74910 Well Begun, Not Yet Done: Vietnam’s Remarkable Progress on Poverty Reduction and the Emerging Challenges WORLD BANK 2012 Vietnam Poverty Assessment Well Begun, Not Yet Done: Vietnam’s Remarkable Progress on Poverty Reduction and the Emerging Challenges World Bank in Vietnam Hanoi, 2012 Acronyms AC Agricultural Census ADB Asian Development Bank ASEAN Association of Southeast Asian Nations CAF Center for Analysis and Forecasting CBN Cost of Basic Needs CPI Consumer Price Index CPRGS Comprehensive Poverty Reduction and Growth Strategy CPS Country Partnership Strategy CSA Country Social Analysis DFID Department for International Development (UK) DOLISA District-level MOLISA staff DPT1 Diptheria, Pertussis, and Tetanus, ï¬?rst immunization EA Enumeration Area EAP East Asia and Paciï¬?c (WB) ELL Elbers, Lanjouw, and Lanjouw FDI Foreign Direct Investment FGT Foster-Greer-Thorbecke FGT0 Poverty headcount FGT1 Poverty gap FGT2 Squared poverty gap GAPAP Governance and Poverty Policy Analysis and Advice GDI Gender Development Index GDP Gross Domestic Product GSO General Statistics Ofï¬?ce HCMC Ho Chi Minh City HCR Headcount Rate HDI Human Development Index HOI Human Opportunity Index ILSSA Institute of Labour, Science, and Social Affairs IMF International Monetary Fund L Large M Medium MCP Monetary Child Poverty (rate) MDCP Multi-dimensional Child Poverty (rate) MDG Millenium Development Goal MICS Multi-Indicator Cluster Survey MOC Ministry of Construction MOET Ministry of Education and Training MOH Ministry of Health MOLISA Ministry of Labor, Invalids, and Social Affairs MPI Ministry of Planning and Investment MPI Multi-dimensional Poverty Index NGO Non-Governmental Organization NHDR National Human Development Report (UNDP) NSS National Sample Survey NTP-PR National Targeted Program for Poverty Reduction NTP-SPR National Targeted Program for Sustainable Poverty Reduction PA Poverty Assessment PAPI Public Administration Performance Index PM Prime Minister POVCALNET PovcalNet, the WB’s online poverty analysis tool PPA Participatory Poverty Assessment PPP Purchasing Power Parity PREM Poverty Reduction and Economic Management PRSP Poverty Reduction Strategy Paper RAFC Rural Agriculture and Fishery Census RCS Ravallion, Chen, and Sangraula RIM Rural Impact Monitoring S Small SCOLI Spatial Cost of Living Index SEDP Socio-Economic Development Plan SEDS Socio-Economic Development Strategy SOE State-owned enterprise SPB Social Policy Bank TFESSD Trust Fund for Environmentally and Socially Sustainable Development UNDP United Nations Development Program UNFPA United Nations Population Fund UNICEF United Nations Children’s Fund USAID United States Agency for International Development VASS Vietnam Academy of Social Sciences VBA Vietnam Bank for Agriculture VDR Vietnam Development Report VHLSS Vietnam Household Living Standards Survey VLSS Vietnam Living Standards Survey VND Vietnam Dong VPHC Vietnam Population and Housing Census WB World Bank WDI World Development Indicators WHO World Health Organization WTO World Trade Organization XL Extra large XS Extra small Acknowledgements This report was prepared in partnership by the World Bank and the Center for Analysis and Forecasting, Vietnam Academy of Social Sciences (VASS), with substantial inputs and comments provided by national researchers and experts as well as international partners, including the United Kingdom (DFID), the United Nations (UNDP, UNICEF, UNFPA, UN Resident Coordinators Ofï¬?ce), the European Commission, Ireland (IrishAid), and Oxfam GB. Work on new poverty monitoring systems was carried out jointly with the Social and Environmental Statistics Department of the General Statistics Ofï¬?ce (GSO), Government of Vietnam, and the Center for Analysis and Forecasting, VASS. Preparation of the report was led by a core team consisting of Valerie Kozel (Task Team Leader) and Nguyen Thang (Director, CAF), Reena Badiani (World Bank), Bob Baulch (RMIT University), Loren Brandt (University of Toronto), Nguyen Viet Cuong (Consultant, NEU), Vu Hoang Dat (CAF), Nguyen Tam Giang (World Bank), John Gibson (Waikato University), John Giles (World Bank), Ian Hinsdale (World Bank), Pham Hung (Consultant, IRC), Peter Lanjouw (World Bank), Marleen Marra (World Bank), Vu Van Ngoc (CAF), Nguyen Thi Phuong (CAF), Paul Schuler (Consultant), Hoang Xuan Thanh (Consultant, Ageless), Le Dang Trung (University of Copenhagen), Phung Duc Tung (IRC), Linh Hoang Vu (World Bank), and Andrew Wells-Dang (Consultant, Oxfam GB). The team from the General Statistics Ofï¬?ce included Nguyen Phong (ex-Director, Social and Environmental Statistics Department), Do Anh Kiem (Director, Social and Environmental Statistics Department), Lo Thi Duc, and Nguyen The Quan. Additional inputs were provided by Paul Van Ufford and the team at UNICEF/Hanoi (on child poverty) and Ingrid Fitzgerald (UN Resident Coordinators Ofï¬?ce, Vietnam) and Michaela Prokop (UNDP/Hanoi) on the Human Development Index and multi-dimensional poverty indicators. The report beneï¬?ted from extensive review and inputs at the concept phase, and the team appreciates the many suggestions received at the World Bank concept review meeting and three early consultations workshops (in Hanoi and HCMC) organized by VASS in 2011. The report beneï¬?ted as well from comments received at two seminars sponsored by the World Bank ofï¬?ce in Hanoi in March and June, 2012, and a technical workshop organized by VASS in June, 2012 to discuss the background papers and an early draft of the report. The team is grateful for comments received at the World Bank decision review in June, 2012, including from peer reviewers: Dominque van de Walle; Michael Woolcock; and Salman Zaidi (all from the World Bank); and Dr. Nguyen Thi Lan Huong (Director, ILSSA). More generally, the team would like to acknowledge comments received throughout report preparation from members of the Vietnam country team as well as staff in East Asia PREM including Mette Bertelsen, Christian Bodewig, Quang Hong Doan, Kari Hurt, Steve Jaffee, Andrew Mason, Nguyen Thi Thu Lan, Trang Van Nguyen, Son Thanh Vo, and Myla Williams, A second and ï¬?nal round of consultation workshops was organized by VASS and the World Bank in HCMC and Hanoi in August, 2012 on the revised draft of the report. The team is grateful for comments and suggestions provided by participants at both workshops, including written comments provided in advance of the HCMC workshop by Dr. Jonathan Pincus (Fullbright Program, HCMC); Dr. Huynh Thi Ngoc Tuyet (former researcher from Southern Institute of Sustainable Development); Dr. Nguyen Hoang Bao (HCMC University of Economics); and Dr. Le Thanh Sang (Southern Institute of Sustainable Development). Written comments were received in advance of the Hanoi workshop from Dr. Le Dang Doanh (former Economic Advisor); Dr. Nguyen Hai Huu (MOLISA); Mr. Do Anh Kiem (GSO); Bert Martens (Oxfam/HK); and Dr Trinh Cong Khanh (CEMA). We are also grateful for comments and suggestions provided at the consultation workshops by Nguyen Tien Phong (UNDP); Pham Quang Ngoc (ADB); Madame Pham Chi Lan (former Vice President of VCCI); and Dr. Dang Kim Son (IPSARD). The team would like to thank the GSO for providing excellent logistical assistance as well as timely access to the 2010 VHLSS and other sources of data. This report is one of many products emerging from the long and fruitful collaboration between the World Bank, VASS, and the GSO on poverty measurement, monitoring, and policy analysis. Guidance for the overall work was provided by Victoria Kwakwa, World Bank Country Director in Vietnam; Sudhir Shetty, Poverty Reduction and Economic Policy Sector Director, and Deepak Mishra, Lead Economist, Vietnam Country Program. Their advice and ongoing support is gratefully acknowledged. The advice of many others, both from inside the World Bank as well as outside, who provided valuable inputs and suggestions throughout the process of preparing the background papers and ï¬?nal report is acknowledged and appreciated. The World Bank in Vietnam’s communications team provided excellent just in time support for dissemination and launch of the ï¬?nal report, with particular thanks to Nguyen Hong Ngan, Vu Lan Huong, and Tran Kim Chi. Tuyet Thi Phung, Lynn Yeargin, Mildred Gonsalvez (all World Bank), and Vu Van Ngoc (CAF) provided excellent administrative support over the course of the project, including the production of the ï¬?nal report. Tuyet Thi Phung and Vu Van Ngoc were responsible for organizing numerous consultation and dissemination events, often working late into the night. Many thanks for your efforts. The team would like to thank DFID for substantial ï¬?nancial support provided under the GAPAP trust fund, including Huong Tran Thi Thien and Renwick Irvine, DFID staff in Hanoi, for their ongoing support in preparing the report. We are also grateful to TFESSD donors for supporting new work on perceptions of inequality. Contents Executive Summary CHAPTER 1 Vietnam’s Growth and Poverty Reduction Record: Remarkable Success, but Big Remaining Challenges 9 A. Introduction 10 B. Vietnam’s economy has grown rapidly and has undergone profound structural transformation 10 C. Progress in reducing poverty has been remarkable by any standard 13 D. Despite this remarkable progress, the task of poverty reduction is not ï¬?nished 20 E. Overview of the report: Vietnam’s old and new poverty reduction challenges 31 CHAPTER 2 Updating Vietnam’s Poverty Monitoring System 36 A. Introduction 37 B. Rethinking Poverty and Poverty Measurement in Vietnam 37 C. Updating Methods for Measuring Poverty 39 D. Constructing a new GSO-WB Poverty Line 47 E. New Poverty Estimates for 2010: GSO-WB and Ofï¬?cial Poverty Methodologies 52 F. Are the New GSO-WB Poverty Lines too High? Are They Consistent with Citizens’ Subjective Views? 54 CHAPTER 3 Poverty Proï¬?le: Establishing the Facts about Poverty and the Poor in Vietnam 63 A. Introduction 64 B. The Poor in Vietnam still Predominately Live in Rural Areas and are Increasingly Concentrated in Upland Regions 66 C. Many of the Poor are Farmers Whose Livelihoods are Primarily Linked to Agriculture 67 D. Ethnic Identity Matters even more for Poverty Today 68 E. Poverty is Still Linked to Low Education Attainment 73 F. Housing and Local Infrastructure have Improved Substantially since the Late 1990s 79 G. Urban Poverty is Low According to GSO-WB Estimates, and Concentrated in Smaller Cities and Towns 80 H. Poverty has Become Less Correlated with Demographic Factors, although Aging is Emerging as an Issue and Child Poverty Remains a Concern 82 I. Poor Households are Still Vulnerable to Weather Shocks 87 J. Limited Coverage is Provided by Existing Poverty Reduction and Social Protection Programs 87 CHAPTER 4 Spatial Dimensions of Poverty: 1999 and 2009 Poverty Maps 93 A. Introduction 94 B. 2009 Poverty Maps 95 C. Inequality and Wealth Maps 103 D. The Evolution of Spatial Poverty, 1999 to 2009 106 E. In what other Ways can Mapping Methods Inform Policy Design and Evaluation? 111 CHAPTER 5 Reducing Poverty among Ethnic Minorities 121 A. Introduction 122 B. Ethnic Minority Poverty Reduction Varies across Regions, among and within Ethnic Groups 123 C. Disparities in Access to Education, Infrastructure, and Public Services Accompany and Reinforce Ethnic Minorities’ Poverty Reduction Outcomes 127 D. The Experiences of Ethnic Households that have already Escaped Poverty Offer Lessons and an Innovative Orientation for Future Policies and Programs 131 E. Ethnic Minority Poverty Reduction begins with an Agricultural Transformation from Semi subsistence to Commercial Production 132 F. Successful Ethnic Farmers are Beginning to Diversify into Non-agricultural Employment, Particularly in Areas with Access to Major Cities or International Markets 134 G. Most Ethnic Minorities Continue to Live in their Communities of Origin 136 H. Ethnic Minority Poverty Reduction Strategies Follow a Series of Steps from Agricultural Specialization to Diversiï¬?cation and Accumulation of Financial, Social, and Cultural Capital 137 I. Prevailing Narratives of Ethnic Minority Livelihoods, Cultures, and Gender Relations are Shifting along with Diversiï¬?ed Development, although some Stereotypes Persist 140 CHAPTER 6 Is Inequality Rising in Vietnam? Perceptions and Empirics 145 A. Introduction 146 B. A Step Back: Why are we Concerned about Inequality? 147 C. Is Inequality of Outcomes Rising in Vietnam? 149 D. Why has Income Inequality Increased in Vietnam? 152 E. Inequalities in Opportunities that Perpetuate Income Differences across Generations 164 F. Inequalities in Connections, Voice, and Influence 170 Annexes Annex 1.1 New qualitative research carried out for the 2012 Vietnam Poverty Assessment 32 Annex 2.1 Differences between “Temporally Comparableâ€? and Comprehensive Welfare Aggregates 56 Annex 2.2: Spatial Cost-of-living Estimates for 2010 VHLSS 58 Annex 2.3 Subjective Poverty in Vietnam 59 Annex 3.1 Overview of Vietnam’s Eight Economic Regions 98 Annex 4.1 The Spatial Distribution of Poverty and the Gains from Spatial Targeting 113 Annex 6.1 Why doâ€? Perceptions of Inequalityâ€? Diverge from Empirical Measures of Inequality?174 Figures Figure 1.1 Growth and Poverty Reduction in Vietnam, 1993-2008 10 Figure 1.2 Progress at Reducing Poverty using GSO-WB and MOLISA Monitoring Systems 14 Figure 1.3 National Poverty Lines Rise with Average Per Capita Consumption: Developing andTransition Countries (2005 PPP) 22 Figure 1.4 Kinh and Ethnic Minorities: Average Annual Rates of Real Growth in Per Capita Expenditures, 1998–2010 26 Figure 1.5 Ethnic Minority Poverty Rates and Changing Composition of the Poor, 1993–2010 27 Figure 1.6 Growth in Income Per Capita by Income Group, 2004-10 28 Figure 1.7 Ratio of Ethnic Minority to Kinh Majoirty Enrolment Rates in Public Schools by Level of Education, 1998 and 2010 29 Figure 1.8 Out-of-pocket Spending per Student, by Education and Expenditure Quintile, 2004 and 2010 30 Figure 2.1 Composition of Per Capita Expenditures, 2010 VHLSS 45 Figure 2.2 Composition of Per Capita Expenditures by Per Capita Expenditure Quintile, 2010 VHLSS 45 Figure 2.3 Nutrition Norms Used to Anchor Poverty Lines in Different Countries 49 Figure 2.4 Measuring Subjective Poverty 54 Figure 2.5 Perceived Sufï¬?ciency of Consumption by Urban and Rural, 2010 55 Figure 3.1 Level and Composition of Poverty by Region, 1998 67 Figure 3.2 Level and Composition of Poverty by Region, 2010 67 Figure 3.3 Household Income by Expanded Quintile, 2010 68 Figure 3.4 Composition of Income by Expanded Quintile, 2010 68 Figure 3.5 Composition of Poor and Better-off Households in 2010, by Ethnicity 69 Figure 3.6 Distribution of Welfare for Kinh and Ethnic Minorities, 2010 71 Figure 3.7 Level and Composition of Poverty by Region, for Kinh/Hoa 71 Figure 3.8 Level and Composition of Poverty by Region, for Ethnic Minorities 71 Figure 3.9 Composition of Income for Extreme Poor, Poor, and Top Quintile in 2010: Comparing Kinh/Hoa and Ethnic Minority Households 73 Figure 3.10 Schooling Achievement by Age Cohort, 1998 and 2010 73 Figure 3.11 Education Achievements by Expanded Quintiles (persons age 21 and older) 75 Figure 3.12 Population Pyramids for Vietnam: 1999 and 2009 82 Figure 3.13 Monetary and Multidimensional Child Poverty in Vietnam, 2006-10 85 Figure 3.14 Multidimensional Child Poverty in Vietnam by Selected Sociodemographic Variables, 2006-2010 86 Figure 3.15 Child Poverty Rate by Domain, 2010 86 Figure 3.16 Distribution of Population on the Ofï¬?cial Poverty List by Expanded Per-Capita Expenditure Quintile, 2010 88 Figure 4.1 Relationship between the Poverty Rate and Gini Index 100 Figure 4.2 Poverty Rate and Proportion of Urban Population 100 Figure 4.3 Poverty Rate and Proportion of Ethnic Minorities 102 Figure 4.4 Poverty Rates, 1999 and 2009 109 Figure 4.5 Progress at Reducing Poverty, 1999-2009 by Poverty Rate in 1999 109 Figure 4.6 Change in Poverty, 1999-2009, Compared to the Initial Gini Index, 1999 109 Figure 4.7 District Poverty: MOLISA compared to Poverty Map Estimates 112 Figure 5.1 Changes in Welfare Levels ( per-capita consumption) for different Ethnic Groups in Vietnam,1998-2010 123 Figure 5.2 Real Per-capita Expenditures for Five Ethnic Categories, 2006-10 125 Figure 5.3 Changes in Net School Enrolment Rates for Kinh and Ethnic Minorities in Rural Areas, 1998-2010 127 Figure 5.4 Net School Enrolment of Selected Ethnic Minority Groups, 2009 128 Figure 5.5 Stunting among Children under Age 5 in Rural Areas, 1998-2010 129 Figure 5.6 Sources of Income for Majority and Minority Households in Rural Areas, 2010 135 Figure 5.7 Sources of Income by Quintile for Minority Households in Rural Areas, 2010 136 Figure 5.8 Paths to Successful Ethnic Minority Development 137 Figure 6.1 Ratio of Mean Per-capita Income by Percentile, 2004-2010 150 Figure 6.2 Mean Per-capita Rural Income per Year by Rural Income Decile, 2004-10 151 Figure 6.3 Theil Decomposition of the Level and Changes in Income Inequality, 2004-10 151 Figure 6.4 Growth by Income Socurce, 2004-2010, Ethnic Minorities 154 Figure 6.5 Growth by Income Source, 2004-2010, Ethnic Majority 154 Figure 6.6 Mean Annual Per-capita Rural Income per Year by Region, 2004-2010 155 Figure 6.7 Sector of Employment for Working-age Individuals in 1998, 2004 and 2010 157 Figure 6.8 Type of Occupation for Working-age Individuals in 1998, 2004 and 2010 157 Figure 6.9 Composition of Income in Urban Areas, 2010 159 Figure 6.10 Composition of Income in Rural Areas, 2010 159 Figure 6.11 Relative Concentration Coefï¬?cients of Different Sources of Income, 2010 160 Figure 6.12 Contribution of different Income Sources to the Gini, 2010 161 Figure 6.13 Per-capita Income per Year by Occupation of the Household Head in Rural and Urban Areas, 2004 and 2010 161 Figure 6.14 Workers Aged 25-30 by Education Level and Job Type 162 Figure 6.15 Hourly Wage and Labor Income Returns to Schooling 163 Figure 6.16 Per-capita Income per Year by Education of most Educated Working-age Household Member, Urban and Rural Households, 2004 and 2010 164 Figure 6.17 Ratio of Enrolments in Primary, Lower Secondary, and Upper Secondary School by Various Groups, 1998 and 2010 165 Figure 6.18 Average Rank in Math Test, by Wealth Quantile, at Ages 5, 8, and 15 Years 167 Figure 6.19 Average Rank in Math Test, by Initial Test Score and Wealth 167 Figure 6.20 Relative Importance of Circumstances for Health Opportunities 170 Figure 6A.1 District-level Expenditure Inequality, 1999 and 2009 175 Figure 6A.2 District-level Expenditure Inequality, 1999 and 2009 Absolute Gini Coefï¬?cients 175 Tables Table 1.1 Two Decades of Progress in Reducing the Number of Poor People 16 Table 1.2 Progress at Reducing Incidence, Depth and Severity of Poverty in Vietnam 17 Table 1.3 Improvements in Non-income Dimensions of Poverty, 1993-2010 18 Table 1.4 Contribution of HDI Components to HDI Growth, 1992-2008 19 Table 1.5 Vulnerability to Poverty Remains High in Vietnam 24 Table 2.1 Comprehensive Consumption Aggregates for the VHLSS, 2004, 2006, 2008, 2010 44 Table 2.2 Temporally Comparable Consumption Aggregates for VHLSS, 2004, 2006, 2008, 2010 44 Table 2.3 Spatial Cost-of-Living Index (SCOLI) for each Region and Sector 47 Table 2.4 Composition of the Reference Food Basket, 1993 and 2010 VHLSS 50 Table 2.5 Poverty Estimates for 2010: Comparing the GSO-WB Methodology and Ofï¬?cial Methodology 52 Table A2.1 Reference Food Basket for Different Population Groups 57 Table A2.2 Subjective Welfare Regression and Variables at Country Means 58 Table 3.1 2010 Poverty Headcount and Composition, by Region and Sector 66 Table 3.2 Poverty Headcount and Composition in 2010, by Sector of Employment of Household Head 67 Table 3.3 Ethnic Minority Poverty: Headcount and Composition in 2010, Region and Sector 69 Table 3.4 Kinh Majority Poverty: Headcount and Composition in 2010, by Region and Sector 70 Table 3.5 Poverty Headcount, Gap, and Severity in 2010, Kinh and Ethnic Minorities 70 Table 3.6 Poverty Headcount and Composition in 2010, by Education of Household Head 74 Table 3.7 Distribution of Completed Education in 2010, by Ethnicity and Expanded Quintiles 75 Table 3.8 School Enrolment Rates (net) for Boys and Girls in 2010, by Expanded Quintiles and Region 76 Table 3.9 Net School Enrolment Rates for Kinh/Hoa and Ethnic Minority Boys and Girls in 2010, by Expanded Quintile 77 Table 3.10 Average Landholdings for Rural Households in 2010, by Consumption Quintile 78 Table 3.11 Percentage of Rural Households without Allocated or Swidden Land 78 Table 3.12 Percent of Rural Households without Allocated or Sweden Land in 2010, by Region and Quintile 78 Table 3.13 Household Ownership Rates of Durables in 1998 and 2010 (Percent) 79 Table 3.14 Percentage of Households with Access to Housing and Neighborhood Amenities in 2010, by Quintile 80 Table 3.15 Poverty by City Size 81 Table 3.16 Percent of Households with Speciï¬?c Characteristics, by City Size 81 Table 3.17 Demographic Characteristics and Scale Economies for the Poor 84 Table 3.18 Percent of Households Experiencing Natural Disasters, 2003-08 87 Table 3.19 Percentage of Households Ofï¬?cially Classiï¬?ed as Poor, by Expanded Quintile, 2010 88 Table 3.20 Coverage of Social Protection and Poverty Reduction Policies by Expanded Quintiles 89 Table 3. 21 Coverage of Social Protection and Poverty Reduction Policies by Urban/Rural and Ethnicity 90 Table 4.1 Poverty Rate, Depth and Severity: Estimates from the 2010 VHLSS and the Small Area Estimation Approach 95 Table 4.2 Per-Capita Expenditure and Poverty Rate by Province and Region 96 Table 4.3 Inequality and Wealth Measures for Provinces in 2009 103 Table 4. 4 Rural Employment and Percent of the Working Population in Sector 110 Table A4.1 Impact on FGT2 of Targeting at Different Levels of Geographic Disaggregation Optimal Targeting Scheme 117 Table A4.2 Impact on FGT0 of Targeting at Different Levels of Geographic Disaggregation Optional Targeting Scheme 117 Table 5.1 Poverty and Median Expenditures of Major Ethnic Groups in Rural Areas, 2009 124 Table 5.2 Access to Public Utilities by Ethnicity in Rural Areas, 2004-10 130 Boxes Box 1.1 How does Vietnam Monitor Progress at Reducing Poverty? 13 Box 2.1 Do India’s New Ofï¬?cial Poverty Lines Measure Up? What are Lessons for Vietnam? 38 Box 2.2 How is Poverty Measured? 40 Box 2.3 How to value Housing Services in the VHLSS 43 Box 3.1 Deï¬?ning Characteristics of Poor Households at the end of the 1990s 65 Box 4.1 Overview of Program 30A 112 Box 5.1 Six “ Pillars of Disadvantageâ€? 112 Box 5. 2 An Ede Coffee “Hotspotâ€? 133 Box 5.3 Pineapples along the Border 135 Box 5.4 Equity in the Khmer Heartland 140 Box 5.5 Emerging Policy Recommendations: Ethnic Minority Poverty 142 Box 6. 1 Emerging Policy Recommendations: Inequality 173 Maps Map 3.1 Spatial Distribution of Poor Minorities 72 Map 3.2 Spatial Distribution of Poor Kinh 72 Map 4.1 Predicted Poverty Rates of Provinces and Districts, 2009 98 Map 4.2 Distribution of Poverty ( Number of Poor People), 2009 99 Map 4.3 Urban and Rural Poverty Rates in 2009 101 Map 4.4 Poverty Rates of Kinh/Hoa and Ethnic Minority Population in 2009 102 Map 4.5 Expenditure Gini Indices, 2009 105 Map 4.6 Ratio of the 90th Expenditure Percentile to the 10th Expenditure Percentile 105 Map 4.7 Proportion of People in the Richest Expenditure Quintile 106 Map 4.8 Provincial Poverty Rates 107 Map 4.9 District Poverty Rates 107 Map 4.10 Distribution of Poverty (number of poor people) in 1999 and 2009 108 Map 5.1 Regional Patters of Poverty and Wealth for Ethnic Minorities 126 EXECUTIVE SUMMARY Vietnam’s record on economic growth and poverty reduction over the last two decades has been remarkable. Using a “basic needsâ€? poverty line initially agreed in the early 1990s1, the poverty headcount fell from 58 percent in the early 1990s to 14.5 percent by 2008, and by these standards was estimated to be well below 10 percent by 2010. Similar progress in the face of steadily rising incomes is evident when assessed by “internationalâ€? standards of $1.25 and $2.00 person/day (2005 PPP). Progress has also been substantial in other dimensions of well-being, ranging from high primary and secondary enrolments to improvements in health status and reduced morbidity and mortality. Vietnam has achieved and in some cases surpassed many of the Millennium Development Goals (MDGs). Figure 1: Economic Growth and Poverty Reduction in Vietnam: Two Decades of Progress 100 18,000 1996Ͳ2000 SEDP 2001Ͳ2005 SEDP 2006Ͳ2010 SEDP 90 16,000 Per capita GDP (Thousand Jan. 2010 VND) 80 14,000 70 12,000 Poverty headcount (%) 60 10,000 50 8,000 40 6,000 30 4,000 20 10 2,000 0 0 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 $1.25/day 2005 PPP HCR $2.00/day 2005 PPP HCR GSOͲWB poverty line HCR Per capita GDP Despite remarkable progress, the task of poverty reduction in Vietnam is not complete. Vietnam’s “basic needsâ€? poverty line, agreed in the early 1990s, is very low by international standards, and the methods used to monitor poverty since the early 1990s are outdated: the poverty standards that applied to low-income Vietnam in the 1990s are no longer relevant to modern day, rising middle- income Vietnam. In addition, although tens of millions of Vietnamese households have risen out of poverty, many have incomes very close to the poverty line and remain vulnerable to falling back into poverty as a result of idiosyncratic shocks and related economy-wide shocks, such as the effects of climate change on rainfall and temperatures, human and animal influenza pandemics, and impacts of the 2008–09 global ï¬?nancial crisis. Economic growth has faltered in recent years as a result of continuing macro instability and sharp bouts of inflation. Despite this, citizens aspirations are rising, and Vietnam’s future development policies must reflect both its new economic realities and citizen’s rising aspirations for greater prosperity and economic security. In important respects, the task of poverty reduction has become more difï¬?cult. Vietnam’s success has created new challenges. The remaining poor are harder to reach; they face difï¬?cult challenges— of isolation, limited assets, low levels of education, poor health status—and poverty reduction has 1 The General Statistics Ofï¬?ce-World Bank (GSO-WB) poverty line was constructed in the late 1990s using data collected in the 1993 Vietnam Living Standards Survey (VLSS); it was presented in the 2000 Vietnam Poverty Assessment entitled Attacking Poverty, carried out by the joint government/donor/NGO Poverty Working Group. 1 become less responsive to economic growth. Ethnic minority poverty is a growing and persistent challenge. Although Vietnam’s 53 ethnic minority groups make up less than 15 percent of the population, they accounted for 47 percent of the poor in 2010, compared to only 29 percent in 1998. Using a new poverty line that better reflects living conditions of the poor (see below), 66.3 percent of minorities are poor in 2010 compared to only 12.9 percent of the Kinh majority population. Rapid structural transformation and Vietnam’s ongoing transition to a market economy have given rise to new patterns of development that bring additional challenges for poverty reduction. Inequality in incomes and opportunities are rising, underpinned by continuing disparities in human development between urban and rural areas and widening disparities within rural areas and across different socioeconomic groups. Poorer areas are still not well connected to markets. While there is good coverage of local infrastructure and basic services in most regions of the country, reliability (for example, of electricity) and quality of services is uneven. The country’s push towards modernization and faster industrialization has had mixed impacts on the overall quality of life in Vietnam. Urbanization is accelerating and a growing number of workers from rural areas are migrating to the cities to work in private industry and services. Many of these jobs are informal and lack the beneï¬?ts historically provided by the public sector and state-owned enterprises. There is a growing demand for young, skilled workers; many older workers do not, however, have the training or skills to compete for jobs in the expanding modern economy. A new Poverty Assessment was launched in 2011 and ï¬?nalized in December, 2012. It was led by the World Bank and the Vietnam Academy of Social Sciences (VASS), working in collaboration with the General Statistics Ofï¬?ce (GSO) and a team of local and international consultants. The Poverty Assessment takes a fresh look at the lives of poor men, women, and children and explores the constraints and opportunities they face today in rising out of poverty. It builds on a rich body of poverty analysis and an excellent base of knowledge from previous reports and aims to do three things. First, it proposes revisions to Vietnam’s poverty monitoring system—via better data, updated welfare aggregates, and new poverty lines—to bring these more in line with economic and social conditions in present-day Vietnam. Second, it revisits the stylized facts about deprivation and poverty in Vietnam, and develops an updated proï¬?le of poverty using data from the 2010 VHLSS and new qualitative ï¬?eld studies. Third, it aims to forge a consensus around some of the key challenges for poverty reduction in the next decade, including changing regional patterns of poverty and wealth, high and persistent poverty among ethnic minorities, and rising inequality in outcomes and opportunities. Improved Systems for Poverty Monitoring Vietnam has used two very different approaches to measure poverty and monitor progress over time. Both were initiated in the early 1990s and have evolved over time. The ï¬?rst approach was developed by the Ministry of Labor, Invalids, and Social Affairs (MOLISA), the agency identiï¬?ed by government in the early 1990s to have primary responsibility for Vietnam’s poverty reduction programs and policies. MOLISA is tasked with proposing ofï¬?cial urban and rural poverty lines at the beginning of each ï¬?ve-year Socio-Economic Development Plan (SEDP) and setting the initial period poverty rate. Using the ofï¬?cial lines, MOLISA is responsible for assessing changes in poverty and updating the ofï¬?cial list of poor households on an annual basis, using a “bottom-upâ€? mix of local surveys and village-level consultations to count the number of poor at local (commune) levels. These local counts are then aggregated up to estimate provincial and national poverty rates. Progress is assessed against poverty reduction targets set in the SEDP. The MOLISA lines were initially based on rice equivalents but since 2005 have been calculated using a Cost- of-Basic-Needs (CBN) methodology similar to the second approach (see below) led by GSO. The ofï¬?cial lines are not adjusted for inflation, but revised in real terms only every ï¬?ve years. MOLISA uses this approach to determine budget allocations and deï¬?ne eligibility for a number of targeted poverty reduction programs (for example, the National Targeted Program for Sustainable Poverty Reduction/NTP-SPR, Program 30a). The second approach is led by the GSO and measures poverty and monitors progress on the basis of nationally representative household surveys. GSO uses two different methods to measure poverty— one based on ofï¬?cial poverty lines (adjusted for inflation) applied to per capita incomes, and one using 2 an approach developed by a joint GSO and World Bank team in the late 1990s. The original GSO- WB poverty line was constructed using a standard Cost-of-Basic-Needs methodology, based on a reference food basket for poor households anchored in caloric norms (2,100 kilocalories per person per day) plus an additional allocation for essential nonfood needs based on consumption patterns of the poor. Unlike Vietnam’s ofï¬?cial poverty lines, the GSO-WB line was kept roughly constant in real purchasing power since the late 1990s, and applied to per capita consumption measured in successive rounds of the Vietnam Living Standards Survey (VHLSS) to estimate changes in poverty over time at the national, urban/rural, and regional level. The GSO-WB line has been used widely in Vietnam and in international fora to monitor changes in poverty since 1998. The national poverty rates reported in Figure 1 are based on the GSO-WB poverty line. The continuing use of the two separate systems for measuring and monitoring poverty, producing widely different poverty estimates, has at times complicated the dialogue between the development community and local researchers (who typically use the GSO-WB approach) and the government (which has tended to use the ofï¬?cial MOLISA approach). While the poverty trends from the two monitoring systems are similar—both show excellent progress--the poverty levels are very different, reflecting differences in methodology as well as differences in intended use. Vietnam’s ofï¬?cial poverty lines and methodology are constrained by resource availability; they are revised every ï¬?ve years in the work-up to the SEDP, and help Vietnam target scarce public resources to those most in need. In contrast, the GSO-WB poverty lines are independent of budget considerations and used only to monitor changes in poverty over time. Updating the GSO-WB Poverty Monitoring System Consistency in methodology and comparability over time are two of the great strengths of Vietnam’s poverty monitoring system. However, by 2009 it was clear that key aspects of the system had become outdated. The methods used to measure household well-being and construct the original basic needs poverty line were based on economic conditions and the consumption patterns of poor households in the early 1990s. Conditions have changed and Vietnam today is very different from Vietnam in the 1990s. In particular, the consumption patterns and living conditions of poor households today are substantially different from those in 1993, the reference period used to calculate the original GSO- WB poverty line. Beginning in 2009, a team from the World Bank worked closely with local and international experts and in collaboration with the GSO to update and improve Vietnam’s poverty monitoring system. The design of the 2010 VHLSS (and subsequent rounds) was improved and a new sample frame developed on the basis of the 2009 Housing and Population Census. The deï¬?nition of the consumption aggregate was updated to make it a more comprehensive measure of well-being, and new spatial cost-of-living indexes (SCOLIs) were calculated using a special survey of consumer prices carried out in conjunction with the 2010 VHLSS. An updated poverty line was constructed using an approach very similar to that of the original GSO-WB poverty line, but based on up-to-date consumption patterns from the 2010 VHLSS. The updated GSO-WB poverty line for 2010 is VND 653,000 per person per month (US $2.26 per person per day, 2005 PPP), which is substantially higher than the original GSO-WB poverty line. The increase reflects improvements in the quality of the food reference basket (fewer calories from rice, more consumption of proteins, vegetables, and fats) and a higher allocation for essential nonfood spending, including housing and durables. The updated “extreme povertyâ€? GSO-WB line is VND 435,000 per person per month (US $1.50, 2005 PPP). These compare to new ofï¬?cial poverty lines (announced in September, 2010) of VND 400,000 per person per month (US $ 1.29, 2005 PPP) for rural areas and VND 500,000 per person per month (US $ 1.61, 2005 PPP) for urban areas. According to the updated GSO-WB poverty line and methodology, 20.7 percent of Vietnam’s population is still poor in 2010, including 27 percent in rural areas and 6 percent in urban areas, and 8 percent of the population remains extremely poor. (Table 1) This compares to an ofï¬?cial poverty rate of 14.2 percent based on Vietnam’s ofï¬?cial urban and rural poverty lines set for the 2011-2016 SEDP. Although the regional distribution of the poor is similar between the two approaches, poverty levels are substantially higher in aggregate according to the GSO-WB methodology. However ofï¬?cial estimates 3 suggest higher poverty in urban areas, also in North Central and South Central coastal regions. The GSO-WB poverty rate is substantially higher in rural areas, in part due to differences between ofï¬?cial poverty lines and the new GSO-WB poverty line, but also due to differences in methodology. The GSO-WB poverty rate is calculated using a nationally representative household survey (the VHLSS) and detailed measures of household welfare; in contrast, MOLISA’s ofï¬?cial poverty rates are calculated at the commune level using a combination of short-form questionnaires and local consultations, then aggregated up from the commune level to province and national levels. Neither methodology is inherently better than the other. Rather, they are designed to serve different and equally valid objectives. The strength of the GSO-WB approach lies in consistent measurement over time and space, also its independence from budgetary or political considerations. It serves an important monitoring function. In contrast, Vietnam’s ofï¬?cial poverty lines and bottom up methodology are intended to help set targets and determine resource allocations for the government’s poverty reduction and social protection programs and policies. Table 1: New Poverty Estimates for 2010 by Region and Urban/Rural Areas WB-GSO Poverty Estimates Ofï¬?cial Poverty Poverty Extreme Poverty Estimates Population Poverty Contribution Poverty Contribution Poverty Contribution Shares (%) Rate (%) (%) Rate (%) (%) Rate (%) (%) All Vietnam 20.7 100 8.0 100 14.2 100 100 (national) Urban 6.0 9 1.5 6 6.9 6 30 Rural 27.0 91 10.7 94 17.4 94 70 Red River 11.4 12 2.8 8 8.4 13 22 Delta (Hanoi) East Northern 37.3 21 17.9 26 24.2 20 11 Mountains West Northern 60.1 9 36.5 14 39.4 9 3 Mountains North Central 28.4 16 9.7 15 24.0 20 12 Coast South Central 18.1 7 5.9 6 16.9 10 9 coast Central 32.8 10 17.0 13 22.2 9 6 Highlands Southeast 8.6 7 3.1 7 3.4 4 18 (HCMC) Mekong Delta 18.7 17 4.8 11 12.6 17 19 Revisiting the Facts about Poverty and the Poor The new GSO-WB poverty line is used to construct an updated proï¬?le of poverty based on the 2010 VHLSS, complemented by new information collected through Participatory Poverty Assessments (PPAs) and qualitative ï¬?eld studies. The poverty rate—deï¬?ned as the proportion of the population living below the poverty line--is a widely understood and frequently reported measure of poverty. But it ignores the fact that all poor people are not the same; some have incomes or consumption levels very close to the poverty line, while others live in much poorer conditions, well below the standards set by the poverty line. The new 2010 poverty proï¬?le differentiates between the total poor (individuals living below the GSO-WB poverty line) and the extreme poor (individuals whose per-capita expenditures are less than the extreme poverty line). In 2010, 20.7 percent of the population are poor and just over a one-third of these (8 percent of the population) are extremely poor. 4 The updated poverty proï¬?le shows that many of the factors that characterized Vietnam’s poor in the 1990s still characterize the poor today: low education achievement and limited job skills, heavy dependence on subsistence agriculture, physical and social isolation, speciï¬?c disadvantages linked to ethnic identity, and exposure to natural disasters and risks. Over the past decade, rising levels of education and diversiï¬?cation into off-farm activities have been powerful forces for poverty reduction. The remaining poor still predominately reside in rural areas and their livelihoods depend on agriculture and related activities. But some of the stylized facts about poverty in Vietnam have changed. Concerns about ethnic minority poverty were only beginning to emerge in the late 1990s; these have become much greater today as the gap continues to widen between minority populations (who make up 15 percent of the population) and the Kinh majority. The report documents great diversity across Vietnam’s 53 ethnic minority groups, and encouraging signs of progress for some minority groups in some regions. But the concentration of minorities among the poor has continued to rise; in 1993, poverty was widespread and minorities comprised only 20 percent of all poor households. By 1998, the share of minorities among the poor had increased to 29 percent, and by 2010 minorities account for 47 percent of the total poor and a resounding 68 percent of the extreme poor. The gap in living standards between ethnic minorities and the Kinh majority is very large: 66.3 percent of ethnic minorities are still poor in 2010 compared to only 12.9 percent of the Kinh, and a resounding 37.4 percent of ethnic minorities are still extremely poor, compared to only 2.9 percent of the Kinh. The majority of poor ethnic minorities continue to live in more isolated and less productive upland regions of Vietnam, and three-quarters of their total income comes from agriculture and allied activities. In contrast, poor Kinh have more diversiï¬?ed labor and earnings portfolios and live in coastal and delta regions. The depth and severity of poverty is much less for poor Kinh as compared to ethnic minorities. Our analysis suggests that agriculture will continue to be an important source of income for many of the poor, including but not limited to ethnic minorities. Compared to many other countries, agriculture land is equitably distributed in Vietnam. Despite the rapid expansion in opportunities for off-farm employment and concomitant income diversiï¬?cation over the last decade, the link between landlessness and poverty has increased, particularly in the Mekong Delta, Our analysis also shows that Vietnamese today are far better educated than they were a decade ago. Primary school completion rates were high already by the end of the 1990s. Since then, there has been a rapid increase in enrolments at lower and upper secondary levels, leading to an increase in the number of students who attend colleges and universities. Lack of education continues to be an important determinate of poverty: in 2010, 46 percent of poor households and 58 percent of extreme poor households are headed by persons who have not completed primary school. Gaps persist between enrolments for children from poor and better-off households. Most primary-school- aged children—rich and poor, minority and majority—are enrolled in school. But enrolments among (poor) minorities drop off at the lower secondary level, and children from lower-income households are much less likely to be enrolled in upper secondary schools than children from better-off households, perpetuating the intergenerational transmission of poverty in Vietnam. Differential enrolments also contribute to rising inequality. According to the 2010 VHLSS, 40 percent of persons 21 years and older in the richest quintile have completed a university degree; in contrast, less than 2 percent in the poorest quintile are university graduates. In fact, more than a quarter of those in the poorest quintile had not even completed primary school by 2010. The impacts of demographic factors on poverty have changed since the late 1990s. Child poverty continues to be a concern, although less so than in the 1990s, when poor rural households had many children and struggled to feed and educate them. As a result of family planning policies initiated in the early 1990s, most households now have only one or two children, and many of the adult children from the erstwhile large families in the 1990s are helping to support their parents and siblings. Aging is a new demographic risk; Vietnam’s population is aging and our analysis suggests that the elderly, particularly those who live alone, may be increasingly at risk of future poverty. Although targeting is good, existing poverty and social protection programs provide only partial coverage and limited beneï¬?ts to poor and at-risk individuals. In 2010, only half of the extreme poor reported that they were eligible to receive beneï¬?ts from the Ministry of Labor, War Invalids, and Social Affairs (MOLISA) poverty reduction programs. 