. The current profile of occupations may become and manual jobs. barriers to development of a high-skill services sector. Occupation skill intensity illustrates higher manual and Underutilized human capital? routine work among Vietnamese youth compared to other youths in the region. O-Net skills scores can be used The Utilization-adjusted Human Capital Index (U-HCI) to measure the skill intensity of occupations, by whether jobs also notes a disconnect between human capital potential employ manual, analytical, cognitive, or interpersonal skills, and its utilization in the labor market. Vietnam’s Human and whether tasks are routine or non-routine (Autor and Capital Index (HCI) as described in Chapter 4 indicates a high Handel, 2013). Based on these categorizations, a profile of job level of human capital potential in Vietnam compared to its sophistication can be illustrated. In Vietnam, younger cohorts LMIC peers or the East Asia and Pacific regional average (Figure (born after 1977) work in jobs with higher skill intensities in 5.19). The index is a measure of potential, but utilization routine manual, and non-routine manual physical tasks than of that potential in the labor market is not guaranteed. The youths in other countries (Figure 5.17). interpretation of Vietnam’s HCI is that a child born today will be 69 percent as productive when she grows up as she These skill intensity trends are consistent with the would be if attaining complete education and health. The higher concentration of younger cohorts in the utilization-adjusted HCI considers the share of employment manufacturing sectors and slower movement into high- in the population and the quality of jobs available to highlight skilled services. According to the sector of occupation, the degree of labor-market underutilization of human capital younger cohorts in Vietnam aged 15–24 are more engaged in in each country, and suggests that realized productivity in the manufacturing than younger cohorts elsewhere (Figure 5.18). labor market is much lower (Pennings, 2020). Vietnamese youth are more engaged in services than youths in Cambodia or Myanmar, but less so than younger cohorts in Considering the utilization of human capital in higher- Indonesia, Philippines, or Thailand. While Vietnamese youth quality jobs, the U-HCI reflects the limits of Vietnamese are moving up the skill ladder in the sense of engaging less in youth in utilizing their human capital potential in the manual and routine jobs, their engagement in jobs requiring workforce. Vietnam’s full U-HCI stands at 0.37, a reduction Figure 5.17. Occupation skill intensities, by age cohort Non-routine Non-routine Non-routine Routine Routine analytical interpersonal manual physical cognitive manual 4.0 3.5 Skill intensity 3.0 2.5 2.0 2012 2012 2012 2012 2012 2002 2006 2008 2010 2014 2016 2018 2002 2006 2008 2010 2014 2016 2018 2002 2006 2008 2010 2014 2016 2018 2002 2006 2008 2010 2014 2016 2018 2002 2006 2008 2010 2014 2016 2018 2004 2004 2004 2004 2004 2000 2000 2000 2000 2000 IDN MNG PHL THA VNM Notes: Based on O-NET occupation skill scores. Cohort born after 1977. Source: World Bank calculations. 146 2022 Vietnam Poverty and Equity Assessment Chapter 5. Sustaining upward economic mobility Figure 5.18. Concentration of manufacturing jobs compared to other countries in the region (15–24-year-old cohort) 20 15 Workers (millions) 10 5 0 2020 2013 2011 2012 2013 2015 2016 2012 2013 2015 2015 2018 2019 2010 2011 2012 2013 2015 2016 2018 2019 2015 2016 2017 2018 2019 2020 2010 2011 2012 2013 2015 2016 2017 2018 2019 2020 2017 2017 2017 2014 2014 2014 2014 2014 2010 Cambodia Indonesia Myanmar Philippines Thailand Vietnam Services Manufacturing Agriculture Source: ILO. Figure 5.19. Comparison of HCI and full of nearly 50 percent from the unadjusted HCI. In its income utilization-adjusted HCI, 2020 group, Vietnam has the largest absolute decline due to the full utilization adjustment and is about average in its income 0.8 0.7 group when the decline is measured in percent terms (Figure 0.6 5.20). After this correction, Vietnam’s full U-HCI is slightly 0.5 lower than the EAP regional average, although still higher 0.4 than the LMIC average. 0.3 0.2 Underutilization of human capital reflects a situation 0.1 whereby a person’s skills and knowledge are not used 0.0 to the fullest to increase productivity. While the HCI Vietnam EAP LMI captures the future supply of a factor of production, the HCI Full utilization-adjusted HCI U-HCI reflects both the factor supply (capturing investments Source: World Bank HCI and Pennings, 2020. in capital), and a productivity index of how efficiently human capital is used in production (Pennings, 2020). The absence of jobs from the demand side that match the supply of skills can drive labor underutilization and reduce realized growth Papers examining the returns in education up to the first and upward economic mobility. Two U-HCI measures are half the decade noted a decline in returns to education in estimated. The first incorporates the basic utilization rate, midway through the recent decade. Expand (Demombynes which is the employment-to-working-age population ratio. et al, 2018; Doan et al 2018; McGuinness et al 2021). Findings The second measure, the full U-HCI, recognizes that not on reasons for the decline were associated with an increase in all jobs are equal in terms of their ability to utilize human the supply of higher educated students, lower demand for capital. Employment in developing countries may be high, educated workers, slower rates of improvements in the quality but it can often be in low productivity agriculture, informal of tertiary education relative to expansion of enrollment, and or low-skilled manufacturing. As such, the full U-HCI an increase in job mismatch. Using data up to 2020, returns incorporates a measure of better employment, in this context are seen to decline further and most notable for tertiary and non-agricultural employees and employers. high-skilled non-manual jobs (Figure 5.21). From the Last Mile to the Next Mile 147 PART 2. THE NEXT MILE IS THE ROAD AHEAD Figure 5.20. Decline in HCI due to ull utilization adjustment 5.4. Conclusion 0.00 Strategies that achieved upward economic mobility over Change in HCI due to full utilization adjustment −0.05 the past decade may not be effective to continue upward economic mobility into the future. The reduction of −0.10 poverty over the last decade was due in large part to structural shifts out of agriculture and into low- and medium-skilled −0.15 manufacturing and services occupations. These off-farm −0.20 opportunities raised household incomes higher than what could be obtained through faming alone and provided a −0.25 generation of new workers with better jobs than their parents. Even in the absence of the goal to reach high-income targets −0.30 by 2045, labor productivity must increase and be driven by Vietnam −0.35 increases in within-sector productivity to sustain current 6.5 7.0 7.5 8.0 8.5 9.0 9.5 10.0 10.5 11.0 11.5 economic growth levels. However, gains from further exits Natural logarithm of GDP per capita, 2019 from agriculture are limited given how much the sector has High income Lower middle income already shrunk as well as population aging. Growth of high- Low income Upper middle income skill jobs has been slower the growth in medium-skill jobs, and the current youth labor force in their 20s is also more engaged Source: World Bank staff illustration based on Pennings 2020. in manual and routine jobs than youths in other neighboring countries. Thus, there is still work to be done to upskill, retrain, and educate new and old workers to realize a high- productivity and high-skill labor transition. ‒ Figure 5.21. Returns to education and skills over time Education levels (grouped) Occupation skill category 100 100 to unskilled manual occupation group to less than primary education group % Change in hourly wages relative % Change in hourly wages relative 80 80 60 60 40 40 20 20 0 0 2010 2012 2014 2016 2018 2020 2010 2012 2014 2016 2018 2020 Primary Lower secondary High-skilled non-manual Low-skilled non-manual Vocational & Upper secondary Tertiary Skilled manual Note: The reported coefficients are calculated from (hourly) log wage regressions restricted to all wage workers controlling for age, sector, urban/rual, region, and gender. Ethnicity is not available in more recent LFS data sets and thus are not included. Source: World Bank calculation using LFS. 148 2022 Vietnam Poverty and Equity Assessment Chapter 5. Sustaining upward economic mobility 5.5. References World Bank. 2009. Job Generation and Growth (JoGG) Decomposition Tool. Understanding the Sectoral Pattern of Growth and its Autor, David H. and Michael J. Handel. 2013. “Putting Tasks to the Employment and Productivity Intensity. Reference Manual and Test: Human Capital, Job Tasks, and Wages”, Journal of Labor User’s Guide Version 1.0. Poverty Reduction and Economic Economics, Vol. 31, No. 2, The Princeton Data Improvement Management (PREM). World Bank Initiative (Part 2, April 2013), pp. S59–S96. ------------- 2017. Trouble in the making – the future of manufacturing Bodewig, Christian, Reena Badiani-Magnusson, Kevin McDonald, led development. https://www.worldbank.org/en/topic/ David Newhouse, and Jan Rutkowski. 2014. Skilling Up Vietnam. competitiveness/publication/trouble-in-the-making-the-future-of- Preparing the Workforce for a Modern Market Economy. manufacturing-led-development Cunningham, Wendy, and Obert Pimhidzai. 2018. Vietnam’s Future ------------- 2019. April 2019 East Asia and Pacific Regional Jobs – Leveraging Mega-Trends for greater prosperity. Economic Update. Demombynes, Gabriel and Testaverde, Mauro, Employment Structure ------------- 2020a. Vibrant Vietnam. Forging the Foundation of a High- and Returns to Skill in Vietnam: Estimates Using the Labor Force Income economy. Survey (March 12, 2018). World Bank Policy Research Working ------------- 2020b. Improving the Performance of Higher Education Paper No. 8364. in Vietnam : Strategic Priorities and Policy Options. World Doan, T., Le, Q. & Tran, T.Q. Lost in Transition? Declining Returns Bank, Washington, DC. © World Bank. https://openknowledge. to Education in Vietnam. 2018. The European Journal of worldbank.org/handle/10986/33681 License: CC BY 3.0 IGO Development Research 30, 195–216. https://doi.org/10.1057/ ------------- 2021a. A Year Deferred – Early Experiences and Lessons from s41287-017-0080-9 COVID-19 in Vietnam. Eckardt, Sebastian; Demombynes, Gabriel; Chandrasekharan Behr, ------------- 2021b. Vietnam - Science, Technology, and Innovation Diji. 2016. Vietnam - Systematic country diagnostic. Washington, Report. Washington, D.C.: World Bank Group. D.C.: World Bank Group. World Bank; Ministry of Planning and Investment of Vietnam. GIZ. 2021. The COVID-19 outbreak and its social impacts. 2016. Vietnam 2035: Toward Prosperity, Creativity, Equity, and Gutierrez, C., C. Orecchia, P. Paci, P. Serneels. 2007. Does Employment Democracy. Washington, DC: World Bank Generation Really Matter for Poverty Reduction? Policy Research Working Paper 4432, The World Bank Group, Washington DC. Macdonald, K. 2018. “Task Composition Globally and Their Returns 5.6. Notes in Select East Asian Countries.” Background Paper to Sustaining the 37  Revival of Middle-Income East Asia. Washington, DC: World Bank. Assumes all growth components the same as actuals for 2012-2018 Mason, Andrew D.; Shetty, Sudhir. 2019. A Resurgent East Asia: except for demographics (working age population) which are equal to Navigating a Changing World. World Bank East Asia and Pacific the 2002-2012 value. Regional Report. Washington, DC: World Bank. 38  Based on the East Asia and Pacific regional roll-out of the World Bank’s McGuinness, Seamus, Elish Kelly, Thi Thu Phuong Pham, Thi Thu Enterprise Surveys in 2015, 10.7 percent of formal firms in Vietnam said Thuy Ha, and Adele Whelan. 2021. “Returns to education in that an inadequately educated workforce was their biggest obstacle, Vietnam: A changing landscape,” World Development, Elsevier, higher than the 6.9 percent average in the East Asia Pacific region. vol. 138(C). Muller. 2008. “Clarifying Poverty Decomposition.” Working Papers in Economics 217, Göteborg University, Department of Economics, revised 17 Nov 2008. Nguyen Duc Thanh and Ohno Kenichi. 2018. Understanding the Labor Market for Productivity Enhancement. Viet Nam Annual Economic Report 2018. Ohnsorge, Franziska; Yu, Shu. 2021. The Long Shadow of Informality: Challenges and Policies. World Bank, Washington, DC. © World Bank Pennings, Steven. 2020. The Utilization-adjusted Human Capital Index (UHCI). Background paper for the Human Capital Index 2020 Update. World Bank Policy Research Working Paper #9375. Pimhidzai, Obert; and Chiyu Niu. 2020. Vietnam Poverty and Shared Prosperity Update: April 2020. – Shared Gains: How high growth and anti-poverty programs reduced poverty in Vietnam. Washington, D.C.: World Bank Group. Internal. From the Last Mile to the Next Mile 149 PART 2. THE NEXT MILE IS THE ROAD AHEAD ANNEX 5.1. Chapter 5 tables Table A.5.1.1. Skill categorization of occupations and expected education level of workers OCCUPATION GROUP SKILL CATEGORIZATION EXPECTED EDUCATION COMPLETION 1: Legislators, senior officials, managers High-skilled non-manual Tertiary (ISCED 5–6) 2: Professionals 3: Technicians and associate professionals 4: Clerical support workers, low-skilled non-manual Low-skilled non-manual 5: Service and sales workers 6: Skilled agricultural and fishery workers, secondary Skilled manual Secondary (ISCED 3–4) (ISCED 3–4) 7: Craft and related trades workers, skilled manual 8: Plant and machine operators and assemblers 9: Elementary occupations, unskilled Primary (ISCED 1–2) Unskilled Primary (ISCED 1–2) 150 2022 Vietnam Poverty and Equity Assessment Chapter 5. SuStaining upward eConomiC mobility ANNEX 5.2. Growth decomposition methodology This methodology is a summary from the World Bank’s Job Structure Tool. See Muller (2008), Gutierrez, et al. (2007), and World Bank (2009) for more detail. Economic growth per capita—or change in value added per capita—can be decomposed into four components: (i) productivity growth, (ii) employment growth, (iii) labor force participation growth, and (iv) changes in working age population. Growth in aggregate per capita value added can be described by growth in its components using the following identity: where: Y = Total value added N = Population E = Employment L = Labor force A = Working age population (15-64 years old) or: where: y = Value added per capita w = Value added per worker e = Employment rate p = Participation rate a = Working age population / total population Growth in value added per capita can be decomposed into the contribution of changes in each of its components: where each represent a marginal contribution of w, e, p, and a, respectively, to while holding other variables constant. Using the Shapley method, the marginal contribution of productivity per worker (w) to the change in per capita value added (y) from , denoted as , is calculated as: From the Last Mile to the Next Mile 151 part 2. the neXt mile iS the road ahead which simplifies to: The first term on the right-hand side, (s,m), is a parameter vector which determines the weights of each term in the square brackets. Under the Shapley approach, the parameter values are given by the formula: where: s = number of variables other than w taken at t = t1 m = Total number of variables. In this case, m=4 (w, e, p, and a), and for the first two terms in the bracket when s is equal to 0 or 3, the coefficient value is 1/4. Equivalently, when s is 1 or 2, the coefficient value is 1/12. or: Thus, the Shapley value of the contribution of w to the change in y is given by: And analogously, the Shapley values of the contributions of e, p, and a to changes in y is given by: 152 2022 Vietnam Poverty and Equity Assessment PART 2. Chapter 6. Building resilience and safety nets to guard against shocks Chapter 6. Building resilience and safety nets to guard against shocks Key Messages • Households are exposed to a range of shocks that can be costly, for the poor can become poverty traps. • Households experience idiosyncratic and covariate shocks that require different systems and policies. • Protection policies are needed to complementation promotive growth strategies to preserve economic and poverty reduction successes. From the Last Mile to the Next Mile 153 PART 2. THE NEXT MILE IS THE ROAD AHEAD 6.1. Introduction - Shocks, described in Chapter 1, about 6 percent of households in 2010 Coping, and the Need to and 2020 reported that their living conditions were worse than five years ago (Figure 6.1). Incidents of sliding backward Build Resilience are thus not entirely rare, which emphasizes the importance of building resilience and providing safety nets to support Guarding against risks is essential to preserve gains longer-term upward economic mobility. This may also explain and prevent households from falling back into poverty why despite great economic progress, poverty and hunger in the event of shocks or disasters that can lead to remain a top citizen concern for the last five years in UNDP poverty traps. COVID-19 has revealed a wider population PAPI surveys. Among respondents (25 percent in 2018) who of economically insecure groups that risk falling into poverty picked poverty to be the top concern, about half stated that traps in the absence of adequate safety nets. These ‘moving the reason was because they were personally concerned about targets’ can sometimes face greater risks because they are out themselves or members of their family falling into poverty. of the line of sight of government by not being registered in the social protection system and the lack of integrated Households reporting stagnant or worsening living national databases. The current social protection system conditions cite reasons related to chronic challenges as covers the chronic poor and other selected groups in Vietnam, well as shocks. Among households who reported having the but gaps remain. An integrated, adaptive, and modern social same or worse conditions (15 percent), nearly half provided protection system is needed to better guard households reasons related to shocks due to droughts, animal disease, against shocks. This chapter discusses the link between risks pests, illness in the household, or accidents (but not including and falling into poverty, the types of shocks that households job loss) (Figure 6.2). These reasons hold broadly consistent experience as well as their coping mechanisms, and finally patterns over the decade. Root causes of other reasons, such how the social protection system can be improved to provide as low income, for example, could also be related to illness in broader safety nets. the household or shocks to crops. Moreover, for households whose conditions worsened rather than stayed the same, Shocks and risks are related to falling into poverty they were significantly more likely to report illness or death in the household as a cause. Thus, providing adequate safety Poorer households are more likely to report that their nets and disaster response can play a large role in preventing living conditions are the same or worse than five years ago. households from falling backwards. Related to the concepts of churning and economic mobility Figure 6.1. Percent of households reporting worsening living conditions compared to five years ago 2010 2020 30 25 25 20 Share of households (%) Share of households (%) 20 15 15 10 10 5 5 0 0 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 (=poorest) (=richest) (=poorest) (=richest) Same Worsened Same Worsened Source: World Bank staff calculations using VHLSS. 154 2022 Vietnam Poverty and Equity Assessment Chapter 6. Building resilience and safety nets to guard against shocks Figure 6.2. Top three reasons why conditions are same or worse than five years ago Accident or disaster (non-environment) Decreased land use Family or friend conflicts Job loss or underemployment Low income High prices of consumer goods Illness or death in the family Natural disaster impacting agriculture Cattle and poultry disease Low selling prices of agricultural products Increased costs of agricultural production 0 10 20 30 40 50 60 Share of affected households (%) 2020 2018 2010 Note: Not all households selected 3 reasons. Agriculture includes forestry and aquaculture. Source: World Bank staff calculations using VHLSS. There are some differences between poor and rich 6.2. Households and households with respect to the reasons provided that their livelihoods did not improve. Poor households are shocks more likely to cite reasons related to agricultural production, This section illustrates different types of shocks that can while richer households are more likely to cite reasons affect households along the entire welfare distribution. related to rising prices. The regional variation also reflects From the examples provided, even if the poor are not more similar differences in responses. For example, households in exposed to risks, with less savings and buffers, shocks can the richer regions of the Southeast and Red River Delta are become poverty traps and thus more consequential to more likely to cite increasing prices, while households in the household welfare and economic mobility. Central Highland more likely cite low agricultural prices and cattle disease. The range of responses lends some caution Examples of two different types of shocks are given: (i) to the interpretation. Richer households may report worse idiosyncratic shocks, or those which only affect specific conditions while still having adequate material comforts but individuals or households; and (ii) covariate shocks, may be unsatisfied with their speed of upward economic or those which affect entire communities, regions, or mobility. Also, rich and poor households report “low countries. Idiosyncratic shocks include illness or accident, income” at similar rates, which adds a subjective element to which usually affect a single individual or household. the interpretation of low incomes. Covariate shocks, such as a natural disaster or an economic shock (as resulting from the COVID-19 lockdowns or the Global and Asian Financial Crises), affect many more people. The distinction can be important because a household’s risk profile varies by the type of shock: susceptibility to illness may depend upon underlying health, access to preventative care, living and working conditions; susceptibility to natural disaster may depend on household location; susceptibility to economic shocks may depend upon the sector a worker is From the Last Mile to the Next Mile 155 PART 2. THE NEXT MILE IS THE ROAD AHEAD in. It is not a mutually exclusive distinction; two households share of out-of-pocket health expenditure stood at 45 percent living in a typhoon-prone area will have a different level of in 2018 (Figure 6.3). While this is slightly lower than the risk exposure if one has a sturdy or unsound house. The same LMIC average (51 percent), it is considerably higher than the category of shock can be either idiosyncratic or covariate; equivalent UMIC and EAP out-of-pocket shares, at 33 and unemployment may come because a single firm conducts a 26 percent, respectively. The high out-of-pocket expenditure small number of redundancies or because an entire sector or share in Vietnam amplifies the risk of impoverishment from country sees widespread job losses due to an economic shock, health shocks. such as tourism during COVID. Nonetheless, the distinction is also important because it has different policy implications. Examining another indicator on the cost burden of Idiosyncratic shocks are generally best dealt with by having health shocks—catastrophic health expenditure— effective social insurance coverage (unemployment insurance, demonstrates the significant risk that health shocks health insurance, old age pensions) while covariate shocks pose to households in Vietnam. Catastrophic health often require a more coordinated government response, such expenditure occurs when co-payments for healthcare result as preemptive disaster risk management and ex-post disaster in a financial catastrophe (WHO, 2005). In 2016, 9.4 percent response, or large social programs such as wage subsidies of the population in Vietnam incurred catastrophic health or direct income support as many countries deployed expenditure, measured as out-of-pocket spending exceeding during COVID. 