5 Emerging Challenges: Changing Spatial Patterns of Poverty and Rising Inequality New poverty maps were developed based on the 2009 Housing and Population Census and the 2010 VHLSS. The maps show that poverty is becoming more concentrated in upland regions of Vietnam, including the North East and North West Mountains and parts of the Central Highlands. (Figure 2) In contrast, complementary household “wealthâ€? maps2 indicate that better off households are primarily concentrated in the Red River Delta (near Hanoi) and Southeast (near Ho Chi Minh City) and in urban centers along the coast. Although poverty rates are low in urban areas, lower income residents struggle to cope with the rising cost of living (including increases in electricity and water tariffs and rising fuel prices), and many work in the informal sector without social protection or employment beneï¬?ts. Urban poverty is most prevalent in Vietnam’s small cities and towns, which lag behind Vietnam’s larger cities in terms of basic infrastructure and public services. Figure 2: Poverty Rates (percent poor) in 1999 and 2009 1999 2009 Ethnic minorities make up 15 percent of the population in Vietnam and nearly half the remaining poor. New poverty maps show that minorities are concentrated in upland regions, with less infrastructure and much poorer connectivity. However location is not the only factor that explains the large gap in living conditions between ethnic minorities and the Kinh: according to Figure 3, even in the same (upland) districts, ethnic minority poverty is substantially higher (by a factor of 4-6 times) than poverty among the Kinh population. The persistent gap contributes to very high levels of inequality in poor regions with substantial minority populations. 4 Individuals in the wealthiest 15 percent of the population 6 Figure 3: Poverty Rates (percent poor) by Ethnicity in 2009 Kinh Ethnic Minorities The Poverty Assessment looks at inequality through two lenses— the ï¬?rst based on empirical analysis of various rounds of the VHLSS and the second drawing on ï¬?ndings from a new qualitative ï¬?eld study of “perceptions of inequalityâ€? that was carried out in sites throughout Vietnam. The perceptions study draws on a number of rich focus group discussions that describe which inequalities are viewed as unacceptable in the eyes of Vietnamese people, and also captures less easily measured inequalities, such as inequalities in connections, voice, and influence. It documents widespread concerns across the population about rising inequality. The quantitative analysis examines the factors driving the rise in inequality, including geographic variations in growth processes, growth in the non-agricultural sector, and disparities in education and ethnic identity. The rise in income inequality is in part a reflection of growth processes that have altered the relative returns to assets, such as education and productive capital in the economy. Growth has interacted with existing inequalities in opportunities—inequalities in education, access to good jobs, patterns of social exclusion, geographic disparities—to increase income inequality and welfare gaps between rich and poor households. The persistent and rising gap between the welfare of ethnic minorities and Kinh majorities also contributes to rising inequality. This study identiï¬?es many new avenues for future research. For example, more work is needed to better understand old and new sources of vulnerability, including urbanization and changing patterns of employment, and new research is needed on aging and health shocks. In addition, a more in-depth analysis of Vietnam’s targeted poverty reduction policies and programs is needed, with particular focus on policies designed to reduce poverty among ethnic minorities, where challenges clearly remain. Although Vietnam has successfully eradicated extreme poverty and hunger in all but a few isolated areas, there are widespread concerns about rising inequality in opportunities and outcomes. New work is needed to better understand these various sources of inequality and, more importantly, to understand what is the appropriate role of public policy in addressing these challenges. 7 Emerging Policy and Program Implications The Poverty Assessment focuses primarily on poverty and inequality diagnostics, and as such aims to support a better informed debate on policy and program responses among stakeholders in Vietnam, including government ministries, the National Assembly, local researchers and research institutes, INGOs and NGOs, international partners and the wider research community. Building on these diagnostics, work is underway with the Vietnam Academy of Social Sciences and other stakeholders in Vietnam to develop a more comprehensive policy framework for poverty reduction. The emerging framework has four areas of policy focus: • First, it is essential for Vietnam to reduce volatility and macro instability, and undertake the complementary structural reforms –-restructuring of the state owned enterprises, reforming the ï¬?nancial sector, raising the effectiveness of public investments and moving to a more transparent and open development process—necessary to put Vietnam back on the path of high and sustained economic growth. But the quality of growth matters as much of the rate of growth. • Second, measures are needed to make Vietnam’s future economic growth more inclusive, for example by supporting productivity and growth in the rural sector through improving connectivity, strengthening skills, improving the investment climate, expanding access to basic services, also better targeting agriculture support measures (e.g. credit, agriculture extension, and market information) to the needs of poor and ethnic minority farmers. Support for labor intensive industries and SMEs in both formal and informal sectors will also contribute to inclusive growth, including better access to credit and training, expanded vocational training for youth in poor and ethnic minority areas, and incentives for local enterprise development to provide more diversiï¬?ed employment options in local communities. The occupational and geographic mobility of labor should be enhanced: migration of rural workers into Vietnam’s rapidly growing cities has been a powerful force for growth and poverty reduction in the past. It is also important to reduce inequality of opportunities, including improving the quality of education and promoting skills development, particularly in rural areas and for ethnic minority groups. Improving governance through greater transparency and accountability will help to increase local participation and reduce inequalities in voice and power that work to undermine inclusive growth. • Third, policies to promote growth must be complemented by effective social insurance and social assistance policies. Vietnam should protect social spending and social assistance in the process of economic restructuring. Social beneï¬?ts and the ofï¬?cial poverty lines used to target these beneï¬?ts should be inflation-indexed, also adjusted to capture differences in the spatial cost of living, including between rural and urban areas, and to properly take into account basket of goods and services speciï¬?c to the poor. Better measures are needed to protect poor and vulnerable households from the rising cost of basic services, particularly rising electricity costs in the context of the planned energy subsidy phase-out. Migrant workers have been hard hit by the rising cost of living in urban areas; they should have equal access to basic services, portable beneï¬?ts (including health insurance), and better access to social protection programs in their new place of residence. • Finally, continuing improvements are needed in Vietnam’s poverty monitoring system so that it provides a reliable source of information for policy making in a rapidly changing economy. To this effect, objective resource-independent poverty lines should be used in parallel with resource-linked targeting lines, and the source and appropriate application of the two types of poverty lines should be communicated clearly to policy makers, practitioners, and the public. The construction of future poverty proï¬?les and poverty estimates should be done in an open and transparent way: more data on poverty, inequality, and social programs should be made publically available to facilitate monitoring of progress by independent experts and the public at large. 8 Chapter 1 Vietnam’s Growth and Poverty Reduction Record: Remarkable Success, but Big Remaining Challenges Vietnam has made remarkable progress at reducing poverty and promoting prosperity over the last two decades. But the task of poverty reduction is not yet ï¬?nished: shared growth, ethnic minority poverty, increasing vulnerability, and rising inequality are the major poverty challenges going forward. 9 A. Introduction 1.1 Vietnam has experienced high and sustained rates of economic growth over the last two decades, driven by a series of market-oriented reforms launched in the late 1980s. Initial progress was led by reforms in the rural economy, which led to a highly egalitarian distribution of agricultural land to rural households and diversiï¬?cation in on-farm activities, reforms that provided the right incentives for increases in farm production and export orientation. In recent years, job creation in the private sector has become a driving force behind Vietnam’s high economic growth, complemented by increased integration of agriculture in the market economy, and further opening of the Vietnamese economy to global trade and investment. Vietnam’s accession to the World Trade Organization (WTO) in early 2007 created opportunities for a new round of reforms, potentially leading to substantial changes in the policy and business environment, with major implications for economic growth and poverty reduction. But these opportunities are accompanied by new challenges and risks; growth has slowed in recent years, and Vietnam has struggled with periods of macro instability and bouts of high inflation. 1.2 Vietnam’s historical growth patterns have been remarkably pro-poor; growth in per capita gross domestic product (GDP) averaged 6.1 percent a year between 1993 and 2008, and poverty fell by an average of 2.9 percentage points a year (ï¬?gure 1.1). Figure 1. 1 Growth and Poverty Reduction in Vietnam, 1993-2008 100 18,000 1996Ͳ2000 SEDP 2001Ͳ2005 SEDP 2006Ͳ2010 SEDP 90 16,000 Per capita GDP (Thousand Jan. 2010 VND) 80 14,000 70 12,000 Poverty headcount (%) 60 10,000 50 8,000 40 6,000 30 4,000 20 10 2,000 0 0 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 $1.25/day 2005 PPP HCR $2.00/day 2005 PPP HCR GSOͲWB poverty line HCR Per capita GDP Source: WB-GSO poverty headcount calculated using 1993 and 1998 VLSS and 2004–2010 VHLSS. Dollar-a-day rates come from Povcalnet. Per capita GDP calculated using GSO population and GDP data. Note: HCR = Headcount Rate of Poverty, that is, incidence of poverty. 1.3 Despite remarkable progress, Vietnam’s task of poverty reduction is not complete, and in important respects, it has become more difï¬?cult. This chapter takes stock of Vietnam’s past record at reducing poverty and improving living conditions—acknowledging remarkable progress judged by any standards—and highlights several remaining and new challenges. It argues that the task of poverty reduction is by no means complete, and that it will become more difï¬?cult with growing affluence and rising aspirations, as Vietnamese society becomes more heterogeneous, market- oriented reforms continue, and Vietnam becomes more integrated into the global economy. B. Vietnam’s economy has grown rapidly and has undergone profound structural transformation 1.4 Comprehensive economic reforms were launched in the second half of the 1980s under Doi Moiand have accelerated over the last two decades. As a result of the reform process, the economy has been liberalized both internally and externally. The passage of the revised Land Law in 1993 and 10 the introduction of the Enterprise Law in 2000 were among the most important milestones in terms of domestic reforms. The accession of Vietnam to the WTO is widely recognized as a key milestone in the country’s external liberalization. Vietnam announced an ambitious plan to restructure the economy and shift into a new growth model in 2011, which is a new and important step in the country’s ongoing transition toward a market economy. 1.5 The Land Law of 1993 marked the continuation of a program of agricultural reforms that were initiated in 1988 with the implementation of Resolution 10. Resolution 10 radically changed the incentive system in the rural sector by recognizing, for the ï¬?rst time, that the household was the basic production unit of Vietnam’s agrarian economy and granted it the needed autonomy. With the aim of consolidating these changes, the 1993 Land Law granted households ï¬?ve basic rights: to transfer, exchange, inherit, rent, and mortgage their land. The law also extended the lease term to 20 years for annual cropland and 50 years for perennial cropland. The implementation of this law resulted in an extensive land titling program in Vietnam. In terms of scale and speed of implementation, it was one of the largest rural titling programs in the developing world (Iyer and Do 2008). Resolution 10 and the Land Law of 1993 together played a crucial role in boosting agricultural growth in the 1990s, thus enabling Vietnam to move from a food deï¬?cit country in the 1980s to one of the world’s largest rice exporters by the end of the 2000s. 1.6 A series of additional policy reforms outside the agricultural sector helped lay the foundation for rapid development of the private sector, whose role was ofï¬?cially recognized by Vietnam’s 1992 constitution. The most important milestone in the process was the Enterprise Law of January 2000. It represented a radical change in approach compared to the preceding Private Enterprise Law and Company Law, both of which were approved in 1990. Private enterprises were allowed to operate prior to 2000, but were subjected to a series of government approvals and controls. With the introduction of the new Enterprise Law, citizens were allowed to establish and operate private businesses with limited intervention from government ofï¬?cials. The most important innovation introduced by the Enterprise Law was the simpliï¬?cation of registration procedures and the associated elimination of a large number of business licenses, which sharply reduced transaction costs for businesses and helped install greater business conï¬?dence. As a result of these reforms, the number of registered enterprises increased by almost 15 times within only 10 years, from 31,000 in 2000 to 460,000 in 2009, according to the Ministry of Planning and Investment. 1.7 External liberalization has been accelerated at all levels—unilateral, bilateral, regional, and multilateral—over the last two decades. Beginning in the late 1980s, tariffs were unilaterally reduced, and numerous quantitative restrictions on trade abolished. Subsequently, Vietnam actively participated in bilateral and regional trade agreements. Membership in the Association of Southeast Asian Nations (ASEAN) in 1995 and its associated Asian Free Trade Area, and the U.S.-Vietnam Bilateral Trade Agreement in 2001, were important steps in the integration process. After 2003, Vietnam accelerated its negotiations for WTO membership and ofï¬?cially acceded to the WTO in January 2007. Becoming a WTO member has had important implications for Vietnam’s development, because of major changes taking place at the border (a reduction in import tariffs and removal of nontariff barriers to trade), beyond the border (greater access to overseas markets and to the WTO’s dispute settlement mechanism), and behind the border (opening of service sectors and distribution systems, changes in legal and regulatory frameworks,and so forth). Implementation of these agreements not only helped promote exports and restructuring in the domestic economy, but became key drivers for reform of key institutional underpinnings of a market economy, including legal and judicial structures. The Common Investment Law of 2005, for example, helped to harmonize treatment and regulation of all types of businesses including domestic ï¬?rms, foreign ï¬?rms, and cooperatives. 1.8 Two decades of reform have helped to sustain high growth in the economy and transform Vietnam in the process. Even with the marked slowdown in economic activity in the last few years in part caused by the international ï¬?nancial crisis, itself a reflection of Vietnam’s growing integration with the rest of the world, the Vietnamese economy has grown at an annual rate of more than 8 percent over the last decade. Today, the Vietnamese economy is four times larger than it was in the early 1990s, and the country now falls into the ranks of lower-middle-income countries. In 2010, per capita gross national income was more than US$3,000 (purchasing power parity [PPP]). 11 1.9 This growth has been accompanied by pronounced structural changes at the aggregate level. Twenty years ago, Vietnam was primarily rural, with nearly 80 percent of the population living in the countryside and only 20 percent residing in cities and towns. Moreover, the urban sector was dominated by two major economic and political hubs, Hanoi in the north, and Ho Chi Minh City in the south. In terms of GDP, slightly more than 40 percent of the economy was generated by agriculture, followed by services and then industry. Growth in the agricultural sector (cropping and farm sidelines) has played an important role in Vietnam’s development success. Nonetheless, its share of GDP has fallen to half of what it was in the early 1990s, and in 2010 contributed 20 percent of GDP. Industry, which includes manufacturing, construction, and utilities, has been the most rapidly growing and dynamic sector and currently makes up 38 percent of GDP. Services contribute 42 percent, modestly higher than the level in 1992. 1.10 These changes in the structure of the economy are largely mirrored in the composition of employment in Vietnam. In 1992, three-quarters of the labor force identiï¬?ed agriculture as their primary source of employment, with only 10 and 15 percent, respectively, in industry and services. Rapid productivity growth in the farm sector has contributed to rising incomes in the countryside; equally important, it has enabled the reallocation of a growing share of labor into even higher-value activities in industry and services. Today, the share of the labor force working in agriculture has fallen below 50 percent, while the share in both industry and services has doubled. 1.11 Accompanying this shift in the composition of employment has been a change in its type, most notably a reduction outside of agriculture in the role of self-employment (largely small, family-run businesses) relative to wage employment. The role of the state in wage employment has also fallen. Overall, however, the state actually employs a slightly larger percentage (upwards of 20 percent) of the labor force than it did in the early 1990s, reflecting the growth in wage employment in the state- owned enterprises sector. Urbanization, aided by increasing migration from the countryside, has also increased, but according to Vietnam’s 2009 census, only 30 percent of the population was classiï¬?ed as urban at that time. This puts urbanization in Vietnam at levels observed elsewhere in Southeast Asia about a decade ago.1 1.12 Thanks to external liberalization, Vietnam’s foreign trade has grown at more than twice the rate of GDP growth, and in 2010 the foreign trade ratio (imports plus exports as a percentage of GDP) was an unprecedented 165 percent. By comparison, and at its peak in China in 2006, it was only 70 percent. The composition of exports has slowly shifted. Exports of oil and agricultural products continue to remain important, but labor-intensive light manufacturing goods now represent the fastest-growing component of exports. Imports of capital machinery and intermediate goods dominate on the other side of the ledger. Export growth has been aided by the run-up in foreign direct investment in Vietnam, which rose from only US$0.5 billion in 1992 to around US$11.0 billion by 2010, with much of this occurring after WTO entry. Rapidly rising wages in China make Vietnam very appealing. Currently, foreign-invested ï¬?rms are the source of half of Vietnam’s nonoil exports. In terms of employment, however, these ï¬?rms still employ less than 2 percent of the labor force. 1.13 In addition to productivity growth, rising rates of investment in the domestic economy have been an important source of growth. This works through two channels—on the demand side, as an important source of growth in expenditure, and on the supply side, through investment’s role in expanding the country’s productive capacities and introducing new technology and know-how into the economy. Between 1992 and 2010, gross capital formation rose from only 17.6 percent of GDP to 38.9 percent, comparable to levels observed in the Republic of Korea; Japan; and Taiwan, China at their peaks. In 2010, the World Bank put domestic savings at 33.2 percent of gross national income. With the government sector typically running ï¬?scal deï¬?cits and state-run ï¬?rms net borrowers, the huge increase in savings is coming from a more than doubling in the savings rates of households and private enterprise. 1.14 Finally, reform and rising incomes have had a profound impact on household demographic behavior and population growth. In the early 1990s, average fertility rates of 3.4 births per woman 1 These numbers may underestimate the reduction of the share of employment in agriculture because of the growth in the countryside of secondary jobs in industry and services. In absolute terms, labor supply to agriculture is likely smaller today than it was when the reforms began. 12 translated into population rates of growth of nearly 2 percent per year. By 2010, fertility had fallen to 1.8, below replacement levels, and population growth to only 1 percent. Over the same period, average household size declined by nearly one person, from 5 to 4. With the sharp drop in fertility, the percentage of the population of working age has increased, pushing labor force participation rates upward from 50 to 60 percent of the entire population. Vietnam’s falling dependency ratio, that is, the ratio of those not working to those in the labor force, has had a direct impact on per capita incomes, and indirectly affected incomes through rising savings rates and investment and the “demographic dividend.â€? C. Progress in reducing poverty has been remarkable by any standard 1.15 Vietnam’s dramatic decline in poverty is evident across a number of different approaches used to monitor progress, whether assessed in terms of national poverty lines or using internationally comparable lines, or using household surveys or bottom-up community-based methods (box 1.1). The absolute number of poor people living in Vietnam has dropped sharply, and reductions in the poverty headcount have been accompanied by notable reductions in the depth and severity of poverty. However, progress has been uneven across regions and ethnic groups and has started to slow. Box 1.1 How does Vietnam Monitor Progress at Reducing Poverty? Vietnam has used two very different approaches to measure poverty and monitor progress. Both were initiated in the early 1990s and both have evolved over time. The ï¬?rst approach was developed and led by the Ministry of Labor, Invalids, and Social Affairs (MOLISA), identiï¬?ed in the early 1990s as the primary government agency responsible for poverty reduction programs and policies. MOLISA is tasked with proposing ofï¬?cial urban and rural poverty lines at the beginning of each ï¬?ve-year Socio-Economic Development Plan (SEDP) and setting the beginning period poverty rate. Using the ofï¬?cial lines and the beginning period poverty rate, MOLISA is responsible for assessing changes in poverty and updating its list of poor households on an annual basis, using a “bottom-upâ€? mix of local surveys and village-level consultations to count the number of poor at local (commune) levels, which are then aggregated up to calculate provincial and national poverty rates. Progress is assessed against poverty reduction targets set in the SEDP. The MOLISA lines were initially based on rice equivalents but since 2005 have been calculated (with technical support from General Statistics Ofï¬?ce[GSO]) using a Cost-of-Basic-Needs (CBN) methodology similar to the second approach (see below) led by GSO. The ofï¬?cial lines are not adjusted for inflation, but are revised in real terms only every ï¬?ve years. MOLISA’s primary objective using this approach is to determine budget allocations and deï¬?ne eligibility for a number of targeted poverty reduction programs (for example, the National Targeted Program for Poverty Reduction, and Program 30a). The second approach is led by the GSO and measures poverty and monitors progress on the basis of nationally representative household surveys. GSO uses two different methods to measure poverty—one based on ofï¬?cial poverty lines (adjusted for inflation) applied to per capita incomes, and one using an approach developed by a joint GSO and World Bank team in the late 1990s and ï¬?rst presented in the 2000 Poverty Assessment. The GSO-WB poverty line is constructed using a standard CBN methodology, based on a reference food basket for poor households anchored in caloric norms (through 2008, 2,100 kilocalories per person per day) plus an additional allocation for essential nonfood needs based on consumption patterns of the poor. Unlike Vietnam’s ofï¬?cial poverty lines, the GSO-WB lines have been kept roughly constant in real purchasing power since the late 1990s, and applied to per capita consumer expenditures measured in successive rounds of the Vietnam Living Standards Survey (VHLSS) to estimate changes in poverty over time at the national, urban/rural, and regional level. The GSO-WB lines have been used widely in Vietnam and in international discussions to monitor changes in poverty since 1993. We use these poverty rates in ï¬?gure 1.1. 13 The share of the population living below Vietnam’s national poverty lines has declined dramatically 1.16 Figure 1.2 shows historical poverty trends based on General Statistics Ofï¬?ce/World Bank (GSO-WB) estimates and ofï¬?cial poverty lines and methods. The continuing use of the two separate systems for measuring and monitoring poverty, producing widely different poverty estimates, has at times complicated the dialogue between the development community and local researchers (who typically use the GSO estimates) and the government (which has tended to use the ofï¬?cial MOLISA estimates). Although the different estimates sometimes caused confusion, the ongoing development and insistence on rigorous approaches to measurement has contributed to a better conceptualization of poverty on the part of government and the policy research community in Vietnam. Moreover the higher poverty rates produced by the GSO methodology, particularly in the 1990s, helped to keep poverty high on the government’s agenda. Figure 1.2 Progress at Reducing Poverty using GSO-WB and MOLISA Monitoring Systems 70 60 50 Heacount rate of poverty (HCR) (%) GSOͲWB poverty HCR Official MOLISA poverty HCR 40 30 20.7 20 14.2 10 0 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 Sources: WB-GSO poverty headcount calculated using 1993 and 1998 VLSS and 2004–2010 VHLSS. MOLISA estimates based on UNDP 2004; Government of Vietnam 2005; MOLISA 2011; and 2011 Vietnam Statistical Yearbook. 1.17 Over time, as the poverty rate fell (narrowing the gap between MOLISA and GSO estimates) and as the poverty estimates produced through the Vietnam Household Living Standards Survey (VHLSS) became increasingly recognized as valid and robust, MOLISA’s poverty estimates have become more aligned with those produced by the GSO. As part of the workup to the 2011–2016 Socio-Economic Development Plan (SEDP), the government agreed formally in Prime Minister’s (PM’s) Decision 60/20102 to separate the two important tasks of (a) targeting poor households for social assistance, on the one hand; and (b) measuring and monitoring poverty over time on the other. The aim is to build on the strengths of both systems. As part of this agreement, the GSO was given formal responsibility for producing national and provincial poverty estimates, based on successive rounds of the nationally representative VHLSS. MOLISA would concentrate on the task of identifying which individual households within provinces, districts, and communes should be included on the 2 PM Decision 60/2010 “On the Issuance of Principles, Criteria, and Norms for the Allocation of Development Investment Funding in the State Budget 2011–2015.â€? 14 MOLISA poverty list, with a ceiling deï¬?ned by the provincial poverty rates proposed by the GSO in consultation with MOLISA. The intention over the longer term is to align MOLISA and GSO’s poverty estimates at the national and provincial levels, with the aggregate number of households on the poverty list determined by GSO’s VHLSS-based measures of poverty based on ofï¬?cial poverty lines. 1.18 As part of this new arrangement, GSO and MOLISA worked together to develop a common methodology for producing the national and provincial poverty estimates, including the construction of new ofï¬?cial urban and rural poverty lines to be used for the period of the 2011–2015 SEDP. The team developed three options for the new ofï¬?cial lines, reflecting different requirements and living standards. The higher options included higher allocations for essential nonfood spending, based on consumption patterns of low-income households in the VHLSS. Following intensive discussion, the government chose the lowest of the three options. While the higher option was preferable on strictly methodological grounds, the government operates under a constrained budget and could not extend beneï¬?ts under the National Target Program for Sustainable Poverty Reduction (NTP-SPR) and other targeted programs to the anticipated large increase in eligible households—the higher-option poverty lines implied national poverty rates of 18 to 20 percent of the population. Given the inevitable tension between resource availability and needs, the MOLISA lines are often referred to as “budgetingâ€? or “planningâ€? lines, and the process of agreeing on ofï¬?cial poverty levels at the start of an SEDP, and annual targets for poverty reduction over the course of SEDP implementation, involve a range of technical, ï¬?nancial, and political considerations. As described in chapter 2, other countries face similar challenges. 1.19 In September 2010, Vietnam announced a new ofï¬?cial poverty rate of 14.2 percent (ï¬?gure 1.2). The ofï¬?cial poverty line for urban areas was raised from VND260,000 per person per month (US$1.34 person per day, 2005 PPP) to VND500,000 per person per month (US$1.61 per person per day, 2005 PPP). The ofï¬?cial line for rural areas was raised from VND200,000 per person per month (US$1.03 per person per day, 2005 PPP) to VND400,000 per person per month (US$1.29 per person per day, 2005 PPP). A second and higher set of ofï¬?cial “near-poorâ€? lines was also approved, allowing the government greater leeway in expanding eligibility criteria when deemed desirable, such as for determining eligibility for health insurance subsidies. The near-poor lines are 30 percent higher than the ofï¬?cial poverty lines—VND650,000 per person per month (US$2.24 per person per day, 2005 PPP) for households living in urban areas and VND520,000 per person per month (US$1.83 per person per day, 2005 PPP) for rural households—and similar in value (and implied national poverty rate) to the higher of the three poverty line options initially proposed. 1.20 The government set ambitious targets for poverty reduction in the 2011–2015 SEDP; poverty at the national level is targeted to fall by 2 percentage points each year between 2011 and 2015, and by 4 percentage points in the poorest communities, including those with high proportions of ethnic minority households. Achieving these targets will require a substantially higher rate of progress than achieved under the previous SEDP, and may be particularly challenging given the slowdown in economic growth and in the absence of substantially higher spending to support pro-poor policies and spending. Progress is monitored closely down to the commune level, and there are strong incentives for local authorities to meet these targets.3 1.21 New poverty estimates for 2011 were released by GSO in Vietnam’s 2011 Statistical Yearbook based on a new household survey (2011 VHLSS) covering nearly 47,000 households. Poverty in 2011 is estimated to have been reduced to 12.6 percent—a 1.6 percentage point reduction between 2010 and 2011. MOLISA released its own set of 2011 poverty estimates on March 28, 20124. According to these ï¬?gures, poverty is estimated to have been reduced to 11.8 percent—a 2.4 percentage point reduction between 2010 and 2011. According to MOLISA’s Decision 375, poverty fell most rapidly in Vietnam’s high-poverty regions—the West Northern Mountains (6.4 percentage points), the North Central Coast (5.7 percentage points), the Central Highlands (3.6 percentage points), and the East 3 Detailed work, including ï¬?eld studies carried out as part of the Poverty Assessment, indicate considerable variation in how resources for poverty reduction are used at the local level. There are incentives to show progress, and in some cases these incentives may cause ofï¬?cials to focus resources on households just below the poverty line (because progress is judged in terms of crossing the poverty line) rather than chronic or extreme poor. 4 MOLISA Decision 375/QÄ?-LÄ?TBXH issued on March 28, 2012. 15 Northern Mountains (3.2 percentage points). Poverty was estimated to fall by only 1.2 percentage points in the Mekong Delta, well below targets set in the SEDP. In response to a new resolution on social protection (Resolution 15) approved by the Central Party Committee in late 2012, MOLISA is developing new average and minimum living standards cut-offs that will provide a more scientiï¬?c basis for beneï¬?t levels linked to future (new) social assistance programs. The methodology used to calculate minimum living standards is similar to that used to calculate the 2010 GSO/WB poverty line. 1.22 For the present, given the differences in 2011 poverty estimates, and pending stronger implementation of agreements reached in PM Decision 60/2010, there is a strong rationale for continuing to use both the MOLISA approach (for targeting) and the GSO approach (for independent monitoring). We return to this issue in Chapter 2. 1.23 As part of the background work for this report, the team worked closely with the GSO to update the GSO-WB poverty line and related methodologies for poverty monitoring, to ensure that Vietnam’s methods for monitoring poverty fully reflect current economic and social conditions. The updated GSO-WB poverty line is VND653,000 per person per month (US$2.24 per person per day, 2005 PPP), which yields a poverty rate of 20.7 percent in 2010 (ï¬?gure 1.2, blue triangle for 2010). Chapter 2 describes proposed changes to the GSO-WB approach including improvements to the VHLSS, updated welfare aggregates, and construction of a revised 2010 GSO-WB poverty line. Note that poverty estimates using the new 2010 methodology are not strictly comparable to poverty estimates from recent rounds of the VHLSS for reasons presented in Chapter 2 and are explicitly set apart in the tables and ï¬?gures in the remainder of this chapter. The fraction of the population living below the international standards of US$1.25 and US$2.00 has also declined 1.24 Vietnam’s own poverty line(s) are clearly better for assessing progress and identifying remaining challenges within the country than international poverty lines. However, PPP-adjusted international poverty lines are often used to compare progress across countries. Vietnam’s progress at poverty reduction is equally impressive judged by international standards of US$1.25 and US$2.00 per person per day (2005 PPP). The poverty headcount fell from 63.7 percent using US$1.25 (2005 PPP) in 1993 to 16.7 percent by 2008, and from 85.7 percent using US$2.00 (2005 PPP) in 1993 to 43.3 percent by 2008, the last year for which comparable poverty rates were published by the World Bank (Table 1.2). Thus, poverty fell by an estimated 3 percentage points per year between 1993 and 2008, albeit with faster progress in the 1990s and ï¬?rst half of the 2000s than in recent years. In total, nearly half Vietnam’s population was lifted out of poverty in less than two decades 1.25 Measured by temporally comparable GSO-WB standards, more than 43 million people were lifted out of poverty between 1993 and 2008. A remarkable reduction in the number of poor men, women, and children living in Vietnam is also conï¬?rmed using PPP-adjusted international poverty lines. Table 1.1 Two Decades of Progress in Reducing the Number of Poor People Poverty standard Number of poor Change (millions) (millions) (% pts) 1993Ͳ 1993Ͳ 1998Ͳ 1993Ͳ 2008, 1993 1998 2008 1998 2008 2008 Annual Official GSOͲWB poverty line: consumption 39.8 28.2 12.3 Ͳ11.5 Ͳ15.9 Ͳ27.4 Ͳ2.9 $1.25/day (2005 PPP): consumption 43.6 37.5 14.3 Ͳ6.2 Ͳ23.1 Ͳ29.3 Ͳ3.1 $2.00/day (2005 PPP): consumption 58.7 59.0 36.9 0.4 Ͳ22.1 Ͳ21.8 Ͳ2.8 Sources: VASS 2010 for 1993–2008 GSO-WB headcount estimates; POVCALNET for 1993–2008 US$1.25 and US$2.00 headcount estimates. Population statistics taken from POVCALNET except for 2010, which come from World Bank Data on Vietnam web page, http://data.worldbank.org/country/vietnam. 16 The depth and severity of poverty have also fallen sharply 1.26 The poverty headcount is a widely understood and widely reported measure of poverty. However, it ignores the fact that all poor people are not the same; some have incomes or consumption levels very close to the poverty line, while others live in much poorer conditions, well below standards reflected in the poverty line. Two additional indicators are used to measure the depth and severity of poverty. The poverty gap (depth) measures the average, across all people, of the gap between the living standards of the poor and the poverty line. The squared poverty gap (severity) is calculated using a similar methodology, but gives greater weight to households whose living standards are further away from the poverty line. 1.27 According to table 1.2, Vietnam has made steady progress in reducing the depth and severity of poverty, whether measured by national or international standards. Living conditions not only have improved for households living near the poverty line, but also for many of Vietnam’s poorest households. Table 1.2 Progress at Reducing Incidence, Depth and Severity of Poverty in Vietnam Sources: VASS, 2010 for 1993–2008 GSO-WB headcount estimates; POVCALNET for 1993–2008 US$1.25 and US$2.00 headcount estimates; Statistics for 2010 calculated by the World Bank using the comprehensive consumption aggregate. Note: Poverty estimates using international poverty lines have not been published yet by the World Bank for Vietnam in 2010. But the rate of poverty reduction is slowing, linked to rising macro instability and slower growth 1.28 High and sustained rates of economic growth have been a key factor in Vietnam’s success at reducing poverty. But the economy has slowed in recent years. Beginning in late 2007, Vietnam has struggled with economic turbulence and inflation, with sharp and persistent increases in the prices of many basic commodities. Many workers lost jobs; others received lower wages and reduced working hours due to reduced demand during the global economic crisis in late 2008 and early 2009. Farmers complain that the costs of agricultural inputs are rising, and proï¬?t margins are reduced. There were again rising food prices and a sharp increase in the costs of electricity and fuel in 2010, which put additional pressure on household budgets. Households in urban and peri-urban areas have been particularly hard hit by high inflation, including rural-to-urban migrants who come to the city in ever growing numbers to seek better jobs and higher pay. Migrants send money home to rural areas; the impacts of higher urban prices are thus also passed on to households living in rural areas through declining remittances (see, for example VASS 2011). Urbanization is increasing at a rapid pace and the face of poverty and sources of vulnerability in urban areas differ in important respects from more traditional poverty concerns in rural areas. Vietnam has also achieved dramatic progress in improving the non-income dimensions of poverty and has met or is likely to meet most of the Millennium Development Goals (MDGs) 1.29 Table 1.3 documents progress along other dimensions of well-being. Vietnamese today are much better educated and arguably better prepared to get jobs in industry or services. In 1998, 25 percent of 17 persons aged 15 to 24 did not complete primary school. By 2010, only 12 years later, the percentage had fallen to only 4 percent, and upper secondary enrolments had nearly doubled (60 percent for girls, 54 percent for boys). Moreover, by 2010, there were more girls enrolled in both levels of secondary school than boys; Vietnam scores remarkably well in terms of gender parity in education. Table 1.3 Improvements in Non-income Dimensions of Poverty, 1993-2010 1993 1998 2010 Education % of 15ͲorͲolder who have not completed primary school 35.5 35.7 14.4 % of 15Ͳ24 who have not completed primary school 23.3 25.4 4.1 Primary enrollment rate (net) Female 87.1 90.7 92.8 Male 86.3 92.1 92.5 Lower secondary enrollment rate (net) Female 29.0 62.1 83.2 Male 31.2 61.3 80.2 Upper secondary enrollment rate (net) Female 6.1 27.4 60.1 Male 8.4 30.0 53.9 Health Immunization, DPT1, % of children ages 12Ͳ23 months 91 94 93 Immunization, measles, % of children ages 12Ͳ23 months 93 96 84 Infant mortality (per 1,000 live births) 34 29 14 Incidence of stunting (low height for age), children under 5 51 34 23 Incidence of underweight (low weight for age), children under 5 37 36 12 Life expectancy at birth (years) 68.1 71.0 74.8 % of poor with health insurance n/a 7.8 71.6 Access to infrastructure and durables % using electricity as main source of lighting 48 77 98 % with access to an improved* water source Rural 76 70 87 Urban 89 89 98 % with access to clean** water Rural 17 29 57 Urban 60 75 89 % with sanitary latrine 19 26 69 Rural 10 14 59 Urban 53 68 92 % of households with durable goods TV 22 56 89 Fan 31 68 85 Refrigerator 4 9 43 Car 0 0 1 Motorbike 11 20 76 ** Clean water is defined to include piped water, bottled water, water from deep wells with pumps, and rainwater. * Improved water sources are defined as clean water sources plus handͲdug, reinforced wells and filtered spring sources. Sources: 2010: immunization, malnutrition, and infant mortality statistics come from various rounds of the MICS; life expectancy from World Bank World Development Indicators database; all others from World Bank 2000. 18 1.30 Vietnamese today are also healthier and live longer than in the 1990s; infant mortality (deaths per 1,000 live births) had fallen to 14 in 2010, which is impressive even by middle-income standards, and life expectancy had risen to 74.8 years. There was also marked improvement in levels of nutrition, although stunting (low height-for-age) remains a concern in some regions of the country and among minority populations. While immunization coverage looks good on the surface—over 90 percent of children begin the recommended series of childhood immunization (for example, DPT1)—the 2010 Multiple Indicators Cluster Survey (MICS) documents immunization completion rates of only 60 percent (GSO 2011). 1.31 Access to infrastructure and local services improved; the number of households connected to the electricity grid increased from 77 percent in 1998 to nearly universal coverage (98 percent) by 2010. However, many households still do not have access to “improvedâ€? water sources,5 particularly in rural areas, or sanitary latrines. But while challenges in these areas remain, there have been dramatic improvements in coverage since 1998. 1.32 Improvements are also notable in housing quality and ownership of durables. By 2010, 89 percent of Vietnamese households owned TVs (compared to 56 percent in 1998), 85 percent owned an electric fan (compared to 68 percent in 1998), 43 percent owned a refrigerator (compared to 9 percent in 1998), and a substantial 76 percent owned at least one motorbike (compared to 20 percent in 1998). If affluence and quality of life are reflected, at least in part, in the consumer durables that people own and use, then there have been dramatic improvements since the late 1990s. 1.33 According to the most recent national Human Development Report (HDR) for Vietnam (UNDP 2011), the country has achieved or is likely to achieve most of the MDG targets by 2015. However, concerns about clean water and sanitation remain (Goal 10), and Vietnam continues to make slow progress toward environmental goals (Goal 9). Progress is also apparent in composite indicators of well-being 1.34 Recent years have witnessed a greater focus on composite indicators of poverty and deprivation in Vietnam, beginning with the Human Development Index (HDI) in the early 1990s, and more recently the Multi-dimensional Poverty Index (MPI) launched in the 2010 Vietnam HDR.6 The MPI builds on earlier work done to measure nonmonetary poverty, such as the approach to measuring child poverty developed by GSO and MOLISA with support from UNICEF, as well as the multidimensional poverty index used in the 2010 Urban Poverty Survey (UNDP 2011). 1.35 Vietnam has seen steady improvements in human development, evidenced by increases in the HDI over time: the HDI value increased 19 percent between 1992 and 2008. With an HDI of 0.728, Vietnam is now comfortably placed among medium human development countries (table 1.4). Table 1.4 Contribution of HDI Components to HDI Growth, 1992-2008 Year HDI Life Contribution Education Contribution Income Contribution Expectancy of Life Index of Education Index of Income Index Expectancy Index to HDI Index to Index to HDI Growth since Growth since since Previous Previous Previous Period (%) ( ) Period (%) ( ) Period (%) ( ) 1992 0.611 0.670 — 0.776 — 0.386 — 1995 0.639 0.690 18.8 0.808 25.9 0.420 55.3 1999 0.651 0.721 86.1 0.803 -13.9 0.430 27.8 2004 0.701 0.782 40.7 0.826 15.3 0.496 44.0 2008 0.728 0.794 15.2 0.830 5.1 0.559 79.7 Contribution to total change 35.2 N.A. 15.9 N.A. 48.95 in HDI 1992–2008 Sources: 2001 Vietnam HDR; HDI, 1992, 1995, 1999, 2004, 2008. Note: HDI = Human Development Index, N.A. indicates not available. 5 See table 1.3 for deï¬?nitions of “cleanâ€? and “improvedâ€? water sources. 6 The Government of Vietnam uses changes in the HDI and in the Gender Development Index as an indicator of progress toward human development and gender equality. Improvement in the HDI rank and value was also included as a target in the current SEDP 2001–2010. The SEDP 2011–2015 refers to improvements in the HDI as an indication of progress toward development goals, while the 2010 national MDG report cites positive change in the Gender Development Index as a sign of progress toward achieving gender equality and women’s empowerment. 19 1.36 The HDI is a composite index and there have been differences in progress for each of the different HDI sub-indices. Strong economic growth between 1992 and 2008 increased the income index by 45 percent. The life expectancy index also saw signiï¬?cant gains, rising by 19 percent between 1992 and 2008. This reflected steady improvements in average life expectancy from 65.2 years in 1992 to 72.7 years in 2008. The education index, which started from a relatively higher base in 1992, saw a slower rate of increase, rising by only 7 percent by 2008. The contribution of the education index to overall growth in the HDI decreased from around 25.9 percent from 1992 to 1995 to 5.1 percent from 2004 to 2008. Thus, since 1992, rising GDP, together with increased life expectancy, have been the main drivers of improvement in Vietnam’s HDI. Slowing gains in life expectancy are to be expected once years of life expectancy reach higher levels. However, slowing gains in the education index may be cause for concern. 1.37 There is a strong correlation between elements of good governance and higher levels of human development. Of the six dimensions of Vietnam’s Public Administration Performance Index (PAPI), public service delivery is most strongly correlated with the HDI, followed by transparency, participation at local levels, and vertical accountability. Similarly, control of corruption is also highly correlated with the HDI (CECODES, FR, CPP, and UNDP 2012). D. Despite this remarkable progress, the task of poverty reduction is not ï¬?nished 1.38 Vietnam has made remarkable progress toward its longstanding goal of eradicating poverty. By the end of the 2006–2010 SEDP, only 9.5 percent of households were estimated to live below Vietnam’s ofï¬?cial poverty lines, and poverty estimates based on the original GSO-WB basic-needs poverty line suggest similar results. Does this mean that the task of poverty reduction is ï¬?nished, except for addressing a few remaining pockets of poverty, and a continuing commitment to look after the poorest and most destitute? 1.39 The task may be ï¬?nished in terms of meeting the most basic food, shelter, and clothing needs of Vietnamese citizens. Vietnam rightly deserves to be recognized for this. But are these the right standards to apply in a rapidly growing, modernizing economy like Vietnam? The remainder of this chapter will discuss why the task of poverty reduction is not ï¬?nished in Vietnam, and indeed has become more difï¬?cult in many respects. 1.40 The task of eradicating poverty is not ï¬?nished because: â—? Standards have changed. By the end of the 2006–2010 SEDP, Vietnam’s system for measuring and monitoring poverty no longer adequately captured the living conditions of the population. The original GSO-WB poverty lines were set in the mid-1990s and do not reflect the consumption patterns or broader aspirations of the population today. â—? Many of the erstwhile poor remain vulnerable to slipping back into poverty. Weather shocks, health shocks, and exposure to other income shocks remain widespread, and in some areas may even be rising. 1.41 Moreover, Vietnam’s rapid pace of development has bred its own challenges. The economy has gone through massive changes since the late 1990s. Workers in their 40s and 50s made schooling and skills training decisions in a much different economy, based on a different set of incentives. Many do not have the skills or training to compete for jobs in today’s rapidly modernizing economy. Even young workers often leave school without adequate training for an expanding skills- based economy. 1.42 The task of eradicating poverty has become more difï¬?cult in other important respects. Growth rates have fallen sharply compared to the ï¬?rst half of the 2000s, and growth is expected to remain sluggish in the foreseeable future. In addition, poverty reduction is becoming less responsive to economic growth. The remaining poor are harder to reach; the easy wins due, for example, to land 20 reforms in the early 1990s, rapid expansion in rural areas into cash crop production, and agricultural diversiï¬?cation have for the most part been realized. The remaining poor are more concentrated in isolated regions and among ethnic minority groups, where structural issues linked to assets and location are binding constraints (for example, poorer-quality land, less education and training, and more limited infrastructure and public services). Poverty reduction policies and programs must reflect these changing realities. 1.43 Vietnam’s ongoing structural transformation to a market economy has given rise to trends that suggest new challenges for poverty reduction. â—? Inequality is back on the agenda. There are widespread concerns among Vietnamese citizens from all walks of life about rising inequality. Recent analysis suggests an increase in income inequality between 2004 and 2010, driven predominantly by growing inequality within rural areas. â—? Continuing disparities in human development contribute to income inequalities. While Vietnam has done a good job on coverage of basic services, quality is uneven, and there are large perceived gaps between better-off and poorer households and regions. With the push toward “socializingâ€? health and education services, access has become more closely linked to incomes, and the burden of out-of-pocket spending for health and education is rising. â—? Vietnam’s cities and towns are growing rapidly, due in part to a massive influx of migrants from rural areas of the country. The cost of living in urban areas is rising, due to rising food costs and to rising demand, higher fuel prices, and water and electricity tariffs. The private sector accounts for an increasing share of the urban labor force, and many continue to work in the informal sector without social protection or employment beneï¬?ts, as was revealed in a number of studies conducted in recent years such as the 2009 Urban Poverty Survey (Haughton et. al. 2010), various rounds of the Vietnam Academy of Social Sciences’ (VASS’s) Rapid Impact Monitoring (RIM) assessments of the global economic crisis (VASS 2009, 2011), and Oxfam-ActionAid’s urban poverty monitoring studies (Oxfam GB/ActionAid 2008, 2011). New forms of vulnerability are developing, in particular among workers in the informal sector and rural migrants in cities like Hanoi and Ho Chi Minh City. Poverty lines used to monitor Vietnam’s progress are low by international standards 1.44 When assessing Vietnam’s performance in recent years, it is important to keep in mind that both ofï¬?cial lines and the original GSO-WB poverty line are low by international standards, and, unlike in many other fast-growing economies, the GSO-WB line has not been revised since it was agreed in the mid-1990s. Using a constant standard to assess progress has many advantages. But most countries raise their standards—and their national poverty lines—as they become more affluent and as the aspirations and expectations of citizens change. Figure 1.3 shows the strongly positive relationship in developing and transition countries between national poverty lines (US$ per month, 2005 PPP) and average per capita expenditures (2005 PPP) (Chen and Ravallion 2008). The overall income elasticity of the national poverty line for countries in the sample is .66, with a substantially higher elasticity for the nonfood component of poverty lines (.91) than the food component (.47). Thus, assessed globally, the economic gradient in national poverty lines is driven more by the gradient in nonfood needs, which account for more than 60 percent of the overall elasticity. This is not surprising; food consumption becomes a much smaller share of total consumption as populations become more affluent. In countries like the United States, for example, even the poor spend only 20 to 25 percent of total expenditures on food. 21 Figure 1.3 National Poverty Lines Rise with Average Per Capita Consumption: Developing and Transition Countries (2005 PPP) 3 0 National poverty line ($/month at 2005 PPP) 0 2 0 0 0 0 0 3 4 5 6 7 Log consumption per person at 2005 PPP Note: Fitted values use a lowess smoother with bandwidth=0.8 Source: Chen and Ravallion 2008. 1.45 The poverty statistics cited in table 1.1 are based on the original GSO-WB poverty line of only US$1.10 per person per month (2005 PPP), which is substantially lower than the US$1.25 per person per day (2005 PPP) “internationalâ€? poverty line calculated by the World Bank and used to measure global progress at reducing poverty. The US$1.25 per person per day international poverty line sets a very low standard; it was constructed by averaging the national poverty lines for the 15 poorest countries in the World Bank’s database of comparator countries7 (Ravallion, Chen, and Sangraula 2008). Higher international poverty lines are typically used for rising middle-income countries. The median poverty line for all developing and transition countries is US$2.00 per person per day (PPP 2005), and the median line for all countries besides the poorest 15 countries is US$2.50 per person per day (PPP 2005). An international poverty line of $4.00 per person per day (PPP 2005) is used for a number of countries in Latin America. Vietnam’s poverty lines are low relative to its rising prosperity and concomitant rising aspirations 1.46 Poverty lines typically increase with economic development because norms change; what was considered an acceptable level of deprivation in the 1990s is no longer acceptable today. Poverty lines also rise because governments have greater capacity and more resources to respond to changing norms. 1.47 Evidence of changing norms is reflected in subjective poverty lines estimated using information reported by households in the 2010 VHLSS on the perceived adequacy of their current levels of consumption. Subjective lines suggest national poverty rates of 20 to 25 percent, substantially higher than current ofï¬?cial poverty estimates (Chapter 2). 1.48 Changing norms and higher aspirations are also captured in a number of qualitative ï¬?eld studies and assessments that have been carried out over the past decade. For example, in the 1999 and 2003 Participatory Poverty Assessments (PPAs) carried out by the World Bank in collaboration 7 Malawi, Mali, Ethiopia, Sierra Leone, Niger, Uganda, Gambia, Rwanda, Guinea-Bissau, Tanzania, Tajikistan, Mozambique, Chad, Nepal, and Ghana. 22 with other donors, international NGOs, and Vietnamese partners, poor respondents deï¬?ned well- being in terms of adequate food, a stable asset endowment (adequate land, labor, and housing), plus nonmaterial aspects such as community respect and freedom from debt and anxiety (ADB 2003; World Bank 1999). Respondents in the more recent 2008 PPA did not refer to hunger or food security, but instead spoke about risks related to rising food prices, concerns about access to employment, and stable jobs (in the face of emerging impacts from the global ï¬?nancial crisis). 1.49 In research on ethnic minority poverty for this report (Annex 1.1), ethnic minority respondents in three regions were asked about indigenous deï¬?nitions of success. The most common response was linked to sufï¬?ciency of basic needs: enough food to eat year-round, clothes to wear, decent housing, and ability to participate in cultural festivals and customs (such as being able to prepare a pig for the Tet festival). Other respondents realized that ideas of success were changing, pointing to increasing material prosperity and connections to the market economy. One minority ofï¬?cial in Muong Khuong district, Lao Cai, said: “In the past it was considered enough to be full and dress warmly (an no, mac am); now people want to eat well and dress beautifully (an ngon, mac dep).â€? Traders mentioned having a larger, cleaner multistoried house as a key indicator of success. Among respondents who have transitioned to trading or other nonagricultural work, the desire for children to be educated and have stable jobs, particularly in the state sector, also formed part of a concept of success. Thus, ideas of well-being, even among poorer Vietnamese, are shifting from satisfaction of basic needs to a higher asset base combined with social status and non-income factors such as health and education. Vietnam increased its ofï¬?cial poverty lines in late 2010, and a revision to the GSO- WB line is proposed in this report 1.50 Despite intense internal debate—many policy makers believe Vietnam should set more ambitious goals in the ï¬?ght against poverty, given its rapid economic growth and vision of itself as a modern industrial society—the new ofï¬?cial poverty lines set in 2010 for the 2011–2016 SEDP are still low by international standards. The new urban line is still well below US$2 per person per day (2005 PPP), and the new rural line is only a little above the US$1.25 per person per day lines applied in the world’s poorest countries. 1.51 As noted, the World Bank is working with the GSO and other local partners to update GSO’s poverty monitoring system, through improvements to the VHLSS household survey; more comprehensive welfare aggregates; and a revised GSO-WB poverty line, using an updated food reference basket (from the 2010 VHLSS), a more comprehensive measure of nonfood spending that includes the flow of consumption from household assets (consumer durables and housing), and new spatial cost-of-living indexes. Despite progress, many households remain vulnerable to falling into poverty in Vietnam, and new sources of vulnerability are emerging as a result of external global events and internal instability 1.52 Although tens of millions of Vietnamese households have risen out of poverty over the last decade, many have incomes very near the poverty line and remain vulnerable to falling back into poverty as a result of idiosyncratic shocks, such as job loss, accidents, death or illness of a household member, or economy-wide shocks, for example, effects of climate change on rainfall and temperatures, human and animal influenza pandemics, and impacts of the recent global ï¬?nancial crisis. The combination of large shocks and many small, often local shocks can be difï¬?cult to manage for poor, near-poor, and even nonpoor households. The strategies that households use to cope with unanticipated shocks, such as reducing spending on health care, selling off assets like land and livestock, and taking children out of school, can themselves have longer-term adverse consequences. At any point in time, apart from the households we observe living below the poverty line, there may be an additional number of households that face the risk of falling back into poverty—that is, households that remain vulnerable to poverty. 1.53 Some studies have equated vulnerability with the near-poor—households whose incomes lie above but still very close to the poverty line. As noted, Vietnam has deï¬?ned near-poor poverty lines 23 that are 1.3 times the ofï¬?cial poverty line. If a similar approach to deï¬?ning the near poor is applied to the 2010 GSO-WB poverty line, there were 13 million near-poor households in 2010 in addition to 18 million poor households. The 2008-2010 VASS poverty report (VASS 2011a) used a different methodology to measure vulnerability-to-poverty. The report analyzed poverty dynamics using a panel data set from the 2002, 2004, and 2006 VHLSS and found that one-fourth of those who were poor in 2002 were chronically poor (poor in all three periods), while the remaining three-fourths experienced temporary bouts of poverty and thus were labeled the transient or stochastic poor. The work found a great deal of churning—households moving above and below the poverty line—over the period, including a number of households that escaped poverty. Ethnic minority households were much more likely to be among the chronic poor. 1.54 Additional evidence is presented below, using a methodology initially developed and applied in a Poverty Assessment for China (World Bank 2009), to assess vulnerability to poverty based on a panel of 1,800 households from the 2004, 2006, and 2008 VHLSS. It constructs an index of vulnerability-to-poverty, deï¬?ned as the share of the population who were poor in at least one year (2004, 2006, or 2008) divided by the average poverty rate across all three years. The results summarized in table 1.5 suggest that a considerable number of households in Vietnam that are not poor in a speciï¬?c year nonetheless remain vulnerable to falling into poverty at some point in time. At the national level, only 7 percent of panel households were among the chronic poor (poor in all three years), despite an end-period (2008) poverty rate of 13 percent. Vulnerability to poverty was particularly high in wealthier areas of the country such as the Red River Delta (where Hanoi is located) and the Southeast (where Ho Chi Minh City is located). It was also surprisingly high in provinces in the South Central Coast and Mekong River Delta. Consistent with VASS ï¬?ndings, upland regions with a high proportion of ethnic minorities evidenced higher rates of chronic poverty. Table 1.5 Vulnerability to Poverty Remains High in Vietnam Consumption poverty (percent) (GSOͲWB) Average VulnerabilityͲ Poor in all 3 Poor in 2 of Poor in 1 of Poor in at Not poor in Headcount, Headcount, Headcount, headcount, toͲpoverty years 3 years 3 years least 1 year any year 2004 2006 2008 2004Ͳ2008 ratio (9) = (4) = [(6)+(7)+(8)] Subgroup (1) (2) (3) (1)+(2)+(3) (5) (6) (7) (8) /3 (10) = (4)/(9) National 7.0 6.7 12.3 26.0 74.0 20.0 13.7 13.0 15.6 1.7 (27) (26) (47) (100) Red River Delta 2.1 5.0 8.5 15.7 84.3 10.9 7.5 6.5 8.3 1.9 (13) (32) (54) (100) East Northern Mountains 10.4 10.3 10.8 31.5 68.5 26.3 17.3 19.0 20.9 1.5 (33) (33) (34) (100) West Northern Mountains 40.5 15.8 16.2 72.5 27.5 59.5 51.4 58.4 56.5 1.3 (56) (22) (22) (100) North Central Coast 10.3 11.5 19.9 41.7 58.3 32.5 25.7 15.6 24.6 1.7 (25) (28) (48) (100) South Central Coast 9.8 8.2 10.0 28.0 72.0 24.0 15.7 16.0 18.6 1.5 (35) (29) (36) (100) Central Highlands 19.1 10.3 3.9 33.3 66.7 31.8 27.9 22.2 27.3 1.2 (57) (31) (12) (100) Southeast 3.1 1.6 6.3 11.0 89.0 8.2 6.2 4.5 6.3 1.8 (28) (14) (57) (100) Mekong River Delta 2.2 4.2 20.0 26.4 73.6 16.9 6.7 11.5 11.7 2.3 (8) (16) (76) (100) Rural 8.8 8.2 14.3 31.3 68.7 24.4 16.6 16.0 19.0 1.6 (28) (26) (46) (100) Urban 0.7 1.6 5.3 7.5 92.5 4.4 3.6 2.5 3.5 2.1 (10) (21) (70) (100) Ethnic minority 34.0 19.4 15.3 68.7 31.3 59.7 49.0 47.5 52.1 1.3 (50) (28) (22) (100) Ethnic majority 2.6 4.6 11.8 19.1 80.9 13.6 8.0 7.4 9.7 2.0 (14) (24) (62) (100) Source: VHLSS tabulations using 2004, 2006, and 2008 panels of households. 24 1.55 Vietnam’s rich body of qualitative research on poverty documents continuing concerns about vulnerability. The 1999 Participatory Poverty Assessment (PPA) identiï¬?ed a number of important sources of vulnerability such as crop failures (weather shocks, insects and other pests, landslides), human disasters (severe illness, death of a laborer, alcoholism, drug addiction), other economic shocks (job loss, death of animals, business failures), and material crisis (damage to homes, theft, and violence). (Vietnam-Sweden Mountain Rural Development Program, ActionAid, Save UK, Oxfam GB 1999) 1.56 Risks were also discussed by respondents in the 2003 and 2008 PPAs. The 2008 PPA (see VASS 2009) highlights the fragile balance between opportunities and risks; households must grasp new economic opportunities in order to move out of poverty, but there are risks inherent in grasping new opportunities, and households may be pushed back temporarily into poverty as a result of setbacks, temporary loss of assets, or changes in family circumstances. Many households raised concerns about rising debt and worries about being caught in a “debt spiral.â€? There is widespread evidence that health shocks have pushed some households back into poverty; affected households report selling assets and taking on extra debt in order to cope with health shocks. 1.57 Activities are underway to monitor the impacts of recent shocks on poverty. Oxfam GB and ActionAid8 carried out an annual program of poverty monitoring in 12 sites in Vietnam (nine in rural areas, three in urban areas) between 2007 and 2011, and VASS (with active participation from development partners) carried out several rounds of a Rapid Impact Monitoring (RIM) study beginning in late 2008. (Oxfam GB/ActionAid 2008-2011; VASS 2011b) Results highlight the importance of occasional and often severe individual risks (for example, health related) coupled with more common seasonal risks that are local-context speciï¬?c (for example, bad weather) in affecting household living conditions. They also document the emerging impacts of “macroâ€? risks such as inflation and global economic crises. Even for the most affected groups, while macro risks worsened existing difï¬?culties (for example, reduced purchasing power), they were found to rarely cause households to relapse into poverty. However, risk and vulnerability were noted as important causal factors in chronic poverty, and were linked to slow poverty reduction among ethnic minority households. Evidence from the RIM and related studies suggests that the 2009 global crisis had a negative but short-lived impact on the living standards of poor households, with particularly adverse effects on Vietnam’s large pool of migrants workers—many of whom work in factories with foreign links (via export production or foreign employers)—and rural households whose livelihoods depend on migrant remittances. 1.58 Three new qualitative ï¬?eld studies were carried out for this report highlight new and old sources of poverty and vulnerability (short summaries are provided in Annex 1.1). Low-income respondents in a study designed to explore “perceptions of inequalityâ€? raised concerns that inflation could widen the gap between the poor and better-off and thereby further reduce opportunities to access education, health care, and other services. Competition for jobs will increase if the economy continues to slow, and good jobs are likely to go to applicants with the right connections or who are willing to pay bribes to potential employers. Concerns about land acquisition have been widely discussed in the press, and were raised again in the perceptions of inequality study as well as a new study carried out jointly by the World Bank and Oxfam to identify the “long-run drivers of poverty reductionâ€? in Vietnam. Erstwhile rural households living in or near urban centers felt vulnerable to having their cultivable land acquired for industrial and other development purposes. Few felt they would be properly compensated for the loss of land, and most saw land acquisition as resulting in an inevitable drop in living standards. A third “positive devianceâ€? study of poverty among ethnic minorities analyzed a range of concerns speciï¬?cally linked to poverty and progress among ethnic minorities. Minorities depend heavily on earnings from agriculture, both crops and animal products, and were particularly vulnerable to weather shocks and other natural disasters, also to commodity and input price volatility. Ethnic minority respondents were acutely aware of the substantial and persistent gap in living conditions between minority and Kinh households, which they attributed to a number of factors including e.g. gaps in opportunities and differences in treatment. 8. This monitoring was conducted for Oxfam GB and ActionAid by the Ageless Consulting Company. 25 Poverty is increasingly concentrated among Vietnam’s ethnic minority populations, who comprise less than 15 percent of the population but nearly half the remaining poor and two-thirds of the extreme poor. 1.59 Vietnam has 54 ofï¬?cially recognized ethnic groups, of whom the Kinh (Viet) are by the far the most numerous, accounting for nearly 74 million people (85.7 percent of the total population) according to the 2009 Population and Housing Census. In 2009, there were ï¬?ve other ethnic groups (the Tay, Thai, Muong, Khmer, and H’mong) with populations of more than 1 million, and another three (the Nung, Dao, and Hoa) whose populations are between 500,000 and 1 million. There are also a number of ethnic groups with populations of less than 5,000 people. With the exception of the Hoa (Chinese), Khmer, and the Cham, most ethnic minority groups live in highland or upland areas, away from the coastal plains and major cities. The largest minority populations are found in the North-West and North-East and the Central Highland regions, although there are also ethnic population clusters in the North-Central, South-Central, and Mekong regions. Figure 1.4 Kinh and Ethnic Minorities: Average Annual Rates of Real Growth in Per Capita Expenditures, 1998–2010 16 14 Kinh/Hoa Ethnic Minorities 12 10 8 6 4 2 0 Red river East West Northern South Central Southeast Mekong Rural Urtan National Delta Northern Northern Cental Cental Highlans River Delta Moutains Moutains Coast Coast Sources: 1998 VLSS and 2010 VHLSS. 1.60 Despite remarkable progress in reducing overall poverty, including a steady reduction in ethnic minority poverty, there remains a substantial and widening gap in living conditions and poverty rates between the Kinh majority and ethnic minorities. This is illustrated in ï¬?gure 1.4, which graphs annualized real rates of growth in per capita expenditures (from the 1998 VLSS and 2010 VHLSS) between 1998 and 2010, by region and ethnicity. Since 1998, per capita expenditures have grown at an average annual rate of 9.4 percent for the Kinh and only 7.4 percent for ethnic minorities. Disparities are largest in some of the poorest and least accessible regions of Vietnam. As discussed in Chapter 6, in recent years growth in income has been uneven across minority households, with higher rates of growth among the better-off. Even the fastest-growing minority households are growing more slowly than the average Kinh households. 1.61 Consistent with differential rates of growth, the concentration of minorities among the poor is rising; in 1993, poverty was widespread and minorities comprised only 20 percent of all poor households (ï¬?gure 1.5). By 1998, the share of minorities among the poor had increased to 29 percent, and by 2010, minorities accounted for 47 percent of the total poor in Vietnam and a 26 resounding 66 percent of individuals in the poorest 10 percent of the population. According to the updated GSO-WB poverty line, 66.3 percent of minorities were poor in 2010 compared to only 12.9 percent of the Kinh. Figure 1.5 Ethnic Minority Poverty Rates and Changing Composition of the Poor, 1993–2010 Composition of Poor by Minority/Majority Poverty Rate for Minority/Majority 100 90 80 80 70 60 60 50 40 40 30 20 21 10 0 0 1993 1998 2004 2006 2008 2010 1993 1998 2004 2006 2008 2010 Ethnic Minorities Kinh/Hoa Ethnic Minorities Kinh/Hoa Sources: 1993, 1998 VLSS; 2004, 2006, 2008, 2010 VHLSS. 1.62 The increasing concentration of minorities among the poor and extreme poor is a serious concern. But not all minorities are poor. There is encouraging evidence of improvements in welfare levels and livelihoods for many minority groups in recent years, and recent analysis of the 2010 VHLSS documents the presence of some better-off ethnic minority households among middle- and upper-income deciles. These issues are explored in greater depth in Chapter 5, which describes encouraging signs of progress in some areas and among some groups, and identiï¬?es important pathways for progress. In recent years, growth has favored the better-off, resulting in rising income inequality 1.63 Past work suggests that Vietnam’s development trajectory was one of growth without an appreciable rise in inequality (VASS 2010). The picture has evolved in recent years, however, and there is growing evidence of rising inequality. A new study of citizen perceptions of inequality carried out as background for this report (Annex 1.1) suggests a widespread sentiment that inequality has risen; the sentiment is shared widely across rural and urban populations, and by both rich and poor. 1.64 The annual rate of growth in real household incomes averaged 8 percent between 2004 and 2010, based on successive rounds of the VHLSS. However, growth since the mid-2000s has been uneven across households, with richer households experiencing stronger growth than poorer households. The variation in growth across households is a reflection of a number of powerful, and potentially opposing, changes in the economic fabric: changes in the returns to education and skills in labor markets, sectoral and occupational transitions, and geographic mobility as individuals leave rural areas in search of work. These forces interact with initial differences in human capital and access to services, as well as “proceduralâ€? and institutional inequalities, such as differences in voice and participation among social groups and access to power and influence, to generate differences in living conditions across the population. 1.65 Figure 1.6 presents a growth incidence curve9 using per capita income and shows growth rates by ranked income group between 2004 and 2010. Real income growth rates over the period varied considerably for households at different points in the income distribution, ranging from around 4 percent for households at the bottom of the income distribution to 9 percent for households at the top. Growth was pro-poor, in as much as it contributed to continued progress toward reducing poverty over the period. However, because growth has favored better-off households, both the relative and absolute gap in incomes between the rich and the poor has risen over time. 27 Figure 1.6 Growth in Income Per Capita by Income Group, 2004-10 45000 9 40000 8 VnDong(Jan2010prices) 35000 7 AnnualizedGrowth 30000 6 25000 5 20000 4 15000 3 10000 2 5000 1 0 0 1 2 3 4 5 6 7 8 9 10 RuralIncomeDecile 2004 2010 AnnualizedGrowth2004Ͳ2010 Source: 2004, 2010 VHLSS. 1.66 The uneven growth process has contributed to rising inequality and is contributing to concerns about increasing social and economic disparities. The Gini index of income inequality has risen modestly from 0.40 to 0.43, adjusted for variations in prices across regions. Inequality in Vietnam in 2010 was comparable to that in other middle-income countries in the region, such as Indonesia and Thailand, although it was lower than in China. This growth has been accompanied by a shift in the share of income from the bottom 60 percent of the population to the top 40 percent. The share of income accruing to the top decile increased by 2 percentage points between 2004 and 2010. To place this ï¬?gure in context—the increase in the share of income going to the top 10 percent was almost as large as the total share of income going to the bottom 10 percent in Vietnam in 2010. Meanwhile, over the same period, the share of income accruing to the bottom 10 percent decreased by 20 percent. Focusing on the top tail of the income distribution, the share of income of the top 5 percent rose from 20.6 percent to 22.5 percent between 2004 and 2010. In this respect, the patterns are similar to those in China and India, where the top 5 percent of income earners earned 20.5 and 21.3 percent of income and consumption, respectively (ADB 2012). 1.67 The trend of rising inequality with economic growth is common across many developing countries in the East Asia and Paciï¬?c region. While rising income inequality may be a manifestation of growth processes that raise overall income and reduce poverty, and can thus be considered a natural consequence of an economic landscape favoring entrepreneurship, innovation, and economic progress, if left unchecked some types of inequalities can lead to rising social tensions and to undermining social cohesion. The “perceptions of inequalityâ€? study documents “acceptableâ€? and “unacceptableâ€? sources of inequality; wealth is acceptable (and admired) if achieved through hard work, luck, or acquiring more and better education. But wealth obtained through illegal means or misuse of power or influence is not acceptable. As Vietnam continues to grow and basic needs poverty is no longer the primary concern, it will be increasingly important to monitor and promote equitable growth processes that ensure all Vietnamese share in beneï¬?ts of rapid development. 9 A growth incidence curve plots the annual rate of growth between two points in time for speciï¬?c percentiles of the income distribution (Ravallion 1997). 28 Disparities in other aspects of human development remain and in some cases appear to be widening 1.68 Vietnam has not only succeeded in raising incomes. Progress in human development has been equally impressive. But as in the case of income growth and poverty reduction, progress has been uneven. Inequalities may undermine growth processes if they are driven by disparities in circumstances—such as ethnicity, gender, and unequal opportunities for acquiring a good education—that ultimately prevent some groups from beneï¬?ting equally in the gains from high growth and development. 1.69 Consider the example of education. Figure 1.7 depicts the ratio of enrolment rates for majority children compared to enrolments for several ethnic minority groups. A ratio of less than 1 indicates that minority children are participating in school at a lower rate than the majority. Although there has been considerable progress since 1998, ethnic minority populations continue to have lower enrolments than the majority, and these differences are substantial at the upper secondary level. Figure 1.7 Ratio of Ethnic Minority to Kinh Majoirty Enrolment Rates in Public Schools, by Level of Education, 1998 and 2010 1 0.9 NetEnrolmentRate 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 1998 2010 1998 2010 1998 2010 Primary LowerSecondary UpperSecondary TayͲThaiͲMuongͲNung KhmerͲCham OtherNorthernUplands CentralHighlands Source: 1998 VLSS, 2010 VHLSS. Incomes matter in terms of access to quality health and education services 1.70 The growing emphasis on “socializationâ€? in the provision of health and education services in Vietnam—which stresses the sharing of social costs and responsibilities between individuals and the state and non-state sectors—means that incomes are beginning to matter more for determining access to basic services. Rising disparities in incomes will contribute to rising social disparities, including disparities in school enrolments (particularly for secondary and higher education) and access to health services. 1.71 A direct consequence of this is that the burden of out-of-pocket health and education expenditures is substantial, particularly for less-well-off households. Analysis based on the VHLSS shows that spending on education rose in real terms between 2004 and 2010 across all levels (ï¬?gure 1.8), and out-of-pocket costs are higher as students move from primary to lower and upper secondary levels. Compared to the poor, better-off households spend substantially more on education in general and in particular on extra courses and after-school tutoring. Given these advantages, it is not surprising that students from wealthier households perform better in the classroom and on standardized tests, and are more likely to obtain higher degrees and training. 29 Figure 1.8 Out-of-pocket Spending per Student, by Education and Expenditure Quintile, 2004 and 2010 12,000 10,000 8,000 6,000 4,000 2,000 0 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 pperSecondaryͲ Up econdaryͲ UpperSe 4 CollegeͲ 2004 geͲ 2010 Colleg 2004 010 20 Tuition ntributiontoSch Con hoolorFund ms,Textbooks,St Uniform tationery ExtraCourses Otherexxpenditures Source: 2004, 2010 VHLSS. 1.72 Research suggests that while ill health is more concentrated among the poor, they are less likely than the better-off to use health services (World Bank 2012). Moreover, the distribution of public spending in the health sector decidedly favors the better-off; spending on commune health centers, utilized by the rural poor, is small compared to spending on government hospitals utilized by the better-off. Concerns have been raised about the impoverishing effects of catastrophic health costs, including that the poor will forego care when faced with serious illnesses. Most of the poor have free health cards, which help to reduce the costs they pay for services, but with concomitant concerns about the quality of care they receive. A number of studies highlight Vietnam’s high out-of-pocket (OPP) health payments; these persist despite improvements in the coverage of the National Health Insurance Scheme as a result of the 2008 Law on Health Insurance. The new law provides fully subsidized health insurance premiums for the poor, and partially subsidized premiums for the near- poor. However health insurance has had a modest impact on reducing out-of-pocket health payments (Lieberman and Wagstaff 2008; Wagstaff 2007) including catastrophic health costs. Households with young children and elderly members have higher exposure to health risks and report higher rates of catastrophic health spending. (Hoang Van Minh et. al. 2012) Urban residents face signiï¬?cant challenges of rising costs and economic instability 1.73 Vietnam has weathered the global economic storm following the ï¬?nancial crisis of 2008–09 better than most countries. Growth hit a decade-low 5.3 percent in 2009, down from a decade-high 8.5 percent just two years before, but in 2010 it bounced back to 6.8 percent. It slipped again to 5.9 percent in 2011, but remained more than 1 point above the average for emerging and developing economies. Growth in 2012 was only 5.7 percent. 