10 percent of household income (Figure 6.4). The reduction in catastrophic health expenditure has been slow over the Health shocks decade, with signs of a potential reversal to this progress in 2016. Rural areas have a higher share of households incurring Households in Vietnam are exposed to idiosyncratic catastrophic health expenditure, but the urban-rural gap has health shocks which can have a significant economic been closing in recent years. impact through changes to household income, medical care spending, and household consumption. Sudden Around a fifth of the population in Vietnam is at health declines can lead to a persistent reduction in earnings risk of incurring catastrophic and impoverishing and other labor market outcomes. The impact on household expenditure for surgical care. Surgical care is a common income depends on who in the household falls ill, whether type of medical care required in the face of particular health the household has multiple income earners, social protection shocks. Data from Harvard Medical School illustrate that the coverage, and the level of financial support available from risk of incurring costs of surgical care exceeding 10 percent of other family and friends (Wagstaff, 2005). Large out-of- income (catastrophic expenditure) has declined by nearly 50 pocket health expenditures following a health shock can percent in the last decade but remains relatively high. In 2020, result in a financial catastrophe for households. The extent the share of population at risk for this type of expenditure was of medical expenditures following health shocks is partly 22 percent, compared to just 10 and 9 percent, respectively, in determined by enrollment in health insurance and its coverage UMIC and EAP countries. Similarly, the share of population levels. Changes to household consumption are determined at risk of impoverishing expenditures, defined as out-of- by the options available for consumption smoothing. In the pocket payments for surgical care that drive people below absence of adequate coping mechanisms, food consumption the extreme poverty threshold of $1.90 PPP per day, was 19 levels may fall to below the recommended daily caloric intake percent in 2020 in Vietnam, compared to 7 percent in UMIC for poor households. and EAP countries. It is worth noting, however, that in 2015, Vietnam’s share of population at risk of impoverishing Out-of-pocket health expenditure in Vietnam has expenditure for surgical care fell below the LMIC average continued to climb in the last decade and exacerbate and has remained below it ever since. Progress has been made health shocks even for those with health insurance. The in improving the financial protection of households from share of out-of-pocket health spending is indicative of the health shocks, but further gains are needed to achieve greater level of financial protection against health shocks. Vietnam’s economic development. 156 2022 Vietnam Poverty and Equity Assessment Chapter 6. Building resilience and safety nets to guard against shocks Figure 6.3. Out-of-pocket health Figure 6.4. Catastrophic health expenditure expenditure remains high has been declining slightly (% of households) 60% 20 16.6 15.2 15.7 Share of total health expenditure 50% 15 13.6 13.7 45% 45% 45% Share of households (%) 12.4 12.6 43% 11.6 41% 41% 42% 10.8 9.9 10.0 9.4 9.5 40% 37% 10 8.8 9.9 7.5 37% 30% 5 20% 0 2010 2011 2012 2013 2014 2015 2016 2017 2018 2002 2004 2006 2008 2010 2012 2014 2016 Vietnam Lower middle income Urban Rural Upper middle income East Asia & Pacific Source: World Development Indicators. Source: World Development Indicators. An aging population will increasingly be exposed The other potential aging shock is growing old without to multiple shocks. enough income to live on—a real risk currently facing three-quarters of Vietnam’s elderly. As people age and Aging can drive two potential shocks for individuals can no longer work, they nonetheless require income to live and households. The first is increased risk of illness. As on. A quarter of the current elderly receive a contributory Vietnamese live longer and grow older, they are increasingly pension, having been either public sector workers or formal exposed to health risks. For example, someone in Vietnam workers in the private sector, both of which have made who is 60 years old can be expected to live another 23 years, pension contributions over their working lives and are able but only 17 of those years will be healthy, with much of the to retire with a reasonable degree of income support. Most disability due to chronic non-communicable diseases (NCDs) of the other three-quarters of elderly have no income support and disabilities associated with aging (World Bank 2021b). beyond whatever savings they may have that they may get from family, while a small number receive a social pension The risk of greater illness is compounded by the high from the government (World Bank, 2021a). out-of-pocket health costs previously discussed, which increase with age and may lead to foregoing of care even Moreover, even for those receiving some form of old-age when seriously ill. As aging leads to greater inpatient and income support, it is either unsustainable or insufficient outpatient needs, so too do the out-of-pocket costs increase. (World Bank, 2021a). Contributions to Vietnam’s pension Out-of-pocket health spending for both types of care rise scheme (VSS) have been less than required to finance significantly with age, reaching around VND 2.0-2.5 million payments to an increasingly aging base, and the scheme will per year on average for those aged 60–79, which is around soon run deficits to finance payments. At the same time, the 15–20 percent of the $3.20 poverty line (World Bank, 2021b). benefits for those being assisted by the government with social Moreover, while they fall for those aged 80 and above, this pensions are small and provide little support; the average ratio is likely because this age group has by far the highest rate of of VSS pensions to social pensions is 18:1. Meanwhile, more foregone care even in the face of serious illness. than half of all elderly receive neither. From the Last Mile to the Next Mile 157 PART 2. THE NEXT MILE IS THE ROAD AHEAD Shocks experienced by households during weak safety nets. The impact of COVID-19 on households COVID-19 and firms illustrated the existing gaps in access to services and social assistance, and the importance of building resilience COVID-19 has likely been the most significant co- and stronger safety nets to guard against poverty traps. variate shock experienced by households since the end of the Vietnam war. Almost 70 percent of households reported Environmental risk and exposure experiencing a negative shock39 between February and June 2020, which covers the duration of the first nationwide Vietnam is one of the most hazard-prone and climate lockdowns in April 2020 (World Bank, 2021b). The crisis risk-affected countries in the world. From 1990 to 2018, affected households across the entire welfare distribution over 26,000 people died directly from epidemics, floods, through a variety of channels. Wealthier households more landslides or storms, with damages estimated at US$14.4 likely experienced income losses from family businesses, and billion. Droughts have caused an additional 7 billion in poorer households experienced losses from farming activities. damages over this period (World Bank and ADB, 2020). Adverse labor impacts occurred broadly across different Nearly a million people have been affected by natural hazards socioeconomic groups and geographic areas. In March 2021, almost every single year since 198541. In addition, urbanization, a little less than one-third of all respondents knew someone industrial production, and a growing consumer class has led to who had lost a job or switched jobs since the beginning of the challenges in waste management, water pollution, congestion, pandemic in February 2020. air pollution, and marine plastics. This compounds existing climate change and environment issues. These risks directly Other household demographics correlate to the reporting affect household economic security and livelihoods. of experiencing negative shocks. For example, households with children were more likely to report experiencing shocks About 10 percent of communes reported experiencing during periods coinciding with occurrences of school an emergency in 2018, including natural disaster, fires, closures, with women more likely to take absences from work and epidemics. About 60 percent of these communes to care for children. Moreover, increased care responsibilities received relief or aid, primarily in the form of direct cash relief were the main driver of the gender difference in lost turnover from the government budget, although information on the from family businesses. Lower family business income from severity of the disaster is not known. Based on UNDP PAPI female respondents compared to men is driven by fewer hours surveys, in 2020 about 14 percent of households experienced a worked. Agriculture households were less likely to report flood over the last five years, 17 percent experienced droughts, shocks than nonagricultural households, except during a and 18 percent were affected by typhoons. Moreover in 2020, period that covered a drought that severely affected farming the share of farming households reporting disruptions to in some regions. their operations was high in the Mekong Delta region (World Bank, 2021b). COVID-19 also illustrates potential scarring effects and that shocks can take time to recover from completely. The Climate change is expected to exacerbate existing risks. World Bank’s COVID-19 monitoring in the early phases of Under a worst-case climate-change scenario, sea levels are set the pandemic illustrated that right before the emergence of the to rise by 30cm and 70cm by 2050 and 2100, respectively. This fourth wave in April 2021,40 about 30 percent of households would result in the flooding of land currently inhabited by in March 2021 still reported lower incomes than the year 20 million people, equivalent to a quarter of the population before (World Bank, 2021b). The challenges brought on by (Rentschler et al., 2020). Climate change can also increase COVID-19 revealed a wider set of vulnerable households, the frequency of droughts, which compounds the severity of beyond the traditional characterization of the chronic poor as saline intrusion, together with rising sea levels. More extreme being mainly from certain regions or ethnic minority groups. weather patterns such as severe typhoons and monsoon Certain groups experienced slower income recovery, including rains also pose risks of flash-flooding and landslides in households without formal sources of income, women, and highland areas. households in the bottom 20th percentile who are closer to the poverty line and vulnerable to falling into poverty given 158 2022 Vietnam Poverty and Equity Assessment Chapter 6. Building resilienCe and safety nets to guard against shoCks The types of environmental risks vary across regions of high levels of air pollution, which is further trapped in place the country, with most areas ranking high in at least by geographical features. The North Central region is more one type of environmental risk (Figure 6.5). Starting affected by heavy rains, which are associated with risks of in the north of the country, high levels of air pollution are flash floods or landslides. Finally, most of the population in concentrated in the Red River Delta where high urban the densely populated central Red River Delta, Southeast density, coal-fueled factories, and coal mines all contribute to region, and the Mekong Delta are exposed to high flood risk. Figure 6.5. Environmental conditions and risks by district Temperature (mean) Air pollution (PM2.5) Precipitation (mean) Temperature (Annual average, Celcius) Air pollution (PM2.5) Precipitation (millimeter) 18.602 28.394 9.87 43.12 2.76 12.84 Percent of the population SPEI drought indicator exposed to high flood risk SPEI drought indicator Percent of the population −0.864 0.864 0 100 Note: District-level map. SPEI is available at the province level. See Table 6.1 for definitions. From the Last Mile to the Next Mile 159 PART 2. THE NEXT MILE IS THE ROAD AHEAD A pressing issue faced by cities is soil subsidence caused by many inhabitants with no choice but to migrate. Between excessive groundwater usage to support growing populations. 2008 and 2018, net migration out of the Mekong Delta region Parts of Ha Noi have been recorded as subsiding at a rate of 68 was around 1 million people (Climate Home, 2018). The mm per year between 2007 and 2011 (Dang et al 2014), while 2015–2016 drought led to the destruction of 160,000 hectares some of Ho Chi Minh City has seen subsidence in excess of 70 of crop. Many households are unable to switch livelihoods to mm per year over the same period (Minh et al., 2015). salt water-tolerant commodities. With increasing frequency and/or intensity of climate events, more households are at risk The Mekong Delta is well known for its share of of losing their livelihoods and homes and falling into poverty, environmental hazards affecting farming livelihoods. with many forced to migrate from the region. While agricultural wages in the Mekong Delta are relatively high and support low poverty rates, wages in all other sectors District-level environmental risk does not always are some of the lowest in the country. Climate change has correlate with poverty rates42 (Figure 6.6), although severely impacted livelihoods in the Mekong Delta, leaving poorer households are less able to cope with shocks when Table 6.1. Environment indicator, definitions UNIT, DESCRIPTION PERIOD OF DATA Average topographical incline Degree 2017 Air pollution PM2.5 concentration (μg/m3) Annual daily average over 2018 Average settlement inundation Meter 2020 flood model; 2015 Global Human depth in 1-in-100 year flood Settlement Layers Temperature Celsius Mean and variance over 2020 Wind speed Meter per second Mean and variance over 2020 Precipitation Millimeter Mean and variance over 2020 Flood risk exposure Share of the population exposed to high risk 2020 flood model and 2020 population data SPEI drought indicator An SPEI of 0 indicates a value corresponding to 50% of the 2016–2021 average cumulative probability of D, according to a Log-logistic distribution." (D value is a threshold for drought, defined as precipitation minus evapo-transpiration). In sum, a higher SPEI corresponds to a higher probability of drought. Figure 6.6. Share of districts with a high environmental risk factor, by district-level poverty rates 100 Share of districts with 80 high risk (%) 60 40 20 0 High poverty Medium poverty Low poverty At lest one out of 6 risks Precipitation (variance) Temperature (variance) Wind (variance) Air pollution (PM2.5) Share of population exposed to high flood risk SPEI drought Source: World Bank staff calculations. 160 2022 Vietnam Poverty and Equity Assessment Chapter 6. Building resilience and safety nets to guard against shocks they do happen. Environmental risks vary by geography, 6.3. What types of shocks which is sometimes related to poverty (Chapter 2), but not always. For example, the Northwest area has high poverty and are households more at is geographically mountainous, which is related to the risk risk from? of landslides, but it is less affected by pollution or drought. Districts can be categorized into high, medium, and low This section examines the relationship between poverty, poverty and environmental risk groups (based on unweighted vulnerability, and risk in greater detail. Analysis presented rankings). While not considered a primary risk due to there seeks to identify which households are “chronically poor,” as being fewer districts with such characteristics, topographical well as the “risk-induced vulnerable” who usually live above incline is also included, as mountainous areas tend to be less the poverty line but sometimes fall below it because of shocks. suitable for agriculture and are at risk of flash floods and Risk-induced insecurity arises from two sources: idiosyncratic landslides, which are hazards specific to traditionally poor shocks and covariate shocks. This section divides risk- areas. Low-poverty districts have higher risks for other factors induced insecurity into these two sources. This can be useful such as air pollution, mainly in the districts around the Red because often vulnerability to idiosyncratic risks is greater River Delta. On average, poor and non-poor districts tend to than vulnerability to covariate risks. While community- have a similar probability of ranking in the top third in at least based insurance mechanisms can be effective in mitigating one out of six risk factors.43 The relative similarity in the share idiosyncratic shocks, where covariate risks matter, insurance of low- and high-poverty districts exposed to at least one risk mechanisms cannot rely on local risk pooling. In such regions, also illustrates the complexity and ubiquity of natural and the need for effective risk management and adaptive social man-made risks in Vietnam. However, whether or not poor protection programs is greater. households are over-exposed, they have fewer coping strategies and safety nets to weather such shocks. These distinctions have important policy implications. Policies required to address chronic insecurity are not exactly Overlaying maps of district-level environmental risk and the same as those needed to address risk-induced insecurity. poverty rates is a useful exercise to gain more powerful The chronic insecure usually live below the economic security insights on the extent of vulnerability from environment line, even in good years, because they lack the human and and climate-related factors. Figure 6.7 illustrates districts physical assets to earn a sufficient livelihood even in good where environmental risk or poverty rates are ranked in the conditions. These households will most likely benefit from top third across all districts. Complementary to Figure 6.5, better livelihoods, cash transfers, and better delivery of basic these bi-variate maps also illustrate the high frequency of risks services to facilitate investments in physical and human across the country. The poorest districts are not often also capital. Those who face risk-induced insecurity consume those with the highest environmental risk (black color). The enough to live above the line, but that this consumption can correlation between districts with high poverty rates and higher be highly variable because of idiosyncratic or covariate shocks, environmental variables is significant for slope, green color, and meaning sometimes they fall below the line. These households average precipitation. do not necessarily need the same investments in assets but do need protection from shocks, which might mean insurance programs to increase resilience. From the Last Mile to the Next Mile 161 part 2. the neXt Mile is the road ahead Figure 6.7. Poverty and environmental risk maps Mean flood deapth Average slope Average PM2.5 Medium poverty High poverty Low poverty Environmental risk (top third) Medium Low Variation in precipitation Variation in wind speed Variation in temperature Variance drought Average percipitation Mean wind speed Mean temperature Mean drought ©Mapbox ©OSM Note: Only districts ranking in the top third of the environmental variable are shown. Poverty rates are grouped into high, medium, and low categories by ranking of district level poverty rates. See Table 6.1. for definitions. Source: World Bank staff calculations. 162 2022 Vietnam Poverty and Equity Assessment Chapter 6. Building resilience and safety nets to guard against shocks In most regions, chronic insecurity is higher than is nearly as important. This regional pattern is the opposite of risk-induced insecurity the chronic insecurity to risk-induced insecurity pattern seen in the last paragraph, suggesting in areas with low chronic In Vietnam the rate of chronic insecurity is twice that insecurity idiosyncratic risk is more important. of risk-induced insecurity, but this varies significantly by region. At the $5.50 line, the rate of chronic insecurity in the country is around 21.7 percent, 1.9 times that of risk- 6.4. Household coping induced insecurity at 11.5 percent (Figure 6.8).44 In other words, while around one in five Vietnamese usually lives below strategies this aspirational economic security line, another one in ten Coping strategies that households rely on during are vulnerable to falling below it sometimes due to shocks. economic crises or shocks range from self-coping, personal Regional outcomes vary widely. With so many chronically networks, to formal external channels. In response to a insecure in the Midlands (51.2 percent), there are fewer people crisis, households can rely on their own means to cope, such left to experience risk-induced vulnerability (13.3 percent). as using personal assets and savings. Changing behaviors such Conversely, in the Red River Delta where chronic insecurity as reducing consumption or engaging in more economic is below 5 percent, risk-induced insecurity (9 percent) is the activities to smooth income are other typical strategies to larger concern, being twice as prevalent. Other regions where deal with shocks. In countries with developed social response risk-induced insecurity is relatively more important than the systems, financial or government institutions can also be a national average are the Southeast and the Mekong Delta. common means of support. Vietnamese households are in general not indebted, which reduces the serious implications In general, idiosyncratic risk contributes to insecurity of moderate declines in income from economic shocks. High more than covariate risk, but this also varies by region. recurring monthly expenses such as housing rent, mortgages, Nationally, idiosyncratic risk contributes to risk-induced car payments, or debt payments are uncommon in Vietnamese insecurity 1.2 times more than covariate risk (Figure 6.9). The society. A low cost of living also ensures that essential items are Red River and Mekong Deltas are regions where idiosyncratic still affordable, and growing food for self-consumption was also risk matters a lot more (1.7 and 1.5 times respectively), while common among poorer households. in the Midlands and North and Central Coast covariate risk Figure 6.8. Rates of chronic insecurity and Figure 6.9. Relative importance of idiosyncratic risk-induced insecurity by region, 2018 and covariate risk by region, 2018 80 5 2.0 70 Ratio of Idiosyncratic/Covariate Risk 4 Ratio of Chronic/Risk 60 1.5 50 3 Percent 40 2 1 30 20 1 10 0.5 0 0 0 All Midlands & Northern Mountains North & Central Coast Central Highlands Southeast Mekong Red River Delta All Midlands & Northern Mountains North & Central Coast Central Highlands Southeast Mekong Red River Delta Chronic Economic insecurity Risk-induced Economic Chronic: Risk Ratio Insecurity Source: VHLSS 2018 and World Bank’s Vulnerability Tool. From the Last Mile to the Next Mile 163 PART 2. THE NEXT MILE IS THE ROAD AHEAD BOX 6.1. Modelling risk and vulnerability among households The distinction between chronic poverty and risk-induced vulnerability depends upon expected household consumption and how much it might vary around that average. Poverty measures from Chapter 1 tell us how many people live below the poverty line at any one time, specifically at the time the household survey was conducted. But they do not indicate what a household’s chances are for being in poverty in the future. This section uses a modelling approach to decompose notions of economic insecurity, drawing upon Skoufias (2019) and Skoufias and Baez (2021)45. For this exercise, consider anyone living below the $5.50 economic security line is considered “economically insecure.” The insecure are then divided into two groups: 1. Households experiencing chronic insecurity are those whose expected or average consumption is always likely to be below the economic security line. 2. Second, the notion of risk-induced insecurity is introduced. This reflects the notion that some people consume enough to live above the line sometimes but that this consumption can be highly variable because of idiosyncratic or covariate shocks, meaning sometimes they fall below the line. This group is referred to as the “risk-induced vulnerable”. Classifying households as chronically insecure or facing risk-induced insecurity means understanding how household consumption varies and setting a risk probability threshold. The analysis in this section uses per capita household consumption as the measure of welfare and the UMIC ($5.50/day 2011PPP) line as that of economic security, reflecting Vietnam’s higher aspirations for the minimum standard of living that all of its citizens should enjoy. In addition, the analysis means specifying: (i) a statistical model of how consumption varies and (ii) a threshold for the probability of falling below the line to be called insecure. The statistical model of the consumption distribution is empirically derived from the variation in the household survey data and described in more detail in Skoufias et al (2021). The probability threshold is a policy choice and depends upon policymakers’ (and society’s) tolerance for risk. Higher thresholds (as used here) will increase the number of households classified as insecure; a higher threshold (or tolerance) for the probability of falling under the welfare threshold will decrease the number of households classified as insecure. These two types of insecurity depend upon how much a household’s consumption varies in different states of the world, whether “normal,” “good,” or “bad.” Households who usually live below the economic security line are chronically insecure; those who are usually above the line but sometimes fall into insecurity. Figure B.6.1.1. illustrates these two concepts with some examples. The orange box is each household’s average (or expected) consumption across different states of the world. The black lines show how much each household’s consumption can vary in these different states. According to Skoufias and Baez (2021), “If a household experiences a negative shock, such as a drought or is a victim of crime, then depending on the severity of the shock, the household would be expected to have consumption somewhere to the left of the average welfare (orange square). If the household experiences a positive shock, such as agricultural prices that are more favorable, then the household would be expected to be somewhere to the right of the expected welfare.” In this figure, households A, D, G, and I have expected consumption below the economic security line and can be classified as chronically insecure. In a “good” year they may be above the line, but on average their consumption is below the line; household A never gets out of insecurity even in the best years. By contrast, households C, F, and J have expected consumption that is above the line, but depending on the realized shocks, in some states of the world they may be below the line. These households are classified as facing risk-induced insecurity.46 Household B, E, and H