1.74 Behind this resilience, however, is a more complicated story of volatility and vulnerability, which plays out in Vietnam’s cities and towns. As export demand fell following the global ï¬?nancial crisis, so did demand for factory labor. Fortunately, the labor market bounced back quite quickly and strongly, in terms of number of working hours and wages in nominal terms. Urban residents were buffeted by inflationary shocks before and after the crisis. In 2008, the GSO reported a price increase of 23 percent overall as Vietnam felt the effects of the global food crisis—with food price inflation registering at 34 percent. Inflation moderated in 2010, but rose again in 2011, to around 18 percent nationally, in both urban and rural areas, with a steeper rise in the price of food and foodstuffs and electricity and fuel. 30 1.75 These events have brought considerable challenges for urban residents, which have been documented in a number of studies and rapid assessments including those by Oxfam/ActionAid, VASS, and the UNDP/GSO cited earlier. For example, 65 percent of households surveyed in the 2009 Urban Poverty Survey reported higher prices for food and essential items as a source of difï¬?culties, making inflation by far the most common factor among job loss, business slowdowns, natural disasters, health shocks, and others (16 percent of households reported job loss or business slowdown as a source of difï¬?culty). On a positive note, a price impact survey undertaken by Oxfam GB and ActionAid in May 2011 found that inflation has not caused families to go hungry or children not to attend school (which may be due to parents giving top priority to their children’s education). Still, serious issues remain. Those living off of savings or ï¬?xed incomes, which are not inflation- indexed, such as pensioners, beneï¬?ciaries of social protection, and those unable to work due to health issues, are vulnerable to the effects of inflation in obvious ways. 1.76 Combined with employment instability like that introduced by the global recession, inflation also poses especially acute issues for migrants who move to urban areas seeking better work. Migrants already tend to face higher prices for accommodation, electricity, and water than local residents and have difï¬?culty accessing social services; they are therefore especially endangered by instability in their livelihoods. Migrants surveyed in Oxfam/ActionAid’s fourth round of participatory monitoring of urban poverty (Oxfam/ActionAid 2011) reported that wage increases have failed to keep pace with price increases; their average monthly expenditures net of savings and remittances increased 87 percent between 2008 and 2011, while monthly income increased only 66 percent. There have been signs of rising labor tension as a result of this dynamic, and a reduction in remittances to rural areas. Instability in urban livelihoods bears not just on urban poverty, but, via this remittance mechanism, on poverty in rural areas, as well. E. Overview of the report: Vietnam’s old and new poverty reduction challenges 1.77 This report takes the view that despite remarkable progress, the poverty reduction task in Vietnam is not complete. The report aims to do three things. 1.78 First, it proposes revisions to Vietnam’s poverty monitoring system in Chapter 2, including improvements to the VHLSS, more comprehensive welfare aggregates, and a new poverty line, with the aim of bringing these more in line with economic and social conditions in present-day Vietnam. Second, Chapter 3 uses the new methodology to revisit the stylized facts about deprivation and poverty in Vietnam, and develops an updated proï¬?le of poverty using data from the 2010 VHLSS and new qualitative ï¬?eld studies. Third, the report selectively analyzes some of the key challenges for poverty reduction in the next decade. Chapter 4 presents new poverty maps based on the 2009 Population and Housing Census and 2010 VHLSS and compares these to earlier poverty maps based on the 1999 census. Chapter 5 focuses on ethnic minority poverty, with the aim of identifying not only the current constraints faced by minority populations but also by documenting important signs of progress. Chapter 6 takes a new look at inequality of outcomes and opportunities, combining analytic work using the VHLSS with ï¬?ndings from the qualitative study of perceptions of inequality. 31 Chapter Annexes Annex 1.1: New qualitative research carried for the 2012 Vietnam Poverty Assessment (1) “Positive devianceâ€? study on ethnic minority poverty This ï¬?eld study, carried out from November 2011 – February 2012, aimed to identify ethnic minority communities that show unusually strong poverty reduction and income growth and identify factors contributing to these positive results. Positive deviance is a methodology that originates in Vietnam, from a 1990s nutrition program led by Save the Children; it has since been applied worldwide by NGOs and researchers (Marsh et al 2004, Ramalingam 2011). Successful families and communities are “positiveâ€? since they escape poverty despite facing the same challenges and obstacles as their neighbors, and “deviantsâ€? (or outliers) because they practice different behaviors from others. The researchers visited ethnic minority communities in Dak Lak province (Ea H’leo district), Tra Vinh province (Chau Thanh and Tra Cu districts) and Lao Cai province (Muong Khuong and Bac Ha districts), conducting semi-structured interviews with over 100 ethnic minority residents and local government ofï¬?cials. Sites were selected using a combination of quantitative analysis and a snowball sample based on expert recommendations and secondary literature. Data from census samples was analyzed to determine rates of poverty reduction (or increase) among ethnic minority respondents only in each province and district over the periods 1999-2006 and 2006-09. Census data was also processed to calculate the mean reported expenditures of ethnic minority respondents (as a proxy for income) by province and district and the percentage of the ethnic minority sample in the top 15 percent of expenditures that resides in each location. A series of qualitative hypotheses was then developed of possible factors leading to poverty reduction and income growth, outlining “provocative propositionsâ€? for qualitative data collection that were explored through interviews and observation in ï¬?eld sites. (2) Identifying Long Run Drivers of Poverty Reduction: The Q-square pilot Oxfam and the World Bank carried out a qualitative pilot study in August, 2011 to identify what have been key long run drivers of poverty reduction over the past two decades in Vietnam. The study was framed around the complementary concepts of opportunities and constraints in assessing income and welfare dynamics at the household and community levels. The longer run aim was to develop a panel data set of households and communities spanning 20 years, drawing on the initial set of communities and households surveyed in the 1992/93 and 1997/98 VLSS. Sites were selected from the 1997/98 VLSS list of districts/communes based on district-level poverty rates and the district-level population of ethnic minorities and Kinh/Hoa. Efforts were made to visit a range of locations, roughly representative of Vietnam’s different regions. In total, the team interviewed 220 households that had been initially surveyed in the VLSS panel, updated household rosters for these households, and held groups discussions with nearly 250 respondents at both village and commune levels. A series of qualitative exercises were carried out including (i) wealth ranking; (ii) time-line exercises are used to explore commune and village histories since 1992 and (iii) card-sorting exercises and mobility diagrams to list and rank opportunities and constraints in the communities over the two decades. Village ofï¬?cials are also asked to discuss their perceptions of how life had changed, what had happened to poverty levels since the early 1990s. Additional life-history interviews and diagrams are conducted with representatives from selected households, focusing on households who had done exceptionally well (and why) or done very poorly (and why). The team also interviewed important ’change agents’ such as local businesses, cooperatives, shops, and projects/programs. (3) Exploring Perceptions of Inequality in Vietnam A ï¬?eld study was carried out in March and April 2012 that aimed to collect and analyze information on perceptions of inequality held by diverse elements of Vietnamese society. The work explored three key areas: (i) perceptions of social and economic disparities within and between different reference 32 groups; (ii) the factors that determine these perceptions, and (iii) the extent to which disaparities have changed over time. Discussions were organized around a number of reference focus groups i.e. better off households, poor households, senior citizens, groups of students as well as working young people, and (in the case of urban areas) rural to urban migrants. Three sentinel groups of sites were selcted -- six locationsn in metropolitan cities, two locations in smaller cities, and seven locations in rural areas. Four overlapping aspects of inequality were higlighted by all groups – inequalities in economic outcomes (incomes, wealth), as well as inquailities in access to education, jobs, and land. Causes of inequality were seen as overlapping and complementary e.g. some rural respondants raised concerns about the poor quality of education in their areas, which contributed to poor skills and unequal access to good jobs. There was strong support for policy measures to ensure equality of opportunities. Many respondents, particularly young, educated people living in urban areas, were tolerant of inequalities in outcomes – for example, ownership of fancy cars, big houses, and the lastest technology – so long as these gains were earned through hard work and legitimate means. Many groups raised concerns about ill-gotten gains, bribery and misuse of power leading to rising inequalities. And there were widespread concerns about ’procedural’ inequalites – the gaps in how systems were supposed to work in principal but failures of systems to work properly in practice e.g. implementation of land compensation policies. 33 References Asian Development Bank. 2003. “Participatory Poverty and Governance Assessment: Central Coast and Highlands Regionâ€?, Hanoi. Asian Development Bank. 2012. Outlook 2012: Confronting Rising Inequality in Asia. Manila: Asian Development Bank. Center for Community Support and Development Studies (CECODES), The Front Review of the Central Committee for the Viet Nam Fatherland Frong (FR), Commission on People’s Petitions of the Standing Committee for the National Assembly of Viet Nam (CPP), and the United Nations Development Program (UNDP). 2012. The Viet Nam Governance and Public Administration Performance Index (PAPI): Measuring Citizen’s Experiences. Hanoi. Chen, Shaohua and Martin Ravallion. 2008. “New Global Poverty Estimates.â€? World Bank Research Digest 3 (1, Fall): 4. Government of Vietnam. 2011. Statistical Handbook. Hanoi. Haughton, J., Nguyen Thi Thanh Loan, and Nguyen Bui Linh. 2010. “Urban Poverty Assessment in Hanoi and HCMC.â€? Joint publication of the UNDP and Vietnam General Statistics Ofï¬?ce, Hanoi. Hoang Van Minh, Nguyen Thi Kim Phuong, Priyanka Saksena, and Chris D. James. “Financial Burden of Household Out-of-Pocket Health Expenditure in Vietnam: Findings from the National Living Standards Survey 2002-2010.â€? Social Science and Medicine 30 (2012): 1-6. Hoang, Xuan Thanh, Nguyen Thu Phuong, Vu Van Ngoc, Do Thi Quyen, Nguyen Thi Hoa, Dang Thanh Hoa, and Nguyen Tam Giang. 2012. “Perceptions of Inequality in Vietnam: A Qualitative Study.â€? Background paper prepared for the 2012 Vietnam Poverty Assessment, Hanoi. Iyer, Lakshmi, and Quy-Toan Do. 2008. “Land Titling and Rural Transition in Vietnam.â€? Economic Development and Cultural Change 56 (3). Leiberman, Samuel and Adam Wagstaff. 2008. Health Financing and Delivery in Vietnam: the Short and Medium Term Policy Agenda. Hanoi: World Bank. Marsh, D., D. Schroeder, K. Dearden, J. Sternin, and M. Sternin. 2004. “The Power of Positive Deviance.â€? British Medical Journal 329 (7475): 1177–1179. Nguyen Tam Giang and Hoang Xuan Thanh. 2012. “Long-run Drivers of Poverty Reduction in Vietnam between 1992 and 2011.â€? Background paper prepared for the 2012 Poverty Assessment, Hanoi. Oxfam GB/ActionAid. 2011. “Participatory Monitoring of Urban Poverty in Vietnam: Fourth Round Synthesis Report 2011,â€? Hanoi. Oxfam GB/ActionAid. 2008. “Participatory Monitoring of Urban Poverty in Vietnam: Synthesis Report 2008,â€? Hanoi. Ramalingam, B. 2011. “A Q&A on Positive Deviance, Innovation and Complexity.â€? February 8. Accessed September 3, 2011. http://aidontheedge.info/2011/02/08/a-qa-on-positive-deviance- innovation-and-complexity/. Ravallion, Martin and Shaohua Chen. 1997. “What Can New Survey Data Tell Us about Recent Changes in Distribution and Poverty?â€? World Bank Economic Review, 11(2): 357-382. Ravallion, Martin and Shouhua Chen. 2007. “China’s Uneven Progress against Poverty.â€? Journal of Development Economics 82: 1-42 Ravallion, Martin, Shoahua Chen, and Prem Sangraula. 2008. “Dollar a Day Revisited.â€? World Bank Research Digest 2(4, Summer): 1-16. 34 Turner, Sarah (2011) “’Forever Hmong’: Ethnic Minority Livelihoods and Agrarian Transition in Upland Northern Vietnam.â€? The Professional Geographer. Vietnam Academy of Social Sciences. 2009. “Participatory Poverty Assessment: 2008 Synthesis Report,â€? Hanoi. Vietnam Academy of Social Sciences. 2011a. Poverty Reduction in Vietnam: Achievements and Challenges. Hanoi. Vietnam Academy of Social Sciences. 2011b. “Rapid Impact Assessment - Vietnam in 2011: Synthesis Report,â€? Hanoi. Vietnam-Sweden Mountain Rural Development Programme, ActionAid, Save the Children (UK), and Oxfam (GB). 1999. A Synthesis of Participatory Poverty Assessments from Four Sites in Vietnam: Lao Cai, Ha Tinh, Tra Vinh, and Ho Chi Minh City. Hanoi: World Bank. UNDP. 2001. Doi Moi Processes and Human Development: Vietnam Human Development Report 2001. Hanoi. UNDP. 2011. Social Services for Human Development: Vietnam Human Development Report 2011. Hanoi. Wagstaff, Adam. 2007. “Health Insurance for the Poor: Initial Impacts of Vietnam’s Health Care Fund for the Poor.â€? Policy Research Paper No. WEPS 4134. Washington DC: World Bank. World Bank. 1995. Vietnam: Poverty Assessment and Strategy. Report No. 13442-VN. Washington DC: World Bank. World Bank. 1999. Vietnam Development Report 2000: Attacking Poverty. Washington DC: World Bank. World Bank. 2003. Vietnam Development Report 2003: Poverty. Hanoi: World Bank. World Bank. 2006. Vietnam Development Report 2007: Vietnam Aiming High. Hanoi: World Bank. World Bank. 2009. From Poor Areas to Poor People: China’s Evolving Poverty Reduction Agenda – an Assessment of Inequality and Poverty. Washington, DC: World Bank. World Bank. 2012. Health Equity and Financial Protection Report: Vietnam. Washington DC: World Bank. 35 Chapter 2 Updating Vietnam’s Poverty Monitoring System Vietnam’s poverty monitoring system was updated to reflect changing economic conditions since the first Vietnam Living Standards Survey was conducted in 1993. New, comprehensive consumption aggregates were created using data from the 2010 Vietnam Household Living Standards Survey (VHLSS). The GSO-WB poverty line was updated using these aggregates: the new line is 653,000 VND per person per month, yielding a national poverty rate of 20.7 percent. 36 A. Introduction 2.1 Vietnam has a robust system for monitoring changes in poverty, based on a long-running system of nationally representative, comparable Vietnam Household Living Standards Surveys (VHLSS); consistent estimates of household welfare; and a poverty line that has been kept constant in real purchasing power since the mid-1990s, when it was agreed between the General Statistics Ofï¬?ce (GSO), the World Bank (WB), and other development partners.10 Consistency in methodology and comparability over time are two of the great strengths of Vietnam’s poverty monitoring system. However, by 2009, it had become clear that key aspects of Vietnam’s poverty monitoring system were outdated. The methods used to measure household welfare and construct the original GSO-WB poverty line were based on economic conditions and the consumption patterns of poor households in the early 1990s. Conditions have changed: Vietnam today is very different from Vietnam in the 1990s. The economy is more diversiï¬?ed and better integrated in the global economy. Connectivity and access to markets have improved, even for households living in more remote rural areas. In addition, the production structure of households has changed: households have access to a much wider array of consumer goods, and they purchase more food from the market rather than producing it at home. Incomes are more diversiï¬?ed, and there has been a rapid shift out of agriculture and into industry and services. These changes affect households across the income distribution. Especially important for determining a poverty line, the consumption patterns of poor households today are substantially different from those of the 1990s. 2.2 This chapter describes revisions and updates to Vietnam’s poverty monitoring system, including improvements to the 2010 VHLSS (and subsequent rounds), revisions to the deï¬?nition of household welfare to make it a more comprehensive measure of well-being, new indexes to adjust for spatial cost-of-living differences, and an update to the original GSO-WB poverty line. The methodology to construct the new poverty line is consistent with the original GSO-WB methodology, but is based on new information from the 2010 VHLSS.11 The revisions described in this chapter result in a higher estimate of poverty for 2010 than the original GSO-WB poverty line would have yielded, and, particularly for rural areas and areas with high numbers of ethnic minority households, higher poverty estimates compared to ofï¬?cial estimates. Reasons for these differences are also discussed. 2.3 The chapter also describes a new methodology for estimating “subjectiveâ€? poverty lines, drawing on experimental questions introduced in the 2010 VHLSS. Poverty estimates based on the subjective poverty line are very similar to those using the updated GSO-WB poverty line. 2.4 The 2010 VHLSS can only give reliable estimates of poverty at the national level, for urban and rural areas and by region. This is due to sample size and design of the sample of the VHLSS, which includes information on both expenditures and incomes. Chapter 3 describes a small-area estimation (poverty mapping) methodology that can be used to estimate poverty at lower levels of disaggregation—in Vietnam’s case, for provinces and districts—and presents new district- and provincial-level poverty maps based on the 2009 Population and Housing Census and 2010 VHLSS. B. Rethinking Poverty and Poverty Measurement in Vietnam 2.5 Poverty is deï¬?ned as unacceptable deprivation in well-being. But well-being can encompass a multitude of dimensions, and there are many different views about what constitutes an acceptable (or unacceptable) standard of living. In many countries, setting (or revising) the poverty line involves active public debate and a careful balancing of political and scientiï¬?c considerations. The enormous 10 The original GSO-WB poverty line was prepared as input to the 2000 “Poverty Assessment Attacking Poverty.â€? 11 A similar methodology was used in 2005 by a team of local and international experts, led by the Ministry of Labour Invalids and Social Affairs (MOLISA), to update Vietnam’s ofï¬?cial poverty lines for the 2006–2010 Socio-economic Development Plan and by MOLISA and GSO more recently to construct ofï¬?cial poverty lines for the 2011–2015 Socio- economic Development Plan. 37 public response, in India and internationally, to the Indian Planning Commission’s announcement of new poverty estimates and revised urban and rural poverty lines provides a recent example of the challenges inherent in updating poverty lines, with some interesting parallels to current discussions in Vietnam. Many in India feel that the new ofï¬?cial poverty lines are far too low (box 2.1). Box 2.1 Do India’s New Ofï¬?cial Poverty Lines Measure Up? What are Lessons for Vietnam? The Indian Planning Commission released a new set of poverty estimates and new poverty lines in March 2012. Many observers believe the new poverty lines are much too low—29 rupees per person per day for rural households (just under US$1.25 2005 Purchasing Power Parity [PPP]) and 32 rupees per person per day for urban households (US$1.65 2005 PPP). The Planning Commission’s new estimates showed a 7-percentage-point drop in poverty, the largest drop since the ofï¬?cial poverty rate was ï¬?rst calculated in 1962. The announcement caused a furor in the Indian and international press: Indian poverty lines have always been low by international standards, and the new lines were seen as a missed opportunity to rectify this. One important criticism is that the nutrition standards embedded even in India’s new lines continue to be based on the sparse diet that the poor consumed in the 1973–74 National Sample Survey (NSS). Like in Vietnam, consumption patterns in India have changed substantially since these standards were set. Another criticism is that India’s new poverty lines do not “constitute an adequate deï¬?nition of poverty because they do not take into account malnutrition, sanitation, drinking water, housing and health needsâ€? (Gill 2012). Vietnam’s updated 2010 poverty lines take full account of housing, durables, nutrition, clean water and sanitation, and health needs. If India is using the same methodology it used in the past, why the big controversy now? Over time, the Indian poverty line has increasingly been used as a cut-off to determine eligibility for India’s social welfare schemes and targeted poverty reduction programs. People who fall below the poverty line are eligible for a range of social beneï¬?ts; states receive funds for some poverty reduction programs (for example, the Public Distribution System, which distributes subsidized rice to poor households) according to the number of residents who fall below the ofï¬?cial poverty line. So where the poverty line is set is not just a statistical artifact, but an important policy decision that determines the eligibility of millions of families for public support. The Government of India cannot afford a poverty cut-off that is too high, and—as the controversy continues—it appears that the people of India will not accept a poverty cut-off that is too low. In a recent article in the Hindustan Times, Abhijit Banerjee, Ford Foundation International Professor of Economics at MIT, suggested that the way out of the current muddle is to have “two different poverty lines: an ethical poverty line to describe the standard we should aspire to … and an administrative poverty line which tells us how to best target our limited resources. As [India] gets richer, perhaps the latter will be raised till it is effectively the same as the former. But right now we don’t want to hurt the poorest [by spreading resources too thinly] in the name of being more aggressive about povertyâ€? (Banerjee 2011). Sources: Banerjee 2011; Gill 2012. 2.6 Vietnam’s ofï¬?cial poverty lines for the 2011–2015 Socio-economic Development Plan are more akin to Banerjee’s concept of an administrative poverty line: they are designed to help target limited public resources to those most in need, and should be judged by that standard. The updated GSO- WB poverty line better captures what Banerjee refers to as an ethical poverty line; it reflects what Vietnam should aspire to achieve. The good news is that compared to the situation in the 1990s, Vietnam’s administrative and monitoring poverty lines are not very far apart. Moreover, the ofï¬?cial poverty lines help to target poverty reduction policies and programs to those most in need, and thus help Vietnam achieve its poverty reduction goals. 38 Capturing Multiple Dimensions of Poverty 2.7 Measuring poverty is a challenging and complicated task, because poverty itself is complex and has many dimensions. This chapter focuses primarily on conventional approaches, based on absolute poverty lines and consumption measures of welfare. While familiar to the public and policy makers in Vietnam, the standard methodology may not fully capture other important dimensions of well-being. For example, households living in large, prosperous cities like Hanoi or Ho Chi Minh City may have access to better-quality schools and health facilities than households in other regions. But students attending higher-quality schools do not necessarily face higher school fees; in fact, households living in areas with poor schools may have to pay more, for instance, on extra tutoring to compensate for quality differences. Poor households that live in areas with low-quality schools but cannot afford to pay more may be at an additional disadvantage not captured in standard poverty analysis. Similarly, two households that look the same in terms of schooling and skills endowments may not earn the same income if one of the households faces discrimination in hiring—due to ethnicity or gender—that limits future prospects. 2.8 A variety of economic and social factors—some subtle and difï¬?cult to capture in standard poverty analysis—must be examined to get a full picture of poverty. Conventional poverty measures provide an important starting point for analyzing other dimensions of poverty. For example, the proï¬?le of poverty presented in Chapter 3 looks explicitly at other dimensions of poverty, for example, deprivations in education and skills, poor health status, and deprivations in access to basic services such as clean water and sanitation. The aim of multitopic surveys of living conditions (like the VHLSS) is to facilitate the measurement and analysis of poverty in multiple dimensions. The Human Development Index (HDI) described in Chapter 1 is a composite measure of well-being, as is the new Child Poverty Index (used in Chapter 3) and the broader Multidimensional Poverty Index (MPI) proposed by several UN organizations. 2.9 Additional information on other dimensions of deprivation experienced by the poor can be identiï¬?ed by soliciting their perceptions and insights through discussions and open-ended interviews. A number of Participatory Poverty Assessments (PPAs) have been carried out over the years in Vietnam, including three new ï¬?eld studies carried out in preparation for this report (see Chapter 1). Findings from qualitative studies are included throughout the report. These studies let the poor themselves give voice and context to the story that emerges from more conventional statistical analyses—poor men and women in Vietnam highlight concerns about lack of skills and education, access to good jobs and stable employment, and access to land and job security. They also speak about poverty in terms of risks—linked to health shocks, aging, and disability; job loss and uncertain wages; and weather shocks that destroy crops and affect rural incomes. Many of the poor are highly indebted, and risk can undermine new economic initiatives. The importance of social identity is also evident; in rural areas, minority status was often equated with being poor. C. Updating Methods for Measuring Poverty 2.10 Two important decisions are required in order to measure poverty: (a) how to measure welfare—in income or expenditure terms, and (b) what poverty threshold or line to use. Both issues have been the subject of debate in Vietnam, among both local researchers and policy makers and in the international community (box 2.2). 2.11 The GSO-WB approach uses per capita expenditures from the VHLSS as a measure of household welfare. The poverty line is constructed using a standard Cost of Basic Needs (CBN) approach, based on the observed consumption behavior of the poor, as reported in the VHLSS. It includes an allowance for food and nonfood spending. The food allowance (or food poverty line) is based on a single reference food basket for poor households, scaled up or down as needed to meet caloric norms and priced using a vector of national food prices. An additional allowance is added for essential nonfood spending, for example, on fuel, housing, schooling, health care, and clothing based on nonfood spending of households whose food spending is equal to the food poverty line (World Bank 1999). 39 Box 2.2 How is Poverty Measured? The poverty rate (or headcount index) is deï¬?ned as the proportion of the population in a speciï¬?c period whose welfare (consumption per capita) falls below the poverty line (ï¬?gure B2.2.1). Figure B2.2.1 Conventional Poverty Measurement Methodology Choice of Welfare Indicator Welfare is typically measured in terms of per capita consumer expenditures or per capita incomes. On a conceptual level, income is a measure of welfare opportunity—the level of well-being a household can afford to purchase at a particular point in time. Consumption can be thought of as a measure of welfare achievement—the level of well-being that a household actually achieves at a point in time. However, incomes are often more variable than expenditures: for example, farmers produce more in years when the weather is good than in years with unseasonable temperatures, droughts, and flooding. Households smooth income variations by saving in good years and dis- saving in bad years. Annual expenditures typically reflect a longer-run concept of income—that is, permanent income—rather than a shorter-run concept of annual income. It is therefore not surprising that income-based poverty statistics can be very different from consumption-based statistics. In the United States, for example, 30 percent of the income-poor own their own home compared to only 15 percent of the consumption-poor, and the food share for the income-poor is only 24 percent compared to 32 percent of the consumption-poor. It is generally assumed that poor households are less likely to own their own home (at least in high-income countries like the United States) and, according to Engel’s law, will spend a higher proportion of expenditures on food. Deï¬?ning the Poverty Line The most commonly used approach to setting poverty lines is the Cost of Basic Needs approach, which is widely applied in countries throughout the world and described in Ravallion (1994, 1998) and Ravallion and Bidani (1994). The Cost of Basic Needs approach consists of ï¬?rst deï¬?ning a basket of food and nonfood items that are adequate for satisfying basic consumption needs of a household, and then calculating the cost of this basket. Conceptually, a Cost of Basic Needs poverty line measures the minimum income necessary for households to purchase a basic needs basket of food and other commodities, so that members have sufï¬?cient food to remain healthy and productive and have the means to participate fully in society. In practical terms, the poverty line is constructed by ï¬?rst deï¬?ning a reference food basket, reflecting consumption patterns of the poor; and anchoring it in an agreed nutrition norm (for example, 2100 calories per person per day), and then adding an allowance for nonfood spending on essential goods (health care, education, housing, and durables) that is consistent with spending patterns of the poor. 40 2.12 Vietnam carried out two Living Standards Surveys in the 1990s—the 1992–93 VLSS and the 1997–98 VLSS—with extensive technical support from international partners. Vietnam then carried out a series of government-ï¬?nanced Vietnam Household Living Standards Surveys (VHLSS) (in 2002, 2004, 2006, and 2008) using a similar approach to the earlier VLSS. The design of the core expenditure and income modules of the VHLSS questionnaires were kept broadly consistent with similar modules of the VLSS modules, with the speciï¬?c and laudable aim of maintaining comparability over time. As noted, comparability has been one of the great strengths of Vietnam’s poverty data. 2.13 But by 2010, strict comparability was coming at too high a cost. The 2010 VHLSS and related welfare aggregates represent a break with the 2002–2008 VHLSS series in three important respects: (a) the 2010 VHLSS was based on a new master sample based on the 2009 Housing and Population Census, including a new set of communes and enumeration areas; (b) the VHLSS household questionnaire was substantially revised (including revisions to the core consumption module) and reduced in length; and (c) an updated methodology was used to construct a more comprehensive consumption (welfare) aggregate. These improvements are summarized here and described in greater detail in Kozel, Hinsdale, and Nguyen (2013). The VHLSS was Improved and Shortened in 2010 2.14 Sampling. The 2002–08 rounds of the VHLSS used a master sample of communes/urban wards drawn from the 1999 Housing and Population Census. In each round of the VHLSS, half of the enumeration areas (villages) and households within the communes were kept and half replaced, with the aim of ensuring stability in poverty measurement. While good for measurement stability, the 2002–08 master sample was substantially outdated by the end of the period. For example, between 2002 and 2008, there was substantial residential development in erstwhile empty areas (for example, “New Cityâ€? on the outskirts of Hanoi), and residential growth in provincial cities and towns, but these new developments were not included in the master sample used for 2002–08 rounds of the VHLSS. 2.15 A new master sample of communes and wards was developed for the 2010 and subsequent VHLSSs based on the 15 percent sample of the 2009 Housing and Population Census. Analysis suggests that the new sample provides better coverage of smaller households in urban areas, and somewhat better coverage of migrant households, many of whom come to work in urban areas for extended periods. Previous rounds of the VHLSS have been criticized for poor coverage of urban migrants, who are often assumed to belong to rural sending households (Pincus and Sender 2008). A recent study of poverty in Hanoi and Ho Chi Minh City (Haughton et al. 2010) indicates that many unregistered short-term urban migrants—who are likely to be undersampled in the VHLSS—may be vulnerable and have lower living standards than longer-term residents. These issues will be explored more systematically in the future; the 2012 VHLSS includes a special module on migrants, focusing in particular on long- and short-term migration for work purposes. 2.16 The sample of households for the 2012 VHLSS will be drawn from the same communes as the 2010 VHLSS, similar to the design of the 2002–08 sample. For 2014 and subsequent years, GSO is strongly advised to (a) update the master sample through careful relisting of enumeration areas on a regular basis, and (b) add new communes to the VHLSS master sample over time, with particular attention to good coverage in peri-urban areas where new population growth is occurring. GSO is also encouraged to explore alternative approaches to improve coverage of urban migrants, through either a more comprehensive sampling methodology or in-depth surveys of migrant populations. 2.17 Questionnaire Design. The VHLSS has been criticized by some researchers for taking too long to administer in the ï¬?eld, with related concerns about data quality and accuracy. In response to these criticisms, many sections of the 2010 questionnaire were shortened. The consumption modules were redesigned to collect information on food and frequent nonfood spending using a ï¬?xed reference period (30 days) rather than a “typical monthâ€? (used in 2002–2008), and a decision was made to administer the VHLSS in four rounds during each survey year.12 Questions designed 12 The decision to move to a ï¬?xed reference period was triggered by difï¬?culties in measuring expenditures and prices during bouts of high inflation (for example, 2008), and an effort to better capture seasonality in consumption patterns. 41 to collect information on labor earnings also used a ï¬?xed reference period (prior month) rather than being based on “typicalâ€? work activities. Additional questions were added to capture Vietnam’s expanding array of social insurance and social assistance programs, and were better measures of remittances and transfers. Improvements were also made to the module on access to poverty programs, including targeting and coverage of beneï¬?ts from targeted poverty reduction programs such as the National Target Program for Sustainable Poverty Reduction. New, more Comprehensive Consumption Aggregates were constructed 2.18 The ï¬?rst step in estimating a poverty line is to construct a welfare aggregate. The consumption aggregates constructed from the VHLSS follow standard practices well established in the literature (Deaton 1997; Deaton and Zaidi 2002). The consumption aggregates includes (a) food consumption, (b) frequent and infrequent nonfood items (personal care and hygiene, clothing, fuel, household goods), (c) education (tuition, books and uniforms, tutoring, and other fees), (d) health (curative and preventive care, health insurance), and (e) utilities (water, electricity, sanitation and trash collection). Two standard imputations are made in constructing the consumption aggregates, (a) the annual flow of services from durables, and (b) the annual value of housing services/imputed rents. 2.19 The poverty line is deï¬?ned on the basis of the welfare aggregate. Any changes in the deï¬?nition of the welfare aggregate will thus require revisions to the poverty line. Different countries use different welfare aggregates for measuring poverty; some countries use income, others use household expenditures. Within countries using household expenditures, there are substantial differences in expenditure aggregates. For example, although many countries include health or education expenditures in the expenditure aggregate, an increasing number of low-income countries in Sub- Saharan Africa do not. If basic health services and primary education services are provided free of charge, they are not captured in household expenditures, however deï¬?ned, unless imputations are made to value the flow of publically provided services. Instead of trying to value these—which is complicated and controversial—additional analysis can be carried out to measure deprivations in human development, as a complement to income- or expenditure-based measures of deprivation. Many countries, particularly as they become more affluent, include the (imputed) value of durables, housing services, and local amenities in the expenditure aggregate. While broad concepts may be similar—welfare is measured through a household-level expenditure aggregate—the great diversity in actual practice makes it difï¬?cult to compare national poverty lines and poverty rates across countries, even when converted into “internationallyâ€? comparable 2005 Purchasing Power Parity (PPP) measures. One reason India’s national poverty line is low in PPP terms is because it is based on a very parsimonious welfare aggregate (box 2.1). 2.20 Two different sets of consumption aggregates have been used for poverty analysis in Vietnam. One set of aggregates (referred to as “temporally comparableâ€?) was designed, as the name suggests, to be strictly comparable with the consumption aggregates initially developed using the 1992–93 VLSS. For example, although new durable goods were added to later rounds of the VHLSS (for example, cell phones, computers), only items available in the 1992–93 VLSS are included in the comparable aggregate. Similarly, estimates of the value of housing services are also based on spending patterns in the 1992–93 VLSS. Because Vietnam’s housing market was very underdeveloped in the 1990s, imputed rents were calculated as a ï¬?xed percentage of total nonfood consumption rather than derived using conventional hedonic methods. This same ï¬?xed percentage (from 1993) was used to calculate the housing component of the consumption aggregate in all subsequent rounds of the VHLSS through 2008. 2.21 The vast majority of research and analytic work using VHLSS data has used the comparable consumption aggregate. The original GSO-WB poverty line, used extensively in the poverty literature for Vietnam, was constructed using the comparable aggregate, and is based on a reference food basket from the 1992–93 VLSS and related spending on a minimum basket of nonfood items, also based on spending patterns of the poor as reported in the 1992–93 VLSS. 42 2.22 Vietnam today is different from Vietnam in the 1990s, and expenditures, including expenditures of low-income households, are far more diversiï¬?ed. Real estate markets are more developed, particularly in urban areas, and many households put considerable investment into housing and land. Vietnam is similar to other fast-growing economies in this respect. Housing values reported in recent rounds of the VHLSS are more reliable than those collected in earlier rounds. 2.23 A second set of “comprehensiveâ€? consumption aggregates was created for the 2004, 2006, 2008, and 2010 rounds of the VHLSS, which aimed to make optimal use of all the expenditure information in a given year, unencumbered by considerations of strict comparability over time. There are a number of minor and major differences between comparable and comprehensive aggregates (see Annex 2.1 for a detailed description). The comprehensive aggregate includes the imputed value for all durables owned by the household and an imputed flow of services from housing. The latter is a particularly important addition (box 2.3). Box 2.3 How to Value Housing Services in the VHLSS Housing is an important component of household welfare, particularly as countries grow and prosper. Investments in housing are rising rapidly in Vietnam—families purchase new houses, and build or add onto existing dwelling units. Housing expenditures—either actual or imputed— should be fully reflected in the consumption aggregate. In countries where housing markets function well, annual rental payments provide a good measure of the value of housing services. Using information on reported rents, a hedonic for housing can be used to impute the value of housing services (based on characteristics of the dwelling unit and neighborhood characteristics) in cases where information on rents is missing (for example, owner-occupied housing, housing supplied by employers). However, Vietnam is an unusual case; rental markets are still thin and there are not enough renters either in early or more recent rounds of the VHLSS to estimate robust hedonic rent equations. Even the 2010 VHLSS includes only 243 households (out of 9,399) who report spending on rents—around 2.6 percent of total households in the sample. In contrast, the 2009 Housing and Population Census reports that 6.4 percent of all households in Vietnam rent their dwelling unit, including 13.2 percent of households living in urban areas. Prior to 2010, the value of housing services was assumed to be a ï¬?xed percentage of nonfood consumption expenditures. Based on shares in 1992–93, the value of housing was set equal to 11.8 percent of nonfood consumption for rural households and 21.4 percent for urban households. In constructing comprehensive aggregates, each household’s annual consumption of housing services is calculated as a ï¬?xed share of the reported sales value of the dwelling unit. This ï¬?xed share is the same for all households and equals 2.88 percent, which is the median ratio of reported annual rent payments to reported dwelling sales value, among the subsample of households who report renting their dwelling. In essence, this method uses the information collected in the 2010 VHLSS about Vietnam’s rental market to approximate the relationship that prevails in Vietnam between rental and ownership values in housing, and then imputes annual consumption of housing services for all households using this relationship. While this method would not be preferable to hedonic estimation given a more comprehensive survey of Vietnam’s renters, it has the virtue of not assuming that a household’s consumption of housing remains a constant proportion of other nonfood consumption over time, an assumption made in the temporally comparable set of aggregates from 1993 to 2008. Derived directly from the reported value of each household’s dwelling, the measure of housing consumption in the comprehensive aggregates is more sensitive to what each household reports about its living situation. The result is that, in 2010, housing averaged 15 percent of total consumption in the comprehensive aggregates compared to 6 percent in the temporally comparable aggregates (table 2.1). Note, however, that the share of housing is much lower for households in the poorest quintile (7.5 percent) and thus does not have a large impact on 2010 poverty rates. Source: Kozel, Hinsdale, and Nguyen 2013. 43 2.24 Tables 2.1 and 2.2 present comparable and comprehensive consumption aggregates for the last four rounds of the VHLSS.12 By 2010, it was clear that the beneï¬?ts of maintaining procedural consistency with 1993 consumption aggregates was substantially outweighed by the resulting loss of information; there is a large and growing gap between the temporally comparable and comprehensive aggregates over time. Going forward, it is recommended that the methodology for estimating consumption aggregates and poverty lines be updated on a more frequent basis. How frequently will depend on Vietnam’s rate of economic progress and how quickly consumption patterns are changing, particularly changes at the lower end of the income distribution, where there is a trade-off between stability and consistency over time and relevance of the methodology to contemporary living conditions. Given how quickly conditions are changing globally and in Vietnam, it is suggested that the methodology be revisited in ï¬?ve (or six) years to assess whether it is providing accurate estimates. Note, however, that despite efforts to ensure procedural consistency, comparisons between the 2010 VHLSS and earlier years using either comparable or comprehensive consumption aggregates must be interpreted with care. As described above, a number of important changes were introduced in the 2010 VHLSS, such as an updated sample frame, a shift to a ï¬?xed reference period in the expenditure module, and a revised deï¬?nition of welfare, which make comparisons difï¬?cult. The 2010 VHLSS and the new GSO-WB poverty lines provide a baseline for consistent poverty monitoring going forward, that is, for the 2012 and future rounds of the VHLSS. Table 2.1 Comprehensive Consumption Aggregates for the VHLSS 2004, 2006, 2008, 2010 Mean consumption Average share of total consumption Expenditure component 2004 2006 2008 2010 2004 2006 2008 2010 Food expenditure 1,753 2,378 2,993 6,515 42 42 38 46 NonͲfood expenditure 1,050 1,449 2,142 3,220 21 21 22 20 Durables consumption 592 767 1,301 1,972 10 10 12 10 Education expenditure 261 334 461 769 5 5 5 4 Health expenditure 297 339 494 722 6 5 5 4 Utilities and electricity 140 183 233 373 3 3 2 2 Housing consumption 1,120 1,390 2,070 3,558 15 15 16 15 Total expenditure 5,212 6,840 9,694 17,129 100 100 100 100 Source: 2004, 2006, 2008, 2010 VHLSS. Table 2.2 Temporally Comparable Consumption Aggregates for VHLSS 2004, 2006, 2008, 2010 Source: 2004, 2006, 2008, 2010 VHLSS. Mean consumption Average share of total consumption Expenditure component 2004 2006 2008 2010 2004 2006 2008 2010 Food expenditure 1,857 2,502 3,153 6,401 49 49 47 54 NonͲfood expenditure 986 1,396 1,987 2,975 20 21 23 21 Durables consumption 518 638 801 1,268 10 9 9 7 Education expenditure 246 330 423 732 5 5 5 5 Health expenditure 290 332 465 680 6 5 6 5 Utilities and electricity 147 191 233 378 3 3 3 3 Housing consumption 351 466 622 988 6 6 7 6 Total expenditure 4,394 5,855 7,683 13,422 100 100 100 100 2.25 Figure 2.1 shows the overall composition of per capita expenditures in the 2010 VHLSS. Spending on food now constitutes less than half of per capita expenditures compared to 57 percent in 1998, and durables and housing make up nearly a quarter of aggregate welfare. 13 These aggregates are in real terms; they have been adjusted to January terms of the survey year and for regional cost- of-living differences. 44 Figure 2.1 Composition of Per Capita Expenditures, 2010 VHLSS Housing 14% Food 46% Other non-food 20% Utilities and electricity 2% Education Durables 4% 10% Health 4% 2.26 Figure 2.2 shows the composition of expenditures, categorized by food, nonfood, durables, housing, and others—broken down by per capita expenditure quintile. Note that the food share falls from 58 percent (in the poorest quintile) to only 32 percent for the wealthiest quintile. In contrast, the poorest individuals spend only 7 percent of their total expenditures on housing and another 7 percent on durables compared to a housing share of 27 percent and a durable share of 12 percent for the wealthiest group of individuals. These gradients are consistent with those of other countries at similar levels of development. Figure 2.2 Composition of Per Capita Expenditures by Per Capita Expenditure Quintile, 2010 VHLSS 100% 7.5 9.9 12.4 15.8 90% 27.1 19.8 80% 20.7 20.3 19.9 70% Housing 17.9 60% 6.6 Other non-food 8.4 50% 9.8 Utilities and electriccity 10.9 40% Education 12.2 Health 30% 58.3 50.3 45.7 41.5 Durables 20% 31.9 Food 10% 0% 1 2 3 4 5 45 Consumption is adjusted for Household Size to Estimate Individual Welfare 2.27 Our objective is to calculate a measure of individual welfare and estimate the number of people who live below the poverty line. But in households, individuals live together, eat together, and often pool their resources. Household surveys like the VHLSS measure expenditures at the household rather than individual level. Different approaches have been used to move from household-level expenditures to individual welfare. One approach is to use equivalence scales and to also adjust for household-level economies of scale. In the absence of a well-deï¬?ned equivalence scale for Vietnam, and building on past practices, household expenditure is converted into per capita terms by simply dividing by household size. The implications of using alternative measures, adjusting for adult equivalencies and household economies of scale, on the poverty proï¬?le are discussed briefly in Chapter 3. Consumption is also Adjusted for Temporal and Spatial Cost Variations 2.28 One of the advantages of the CBN methodology is that it anchors the poverty line at a ï¬?xed level of well-being, and consequently allows for consistent poverty comparisons. However, households living in different regions of the country may face different prices for similar consumer goods due to differences in transport, storage, and marketing costs. For example, consumers pay more per kilogram to purchase rice in a market in Ho Chi Minh City than they pay to purchase the same quality of rice in a rural district in the Mekong Delta, where the rice is grown. In contrast, laundry soap may cost more in rural areas than in cities, where it is produced and packaged. Prices also change over time due to inflation and other factors. 2.29 Some countries (for example, Indonesia and Mozambique) account for inflation and spatial cost-of-living differences by constructing poverty lines for different regions, based on region-speciï¬?c prices and (sometimes) region-speciï¬?c consumption baskets. In keeping with past practice, a single national GSO-WB poverty line was constructed using information from the 2010 VHLSS. The new GSO-WB poverty line is applied to spatially and temporally adjusted (that is, real) per capita expenditures to calculate poverty rates. 2.30 Temporal adjustments are straightforward; the consumption aggregates described in table 2.1 have been deflated to January of each survey year (for example, 2004, 2006, 2008, 2010) using the GSO’s ofï¬?cial Consumer Price Index (CPI) deflators for rice, other foods, and nonfoods. Previous to 2010, spatial adjustments were made using regional CPI deflators provided by the GSO. For 2010, new spatial cost-of-living indexes (SCOLIs) were estimated and are used instead of regional CPI deflators to calculate poverty rates. 2.31 There are three reasons why prices collected for the CPI are not well-suited to measuring spatial differences in the cost of living. First, CPI prices are collected on a frequent basis in outlets where a wide range of consumer goods are available and shopping volumes are high. These are typically located in urban and peri-urban areas. But many of the rural population (including poor households) shop in local markets near where they live. Second, the speciï¬?cation of items whose prices are collected for the CPI is not the same across provinces. Vietnam’s CPI price collection system maintains temporal consistency (prices for the same items are collected over time in each location) but not spatial consistency (the items in the basket may be slightly different in each location). For example, prices of higher-end cotton shirts may be surveyed in large urban areas, while prices for lower-cost polyester shirts are surveyed in smaller towns or rural areas. Regional variations in the speciï¬?cation of items may reflect quality differences rather than only capturing price differences for an identical good. Third, a CPI and SCOLI have different objectives, and the differences make it difï¬?cult for the two indexes to rely on the same set of price data. The CPI aims to give equal weight to every Vietnamese dong spent; it is used as a deflator to ensure the real value of currency remains unchanged. Consequently the expenditure patterns of wealthier households have more weight in a CPI because they spend more money, and the CPI price collection system targets outlets with a high volume of purchases. 46 2.32 In contrast, a SCOLI is population-weighted rather VND-weighted; the SCOLI is estimated using the prices paid by the average individual from each area, and prices are aggregated into a population-weighted index that treats everyone equally. In short, compared to the CPI, a SCOLI requires different budget shares for aggregating items into an index, a different set of outlets for price collection, and different weights to aggregate information on individuals to form regional averages. 2.33 Regional adjustments were based on regional CPI indexes in earlier rounds of the VHLSS. However, for 2010, adjustments were made for regional cost-of-living differences using market price data from a SCOLI ï¬?elded in conjunction with the second and third rounds of the 2010 VHLSS. The approach is described in Annex 2.2. 2.34 The 2010 SCOLI ranges between 0.7 and 1.0 (table 2.3). The Mekong Delta has the lowest overall cost of living and the Red River Delta (which is also the base region) has the highest cost of living. In all but two of the six regions, the SCOLI shows only a small difference in the cost of living between urban and rural sectors. The two exceptions are the Red River and South East regions, where the urban cost of living is approximately 20 percent higher than the rural cost of living, largely reflecting the higher estimated cost of accommodation services in the metropolitan areas of Hanoi and Ho Chi Minh City. Apart from these two exceptions, the variation in the cost of living is greater across regions than it is between the urban and rural sectors within a region. Table 2.3 Spatial Cost-of-Living Index (SCOLI) for each Region and Sector Region Urban Households Rural Households Red River 1.00 0.79 Midlands & Northern Mountains 0.81 0.79 Northern & Central Coast 0.78 0.71 Central Highlands 0.83 0.78 South East 0.97 0.77 Mekong Delta 0.74 0.70 Note: Calculations are based on a Törnqvist index applied to regional average prices that are pooled over the two rounds of SCOLI data collection, and using person-weighted average budget shares, with housing values based on the hypothetical values reported by all survey respondents. D. Constructing a new GSO-WB Poverty Line 2.35 The poverty line consists of two components, a food poverty line and an additional allocation to account for essential nonfood needs. The food poverty line is estimated in three steps. First, a reference food basket is deï¬?ned that reflects the consumption patterns of the poor; second, quantities are adjusted to reach an agreed nutrition norm; and third, the cost of purchasing the adjusted reference basket is calculated. An allowance for essential nonfood needs is estimated using an Engel’s curve regression and is then added to the food poverty line in order to construct the total poverty line. Deï¬?ning the Reference Food Basket 2.36 The reference food basket used to construct the original GSO-WB poverty line is anchored in the food consumption patterns of poor households14 in the 1993 VLSS. The reference food basket for the updated GSO-WB poverty line is anchored in food consumption patterns of poor households in the 2010 VHLSS. 14 The methodology is described in Annex 2 of the 2000 “Vietnam Development Report: Attacking Poverty.â€? (World Bank 1999). Food consumption of the 3rd quintile of households, ranked nationally based on per capita expenditures, was used to construct the reference food basket. 47 2.37 Deï¬?ning the reference basket is an iterative process; we do not know in advance which households are poor (the method is described in Pradhan et al. 2001)15. Households were ranked according to SCOLI-adjusted and temporally adjusted per capita expenditures (henceforth referred to as “realâ€? per capita expenditures) from least well-off to most well-off, and the poor were initially deï¬?ned as those in the bottom 2.5 percent to 20 percent of the real per capita expenditure distribution. This initial reference basket ultimately became the ï¬?nal reference basket; the 2010 poverty rate, based on an updated GSO-WB poverty line, was close to 20 percent. 2.38 Analyses were carried out to assess the stability of the poverty line food basket across different reference groups; food consumption patterns of the bottom 2.5 to 20 percent (bottom quintile) of individuals were compared with the bottom 2.5 to 10 percent (bottom decile). The initial 2.5 to 20 percent reference group was further divided to compare (a) food baskets for bottom-quintile ethnic minorities and bottom-quintile majorities, and (b) food baskets for bottom-quintile urban and bottom- quintile rural households (Annex table 2.1). 2.39 Food consumption patterns were similar when comparing the poorest 10 percent and the poorest 20 percent of the population. Similarly, the consumption patterns of poor minority households were on average quite similar to consumption patterns of poor majority households. Dietary patterns, however, were different for urban and rural households in the 2.5 to 20 percent reference group: urban poor households consumed less rice and higher-priced calories (meats, oils), and were more likely to consume food and drinks outside the home. Although the GSO-WB poverty line is based on a single national reference basket for poor households, Vietnam’s ofï¬?cial poverty lines use different reference baskets for urban and rural households. A number of other countries, including, for example, Indonesia, Mozambique, Papua New Guinea, and Russia, deï¬?ne regional reference baskets that reflect local preferences and tastes. The problem with using different reference baskets, particularly for urban and rural areas, is that the different baskets often reflect diets of different quality, so the poverty line for urban areas (based on consumption patterns of urban households) may give a superior standard of living compared to the poverty line for rural areas (based on consumption patterns of rural households). In 2010, only a small fraction (9 percent) of the poor reference group actually lived in urban areas. Given this, coupled with concerns about avoiding quality differences (that is, setting a higher standard of living for urban households), a single national reference food basket was again used to construct the new GSO-WB poverty line. 2.40 In line with standard CBN practice, food quantities in the reference basket are scaled up to an “acceptableâ€? nutritional norm, holding constant the relative composition of the reference basket (that is, all quantities are scaled up by the same factor). But what constitutes an acceptable norm? International experience shows that countries anchor their poverty lines in very different caloric norms, ranging from a low of 1,800 Kcals for India (GOI 2009) to more than 2,700 Kcals for some countries in Africa. 2.41 The original GSO-WB poverty line was anchored in a caloric norm of 2,100 Kcals per person per day. However, the composition of the Vietnamese population has changed since the early 1990s, when the 2,100 Kcals norm was set. The share of young children in the population (who consume less food) has decreased and the share of adults (who consume more) has increased. A new caloric norm of 2,230 Kcals per person per day was estimated using age- and gender-speciï¬?c caloric requirements for the Vietnamese population developed by the Nutrition Institute in the Ministry of Health (MOH 2006), and weighted by the relevant age-gender composition of the national population in the 2010 VHLSS. These new norms compare well with international practice (ï¬?gure 2.3). 15 We restrict the group to the bottom 2.5 percent to 20 percent to avoid potential problems with outliers and measurement error. 48 Figure 2.3 Nutrition Norms Used to Anchor Poverty Lines in Different Countries 3500 3000 2500 2000 1500 1000 500 0 Nicaragua Egypt Uganda India Ecudador Honduras Senegal Mexico Indonesia Bangladesh SriLanka Mozambique ElSalvador Guatemala Paraguay Vietnam Panama Iraq Malawi SierraLeone Cameroon Chile Jordan TimorLeste SouthSudan Colombia Sources: UN Statistics Division 2005; World Bank staff estimates. 2.42 Table 2.4 compares the calorie and expenditure composition of the 1993 reference food basket used to estimate the original GSO-WB poverty line with the new food basket use to construct the 2010 GSO-WB poverty line. The original food reference basket was heavily dominated by rice (79 percent of calories, 46 percent of food spending). The 2010 basket is more diversiï¬?ed; although rice continues to be important in the food consumption of the poor (66 percent of calories, 30 percent of food spending), their consumption patterns have become more diversiï¬?ed to include, for instance, pork and other meats and seafood, vegetables and fruits, more oils, and more calories from meals eaten outside the household. Rice calories are very cheap; calories from pork, oils, and seafood are more expensive. The cost of the 2010 reference basket will be higher than the original 1993 reference basket. In addition, there has been a substantial increase in the non-quantiï¬?ed share of consumption, that is, food reported under “otherâ€? categories and meals eaten outside the household. More than 95 percent of food consumption was recorded under quantiï¬?ed items in the 1998 VLSS compared to less than 80 percent in the 2010 VHLSS. An extended list of food items was included in the 2012 VHLSS, with the aim of getting better (more quantiï¬?ed) measures of food consumption (table 2.4). 49 Table 2.4 Composition of the Reference Food Basket, 1993 and 2010 VHLSS 1993 2010 Average Average Average Average share of share of total share of share of total total food total food Food item calories expenditure calories expenditure Plain rice (including fragrant and specialty rice) 78.9 46.5 66.4 30.5 Sticky rice 2.7 2.3 4.2 2.5 Maize (in seed equivalent) 1.0 0.4 1.6 0.4 Cassava (in freshͲtype equivalent) 1.9 0.9 1.0 0.3 Potato of various kinds (in freshͲtype equivalent) 1.6 2.5 0.3 0.3 Wheat grains, bread, wheat powder 0.3 0.4 0.3 0.3 Flour noodle, instant rice noodle/porridge 0.3 0.7 1.3 1.6 Fresh rice noodle, dried rice noodle 0.4 0.5 Vermicelli 0.1 0.2 Pork (in equivalent of the pork type with removed fat) 2.4 9.3 4.0 11.1 Beef 0.1 0.8 Buffalo meat 0.0 0.5 0.0 0.2 Chicken meat 0.7 5.1 0.9 5.1 Duck and other poultry meat 0.1 0.7 0.2 1.0 Other types of meat 0.0 0.3 Processed meat 0.1 0.6 Lard, cooking oil 1.8 1.5 4.2 2.5 Fresh shrimp, fish 1.3 8.3 1.4 6.9 Dried and processed shrimps, fish 0.3 1.2 Other aquatic products and seafood (crabs, snails,...) 0.1 0.5 Eggs of chickens, ducks, Muscovy ducks, geese 0.0 0.3 0.7 1.7 Tofu 0.4 0.9 0.6 1.3 Peanuts, sesame 0.7 0.8 0.5 0.4 Beans of various kinds 0.4 0.6 0.3 0.3 Fresh peas of various kinds 0.1 0.4 Morning glory vegetables 0.6 2.2 0.5 1.1 Kohlrabi 0.3 1.0 0.1 0.2 Cabbage 0.2 1.0 0.1 0.4 Tomato 0.1 0.7 0.0 0.4 Other vegetables 0.7 3.3 Orange 0.0 0.2 0.0 0.2 Banana 0.7 1.2 0.6 0.6 Mango 0.0 0.3 0.0 0.2 Other fruits 0.4 1.5 Fish sauce 0.3 2.0 0.2 1.1 Salt 0.0 0.5 0.0 0.3 MSG 0.0 0.8 0.0 0.3 Glutamate 0.0 1.3 Sugar, molasses 1.3 1.3 1.3 1.2 Confectionery 0.6 1.0 Condensed milk, milk powder 0.0 0.1 0.2 0.7 Ice cream, yoghurt 0.0 0.2 Fresh milk 0.1 0.5 Alcohol of various kinds 1.3 1.8 Beer of various kinds 0.8 0.9 0.1 0.3 Bottled, canned, boxed beverages 0.1 0.2 Instant coffee 0.0 0.2 Coffee powder 0.0 0.1 Instant tea powder 0.0 0.1 Other dried tea 1.0 6.3 0.4 1.1 Tobacco 0.0 2.3 Betel leaves, areca nuts, lime, betel pieces 0.0 0.1 Outdoors meals and drinks 3.3 5.9 Other food and drinks 1.0 2.6 50 Calculating the Food Poverty Line 2.43 The food portion of the CBN poverty lines is deï¬?ned as the cost of purchasing the (scaled) reference food basket. There are three sources for food prices that could be used to estimate the food portion of the poverty line: (a) unit values (reported value of food consumption divided by reported quantities) calculated from the 2010 VHLSS survey, (b) food prices collected by the GSO Price Department for the CPI, and (c) food prices collected through the SCOLI survey. 2.44 The original GSO-WB food poverty line was based on CPI food prices provided by the Price Department. However, Vietnam’s new ofï¬?cial poverty lines are calculated using unit values from the 2006 VHLSS and adjusted for inflation. Both the SCOLI and CPI prices cover only a subset of food items in the 2010 VHLSS. Unit values (real or imputed in the case of non-quantiï¬?ed consumption) are available for all food items in the VHLSS and, moreover, can be estimated speciï¬?cally for low-income households, thus reflecting what the poor actually purchase (quality, brand) and what they pay. There are mixed views in the literature (Deaton 1988, 1997; Deaton and Tarozzi 2005) about whether unit values are adequately well speciï¬?ed to be used as prices. Even well-deï¬?ned items in the household consumption module, such as rice, are available in a range of qualities, and prices vary between urban and rural areas and among regions. Limiting unit values to a group of poor households will help control for quality differences, which are usually linked to income levels (for example, wealthier households tend to purchase higher-quality/more expensive rice). 2.45 Consistent with the methodology used to estimate Vietnam’s ofï¬?cial poverty lines, the new GSO-WB food poverty line is calculated using mean unit values for food purchases by poorer households (bottom 2.5 to 20 percent) reported in the 2010 VHLSS. National food poverty lines are estimated for each round of the 2010 VHLSS (June, October, December) using the national reference food basket and food prices (unit values) from each round, and adjusted for inflation and averaged to construct a national food poverty line in January 2010 VND. 2.46 The new GSO-WB food poverty line for 2010 is VND 343,000 per person per month (VND 4,116,000 per person per year). Calculating the Total Poverty Line, including Food and Essential Nonfood Spending 2.47 In addition to food, an allowance must be added for essential nonfood spending such as for fuel, housing, schooling, health care, clothing, and other daily needs. However, estimating the nonfood component of the poverty line is not as straightforward as estimating the food poverty line, because there is no easily deï¬?ned “normâ€? for nonfood expenditures in the way that caloric norms can be used to deï¬?ne food needs. 2.48 The CBN approach looks to the actual expenditure patterns of the poor in the 2010 VHLSS with the aim of estimating (a) an “austereâ€? allowance for nonfood needs, based on the typical value of nonfood spending by households whose total expenditure just equals the cost of the food poverty line; and (b) “minimal but adequateâ€? allowance for nonfood needs, based on the typical value of nonfood spending by households whose food spending actually reaches the cost of the food poverty line, so that basic food needs are fully met. 2.49 An Engel curve looks at the relationship between the share of spending on food and total per capita expenditures. According to Engel’s law, the food share decreases as expenditures (welfare) rise. The average food share for each group of households can be calculated using an Engel curve regression (Ravallion and Bidani 1994) as follows: ݂ሺ‫ݕ‬௜ ሻ ‫ݕ‬௜ ൌ ß™ ൅  ߚଵ Ž‘‰ ቀ ௙ á‰? ൅  ß› ᇱ ൫݀௧ െ  Ý€Ò§ ൯ ൅  ‫݈ܽݑ݀݅Ý?Ý?ݎ‬௜ ‫ݕ‬௜ ܾ ௙ሺ௬೔ ሻ à³” ௬ where  is the food budget share, α is a national intercept, ቀ௕೑ á‰? is total (nominal) expenditure ௬೔ divided by the food poverty line, and dt is a vector of demographics with mean d. 2.50 In keeping with international practice, we propose to use the upper-bound poverty line (that is, with “minimal but adequateâ€? allowance for nonfood) as the new GSO-WB poverty line, which is thus 51 deï¬?ned as the food poverty line divided by Engel’s coefï¬?cient estimated from the regression (.525)16: ܾ௙ ß™ ‫כ‬ The new poverty line assumes the nonfood spending of a typical household at the point on the Engel curve where actual food expenditure is equal to the food poverty line. 2.51 The new GSO-WB poverty line is therefore deï¬?ned as: VND 653,000 per person/month, which equals VND 343,000 (food poverty line) /.525. E. New Poverty Estimates for 2010: GSO-WB and Ofï¬?cial Poverty Methodologies 2.52 New poverty estimates based on the new GSO-WB poverty lines and consumption aggregates described in this chapter are presented in table 2.5. For purposes of comparison, the table also presents Vietnam’s ofï¬?cial household-level poverty estimates for 2010,17 based on ofï¬?cial poverty lines of VND 400,000 person/month (rural) and VND 500,000 person/month (urban). The GSO-WB poverty rates are higher overall—20.7 percent compared to 14.2 percent—which is not surprising because the GSO-WB poverty line (VND 653,000 person/month) is higher than the ofï¬?cial poverty lines. Comparing the two estimates for 2010, ofï¬?cial estimates suggest higher rates of poverty in the North Central and South Central coastal regions compared to GSO-WB estimates, and slightly lower rates in the Central Highlands and Southeast region. Differences in poverty estimates for the Southeast primarily reflect the fact that the SCOLI measured a higher cost of living in the Southeast compared to the CPI-based regional deflator. Overall, the GSO-WB estimates suggest lower poverty rates in urban areas than ofï¬?cial estimates. Table 2.5 Poverty Estimates for 2010: Comparing the GSO-WB Methodology and Ofï¬?cial Methodology GSOͲWBPoverty Rate OfficialPovertyRate Contributionto Contribution to Incidence(%) total(%) Incidence (%) total(%) AllVietnam(national) 20.7 100 14.2 100 Urban 6.0 9 6.9 14 Rural 27.0 91 17.4 86 RedRiverDelta(Hanoi) 11.4 12 8.4 13 EastNorthernMountains 37.7 21 24.2 20 WestNorthernMountains 60.1 9 39.4 9 NorthCentralCoast 28.4 16 24.0 20 SouthCentralCoast 18.1 7 16.9 10 CentralHighlands 32.8 10 22.2 9 Southeast(HCMC) 8.6 7 3.4 4 Mekong Delta 18.7 17 12.6 17 16 Where α* is deï¬?ned as α*= α+ β1 log(1/α*). 17 Ofï¬?cial estimates reflect the number of households on the poverty list and not the number of individuals on the poverty list. To the extent that poor households are larger on average than nonpoor households, ofï¬?cial estimates of the share of individuals below the poverty line would be higher than the share of households. 18 Each round of the VHLSS includes around 46,000 households. Detailed information on household income is collected for all households, but consumption information is collected for only 20 percent of households (three in each enumeration area), or 9,400 households in total. Only unit record data from the 20 percent sample (income + consumption) are released to the public. 52 2.53 Although the methodologies are broadly similar (both use a CBN approach based on spending behavior of the poor in the VHLSS), the new GSO-WB poverty line is higher than ofï¬?cial lines for the following reasons: â—? Ofï¬?cial lines were ï¬?nalized in late 2010, before the 2010 VHLSS data were available and are thus based on a food reference basket and consumption behavior of poor households in the 2006 VHLSS. As noted, the 2010 VHLSS is different from the 2006 VHLSS in a number of important respects, including sampling and design of the questionnaire. â—? Ofï¬?cial poverty lines were estimated using the temporally comparable consumption aggregates rather than comprehensive consumption aggregates. As demonstrated in table 2.1, the comprehensive aggregate is higher due especially to the inclusion of more types of durable goods and, most importantly, a better measure of the value of housing services. But using the new measure of housing services does not in itself lead to a higher poverty rate. We tested a modiï¬?ed comprehensive consumption aggregate that included a value of housing calculated using the original GSO-WB method, and then calculated new poverty lines and poverty rates. The “old housing methodâ€? poverty rate was 21.3 percent, slightly higher than the “new housing methodâ€? poverty rate. â—? Although food poverty lines are similar in the ofï¬?cial and GSO-WB approaches, a decision was made to use a lower allocation for essential nonfood spending for the ofï¬?cial poverty lines than indicated in the VHLSS data (see discussion in Chapter 1). 2.54 There are other important differences between the two methodologies that might result in different poverty rates in the aggregate and across regions. For example: â—? Ofï¬?cial poverty rates for 2010 were calculated on the basis of per capita incomes in the full VHLSS,19 with some adjustments at provincial levels following discussions with MOLISA. As described in box 2.2, income-based poverty estimates are typically different (and yield a different poverty proï¬?le) than consumption-based estimates. â—? Income-based poverty rates were adjusted for spatial cost-of-living differences using a CPI-based regional deflator rather than the SCOLI. Consumption-based poverty rates were re-estimated using CPI-based spatial cost-of-living adjustments instead of the SCOLI. The impact was small and worked to raise the poverty rate (to 21.5 percent) rather than lower it. 2.55 Neither set of lines is inherently better than the other. As noted in Chapter 1, they are designed to serve different purposes. The strength of the GSO-WB approach lies in consistent poverty monitoring and its independence from budgetary or political considerations. In contrast, Vietnam’s ofï¬?cial poverty lines are primarily intended to help set targets and related resource allocations for targeted poverty reduction programs and policies under Vietnam’s 2011–2015 Socio-Economic Development Plan. In this sense, they are administrative lines, necessarily constrained by resource availability. In response to a new directive on social protection (Resolution 15), MOLISA is developing new measures of average and minimum living standards, which will be used to identify potential beneï¬?ciaries of social assistance and social insurance policies and programs. 2.56 Ofï¬?cial lines were used in carrying out the 2010 Poverty Census in Vietnam. Local surveys were used to identify poor and near-poor households (using short forms, proxy-means-test scorecards, and short income questionnaires), combined with village-level discussions to determine which households had incomes below the ofï¬?cial poverty lines and were eligible to be on the poor list (Prime Minister’s Directive No. 1752/CT-TTg). These lists are being updated annually, again using a mix of survey methods and village-level discussions, often applied differently across the 10,000 or so communes in Vietnam. Analysis suggests that many of those included on the lists are poor, but not all poor households are included on the list (Chapter 3). In short, errors of exclusion are a greater concern than errors of inclusion. 19 Each round of the VHLSS includes around 46,000 households. Detailed information on household income is collected for all households, but consumption information is collected for only 20 percent of households (three in each enumeration area), or 9,400 households in total. Only unit record data from the 20 percent sample (income + consumption) are released to the public. 53 F. Are the New GSO-WB Poverty Lines too High? Are They Consistent with Citizens’ Subjective Views? 2.57 An alternative methodology for estimating subjective poverty lines that has received growing attention in the literature (Kapteyn 1994; Ravallion 2012; Ravallion and Lokshin 2002) was also applied in Vietnam based on additional questions added to the 2010 VHLSS to elicit households’ own assessment of whether their consumption of important items, such as foods, foodstuffs, electricity, water, clothing, and housing, was sufï¬?cient to meet their needs. (See Annex 2.3 for technical details, and Marra 2012.) For example, the following question was intended to assess adequacy of food (for example, rice, basic food grains, and staples) and foodstuffs (for example, meats, vegetables, condiments): 2.58 The intuition behind subjective poverty lines is straightforward: households whose observed incomes are above the subjective poverty line (that is, marked in red in ï¬?gure 2.4, panel A) feel they have enough or more than enough income to meet their needs, while households with observed incomes below the subjective line consider their incomes inadequate to meet their needs. The approach used here is slightly different and is based on perceptions of the adequacy of speciï¬?c items, for example, foodstuffs. In the case of foodstuffs, panel B shows that, in 2010, poorer households (deciles 1 and 2) were much less likely than better-off households to say their consumption of foodstuffs was sufï¬?cient. Figure 2.4 Measuring Subjective Poverty Panel A, Stylized Case Panel B Based on the 2010 VHLSS 2.59 Overall, responses to these questions suggest that less than 5 percent of the households in the 2010 VHLSS felt they had consumed insufï¬?cient amounts of food in the 30 days preceding the survey. Acute hunger is no longer a major issue for Vietnam. However, 11.5 percent of households indicated insufï¬?cient consumption of foodstuffs, and the percentage was signiï¬?cantly higher in rural than in urban areas—14 percent compared to 5 percent (ï¬?gure 2.5). A surprisingly high percentage of households (25 percent in rural areas) reported they were not able to consume sufï¬?cient electricity in the 30 days before the survey. This almost certainly reflects supply-side problems with the quality and availability of electricity in 2010 rather than concerns about affordability; 2010 was a drought year in many parts of Vietnam, and load-shedding and brownouts were widespread. 54 Figure 2.5 Perceived Sufï¬?ciency of Consumption by Urban and Rural, 2010 100% 90% 80% 70% ble Not applicab 60% More than sufficient 50% Sufficient 40% 30% Insufficient 20% 10% 0% ral Urban Rura Rur l Urban Rural Urban Rural U al Urban Rural U Urban Rural Urban Food odstuff Foo ctricity Elec Wa ater Housing ng Clothin Source: 2010 VHLSS. 2.60 Perceptions of sufï¬?ciency also differed across regions. Households in poorer regions (for example, Northern Mountains, Central Highlands) were more likely to report insufï¬?cient levels of consumption. Concerns about insufï¬?cient electricity were particularly high in regions in the north of Vietnam. 2.61 The responses to these questions can be used to calculate a subjective poverty line, following an approach proposed in Pradhan and Ravallion (2000). The perceived sufï¬?ciency of consumption is regressed against characteristics of the household such as total consumption, size, gender composition, age, and education of members. Different regression models were used to test for the sensitivity of results. Based on regression results, subjective poverty lines were calculated as the minimum total expenditure needed by a household to meet sufï¬?cient (foodstuff) consumption needs. (Annex 2.3 provides a more detailed description of the derivation of subjective poverty lines.) 2.62 Subjective poverty lines for 2010 ranged from a high of VND 888,000 per person per month to a low of VND 616,000 per person per month depending on the exact speciï¬?cation of the regression model. All estimates of subjective poverty lines were higher than Vietnam’s ofï¬?cial poverty lines, and nearly all were higher than the new GSO-WB poverty line (VND 653,000 per person per month). Most lines were clustered in the range of VND 700,000 to VND 800,000. 2.63 Estimates of subjective poverty lines suggest that the updated GSO-WB poverty lines and related poverty estimates do indeed reflect the aspirations and perceptions of the Vietnamese population. 55 Table A2.1 Reference Food Basket for Different Population Groups 2.5Ͳ20th 2.5Ͳ10th 2.5Ͳ20th percentile Reference Group: percentile percentile Ethnic Ethnic Subpopulation: (all) (all) minorities majority Urban Rural Food item Plain rice (including fragrant and specialty rice) 66.4 69.1 64.2 68.2 63.1 66.7 Sticky rice 4.2 4.4 7.9 1.1 1.2 4.5 Maize (in seed equivalent) 1.6 2.6 2.7 0.6 1.1 1.6 Cassava (in freshͲtype equivalent) 1.0 1.4 1.9 0.2 0.3 1.0 Potato of various kinds (in freshͲtype equivalent) 0.3 0.2 0.3 0.3 0.3 0.3 Wheat grains, bread, wheat powder 0.3 0.2 0.2 0.4 0.5 0.3 Flour noodle, instant rice noodle/porridge 1.3 1.0 1.1 1.4 1.9 1.2 Fresh rice noodle, dried rice noodle 0.4 0.3 0.3 0.6 0.6 0.4 Vermicelli 0.1 0.1 0.0 0.1 0.1 0.1 Pork (in equivalent of the pork type with removed fat) 4.0 3.6 4.0 4.1 4.3 4.0 Beef 0.1 0.1 0.1 0.1 0.1 0.1 Buffalo meat 0.0 0.0 0.1 0.0 0.0 0.0 Chicken meat 0.9 0.8 1.0 0.8 0.9 0.9 Duck and other poultry meat 0.2 0.1 0.1 0.2 0.1 0.2 Other types of meat 0.0 0.0 0.0 0.0 0.1 0.0 Processed meat 0.1 0.1 0.1 0.1 0.1 0.1 Lard, cooking oil 4.2 3.9 4.0 4.3 4.4 4.1 Fresh shrimp, fish 1.4 1.2 0.8 1.9 1.8 1.4 Dried and processed shrimps, fish 0.3 0.3 0.4 0.3 0.3 0.3 Other aquatic products and seafood (crabs, snails,...) 0.1 0.1 0.1 0.1 0.1 0.1 Eggs of chickens, ducks, Muscovy ducks, geese 0.7 0.6 0.5 0.8 0.8 0.7 Tofu 0.6 0.6 0.6 0.7 0.6 0.6 Peanuts, sesame 0.5 0.4 0.5 0.6 0.5 0.5 Beans of various kinds 0.3 0.2 0.3 0.2 0.3 0.2 Fresh peas of various kinds 0.1 0.1 0.1 0.1 0.1 0.1 Morning glory vegetables 0.5 0.5 0.4 0.7 0.6 0.5 Kohlrabi 0.1 0.1 0.0 0.1 0.1 0.1 Cabbage 0.1 0.1 0.1 0.1 0.2 0.1 Tomato 0.0 0.0 0.0 0.1 0.1 0.0 Other vegetables 0.7 0.6 0.7 0.6 0.8 0.6 Orange 0.0 0.0 0.0 0.0 0.1 0.0 Banana 0.6 0.6 0.6 0.5 0.5 0.6 Mango 0.0 0.0 0.0 0.0 0.0 0.0 Other fruits 0.4 0.3 0.3 0.5 0.6 0.4 Fish sauce 0.2 0.1 0.1 0.2 0.2 0.1 Salt 0.0 0.0 0.0 0.0 0.0 0.0 MSG Glutamate Sugar, molasses 1.3 1.0 0.8 1.7 1.6 1.3 Confectionery 0.6 0.6 0.6 0.7 0.8 0.6 Condensed milk, milk powder 0.2 0.1 0.1 0.2 0.2 0.2 Ice cream, yoghurt 0.0 0.0 0.0 0.0 0.1 0.0 Fresh milk 0.1 0.0 0.0 0.1 0.1 0.1 Alcohol of various kinds 1.3 1.3 1.7 0.9 1.0 1.3 Beer of various kinds 0.1 0.0 0.0 0.1 0.1 0.0 Bottled, canned, boxed beverages 0.1 0.1 0.0 0.1 0.2 0.1 Instant coffee Coffee powder 0.0 0.0 0.0 0.1 0.1 0.0 Instant tea powder Other dried tea 0.4 0.3 0.3 0.4 0.3 0.4 Tobacco Betel leaves, areca nuts, lime, betel pieces Outdoors meals and drinks 3.3 2.1 2.1 4.3 7.6 2.9 Other food and drinks 1.0 0.8 0.8 1.1 1.3 0.9 Source: 2010 VHLSS. 