are not expected to fall below the line under normal shocks and are thus not identified as insecure. Continued > 164 2022 Vietnam Poverty and Equity Assessment Chapter 6. Building resilience and safety nets to guard against shocks BOX 6.1. Modelling risk and vulnerability among households (Continued) Figure B.6.1.1. Stylized examples of chronic insecurity and risk-induced insecurity A B C D Household E F G H I J Poverty line Welfare Mean and Variance of welfare Observed welfare Note: Adapted from Skoufias et al (2021). Self-coping strategies vary between poor and while richer households coped by relying on personal saving rich households. and assets and reducing consumption. In earlier studies on coping behavior during the global financial crisis, poor Self-coping strategies used by households when facing households having fewer assets either did not employ any shocks include relying on oneself or family or friends coping strategies, sought external financial assistance from for support. This was evident during COVID-19 when family and friends, or sought additional income-generating households were self-insuring and remained self-sufficient, opportunities (Tran, 2015). These differences were similarly despite the lack of widespread social relief measures. observed during COVID-19. While these coping strategies Reducing consumption was the most common household were employed during COVID-19, the impact of different coping strategy, followed by borrowing from friends. Other shocks such as health or disaster events are likely to bring strategies were less commonly used: growing food, relying on about different coping strategies. savings, loans from an institution, or receiving assistance from family members. Formal channels of assistance from financial The poor are less likely to have savings institutions or government were the least common. The low level of reliance on formal public support reflects issues in the Savings enable households to smooth consumption design and implementation of the fiscal response. in response to income shocks. Vietnamese save to buffer against shocks, in particular illness, given the high cost of out- Coping strategies vary between poor and rich of-pocket health expenses. Three in four workers are in farm households. While poor households rely on external sources, or non-farm informal employment and therefore not covered rich households are more able to rely on their own means, by the formal social security system. These households largely again highlighted during COVID-19. Poor households turned depend on their own consumption-smoothing mechanisms, to family and friends for financial assistance during this time, such as savings, in the event of a negative income shock. Even From the Last Mile to the Next Mile 165 PART 2. THE NEXT MILE IS THE ROAD AHEAD among households that save, savings intended for emergencies Borrowing through personal networks is more may not be sufficient to cope with such shocks when they common than through financial institutions. arrive. Survey data from the Global Findex database show that 70 percent of respondents (aged 15 and older) reported being Financial inclusion is limited for certain vulnerable able to come up with emergency funds, but only 14 percent groups due to household-specific constraints. The share of reported savings to be the main source (2017). The main adults with an account at a financial institution increased from source of emergency funds is likely to be income from work. 26 percent in 2017 to 42 percent in 2019 (Figure 6.10). This This ability to earn an income, however, is severely impacted varies by socioeconomic groups along differences in wealth during shocks such as economic crises or illness. and education levels. While 83 percent of rich households have an account, only 17 percent of poor households do. This There is evidence that level of income per capita drives highlights that financial inclusion is not necessarily due to household savings. Households in the bottom income limited supply but rather to household-specific constraints to quintile, particularly those in the bottom decile, persistently access that disproportionately affect the vulnerable. have negative savings rates. Notably, almost one in three households had no or negative savings in 2010. Analysis at a Having large family networks is a significant more disaggregated level finds that this varied greatly across determinant of financial resilience. Family members income groups: it was 68.8 percent and 44 percent in the tend to be more altruistic toward each other, share common bottom and second-to-bottom quintiles, respectively, in the obligations, and have stronger contract enforcement (Cox same year, thus significantly higher than the average. This and Fafchamps, 2008; Hamilton, 1964; La Ferrara, 2011). pattern held across all the years when VHLSS was conducted Vietnamese society places great importance on family ties, in the last two decades (Nguyen, 2022). While the poor roles, and responsibilities. Families are essential and core social are usually beneficiaries of targeted assistance programs, units. This can be seen in the large share of households (50 households in the second-lowest income deciles are normally percent) that engage in family businesses.47 Family units also not. Therefore, not only the poor, but also those non-poor have a significant place in administrative processes. The family with negative or modest positive savings rate should be given registry book, passed down for generations, is an essential adequate support in time of economic shocks. The same for life events such as marriages, opening bank accounts, figure also shows that higher income households have higher and so on. saving rates and households in the bottom income quintile have had negative savings rates since 2010. This was largely Borrowing from personal networks is more common driven by households in the bottom decile, approximately the than borrowing from financial institutions (Figure poor according to MOLISA’s poverty line. In the better years 6.11). In developing countries, borrowing from family and of 2016 and 2018, savings of households in the second-lowest friends is a common strategy in the presence of unexpected income decile, approximately MOLISA’s near poor, turned shocks (Demirguç-Kunt and Klapper, 2013; Pearlman, 2010). positive, albeit modestly. Coping strategies during COVID-19 Relying on financial assistance from family and friends has also highlighted differences in the ability of poor and rich been more commonly seen in poor households compared households to utilize savings to cope with shocks. During to rich as a coping strategy during COVID-19 (World Bank, this time, rich households are much more likely to tap into 2021b). While used less frequently than borrowing from savings compared to poor households (World Bank, 2021b). friends and family, poor households borrowed from financial With a lower level of assets and savings, poor households are institutions more frequently than tapping personal savings. also less likely to use formal savings channels. The growth rate These financial institutions include State Policy Banks and of household deposits over the medium term also slowed and farming cooperatives and other noncommercial banks, which may indicate changes to the saving behavior of households. provide loans to agricultural and poor households. Overall, this is consistent with previous research that poor households tend to rely on external sources, whereas rich households are better able to cope within their own means. 166 2022 Vietnam Poverty and Equity Assessment Chapter 6. Building resilience and safety nets to guard against shocks Figure 6.10. Use of financial services has improved Figure 6.11. Borrowing from family and on some dimensions and less in others friends is the most common source 50 80 42.3 70 60.2 40 60 Share of respondents (%) Share of respondents (%) 48.4 30 50 26.4 42.0 24.0 40 20 30 20 10 3.9 3.2 10 0 0 Has an account Has a debit card Has a credit card From a financial From family From an informal institution or friends savings club 2017 2019 2017 Source: Based on 2017 data from the Global Findex database and World Bank 2019. Source: Based on 2017 data from the Global Findex database. Notes: The FINDEX target population is the population of individuals aged 15 Notes: The FINDEX target population is the population of individuals aged 15 years and older. years and older. 6.5. Social Protection the system is less prepared to support individuals in the face programs in Vietnam of idiosyncratic shocks. This was evident during COVID-19 (See Box 6.2). today • Implementation issues are an important constraint. The application process can inhibit sign-up. A lack of a national As the previous section showed, households primarily social registry means responses to large shocks can quickly rely on their own resources and personal networks to scale vertically, providing additional payments to existing cope through shocks. The social protection system can play beneficiaries, but not horizontally, providing new payments a larger role in guarding against shocks and helping households to new beneficiaries. A largely cash-based transfer system avoid poverty traps, but it currently has some key limitations. means assistance may not get to households as quickly as they need it. • Social protection is underfunded, and programs are • Some risks are addressed, but coverage is limited. Social fragmented. Vietnam underinvests in social protection insurance covers unemployment and old age income, but relative to peers, meaning coverage will be too low, benefits only formal workers are covered; with most of the country’s insufficient, or both. Furthermore, there are design issues. workers being informal, the gaps in coverage are significant. There is no flagship program around which social assistance • Some risks are not adequately addressed. For example, is built but instead a number of smaller and uncoordinated natural disasters—which will increase in frequency and/ programs. This means economies of scale are missed, or intensity with climate change—are not prepared for as exacerbating the lack of funding, and there is no central well as they could be, nor responded to as quickly and as design to make sure that the mix of schemes matches the comprehensively as needed to stop households from resorting changing risk profile of households and shocks. At the same to ineffective coping mechanisms. Much of the country has time, decentralization further disrupts efforts at coordinating health insurance, but out-of-pocket costs undermine its social protection. Large budgets are channeled toward area- effectiveness. based targeting programs and investments, which gives local governments flexibility in how they allocate resources, but From the Last Mile to the Next Mile 167 PART 2. THE NEXT MILE IS THE ROAD AHEAD BOX 6.2. Vietnam’s early COVID-19 response Vietnam’s initial fiscal support in 2020 for households in response to the COVID-19 crisis was smaller than its regional peers. Vietnam’s income support to households included both topping up payments to existing social assistance beneficiaries and making payments to new beneficiaries not previously covered. However, the amount of spending early in the pandemic was significantly smaller than in other East Asian countries, and total COVID-19 spending in support of households lagged all major regional peers, such as China, Indonesia, Malaysia, the Philippines, and Thailand (World Bank, 2021c). Consequently in 2020, although employment and wage losses were lower in Vietnam than in other countries, the gap between those losses and the support that households did receive was proportionally larger than in all other countries except for the Philippines, where households experienced very high losses (Figure B.6.2.1). Figure B.6.2.1. Vietnam had the second-largest Moreover, implementation issues meant that the size relative gap between household losses and and coverage of household income support was even support, East Asia and Pacific smaller than intended and was not pro-poor. The initial 10 COVID-19 package was to cover 5 million informal workers Percent of gross domestic income (%) affected by lockdowns. Due to the lack of a social registry or 8 other processes to identify eligible people, only 1 million new 6 beneficiaries were enrolled (World Bank, 2021b). Moreover, households receiving new assistance were not generally poorer. 4 Sixty-one percent of households receiving new COVID-19 assistance were from the richest 60 percent of households and 2 39 percent from the poorest 40 percent, indicating near equal chances of being supported for households across the 0 distribution (World Bank, 2021b). Nor was this because Vietnam Mangolia Thailand Malaysia Indonesia Philippines poorer households did not know how to apply; the poorest Earnings loss Support to households 40 percent of households made up 43 percent of applicants. Source: World Bank 2021c Benefit levels were also relatively low, limiting effectiveness for those who did receive support. Benefits were for a maximum of three months and relatively low. National Devotees received the highest benefits, about VND 1.5 million over three months; in comparison, minimum monthly wages are about VND 4 million per month. The estimated impact of this support was minimal (World Bank, 2021b). The second support package in response to the Delta variant sweeping Vietnam in 2021 was smaller than that of the first package. The package was designed to be implemented for a period of 8 months between May-December 2021, longer than the duration of the first package of 3 months. Individuals falling under national devotees, social assistance beneficiaries, and members of the poor and near-poor were no longer eligible for support, while contracted employees in quarantine, children infected or quarantined, and employees in certain occupations received financial support. Affected contract employees who were pregnant or had young children and employees in certain occupations received additional amounts based on the number of eligible individuals. Changes reduced aid to the pre-existing vulnerable and expand aid to formal workers (who are less likely to be poor, although may well be affected by the crisis). Thus, aid in the form of household cash transfers saw changes between the first and second rounds of national support packages, potentially disadvantaging poorer households. 168 2022 Vietnam Poverty and Equity Assessment Chapter 6. Building resilience and safety nets to guard against shocks The social assistance system requires Indonesia, and the Philippines. Different programs have their modernization. own delivery systems (registration, eligibility determination, payments) and are very decentralized, leading to significant The social assistance system has evolved, but further variation across different parts of the country. modernization is needed. The system has been transformed from a basic income support system of last resort for those This fragmentation is exacerbated by the fact that unable to work to a system capable of addressing multiple several poverty-reduction programs work in parallel to challenges of a more diverse set of beneficiaries (see Box 6.3 the social assistance system. These additional programs for a description of Vietnam’s social protection system today). include the National Target Poverty Reduction Program There has been a gradual transition to a quasi-universal social (NTPRP), Program 135, and the Sustainable Poverty pension (albeit for the very old) and support for people Reduction Support Program (SDRSP). These programs are with disabilities. Some efforts have been made to improve predominantly area-based but also include some household- the relatively rudimentary delivery system that continues based social assistance. For instance, poor, ethnic minority to rely on manual, paper-based administrative processes. households facing housing challenges are provided a lump Social assistance coverage has expanded significantly from sum for housing, production land, and clean water assistance. 200,000 beneficiaries in 2000 to more than 3.1 million in Another subset of programs provides assistance to poor and 2020. Around 1.8 million of the beneficiaries are elderly, ethnic minority households through cash or in-kind support, reflecting the importance of the social pension scheme, which such as the kerosene subsidy for poor households of ethnic has universal coverage above age 80 but is means-tested for minorities residing in areas without grid connection. Limited those aged 60-79. Vietnam did not always perform relatively active labor market programs also operate without any link to poorly in social assistance. A decade ago, Vietnam’s social social assistance programs.48 assistance coverage was in line or better than many East Asia and Pacific neighbors. However, since then it has started to Despite this complexity, Vietnam spends relatively lag. In particular Malaysia and Indonesia’s social assistance little on its safety nets, which consequentially provides coverage has increased the most. Both countries used to lag benefits that are low relative to global benchmarks. The behind Vietnam but now have wider coverage. (World Bank decentralized and fragmented nature of the system makes it and MPI, 2016). difficult to calculate the overall spending on social assistance.49 This is relatively low compared to other middle-income The administration of social assistance programs is countries and is the result to a great extent of low benefit levels. complex and can be difficult to navigate for potential In recent years, benefits have been allowed to shrink in real beneficiaries. According to MOLISA, there are 8 Laws, terms with social assistance benefits frozen in nominal terms 1 Master Plan, 14 Decrees of Government, 37 Decisions from 2015 to 202150. This is reflected in household surveys of the Prime Minister, and 13 Circulars of Ministries that that show that benefits are spread thinly and have a relatively regulate issues related to social assistance. Several of the newer small impact on poverty. Figure 6.12 shows that benefits as a programs were introduced in a piecemeal fashion to mitigate share of consumption of households in the bottom quintile the impacts of certain reforms. There is no flagship program of the distribution are much lower in Vietnam than in other to provide a “backbone” program in contrast to those found middle-income countries. in other middle-income countries in the region such as China, From the Last Mile to the Next Mile 169 PART 2. THE NEXT MILE IS THE ROAD AHEAD Figure 6.12. Social assistance benefit levels in Vietnam compared to other countries 25 Benefits as a share of welfare among the lowest quintile 20 15 10 5 0 LIC LMIC UMIC HIC World Vietnam Source: WB ASPIRE 2018; Dutta and Sen (2018). Excludes health waivers. BOX 6.3. Vietnam’s Social Protection System Figure B.6.3.1. depicts the various elements of Vietnam’s social protection system today. It has evolved rapidly in the last two decades in terms of both the number and coverage of programs. The current system is based on Resolution 15, adopted by the Central Committee of the Communist Party of Vietnam in 2012. The resolution commits to developing the social protection system through expansion of social assistance, social services, social insurance, and poverty reduction programs. In 2017, the government approved the Master Plan for Social Assistance Reform and Development (MPSARD), which assessed the performance of the social assistance system and identified ways of making it more comprehensive and effective at addressing vulnerability and exclusion in the country. It also set out specific goals for expanding the coverage of existing social assistance programs, including expanding the coverage and benefits for the elderly, people with disabilities, children in need (orphans, abandoned children, severely disabled children, and especially disadvantaged children) or those facing emergencies or difficult circumstances. Finally, it proposed the phased introduction of a new policy to provide cash transfers to infants and young children under 4. Figure B.6.3.1. Key Elements of Vietnam’s Social Protection System Resolution 15 on Social Policies for 2012-2020 Social Protection Employment, Social Assistance Social Insurance Incomes and Minimum Basic Social Services Poverty Reduction Other Productive Employment Programs Compulsory and voluntary Insurance Social Care Centers and Care Services Other Programs (Supplement SI) Health Insurance and Health Unemployment Insurance Public Works Training Emergency Assistance Regular Allowances Vocational Training Information Access Clean Water Education Housing 170 2022 Vietnam Poverty and Equity Assessment Chapter 6. Building resilience and safety nets to guard against shocks Social insurance is a key tool to support The social insurance system also addresses short-term demographic change. risks such as unemployment or disability. Again, since coverage is limited to formal sector workers, most workers have The main objective of the social security pillar—which no recourse when subjected to these shocks. The COVID-19 in Vietnam consists primarily of its social insurance and crisis has exposed this major gap in the social protection health insurance systems—is to smooth individuals’ system, as millions of non-poor and mostly urban informal consumption across their lifetimes as they confront sector workers in sectors such as transport and tourism different risks. The pension system will be increasingly suddenly lost their incomes. While most of these workers important for Vietnam due to its rapid demographic and their households were covered by health insurance, they aging and, along with social pensions, will be a key tool had no insurance against the impact of the pandemic on their in managing the social consequences of this demographic employment and incomes. change (World Bank, 2021b). The existing system pays very generous pensions to public sector workers and more modest Lack of adaptive social protection to guard pensions to private sector workers. The retirement age has against shocks not changed for decades, while life expectancy at retirement age has risen sharply. In a positive step, the government has The response to the pandemic in 2020 also highlighted recently announced that the retirement age will rise gradually the weaknesses in social protection delivery systems, from 50 to 60 for women and from 55 to 62 for men. This specifically their inability to respond quickly to change should increase labor force participation and make the covariate shocks. Initially, the first relief package targeted 5 pension scheme more sustainable in the long run. Financial million informal sector workers, but reached only around 1 sustainability is important given that the scheme benefits million. A significant part of the problem was the inability to workers in the upper part of the income distribution and leverage administrative data and the national ID database in deficits could crowd out spending on redistributive programs. the way many other countries were able to do (Johnson and Palacios, forthcoming). Instead, the relief was implemented The biggest challenge, however, is coverage. The primary using manual processes that were difficult for potential scheme of the pension system is a compulsory insurance scheme beneficiaries to navigate, burdened local officials, and were that is part of Vietnam Social Security (VSS). In 2020, this difficult to monitor at the national level. As a result, far less scheme covered about one quarter percent of the working age of the population received Vietnam’s 2020 COVID cash population (14.6 million people). Its coverage extends almost response than in other countries in the region. Moreover, exclusively to workers in the formal sector—a serious constraint among respondents who received monetary assistance, the given the country’s high informality rate. To address this mode of payment was universally cash, with no instance of shortcoming, a voluntary contributory scheme was established digital payments (World Bank, 2021b). in the 2006 Social Insurance Law and SI Law amendments. However, this scheme has failed to attract a significant number In the context of the second COVID-19 relief package of informal sectors workers and covered only about 1.2 million that the government launched in September 2021, the workers by 2020. While coverage has risen gradually over time, Hue City government conducted a pilot. Its objective the pace of expansion is too slow to ensure that most of the was to register new beneficiaries using a mobile application elderly have a source of income in the next few decades. The created for this purpose. The digitalization of the manual government has set a coverage target of 60 percent of the labor registration process would reduce queues and physical force by 2030, but based on international experience and interaction with government officials who would approve or Vietnam’s own experience with health insurance, this target is reject applications based on certain criteria. Those approved as unlikely to be achieved without significant subsidies to make it beneficiaries would opt for a mode of payment that included affordable for informal sector workers. digital payment. District and provincial officials would be able to monitor progress in real time. As of November 2021, around 14,000 beneficiaries had been registered. From the Last Mile to the Next Mile 171 PART 2. THE NEXT MILE IS THE ROAD AHEAD Vietnam is an outlier in that it makes the majority of mass effort has been made with the intention of distributing payments in cash. The government has issued a decree 50 million IDs based on this database. The fragmented mandating a shift to digital payments for social assistance.51 social assistance databases, however, are highly decentralized, Figure 6.13 compares Vietnam to other countries based on as are the processes of registration and determination of the 2017 FINDEX survey. More recent government figures eligibility. Advanced systems such as the ISAS in Turkey show that more than 90 per cent of social assistance payments (see Chapter 8) have been able to establish data sharing are made in cash. The government has conducted several protocols with digital authentication using the unique ID e-payment pilots and has expanded digital payments for all number to make the registration, enrollment, and grievance social assistance beneficiaries in two provinces. Moreover, the redressal processes more efficient for both government and Ministry of Labor, Invalids and Social Affairs (MOLISA) has beneficiaries. Most importantly, it would make it possible to committed to scaling this up to the national level. Moving to respond quickly to both idiosyncratic and co-variate shocks. digital payments into bank accounts could not only increase Combined with institutional and financial arrangements, the efficiency of government-to-person (G2P) payments but this would help Vietnam move toward a truly adaptive social has the potential to dramatically increase financial inclusion, protection system. especially for women. Adequacy of minimum basic social services The government has also recognized the importance of data governance generally and social protection in Limited recent studies point to the positive impact of particular. The National Plan envisions linking six major health insurance on catastrophic health expenditure and databases, including the National Citizen database managed impoverishment in Vietnam. Studies from earlier in the by the Ministry of Public Security and the social insurance decade found that health insurance had very limited impact on database managed by the VSS.52 The latter is being centralized healthcare costs incurred by households (Nguyen et al., 2016; as part of the VSS ICT strategic plan. In principle, the MPS Ahmed et al., 2018; and Van Minh H et al., 2012). Vietnam database contains the demographic and biometric information implemented several key health insurance reforms in recent required to issue a unique national ID number. In 2021, a decades, notably in 2009 and 2015. The latest reforms in 2015 Figure 6.13. Vietnam lags other countries shifting to digital payments 100 90 80 70 60 50 40 30 20 10 0 Russian Federation Vietnam Mexico Egypt, Arab Rep. Thailand Ukraine United States Iran, Islamic Rep. Sweden Cambodia Ethiopia India China Kenya Turkey South Africa Mongolia Latvia Norway Received govt. payments in the past year (% age 15+) Received govt. payments: in cash only (% payment recipients, age 15+) Received govt. payments: into a financial institution account (% payment rerecipients, age 15+) Source: FINDEX survey (2017). 172 2022 Vietnam Poverty and Equity Assessment Chapter 6. Building resilience and safety nets to guard against shocks were aimed at increasing participation in health insurance, Modernizing the social protection system is needed. A encouraging participation through household subscription, decade ago, Vietnam’s social assistance coverage was in line and expanding entitlements for the insured (Thuong et al., with or higher than many of its East Asia and Pacific neighbors. 2021). A more recent study using the VHLSS found that However, it is now lagging behind them, as other countries the health insurance reforms in Vietnam has contributed to have since made more progress. In particular Malaysia and lowering the incidence of catastrophic health expenditure Indonesia’s social assistance coverage has increased, and and medical impoverishment (Thuong, 2021). More up-to- now both countries have wider coverage than Vietnam. The date studies are needed to ascertain the impact of the current following are some needed changes: health insurance system, given existing high levels of out-of- pocket payments. • More effective social assistance for poorer households: ‒ Increases in coverage and benefit levels to achieve greater poverty and inequality reduction; 6.6. Conclusion ‒ Increases in overall spending while consolidating the currently fragmented mix of programs to achieve While the odds of falling into extreme poverty are low, greater efficiency. about 2 million people still fell from a higher economic • Greater coverage of social insurance to protect all households class into poverty from 2016 to 2018. In 2010 and from all risks: 2020, about 6 percent of households also self-reported that ‒ Expanding coverage to non-poor informal workers their living conditions were worse than five years ago. These by blurring the line between social assistance and households report reasons for a lack of progress that can be insurance; workers make the contributions they can attributed to shocks such as job loss, accidents, illness, or afford while the state subsidizes the remainder. environmental shocks affecting agricultural output. Selected • A more flexible and adaptable delivery chain for both social reviews of shocks illustrate that households across the entire assistance and insurance: welfare distribution can be exposed either due to economic ‒ Better data collection and use to determine who is in impacts from COVID-19 or for environmental or health need and how this changes over time; reasons. However, for the poor who lack sufficient savings ‒ Adoption of digital payment systems so that and have fewer coping strategies, any exposure to shocks is payments occur quickly and safely, reaching those riskier and more consequential. who need them when they need them; ‒ Movement toward an adaptive social protection Thus, safety nets and insurance can play a role in system which can scale up when shocks strike: top- preserving economic gains and helping households avoid ups for existing beneficiaries, expansion to those falling into poverty traps. However, the current social newly in need. protection system is outdated and is not adaptive to offer adequate protection and safety nets especially for ‘moving targets,’ that is, those who are not registered as poor but can still be at risk of falling into poverty. Recent employment shocks from COVID-19 clearly illustrated the challenges to getting cash support into the hands of informal workers or labor migrants, who can be economically insecure but not registered in the social protection system. From the Last Mile to the Next Mile 173 PART 2. THE NEXT MILE IS THE ROAD AHEAD 6.7. References DC. © World Bank. https://openknowledge.worldbank.org/ handle/10986/34639 License: CC BY 3.0 IGO. Ahmed et al. 2018. Catastrophic healthcare expenditure and Skoufias, E. 2019. “Poverty and Vulnerability in the Ethiopian impoverishment in tropical deltas: evidence from the Mekong Delta Lowlands: Building a More Resilient Future”, Washington D.C.: region. https://doi.org/10.1186/s12939-018-0757-5 World Bank Group. Report of the World Bank Group and DfID. Climate Home (2018). Climate change is driving migration from World Bank, Washington D. Vietnam’s Mekong Delta. https://www.climatechangenews. Skoufias, E. and Baez, J. 2021. “Operationalizing Vulnerability to com/2018/01/11/climate-change-driving-migration-vietnams- Poverty”, World Bank Working Paper. mekong-delta/ Thuong, Nguyen Thi Thu. Catastrophic health expenditure and Cox D., and M. Fafchamps. 2008. “Extended Family and Kinship impoverishment in Vietnam: Evidence from national Household Networks: Economics Insights and Evolutionary Direction.” In Living Standards Surveys 2008–2018. 2021. https://doi. Handbook of Development Economics, Vol. 4, edited by T. P. Schultz org/10.1016/j.eap.2021.10.009 and J. Strauss. 3711–84. Amsterdam: Elsevier. Van Minh, Hoang et al. 2013. “Financial burden of household out- Dang, V. K., Doubre, C., Weber, C., Gourmelen, N., and Masson, of pocket health expenditure in Viet Nam: findings from the F. (2014). “Recent land subsidence caused by the rapid urban National Living Standard Survey 2002-–2010.” doi:10.1016/j. development in the Hanoi region (Vietnam) using ALOS InSAR socscimed.2012.11.028 data”, Nat. Hazards Earth Syst. Sci., 14, 657–674 Wagstaff, Adam. 2005. The Economic Consequences of Health Demirguç-Kunt, Asli, and Leora Klapper. 2013. “Measuring Financial Shocks. Policy Research Working Paper; No. 3644. World Bank, Inclusion: Explaining Variation in Use of Financial Services across Washington, DC. © World Bank. https://openknowledge. and within Countries.” Brookings Papers on Economic Activity worldbank.org/handle/10986/8307 2013: 279–340. The World Bank Group and the Asian Development Bank. 2020. Dutta, Puja Vasudeva, and Shonali Sen. 2018. Vietnam: Building an Climate Risk Country Profile: Vietnam (2020). effective and sustainable social assistance system. World Bank note. World Bank. 2016. Vietnam 2035: Toward Prosperity, Creativity, Hamilton, W. 1964. “The Genetical Evolution of Social Behaviour I.” Equity, and Democracy. Washington, DC: World Bank Journal of Theoretical Biology 7: 1–16. ------------- 2021a. Vietnam: Adapting to an Aging Society. Washington, Johnson, Doug and Robert Palacios (forthcoming). Scaling up cash DC: World Bank. © World Bank. transfers during the COVID-19 pandemic: what might explain ------------- 2021b. A Year Deferred – Early Experiences and Lessons from differences in coverage? COVID-19 in Vietnam. La Ferrara, E. 2011. “Family and Kinship Ties in Development: An ------------- 2021c. World Bank East Asia and Pacific Economic Update, Economist’s Perspective.” In Culture, Institutions and Development: April 2021: Uneven Recovery. Washington, DC: World Bank. © New Insights into an Old Debate, edited by J. P. Platteau and R. World Bank. Peccoud. New York: Routledge. World Health Organization. 2005. Designing Health Financing Systems Minh DHT, Van Trung L, Toan TL. (2015). “Mapping Ground to Reduce Catastrophic Health Expenditure. Technical Briefs for Subsidence Phenomena in Ho Chi Minh City through the Radar Policy Makers. https://www.who.int/health_financing/pb_2.pdf Interferometry Technique Using ALOS PALSAR Data”. Remote Sensing, 7(7):8543-8562 Narloch, Ulf, Bangalore, Mook. 2016. “Environmental Risks and Poverty: Analyzing Geo-Spatial and Household Data from Vietnam”. Policy Research Working Paper; No. 7763. World Bank, 6.8. Notes Washington, DC. World Bank. 39  Shocks can be non-COVID-19 related, such as illness or death in Nguyen, H., Ivers, R., Jan, S. and Pham, C., 2017. Analysis of out-of- the family. pocket costs associated with hospitalised injuries in Vietnam. BMJ 40  Before the 4th wave, Vietnam had COVID-19 under control. However Global Health, 2(1), p.e000082. in April 2021, the Delta variant entered the country that was almost Nguyen, Thang. 2022. Household savings in Vietnam: Evidence from completely unvaccinated. In one month, the number of cases exceeded the Vietnam Household Living Standard Survey. Mimeo. the number of cases from the entire previous year. At the time of this Pearlman, S. 2010. Flexibility matters: Do more rigid loan contracts report, developments are still unfolding. reduce demand for microfinance? CFA Working Paper No. 2010/10. 41  https://climateknowledgeportal.worldbank.org/country/vietnam/ Rentschler, Jun; de Vries Robbé, Sophie; Braese, Johannes; Nguyen, vulnerability Dzung Huy; van Ledden, Mathijs; Pozueta Mayo, Beatriz. 2020. 42  However, some past results find that the poor are more exposed Resilient Shores : Vietnam’s Coastal Development Between Opportunity and Disaster Risk. World Bank, Washington, to environment risk, and that there is a positive correlation between 174 2022 Vietnam Poverty and Equity Assessment Chapter 6. Building resilience and safety nets to guard against shocks districts being high-risk and having higher poverty rates using data on environmental risk for 2000-2010 and a poverty map for 2010 (Narloch and Bangalore, 2016). 43  The six environmental risks or uncertainties considered include air pollution, and variation in wind, precipitation, and temperature, the share of the population exposed to high flood risk, and the SPEI drought indicator. 44  The analysis in this section relies upon commune-level variables to help estimate the degree of idiosyncratic and covariate risk. Data are not available for all communes, so not all of the VHLSS is used in this particular analysis. 45  In addition, the modelling is implemented using a vulnerability tool developed by the Equity Policy Lab of the World Bank. 46  Note, Skoufias and Baez use the term “poverty-induced vulnerability” where this report uses “chronic poverty.” 47  World Bank COVID-19 monitoring surveys. 48  Employee Service Centers (ESC) manage the unemployment insurance program and administer vocational training and job placement programs. However, these services are minimal in most of the 98 ESCs and focused mainly on formal sector workers with low skill levels. See World Bank (2020). 49  The most recent estimate of 0.66 percent of GDP is for 2013. See Dutta and Sen (2018). 50  They were raised from VND270,000 to 350,000 based on Decree 20 in March 2021. 51  Government of Vietnam, Decree 20/2021. 52  The others are the National Land Database (managed by Ministry of Natural Resources and Environment); National Database on Business Registration; National Database on General Statistics of Population (managed by Ministry of Planning and Investment); National Database on Finance (managed by Ministry of Finance). From the Last Mile to the Next Mile 175 PART 2. Chapter 7. Fiscal policy, financing future investments and building the middle class Chapter 7. Fiscal policy, financing future investments and building the middle class Key Messages • Fiscal policy can play a critical role in both driving Vietnam toward high-income status and doing so in an inclusive manner help 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 Last Mile and Next Mile constraints today, such as a modern social protection system and strengthening of National Targeting Programs (NTPs). 176 2022 Vietnam Poverty and Equity Assessment Chapter 7. Fiscal policy, financing future investments and building the middle class 7.1. Introduction – This chapter asks how different households benefit measuring fiscal impacts (or do not) from fiscal policy in Vietnam. Using the Commitment to Equity (CEQ) framework (see Box 7.1) Fiscal policy is one of the few instruments that allows two key questions to be answered. First, who pays a governments can use to reduce inequality in the particular tax or receives a particular benefit? For example, short term. It is not just that fiscal policy finances public how much VAT is paid by poorer households and how investments which can promote growth as well as reduce much by richer households, both in Vietnamese dong and poverty and inequality in the long term; it can also affect as a percentage of their market income? Second, what is the household income distribution today. This chapter examines net impact of all taxes and transfers on different households? the relationship in Vietnam immediately before COVID-19 For example, which households pay more in taxes than they between fiscal policy and poverty and inequality reduction. receive in benefits, and which pay less? How does this net Different households pay various taxes and benefit from fiscal impact on households affect poverty and inequality in public spending in different ways. The net effect determines Vietnam? 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. BOX 7.1 The Commitment to Equity methodology To determine the impact of fiscal policy on household welfare, the Commitment to Equity (CEQ) framework of fiscal incidence analysis is used (Lustig, 2018).53 Under this framework, household income is assessed at different stages, as outlined in the figure below. Figure B.7.1.1. Definition of income concepts and the role of fiscal instruments + Market Income Contributory social insurance old-age pensions (net of contributions) Market Income Plus Pensions − (Pre-fiscal income) Direct taxes: PIT, property tax, border tax + Net Market Income Direct cash transfers: NAF; bread conpensation scheme, Zakat, non-contributory pensions, others. + − Disposable Income Indirect subsidies (direct and Indirect taxes: GST; SST tobacco, indirect impacts): electricity; water alcohol, sodas, mobile phones) Consumer Income In-kind transfers net of user fees: health; education + Final Income Source: Adapted from Lustig (2018). Continued > From the Last Mile to the Next Mile 177 PART 2. THE NEXT MILE IS THE ROAD AHEAD BOX 7.1. The Commitment to Equity methodology (Continued) As a starting point, a household’s market income is the total income it received from wages and salaries, rents and dividends, private transfers and remittances, and pension income.55 This is the income a household generates before it encounters the fiscal system and is also called its pre-fiscal income. Some households pay personal income taxes, property taxes, and capital gains taxes, which reduces their market income. Some households receive direct transfers from the government as part of the social safety net, which increases their market income. The net effect after direct taxes and direct transfers is a household’s disposable income: how much money it has to spend on goods and services, or to save. Second, when a household buys goods and services, it pays indirect taxes (such as VAT or special excises on particular goods), which means its disposable income buys less, but it might also benefit from indirect subsidies (such as cheaper electricity) which means its disposable income buys more. How much of different goods and services a household can afford to buy, after considering both indirect taxes and subsidies, is called consumable income. When considering only cash-based fiscal instruments, this is also a household’s post-fiscal income. Finally, a household may also use public services such as send their children to a public school or visit a health center or hospital. In this case they are benefitting from public spending not in a cash sense but in-kind. Including this non-cash spending results in a household’s final income (the post-fiscal income if non-cash spending is included). (See Box 7.2 for a discussion of data limitations and Annex 7.1 for a summary of key modelling assumptions.) 5 4 A key objective of this chapter is to ask whether pre- Revenues COVID-19 fiscal policy in Vietnam made the household income distribution more equal as it moved from Just over half of Vietnam’s tax revenues are included market income to consumable and final income, and in the analysis. Tax revenues accounted for 18.4 percent of which instruments contribute to any such effect. The GDP in 2018 (Table 7.1). The items in bold indicate taxes results are put in international perspective and insights from which are included in the current analysis, representing 57 international experience are summarized, while specific policy percent of total tax revenues. Indirect taxes (such as VAT and recommendations are discussed in the concluding chapter of excises) make up half of tax revenue, while direct taxes such as the report. The data and methodology used in to produce corporate and personal income tax contribute less; corporate these results are discussed further in Wai-Poi et al (2022). taxes are still an important but declining revenue source. Indirect taxes include VAT and special excises on 7.2. Vietnam’s pre- tobacco and alcohol. Vietnam’s standard VAT rate is 10 COVID-19 fiscal policies: percent, but a number of goods and services are subject to a reduced 5 percent rate and others are exempt. An excise rate how is money raised and of 75 percent of the pretax factory price is applied to domestic how is it spent? tobacco products, with VAT then levied on the retail price. Tobacco manufacturers and importers pay an additional 2 This section provides an overview of fiscal revenues and percent of the pretax ex-factory price, earmarked to finance expenditures in Vietnam for 2018, which is calibrated to the Fund for Prevention and Control of Tobacco Harms; as the 201855 VHLSS household survey data used to determine of May 2019, the total tax rate for domestic cigarettes equaled fiscal incidence. The treatment here is cursory and intended 95 percent of the factory price.56 as a high-level overview of revenue and spending magnitudes and breakdowns, and does not provide a detailed description by tax instrument or social sector; see the companion fiscal incidence report for more details (Wai-Poi et al., 2022). 178 2022 Vietnam Poverty and Equity Assessment Chapter 7. Fiscal policy, financing future investments and building the middle class Personal income tax (PIT) has seven tiers and with Expenditures progressive rates increasing by tier; capital income taxes vary by source of income. Vietnam’s PIT framework does This chapter focuses on central government spending; not have a tax-free threshold, but there are significant personal data limitations mean important sub-national spending and dependent deductions and allowances, and rates are on social assistance is excluded, which is material in a relatively low for lower tiers before increasing to a top rate highly decentralized fiscal system such as Vietnam’s. of 35 percent. Taxes on capital income vary by source but in Revenues in Vietnam are largely centralized, allowing this all cases not exceeding 20 percent (for capital gains and some chapter to capture a considerable deal of them. However, property taxes) and in most cases significantly lower. spending in Vietnam is quite decentralized, meaning that the focus on central government spending in the current analysis does miss important social spending decisions which occur Table 7.1. Vietnam government revenues, 2018 at the local level, in particular social assistance. Nonetheless, both central and local health and education expenditures TRN DONG % GDP have been included in this chapter, and they account for the Total revenues & grants 1,364 24.6 majority of social spending analyzed. Data limitations prevent Taxes 1,022 18.4 a more comprehensive review. Box 7.2 discusses spending Oil revenues 66 1.2 decentralization in Vietnam. CIT 49 0.9 Vietnam’s total central public expenditures in 2018 Natural resources tax 17 0.3 were around 19 percent of GDP;57 of this, 6.8 percent Non-oil taxes 956 17.2 of GDP was spent on social and non-social expenditures Taxes on goods & services 509 9.2 included in the analysis (Table 7.2). Education was the VAT 344 6.2 largest expenditure, followed by health. Direct cash transfers Excise 96 1.7 comprise a small part of total expenditures, 11 percent of Natural resources tax 38 0.7 those analyzed. Administrative data on electricity subsidies were not available but have been estimated at 26 trillion dong Environment 47 0.8 protection tax in the current work. Direct taxes 296 5.3 Household composition CIT 202 3.6 PIT 94 1.7 The composition of a household has a strong influence Taxes on international trade 87 1.6 over how much it is likely to benefit from certain Other taxes 64 1.2 types of spending. For example, households with school- Registration tax 32 0.6 aged children will benefit from education spending when Agricultural land use tax 0 0.0 kids go to school, and those with seniors are more likely to receive pensions. Households with children or elderly Land & housing tax 2 0.0 are also potentially eligible for social assistance targeted at Lottery 29 0.5 these particular categories. Households with neither cannot Non-taxes 334 6.0 benefit from these types of social spending. Figure 7.1 shows Fees & charges 38 0.7 how household composition changes across the income Land rents 28 0.5 distribution. Poorer households tend to be younger, with LUR assignment 148 2.7 more children. As households get richer, they have fewer children, with half or more of the richest two deciles having Other non-taxes 121 2.2 none at all. At the same time, there are almost no elderly in Grants 8 0.1 the poorer half of the distribution (and relatively few in any Source: MOF, GSO, IMF and World Bank staff calculations. household, although this will change as the country ages). From the Last Mile to the Next Mile 179 PART 2. THE NEXT MILE IS THE ROAD AHEAD Table 7.2. Vietnam central government Education expenditures, 2018 Differences in demographics and enrollment rates across   TRN DONG % GDP the income distribution drive the incidence of public Total expenditures 1,435 25.9 education spending. Poorer households generally have Expenses 1,051 19.0 more children to benefit from public education, but richer Total spending analyzed 376 6.8 households’ children are more likely to stay enrolled at higher Direct Transfers 40 0.7 (and more expensive) education levels. For example, nearly two million pre-school and primary students come from the Education 220 4.0 poorest income decile of households, compared to less than Pre-school 34 0.6 half a million from the richest decile, and lower secondary Primary 63 1.1 students from poorer households significantly outnumber Lower secondary 49 0.9 richer ones. However, upper secondary students are roughly Upper secondary 22 0.4 evenly split between the bottom and top half of the income TVET 31 0.6 distribution while the upper half dominates tertiary enrollment (Wai-Poi et al., 2022). Tertiary 21 0.4 Health 89 1.6 There are also significant private expenditures on In-patient 43 0.8 education even at compulsory levels and in public schools, Out-patient 46 0.8 which offsets the benefit of public spending. The richest Subsidies 26 0.5 quintile far outspends the poorest households in education, Electricity 26 0.5 primarily in the categories of extra courses and tuition, Notes: Education total expenditures are based on the 2018 administrative data. spending almost four times as much than the lowest quintile Breakdowns by education level are based on 2013 UNESCO data, pro-rated to the (Wai-Poi et al., 2022). For the purposes of this chapter, out-of- 2018 total expenditures. Modelled estimates come close to the total level (modelled pocket expenses which are received by the school are treated is 229 trillion compared to 220 trillion admin) and pro-rated levels except for TVET as co-payments—effectively a tax—as they reduce the value (overestimated) and tertiary (underestimated). Health total expenditures are based of the education benefit, and the payments that go into the on the 2018 administrative data; WHO 2018 data is 154 trillion. The table uses fiscal system. Expenses which go to the private sector, such as 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 for uniforms or books, are excluded from the fiscal incidence accounts for 92 percent of administrative expenditure analyzed. analysis; they are legitimate expenses but because they are not Source: Ministry of Finance, Ministry of Education, WHO Global Health paid to a public entity, they are not part of the fiscal system. Expenditure Database, GSO, IMF, and World Bank staff calculations. Figure 7.1. Demographic composition of households by per capita consumption decile 100% Share of households (%) 80% 60% 40% 20% 0% 1 2 3 4 5 6 7 8 9 10 Income decile No children or elderly No children, with elderly 1 child 2 children 3+ children Source: 2018 VHLSS and World Bank calculations. 