56 Chapter Annexes Annex 2.1: Differences between “Temporally Comparableâ€? and Comprehensive Welfare Aggregates Temporally Comparable Comprehensive Food Excludes consumption of tobacco Includes consumption of all 54 and betel nut. Assumes food items food items in VHLSS. Assumes listed in section 5A2 but not listed the only food items consumed in 5A1 were consumed during Tet/ during Tet/holidays were those holidays. Tet/holidays considered listed in section 5A1. Tet/holidays 15.2 days long considered 14 days long. Durables Excludes consumption of certain Includes all types of durables durables: printers, photocopiers, in 2010 VHLSS, but does not mobile phones, microwaves, impute consumption for durables blenders, other transport. Imputes acquired more than 10 years using depreciation rates from prior. Imputes using depreciation 1998 VLSS and real interest rate rates calculated from 2010 of 5 percent. VHLSS data and real interest rate of 5 percent. Housing Imputes housing consumption Imputes housing consumption as 11.8 percent of other nonfood as 2.88 percent of reported consumption for rural households housing values. 2.88 percent is and 21.4 percent for urban the median ratio of rental income households. to housing values for the 2.6 percent of households in the 2010 VHLSS who are renters. Education Equals total expenditures related Also includes supplemental to compulsory school subjects. expenditure on education, e.g., for tutors, typing classes, etc. Health Equals spending on curative Also includes spending on health and preventive care, including insurance. out-of-pocket costs of inpatient and outpatient health services, expenditures for nonprescription medicine, and expenditure on medical tools. Utilities: Electricity, Water, Simple sum of reported spending. Same. Garbage Other nonfood items (e.g., Excludes spending on parties and clothing, fuel, kitchen celebrations, and consumption of items, services, etc.) self-produced daily nonfood items from section 5B1. Temporal deflator GSO’s rice, nonrice food, and Same. nonfood monthly CPI. Spatial deflator GSO’s regional CPI. 2010 SCOLI. 57 Annex 2.2: Spatial Cost-of-living Estimates for 2010 VHLSS A detailed price survey of 64 items was conducted in the main market in all communes in the October 2010 round of the VHLSS sample (n = 1049) and in half the communes in the December 2010 round (n = 539). The 64 items included 45 speciï¬?cally identiï¬?ed foods (including outdoor meals), and another 19 specially identiï¬?ed nonfoods, including some durable goods and services. It was important to ensure consistency over space in the list of 64 items and to avoid problems with missing observations. Surveyors were given detailed speciï¬?cations (aided by photographs to ensure standardization) and were instructed to take two observations on the price of the detailed speciï¬?cation and to record whether that particular speciï¬?cation was the most common one in the market. A particular size, and brand name (for packaged goods), was speciï¬?ed to avoid variation due to either bulk discounting or quality discounting. In almost 80 percent of the market-item combinations, the speciï¬?cation listed in the questionnaire was indeed the most common; it was available but not the most common in approximately 5 percent of markets. To deal with the missing prices problem in the remaining market-item combinations, surveyors also collected the price of the most commonly available speciï¬?cation that was not the target speciï¬?cation. The price of the target speciï¬?cation was regressed against the prices of the alternate speciï¬?cations (using brand name ï¬?xed effects, or for unbranded items, creating quasi-brands by dividing into intervals based on their unit prices) and a set of regional ï¬?xed effects. The regressions were used to impute the price of the target speciï¬?cation in about 10 percent of markets. District or province average prices were used to impute the missing commune-level prices in the remaining few cases. There are a number of different indexes that are used to adjust for cost-of-living differences. The Consumer Price Index (CPI) is typically based on a Laspayres index. For purposes of the SCOLI, new prices were combined with regional budget shares from the 2010 VHLSS in order to calculate a Törnqvist price index. The Törnqvist index is the geometric average of the price relativities between region i and the base region, weighted by the arithmetic average of the budget shares for the two regions. à­Ž à­©à­¨ ൅  ୧୨ ୧୨ ܶ ൌ Ý?‫݌ݔ‬ሾà·? ൬ ൰ ÂŽÂ? ቆ ቇሿ Í´ à­©à­¨ ୨ୀଵ where P denotes prices in each region and S is the budget shares. The Törnqvist index speciï¬?cally accounts for the fact that consumers will substitute away from items that are expensive in their own region, relative to the base region, by using the budget shares of both the base region and the own region when weighting the price relativities. Technically, it closely approximates a true cost-of-living index for any arbitrary utility function, whereas the Laspeyres index (used for the CPI) is an exact measure of the cost-of-living index only when items are consumed in ï¬?xed proportions, without allowing for substitutions. Because only 64 items had prices obtained in the SCOLI survey, while there are over 100 consumption items listed in the VHLSS (including the consumption of housing services and the service flow from durables), a mapping on prices to budget shares was formed, where the price relativities for some closely related items were used as a proxy for the missing price relativities for other items. Two exceptions were for utilities, where the trimmed median unit value of electricity tariffs in each region and sector was used as the proxy to form a price relativity and flow of accommodation services from dwellings. For the imputed rents, detailed econometric analysis of the housing section of the VHLSS questionnaire was undertaken, to estimate a hedonic house value equation, which allowed for regional differences in the cost of constant-quality housing service. 58 Annex 2.3: Subjective Poverty in Vietnam It is often argued that as countries develop and become less poor, societies’ standards also evolve. Even if the basic point of departure is to measure poverty with an “absoluteâ€? poverty line that is held ï¬?xed in real terms over time, societies will need to update this poverty line from time to tome so it remains relevant to a country’s speciï¬?c circumstances. As noted in chapter 2, as countries grow their national poverty lines increase over time. Regardless of how carefully an absolute poverty line is developed, it is not possible to avoid some degree of arbitrariness. Challenges in setting a poverty line are groups by Ravallion (2012) into (i) a referencing problem, including the choice of the reference group and basket, and (ii) an identiï¬?cation problem that involves translating households’ utility function into the measurable expenditure space. An alternative method for analyzing poverty that has received growing attention builds on subjective welfare questions included in household surveys. A subjective poverty line built up from such questions can offer an alternative entry point into the derivation of the poverty line, also help with the interpretation of the conventionally derived, Cost-of-Basic-Needs (CBN) poverty line. This subjective poverty line exercise is particularly interesting in the context of Vietnam given the proposed update to the 2010 CBN poverty line. Van Praag (1968) introduced subjective welfare assessment by constructing utility functions based on respondents’ answers to the question of how much income they regarded as “very bad,â€?, “bad,â€? and so forth, to “very good.â€? A similar method, the Minimum Income Question (MIQ), asks about the minimum income that respondents perceive to be necessary “to make ends meetâ€? (Kapteyn 1994). However, applicability of the MIQ methodology to the poorest countries has been debated (Deaton and Zaidi 2002; Pradhan and Ravallion 2000; Ravallion and Lokshin 2000). Pradhan and Ravallion (2000) propose an adaptation to Kapteyn’s method by asking households if their consumption of food (and other things) has been adequate to “meet their needs.â€? The 2010 VHLSS included a set of similar questions, allowing us to follow a similar estimation methodology. The exact framing of the question, asked of the household head, is the following: The same question as above is asked about “waterâ€? “electricityâ€? “housingâ€? “clothing & footwearâ€? Out of total respondents to the 2010 VHLSS consumption section, 440 reported insufï¬?cient food consumption, 8,218 reported just sufï¬?cient food, and 686 indicated that their food consumption was more than sufï¬?cient (54 households did not respond). Satisfaction with adequacy of foodstuff consumption (including higher-cost calories from meat, vegetables, oils, and condiments) was less: 1,079 respondents reported inadequate consumption of foodstuffs, 7,580 indicated sufï¬?cient consumption, and 678 claimed their consumption was more than sufï¬?cient. To calculate a subjective poverty line, we follow Pradhan and Ravallion (2000) in regressing perceived sufï¬?ciency of consumption on household expenditure and household (head) characteristics, using the sufï¬?ciency of foodstuff as the dependent variable. “Not Applicableâ€? responses were excluded, and the other three categories are subjected to an ordered probit regression including actual household consumption, household size, and characteristics of the household head. Regression coefï¬?cients, presented in table A2.1, were also used in calculating a range of subjective poverty lines, including those reported in the chapter. 59 Table A2.1 Subjective Welfare Regression and Variables at Country Means Regression Results Means Variables Coefï¬?cient S.E. of Mean S.D. Log total household expenditure 0.717*** 0.029 10.978 0.731 Log household size -0.475*** 0.049 1.435 0.381 Household head is female -0.092** 0.040 0.220 0.414 Household head has a wage job -0.172*** 0.031 0.407 0.491 Household has at least one widow(er) -0.040 0.042 0.186 0.389 Highest grade household head 0.022*** 0.005 7.313 3.683 Household head is registered within the commune 0.046 0.034 0.256 0.437 Household head is of ethnic majority (Kinh) 0.516*** 0.044 0.854 0.353 Share of household < 18 years old 0.206*** 0.078 0.256 0.206 Share of household > 59 years old 0.009 0.093 0.072 0.175 Log land area owned by household 0.029*** 0.005 4.859 3.757 Urban -0.148*** 0.041 1.297 0.457 Cutoff 1 6.264*** 0.277 Cutoff 2 9.327*** 0.289 Number of observations 9,337 Pseudo R2 0.139 Note: The dependent variable is “perceived sufï¬?ciency of foodstuff consumptionâ€? with the following answer codes: 1 = insufï¬?cient, 2 = sufï¬?cient, and 3 = more than sufï¬?cient (“not applicableâ€? is recoded as missing). The results are from an ordered probit regression. The natural logarithm is used for the log variables. Signiï¬?cance levels are *** 0.01, **0.05, * 0.1. The means of the variables and the regression are both weighted by population weights. 60 References Banerjee, Abhijit. 2011. “Draw the right line.â€? Hindustan Times, October 24. Accessed May 2012. http://www.hinustantimes.com/StoryPage/Print/761099.aspx. Bertrand, M., and S. Mullainathan. 2001. “Do People Mean What They Say? Implications for Subjective Survey Data.â€? American Economic Review, Papers and Proceedings 91 (2): 67–72. Conti, G., and S. Pudney. 2011. “Survey Design and the Analysis of Satisfaction.â€? Review of Economics and Statistics 93 (3): 1087–1093. Deaton, A., and O. Dupriez. 2011. “Spatial Price Differences within Large Countries.â€? Princeton University Working Papers, Princeton, NJ. Deaton, A., and S. Zaidi. 2002. “A Guide to Aggregating Consumption Expenditures.â€? Living Standards Measurement Study Working Paper No. 135, World Bank, Washington, DC. Deaton, Angus. 1988. “Quality, Quantity, and Spatial Variation in Price.â€? American Economic Review 78 (3): 418–30. Deaton, Angus. 1997. Analysis of Household Surveys: A Microeconometric Approach to Development Policy. Washington, DC: The Johns Hopkins University Press and World Bank. Deaton, Angus, and Alessandro Tarozzi. 2005. “Prices and Poverty in India.â€? In The Great Indian Poverty Debate, ed. Angus Deaton and Valerie Kozel. New Delhi: Macmillan, Chapter 16, pp. 381– 411. Gill, Nikhila. 2012. “Has Poverty Really Dropped in India?â€? New York Times, March 21. Accessed May 2012. http://india.blogs.nytimes.com/2012/03/21/has-poverty-really-dropped-in-india/. Government of India. 2009. “Report of the Expert Group to Review the Methodology for Poverty Estimation.â€? Planning Commission, Government of India, New Delhi. Hansen, H., and T. Nguyen, eds. 2006. Market, Policy, and Poverty Reduction in Vietnam. Hanoi: Vietnam Academy of Social Sciences, Vietnam Cultural Information Publishing House. Haughton, J., Nguyen Thi Thanh Loan, and Nguyen Bui Linh. 2010. Urban Poverty Assessment in Hanoi and HCMC. Hanoi, joint publication of the UNDP and General Statistics Ofï¬?ce. Kapteyn, A. 1994. “The Measurement of Household Cost Functions: Revealed Preference versus Subjective Measures.â€? Journal of Population Economics 7 (4): 333–350. Kapteyn, A., and B. Van Praag. 1976. “A New Approach to the Construction of Family Equivalence Scales.â€? European Economic Review 7: 313–335. Kozel, Valerie, Ian Hinsdale, and Nguyen Phong. 2012. “Updated Methodologies for Poverty Monitoring in Vietnam.â€? Background paper prepared for the 2012 Vietnam Poverty Assessment, World Bank, Washington, DC. Krueger, A. B., and D. Schkade. 2008. “The Reliability of Subjective Well-being Measures.â€? Journal of Public Economics 92 (8–9): 1833–1845. Lokshin, M., N. Umapathi, and S. Paternostro. 2006. “Robustness of Subjective Welfare Analysis in a Poor Developing Country: Madagascar 2001.â€? Journal of Development Studies 42 (4): 559–591. Marra, M. 2012. “Estimating Subjective Poverty Lines for Vietnam.â€? Background note prepared for the 2012 Vietnam Poverty Assessment, World Bank, Washington, DC. Ministry of Health. 2006. “Proposed Nutrition Needs for the Vietnamese.â€? Ministry of Health, Hanoi. Phung, D. T. 2005. “Determination of a Consistent Poverty Line for Vietnam.â€? Environmental and Social Department, General Statistics Ofï¬?ce, Government of Vietnam. 61 Pincus, J., and J. Sender. 2008. “Quantifying Poverty in Vietnam: Who Counts?â€? Journal of Vietnamese Studies 2 (1) (January): 108–150. Pradhan, M., and M. Ravallion. 2000. “Measuring Poverty Using Qualitative Perceptions of Consumption Adequacy.â€? The Review of Economics and Statistics 82 (3): 462–471. Pradhan, M., M. Suryahadi, S. Sumarto, and L. Pritchettt. 2001. “Eating Like Which Jone’s? An Iterative Solution to the Choice of a Poverty Line Reference Group.â€? The Review of Income and Wealth, Series 47 (4): 473–487. Ravallion, Martin. 1998. “Poverty Lines in Theory and Practice.â€? Living Standards Measurement Study Working Paper 133, World Bank, Washington DC. Ravallion, M. 2012. “Poor, or Just Feeling Poor? On Using Subjective Data in Measuring Poverty.â€? World Bank Policy Research Working Paper 5968, World Bank, Washington, DC. Ravallion, M., and B. Bidani. 1994. “How Robust Is a Poverty Proï¬?le?â€? The World Bank Economic Review 8 (1) (January): 75–102. Ravallion, M., and M., Lokshin. 2002. “Self-Rated Economic Welfare in Russia.â€? European Economic Review 46 (8) (September): 1453–1473. Taylor, M. P. 2006. “Tell Me Why I Don’t Like Mondays: Investigating Day of the Week Effects on Job Satisfaction and Psychological Well-being.â€? Journal of the Royal Statistical Society: Series A (Statistics in Society) 169 (1): 127–142. United Nations Statistics Division. 2005. Handbook on Poverty Statistics: Concepts, Methods, and Policy Use, Special Project on Poverty Statistics. New York: United Nations. Van Praag, B., and M. Warnaar. 1997. “The Cost of Children and the Use of Demographic Variables in Consumer Demand.â€? In Handbook of Population and Family Economics, ed. Mark Rosenzweig and Oded Stard. Amsterdam: North-Holland, Chapter 6, pp: 241–273. World Bank. 1999. Vietnam Development Report 2000: Attacking Poverty. Washington, DC: World Bank. 62 Chapter 3 Poverty Proï¬?le: Establishing the Facts about Poverty and the Poor in Vietnam A new poverty proï¬?le is presented that characterizes the poor and the extreme poor and compares them with the rest of society along a number of key dimensions including geographic location, ethnicity, sector of employment, income sources, educational attainment, ownership of durable goods, landholdings, household amenities, child poverty, and coverage under social protection and poverty reduction programs and policies. Statistical analysis is complemented by a rich body of qualitative research. The poor in Vietnam today are similar in important respects to the poor in the late 1990s. Among other factors, poverty is linked to rural and upland locations, agricultural livelihood, ethnic identity, low educational attainment, exposure to risk and rising vulnerability. 63 A. Introduction 3.1 Poverty reduction remains a challenge in Vietnam, albeit one that has changed dramatically in scope and nature over the last two decades. This chapter revisits the basic facts about poverty and the poor in Vietnam. It takes stock of what we know about poverty today and draws comparisons with the situation of the poor in the late 1990s, with the aim of highlighting both important areas of progress and remaining and new challenges. The chapter presents a new proï¬?le of the poor, using the 2010 General Statistics Ofï¬?ce-World Bank (GSO-WB) poverty line and more comprehensive measures of household welfare proposed in Chapter 2. The analysis is primarily based on the 2010 Vietnam Household Living Standards Survey (VHLSS), but also draws selectively on earlier rounds of the Vietnam Living Standards Survey (VLSS), (particularly the 1998 VLSS), and other sources, such as recent Participatory Poverty Assessments and qualitative ï¬?eld studies, 2009 poverty maps, and other supplementary data sets. 3.2 A poverty line only discriminates between poor and non-poor households. It ignores the fact that not all poor people are the same; some have incomes or consumption very close to the poverty line, while others live in much poorer conditions. Nor are the non-poor homogeneous; some live near the poverty line (referred to as the “near-poorâ€? in Vietnam) while others are much more prosperous. The analysis presented in this chapter recognizes the broad economic diversity among poor and non-poor households in Vietnam. At the lower end of the welfare distribution, we distinguish between the “extreme poorâ€? (per-capita expenditures below two-thirds of the poverty line) and “poorâ€? (per- capita expenditures below the poverty line). The remainder of the population is analyzed on the basis of per-capita expenditure quintiles and deciles. Speciï¬?cally: â—? Individuals are ranked by per-capita expenditures from least well-off to most well-off, then divided into ï¬?ve equally-sized population groups (for quintiles) and ten equally sized population groups (for deciles). Quintile 1 comprises the poorest 20 percent of the population, and quintile 5 comprises the wealthiest 20 percent. Similarly, decile 1 comprises the poorest 10 percent of the population and decile 10 the wealthiest 10 percent. â—? Individuals are also categorized into expanded per-capita expenditure quintiles, where the poor are classiï¬?ed into two groups (all poor and extreme poor) and the non-poor are classiï¬?ed by the standard per-capita expenditure quintiles. Expanded quintiles thus comprise six groups: o The extreme poor: individuals whose per-capita expenditures are less than two-thirds of the poverty line (poorest 8 percent of the population) o All poor: individuals whose per-capita expenditures are below the poverty line (poorest 20.7 percent of the population) o And quintiles 2 through 5 (as above). 3.3 In the context of the 2006-2010 Socio-Economic Development Plan (SEDP), the Ministry of Labour, Invalids and Social Affairs (MOLISA) introduced a “near-poorâ€? classiï¬?cation, which includes households whose per-capita income lies between the poverty line and 1.3 times the poverty line. If this deï¬?nition is applied to the 2010 GSO-WB poverty line, roughly three-quarters of individuals in quintile 2 would fall into the near-poor group. 3.4 As a follow-on to the Millennium Development Goals, the World Bank is proposing to launch a new global initiative designed to accelerate the rate of poverty reduction among the poorest and most destitute and to promote shared prosperity over the next decade. Research from countries throughout the world shows that the poorest and most destitute are more difï¬?cult to reach than those living close to the poverty line; they face a structural barriers and speciï¬?c constraints, and better policies and programs are needed to address these speciï¬?c challenges. In many countries, including Vietnam, the extreme and destitute poor are falling further behind. This chapter develops proï¬?les of the extreme poor as well as the total poor, and recognizes that many of the near-poor (quintile 2) remain vulnerable to falling (back) into poverty. 64 3.5 In constructing the poverty proï¬?le, households and individuals are also categorized by socioeconomic group (ethnic minority, Kinh majority), sector (urban, rural), and economic region. The Government of Vietnam has identiï¬?ed eight economic regions encompassing 63 provinces, more than 680 districts, and two major urban areas (Hanoi and Ho Chi Minh City). Annex 3.1 provides a description of the eight economic regions including the North East region, North West region, the Red River Delta (which houses Hanoi), the North Central Coast, the South Central Coast, the Central Highlands, the South East (which houses HCMC), and the Mekong River Delta. The North East and North West are mountainous regions where the majority of Vietnam’s ethnic minorities reside. Ethnic minorities also live in upland areas of central and southern regions, particularly the Central Highlands. The two deltas (Red River, Mekong) are major rice growing regions, and the majority of Vietnam’s rice exports come from the Mekong River Delta. The Stylized Facts about Poverty and Poor Households at the End of the 1990s 3.6 The Vietnam Development Report 2000: Attacking Poverty (World Bank 1999) described the key characteristics of poor households at the end of the 1990s, drawing on the 1993 and 1998 VLSS combined with a series of Participatory Poverty Assessments (PPAs) carried out in 1999. These early PPAs stressed core poverty concerns like hunger; lack of productive assets; high exposure to adverse shocks like drought, flooding, and illnesses; and concerns about social marginalization and isolation (particularly for ethnic minority groups). Many poor households struggled to feed and educate large families, and child poverty was widespread. Landlessness was rising, and there were limited options for off-farm employment (box 3.1). Box 3.1 Deï¬?ning Characteristics of Poor Households at the end of the 1990s By the end of the 1990s, the key deï¬?ning characteristics of poor households included: â—? The poor lived in rural areas and were predominantly farmers with low levels of educational attainment, limited access to information, and low function skills. In 1998, nearly four-ï¬?fths of the poor were agriculture households. â—? Poor households had small landholdings, and landlessness was increasing, especially in the Mekong Delta. Households that were unable to make a living from the land found few opportunities for stable off-farm income generation. There was an urgent need for reforms to stimulate demand for off-farm employment. â—? Households with many children or few laborers were disproportionately poor and were particularly vulnerable to rising and variable health and education costs. Newly formed households went through an initial phase of poverty, often aggravated by limited access to land. Poor households were also frequently caught in a debt trap. â—? Poor households were vulnerable to seasonal hardship and household-speciï¬?c and communitywide shocks and some were socially and physically isolated. â—? Poverty among ethnic minority groups had declined, but not as rapidly as for the majority population. Ethnic minorities faced many speciï¬?c disadvantages that could best be addressed through an Ethnic Minority Development Program. â—? Migrants to urban areas who were poor and who had not secured permanent registration faced difï¬?culties accessing public services and some felt socially marginalized. Further work was needed to identify the best way to help these groups. â—? Children were overrepresented in the poor population; they were less able to attend school and were trapped in a cycle of inherited poverty. Many felt insecure and uncertain about their future. Source: World Bank 1999. 65 Many of these Stylized Facts are still True Today 3.7 Although poverty has fallen dramatically, many of the factors that characterized the poor in the 1990s still characterize the poor today: low education and skills, heavy dependency on subsistence agriculture, physical and social isolation, speciï¬?c disadvantages linked to ethnic identity, and exposure to natural disasters and risks. Those that moved out of poverty acquired more schooling and job skills, diversiï¬?ed out of agriculture and into manufacturing and services, and reduced exposure to seasonal hardships and shocks through income diversiï¬?cation and migration. But some of the stylized facts have changed. For example, issues such as ethnic minority poverty that were only emerging as concerns in the late 1990s are much greater concerns today. Other issues, like poverty and vulnerability among migrants in urban areas, have become lesser concerns. Although income poverty remains very low in Vietnam’s cities and towns, there is evidence that new forms of poverty are arising: urban households are particularly vulnerable to sharp bouts of inflation and a rising cost of living. Risk remains an important feature of the rural economy as well, including weather-related risks and the emerging impacts of climate change for agriculture. B. The Poor in Vietnam still Predominately Live in Rural Areas and are Increasingly Concentrated in Upland Regions 3.8 As shown in table 3.1, an estimated 20.7 percent of the population was poor in 2010 and 8 percent was extremely poor. Poverty remains a rural phenomenon in Vietnam; more than 90 percent of the poor and 94 percent of the extreme poor live in rural areas. The poor in urban areas for the most part live in smaller cities and towns (Section G). However, qualitative studies complete for this report and recent research on urban poverty (Haughton et. al. 2010) suggest that urban low-income households are impacted by other (non-income) dimensions of poverty, such as poor sanitation, lack of adequate housing, limited coverage of social insurance, increasing exposure to risk, and continuing vulnerability to poverty. Table 3.1 2010 Poverty Headcount and Composition, by Region and Sector Poverty ExtremePoverty Shareof Contribution Contribution totalpop Index(%) tototal(%) Index(%) tototal(%) (%) National 20.7 100.0 8.0 100.0 100.0 RedRiverDelta 11.4 12.3 2.8 7.8 22.3 EastNorthernMountains 37.7 20.8 17.9 25.8 11.5 WestNorthernMountains 60.1 9.1 36.5 14.4 3.2 NorthCentralCoast 28.4 16.5 9.7 14.6 12.0 SouthCentralCoast 18.1 7.4 5.9 6.3 8.5 CentralHighlands 32.8 9.5 17.0 12.9 6.0 Southeast 8.6 7.2 3.1 6.9 17.5 MekongRiverDelta 18.7 17.1 4.8 11.4 19.0 Rural 27.0 91.4 10.7 94.4 70.3 Urban 6.0 8.6 1.5 5.6 29.7 Source: 2010 VHLSS. 3.9 The spatial distribution of poverty has changed over time. In the 1990s, poverty was widespread in Vietnam. Although poverty rates were higher in some regions than others, (for example, in sparsely settled provinces in the Northern Mountains and Central Highlands), the majority of the poor lived in the 66 more densely settled Delta regions (ï¬?gure 3.1). Poverty fell throughout Vietnam between 1998 and 2010, but it fell more rapidly in fast-growing regions around Hanoi and Ho Chi Minh City (that is, the Red River Delta and the Southeast). Uneven progress has resulted in substantial changes in the spatial distribution of poverty, with the remaining poor becoming more concentrated in the upland areas in the north of Vietnam and in the Central Highlands (ï¬?gure 3.2). Chapter 4 uses poverty mapping methods to look at the spatial distribution of poverty at lower levels of spatial disaggregation (provinces and districts). Figure 3.1 Level and Composition of Figure 3.2 Level and Composition of Poverty by Region, 1998 Poverty by Region, 2010 80 80 70 70 WBͲGSO poverty rate WBͲGSO poverty rate Contribu on to total Contribu on to total 60 60 50 50 Na onal WBͲGSO poverty rate: 37.4 40 40 30 30 Na onal WBͲGSO poverty rate: 20.7 20 20 10 10 0 0 Red River Delta East Northern West Northern North Central South Central Central Southeast Mekong River Red River Delta East Northern West Northern North Central South Central Central Southeast Mekong River Mountains Mountains Coast Coast Highlands Delta Mountains Mountains Coast Coast Highlands Delta Source: 1998 VLSS. Source: 2010 VHLSS. C. Many of the Poor are Farmers Whose Livelihoods are Primarily Linked to Agriculture 3.10 The poor in Vietnam are still predominately farmers; 32.9 percent of agricultural households live below the poverty line,20 which is nearly three times higher than the national poverty rate, and agricultural households make up 65 percent of the poor and 73 percent of the extreme poor compared with a population share of only 41 percent (table 3.2). Agricultural households also contribute disproportionately to the poverty gap and poverty severity. Table 3.2 Poverty Headcount and Composition in 2010, by Sector of Employment of Household Head Poverty ExtremePoverty Shareof Contribution Contribution totalpop Index(%) tototal(%) Index(%) tototal(%) (%) National 20.7 100.0 8.0 100.0 100.0 Employmentofhouseholdhead: Notemployed 13.2 9.1 5.3 9.6 14.4 Agriculture 32.9 64.8 14.1 72.5 40.9 Familybusiness 5.9 4.4 1.2 2.3 15.4 Employedforwagesin: Industry&manufacturing 13.2 4.0 2.7 2.1 6.3 Construction 19.3 7.7 5.1 5.3 8.3 Services 14.0 10.0 4.4 8.2 14.9 Source: 2010 VHLSS. 20 Deï¬?ned as households where the head’s main job is in agriculture. 67 3.11 The level and composition of household income across the expanded per-capita expenditure quintiles is described in ï¬?gure 3.3. The height of each bar reflects the average level of per-capita income for each group. Figure 3.4 looks in greater detail at the composition of income for each group, broken down by income from agriculture sources (crop cultivation, livestock, forestry, aquaculture, and agriculture wages), nonfarm family enterprises, non-agriculture wages, social transfers, domestic and overseas remittances, and other sources. According to ï¬?gure 3.4, poor households derive roughly half their income from agricultural activities, including agricultural wages. However, what differentiates the incomes of the poor from wealthier households is not the level of income from agricultural activities; crop incomes are surprisingly equal across wealth quintiles, reflecting Vietnam’s broadly egalitarian distribution of agriculture land. What differentiates the incomes of the poor from wealthier households is, instead, the extent to which households have successfully diversiï¬?ed into off-farm activities. Progress in the 1990s was driven by on-farm diversiï¬?cation, for instance into cash crops, livestock, and (in some parts of the country) ï¬?sh and shrimp farming (World Bank 1999). But progress in recent years has been driven by diversiï¬?cation into business and trading and, even more importantly, by salaried employment in industry and manufacturing and jobs in the service sector. Even the extreme poor have income sources outside agriculture, although as shown in the next section, this differs for poor minority households compared to poor minorities. Figure 3.3 Household Income by Expanded Figure 3.4 Composition of Income by Quintile, 2010 Expanded Quintile, 2010 Level of household incomes, million VND Composition of household income (percent) (January 2010) Source: 2010 VHLSS. D. Ethnic Identity Matters even more for Poverty Today 3.12 Although Vietnam’s 53 ethnic minority groups make up only 15 percent of the total population, they account for nearly half (47 percent) of the total poor and 68 percent of the extreme poor in Vietnam. (Figure 3.5). Although living conditions for many minorities have improved since the late 1990s, the concentration of minorities among the poor has nonetheless increased dramatically —by 25 percentage points for the extreme poor (from 43 percent in 1998 to 68 percent in 2010) and 19 percentage points for the poor (from 28 percent in 1998 to 47 percent in 2010). 68 Figure 3.5 Composition of Poor and Better-off Households in 2010, by Ethnicity Ethnic minori es Majority 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 1998 2010 1998 2010 1998 2010 1998 2010 1998 2010 1998 2010 Extreme poor All poor Quin le 2 Quin le 3 Quin le 4 Quin le 5 Sources: 1998 VLSS and 2010 VHLSS. 3.13 Despite progress, as shown in the Table 3.3, 66.3 percent of minorities still lived below the poverty line and 37.4 percent lived below the extreme poverty line in 2010. In comparison, only 12.9 percent of the Kinh majority population was still poor and 2.9 percent lived below the extreme poverty line in 2010. (Table 3.4) Because the Kinh make up a much larger share of the population in Vietnam, they still account for just over half (53 percent) of the total poor in Vietnam. Table 3.3 Ethnic Minority Poverty: Headcount and Composition in 2010, Region and Sector Poverty Extreme Poverty Share of Contribution Contribution total pop Index (%) to total (%) Index (%) to total (%) (%) National 66.3 100.0 37.4 100.0 100.0 Red River Delta 13.1 0.2 0.0 0.0 1.0 East Northern Mountains 64.8 35.4 34.9 33.9 36.2 West Northern Mountains 72.8 18.9 45.5 20.9 17.2 North Central Coast 71.2 14.0 34.8 12.1 13.0 South Central Coast 78.4 5.3 50.7 6.1 4.5 Central Highlands 76.6 15.2 50.4 17.7 13.1 Southeast 46.4 3.5 22.2 3.0 5.0 Mekong River Delta 50.4 7.6 23.3 6.2 10.0 Rural 68.9 95.5 39.3 96.8 91.9 Urban 36.5 4.5 14.8 3.2 8.1 Source: 2010 VHLSS. 69 Table 3.4 Kinh Majority Poverty: Headcount and Composition in 2010, by Region and Sector Poverty Extreme Poverty Share of Contribution Contribution total pop Index (%) to total (%) Index (%) to total (%) (%) National 12.9 100.0 2.9 100.0 100.0 Red River Delta 11.4 22.9 2.8 24.7 26.0 East Northern Mountains 14.4 8.0 3.3 8.2 7.2 West Northern Mountains 10.7 0.6 1.3 0.3 0.8 North Central Coast 20.4 18.6 4.9 19.8 11.9 South Central Coast 13.0 9.2 2.1 6.5 9.2 Central Highlands 12.4 4.6 1.5 2.4 4.8 Southeast 6.9 10.5 2.3 15.3 19.7 Mekong River Delta 16.1 25.5 3.3 22.7 20.5 Rural 17.0 87.7 3.9 89.1 66.6 Urban 4.8 12.3 1.0 10.9 33.4 Source: 2010 VHLSS. 3.14 Looking beyond the headcount, the poverty conditions experienced by ethnic minority poor are more severe than the conditions experienced by poor Kinh households. Minorities are more heavily concentrated among the extreme poor, as illustrated in table 3.5, and both the depth and severity of poverty are substantially higher for minorities. These differences are illustrated graphically in ï¬?gure 3.6: the distribution of welfare (per-capita expenditures) for minorities who fall below the poverty line is skewed to the left and the overall distribution has a much thinner “tailâ€? than the distribution of welfare for Kinh majorities. In contrast, poor Kinh have welfare levels much closer to the poverty line than poor ethnic minorities. Table 3.5 Poverty Headcount, Gap, and Severity in 2010, Kinh and Ethnic Minorities Headcount PovertyGap PovertySeverity Contributio Contributio ntototal ntototal Contribution Index(%) (%) Index(%) (%) Index(%) tototal(%) Poor: Kinh/Hoa 12.9 53.3 2.7 39.7 0.9 31.1 Ethnicminorities 66.3 46.7 24.3 60.3 11.3 68.9 Extremepoor: Kinh/Hoa 2.9 31.5 0.5 21.5 0.1 15.1 Ethnicminorities 37.4 68.5 9.7 78.5 3.7 84.9 Source: 2010 VHLSS. 70 Figure 3.6 Distribution of Welfare for Kinh and Ethnic Minorities, 2010 3.15 There are important differences in the spatial distribution of Kinh and ethnic minority populations in Vietnam. Minority populations remain heavily concentrated in the East and West Northern Mountains, in the Central Highlands, and (to some extent) in the North Central Coast. In contrast, the Kinh population is concentrated in large cities (including Hanoi and Ho Chi Minh City), the Red River and Mekong deltas, and in lower elevations along the coast and inland areas. The spatial distribution of poverty tends to follow the spatial distribution of their respective populations: poor Kinh households are concentrated in the deltas and in provinces along the North Central Coast. In contrast, most poor minority households live in upland areas, with the West Northern Mountain region and Central Highlands accounting for a somewhat higher share of poor ethnic minorities than their share in the population. Notably, across all locations (with the exception of Red River Delta, where very few ethnic minorities reside), poverty rates among ethnic minorities average between four and seven times higher than poverty rates among the Kinh (ï¬?gures 3.7 and 3.8). Majorities living in minority areas have substantially better living conditions on average than the minorities living in these same areas. Figure 3.7 Level and Composition of Poverty Figure 3.8 Level and Composition of by Region, for Kinh/Hoa Poverty by Region, for Ethnic Minorities 80.0 0 80.0 70.0 0 70.0 Incidence Incidence 60.0 0 60.0 Contribu on to total Contribu on to total 50.0 0 50.0 40.0 0 40.0 30.0 0 30.0 20.0 0 20.0 10.0 0 10.0 0.0 0 0.0 Red Riv ver East West North South Central Southeast Mekong M Red River East West North South Central Southeast Mekong M Delta Northern Northern Central Central Highlands River Delta Delta Northern Northern Central Centrall Highlands River Delta s Mountains Mountains Coast Coast Mountains Mo ountains Coast Coast Source: 2010 VHLSS. Source: 2010 VHLSS. 3.16 Maps 3.1 and 3.2 illustrate the strong spatial segregation between poor minority and poor majority households in Vietnam. Poor minorities are heavily concentrated in the East and West Northern Mountains, upland areas in the North Central Coast, and the Central Highlands. In contrast, 71 poor people from the majority population are concentrated in the Red River Delta, along coastal regions, and in the Mekong Delta. Map 3.1 Spatial Distribution of Poor Minorities Map 3.2 Spatial Distribution of Poor Kinh Sources: Cuong et al. 2012. 3.17 There are important differences in livelihood strategies and employment patterns between poor majority and minority households (Figure 3.9). Poor minorities earn three-quarters of their total income from agriculture and allied activities, including wage employment in agriculture. In contrast, poor majority households earn only 42 percent from agriculture and allied activities and a much higher share from off-farm activities, both salaried non-farm employment and family enterprises. Forestry is important for minorities, but much less so for poor majorities, in large part reflecting differences in residential patterns. Notably, the composition of income is similar between ethnic minorities and majorities in the wealthiest quintile. 72 Figure 3.9 Composition of Income for Extreme Poor, Poor, and Top Quintile in 2010: Comparing Kinh/Hoa and Ethnic Minority Households Source: 2010 VHLSS. E. Poverty is Still Linked to Low Education Attainment 3.18 Vietnamese today are far better educated than they were a decade ago. Primary completion rates were high already by the end of the 1990s, as evidenced in the ï¬?rst panel of Figure 3.10. Since then, the other panels illustrate the rapid increase in enrolments at lower and upper secondary levels, leading to an increase in the number of students who attend colleges and universities. However, lack of education continues to be an important determinate of poverty, and this was highlighted by respondents in both urban and rural areas as a cause of rising inequality (Chapter 6). Figure 3.10 Schooling Achievement by Age Cohort, 1998 and 2010 Completed Primary Completed Lower Secondary 100 100 1998 1998 90 90 2010 2010 80 80 70 70 60 60 Percent Percent 50 50 40 40 30 30 20 20 10 10 0 0 21-25 26-30 31-35 36-40 41-45 46-50 51-55 56-60 61+ 21-25 26-30 31-35 36-40 41-45 46-50 51-55 56-60 61+ Age Age Completed Upper Secondary Completed University 100 100 1998 1998 90 90 2010 2010 80 80 70 70 60 60 Percent Percent 50 50 40 40 30 30 20 20 10 10 0 0 21-25 26-30 31-35 36-40 41-45 46-50 51-55 56-60 61+ 21-25 26-30 31-35 36-40 41-45 46-50 51-55 56-60 61+ Age Age Source: 1998 VLSS, 2010 VHLSS. 73 3.19 As shown in Table 3.6, individuals living in households whose head did not complete primary school have the highest poverty rate in 2010 (nearly 40 percent or twice the national average) as well as the highest extreme poverty rate (nearly 19 percent or two-and-a-half times the national average). The inverse relationship between education and poverty has become stronger over time: in 1998, households whose heads had completed primary or less schooling accounted for 55 percent of the total poor. By 2010, they accounted for 75 percent of the poor. Rising levels of education coupled with rapid income diversiï¬?cation has been a powerful force for poverty reduction in Vietnam since the late 1990s. Table 3.6 Poverty Headcount and Composition in 2010, by Education of Household Head Poverty ExtremePoverty Shareof Contribution Contribution totalpop Index(%) tototal(%) Index(%) tototal(%) (%) National 20.7 100.0 8.0 100.0 100.0 Householdhead's  highesteducational qualification: None 39.8 46.1 19.3 58.1 24.0 Primary 23.5 28.5 7.9 25.0 25.1 Lowersecondary 15.3 18.4 4.2 13.2 24.9 Uppersecondary 8.7 4.2 2.1 2.6 9.9 Vocational 5.8 2.6 0.8 0.9 9.4 Highereducation 0.7 0.2 0.1 0.1 6.6 Source: 2010 VHLSS. 3.20 Table 3.7 describes the distribution of education for persons 21 years and older across expanded per-capita expenditure quintiles, illustrating in yet another way the strong relationship between rising levels of education and increasing wealth in Vietnam. By 2010, 40 percent of persons 21 years and older in the richest quintile had completed a university degree; in contrast, less than 2 percent in the poorest quintile were university graduates. In fact, more than a quarter of those in the poorest quintile had not even completed primary school by 2010. 3.21 Table 3.7 also highlights the gaps in education between ethnic minorities and Kinh majorities. Even among the poor, minorities are substantially less educated than their Kinh economic peers: for example, 39 percent of poor minorities had not completed primary school compared to only 16 percent of poor Kinh majorities. Achievement gaps are in part due to a historical legacy of lower education achievement among many minority populations, but also reflect lower (albeit increasing) current enrolment rates. Figure 3.11 illustrates the relationship between education and total per- capita expenditures for Kinh and minorities documented in Table 3.7. 