180 2022 Vietnam Poverty and Equity Assessment Chapter 7. Fiscal policy, financing future investments and building the middle class Health Out-of-pocket health expenditures are high in Vietnam, reducing the net value of public health services. Despite In 2020, health insurance covered 91 percent of population the high level of health insurance coverage, out-of-pocket costs (MOH, 2021). Vietnam introduced new reforms to the Health remained high at 45 percent of current health expenditure in Insurance Law in 2015 to improve universal health coverage. 2018 (see Chapter 6). In 2016, 9 percent of the population Support was shifted from supply-side to demand-side subsidies on incurred catastrophic health expenditure measured as out- health insurance premiums for various groups such as poor and of-pocket spending exceeding 10 percent of household near poor, among others. These reforms were aimed at expanding consumption or income. As with education expenses, those participation in health insurance, promoting enrollment in which are paid to a public entity are treated as co-payments or a health insurance through household subscription, and increasing tax. Private expenses such as medicines from a pharmacy would entitlements to the insured (Thu et al., 2020). As a consequence not be included. of the reforms, most households are covered and coverage does not vary significantly across the income distribution, although the nature of coverage does (Wai-Poi et al., 2022). BOX 7.2. Limitations to CEQ methodology in Vietnam Decentralization is an important feature of fiscal spending and decision making in Vietnam. Vietnam has 63 provinces that differ significantly in size, economic structure, and level of development. Their average size is relatively small compared to that in many other countries, with an average of 1.5 million inhabitants per province and 8,500 per municipality. These ratios are much smaller than the OECD average (3 million inhabitants per region and 37,800 inhabitants per municipality) and in other more populous developing countries, such as China and India, but also Malaysia and Mexico (OECD, 2020). In turn, this means that smaller provinces are not able to take advantage of economies of scale and efficiently solve problems because of externalities (Xu, 2011). Along with an increasing number of provinces, Vietnam has embarked on extensive spending decentralization since the mid-1990s; current levels are significantly higher than those of its peers. Within the set of rules and norms applied across the country, subnational governments have at their discretion an increasing share in total expenditures, from 26 percent in 1992 to close to 60 in 2020. They are also responsible for over three-quarters of the country’s public investment. The subnational share of spending is higher in key sectors such as education (90 percent), health (80 percent), and transport (65 percent), some of which are key for fiscal incidence analysis. Subnational spending as a share of total government spending is now significantly higher than the international average (24 percent) and the lower middle-income average (20 percent). Subnational investment spending is almost double the averages in the unitary countries (34 precent) and overall (39 percent); see figure below. There has not been a corresponding increase in subnational revenues; “balancing” transfers from central government make up the growing financing gap. The proportion of decentralized revenue to total local revenue has declined over time as there is very little revenue autonomy at the subnational level. Balancing transfers in turn have constituted a growing share of local financing; 47 out of 63 provinces are dependent on balancing transfers from the central government. The current balancing transfers system is equity-based (using “registered population” as the key criterion for allocation). As resources are increasingly transferred to the subnational levels, the central government lacks necessary resources for nationally important infrastructure investments. At the same time, the leading urban areas also lack the resources required to meet the infrastructure needs to accommodate economic growth. Fiscal decentralization creates challenges to obtaining provincially disaggregated expenditure and revenue data. A subnational fiscal data based was created in 2014 and informed the World Bank’s previous household fiscal incidence analysis study (World Bank, 2017). However, this database has not been updated, which affects the analysis in this chapter. From the Last Mile to the Next Mile 181 PART 2. THE NEXT MILE IS THE ROAD AHEAD 7.3. Vietnam’s pre- 5.2 percent of the poorest decile’s average income to COVID-19 fiscal incidence: 4.5 percent for the richest. The progressive nature of the aggregate fiscal cash impact is driven by direct taxation, who pays and who part of which is personal income tax but most of which are benefits? contributions to social insurance, particularly social security contributions. Also of note is that direct transfers are Net impact of fiscal policy on households by relatively small albeit concentrated in the poorest decile and income decile representing just over 10 percent of their income on average. Most households pay more by way of taxes and co- When in-kind health and education spending is payments than they receive in cash benefits. This section included, the poorest four deciles are net beneficiaries of presents the key results showing which households taxes and public spending, with the poorest decile receiving pay which taxes and receive which benefits. Figure 7.2 net benefits of 36 percent of market income (Figure 7.3). considers the aggregate impact of fiscal policy, excluding The majority of the in-kind benefits comes from education, non-cash in-kind health and education spending. The first which is worth around 13–22 percent of market income feature of note is that all but the poorest decile of households for the poorest two deciles. However, even when health and are net contributors in cash terms into the fiscal system (and education spending is included, the net payment into the even the first decile sees net benefits of less than 5 percent of fiscal system for richer households remains relatively similar. their already low incomes). This contribution represents 10 In education, this reflects the number of richer households percent or less of household market income for deciles 2-4 who do not have children (or who might be enrolled in private and then rises above 10 percent for all other deciles, reaching schooling). In health, it reflects high out-of-pocket costs to 23 percent and 20 percent for the richest and next richest 10 access public health; these reduce the benefit of health services percent of people, respectively. Indirect taxation—largely to all households but actually outweigh the value of benefits VAT – is modestly regressive in relative terms, ranging from for the richest decile. Figure 7.2. Impact of fiscal policies by per capita consumption decile (excluding in-kind spending Panel A: Absolute incidence (total tax collected and spending received) Panel B: Relative incidence (tax collected and spending received as a percentage of household market income) 500 15 Monthly VND (thousands) 0 10 Percent of market income 5 −500 0 −1,000 −5 −10 −1,500 −15 −2,000 −20 −25 −2,500 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 Per capita market income decile Per capita market income decile SI Direct taxes VAT Excise Environment Direct transfers Subsidies Net effect Note: See Annex 7.1 for key assumptions. Source: 2018 VHLSS and World Bank calculations. 182 2022 Vietnam Poverty and Equity Assessment Chapter 7. Fiscal policy, financing future investments and building the middle class Figure 7.3. Impact of fiscal policies by per capita consumption decile (including in-kind spending) Panel A: Absolute incidence (total tax collected and spending received) Panel B: Relative incidence (tax collected and spending received as a percentage of household market income) 1000 40 30 500 Monthly VND (thousands) 20 Percent of market income 0 10 −500 0 −1000 −10 −20 −1500 −30 −2000 −40 −2500 −50 −3000 −60 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 Per capita market income decile Per capita market income decile SI Direct taxes VAT Excise Environmental Direct transfer Subsidies Education Education OOP Health Health OOP Net effect Note: See Annex 7.1 for key assumptions. Source: 2018 VHLSS and World Bank calculations. Given the pattern of benefit and burden across the was 7.5 percent. Figure 7.4 shows how each fiscal instrument household income distribution, what is the impact contributes to poverty reduction or increases poverty. Direct of fiscal policy on poverty and inequality? cash transfers contribute 1.1 points of poverty reduction (that is, it reduces poverty by 1.1 points). Direct taxes— The short-term value of transfers received by the poor mostly social security and health insurance contributions— are less than the taxes they pay – largely indirect taxes reduce this contribution by 0.4 points, meaning that on consumption. However, the longer-term benefits they poverty measured at disposable income is 0.6 points lower get from public spending on health and education are likely than at market income. However, while electricity subsidies to be greater than the taxs paid now. At market incomes, the make a minor contribution to poverty reduction, excises, LMIC ($3.20/day 2011PPP) poverty rate in Vietnam in 2018 environmental taxes, and particularly VAT more than offset Figure 7.4. The impact of fiscal policy on poverty by fiscal instrument Direct Direct taxes Indirect taxes 1.2 transfer Indirect subsidies Percentage Point Reduction 1.0 0.8 0.6 0.4 0.2 0.0 −0.2 −0.4 −0.6 Cash transfer PIT Other direct SSC/HI After direct Electricity VAT Excise Environmental After direct taxes and indirect subsidy tax and indirect taxes and transfers taxes (consumable income) Note: Marginal effects on poverty reduction are presented. A positive bar represents contribution to poverty reduction. A negative bar represents an increase in poverty. 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. Source: 2018 VHLSS and World Bank calculations. From the Last Mile to the Next Mile 183 PART 2. THE NEXT MILE IS THE ROAD AHEAD the impact of direct transfers and result in overall poverty Impact and cost-effectiveness of fiscal policy being higher after all fiscal policies are considered. That is, instruments on inequality poverty when measured with market incomes is 7.5 percent; after cash transfers and direct taxes it falls to 6.9 percent; Vietnam raises more revenue from the least progressive after electricity subsidies and indirect taxes it increases to 8.0 taxes and spends the most on the least progressive percent, or 0.5 points higher than before fiscal interventions. expenditures, indicating the potential for fiscally neutral However, spending on health and particularly education not reforms which would reduce inequality further. The only benefits poorer households more today, as discussed previous section examined how much each fiscal instrument next, it also benefitting them in the long-term. The returns increased or reduced inequality. The following chart considers to education in terms of better adult health outcomes and the progressive instruments and asks which is the most cost- higher earnings are well-documented and poorer children effective in reducing inequality. That is, given the money disproportionately benefit (for example, see Holla et al 2021 spent on each instrument (or the revenue raised), how much and World Bank (forthcoming)). is inequality reduced? Non-progressive instruments such as indirect taxes are excluded. In Figure 7.5, personal income The progressive nature of in-kind spending, particularly taxes decreased inequality by about 0.5 points and other on education, means that the fiscal system reduces direct taxes by 0.2 points while the impact of social insurance household inequality. While the fiscal system results in a contributions (e.g., social security, health and unemployment small increase in poverty, it does help redistribution, and the insurance) was much larger (1.6 points). Figure 7.6 presents Gini Index measure of inequality is reduced by nearly 3 points the results in a different light. First, a cost-effectiveness index after accounting for all cash-based policies and by 5 points after is constructed (on the left-hand axis), which is the decrease including non-cash health and education spending (Figure in inequality per dong raised (taxes) or spent (spending).58 7.5). The key progressive instruments are direct taxes and Second, this is compared to the size of spending (or revenue transfers, which reduce inequality by nearly 3 points, driven in the case of a tax) on that instrument (right-hand axis). mostly by social security and health insurance contributions, Personal income taxes reduce inequality the most per dong and education spending, which contributed nearly another 2 of revenue raised. Other direct taxes are half as effective at points of inequality reduction. Indirect transfers (electricity reducing inequality, and social insurance contributions are subsidies) and indirect taxes were largely neutral and had little only one-third as cost-effective. However, social insurance effect on inequality. contributions are responsible for five times more revenue contributions as both personal income and other direct Figure 7.5. The impact of fiscal policy on inequality by fiscal instrument 6.0 5.0 Percentage Point Reduction 4.0 3.0 2.0 1.0 0.0 Cash PIT Other direct SSC/HI After direct Electricity VAT Excise Environmental After Education Health After all transfer taxes and indirect subsidy tax direct and taxes and taxes and transfers indirect taxes transfers (consumable (final income) income) Note: Marginal effects on poverty reduction are presented. A positive bar represents contribution to poverty reduction. A negative bar represents an increase in poverty. 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. Source: 2018 VHLSS and World Bank calculations. 184 2022 Vietnam Poverty and Equity Assessment Chapter 7. Fiscal policy, financing future investments and building the middle class taxes combined, and so have the largest absolute impact on Capital Index (World Bank, 2020) can be disaggregated across inequality (Figure 7.6). If the revenue mix were rebalanced to household welfare quintiles (D’Souza, Gatti, and Kraay, bring more in from direct taxes and less in from social security 2019). Figure 7.7 shows how well each quintile scores on the contributions, inequality would be further reduced at no different HCI components and overall relative to the richest fiscal cost. Alternatively, if Vietnam were looking to increase quintile.59 For example, the poorest quintile stays in school tax revenues, broadening the income tax base would both for only 80 percent of the time that the richest quintile does. bring in more money and reduce inequality at the same time. Moreover, the average test scores for the poorest quintile are Similar results can be seen on the expenditure side. Direct only 87 percent those of the richest quintile, so even when cash transfers are nearly three times more effective at reducing they do stay in school, they do not learn as much. As a inequality than less well-targeted electricity subsidies, yet consequence, when their years of schooling are adjusted for almost as much is spent on the subsidies as the transfers. how much they learn, the poorest quintile has only 71 percent of the educational attainment as the richest quintile. Stunting Value of in-kind health and education spending outcomes are even more unequal. can also take into account quality Adjusting the value to a household in a particular Poorer children do not benefit from education and quintile of health and education spending by their health spending to the same extent as richer children. Just human capital outcomes means inequality is reduced by over two-fifths of the total reduction in inequality through 15 percent less. The value of health and education spending Vietnam’s fiscal system comes from health and education to each household can then be adjusted according to that spending (Figure 7.5). These benefits are in-kind services household’s S-HCI quintile result relative to that of the received, and the benefit level has been set at the cost to the richest quintile, reflecting the lower human capital outcomes government of delivering these services. However, the value to that household achieves. After making this adjustment, the households may not be the same as the cost of providing them. contribution of health and education benefits to inequality Quality may vary from province to province or between urban reduction falls by around one-third, from 2.3 points to 1.6 and rural locations, for example. Some classrooms may have points, and the overall redistributive effect of Vietnam’s fiscal better equipment than others, smaller teacher-pupil ratios or system declines by 15 percent, to around 4.4 points in total more textbooks. To test how sensitive this key contributor to (Figure 7.8). 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 Figure 7.6. Magnitude of different fiscal instruments and their cost-effectiveness at reducing inequality 15 8% 7% 12 Cost-effectiveness Index 6% Size relative to GDP 9 5% 4% 6 3% 2% 3 1% 0 0% PIT Other direct Social insurance Cash transfer Electricity Education Health taxes subsidies Effectiveness % GDP Note: Cost-effectiveness index is the marginal contribution to inequality reduction (points of Gini changed) divided by total spending. Source: 2018 VHLSS and World Bank calculations. From the Last Mile to the Next Mile 185 PART 2. THE NEXT MILE IS THE ROAD AHEAD Figure 7.7. Socioeconomic-disaggregated Human Capital Index (S-HCI) by component 100 Quintile score relative to Quintiles 5 80 60 40 20 0 Q1 Q2 Q3 Q4 Q5 Q1 Q2 Q3 Q4 Q5 Q1 Q2 Q3 Q4 Q5 Q1 Q2 Q3 Q4 Q5 Q1 Q2 Q3 Q4 Q5 Q1 Q2 Q3 Q4 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. Figure 7.8. The impact of fiscal policy on inequality after adjusting value of in-kind benefits for S-HCI differences 6 5 4 Points 3 2 1 0 Direct taxes and Indirect taxes and In-kind benefits Discounting education Discounting health Final impact transfers transfers Note: Marginal effects on inequality reduction are presented. A positive bar represents contribution to inequality reduction. A negative bar represents an increase in inequality. Health and education discounting are as per the note to Table 7.2. Source: 2018 VHLSS, D’Souza, Gatti and Kraay (2019) and World Bank calculations. A longstanding and persistent issue with general Equalizing opportunities requires public policies education is financing inequity for public schools, and investments that compensate for the gap in especially those in the most lagging ethnic minority private investments between children of privileged and mountainous areas. Schools in remote and poor areas and underprivileged parents (Narayan et al., 2018). are on average underfunded. Ethnic minorities and low- Government has a role to play in leveling the playing field income students are more likely to attend low-quality school so that Vietnamese children regardless of circumstances can facilities that lack qualified teachers, instructional materials, grow up with equal access to educational opportunities. If technology, critical facilities, and physical maintenance. The the circumstances of a child’s upbringing are linked to their absence of these facilities negatively impacts learning ability opportunities in life, then certain groups of children will not and exacerbates existing inequities in student access and be able to realize their full potential. Globally, as well as for learning outcomes. East Asia and Pacific countries, public spending on education 186 2022 Vietnam Poverty and Equity Assessment Chapter 7. Fiscal policy, financing future investments and building the middle class increases with income levels, and relative economic mobility pro-growth and pro-poor investments for the future. Chapter tends to be higher as well in countries with higher spending 8 further discusses key fiscal policy priorities. as a share of GDP. The evidence thus suggests that richer countries tend to have higher relative mobility in education, High-income countries tend to have the most on the average, because they tend to invest more (relative to progressive mix of fiscal policy for the poor the size of their economy) in human capital development to equalize opportunities (Narayan et al., 2018). 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 7.4. Conclusion - Insights help further reduce poverty (Figure 7.9). The previous from international section showed that Vietnam’s fiscal policy in 2018 increased short-term poverty slightly by 0.6 points once all direct and experience indirect taxes and transfers are taken into account, exluding the many beneficial long-term poverty reduction effects of Vietnam can learn from countries that achieve greater public investments in health and education. At the same redistribution through fiscal policy. In international time, inequality is reduced by 2.9 points when only cash comparison, particularly compared to other lower middle- instruments are considered and by 5.2 points when in-kind income countries, Vietnam is one of the better performers spending on health and education are also included. How with respect to poverty and around average with respect do these results compare to other lower-middle-income to inequality. This section concludes by benchmarking countries? The small increase in poverty from fiscal policy is Vietnam’s performance with other countries in the CEQ a relatively good outcome for a lower-middle-income country database, and then discussing how some countries are able to based on the methodology, which average a 1.5 point increase achieve more pro-poor outcomes with different combinations and some of which see 5 or 7 point increases. of fiscal instruments. Fiscal policy can be used to both reduce inequality in Vietnam today and finance sustainable Figure 7.9. Impact of fiscal policy on poverty by country income level 8 6 4 Percentage Point Change 2 0 −2 −4 −6 −8 −10 −12 Armenia Iran, Islamic Rep. Ecuador Mexico Jordan Russian Federation Mexico Mexico Belarus Russian Federation Peru Peru Venezuela, RB Dominican Republic El Salvador Ghana Sri Lanka Indonesia South Africa Colombia Argentina Costa Rica Colombia Guatemala Albania Tunisia Honduras Vietnam Bolivia Nicaragua Tanzania Uganda Burkina Faso Poland Panama Chile Uruguay Paraguay Ethiopia Brazil High income Upper middle income Lower middle income Lower income Market to Disposable Market to Consumable Note: Pensions treated as deferred income. Poverty is measured at the relevant World Bank international poverty line for each country’s income level: $1.90 for low- income countries, $3.20 for lower middle-income countries, $5.50 for upper middle-income and high-income countries. Source: CEQ and World Bank databases and World Bank calculations. From the Last Mile to the Next Mile 187 PART 2. THE NEXT MILE IS THE ROAD AHEAD The impact of fiscal policy on reducing inequality is Some degree of indirect taxation is still included in the price, as more mixed across economies from different levels of inputs into production are often bought on the formal market development. The reduction in inequality from Vietnam’s (such as electricity), but VAT is not applied to the final good, fiscal policy is around average for lower- middle-income meaning the effective tax rate is lower. As Bachas, Gadenne, and countries. When excluding in-kind health and education Jensen (2020) show for 31 developing countries, the share of spending, only Guinea does not see inequality decrease; in all household consumption spent in the informal sector declines other countries inequality declines. The 2.9-point decrease in with income; the rich face an effective rate twice as high as that Vietnam ranks in the middle of lower-middle-income countries of the poor on average. This phenomenon reverses the revenue- (Figure 7.10 Panel A). Similarly, when in-kind spending is equity trade-off of exemptions and preferential rates: greater included (Figure 7.10 Panel B), Vietnam’s ranking among these informality means more progressive indirect taxation but lower countries remains unchanged, although it gets closer to the overall collection rates. best performers. Progressive combinations of indirect taxes and Indirect taxation tends to be main source of direct transfers burden for the poor The mix of taxes used by a country has a strong influence Of the 33 low, lower and upper-middle-income countries on how progressive its fiscal policy will be. Direct taxes, out of the 48 countries for which data are available such as corporate and personal income taxes, property in the CEQ cross-country database, the increase in taxes, and inheritance taxes, are powerful tools for reducing poverty occurs when moving from disposable income inequality. They are paid largely by richer households, to consumable income (Figure 7.9).60 That is, the burden of directly reducing the inequality arising from market incomes. indirect taxes such as VAT and excises can cause households to But part of the revenues can also be used to finance pro- become impoverished if there is not an offsetting benefit from poor investments, such as in human capital and physical direct cash transfers. However, in all high-income countries infrastructure, thus indirectly reducing inequality in the except for Romania and around half of upper middle-income future. Indirect taxes, such as VAT and excises, are much less countries, poverty is reduced at both the disposable and progressive (as has been seen to be the case in Vietnam) and consumable income stage; direct transfers are well targeted often regressive, representing a greater percentage of market enough with sufficient coverage generosity to more than incomes for poorer households. However, collecting personal offset the effect of indirect taxation. income tax requires significant tax administrative capacity. As a consequence, most developing countries rely largely A reliance on indirect taxation initially presents a trade- on increasing indirect taxation as they increase tax revenues off with respect to revenue collection and equity (Fuchs, as a percentage of GDP and move to a higher income level. Sosa, and Wai-Poi, 2021). Some countries apply extensive Fuchs, Sosa, and Wai-Poi (2021) show that as countries move VAT exemptions or preferential rates to staples such as food from low income to lower-middle to upper-middle-income and clothing. These exemptions can be quite meaningful to levels, they largely increase their tax revenues not through poorer households for whom these items make up a greater progressive personal income tax but instead through greater share of their consumption baskets. At the same time, richer levels of less progressive indirect taxation (Figure 7.11). households also consume these goods and in greater amounts, Even non-OECD high-income countries are more reliant meaning that richer households usually benefit more from on indirect taxation relative to direct taxation. It is only in the often considerable tax expenditures. Countries with such OECD high-income countries that direct taxation outweighs exemptions have more neutral or even progressive indirect indirect taxation in the revenue mix. Consequently, for taxation incidence but lower revenue collection, as is discussed upper-middle-income countries looking to transition to high- in Fuchs et al. (2021) and Wai-Poi et al. (2022). income status, strengthening tax administrative capacity and broadening the personal income tax base is a sound way to The degree of informality in an economy should also be both increase revenue collection and reduce inequality. For considered. Indirect taxes are usually not applied to purchases other developing countries, greater revenues are realistically of goods and services at informal locations, such as street stalls. going to be collected through indirect taxes. 188 2022 Vietnam Poverty and Equity Assessment Point change Point change −25 −20 −15 −10 −5 0 −8 −7 −6 −5 −4 −3 −2 −1 0 1 2 Poland Poland Croatia Romania Panama High income Croatia Uruguay Uruguay High income Chile Chile Romania Panama Nambia South Africa Argentina Argentina South Africa Iran Note: Pensions treated as deferred income. Costa Rica Nambia Brazil Mexico Mexico Mexico Iran Ecuador Mexico Brazil Mexico Mexico Ecuador Russia Belarus Armenia Colombia Albania Upper middle income Colombia Belarus Venezuela Costa Rica Upper middle income Russia Dominican Republic Albania Venezuela Dominican Republic Colombia Paraguay Jordan Armenia Colombia Peru Peru Peru Peru Indonesia Paraguay Indonesia Source: CEQ and World Bank databases and World Bank calculations (see Rodriguez and Wai-Poi, 2020). Guatemala Jordan Guatemala Ukarine Ukarine Tunisia Tunisia Kenya Tanzania El Salvador Kenya Bolivia El Salvador Panel B: Including in-kind spending on health and education Panel A: Excluding in-kind spending on health and education Vietnam Vietnam Tanzania Egypt Nicaragua Nicaragua Figure 7.10. Impact of fiscal policy on inequality by country income level Egypt Ghana Ghana Honduras Lower middle income Lower middle income Sri Lanka Comoros Bolivia Honduras Comoros Sri Lanka Ethiopia Burkina Faso Uganda Uganda Lower income Burkina Faso Lower Ethiopia income Fiscal policy, financing future investments and building the middle class Chapter 7. From the Last Mile to the Next Mile 189 PART 2. THE NEXT MILE IS THE ROAD AHEAD Figure 7.11. Tax mix by country income level, direct and indirect taxes 15 12 Percentage of GDP (%) 9 6 3 0 Dir. Ind. Dir. Ind. Dir. Ind. Dir. Ind. Dir. Ind. HIC, OECD HIC, Non-OECD UMIC LMIC LIC PIT Payroll CIT Property VAT Excises Trade Note: Dir.: direct taxes (PIT, CIT, Payroll Tax, Property Tax). Ind.: indirect taxes (VAT, Excises, Trade). Source: Fuchs, Sosa and Wai-Poi (2021), authors’ calculations based on data from IMF WoRLD database. Non-progressive taxation can be combined with Reducing household disparities through health progressive transfers to achieve a net progressive fiscal and education spending system. As this chapter has made clear, the impact on poverty and inequality of taxes should not be considered in isolation. Vietnam’s health and education spending, while All taxes will place a burden on at least some households. It progressive, reduces inequality less relative to cash is the net effect of these taxes and the spending they finance transfers than in other countries; in part this is because which determines the extent to which a fiscal system reduces of relatively low spending levels on health. Almost all poverty and inequality. The most effective instrument for countries reduce inequality through public investments reducing inequality is targeted direct cash transfers. In almost in education and health. The issues with quality and value all countries, poorer households enjoy a greater share of cash of such spending to households notwithstanding, in most transfer spending than richer households (Fuchs et al., 2021) countries spending on human capital accounts for the which represents an even greater percentage of their market majority of the inequality-reducing effect of fiscal policy income. Thus a combination of increased indirect taxes— (Figure 7.12.). Vietnam is relatively unusual in that less than perhaps through the removal of inefficient exemptions half (around two-fifths) of the fiscal impact on inequality and preferential rates—can both raise revenues and reduce comes from health and education spending. In part this poverty and inequality if it is combined with an expansion of reflects relatively low spending levels in these sectors. For a targeted safety net.61 example, while Vietnam’s education spending (4.0 percent of GDP) is broadly in line with that of upper middle income Moreover, fiscal policy can reduce inequality without (3.9 percent) and middle income (3.7 percent) averages, it necessarily spending more. Inefficient spending on energy is lower in health (2.3 percent compared to 3.3 percent for and food subsidies can be better directed toward targeted upper middle income countries and 2.8 percent for middle direct transfers. A fraction of the subsidy spending can achieve income ones).62 greater reductions in poverty and inequality when spent as targeted cash transfers (Fuchs, Sosa, and Wai-Poi, 2021). 190 2022 Vietnam Poverty and Equity Assessment Percentage point change in Gini −25 −20 −15 −10 −5 0 Sri Lanka 5 Ethiopia Jordan Guatemala Honduras Indonesia Uganda Comoros Ghana Burkina Faso Note: Pensions as deferred income treatment. Peru Peru Egypt Armenia Nicaragua Paraguay Tanzania Vietnam Dominican Republic Albania Russia Bolivia El Salvador Venezuela, RB MI to CI Colombia Colombia Kenya Belarus Source: CEQ and World Bank databases and World Bank calculations (see Rodriguez and Wai-Poi, 2020) Romania CI to FI Ecuador Chile Tunisia Mexico Mexico Iran Mexico Uruguay Ukraine Panama Brazil Figure 7.12. Reduction in inequality from cash and non-cash fiscal instruments Croatia Costa Rica United States Poland South Africa Argentina Nambia Fiscal policy, financing future investments and building the middle class Chapter 7. From the Last Mile to the Next Mile 191 PART 2. THE NEXT MILE IS THE ROAD AHEAD 7.5. References N. Ramasubbaiah, and Stefan Thewissen. 2018. Fair Progress? Economic Mobility across Generations around the World. Bachas, Pierre; Gadenne, Lucie; Jensen, Anders. 2020. Informality, Washington, DC: World Bank. Consumption Taxes and Redistribution. Policy Research Working OECD. 2020. Multi-dimensional Review of Viet Nam: Toward Paper; No. 9267. World Bank, Washington, DC. https:// an Integrated, Transparent and Sustainable Economy, OECD openknowledge.worldbank.org/handle/10986/33851 Development Pathways, OECD Publishing, Paris, https://doi. Dsouza, R., Gatti, R.V., Kraay, A.C. 2019. A Socioeconomic org/10.1787/367b585c-en. Disaggregation of the World Bank Human Capital Index. Policy Phillips, D., Warwick, R.. Goldman, M., Goraus, K., Inchauste, G., Research Working Paper; no. WPS 9020 Washington, D.C.: Harris, T. and Jellema, J. 2018. “Redistribution via VAT and World Bank Group. http://documents.worldbank.org/curated/ Cash Transfers: An Assessment in Four Low and Middle Income en/752571569852714031/A-Socioeconomic-Disaggregation-of- Countries”, Commitment to Equity (CEQ) Working Paper Series the-World-Bank-Human-Capital-Index 78, Tulane University, Department of Economics. Duggan, Stephen. 2001. “Educational Reform in Viet Nam: A Process Rodriguez, L and Wai-Poi, M. 2020. Fiscal Policy, Poverty and Inequality of Change or Continuity?”, Comparative Education Vol. 37, No. 2 in Jordan: The Role of Taxes and Public Spending. World Bank, (May, 2001), pp. 193–212. Washington, DC. https://documents.worldbank.org/en/publication/ Flessa, Steffen, and Nghiem Tran Dung. “Costing of services of documents-reports/documentdetail/113451615905549990/fiscal- Vietnamese hospitals: identifying costs in one central, two provincial policy-poverty-and-inequality-in-jordan-the-role-of-taxes-and- and two district hospitals using a standard methodology.” The public-spending-policy- summary international journal of health planning and management 19, no. Thu Thuong NT, Van Den Berg Y, Huy TQ, Tai DA, Anh BNH. 1 (2004): 63-77. Determinants of catastrophic health expenditure in Vietnam. Fuchs Tarlovsky, Alan; Gonzalez Icaza, Fernanda. 2019. The Welfare and International Journal Health Planning and Management. 2020;1– Distributional Effects of Increasing Taxes on Tobacco in Vietnam. 18. https://doi.org/10.1002/hpm.3076 World Bank, Washington, DC. © World Bank. Xu, Cheng-Gang. 2011. The Fundamental Institutions of China’s Fuchs, A., Sosa, M. and Wai-Poi, M. 2021. “Progressive Domestic Reforms and Development. Journal of Economic Literature. 49. Resource Mobilization for a COVID-19 Recovery”, Briefing Note, 1076-1151. 10.2307/23071664. Washington, DC: World Bank Group. Wai-Poi, Matthew, Michal Myck, Kajetan Trzcinski, Monika Oczkowska, Holla, Alaka; Magdelena Bendini, Lelys Dinarte and Iva Trako. 2021. and Jon Jellema. 2022. Vietnam CEQ Technical working paper. “Is Investment in Preprimary Education Too Low? Lessons World Bank, Government of Vietnam. 2017. Vietnam Public from (Quasi) Experimental Evidence across Countries”, World Expenditure Review: Fiscal Policies toward Sustainability, Efficiency, Bank Policy Research Working Paper 9723, Washington, DC: and Equity. World Bank, Washington, DC. © World Bank. (https:// World Bank. openknowledge.worldbank.org/handle/10986/28610) Horton, M. and El-Ganainy, A. 2020. Fiscal Policy: Taking and Giving World Bank. 2017. Taking Stock – An Update on Vietnam’s Recent Away. Finance & Development, Feb 2020. https://www.imf.org/ Economic Developments. Special Focus: Improving Efficiency and external/pubs/ft/fandd/basics/fiscpol.htm Inchauste, G. and Equity of Public Spending. December 2017. Lustig, N. (2017). Overview: Fiscal Policy and Redistribution, in ------------- 2020. The Human Capital Index 2020 Update: Human Inchauste, G. and Lustig, N. The Distributional Impact of Taxes Capital in the Time of COVID-19. World Bank, Washington, and Transfers: Evidence from Eight Developing Countries. Directions DC. © World Bank. https://openknowledge.worldbank.org/ in Development—Poverty. Washington, DC: World Bank. https:// handle/10986/34432 openknowledge.worldbank.org/handle/10986/27980 ------------- 2021a. Recommendations for the Vietnam Tax Reform Lustig, N. (Ed.) 2018. Commitment to Equity Handbook. Estimating Strategy 2021-2030. World Bank, Washington, DC. © World Bank. the Impact of Fiscal policy on Inequality and Poverty. New Orleans ------------- 2021b. East Asia and Pacific Economic Update April 2021. and Washington, D.C.: CEQ Institute at Tulane University and Brookings Institution Press. ------------- 2021c. A Year Deferred – Early Experiences and Lessons from COVID-19 in Vietnam. Ministry of Finance and World Bank. 2021. Revisiting the Impact of Government Spending and Taxes on Poverty and Inequality in ------------- forthcoming. Poverty and Shared Prosperity Report 2022, Indonesia. World Bank, Washington, DC. Ministry of Health, 2021. https://moh.gov.vn/hoat-dong-cua-lanh- dao-bo/-/asset_publisher/TW6LTp1ZtwaN/content/het-nam- 2020-co-87-96-trieu-nguoi-tham-gia-bhyt-bao-phu-90-85-dan-so Accessed 15th November 2021. Narayan, Ambar, Roy Van der Weide, Alexandru Cojocaru, Christoph Lakner, Silvia Redaelli, Daniel Gerszon Mahler, Rakesh Gupta 192 2022 Vietnam Poverty and Equity Assessment Chapter 7. Fiscal policy, financing future investments and building the middle class 7.6. Notes 53  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 (2018). 54  Pensions can either be treated as deferred income–a person makes contributions when they are working and this is in effect saving, and then they draw down on this income when they retire—or they can be treated as government taxes and transfers—contributions are treated as a tax and payments are treated as a transfer. In the case of Vietnam, the baseline treats pensions as deferred income. See the companion CEQ report for a more detailed discussion of the results when pensions are treated as taxes and transfers. 55  2020 VHLSS data does not provide a pre-COVID baseline. 56  See Fuchs and Gonzalez Icaza (2019) for a more detailed discussion. 57  Excluding acquisition of non-financial assets. 58  Technically, the change in Gini (points) per 1,000 billion dong of expenditure (or revenue in the case of taxes). 59  The welfare quintiles are not based on a monetary aggregate such as consumption or income but are rather a generalized welfare measure based on a number of proxies for household welfare such as household demographics, asset ownership, and housing quality, depending on the underlying data source, condensed into a single welfare index (technically, the first component of a Principal Component Analysis). 60  In only nine countries does poverty increase when moving from market income to disposable income. 61  Phillip et al. (2018) show that even removing VAT exemptions and spending 75 percent of proceeds on an untargeted universal cash transfer both raises revenues and reduces inequality. 62  Based on latest data from World Development Indicators; middle income averages are for both lower and upper middle income countries. From the Last Mile to the Next Mile 193 PART 2. THE NEXT MILE IS THE ROAD AHEAD ANNEX 7.1 Indirect taxes and subsidies The Input-Output (IO) table used in the simulation of environmental taxes has highly aggregated categories, Key assumptions in requiring blunt assumptions about the pass-through of these modelling tax and transfer taxes. All information about the proportion of the output of a given sector that can be attributed to the production of the incidence in Vietnam taxed product has been used. Direct taxes Administrative data on electricity subsidy spending is not available. The current analysis applies the difference between Households that have income from formal non-agricultural the “cost-of-return” for one kWh of electricity and the business activities are currently taxed along the progressive subsidized price for one kWh. personal income tax (PIT) rates. If these types of incomes should be taxed according to the 0.5–5% business income tax In-kind transfers rates, more details would be needed on the criteria according to which these rates are assigned. In the case of both health and education, the aggregate annual public expenditure less household expenditures on Voluntary health insurance contributions for the self- public services is used as the measure for the total value of in- employed are randomly assigned to those with a voluntary kind transfers. Data from past years on costs of each level of health insurance card and calibrated to the number of education or each kind of treatment are used, then uprated voluntary contributors in the administrative data; pensioners while maintaining their proportion to the total government pay the full 4.5 percent contributions (employer plus expenditures in 2018. This allows a distinction between employee rates). expensive public provision of treatments or education and assignment to particular households in the data. Formality for PIT purposes is assigned to employees who report having paid leave, pay SSC, and have an employment Steffen Flessa and Nghiem Tran Dung (2004) estimate contract; self-employed with a registered business are the costs of each kind of health intervention in Vietnam considered formal. Only formal workers pay PIT. The number (inpatient vs. outpatient at various types of facility); these of PIT taxpayers in 2018 is not available. values are uprated to match the total public spending on healthcare in 2018. Gifts are taxed according to formal regulations (gifts from outside the family above 10 million per year). It is unlikely Data on aggregate public expenditures and the number of that all gifted assets are declared; lack of data on the number of students enrolled at each level are combined with data from payers/aggregate revenues prevents more careful calibration. past years on expenditures on each level of education to uprate these values while maintaining their proportionality to total Land value is used as a proxy for land size when assigning the government expenditures in 2018. Data on public spending rate of the non-agricultural land taxes. Information on how per student could improve accuracy. many households pay what rate and the aggregate revenues coming from households is not available. Cash transfers All cash transfers are taken directly from the data. Not all programs are included due to lack of data. 194 2022 Vietnam Poverty and Equity Assessment Chapter 8. Policies for the way forward Chapter 8. Policies for the way forward Policies are discussed following the Last Mile to Next Mile framework of this report. Over the last decade, rapid economic growth was broadly inclusive and livelihoods in Vietnam improved dramatically. The rapid speed of developmental change, however, left some behind who did not have the opportunity to join the most vibrant sectors of the economy and has also created a large class of the population that are not poor but also not yet middle class. Thus, the poverty and equity agenda is no longer only about raising minimum living standards and tackling chronic poverty; it is also about creating new and sustainable economic pathways for a more aspirational population. The emergence of COVID-19 added to challenges looming in regard to skills, productivity, climate change, and an aging society. Whether these constraints to Last Mile poverty reduction and Next Mile high-income development challenges end up being short-term growing pains or long-term bottlenecks to the welfare trajectory of Vietnam’s households will depend in part on policy and prioritization. From the Last Mile to the Next Mile 195 8.1. Addressing a new those with limited access to productive agricultural resources context and Last Mile to migrate to other sectors. Commercialization and better value-chain integration would help maintain agriculture that Chronic Challenges is more competitive and productive. This process needs to be managed with better policies to provide the right skills, knowledge, and technological options for those remaining to be competitive, while also supporting those transitioning to Many of the principal constraints to Last Mile chronic other sectors. poverty reduction challenges remain similar to those that existed at the beginning of decade. These constraints Ethnic minority inclusion policies for further and evidence-based policy recommendations are summarized labor market integration in Table 8.1. Several key policies that are important for continuing tackling Last Mile poverty reduction challenges Poverty among ethnic minorities is still high, but are further summarized. Readers can also refer to a selection progress can be seen in larger numbers of workers of reports that discusses these issues and recommendations in transitioning out of agriculture and a steady pace of detail (World Bank, 2018; 2019a; 2019b and Pimhidzai and poverty reduction over the last decade. To continue Niu, 2020). improving their participation in the labor market, laws protecting the rights of ethnic minorities can be strengthened. Moreover, COVID-19 highlighted new issues that The Labor Code and related laws could be strengthened to require policy attention. A particular challenge that was prevent exploitation and discrimination of ethnic minorities revealed was the need to provide safety nets and options for (World Bank, 2021c). Equal opportunity legislation needs groups of people who may not be poor or registered in the to be consolidated with affirmative actions with respect to social protection system but who face economic insecurity. A non-discrimination for ethnic minorities in the workplace, second emerging area of concern is the accumulated effect of especially in the foreign-invested sector. One rationale inequality, learning losses, and income and employment losses behind the growing opportunities for ethnic minorities in on poor and low-income households (World Bank, 2020f). the labor market is the low labor cost in some labor-intensive sectors. However, ethnic minority youths in these sectors Modernize the agriculture sector and move up have encountered difficulties and to some extent unequal the value chain treatment. Strengthening the enforcement of the Labor Code and related legislation in this context is needed but equally For households remaining in agriculture, farm incomes challenging because many stakeholders, especially at the local and productivity need to increase for this activity level, might be resistant to change in order to maintain their to be an adequate income source. Improving access to comparative advantages in attracting private investment. productive resources, such as land, improved technologies, and extension advisory services, as well as better linkage to Furthermore, understanding and recognizing the favorable input and output markets, are critical to sustain and heterogeneity in ethnic minority development is needed improve agricultural productivity. Access to new knowledge to ensure that resources can be better targeted to improve and innovations, including the application of appropriate development outcomes. Among different ethnic minority digital technologies, would support productivity growth by groups, there are significant variations in socioeconomic substituting for labor intensity, driving down production development. These range from differences in educational costs, improving the quality of produce, and broadening attainment, main sector of employment, to living conditions profit margins for farmers. The type of agriculture may and poverty rate, among others. In identifying the needs and be different, more specialized, and better integrated across areas of support for ethnic minorities, governments should the value chain. This will require new skills and knowledge seek to understand the factors that explain these differences in beyond predominantly primary-level agriculture. The socioeconomic outcomes. provision of new skills will also increasingly be critical to allow 196 2022 Vietnam Poverty and Equity Assessment Chapter 8. Policies for the way forward Table 8.1. A new context and Last Mile policy summary ADDITIONAL RECOMMENDATIONS REFERENCES Small shareholder For households remaining reliant on agricultural earnings, addressing sub-optimal farmland World Bank, 2019a farmers exhibit low use patterns is key to unlocking the agricultural potential of the poor by aligning land use productivity and with comparative advantages in specific areas. This requires a bolder shift of production earnings from rice and maize to more profitable annual and perennial crops. Key policies to achieving this are: • Enabling the poor to invest in more profitable crops that require costly initial investments, intermediate inputs, or hiring of labor by improving access to credit or land rights. • Strengthening land rights by issuing land titles, which could help increase household access to credit by using land as collateral. • Providing targeted services to the poorest households to improve farm management and business skills to boost agricultural productivity and profits. The poor primarily live Earnings and productivity in household enterprises and family farming can also be World Bank, 2019a in more remote areas improved by strengthening their linkages to the broader economy, with three key measures: • Integrating lagging areas into the broader national economic network to expand their market potential; • Facilitating business linkages between household enterprises and small and medium enterprises; and • Encouraging the agriculture sector to diversify into high-value-added crops and local value chains. Lack of labor market Labor market participation plays an increasingly important role for household income World Bank, 2019b inclusion for ethnic generation and poverty reduction. This aspect should be more central to poverty reduction minorities policies for ethnic minorities. Policies should address constraints to obtaining off-farm opportunities such as low educational attainment, language barriers, or gender-biased household divisions of labor. Strengthening Implementation of National Targeting Programs (NTPs) can be strengthened in four areas, Pimhidzai and Niu area-based targeting namely 2020 financing • Ensuring resource additionality to the lagging communes; • Ensuring allocation of resources based on deprivation at the commune level, so that more deprived communes receive more investments; • Earmarking resources across sub-sectors to ensure that adequate resources are devoted to improving quality of human development services and livelihood interventions; and • Strengthening commune-level monitoring by deploying online data collection and aggregation tools to construct a centralized NTP database. A new context after Two important lessons from COVID-19 are in regard to social protection modernization, World Bank, 2021f COVID-19 and learning losses. • Both the household and firm COVID-19 relief packages faced some implementation challenges on the ground. In the short term, technology can be leveraged to facilitate self-registration and online registration to identify informal workers needing assistance. Digitalization of payment mechanisms would help make payments more efficient, transparent, and faster. • Children in more remote regions had less access to online learning during school closures. More investment in distance learning and education continuity can minimize learning losses. 