74 Table 3.7 Distribution of Completed Education in 2010, by Ethnicity and Expanded Quintiles (persons age 21 and older) Lower Upper Higher None Primary secondary secondary Vocational education National Extreme Poor 37.1 28.3 23.4 9.3 1.2 0.7 All Poor 26.7 29.7 28.7 12.3 1.3 1.4 Quintile 2 12.4 26.6 34.7 20.7 3.4 2.3 Quintile 3 6.6 21.6 31.8 27.0 6.1 6.9 Quintile 4 4.7 14.2 23.1 30.3 9.8 17.8 Quintile 5 2.0 7.7 15.6 25.6 9.2 40.0 Rural 13.1 23.1 30.6 21.9 4.7 6.7 Urban 4.7 12.5 17.6 25.9 9.0 30.3 National 10.6 20.0 26.7 23.1 5.9 13.7 Majority Extreme Poor 21.7 25.1 33.6 16.1 2.5 1.0 All Poor 16.4 31.2 34.5 14.2 1.8 2.0 Quintile 2 10.7 26.2 36.0 21.2 3.3 2.6 Quintile 3 6.3 21.6 32.2 27.0 6.0 6.9 Quintile 4 4.5 14.6 23.4 30.3 9.8 17.4 Quintile 5 2.0 7.8 15.7 25.6 9.0 39.9 Ethnic minorities Extreme Poor 44.2 29.8 18.7 6.1 0.6 0.6 All Poor 38.6 28.0 21.9 10.1 0.9 0.6 Quintile 2 23.3 28.5 25.8 17.5 3.9 0.9 Quintile 3 12.2 21.5 25.3 26.1 8.2 6.8 Quintile 4 9.3 7.2 18.3 29.0 10.0 26.3 Quintile 5 4.2 1.7 9.2 23.0 17.1 45.0 Source: 2010 VHLSS. Figure 3.11 Education Achievements by Expanded Quintiles (persons age 21 and older) Kinh/Hoa Ethnic Minorities Source: 2010 VHLSS. 75 3.22 High levels of current enrolments indicate that future generations of workers will be better prepared to participate in Vietnam’s modernizing economy than previous generations. However, gaps in enrolments between children from poor and better-off households have persisted (Table 3.8), including gaps between enrolments for Kinh and ethnic minority children. (Table 3.9) Most primary-school-aged children—rich and poor, minority and majority—are enrolled in school. But enrolments among (poor) minorities drop off at the lower secondary level, and children from lower- income households are much less likely to be enrolled in upper secondary schools than children from better-off households. Chapter 6 analyzes the links between education and rising inequality, including the role of inequality in opportunities (especially education) in perpetuating poverty across generations. Table 3.8 School Enrolment Rates (net) for Boys and Girls in 2010, by Expanded Quintiles and Region Primary Lower Secondary Upper Secondary Male Female Total Male Female Total Male Female Total Extreme Poor 91.6 88.8 90.2 62.2 70.8 66.6 16.4 28.1 22.9 All Poor 90.2 90.2 90.2 68.6 75.6 72.2 28.1 36.1 32.4 Quintile 2 93.7 92.6 93.2 77.5 82.6 79.9 50.0 56.5 53.0 Quintile 3 94.1 92.9 93.5 84.9 85.5 85.2 58.1 62.5 60.3 Quintile 4 92.5 93.7 93.1 90.5 90.4 90.5 66.0 73.6 69.5 Quintile 5 93.3 97.6 95.3 86.1 90.3 88.0 76.2 85.6 80.9 Red River Delta 95.0 93.5 94.3 89.6 91.9 90.6 69.2 67.2 68.2 East Northern Mtns 93.0 90.9 91.9 85.2 83.0 84.1 56.0 60.7 58.3 West Northern Mtns 93.3 93.9 93.6 80.9 65.5 74.2 47.4 38.8 42.7 North Central Coast 90.9 91.1 91.0 83.8 87.6 85.8 54.7 58.9 56.8 South Central Coast 92.1 90.7 91.4 89.5 86.4 88.1 58.4 69.6 64.0 Central Highlands 95.4 87.7 91.9 67.3 78.2 73.1 45.6 52.5 49.3 Southeast 90.3 97.9 94.1 76.1 81.8 78.4 52.8 63.1 58.0 Mekong Delta 91.4 92.7 92.0 66.1 76.5 71.2 39.2 50.5 44.1 Rural 92.4 91.9 92.2 78.9 82.8 80.7 49.3 54.5 51.8 Urban 92.9 95.2 94.1 83.5 85.0 84.2 68.8 76.2 72.5 National 92.5 92.8 92.6 80.0 83.3 81.5 53.9 60.1 57.0 Source: 2010 VHLSS. 3.23 Gender gaps in minority school enrolments have received a lot of attention in Vietnam. These gaps have closed at the primary level but persist at the secondary level and above. However, reverse gender gaps—substantially higher enrolments for girls compared to boys at the secondary level— have started to emerge at the secondary level, particularly among children from poor (majority) households and in the Central Highlands, the Southeast, and the Mekong Delta. Concerns have been raised that boys from poor households are leaving school earlier than girls to take up jobs in the service sector and manufacturing, “pushedâ€? by poverty and economic imperatives and “pulledâ€? by expanding employment opportunities in nearby cities and towns. While leaving school after six or eight years of education may make sense given short-run incentives, education choices made today will follow children for the rest of their lives. These young workers may not have the education and skills to get good jobs in the future as Vietnam’s economy continues to grow and modernize, and Vietnam’s economic development will be constrained by the lack of an educated and skilled labor force. 76 Table 3.9 Net School Enrolment Rates for Kinh/Hoa and Ethnic Minority Boys and Girls in 2010, by Expanded Quintile Primary Lower Secondary Upper Secondary Male Female Total Male Female Total Male Female Total Majority Extreme Poor 92.4 96.4 94.5 69.7 94.1 81.8 27.6 48.5 39.9 All Poor 88.3 94.2 91.0 71.9 85.8 79.5 34.2 46.4 40.8 Quintile 2 93.2 92.1 92.7 75.7 84.2 79.6 50.7 57.7 54.0 Quintile 3 93.8 93.0 93.4 85.2 85.7 85.4 58.1 63.3 60.7 Quintile 4 92.4 94.6 93.5 91.0 90.5 90.7 66.7 75.4 70.7 Quintile 5 93.2 97.5 95.3 86.0 90.2 87.9 76.8 85.3 81.0 Ethnic minorities Extreme Poor 91.4 86.1 88.7 59.4 62.5 61.0 12.4 19.2 16.1 All Poor 92.5 86.5 89.3 65.5 63.1 64.4 22.4 26.3 24.5 Quintile 2 97.4 96.1 96.8 90.1 72.2 81.6 46.1 48.3 47.1 Quintile 3 100.0 90.5 95.4 78.0 82.1 80.3 57.9 43.4 53.1 Quintile 4 94.5 74.9 85.5 80.1 88.9 84.4 58.4 41.2 52.3 Quintile 5 100.0 100.0 100.0 100.0 100.0 100.0 25.7 100.0 75.1 Source: 2010 VHLSS. 3.24 There are many reasons why children from poor and ethnic minority households do not stay in school. High out-of-pocket costs are one factor (Chapter 1). Location is another. In upland regions, particularly in the Northern Mountains, upper secondary schools are often located at some remove from rural communities, and students are forced to board rather than commute to school each day from their homes. Background qualitative studies carried out for this report also highlight widespread concerns about the poor quality of schools in some rural areas. Vietnamese Farmers have Small Landholdings and Landlessness is Rising 3.25 An early and strong commitment by the government to distribute land use rights equitably among farmers in Vietnam has resulted in a pattern of land distribution that remains remarkably equitable by international standards. Rural growth and on-farm diversiï¬?cation were the driving forces for poverty reduction in the 1990s. Most rural households continue to have small landholdings and, in recent years, few households were able to substantially improve their living conditions through expanded cultivation of annual crops. A high percentage of Vietnamese farmers continue to grow rice, in part driven by state restrictions on the use of land. Land use restrictions are primarily in place for rice production, and affect land in the Mekong and Red River Deltas (Markussen, Tarp, and van den Broeck 2009). Except in the Mekong Delta, rice is grown primarily for own consumption rather than as a source of cash income. 72 percent of poor households in Vietnam grew rice according to the 2008 VHLSS; 90 percent of this rice consumed at home, and only 18 percent of poor households were net sellers of rice. Instead, rising wealth among rural households is linked to on-farm diversiï¬?cation into cash crops, and even more important, diversiï¬?cation into off-farm activities. The last decade is notable for rapidly expanding opportunities for stable off-farm income generation, including in industrial centers and nearby towns. 3.26 Less-well-off rural households cultivated, on average, more land than better-off rural households in 2010 (Table 3.10). However, these statistics should be interpreted with care; much of the land cultivated by ethnic minorities is in upland regions and often of lower quality due to sloping and rocky terrain and lack of dependable irrigation. Better-off households cultivate more perennial cropland, which is used for commercial activities (including coffee, an important cash crop). 77 Table 3.10 Average Landholdings for Rural Households in 2010, by Consumption Quintile Quintile 1 2 3 4 5 All land (sq. m.) 8235 6049 5901 5723 5608 of which: Annual crop land 3765 3322 2927 2826 2302 Perennial land 698 1031 1145 1640 2463 Source: 2010 VHLSS. 3.27 The proportion of landless rural households has risen in all regions since the late 1990s (Table 3.11). However, with the exception of the Mekong Delta, landlessness is not associated with higher poverty. In fact, initial analysis suggests a positive relationship between rural landlessness and wealth in most regions in the north of Vietnam. (Table 3.12). But 54 percent of the rural poor living in the Southeast region and 48 percent of the rural poor living in the Mekong Delta are landless (landless rates among extreme poor are similar). Concerns have been raised over the years about the links between landlessness and poverty. Some were concerned that legislation allowing the opening up of land markets in the late-1990s would encourage poor farmers to sell land for quick proï¬?ts, leaving them without adequate means of livelihoods; others argued that land markets would promote greater efï¬?ciency. (Ravallion and Van der Walle 2008a, 2008b) The picture is mixed. Respondents living in Tra Vinh province in the Mekong Delta interviewed for the positive deviance study (Chapter 1) noted expanding opportunities for “land-poorâ€? households in the Mekong and Southeast to diversify into higher paid off-farm activities. However, off-farm diversiï¬?cation requires relevant education and skills. Although young workers can acquire these skills, the situation is more complicated for households with older workers. More work is needed to understand the complex links between landlessness and poverty in Vietnam’s southern provinces. Table 3.11 Percentage of Rural Households without Allocated or Swidden Land, 1993 1998 2010 Northern Mountains 2.0 3.7 8.1 Red River Delta 3.2 4.5 13.4 North Central Coast 3.8 7.7 15.5 South Central Coast 10.7 5.1 19.7 Central Highlands 3.9 2.6 17.3 Southeast 21.3 28.7 58.9 Mekong Delta 16.9 21.3 33.6 National 8.2 10.1 22.5 Source: 1993 and 1998 ï¬?gures taken from the World Bank 2000 Vietnam Development Report, table 2.4. 2010 ï¬?gures are World Bank estimates from 2010 VHLSS. Note: Swidden land is land cleared for cultivation by cutting and burning the vegetation. “Landâ€? includes annual cropland, perennial cropland, forestry land, water surface, and shifting-cultivation farmland. It excludes gardens, ponds, and land classiï¬?ed as “other.â€? Table 3.12 Percent of Rural Households without Allocated or Sweden Land in 2010, by Region and Quintile Quintile Extremepoor 1 2 3 4 5 RedRiverDelta 2.2 4.6 4.8 7.9 14.6 30.5 EastNorthernMountains 0.7 2.2 4.8 9.6 20.9 31.4 WestNorthernMountains 0.5 0.6 5.3 5.5 38.7 56.9 NorthCentralCoast 7.9 7.9 9.9 14.9 22.6 52.0 SouthCentralCoast 2.5 10.6 14.6 16.7 21.7 25.3 CentralHighlands 13.2 9.6 17.0 27.6 21.1 23.9 Southeast 43.4 53.9 43.4 53.6 56.5 68.5 Mekong RiverDelta 50.3 47.5 29.0 29.7 30.6 34.9 Source: 2010 VHLSS. 78 F. Housing and Local Infrastructure have Improved Substantially since the Late 1990s 3.28 Housing conditions are an important measure of quality of life, both as ends in themselves and as means toward achieving better living standards. For example, access to sanitation interacts with health care, good nutrition, and water supply to influence the health of individuals. Homes built with more durable building materials provide safer shelter and reduce labor costs for repairs and new construction. 3.29 Vietnam has achieved widespread improvements in the quality of housing and access to infrastructure in recent years. These are evident in recent rounds of the VHLSS, and were also reported in supporting ï¬?eld studies. For example, respondents in the long-run drivers of poverty reduction study (Nguyen and Hoang 2012) describe substantial improvements in rural infrastructure since the early 1990s and increased access to associated social and economic services, markets, and information. These include better road and bridge access for rural communes and remote villages, new irrigation facilities, and a rapid expansion of media services and technologies into rural areas. Associated with this, many households have invested in new types of assets that improve mobility and access to information, including motorbikes, TVs, mobile phones, and even computers in urban areas. These widespread improvements in economic and social infrastructure have resulted from the combined efforts of many government infrastructure investment programs across the different infrastructure sectors, and provide a good foundation for growth of the rural economy and continued reductions in rural poverty in the coming years. 3.30 Although the poor still own fewer durable goods than better-off households, the comparative statistics in table 3.13 indicate substantial increases in durable goods ownership since 1998. For example, in 2010, 51 percent of the poor owned a motorbike compared to 2 percent in 1998; 74 percent owned a TV compared to 30 percent in 1998; and 46 percent owned a rice cooker or electric stove compared to 1 percent in 1998, and 37 percent owned a mobile phone. The extreme poor owned very little in 1998, but by 2010, 40 percent owned a motorbike, 61 percent owned a TV, 28 percent owned a rice cooker or stove, and 24 percent owned a mobile phone. Wider access to transport, TVs, and mobile phones has improved the spread of information and helped the poor to become less socially isolated and more integrated with the wider economy. Table 3.13 Household Ownership Rates of Durables in 1998 and 2010 (Percent) National Poor ExtremePoor 1998 2010 1998 2010 1998 2010 Car 0.2 1.3 0.0 0.0 0.0 0.0 Motorbike 20.3 75.9 2.4 50.9 0.4 39.6 Mobilephone ͲͲ 69.8 ͲͲ 37.1 ͲͲ 24.2 TV 55.7 89.3 30.2 73.6 11.9 61.3 Computer 0.7 16.8 0.0 0.3 0.0 0.4 Refrigeratororfreezer 9.0 42.6 0.0 5.3 0.0 2.2 Airconditioner 0.7 8.2 0.0 0.1 0.0 0.2 Electricfan 68.4 85.2 45.9 65.2 26.3 49.4 Ricecookerorelectricstove 19.3 77.6 1.1 45.6 0.0 28.3 Source: 2010 VHLSS. 3.31 Despite improvements, many of the poor still do not have access to clean water (36 percent of households in the bottom quintile, 14 percent in the second quintile) or adequate sanitation (21 percent of households in the bottom quintile and 8 percent in the second quintile do not have flush or semi-flush toilets). Although Vietnam has done a remarkable job at making electricity widely available (more than 95 percent of households are connected to the grid) and improving the reliability of supply, 11 percent of households in the bottom quintile are still not connected to the electricity grid. Many of the households without access to clean water, adequate sanitation, and electricity are ethnic 79 minorities living in less accessible upland regions of Vietnam (Table 3.14). As described in Chapter 1, these households are deprived not only in income terms, but also in terms of access to public goods and services. Table 3.14 Percentage of Households with Access to Housing and Neighborhood Amenities in 2010, by Quintile Quintile Total 1 2 3 4 5 Tap water 7.5 13.3 21.7 32.8 59.2 26.9 Clean (nontap) water 56.4 72.8 71.2 62.3 39.7 60.5 Flush toilet 12.8 31.2 48.4 67.6 88.7 49.7 Semi-flush toilet 66.0 61.3 46.8 30.7 10.9 43.1 Solid house 12.0 19.7 26.9 34.5 62.5 31.1 Semisolid house 64.9 66.2 64.7 60.7 36.3 58.6 Household with electricity 89.0 97.9 99.4 99.3 99.6 97.0 Source: VHLSS 2010. G. Urban Poverty is Low According to GSO-WB Estimates, and Concentrated in Smaller Cities and Towns 3.32 The poverty rate in urban areas is only 6 percent compared to 27 percent in rural areas. Because only 30 percent of the Vietnamese population lives in urban areas, the urban poor comprise only 8.6 percent of the total poor in Vietnam. 3.33 Although poverty in Vietnam is primarily a rural phenomenon, understanding and addressing urban poverty is increasingly important. Vietnam is urbanizing rapidly; the urban population grew by 3.4 percent per year between 1999 and 2009 compared to an annual population growth rate of only 0.4 percent in rural areas. The urban population is forecast21 to reach 45 percent of the total population by 2020—a major increase over the 30 percent registered in the 2009 Housing and Population Census. In light of this rapid change, it is vital to better understand the factors that influence the living conditions of low-income urban households, including how poverty is distributed across urban areas. 3.34 City size is one important correlate of poverty. The sample size of the 2010 VHLSS is too small to estimate poverty rates for different types of cities. Instead, the poverty mapping methods described in Chapter 4 were used to estimate poverty rates by city size, ranging from very large “special citiesâ€? (for example, Hanoi and Ho Chi Minh City) to small Class 5 cities, which include district towns with 4,000 or fewer inhabitants. Table 3.15 presents poverty statistics by city size ranging from extra- large (that is, Hanoi and Ho Chi Minh City) to extra-small Class 4 and 5 towns. 3.35 Poverty levels decrease with city size; if measured by the 2010 GSO-WB poverty line,22 only 1.9 percent of the population in the largest cities is poor, while the poverty rate in the smallest cities is 11.2 percent. Poverty depth (the poverty gap) and poverty severity (the squared poverty gap) also decrease with city size. The urban poor are overwhelmingly concentrated in small cities and towns; small and extra small cities account for only 43 percent of the urban population but over 70 percent of the urban poor. Conversely, 32 percent of Vietnam’s urban population lives in Hanoi and Ho Chi Minh City, but only 11 percent of the urban poor live in these two cities. 21 Ministry of Construction plan, as part of Decree 10/1998/QD-TTg, 1998. 22 Several of Vietnam’s largest cities have developed their own poverty lines; for instance, Hanoi recently announced a new poverty line of 750,000 VND per person per month for the 2011–2015 Socio-Economic Development Plan, and the poverty line used by Ho Chi Minh City is 1,000,000 VND per person per month. 80 Table 3.15 Poverty by City Size Extra- Extra- Large Medium Small Rural Large Small Special City class Class 1 Class 2 Class 3 Class 4, 5 City Number of cities in category 2 7 14 45 634 Average population (000) 4,075 467 225 86 11 % of total population 9.5 3.8 3.7 4.5 8.1 70.4 % of urban population 32.1 12.9 12.4 15.3 27.3 Poverty rate (%) 1.9 3.8 4.2 5.8 11.2 25.6 Poverty gap (%) 0.4 0.6 0.7 1.1 2.4 6.8 Share of urban poor (%) 11.0 8.8 9.2 5.9 55.0 Source: World Bank estimates. 3.36 Smaller cities can be thought of as more “ruralâ€? than larger cities; urban poverty is concentrated in the more “rural-likeâ€? urban areas. This is consistent with the stylized facts presented earlier in the chapter; the poor in Vietnam overwhelmingly live in rural areas. And indeed, smaller cities are more rural-like than larger cities in more aspects than just population. Table 3.16 provides an overview of housing and local services, also education levels of urban residents, categorized by city size and for rural areas. Although access to electricity is universal across all city types, smaller cities lag the larger ones in most other basic services. Use of gas for cooking is less common, use of ï¬?rewood for cooking is more common, and access to piped water is much less common in smaller cities and towns. In fact, a group of smaller cities report having no access to piped water at all. Similarly, fewer households in small cities have flush toilets and substantial numbers use ï¬?rewood instead of gas for cooking. Smaller cities and towns also lag larger cities in the education level of the household head. Table 3.16 Percent of Households with Speciï¬?c Characteristics, by City Size Extra Large Large Medium Small Extra Small Rural Primary education 20.2 21.8 20.7 23.7 26.2 30.0 Secondary education 19.0 21.0 20.5 20.1 22.6 27.0 Tertiary education 49.7 41.7 46.5 40.1 30.6 14.9 Dwelling walls of solid material 98.2 90.6 92.4 86.7 79.9 69.5 Dwelling walls of semisolid material 1.2 4.5 5.0 8.4 11.9 16.0 Dwelling walls of temporary material 0.6 4.9 2.6 4.9 8.2 14.5 Dwelling roof of solid material 35.1 21.5 25.2 19.5 17.9 13.4 Dwelling roof of semisolid material 6.0 11.5 18.1 20.7 26.6 39.6 Dwelling roof of temporary material 58.8 67.0 56.8 59.8 55.5 47.1 Has flush toilet 99.3 89.6 92.7 82.9 69.6 38.8 Has other kind of toilet 0.5 9.9 5.0 14.6 24.9 50.4 Has no toilet 0.2 0.5 2.3 2.5 5.5 10.9 Drinks water from pipe 74.2 74.3 75.5 57.2 33.6 8.0 Drinks water from well 25.3 15.9 21.3 35.6 52.2 58.3 Drinks water other source 0.6 9.9 3.2 7.2 14.2 33.8 Uses electricity for lighting 99.7 99.7 99.8 99.6 99.0 94.1 Uses electricity for cooking 2.1 1.4 1.1 1.9 1.8 1.5 Uses gas for cooking 89.3 70.7 75.5 66.9 55.6 22.9 Uses ï¬?rewood for cooking 0.7 12.0 7.2 15.7 32.2 64.6 Source: World Bank estimates from 2009 Population Census. Note: Education level is highest attainment of the household head. 81 H. Poverty has Become Less Correlated with Demographic Factors, although Aging is Emerging as an Issue and Child Poverty Remains a Concern 3.37 Compared to the 1990s, demographic factors such as high dependency ratios and female headship have become less linked to poverty. Comparisons between 1999 and 2009 population “pyramidsâ€? for Vietnam (GSO 2010) highlight the sharp reduction in the proportion of children in the population and an increase in the proportion of older adults. Recent qualitative studies (e.g. the long- run drivers of poverty reduction study; Nguyen and Hoang 2012) identify important links between changing household structures and the dynamics of income and well-being. The nationwide family planning campaign, active since the late 1980s, were widely acknowledged at all ï¬?eld sites as having made an important contribution to poverty reduction. Most couples (nearly 80 percent according to the 2010 VHLSS) now have only two children, which helps reduce household spending on basic services like education and health and allows for more “qualityâ€? spending on children. 3.38 The long-run drivers study, with its two-decade reference period, also identiï¬?ed several positive impacts for families that had more children. The Vietnamese economy has been expanding and creating new jobs. Although poor rural households struggled to raise and educate children born in the 1980s and early 1990s, these children are now grown, and many are working in off-farm activities or have migrated to work in urban areas. Rather than being a burden, they contribute to supporting their parents and younger siblings who stay home. Figure 3.12 Population Pyramids for Vietnam: 1999 and 2009 2009 85+ 80-84 1999 75-79 70-74 65-69 60-64 55-59 50-54 45-49 40-44 35-39 30-34 25-29 20-24 15-19 10-14 5-9 0-4 7 6 5 4 3 2 1 0 1 2 3 4 5 6 7 Male % Female Source: GSO 2010. 82 3.39 Female-headed households with children were identiï¬?ed in a number of sites as more vulnerable to and at risk for poverty, in large part because they were dependent primarily on the earnings of the female household head. Many respondents felt that two parents are required to work to support a family in Vietnam. Moreover, men in rural areas are better paid than most women because they take on different (heavier and more dangerous) tasks. Single mothers struggle with the lack of adequate daycare facilities, particularly in rural areas, and may not receive support from extended family. 3.40 Aging is another important source of vulnerability. Vietnam has a high proportion of widows; according to the 2010 VHLSS, 19 percent of households include a widow, and 12.5 percent are currently headed by a widow. The proportion of widows in an age cohort rises sharply with age: 47.6 percent of women aged 66-70 are widowed compared to only 9.7 of men in the same age cohort; 67.6 percent of women aged 76-80 are widowed compared to 24.5 percent of men in the cohort. Participatory Poverty Assessments (PPAs) and recent qualitative studies carried out, for instance, by Oxfam, highlight the vulnerability of households headed by elderly persons, and in particular widows, in part linked to the limited coverage of social insurance and pensions for Vietnam’s aging population (UNFPA 2011). Vulnerability linked to aging is a growing challenge in Vietnam, and additional research on the links between poverty, vulnerability, and aging is needed. Aging and Economies of Scale in Consumption 3.41 New work on aging and household economies of scale and composition was carried for this report to address the concern that conventional poverty proï¬?les based on per-capita consumption tend to underreport poverty among small households (particularly those with only elderly members) and over-report poverty among large households (including those with many children). The study explores different methods to adjust for economies of scale (size) in household welfare (measured in terms of per-capita consumption). While some types of consumption such as food are more directly a function of household size (although young children eat less than adults), other types like electricity and housing are ï¬?xed costs and less directly linked to household size. To adjust for economies of scale, individual welfare is redeï¬?ned as Ü» ‫ כݕ‬ൌ ሺܰሻà°? Where Y is total household expenditures, N is the number of household members, and θ is a scale parameter, which ranges from 1 to 0. When θ = 1, individual welfare is equal to per-capita expenditures (no economies of scale). Engel curve analysis undertaken as part of the study suggest that moderate scale economies hold for Vietnam (that is, θ = .68). 3.42 Table 3.17 presents poverty rates for different demographic groups and different household demographic compositions using conventional per-capita expenditure measures (θ = 1) and moderate (θ = 0.8) and more substantial (θ = 0.6) adjustments for economies of scale. Using conventional measures, we see the standard results: higher poverty than the national average for minority households and for large households with more dependents (two or more children). Households with three or more children (around 10 percent of households in 2010) are more likely to be poor even after adjusting for economies of scale. Child poverty, therefore, remains an important concern. In addition, although low in absolute numbers at present, small households with elderly members emerge as a new group of vulnerable/poor as we adjust progressively for economies of scale. The number of these households is likely to increase as the population ages and Vietnam becomes more urbanized. Ongoing efforts to develop a modern social protection system for Vietnam should keep (single) elderly and widow/widower households in sight as target populations deserving special attention. 83 Table 3.17 Demographic Characteristics and Scale Economies for the Poor Percent Poor % Household Population size θ=1 θ = 0.8 θ = 0.6 All households 100.0 4.5 20.7 21.2 No widow 81.0 4.4 20.3 20.5 With widow 19.0 4.8 23.6 24.1 25.2 Female-headed 24.8 4.0 14.9 16.5 18.2 Male-headed 75.2 4.6 22.6 22.5 23.0 Widow-headed 12.5 4.1 21.5 23.2 26.0 Ethnicity = Kinh 82.2 4.4 13.2 13.4 14.3 Ethnicity = not Kinh 17.8 5.1 62.2 63.0 62.9 Household Composition Single adult 0.7 1.0 4.0 11.3 19.9 Single elderly/widow/ widower 0.7 1.0 14.9 29.6 51.1 2 adults 3.8 2.0 6.8 10.9 16.9 Single parent 0.6 2.0 21.4 26.7 34.5 2 elderly 1.2 2.0 22.3 31.9 46.0 Other 2-member household 1.2 2.0 17.0 23.6 34.3 Nuclear 1 child 6.5 3.0 14.0 16.8 19.3 Nuclear 2 children 14.0 4.0 25.1 26.8 28.3 Nuclear 3+ children 5.3 5.3 47.3 45.1 42.9 Extended family no children 20.4 3.9 8.7 9.7 11.1 Extended family 1 child 19.9 4.8 15.0 14.8 15.1 Extended family 2 children 12.0 5.6 26.2 24.0 22.2 Extended family 3+ children 4.7 7.5 56.3 52.4 46.7 Joint family no elderly 6.0 5.7 29.9 26.4 24.0 Joint family with elderly 3.0 6.0 20.9 18.4 17.0 Source: World Bank estimates. Child Poverty Rates Remain High, and Children Face Multiple Deprivations that could Impact their Long-term Development 23 3.43 Children face a higher risk of poverty than adults, and poverty affects them differently. They have different dietary requirements, for example, and the role of education is vital at this stage of life. A child-speciï¬?c approach to measuring poverty can highlight and emphasize those needs that are especially crucial for children and their development, and enable more effective poverty reduction objectives, strategies, and policies. 3.44 The most common approach to measuring child poverty examines income and/or expenditures at a household level. According to the 1998 VLSS, 47.2 percent—nearly half—of all children lived below the original GSO-WB poverty line. By 2010, this ï¬?gure had fallen to 29.2 percent. Extreme child poverty fell more slowly—from 16.8 percent in 1998 to 12.5 percent in 2010. Furthermore, in households with three or more children, child poverty remains high, as noted in the previous section. But the monetary approach to measuring child poverty reflects only one dimension of well-being, and does not capture the intra-household distribution of resources. The conventional methodology has thus been extended to assess child poverty along additional dimensions. 23 Information in this section was provided by UNICEF/Hanoi. 84 3.45 In 2008, MOLISA and UNICEF developed a Vietnam-speciï¬?c multidimensional poverty measurement approach, based on the Convention on the Rights of the Child. The approach incorporates eight poverty domains, including deprivations in education, nutrition, health, shelter, water and sanitation, child labor, leisure, and social inclusion and protection. Poverty prevalence can be calculated for any one of these domains, and a multidimensional child poverty rate (MDCP) constructed to measure the percentage of children who are poor in at least two domains. This methodology has been applied to the 2006, 2008, and 2010. 3.46 UNICEF’s monetary child poverty rate (MCP) measures the proportion of children living in households whose welfare levels fall below the GSO-WB poverty line. In contrast, the MDCP identiï¬?es the proportion of children suffering from deprivation in at least two of the eight selected domains. The MDCP is systematically higher than the MCP, indicating that around one-third of children living in Vietnam—or an estimated 7 million children—are considered multidimensionally poor, in contrast to around one in ï¬?ve who are poor according to conventional income or expenditure criteria. (ï¬?gure 3.13) Figure 3.13 Monetary and Multidimensional Child Poverty in Vietnam, 2006-10 Source: 2006, 2008, 2010 VHLSS. 3.47 A deeper analysis of the degree of overlap between the MCP and the MDCP reveals that the methods identify different groups of children. While some children are identiï¬?ed as poor according to both methods, there is also a group that is only identiï¬?ed as poor by the multidimensional approach, and likewise for the monetary approach. Using the 2006 VHLSS data, GSO and MOLISA estimate that 18 percent of children are captured exclusively by the MDCP and would not have been considered poor by the MCP. This result underlines the stark difference between child and overall poverty and the importance of a multidimensional measure to complement the standard monetary measurement of poverty. 3.48 Figure 3.14 indicates the disparities that exist among subgroups of the national population. The MDCP declined for both ethnic categories from 2006 to 2010, but children from ethnic minority households are still almost three times more likely to be multi-dimensionally poor than their Kinh/Hoa peers. The ï¬?gures also provide evidence of a signiï¬?cant urban-rural divide; children in rural areas are twice as likely to be multi-dimensionally poor than children in urban areas. While child poverty in rural areas has shown some decline in recent years, the MDCP indicates that urban poverty is rising. 85 Figure 3.14 Multidimensional Child Poverty in Vietnam by Selected Sociodemographic Variables, 2006-2010 Source: 2006, 2008, 2010 VHLSS. 3.49 Figure 3.15 provides a breakdown by domain of the MDCP for 2010. Health, water and sanitation, and leisure are clearly the domains of most concern. These ï¬?gures indicate that more than one in three children aged 2 to 4 (36.7 percent) was not fully immunized and had not visited a health facility in the prior 12 months (health); almost two out of ï¬?ve aged 0 to 15 (39.2 percent) lived in dwellings without hygienic sanitation or safe drinking water (water and sanitation); and more than two out of three children aged 0 to 4 did not have any toys or books (leisure). Figure 3.15 Child Poverty Rate by Domain, 2010 Source: 2006, 2008, 2010 VHLSS. 86 I. Poor Households are Still Vulnerable to Weather Shocks 3.50 Located in one of the earth’s ï¬?ve typhoon centers, Vietnam is prone to natural disasters, including frequent tropical storms and flooding. The 2008 VHLSS collected information on whether households had experienced weather shocks between 2003 and 2008 and the types of shocks. Results are presented in Table 3.18. Households in rural areas are much more likely to experience weather shocks than their urban counterparts, and the poor are more exposed to shocks than the nonpoor. Households in the Central Highlands are more likely than those in any other region to experience droughts, while those in the Central Coastal regions are most likely to experience storms or flooding. (Le, Nguyen, and Phung 2012). Table 3.18 Percent of Households Experiencing Natural Disasters, 2003-08 Other forms Flood, of extreme Drought storm Landslide weather National 6.7 12.9 0.7 15.2 Rural 8.6 15.5 0.9 19.4 Urban 1.8 6.3 0.1 4.3 Red River Delta 2.6 10.3 0.4 28.6 East Northern Mountains 9.4 7.0 1.7 23.0 West Northern Mountains 8.1 14.3 1.3 22.6 North Central Coast 15.8 29.3 1.1 30.3 South Central Coast 7.3 25.9 0.4 7.4 Central Highlands 19.2 10.9 0.4 4.9 Southeast 2.9 5.1 0.3 1.3 Mekong River Delta 3.5 10.2 0.5 1.4 Poor 14.2 17.9 1.2 22.9 NonͲpoor 5.6 12.2 0.6 14.1 Source: 2008 VHLSS. J. Limited Coverage is Provided by Existing Poverty Reduction and Social Protection Programs 3.51 This report focuses on diagnostics. Follow-up work on policy and program implications is planned, including on the design and targeting of social protection and poverty reduction policies and programs. Access to poverty reduction programs and policies is an important aspect of well-being for low-income households. But concerns have been raised about both the targeting and coverage of Vietnam’s existing poverty reduction programs. These issues are examined briefly using information collected in the 2010 VHLSS: each round of the survey includes information on whether households have been formally classiï¬?ed as poor—that is, whether they are on the ofï¬?cial MOLISA poverty list—and thus eligible for beneï¬?ts under existing government programs, most notably the National Targeted Program for Sustainable Poverty Reduction (NTP-SPR). Each round of the VHLSS also includes information on whether the household received program beneï¬?ts. This information can be used to assess coverage and targeting of Vietnam’s poverty programs. 87 3.52 Analysis suggests that coverage is problematic (a substantial number of households that should be on the poverty list are not) but targeting is less of a concern (most households on the list are from the poorest groups). Note, however, that the 2010 VHLSS data were collected before the government implemented the poverty census for the 2011–2015 Socio-Economic Development Plan and used this information to update the ofï¬?cial poverty list. Thus, while the ofï¬?cial poverty rate for 2010 is 14.2 percent, only 10.6 percent of households surveyed in the 2010 VHLSS reported being on the (old) MOLISA poverty list. 3.53 Table 3.19 shows the percentage of households (by expanded expenditure quintile) that reported being classiï¬?ed as poor by commune authorities, and are thus on the ofï¬?cial MOLISA poverty list. 8 percent of individuals in the 2010 VHLSS are classiï¬?ed as extreme poor by the updated GSO-WB poverty line. However, only 52 percent these households said they were on the ofï¬?cial poverty list. Similarly, 20.7 percent of individuals were classiï¬?ed as poor using the updated GSO-WB poverty line, but only 36 percent of these households said they were on the ofï¬?cial poverty list. Thus coverage is low, but leakage of beneï¬?ts to the non-poor is modest; only 12.2 percent of households in the second quintile and 6.3 percent of households in the third quintile said they were on the ofï¬?cial poverty list. Table 3.19 Percentage of Households Ofï¬?cially Classiï¬?ed as Poor, by Expanded Quintile, 2010 2010 Extreme poor 52.0 All poor 36.0 Quintile 2 12.2 Quintile 3 6.3 Quintile 4 2.6 Quintile 5 0.4 3.54 Figure 3.16 describes in greater detail how households on the poverty list are distributed across the welfare distribution. The great majority—nearly 70 percent—of households are also classiï¬?ed as poor using the GSO-WB poverty line. Only 11.5 percent of those ofï¬?cially classiï¬?ed as poor are in the upper half of the welfare distribution. While there is room for improvement, these targeting results are better than in many other countries, where program beneï¬?ts are frequently captured by better-off households and rural elites. This being said, there are clearly problems with program coverage, including coverage of the poorest households. Deeper analysis of coverage and targeting at the regional level indicates that coverage is lower in high-poverty provinces, such as in the North West and North East, and higher in some better-off provinces and urban areas. MOLISA may face pressure to spread program beneï¬?ts more equitably across provinces; given the increasing concentration of the poor in high-poverty regions, this would lead to reduced program coverage. Figure 3.16 Distribution of Population on the Ofï¬?cial Poverty List by Expanded Per-Capita 80 70 60 Percent 50 40 30 20 10 0 88 Expenditure Quintile, 2010 3.55 Table 3.20 looks in detail at the coverage of Vietnam’s various social protection and poverty reduction policies for households classiï¬?ed by expanded expenditure quintile (Nguyen and Vu 2012). Coverage rates are low in general and social insurance programs are not well targeted to the poor. Few households reported receiving vocational training in 2010. Analysis of the coverage of social assistance measures presents a more nuanced picture. Many of the policies included under the National Target Program for Sustainable Poverty Reduction are well-targeted toward the poor (for example, education fee reductions and subsidies, production support, food support) but, consistent with the analysis above, the coverage of these programs is very low. In general, less than a third of the extreme poor were covered by these poverty reduction policies in 2010. Health coverage (free health cards) is better, but beneï¬?ts accrue to households across the welfare distribution. Table 3.20 Coverage of Social Protection and Poverty Reduction Policies by Expanded Quintiles Percentage of People in Households Extreme All Quintile Quintile Quintile Quintile Receiving: Total Poor Poor 2 3 4 5 All transfers and programs 72.6 88.8 77.2 68.1 67.8 70.6 74.5 All social insurance 32.1 11.2 14.3 20.4 28.0 41.1 58.1 Employment subsidy 1.5 1.2 0.8 1.3 1.6 1.8 1.7 Pension 9.2 2.9 2.2 5.4 7.0 11.6 19.5 Having social insurance 26.7 7.5 11.9 15.6 23.4 34.1 50.0 Vocational training 0.1 0.2 0.3 0.2 0.0 0.0 0.0 All social assistance 56.6 87.4 72.0 60.6 54.7 47.9 41.0 Allowances for veterans, merit households 4.0 2.9 2.8 5.2 4.8 4.6 2.6 Allowances for policy households 4.9 11.8 8.8 5.0 4.1 3.3 1.6 Health subsidy allowances 32.7 29.6 31.3 34.3 34.9 29.8 33.7 Education subsidy allowances 8.3 36.0 15.0 7.6 4.0 4.2 2.3 Allowance for recovery from disaster, ï¬?re 4.9 7.4 6.7 7.4 5.7 3.8 1.0 Loan from Vietnam Bank for Social Policies 13.1 33.7 25.6 14.2 10.3 8.6 3.2 Health program 12.0 54.7 29.3 11.9 5.2 2.3 0.7 Education fee reduction and exemption 5.5 25.8 14.9 5.4 1.9 0.7 0.1 Housing program 1.1 4.4 2.9 1.3 0.4 0.2 0.0 Cultivation land for ethnic minorities 0.1 0.1 0.5 0.3 0.0 0.0 0.0 Agricultural extension 7.8 25.5 14.4 7.3 6.1 4.7 1.9 Clean water 1.9 9.1 4.5 2.1 0.6 0.5 0.2 Food supports 5.2 24.9 10.4 5.6 2.0 1.9 0.2 Production support 9.0 27.9 14.5 9.0 8.0 5.6 2.1 Source: Nguyen and Vu 2012. 3.56 Table 3.21 presents similar estimates stratifying for urban versus rural households, also for Kinh majorities and ethnic minorities. Minorities report substantially lower coverage of social insurance programs, albeit greater access to NTP-SPR support, and greater access to social assistance 89 programs more generally. Higher coverage is not surprising given the very high poverty rates among ethnic minorities. Table 3. 21 Coverage of Social Protection and Poverty Reduction Policies by Urban/Rural and Ethnicity Percentage of People in Households Receiving: Total Urban Rural Kinh/Hoa Ethnic Minorities All transfers and programs 72.6 75.3 71.5 70.3 86.1 All social insurance 32.1 56.2 22.0 35.2 14.0 Employment subsidy 1.5 2.0 1.3 1.6 0.8 Pension 9.2 17.9 5.5 10.1 4.0 Having social insurance 26.7 48.9 17.3 29.3 11.0 Vocational training 0.1 0.0 0.1 0.0 0.6 All social assistance 56.6 44.0 61.9 52.2 82.0 Allowances for veterans, merit households 4.0 2.6 4.6 4.2 2.4 Allowances for policy households 4.9 2.3 5.9 4.1 9.4 Health subsidy allowances 32.7 31.9 33.0 33.0 30.7 Education subsidy allowances 8.3 3.5 10.3 4.1 32.7 Allowance for recovery from disaster, Fire 4.9 1.3 6.4 4.8 5.6 Loan from Vietnam Bank for Social Policies 13.1 6.8 15.8 9.7 33.2 Health program 12.0 3.4 15.6 6.4 44.1 Education fee reduction and exemption 5.5 1.8 7.1 3.2 18.8 Housing program 1.1 0.2 1.5 0.4 4.8 Cultivation land for ethnic minorities 0.1 0.0 0.2 0.0 0.8 Agricultural extension 7.8 1.1 10.6 4.7 25.9 Clean water 1.9 0.2 2.7 0.6 9.7 Food supports 5.2 1.4 6.8 2.8 19.1 Production support 9.0 1.4 12.1 6.0 26.2 Source: VHLSS 2010. Notes: Program coverage is the portion of population in each group that receives the transfer. Speciï¬?cally, coverage is (number of individuals in the group who live in a household where at least one member receives the transfer) / (number of individuals in the group). Program coverage is calculated setting as the expansion factor the household expansion factor multiplied by the household size. Source: Nguyen and Vu 2012. 90 Chapter Annexes Annex 3. 