197 2022 Vietnam Poverty and Equity Assessment From the Last Mile to the Next Mile 197 Use technology to reduce economic distance to the commune level. Commune-level information on remote areas achievement of criteria and expenditures is not always readily available at the central level, which hampers both monitoring Leveraging digital technologies, integrating network and targeting of investments. This should be improved economies, and reducing the cost of migration can help by leveraging digital data collection to create a centralized reduce economic distance to rural areas (World Bank, database of investments, outputs, and outcomes to promote 2019a). The growth of non-agriculture sectors is weaker in evidence-based decision making. rural and low-density areas. A critical factor holding back the poor from being engaged in more off-farm activities Be observant of existing and widening monetary is connectivity. These rural economies can become better and non-monetary gaps digitally linked and integrated with the wider economy to expand economic opportunities. COVID-19 highlighted existing inequalities and differences in coping and adaptation. Education outcomes Implementation of National Targeting in Vietnam vary by socioeconomic status, and progress in Programs (NTPs) can be improved to maximize reducing stunting had also stagnated. Women, those in the effectiveness informal sector, and households in the bottom 20 percent experienced the slowest household income recovery between NTPs can provide incremental resources to lagging June 2020 and March 2021. In terms of coping during areas, especially to more deprived communes. The COVID-19, poor households were more reliant on external poorest communes face trade-off across many needs and sources such as borrowing, while rich households were better failed to close the gap in access to services relative to better off able to cope on their own, by tapping into savings. communes. New NTPs should (i) ensure additionality instead of substitution of resources to disadvantaged communes and COVID-19 created unequal impacts which can lead to (ii) allocate resources to reduce deprivation across communes. longer-term implications. Women bear a larger share of Resources can be further prioritized toward investments to caregiving responsibilities, and their labor market activities improve the quality of human development service delivery. were more adversely impacted than men’s. Informal workers Investments under previous NTPs were heavily focused on have the fewest safety nets and experienced the most challenges infrastructure and less on livelihoods, health, and education. when seeking government cash support. Continuity of Thus, while connectivity infrastructure improved, commune education was uneven during COVID-19, and the pandemic leaders still identify poor quality of school facilities and has potentially widened gaps in human capital formation poorly equipped health centers as key challenges. Ethnic because of the uneven capacity of schools across the country. minority policies should also be more targeted since there is Lost education is unlikely to be recovered, with consequences high variation in the outcomes of ethnic minorities across the for lifetime wages; sold assets cannot produce future incomes, 53 minority groups. Future NTPs must earmark resources and employment scarring is also associated with lower lifetime for (i) livelihood interventions, especially promoting value earnings. Larger businesses and wealthier households were chain participation and (ii) human development investments, able to make investments to reap larger sales from digital especially those promoting progression to tertiary education orders, which may lead to widening inequality down the and facilitating the transition from school to work. It could road. Minimizing future disparities will require forward- also provide unconditional commune block grants that would looking policies and improving existing support systems. give communes greater control over investment. International experience suggests that reducing inequality can help accelerate economic growth (Ferreira et al., 2013; World Monitoring and evaluation tools can be strengthened, Bank, 2005; 2016b). including output and expenditure reporting at 198 2022 Vietnam Poverty and Equity Assessment Chapter 8. Policies for the way forward BOX 8.1. Gender Dimensions in poverty reduction and inclusion efforts Due to intersectionality of gender inequality and ethnicity, ethnic minority women are doubly disadvantaged in various domains. Ethnic minority women face social and economic inequalities compared to ethnic minority men and compared to women nationwide. These are most visible in (i) employment; (ii) in education; (iii) health; and (iv) infrastructure (World Bank, 2022). Focus group discussions revealed that ethnic minority women from Dien Bien province face more disadvantages in terms of language, culture, customs and traditions than the unskilled general labor force (Buchhave, Cunningham, Nguyen, and Weimann-Sandig, 2020). Furthermore, ethnic minority women remain underrepresented in positions of decision making at the central and local levels. By 2017, only 17.5 percent, 12.3 percent and 10 percent of ethnic minority women were leaders and managers at the central, provincial, and district levels respectively. Increased attention to promoting joint land titling. To increase poverty reduction and shared prosperity it is relevant to speed up the share of land titles that are signed jointly by men and women. Research by the ABP2-Gender Theme finds that the provision for joint land titling in the 2003 Land Law has positive impact on the empowerment of women, household income, expenditure, and access to credit. Households that have joint-titled Land-Use Rights Certificates (LURCs, naming both members of a married couple as owners of land use rights, increase their expenditure by between 1.6 percentage points (for agricultural land) and 2.5 percentage points (for residential land). For residential land, having a joint-titled LURCs increases formal credit levels by 30 percent and informal credit levels by 17.33 percent. A joint-titled LURC for residential land also increases the share of non-farm business income by 1.8 percentage points (Buchhave, Nguyen, Nguyen, 2020). Women named in LURCs for agricultural and residential land are more likely to have held non-farm jobs in the previous year by 1.76 and 3.12 percent-age points, respectively. For a woman in an ethnic minority, being named in an agricultural LURC increases her likelihood of having wage employment by 3.74 percentage points and nonfarm employment by 3.62 percentage points. The country and its people are however not reaping the benefits of LURCs to the extent possible because the conversion to single to joint land titles are progressing slowly since they are left to the initiative mainly of individual household members, who are either not aware of the benefit and/or process. This evidence has informed policy dialogue within the Government and latest in June 2021, the Minister of MoNRE dispatched an official letter (No. 3362/BTNMT-TCQLDD) to provincial people’s committees requesting for enhancing the application for joint land titling. BOX 8.2. Data for better lives in Vietnam Vietnam scored 66 on the Statistical Performance Index (SPI) which places it in the middle quintile in terms of statistical performance. The SPI framework, developed by the World Bank Group, assesses the maturity and performance of national statistical systems in five key areas: data use, data services, data products, data sources, and data infrastructure (Dang et al, 2021). Between 2018 and 2019, Vietnam’s scores on the SPI have declined marginally. From the Last Mile to the Next Mile 199 BOX 8.2. Data for better lives in Vietnam (Continued) Figure B.8.2.1. Statistical Performance Index scores, 2019 100 80 60 Score 40 20 0 Pillar 1 - Data Pillar 2 - Data Pillar 3 - Data Pillar 4 - Data Pillar 5 - Data SPI Overall score Use Services Products Sources Infrastructure Lower middle income East Asia and Pacific Vietnam Source: World Bank (2021a, 2021b). While Vietnam performs fairly well compared to its peers on the SPI, the country still suffers from several shortcomings in its statistical practices: 1. The Vietnam Household Living Standard Survey (VHLSS)—the main socioeconomic survey—has only 9,000 observations for the expenditure module. Thus, the sample is not representative at the provincial level, which is the main policy and spending administrative unit in Vietnam. In Indonesia, for example, the sample is representative at the district level (of which there are 500), thus enabling proper monitoring and analysis of household expenditure. 2. Currently, there are few health surveys available. Understanding health-related issues such risk factors, health behaviors, and non-health determinants such as socioeconomic status is essential to improving the health and well-being of citizens and in turn the development of the country. 3. Access to data remains limited. In 2020, Vietnam was ranked 91st out of 187 countries on the Open Data Inventory. This ranking reflected a mixed performance in terms of coverage and openness. The current Law on Statistics (2015) has reduced openness on micro data sharing, which is a step backward from the previous legislation approved in 2003. This has resulted in the discontinuation of the micro data portal hosted by the General Statistics Office (GSO). Open data needs to be widened to improve the value of data. When data are not widely available to the general public, its benefits are not fully realized in ways that promote social and economic development. Overall, data can be used more effectively to improve development outcomes in Vietnam. This involves improving on aspects described above, but also requires innovation to stay relevant and provide the needed data for evidence-based policymaking as Vietnam makes strides toward its high-income aspirations. 200 2022 Vietnam Poverty and Equity Assessment Chapter 8. Policies for the way forward 8.2. Reaching Next Mile One of the urgent tasks for the future of education Aspirations in Vietnam is to develop a bold vision and long-term strategies to improve equitable access to education by 2030. Education policies should address socioeconomic inequality in education and skill development opportunities, The Next Mile is the journey to upper and high-income which have longer-term implications for intergenerational country living standards. For society, this means creating mobility and socioeconomic development. Higher private more economic opportunities to build a strong middle class expenditures in education are related to greater education while at the same time expanding support to low-income completion, but its role in pupils’ academic achievement in and economically vulnerable households. The millions lower secondary education also signals inadequate teaching and who have escaped poverty over the last decade now need to puts poor children at a disadvantage. Reforms to the curricula continue climbing up the economic ladder, be provided safety and learning hours are necessary to close this gap. At the general nets to prevent them from falling back into poverty, and be education level, equity in school completion can be improved equipped with the human capital and skills to engage in more by ensuring that students from different backgrounds and with productive and sophisticated work. Policy actions for the different levels of capacity before entering the formal schooling Next Mile include investing in skills for the future, investing system have adequate school readiness, providing financial and in higher-quality education, modernizing social protection non-financial incentives, minimizing social barriers that impede systems for idiosyncratic shocks, and leveraging fiscal policy education progress, and developing an approach on school- to fund inclusive investments. Table 8.2 summarizes the main parent contracts. Additionally, evidence suggests that dropouts recommendations to achieve Next Mile aspirations with regard at secondary schools are higher for children in communes to promoting education, social protection, and fiscal reforms further away from the district center, pointing to challenges while narrowing inequality and promoting the middle-class in access to tertiary education for pupils in remote areas. Their and lifting low-income households to higher levels of income. remoteness increases the cost for post-secondary education, which could be mitigated by raising education aid. Equitable human capital formation Improving higher education Break intergenerational poverty traps with and labor market matching equitable education Continuing improvements in higher education access and quality Investing in human capital can enable Vietnam to exploit emerging opportunities while maintaining Vietnam’s access to higher education, as measured by its competitiveness in a dynamic global marketplace. the gross enrolment rate (GER), is below 30 percent, one Multiple demographic trends highlight the importance of the lowest among the East Asian countries. Reasons of broad-based investment in human capital, including for low level of access include: (a) absence of a clear financing declining fertility rates, rising life expectancies, and a shrinking plan to achieve the originally set quantitative targets, (b) a working-age population. As the dependency ratio increases, fragmented tertiary education system of universities, colleges, a highly productive workforce will be critical to support and VET sectors managed by multiple ministries, (c) an the elderly (World Bank and MPI, 2016). COVID-19 has inconsistent regulatory framework that did not encourage brought additional challenges to human capital development, private sector expansion even though a high target had been affecting nutritional attainment, health and educational set, (d) insufficient student financial aid coverage for low- continuity, and outcomes (World Bank, 2021f). Investment income students, (e) under-development of alternative modes in human capital is now more important than ever, to prevent of education including e-learning and MOOCs education, a reversal of progress already made, as well as preparing future and (f) low quantity and quality pipeline of secondary school generations for a rapidly evolving economic landscape. graduates due to lack of access and learning for children from disadvantaged backgrounds (World Bank, 2020). From the Last Mile to the Next Mile 201 Table 8.2. Next Mile policy summary ADDITIONAL RECOMMENDATION REFERENCES Equitable early human capital formation Better education for all Expanding access to high-quality education across groups is vital to reduce inequalities and World Bank, 2018 increase access to better paying jobs for all. This can be achieved by: • Reforming the structure of the school day to increase instruction hours: Tutoring and differences between quality of schools in poor and non-poor communities explain the variation in academic achievement at lower secondary level, which determines progression to tertiary education. This signals that inadequate teaching places poor children at a disadvantage. This could be addressed by increasing teaching hours in school. • Revision the curricula and pedagogical approach - Teaching and testing should place more emphasis on developing problem solving and critical thinking, the skills that employers deem lacking in Vietnam Sustaining Upwards Economic Mobility through improving higher education, skills, and labor market outcomes Improve higher education Higher education will play a role in the development of a high-skill and high-productivity World Bank, 2020, and skill formation for a growth pathway. Improvements can still be made in access and quality of higher education. 2021e new workforce • Improve access – gaps in education are pronounced across income groups as well as ethnicity. Reducing disparities include addressing insufficient academic readiness, increasing financial aid to disadvantaged students, outreach programs to disadvantaged communities. • Improve skills and quality in particular in science, technology, and innovation. • Improving relevance by creating better networks between education institutions and employers. Improve job transitions Creating upward economic mobility pathways for existing and aging low and medium- World Bank, 2021c, and labor mobility skilled workers is also important. 2021g • Facilitate knowledge dissemination of job opportunities and job matching. • Provide older workers with reskilling opportunities. • These labor market interventions should have a focus on assisting vulnerable groups. Promote firm Improving the capabilities of firms to realize high-income country status ambitions. World Bank, 2021e development and creation of better jobs • Governments can promote SME capabilities by more widely sharing know-how and information on adopting new technologies and best practices. • Investments can be made to support technology generation, research, commercialization, and transfer of new innovations. Modernization of the social protection system is needed to provide responsive safety nets to the economically insecure Adapt to a hybrid model Addressing coverage gaps Nguyen and to address coverage gaps O’Keefe, 2019 especially for informal • A key missing group is the non-poor informal sector. To better target informal workers, workers among others, the system can move away from a static pool of beneficiaries and become more agile in targeting by setting up a dynamic system based on eligibility determination process using an integrated database system. • Increasing coverage costs money, but better identification can also help by identifying who can contribute, and being more precise about who needs social assistance cash support. Modernization is needed Various aspects of the social protection system can be further modernized: Nguyen and across different aspects: O’Keefe, 2019 integration, digitalization, • An integrated and centralized data system including deduplication and cleaning of and e-payments databases and clarification of single source of truth. • Movement to e-payments and other government-to-person (G2P) payments. 202 2022 Vietnam Poverty and Equity Assessment Chapter 8. Policies for the way forward ADDITIONAL RECOMMENDATION REFERENCES • Introduction of a data exchange framework, data exchange platform, and APIs with Nguyen and privacy by design incorporated and personal data protection legislation in place. O’Keefe, 2019 • Digitalize and reform business processes including on-line registration of new beneficiaries and digital payments (eventually into transactional accounts of beneficiary’s choice). Financing for a growing middle-class Over the longer-term, Sophisticated administration and a larger tax base is needed to begin a transition to more Bachas, Sosa and shift towards more progressive revenue generation, requiring Wai-Poi progressive revenue (forthcoming) generation • Investments in tax administrative capacity • Broadening greater formalization of the economy Adequate spending on As countries get richer and raise sufficient tax revenues, they can increasingly fund World Bank, 2019c equitable education and progressive transfers and investments in health and education and 2020 social protection Bachas, Sosa and Wai-Poi (forthcoming) In the short-run, a Building a tax base and formalization takes time, in the short-run revenues can be raised by World Bank, 2021d combined VAT and social some measures including assistance reform can both Fuchs, Gonzalez, raise revenues and reduce • Elimination of regressive tax expenditures of VAT exemptions and preferential rates and Paz, 2019 poverty and inequality • Increased use of property taxes; property is immobile and such taxes are progressive Fuchs and Gonzalez, • Increased use of health taxes which both address negative health externalities and reduce 2019 future health spending as well as raise revenues in a progressive manner Bachas, Sosa and • The burden of higher VAT can be offset by spending a portion of the new revenues Wai-Poi on expanding the social safety net, meaning the combined effect is to raise revenue and (forthcoming) reduce poverty and inequality Improving the relevance and quality of tertiary Increase collaboration between universities education curriculum and staffing can help reduce skill and industry gaps and improve perceptions by businesses who report difficulties recruiting for certain skills. In particular, Stronger engagement with employers on skills businesses find it difficult to recruit for leadership and development is instrumental in aligning education with managerial, socio-emotional, or job-specific technical skills. labor demand. Surveys of firms and TVET institutions To enhance curriculum quality, policies can be pursued alike show limited engagement of the private sector in either to improve funding of university research, improve staff shaping, financing, or providing vocational training. While quality and qualifications, design programs to recruit and policies have been put in place to encourage greater employer retain high quality staff, and to convert higher education engagement in skills development, mechanisms for firms teaching status from administrative staff to public servants. to access incentives are complex and, therefore, firms do The quality of curriculum can also be improved through not take advantage of them. There is a need to simplify and centrally coordinated investments to increase international further encourage firms to provide more formal technical and accreditation of programs, increase international exchange of vocational training. On-the-job training linked to employment students and staff, and the internationalization of curriculum. placement is essential and could be useful to help youths find their first jobs and mature workers to transition into new jobs. Greater links between universities, vocational schools, and employers are needed to ensure that graduates gain relevant skills for the labor market. From the Last Mile to the Next Mile 203 Recruitment techniques can be improved to reach the Migrant workers need assistance to help them access best qualified candidates. Job matching is not competitive better basic services, which will encourage mobility and since the primary pathway to obtaining employment is increase access to non-farm employment opportunities. through personal networks (Nguyen and Kenichi, 2018). Migrants face housing challenges (43 percent), followed by Even when searching for experts, as many as 66 percent of lack of income (38 percent) and inability to find a job (34 enterprises rely on personal contacts. Successful matching percent), according to the 2015 Internal Migration Survey through employment service centers (ESC) results in a small (GSO and UNFPA, 2016). The majority of migrant workers share of new jobs. Focus group studies have also shown live in low-quality, crowded rental dwellings, which, during that employers feel ESC fees are too expensive, such as for the COVID-19 pandemic, has exacerbated their risks of advertising. From the labor supply side, ESCs are also not well contracting the virus. The implementation of social assistance utilized: only 0.2 percent of young workers used these centers packages for those affected by the COVID-19 pandemic has in 2015. More recent studies also find that workers find ESCs also highlighted challenges in identifying and registering to be ineffective (Buchhave et al., 2020). low-income migrant workers, who are concentrated in the informal sector and are more vulnerable to adverse economic Effective options should exist for upskilling and shocks. Interventions to promote labor mobility could aim reskilling the current labor force at lowering the social challenges to migrant workers, namely, unequal access to public services in the host communities, Upskilling and reskilling the labor force are instrumental including housing, healthcare, childcare and education, as for Vietnam to fully harness the potential of its human well as elder care in rural areas (Giles and Huang, 2020).63 capital and prepare its workers for higher-quality jobs. Furthermore, infrastructure interventions aimed at lowering Despite significant progress in educational enrollment and transportation costs, improving ICT connectivity, and attainment, there still exist skills shortages, mismatches easing housing constraints would also encourage migration, between education and occupation, and unequal access to especially from remote low-density regions. education and training opportunities across gender, location, ethnic groups, and socioeconomic backgrounds. Vietnam’s Another important area of focus is reducing the burden most common occupations are semi-skilled. Technical and of child and elderly care and enhancing flexible working vocational education and training (TVET) institutions have arrangements to help women get into more and better a role to play in both upskilling and reskilling the labor force, jobs. These objectives can be achieved by (i) expanding but they need essential reforms to become modern, results- childcare facilities, (ii) promoting participation in the digital oriented, and responsive to labor market needs. Around marketplace, and (iii) ensuring that women are named on the world, TVET plays a key role not only as part of post- both agricultural and residential land use rights certificates. secondary training options for youth but also for reskilling of the workforce, as reskilling is needed more and more with accelerating technological change. Modernizing social …all with a focus on vulnerable groups protection to guard against shocks For disadvantaged and vulnerable groups, lowering financial barriers is critical, particularly in the context A hybrid social protection model can help close of increasing tuition fees, as more TVET and higher the coverage gap education institutions are being granted financial autonomy. Introducing universal, well-designed, income- COVID-19 highlighted the need to make more progress contingent loans, supplemented by an expansion of scholarship on addressing the coverage gap. The economic impact programs to the most economically disadvantaged, could of COVID-19 was disproportionately felt by the non-poor significantly improve equitable access to tertiary education informal sector in many countries, including Vietnam. and reduce dropouts. This led to many ad hoc attempts to expand social assistance and cash transfer programs to this population. 204 2022 Vietnam Poverty and Equity Assessment Chapter 8. Policies for the way forward Some countries were able to do so quickly by leveraging a A key to ensuring that more households are covered variety of administrative databases, but most faced serious against a wide range of risks is extending coverage to challenges in their horizontal expansions and were unable the ‘missing middle’ Figure 8.1 visualizes the gap in social to reach many affected households (Johnson and Palacios, protection coverage which is often referred to as the missing forthcoming). This highlighted a major gap in many social middle. Closing the gap implies blurring the line between social protection systems that focus on a static pool of social assistance and social insurance and ensuring that the non-poor assistance beneficiaries on the one hand and the relatively informal sector is incorporated into administrative databases small formal sector covered by social insurance on the other. and that those affected by shocks can be quickly identified. While the magnitude of this gap became more evident in In the last 20 years, many countries including Vietnam have 2020–2021, the vulnerability of this large group to various effectively taken this route toward universal coverage in health types of shocks was a reality before the pandemic and will be by subsidizing contributions to social health insurance schemes. again afterwards. A hybrid social assistance-social insurance means that non- poor informal workers make some contributions to covering their risks while the government makes up the difference; this contrasts with true social assistance (where the poor receive Figure 8.1. Addressing the social protection coverage gap A. Stylized state of social protection in developing country B. Policy to fill the current gaps in protection Market insurance Encourage more is scant, expensive, supply and demand of and/or ineffective voluntary insurance Social insurance Ensure social Level of protection Level of protection Risk-pooling instruments combines actuarial insurance is for preventing poverty are and equity actuarially fair objectives with tightly rationed, excluding distortive and many poor and vulnerable rationing effects people Eliminate exclusion Expand beyond narrow Shift equity elements to poverty-targeting financing from broader-based (general revenues financed) Income or consumption Income or consumption C. Comprehensive insurance assistance Raise the realtve importance of voluntary to mandtory insturments through nudging Level of protection Taper-away assistance and subsidies from the top of the distribution, through tax systems Income or consumption Source: Packard et al. (2019). From the Last Mile to the Next Mile 205 non-contributory support) and social insurance (where those manner may be useful in situations where there is little change who can afford it make premium payments themselves). The in household status, but in rapidly changing environments, result is that Vietnam had achieved health insurance coverage of however, this approach produces information that can quickly close to 90 percent of the population, with most of this increase become outdated. This was demonstrated in a number of happening in the last decade. countries early on during the COVID-19 pandemic. Many Vietnamese who are neither poor nor covered A better approach is to set up a dynamic system based on by formal social insurance can afford to make some an eligibility determination process that allows for rapid contributions to enter the system, even if they are and digital application and registration processes and partly subsidized by the government. To make such a interoperable databases linked via unique individual and hybrid policy more affordable, it is useful to differentiate household identifiers. Rather than periodically collecting subsidies according to the ability of informal sector workers data from households that quickly becomes obsolete, linking to contribute. Figure 8.1 assumes that the subsidy is gradually various administrative databases where data are regularly phased out at the upper end of the income distribution. This updated and linked by common identifiers can facilitate rapid is similar to Vietnam’s policy of differentiating the health and accurate responses to different shocks. Combined with insurance subsidy for the poor and near poor. However, online applications, this federated data model has proven to poorer households will need support without making be more agile both in crises and in normal times. contributions. An often-cited example is Turkey’s Integrated Social If well-targeted, reaching more beneficiaries doesn’t have Assistance System (ISAS).64 It draws information in real time to be expensive. As noted in Chapter 1, a transfer of 24.6 from two dozen databases and registries to help determine trillion VND would be sufficient to eradicate consumption- eligibility for dozens of social programs. Applicants enter their based poverty in 2016 if perfectly targeted to households national ID number through a website triggering the process. who were poor based on the LMIC ($3.20/day 2011PPP) Local government councils play a role in verifying certain poverty line. As a comparison, the NTP-NRD and SPR-135 information, but most of the criteria are based on objective budget over the period 2016–2020 was 804 and 20.6 trillion, information taken from databases such as income tax, property, respectively. With a more robust data system, beneficiaries assets, and others (see MoSFP and World Bank, 2018). This has can be better identified and targeted. One key to improving reduced the processing time dramatically. Before the pandemic, the coverage of the existing non-contributory social assistance this system allowed the Turkish government to differentiate programs is to make eligibility assessment more dynamic, health insurance contribution subsidy levels and determine so that those truly in need—a status that changes for better eligibility for social assistance benefits. During the COVID-19 and for worse over time—are receiving assistance. Advances pandemic in 2020, it allowed for rapid, electronic registration in availability of data and technology are helping to drive of millions of households in just a few weeks. not only the dynamic nature of social registries but also their accuracy in assessing household welfare (Grosh et al., 2022). Vietnam now has most of the building blocks needed to move to this kind of federated system. The mass A centralized, digital, and integrated data system registration campaign for the national ID and the subsequent is necessary to be more agile and responsive mandate to link the unique National ID number with major databases early in 2022 has set the stage for effective A modern social protection system requires the ability digitalization of social protection. It will require a series of to target effectively in a more agile way. Especially in actions over several years with the collaboration of multiple countries with high informality, workers and households are government stakeholders. Next steps include the following: constantly experiencing changes in their employment status and being affected by different idiosyncratic and occasionally • Introduction of common data formats and data dictionary covariate shocks. Most countries rely on periodic assessments across relevant databases of the needs and conditions of households to determine • Deduplication and cleaning of databases and clarification of their eligibility for different programs. Data collected in this single source of truth 206 2022 Vietnam Poverty and Equity Assessment Chapter 8. Policies for the way forward • Introduction of a data exchange framework and data exchange Progressive taxation requires investments in tax platform and APIs with privacy by design incorporated and administrative capacity and greater formalization of the personal data protection legislation in place economy. Key priorities for tax administrative capacity reform • Digitalization and reform of business processes including include institutional reform, in particular the merger of the on-line registration of new beneficiaries and digital payments General Department of Taxation’s (GDT) strategic and risk (eventually into transactional accounts of beneficiary’s management functions; review of tax administration business choice) processes and design/implementation of a comprehensive IT system; addressing potential taxation of the growing digital economy; enhancing GDT staff capacity; and creating a Adapting fiscal policy to unified collection system for personal income tax and social help reduce poverty and insurance contributions World Bank (2021d). In addition to greater capacity to collect income and property taxes from inequality is a long-term existing formal workers and firms, this formal tax base needs to process be expanded by bringing more informal workers and informal firms into the formal economy. Formal workers not only pay Fiscal policy can play a critical role in both driving income tax; they also contribute to social insurance schemes. Vietnam toward high-income status and doing so in an At the same time, they earn higher wages and have better inclusive manner to assist the movement of people into protection, meaning that both workers and state revenues are a prosperous middle class. It can achieve this in two ways. better off. First, it can help finance the required investments needed for the country and its workers to become more productive and Vietnam could also use fiscal policy more proactively higher earners, such as those discussed in this chapter so far: to help households manage risk, by financing a modern modernization of agriculture, better skills and higher quality social protection system and cushioning shocks in a education, a more robust digital backbone, and accompanying countercyclical fashion. Households face risks on both services. Second, it can also finance policies which can address the macro and the micro level (Chapter 6). A critical policy Last Mile and Next Mile constraints today, such as a modern for helping households manage these risks is a modern social social protection system and strengthening of NTPs. Finally, protection system, including targeted social assistance to lift the revenues required to finance these investments can the remaining poor out of poverty; social assistance and social themselves be more or less progressive. insurance protect the gains already made by much of the rest of the country and help them continue to climb into the Fiscal policy can be used in a progressive manner middle class; and an adaptive social protection system which can scale up support to existing and new beneficiaries when As countries grow richer, they increasingly raise tax larger shocks occur. Vietnam needs to bring social protection revenues to finance public spending and investments spending in line with international norms to develop such a and do so in an increasingly progressive manner; system. However, fiscal policy can also manage household risk Vietnam should follow this path. As Chapter 7 has shown, at a macro level by being used in a countercyclical fashion. developing countries rely on easier to collect but neutral or even regressive indirect taxation. However, as upper middle- Fiscal policy not only needs to raise adequate revenues in income countries look to make the transition to high-income a progressive manner; it also needs to spend in the right status, they need to broaden their tax base mix and increasingly way. All of the recommendations in this report require some use more progressive direct taxes such as personal income and public investment. Greater revenues will create more fiscal property taxes; high-income OECD countries are the only space for such investments. In addition, additional space can ones to collect the majority of their tax revenues from direct be created now by redirecting spending from inefficient and taxation. This supports an inclusive economy both directly inequitable expenditures such as electricity subsidies, which by collecting revenues from those most able to afford it and encourage wasteful energy use and contribute to climate indirectly by financing inclusive investments. change while mostly benefiting richer households. From the Last Mile to the Next Mile 207 Education financing should be targeted to as food and other items which make up a greater share of disadvantaged students the consumption basket of the poor, richer households also consume these goods, and in greater amounts, so the large Investments should be made on proven interventions to majority of these tax breaks goes to them. improve access to quality education for disadvantaged students through better targeted funding. Adequate Research has shown that expanding both the VAT education financing would require (i) higher state budget tax base and social protection schemes in tandem has expenditure for education, (ii) re-allocation of the state budget the potential to both raise tax revenues and reduce on education across education subsectors toward a healthier poverty. While most of the tax expenditures spent on VAT share for tertiary education without compromising general exemptions benefit richer households, they are nonetheless education, and (iii) efficient use of funding. The government important to the poor. The overall benefit to the poor may has been accelerating deployment of additional state budget be small in absolute Vietnamese dong, but they represent a for education, including investment in improving physical greater percentage of poorer households’ meager incomes infrastructure of schools serving students from the lagging than they do of those of richer households, meaning that ethnic minorities and mountainous areas. However, in VAT exemptions are an expensive but progressive instrument the coming years, the government should adopt a broader which help mitigate the tax burden on poorer households. concept of how needy students are identified, in addition to However, if part of new revenues raised from eliminating such the current considerations of ethnicity, socioeconomic status, exemptions are transferred to households as cash benefits, and gender. In particular, there needs to be a focus on low poverty and inequality can be reduced while still reserving performing students regardless of background. Additional some of the new revenue for other purposes. Such a result has funding may support proven education interventions in early already been shown for six lower-middle-income countries years such as the transition between lower secondary to upper (Warwick et al 2022), such as Vietnam. If the transfers were secondary schools, or provision of parenting supports and targeted more narrowly at the bottom half of the distribution, career guidance among other key areas. similar results could be achieved at a lower cost. Proper financing requires a more balanced approach In addition, a range of new taxes can be explored, a between decentralization and centralization as well as a number of which have positive spillover effects. Non- stronger accountability mechanism. Given the increasing traditional taxes which would help broaden Vietnam’s tax decentralization in education financing, it is imperative for base include taxes on digital transactions. In addition, some the funding allocation process at the central level to take into taxes not only increase revenue; they also discourage behaviors account locality-specific variations, identifying disparities by with negative consequences. Health taxes on products such as regions or local areas, including drilling down to individual alcohol, tobacco, and sugar-sweetened beverages can be new ethnic groups and schools as necessary. At the provincial sources of revenue. They can also reduce the harmful effects and district level, funding allocation should consider how to health of over-consumption of these goods, which benefits best to meet the needs of specific ethnic groups from the low individual health and also reduces public health expenditures socioeconomic backgrounds whose learning outcomes are in the future, creating further fiscal space for other purposes. substantially lower than the average for all students. Moreover, taken over a lifecycle, these taxes have been shown to be progressive, placing less burden on poorer households, In the short term, revenues can be raised through including in Vietnam (Fuchs, Gonzalez, and Paz, 2019; Fuchs VAT and Gonzalez, 2019). See World Bank (2021d) for a discussion on additional key tax reforms for Vietnam including personal Building tax capacity and formalizing the economy income tax and property tax reforms. takes time. In the shorter term, eliminating regressive tax expenditures of VAT exemptions and preferential rates would raise significant revenue. Vietnam’s VAT schedule includes many exemptions and lower rates for a range of items (World Bank, 2021d). While these are often on staples such 208 2022 Vietnam Poverty and Equity Assessment Chapter 8. Policies for the way forward 8.3. References Nguyen, Nga Nguyet, and Philip B. O’Keefe. 2019. A Vision for the 2030 Social Protection System in Vietnam (English). Washington, Bachas, Pierre, Mariano Sosa and Matthew Wai-Poi (forthcoming) D.C.: World Bank Group. “Fiscal Policy and its Short-term Distributional Impact: Patterns Packard, Truman G., Ugo Gentilini, Margaret Ellen Grosh, Philip and lessons from international experience” in Jed Friedman and B. O’Keefe, Robert J. Palacios, David A. Robalino, and Indhira Ruth Hill (eds.) (forthcoming) Poverty and Shared Prosperity Vanessa Santos.2019. Protecting All: Risk Sharing for a Diverse and Report 2022, World Bank Diversifying World of Work. Washington, DC: World Bank Group. Buchhave, Helle; Wendy Cunningham, Giang Tam Nguyen, and Pimhidzai, Obert and Chiyu Niu. 2020. Shared Gains: How High Nina Weimann-Sandig. 2020. Perceptions of Gender Disparities in Growth and Anti-poverty Programs Reduced Poverty in Vietnam. Vietnam’s Labor Market. Washington, DC: World Bank. Vietnam Poverty and Shared Prosperity Update. Buchhave, Helle; Giang Tam Nguyen; Cuong Viet Nguyen. 2020. Benefits Warwick, Ross, Tom Harris, David Phillips, Maya Goldman, Jon of Joint Land Titling in Vietnam. Washington, DC: World Bank. Jellema, Gabriela Inchauste and Karolina Goraus-Tanska. 2022. Card, David, Jochen Kluve, and Andrea Weber. 2018. What Works? A The Redistributive Power of Cash Transfers vs VAT Exemptions: A Meta Analysis of Recent Active Labor Market Program Evaluations. Multi-Country Study. World Development. Volume 151. Journal of the European Economic Association, European Economic World Bank. 2018. Climbing the Ladder: Poverty Reduction and Shared Association 16(3): 894–931. Prosperity in Vietnam. Washington, DC: World Bank. Dang, Hai-Anh H., Mustafa Dinc, Juderica Zilla Josephine Dias, ----------- 2019a. Better Opportunities for All: Vietnam Poverty and Shared Hiroko Maeda, John James Pullinger, Umar Serajuddin, Brian Prosperity Update. Hanoi: World Bank. https://openknowledge. William Stacy, and Dereje Ketema Wolde. 2021. Measuring the worldbank.org/handle/10986/33545 License: CC BY 3.0 IGO. Statistical Performance of Countries: An Overview of Updates to ----------- 2019b. Drivers of Socio-economic Development. Among the World Bank Statistical Capacity Index (English). Washington, Ethnic Minority Groups in Vietnam. Washington, DC: World Bank. DC: World Bank Group. ----------- 2020. Improving the Performance of Higher Education in Fuchs, A., M. Sosa, and M. Wai-Poi. 2021. Progressive Domestic Vietnam: Strategic Priorities and Policy Options. Washington, DC: Resource Mobilization for a COVID-19 Recovery. Briefing Note. World Bank. Washington, DC: World Bank Group. ----------- 2021a. Vietnam Country Report (Statistical Performance Index Fuchs, Alan, Fernanda González Icaza, and Daniela Paz. 2019. Team). May 25, 2021. Retrieved on March 11, 2022 from https:// Distributional Effects of Tobacco Taxation: A Comparative Analysis. datanalytics.worldbank.org/SPI/ Policy Research Working Paper No. 8805. Washington, DC: World ----------- 2021b. World Development Report 2021: Data for Better Bank. https://openknowledge.worldbank.org/handle/10986/31534 Lives. Washington, DC: World Bank. Fuchs Tarlovsky, Alan and Fernanda Gonzalez Icaza. 2019. The Welfare ----------- 2021c. Vietnam: Adapting to an Aging Society. Washington, and Distributional Effects of Increasing Taxes on Tobacco in DC: World Bank. Vietnam. Washington, DC: World Bank. https://openknowledge. worldbank.org/handle/10986/32062 ----------- 2021d. Recommendations for the Vietnam Tax Reform Strategy 2021-2030. Washington, DC: World Bank. Giles, John and Huang Yang. 2020. Migration and Human Capital Accumulation in China. IZA World of Labor 476. ----------- 2021e. Vietnam: Science, Technology, and Innovation Report 2020. Washington, DC: World Bank. Grosh, Margaret, Philippe Leite, Matthew Wai-Poi, and Emil Tesliuc. 2022. Revisiting Targeting in Social Assistance: A New Look at Old ----------- 2021f. A Year Deferred - Early Experiences and Lessons from Dilemmas. Renouf Online Bookstore. COVID-19 in Vietnam (English). Washington, DC: World Bank. GSO and UNFPA. 2016. The 2015 Internal Migration Survey: Major ----------- 2021g. Training Assessment Project Vietnam Country Report. Findings. ----------- 2022. Reducing Poverty among Ethnic Minorities in Vietnam: Johnson, Doug and Robert Palacios (forthcoming). Scaling up Cash Priority Action Areas and Lessons Learnt (internal document). Transfers during the COVID-19 Pandemic: What might explain differences in coverage? 8.4. Notes Lustig, N. (Ed.) 2018. Commitment to Equity Handbook. Estimating the Impact of Fiscal policy on Inequality and Poverty. New Orleans and Washington, DC: CEQ Institute at Tulane University and 63  The need to take care of the elderly discourages migration for both Brookings Institution Press. men and women and contributes to reverse migration. Adult children MoSFP and World Bank. 2018. Turkey’s Integrated Social Assistance with elderly parents in rural Vietnam are less likely to migrate. System. Available at: http://documents.worldbank.org/curated/ 64  en/515231530005107572/pdf/Turkey-SA-summary.pdf Other good examples include Chile’s Social Protection Information Nguyen, Duc Thanh and Ohno Kenichi. 2018. Understanding the System and Australia’s Centrelink. Labor Market for Productivity Enhancement. Viet Nam Annual Economic Report 2018. From the Last Mile to the Next Mile 209 With support from: 8 Dao Tan Street, Ba Dinh District, Hanoi, Vietnam Telephone: (84-24) 3774 0100 Fax: (84-24) 3774 0111 Website: www.dfat.gov.au 63 Ly Thai To Street, Hoan Kiem, Hanoi, Vietnam Telephone: (84-24) 3934 6600 Fax: (84-24) 3935 0752 Website: www.worldbank.org.vn @WorldBankVietnam @WB_AsiaPacific