1 Overview of Vietnam’s Eight Economic Regions Vietnam’s eight regions include the North East, the North West, the Red River Delta, the North Central Coast, the South Central Coast, the Central Highlands, the South East, and the Mekong River Delta. The North East lies to the north of the Red River Delta. It includes nine provinces, with a population of 8.2 million. The Viet (Kinh) people make up the majority, with the exception of where a number of minority groups reside. Economic development in the region is mainly based on mining, especially coal and various minerals, forestry, perennial crops, vegetables, and tourism at sites like Ba Be lake, Tam Dao, and Ha Long Bay. The North West is in the mountainous northwestern part of the country, bordering China and Laos. It covers six provinces, with a population of 4.2 million. The Thai people make up the majority, but more than 20 other ethnic groups live in North West region. High mountains make communications difï¬?cult. The economy is based on agriculture and industrial crops such as tea and maize. The soil contains various minerals that have not yet been exploited. The Red River Delta’s population is 18.8 million inhabitants, a majority of which (96.2 percent) are Viet people who live in 10 provinces. The region is the economic, political, and cultural center of the country, with the capital Hanoi and the port of Haiphong. The economic engines are industrial production and services. It is also the second- largest rice producer of the country. The North Central Coast has about 10.1 million inhabitants consisting of 25 ethnic groups the majority of which are Viet people. The region is located between the Lao border and a long coastal line. It offers good conditions for overseas trading and tourism. The South Central Coast encompasses eight provinces with a combined population of 8.9 million. The majority of the population are Viet people, but other minorities include Bana, Cham, and RaGlai. Economic development is mainly based on industrial production, especially in Da Nang and Khanh Hoa provinces, and in new industrial centers, namely the Chu Lai economic zone and the Dung Quat economic zone (with the Dung Quat reï¬?nery). The long coastline offers good potential for the development of the marine economy in the region. The Central Highlands region has a population of 5.3 million that is ethnically dominated by the Bana, Coh, Ede, and Giarai. It shares a border with Cambodia and Laos and covers the poorest areas of the country, with sluggish economic development and weak infrastructure. Its fertile soil is good for industrial crops such as coffee, pepper, and rubber. The South East consists of seven provinces and 14.9 million people, of which Viet people are the majority and Cham and Kh’mer are the main ethnic minorities. This region is the most economically developed and also the most urbanized region in Vietnam, with the economic hub Ho Chi Minh City. Other provinces of the region such as Binh Duong, Dong Nai, and Ba Ria-Vung Tau are industrialized and contribute signiï¬?cantly to economic development in the region. The Mekong River Delta includes 13 provinces and 17.3 million people of which Viet is the main group and Hoa and Khmer the minorities. It is the largest rice-growing area and produces half of Vietnam’s total rice production. In addition, the region is home to a large aquacultural industry of catï¬?sh and shrimp and a variety of fruits. 91 References GSO (General Statistics Ofï¬?ce of Vietnam). 1998. “Decision to Approve the Orientations of the Master Plan for the Development of Vietnam’s Urban Centers till 2020.â€? Decree 10/1998/QD-TTg. Hanoi, January 23, 1998. GSO (General Statistics Ofï¬?ce of Vietnam). 2011. “Report on Multidimensional Child Poverty in Vietnam.â€? Prepared jointly by UNICEF and GSO, Hanoi, September. GSO (General Statistics Ofï¬?ce of Vietnam), 2010. “Migration and Urbanization in Vietnam: Patterns, Trends and Differntials.â€? Prepared with support from UNFPA based on 2009 Housing and Population Census, 15% sample. Hanoi. Haughton, J., Nguyen Thi Thanh Loan, and Nguyen Bui Linh. 2010. “Urban Poverty Assessment in Hanoi and HCMC.â€? Joint publication of the UNDP and Vietnam General Statistics Ofï¬?ce, Hanoi. Le, T. D., C. V. Nguyen, and T. D. Phung. 2012. “Natural Shocks, Vulnerability to Poverty in Vietnam.â€? Background paper for the 2012 Vietnam Poverty Assessment, Hanoi. Markussen, T., Finn Tarp, and K. van den Broeck. 2009. “The Forgotten Property Rights: Restrictions on Land Use in Vietnam.â€? Discussion Paper No. 09-21, Department of Economics, University of Copenhagen, Copenhagen. Nguyen, Cuong Viet and Linh Vu. 2012. “Poverty Targeting and Social Protection Strategies in Vietnamâ€?. Background paper prepared for the 2012 Vietnam Poverty Assessment, Hanoi. Nguyen Tam Giang and Hoang Xuan Thanh. 2012. “Long-run Drivers of Poverty Reduction in Vietnam between 1992 and 2011.â€? Background paper prepared for the 2012 Vietnam Poverty Assessment, Hanoi. Ravallion, Martin, and Dominique van de Walle. 2008a. Land in Transition: Reform and Poverty in Vietnam. New York: Palgrave Macmillan; Washington, DC: World Bank. Ravalliion, Martin and Dominiqe van de Walle. 2008b. “Land and Poverty in Reforming East Asiaâ€?. Finance and Development 45(3): 38-41. UNFPA (United Nations Population Fund). 2011. “The Aging Population in Vietnam: Current Status, Prognosis, and Possible Policy Responses.â€? United Nations Population Fund, Hanoi. World Bank. 1999. Vietnam Development Report 2000: Attacking Poverty. Washington DC: World Bank. 92 Chapter 4 Spatial Dimensions of Poverty: 1999 and 2009 Poverty Maps New poverty and inequality maps were created using Vietnam’s 2009 Population and Housing Census in combination with the 2010 Vietnam Household Living Standards Survey. Poverty rates are highest in rural, inland, upland areas, and especially for ethnic minorities. Regions with high poverty are also characterized by high inequality. Poverty is becoming more spatially concentrated over time. 93 A. Introduction 4.1 Household surveys are an important source of information on poverty and living conditions. But there is also widespread demand for information on poverty at a more disaggregated level, such as districts, communes, and villages, than is typically available through national household surveys. Knowing where poor people live is important information for designing effective poverty reduction policies and programs, including targeted poverty reduction programs and policies to promote infrastructure investment and improve access to public goods and services in poor areas. 4.2 Spatial targeting requires reliable information on poverty outcomes at the local level. The Ministry of Labor, Invalids and Social Affairs’ (MOLISA’s) system for determining eligibility for support under the National Target Program for Sustainable Poverty Reduction and other social programs uses a bottom-up process of local surveys combined with village-level discussions to produce poverty estimates at the commune level. But analysis suggests that coverage is uneven and there is a need to improve information on poverty outcomes at the local level (Nguyen et al. 2012). Estimation of poverty for small geographical units (for example, districts and communes) is data intensive. While household surveys like the Vietnam Household Living Standard Survey (VHLSS) collect detailed information on household incomes and expenditures, the sample sizes are too small to yield reliable estimates of poverty at the district or commune level. In contrast, Vietnam’s decennial Population and Housing Censuses do not suffer from small-sample problems; they cover the whole population. Censuses also collect valuable information on individual and household characteristics that provide insights into living standards. But the Census does not collect the detailed information on income or expenditures needed to directly measure poverty. 4.3 Small area estimation techniques (often referred to as poverty mapping methods) have been developed to estimate poverty at the small-area level. One popular approach, introduced by Elbers, Lanjouw, and Lanjouw (2002, 2003), combines household survey data and census data at the unit record level. The approach exploits a census’s coverage of the entire population and the household survey’s detailed information on income and expenditure. First, an expenditure (or income) model is estimated using the household survey data. The dependent variable is expenditure (or income), and the explanatory variables are a set of household and community characteristics that are comparable and that are available in both the household survey and the census. Subsequently, the parameter estimates from the expenditure model are applied to the census data in order to predict expenditure for all households in the population. From there it is a straightforward procedure to estimate poverty measures in small areas such as communes and districts. 4.4 The small area estimation method has been applied in a large number of countries to produce maps not only of poverty measures but also of other welfare indicators (see Bedi, Coudouel, and Simler [2007] for review of applications). In Vietnam, a number of poverty maps have been developed in the past using the Elbers, Lanjouw, and Lanjouw small area estimation method. Minot, Baulch, and Epprecht (2003) combined the 1993 Vietnam Living Standard Survey (VLSS) and the 1994 Agricultural Census to estimate poverty at the local level in rural areas of Vietnam. Minot, Baulch, and Epprecht (2003) constructed a poverty map using the 1998 VLSS and a 33 percent sample of the 1999 Population and Housing Census. Nguyen (2009) applied the 2002 VHLSS to the 33 percent sample of the 1999 Population and Housing Census to produce a poverty map for 2002. Nguyen et al. (2010) further updated the rural poverty map for 2006 using the 2006 VHLSS and the 2006 Rural Agriculture and Fishery Census. 4.5 The General Statistics Ofï¬?ce (GSO) completed a new census of the population in 2009 and a new round of the Vietnam Household Living Standards Survey in 2010. These datasets were used to construct new poverty and inequality maps for Vietnam. This chapter documents these new estimates of poverty at the province and district level 24 of Vietnam, using the updated 2010 poverty line and 24 It is not feasible to produce reliable commune-level poverty estimates using the 15 percent sample of the 2009 Population and Housing Census. These will be done at a later date if GSO makes the unit record data available for the full 2009 census. 94 comprehensive consumption aggregates described in Chapter 2. The estimates are based on the 15-percent sample of the 2009 Population and Housing Census. In addition, poverty is estimated at the provincial and district level for different groups including rural, urban, Kinh/Hoa, and ethnic minority subpopulations. Estimates of provincial- and district-level inequality are also presented, as is a complementary set of “wealth maps,â€? that is, maps that show which provinces and districts account for the wealthiest 15 percent of the Vietnamese population. 4.6 The chapter then turns to an assessment of spatial changes in poverty based on the 1999 and 2009 poverty maps. Although poverty at the national level has fallen substantially over this period, the rate of progress has not been uniform across all localities. Against a background of substantial aggregate growth and poverty reduction, poverty today has become more concentrated in certain regions of the country and within certain socioeconomic groups. Building on these ï¬?ndings, the mapping methodology is used to assess whether the 62 “poorest districtsâ€? identiï¬?ed under Program 30A are indeed among the poorest in Vietnam. Initial ï¬?ndings from policy simulations to assess the gains from spatial targeting in 2010 compared to 1999 are also briefly described. The policy message emerging from both exercises is that spatially targeted poverty reduction policies, including, for example, area-based schemes, will continue to play an important role in Vietnam. B. 2009 Poverty Maps 4.7 Small area estimation methods are used to construct per capita expenditure-based poverty rates for regions, provinces, and districts in Vietnam. Table 4.1 presents regional estimates of the poverty rate and per capita expenditure that are computed directly using per capita expenditure data of the 2010 VHLSS and those estimated from the poverty mapping method. The 2012 VHLSS is representative at the regional level, and the regional poverty rate directly estimated from expenditure data can be thus regarded as the benchmark against which to compare the poverty map estimates. Table 4.1shows that estimates of the poverty rate are quite similar across the two approaches. Table 4.1 Per Capita Expenditure and Poverty Indexes Estimates from the 2010 VHLSS Estimates from Small Area Estimation Method Per Capita Per Capita Expenditure P0 P1 P2 Expenditure P0 P1 P2 (thousand VND) (thousand VND) Northern 10,927.1 44.87 0.1558 0.0701 10,826.4 43.85 0.1483 0.0679 Mountain (250.2) (1.54) (0.0069) (0.0042) (340.9) (1.76) (0.0082) (0.0046) Red River 21,546.0 11.95 0.0265 0.0088 20,515.2 10.65 0.0203 0.0060 Delta (605.6) (0.85) (0.0025) (0.0010) (592.2) (1.02) (0.0025) (0.0009) Central 14,222.6 23.73 0.0635 0.0251 14,002.1 22.48 0.0520 0.0180 Coast (267.3) (1.33) (0.0051) (0.0028) (268.7) (1.05) (0.0031) (0.0013) Central 13,069.0 32.74 0.1149 0.0542 12,931.0 33.29 0.1146 0.0536 Highlands (490.9) (2.75) (0.0128) (0.0077) (351.8) (1.25) (0.0056) (0.0032) South 24,297.4 7.02 0.0172 0.0064 23,350.9 7.07 0.0139 0.0043 East (935.9) (0.96) (0.0036) (0.0018) (844.9) (0.84) (0.0020) (0.0007) Mekong 14,858.2 18.71 0.0425 0.0143 14,497.9 17.45 0.0359 0.0112 River Delta (265.8) (1.10) (0.0033) (0.0015) (280.7) (1.08) (0.0029) (0.0011) Source: Estimation based on the 2009 Vietnam Population and Housing Census and the 2010 VHLSS Note: Standard errors are in parentheses. P0 is the poverty headcount, P1 is the depth of poverty, P2 is the severity of poverty. 95 4.8 Table 4.2 presents estimates using poverty-mapping methods of the mean of per-capita expenditure and the estimated poverty rate, and the absolute number of poor people and the contribution to national poverty for all 63 provinces in Vietnam. Lai Chau, Ha Giang, and Dien Bien are the three poorest provinces, with a poverty rate of more than 70 percent. As expected, Hanoi and Ho Chi Minh City are the least-poor cities, followed by Da Nang, Hai Phong, Quang Ninh, Binh Duong, and Ba Ria-Vung Tau. Similar estimates were made for Vietnam’s 668 districts and, along with provincial estimates, are presented in the ï¬?gures and maps that follow (Nguyen et al. 2012). Table 4.2 Per-Capita Expenditure and Poverty Rate of Provinces Province Number Share Per Capita Poverty Rate Number Share of People in Expenditure (%) of Poor in Total Total (thousand VND) People Poverty Pop. (%) Mean Std. Err. Mean Std. Err. Northern Mountain Ha Giang 724,352 0.84 7422.7 448.1 71.46 2.99 517,586 3.07 Cao Bang 510,884 0.60 9,325.7 515.1 53.11 3.26 271,348 1.61 Bac Kan 294,660 0.34 10,136.1 792.0 45.97 5.32 135,448 0.80 Tuyen Quang 725,467 0.85 11,238.3 917.9 39.95 5.41 289,798 1.72 Lao Cai 613,074 0.71 9,711.5 817.8 56.77 3.90 348,018 2.06 Dien Bien 491,046 0.57 7,625.9 611.7 71.06 3.65 348,953 2.07 Lai Chau 370,134 0.43 6,809.2 465.3 76.41 2.99 282,805 1.68 Son La 1,080,641 1.26 8,326.0 590.3 63.60 4.02 687,305 4.08 Yen Bai 740,904 0.86 10,621.9 794.5 45.33 4.72 335,860 1.99 Hoa Binh 786,963 0.92 10,439.0 675.5 47.31 4.23 372,330 2.21 Thai Nguyen 1,124,785 1.31 14,170.5 1,117.1 21.99 3.42 247,386 1.47 Lang Son 731,886 0.85 10,292.1 715.1 45.69 4.29 334,364 1.98 Bac Giang 1,555,720 1.81 12,823.4 889.4 23.83 4.33 370,722 2.20 Phu Tho 1,313,926 1.53 13,535.9 806.9 23.62 3.20 310,380 1.84 Red River Delta Ha Noi 6,448,837 7.52 29,344.6 1,375.7 4.94 0.89 318,488 1.89 Quang Ninh 1,144,381 1.33 18,538.0 1,243.9 12.12 1.81 138,656 0.82 Vinh Phuc 1,000,838 1.17 15,743.1 869.0 11.99 2.83 119,989 0.71 Bac Ninh 1,024,151 1.19 17,590.4 1,145.4 10.19 2.37 104,327 0.62 Hai Duong 1,703,492 1.99 15,261.3 827.5 14.84 2.73 252,716 1.50 Hai Phong 1,837,302 2.14 20,316.9 1,140.2 7.93 1.62 145,625 0.86 Hung Yên 1,128,702 1.32 16,063.4 812.6 12.78 2.36 144,273 0.86 Thai Bình 1,780,953 2.08 13,578.2 873.7 18.95 3.86 337,435 2.00 Ha Nam 785,057 0.92 14,269.8 1,011.8 16.56 4.07 130,009 0.77 Nam Dinh 1,825,770 2.13 14,866.4 814.6 14.04 2.70 256,321 1.52 Ninh Bình 898,458 1.05 14,955.3 878.3 15.28 3.33 137,314 0.81 Central Coast Thanh Hoa 3,400,238 3.96 13,118.2 474.9 26.48 2.09 900,393 5.34 Nghe An 2,913,054 3.40 13,356.4 576.6 26.74 2.57 778,900 4.62 Ha Tinh 1,227,554 1.43 13,222.9 578.5 21.55 2.97 264,499 1.57 Quang Binh 846,924 0.99 13,847.2 798.8 23.20 4.14 196,475 1.17 Quang Tri 597,984 0.70 12,567.1 621.0 29.55 3.15 176,710 1.05 96 Province Number Share Per Capita Poverty Rate Number Share of People in Expenditure (%) of Poor in Total Total (thousand VND) People Poverty Pop. (%) Mean Std. Err. Mean Std. Err. Thua Thiên Hue 1,087,578 1.27 14,453.7 955.1 19.43 3.03 211,283 1.25 Da Nang 887,068 1.03 23,087.9 1,311.7 2.39 1.05 21,218 0.13 Quang Nam 1,419,502 1.65 12,703.2 528.7 23.47 2.73 333,146 1.98 Quang Ngãi 1,217,159 1.42 12,955.1 573.2 23.65 2.80 287,827 1.71 Binh Dinh 1,485,943 1.73 14,498.9 834.9 16.68 3.16 247,882 1.47 Phú Yên 861,993 1.00 13,377.2 793.1 22.08 3.47 190,348 1.13 Khanh Hoa 1,156,902 1.35 16,778.1 1,244.5 15.51 2.87 179,462 1.06 Ninh Thuan 564,128 0.66 11,626.1 799.1 34.52 4.36 194,759 1.16 Binh Thuan 1,169,450 1.36 13,428.5 693.8 21.44 3.04 250,692 1.49 Central Highlands Kon Tum 430,036 0.50 11,112.5 796.7 47.58 3.37 204,624 1.21 Gia Lai 1,272,791 1.48 11,222.1 439.8 43.34 2.07 551,632 3.27 Dak Lak 1,728,380 2.01 13,445.5 639.8 30.32 2.03 524,104 3.11 Dak Nong 489,441 0.57 11,719.4 500.0 32.50 2.83 159,063 0.94 Lâm Dong 1,186,786 1.38 15,173.1 687.8 21.96 1.97 260,629 1.55 South East Binh Phuoc 874,961 1.02 14,370.4 849.9 17.20 3.58 150,477 0.89 Tay Ninh 1,066,402 1.24 15,459.4 737.6 11.78 2.51 125,615 0.75 Binh Duong 1,482,635 1.73 18,378.5 1,168.5 7.82 2.10 115,901 0.69 Dong Nai 2,483,210 2.89 17,293.1 1,129.8 11.73 2.21 291,223 1.73 Ba Ria - Vung Tau 994,836 1.16 18,704.2 1,336.3 9.97 2.22 99,206 0.59 Ho Chí Minh 7,123,340 8.30 29,431.0 1,342.5 2.94 0.51 209,427 1.24 Mekong River Delta Long An 1,436,913 1.67 16,334.8 703.5 10.97 1.64 157,596 0.93 Tien Giang 1,670,215 1.95 16,578.6 875.9 9.53 2.14 159,215 0.94 Ben Tre 1,254,588 1.46 16,022.7 745.8 10.00 2.00 125,506 0.74 Tra Vinh 1,000,932 1.17 13,507.1 688.8 22.28 3.09 222,988 1.32 Vinh Long 1,028,365 1.20 16,038.5 887.7 11.76 2.26 120,947 0.72 Dong Thap 1,665,420 1.94 13,820.8 605.6 15.58 2.42 259,532 1.54 An Giang 2,144,772 2.50 13,739.4 595.5 18.22 2.50 390,808 2.32 Kiên Giang 1,683,149 1.96 13,057.1 580.7 24.02 2.62 404,319 2.40 Can Tho 1,187,088 1.38 17,911.6 1,029.2 11.70 1.97 138,868 0.82 Hau Giang 756,625 0.88 13,369.3 690.7 19.68 3.41 148,915 0.88 Soc Trang 1,289,441 1.50 12,561.6 604.5 27.28 3.10 351,709 2.09 Bac Liêu 856,249 1.00 12,533.0 670.7 23.30 3.74 199,528 1.18 Ca Mau 1,205,107 1.40 12,456.9 682.5 26.36 3.48 317,609 1.88 Sources: Estimation based on the 2009 Population and Housing Census and the 2010 VHLSS. 97 4.9 Map 4.1 shows the spatial distribution of poverty by provinces and districts in 2009. Poverty rates are highest in the mountainous Northern areas and lowest in the Mekong and Red River Deltas. Disaggregating down to the district level reveals a greater degree of heterogeneity in terms of both pockets of extreme poverty and pockets with particularly low levels of poverty. As discussed later in the chapters, such heterogeneity across sub-national localities translates into gains from spatial targeting of resources for poverty reduction. Map 4.1 Predicted Poverty Rates of Provinces and Districts, 2009 Panel A Province Panel B District Source: Estimation based on the 2009 Population and Housing Census and the 2010 VHLSS. 4.10 Map 4.2 graphs the density of the poor across the country. Because of their large populations, the Mekong and Red River Delta regions still account for a signiï¬?cant number of poor people living in Vietnam. However, as shown below (map 4.10), the picture in 2009 is much less accentuated than at the time of the preceding census, and as such indicates a clear attenuation of the pattern described in earlier studies of poverty in Vietnam (see Minot, Baulch, and Epprecht 2003) where the distribution of the number of poor people was inversely correlated with the spatial distribution of poverty rates. In the late 1990s, the incidence of poverty was highest in more sparsely populated localities and these thus accounted for only a modest fraction of the poor. Today, although poverty rates remain spatially concentrated, the distribution of poor people is more evenly spread across the country. Consequently Vietnam’s poorest communities now account for a larger share of the poor population. 98 Map 4.2 Density of Poverty ( Number of Poor People), 2009 Source: Estimation based on the 2009 Population and Housing Census and the 2010 VHLSS. Inequality is higher in poorer regions 4.11 In Vietnam, there is a positive relationship between poverty and inequality (measured by the Gini index). A more equal distribution in well-being is associated with a lower poverty rate (ï¬?gure 4.1) at the district and province level, while regions with high poverty rates tend to be more unequal. This result is in large part driven by persistent gaps in well-being between ethnic minorities and Kinh majorities (see below, also Chapter 5). However, there remains a great deal of heterogeneity in inequality outcomes, particularly when results are disaggregated to the district level. 99 Figure 4.1 Relationship between the Poverty Rate and Gini Index Panel A: Provinces Panel B: Districts 100 80 80 60 60 Poverty rate Poverty rate 40 40 20 20 0 0 .25 .3 .35 .4 .45 .2 .25 .3 .35 .4 .45 Gini index Gini index Source: Estimation based on the 2009 Population and Housing Census and the 2010 VHLSS. Relationship between Poverty and other Characteristics 4.12 Although Vietnam remains a rural country, urbanization has been accelerating in recent years. About 30 percent of people now reside in urban areas (GSO 2011). Overall, urban areas tend to have lower poverty, and poverty tends to decrease as the urban population share increases (Ravallion, Chen, and Sangraula 2007). Figure 4.2 shows that poverty is negatively correlated with the urban population share at the provincial and district level but, again, with considerable geographic variability. Figure 4.2 Poverty Rate and Proportion of Urban Population Panel A Provinces Panel B Districts 100 80 80 60 60 Poverty rate Poverty rate 40 40 20 20 0 0 0 20 40 60 80 0 20 40 60 80 100 Percentage of urban population Percentage of urban population Source: Estimation based on the 2009 Population and Housing Census and the 2010 VHLSS. 4.13 Despite the ongoing urbanization process, poverty in Vietnam is still largely a rural phenomenon; consistent with the updated poverty proï¬?le presented in Chapter 3, results using the poverty mapping approaches conï¬?rms that 95 percent of the poor live in rural areas. Map 4.3 compares poverty rates in urban and rural areas both at province and district levels. Urban poverty is found to be uniformly lower, and there are substantial differences in poverty rates between urban and rural areas within a given province or district. As discussed in Chapter 3, 70 percent of the urban poor live in smaller cities and towns, rather than Vietnam’s large (special, Class 1 and 2) cities. 100 Map 4.3 Urban and Rural Poverty Rates Panel A Urban Provinces and Districts Panel B Rural Provinces and Districts 4.14 Analysis based on mapping methods also conï¬?rms that poverty has become increasingly concentrated among ethnic minority populations, and there is a strong correlation between the share of ethnic minorities in the population and the poverty rate, at both the province and district levels (ï¬?gure 4.3).25 25 The mapping methodology may underestimate ethnic minority poverty, because it assumes that minorities receive the same returns to their endowments as the Kinh majority. Studies suggest that minorities not only have lower levels of assets, but also receive lower returns on their assets (Baulch and Dat 2012). Estimates presented here and in Chapter 3 provide lower bound estimates of geographically disaggregated poverty levels. 101 Figure 4.3 Poverty Rate and Proportion of Ethnic Minorities Panel A Provinces Panel B Districts 100 80 80 60 60 Poverty rate Poverty rate 40 40 20 20 0 0 0 20 40 60 80 100 0 20 40 60 80 100 Percentage of ethnic minority population Percentage of ethnic minority population Source: Estimation based on the 2009 Population and Housing Census and the 2010 VHLSS. 4.15 Consistent with Chapter 3, Vietnam’s poor are increasingly concentrated in the Northern Mountains and Central Highlands, where there are high proportions of minorities in local populations. Map 4.4 The Poverty Rate of Kinh/Hoa and Ethnic Minority People Panel A Kinh/Hoa People Panel B Ethnic Minority People 102 C. Inequality and Wealth Maps 4.16 We employ two measures of inequality, the Gini index and the ratio of the 90th-to-10th expenditure percentile (a measure of “absoluteâ€? inequality). Provincial results are presented in table 4.3. Provincial- and district-level estimates are presented in the ï¬?gures and maps that follow, and elsewhere (Nguyen et al. 2012). 4.17 Consistent with table 4.3, maps 4.5 and 4.6 illustrate that inequality of expenditures tends to be higher in provinces and districts with low average expenditures. Districts with high poverty rates in the Northern Mountains (these also have a high percentage of minorities) have higher expenditure inequality than other regions. This ï¬?nding is noteworthy in light of the common (often implicit) view in Vietnam that everyone in poor communities is similarly poor. But the ï¬?nding also resonates with other empirical studies of inequality (see Elbers et al. 2004). While there may be poor localities where everyone is similarly poor, more in-depth analysis at the commune level (see targeting simulations described in Annex 4.1) suggests there is still substantial inequality at low levels of geographic disaggregation. Communes in Vietnam typically consist of four to six villages; empirical work suggests that villages tend to be more ethnically and economically homogeneous than communes. Table 4.3 Inequality and Wealth Measures for Provinces Provinces Gini Index Ratio of 90th to 10th Percentage of People Expenditure Percentile in the Richest 20% Mean Std. Err. Mean Std. Err. Mean Std. Err. Northern Mountain Ha Giang 0.374 0.018 4.93 0.35 3.55 0.89 Cao Bang 0.351 0.016 5.10 0.40 4.73 1.14 Bac Kan 0.321 0.018 4.21 0.32 5.31 1.62 Tuyen Quang 0.329 0.021 4.38 0.37 7.54 2.13 Lao Cai 0.397 0.019 6.12 0.53 7.38 1.99 Dien Bien 0.404 0.023 5.82 0.56 4.51 1.29 Lai Chau 0.376 0.017 4.82 0.29 2.99 0.80 Son La 0.360 0.013 4.82 0.27 4.20 1.02 Yen Bai 0.354 0.019 5.20 0.46 7.24 1.91 Hoa Binh 0.345 0.018 4.70 0.35 6.83 1.57 Thai Nguyen 0.308 0.021 4.11 0.42 13.33 3.44 Lang Son 0.325 0.018 4.31 0.32 5.77 1.69 Bac Giang 0.281 0.012 3.60 0.22 8.55 2.29 Phu Tho 0.305 0.013 4.01 0.26 11.30 2.21 Red River Delta Ha Noi 0.382 0.013 6.02 0.40 49.03 2.16 Quang Ninh 0.324 0.015 4.50 0.34 25.76 3.65 Vinh Phuc 0.275 0.012 3.47 0.19 15.81 2.73 Bac Ninh 0.297 0.014 3.85 0.26 22.08 3.55 Hai Duong 0.289 0.013 3.63 0.18 14.49 2.33 Hai Phong 0.322 0.014 4.32 0.28 30.29 3.26 Hung Yên 0.290 0.012 3.68 0.21 16.96 2.49 Thai Bình 0.271 0.014 3.36 0.19 9.40 2.33 Ha Nam 0.273 0.015 3.41 0.23 11.33 2.95 Nam Dinh 0.271 0.014 3.40 0.19 12.97 2.50 Ninh Bình 0.283 0.016 3.57 0.24 13.63 2.55 Central Coast Thanh Hoa 0.316 0.011 3.95 0.15 10.11 1.15 Nghe An 0.328 0.016 4.15 0.21 10.88 1.33 103 Provinces Gini Index Ratio of 90th to 10th Percentage of People Expenditure Percentile in the Richest 20% Mean Std. Err. Mean Std. Err. Mean Std. Err. Quang Binh 0.322 0.017 3.99 0.26 11.75 1.81 Quang Tri 0.323 0.012 4.42 0.25 9.45 1.51 Thua Thiên Hue 0.305 0.016 3.90 0.29 13.22 2.80 Da Nang 0.283 0.011 3.63 0.21 40.11 4.16 Quang Nam 0.281 0.009 3.55 0.17 8.04 1.42 Quang Ngãi 0.290 0.012 3.76 0.20 8.72 1.58 Binh Dinh 0.293 0.015 3.57 0.23 12.42 2.28 Phú Yên 0.297 0.015 3.60 0.22 9.69 2.02 Khanh Hoa 0.325 0.017 4.44 0.35 20.18 3.50 Ninh Thuan 0.313 0.015 4.19 0.30 7.28 1.92 Binh Thuan 0.287 0.012 3.64 0.19 10.02 1.91 Central Highlands Kon Tum 0.414 0.011 7.60 0.47 9.97 2.04 Gia Lai 0.374 0.008 6.18 0.24 8.87 1.16 Dak Lak 0.356 0.011 5.34 0.25 12.50 1.70 Dak Nong 0.307 0.007 4.44 0.15 7.03 1.19 Lâm Dong 0.337 0.010 4.98 0.23 16.80 2.00 South East Binh Phuoc 0.294 0.009 3.53 0.16 11.53 1.91 Tay Ninh 0.287 0.008 3.35 0.14 13.49 1.79 Binh Duong 0.300 0.008 3.62 0.15 22.47 3.65 Dong Nai 0.319 0.014 3.93 0.27 19.47 3.27 Ba Ria - Vung Tau 0.331 0.015 4.14 0.28 23.46 3.70 Ho Chí Minh 0.357 0.009 4.73 0.18 51.17 2.87 Mekong River Delta Long An 0.285 0.009 3.57 0.13 17.55 2.15 Tien Giang 0.277 0.010 3.46 0.14 18.18 2.72 Ben Tre 0.269 0.009 3.36 0.13 16.29 2.33 Tra Vinh 0.294 0.009 3.76 0.15 10.49 1.80 Vinh Long 0.284 0.011 3.58 0.17 16.81 2.66 Dong Thap 0.261 0.007 3.18 0.10 9.59 1.60 An Giang 0.278 0.009 3.39 0.13 9.98 1.49 Kiên Giang 0.293 0.010 3.72 0.14 9.43 1.48 Can Tho 0.328 0.017 4.29 0.33 22.59 2.76 Hau Giang 0.271 0.008 3.39 0.12 9.22 1.70 Soc Trang 0.298 0.011 3.79 0.16 8.44 1.46 Bac Liêu 0.271 0.010 3.32 0.13 7.25 1.56 Ca Mau 0.288 0.012 3.58 0.17 7.76 1.63 Source: Estimation based on the 2009 Population and Housing Census and the 2010 VHLSS. 104 Map 4. 5 Expenditure Gini Indices Panel A Provinces Panel B Districts Source: Estimation based on the 2009 Population and Housing Census and the 2010 VHLSS. Map 4.6 Ratio of the 90th Expenditure Percentile to the 10th Expenditure Percentile Panel A Provinces Panel B Districts Source: Estimation based on the 2009 Population and Housing Census and the 2010 VHLSS. 105 4.18 Map 4.7 shows the locations of the wealthiest 20 percent of households in Vietnam—the so- called middle class and rich. As expected, individuals in the top quintile of the per-capita expenditure distribution are spatially concentrated in the Delta regions, especially in Hanoi and Ho Chi Minh City and in the immediate surrounding areas. Map 4.7 Proportion of People in the Richest Expenditure Quintile (%) Panel A Provinces Panel B Districts Source: Estimation based on the 2009 Population and Housing Census and the 2010 VHLSS. D. The Evolution of Spatial Poverty, 1999 to 2009 4.19 Chapter 1 documents Vietnam’s rapid reduction in poverty since the early 1990s based on a range of poverty lines applied to successive rounds of the VHLSS. However, the VHLSS is only representative at higher levels of spatial aggregation, that is, by region and urban and rural sector. The 2009 poverty maps can be compared with 1999 poverty maps to measure progress at the provincial and districts levels, also to look at changes in the spatial distribution of poverty over time. This section describes spatial patterns of poverty, albeit leaving for future work indepth analysis of the causal mechanisms that underpin these patterns. 4.20 Comparisons of maps 4.8 and 4.9 show that poverty fell most rapidly between 1999 and 2009 in the provinces and districts in the two Deltas. Provinces and districts in the Northern Mountains and Central Highlands experienced substantially lower rates of poverty reduction. District-level maps highlight the variation within provinces, such as in the Central Highlands. 106 Map 4.8 Provincial Poverty Rates Panel A 1999 Panel B 2009 Source: Estimation based on the 2009 Population and Housing Census and the 2010 VHLSS. Note: the 1999 poverty rates are obtained from Minot, Baulch, and Epprecht (2003). Map 4.9 District Poverty Rates Panel A 1999 Panel B 2009 Source: Estimation based on the 2009 Population and Housing Census and the 2010 VHLSS. Note: the 1999 poverty rates are obtained from Minot, Baulch, and Epprecht (2003). 107 4.21 Areas with high incidence of poverty are not necessarily the areas with the highest numbers of poor people. For example, many provinces in the Northern Mountains have a high incidence of poverty but have low population densities, and thus account for a small share of the total poor in Vietnam. Map 4.10 shows the density of the poor across the country in 1999 and 2009. In 1999, the poor were highly concentrated in the Red River Delta and Mekong River Delta; these areas had moderate poverty rates but high population densities. By 2009, however, poverty had become less spatially concentrated. The number of poor decreased remarkably in the two Delta regions, but much less in the Northern Mountains and Central Highlands. Map 4.10 Poverty Density (Number of Poor) Panel A 1999 Panel B 2009 Source: Estimation based on the 2009 Population and Housing Census and the 2010 VHLSS Note: the 1999 poverty rates are obtained from Minot, Baulch, and Epprecht (2003). 4.22 Nearly all provinces and districts experienced a decline in poverty between 1999 and 2009 (ï¬?gure 4.4). But the rate of progress was slower in areas that had very high or very low rates of poverty in 1999, and much faster in areas that started the period in the middle ranges (that is, with a headcount of 25 to 55 percent) (ï¬?gure 4.5). 4.23 Provinces with lower levels of inequality in 1999 also in general achieved a larger reduction in poverty. This largely reflects the growing gap between Kinh and ethnic minority households; high inequality areas typically had a high proportion of ethnic minorities (ï¬?gure 4.6). 108 Figure 4.4 Poverty Rates, 1999 and 2009 Panel A Provinces Panel B Districts 100 100 80 80 The 2009 Poverty rate The 2009 Poverty rate 60 60 40 40 20 20 0 0 0 20 40 60 80 100 0 20 40 60 80 100 The 1999 Poverty rate The 1999 Poverty rate Source: Estimation based on the 2009 Population and Housing Census and the 2010 VHLSS. Note: the 1999 poverty rates are obtained from Minot, Baulch, and Epprecht (2003). Figure 4.5 Poverty Reduction, 1999-2009, and Poverty Rate, 1999 Panel A Provinces Panel B Districts 40 60 Poverty reduction (percentage points) Poverty reduction (percentage points) 30 40 20 20 10 0 0 -20 0 20 40 60 80 0 20 40 60 80 100 The 1999 Poverty rate The 1999 Poverty rate Source: Estimation based on the 2009 Population and Housing Census and the 2010 VHLSS. Note: the 1999 poverty rates are obtained from Minot, Baulch, and Epprecht (2003). Figure 4.6 Change in Poverty, 1999-2009, Compared to the Initial Gini Index, 1999 Panel A Provinces Panel B Districts 60 40 Poverty reduction (percentage points) Poverty reduction (percentage points) 30 40 20 20 10 0 0 -20 .24 .26 .28 .3 .32 .34 .2 .25 .3 .35 .4 The 1999 Gini index The 1999 Gini index Source: Estimation based on the 2009 Population and Housing Census and the 2010 VHLSS Note: the 1999 poverty rates are obtained from Minot, Baulch, and Epprecht (2003). 109 Contribution of the Rural Nonfarm Sector to Poverty Reduction 4.24 A number of factors are responsible for differential rates of progress across provinces and districts in Vietnam, and new work is underway to better understand some of the key drivers of progress over the last decade. Income and employment diversiï¬?cation has been a strong force for growth and poverty reduction. Much attention has been paid to diversiï¬?cation linked to rural-to-urban migration and the role of remittances. In a number of other countries, the expansion of the rural nonfarm sector has been shown to play a beneï¬?cial role in rural development and improving the lives of the poor. The rural nonfarm sector can help absorb excess agricultural labor, provide insurance against agricultural shocks, reduce rural-to-urban migration and, more generally, promote a more equitable distribution of income (see, for example, Ferreira and Lanjouw 2001; Lanjouw and Lanjouw 2000; Oseni and Winters 2009). 4.25 Between 1999 and 2009, a major shift occurred in rural occupations in Vietnam. While in 1999, more than 81 percent of the working population worked in agriculture, by 2009, this has dropped to about 71 percent. The growth of the rural nonfarm sector has been primarily due to expansion in the number of of low-skilled blue collar occupations in the construction, manufacturing, trade, and food preparation sectors. More than half of the increase in fast-growing blue collar nonfarm industries in rural Vietnam is the result of an expanding construction sector (table 4.4). Table 4. 4 Rural Employment and Percent of the Working Population in Sector Description 1999 (%) 2009 (%) Farm All agriculture and forestry and ï¬?shing 81.4 71.2 Nonfarm Self-employed nonfarm, nonfarm wage 18.6 28.8 labor, rural-urban commuters White-collar nonfarm Finance, consulting, science, 5.9 5.8 government, television, healthcare, education, Communist party Blue-collar nonfarm Mining, processing, construction, 12.6 23.0 reparation, trading, food preparation, transportation, cleaning Construction All construction, site preparation, building 1.6 7.5 activities Other blue-collar nonfarm All other blue-collar nonfarm jobs 11.0 15.5 Source: 1999 and 2009 Vietnam Population and Housing Censuses. 4.26 Results from the district-level poverty maps, augmented with data from the 1999 and 2009 Population and Housing Censuses, were used to explore the determinates of rural nonfarm diversiï¬?cation and its contribution to poverty reduction. Proximity to an urban center was found to stimulate rural nonfarm employment, in particular, proximity to large cities (Lanjouw and Marra, 2013). In terms of economic signiï¬?cance, the nonfarm sector of rural districts that are on average 10 kilometers further removed from the nearest city grew 1.63 percentage points more slowly between 1999 and 2009. Although the absolute magnitude may seem small, providing jobs for around 2 percent of the working population for every 10 kilometers of urban proximity is substantial. In addition, analysis suggests that growth in the rural nonfarm sector did indeed contribute to poverty reduction between 1999 and 2009; the poverty headcount was reduced by .0186 (1.86 percent) for a 10- percentage-point increase in the growth in the nonfarm sector. A similar picture emerges when we consider reductions in the severity of poverty (P1), and even the poorest of the poor, captured in reductions in the squared poverty gap (P2), were found to beneï¬?t from an expanding nonfarm sector. These ï¬?ndings stand in contrast to Hoang et al. (2012), whose ï¬?ndings suggest that the very poor do not beneï¬?t from expansion in the rural nonfarm sector because they lack the education and skills to access nonfarm jobs. It is clearly important to look beyond the household level to understand the potential indirect labor market effects of an expanding nonfarm sector. 110 E. In what other Ways can Mapping Methods Inform Policy Design and Evaluation? 4.27 This chapter has documented changing patterns in the spatial distribution of poverty between 1999 and 2009. But what do these imply for the design of policy? A series of simulations were carried out to assess how much the spatial disaggregation provided by poverty maps can help to improve area-based targeting schemes in Vietnam (details provided in Annex 4.1). The simulations are based on a hypothetical transfer scheme that aims to minimize poverty at the national level (focusing on the squared poverty gap, or severity of poverty) by using spatial targeting at different levels of geographic disaggregation, that is, province, district, and commune. The initial results clearly show that in both 1999 and 2009 there were potentially large gains in targeting performance by disaggregating to the local level. An important corollary of these ï¬?ndings is that the beneï¬?ts from spatial targeting become increasingly evident as more and more disaggregated data on poverty are considered. The simulations show that a given impact on poverty can be achieved at considerably less expense with detailed spatial targeting than with a uniform transfer. 4.28 A second key ï¬?nding is that the beneï¬?ts from spatial targeting, at any level of geographic disaggregation, are more clearly evident in 2009 than 1999. This ï¬?nding follows directly from the evidence presented in the previous section on the changing spatial distribution of poverty in Vietnam over time. As Vietnam has prospered, moderately poor households living in relatively well-off areas in 1999 (for example, Red River Delta) were able to cross the poverty line, so that by 2009 such relatively well-off areas no longer contributed as much to overall poverty. Poverty has become more concentrated in poor districts. For policy makers, this is an important ï¬?nding, because it indicates that there may be a stronger rationale for using area-based targeting to reach the poor today than was previously the case. 4.29 But these ï¬?ndings should be viewed as illustrative only. They do not take account of important practical and political considerations such as how the hypothetical transfers would be ï¬?nanced, the costs of administering such a scheme, possible behavioral responses of households, and the possibility of local capture linked to power and influence. The anticipated albeit hypothetical gains from targeting must be juxtaposed against the potential costs and political-economy considerations, and should be scrutinized against other possible policy objectives. In practice, a combination of geographic targeting between villages and means-tested targeting on poor households within villages is likely to be the best way forward for Vietnam. 4.30 We close this chapter with a brief assessment of the targeting performance of Program 30A, one of MOLISA’s newer area-based targeted poverty reduction programs. A welfare ranking of districts is drawn up, based on criteria developed by MOLISA (incorporating information on income, as opposed to expenditures, and other indicators of well-being), and the poorest 62 districts on the list are singled out for speciï¬?c policy interventions (box 4.1). Mapping methods were used to see whether the 62 poorest districts identiï¬?ed by MOLISA’s criteria are also the poorest as measured by the per-capita expenditure criteria underpinning the Vietnam poverty map for 2009. Figure 4.7 illustrates the close correlation between the two approaches; the districts targeted by MOLISA are also among the poorest identiï¬?ed by the independent mapping methodology. Spatial targeting in Vietnam is not only warranted on empirical and conceptual grounds, but appears administratively and logistically feasible, as evidenced by one well-established program. 111 Box 4.1 Overview of Program 30A Program 30A, named after Prime Minister Decision 30A in 2008, is a comprehensive poverty reduction program targeted at 61 (now 62) of the country’s poorest districts through 2020. These districts lie in 20 provinces throughout the country, but most of the districts are located in the northeastern mountainous region. The program focuses on four primary areas: (a) increasing income through production, job creation, and labor exports; (b) improving education standards; (c) improving the quality of local administrators; and (d) investing in infrastructure. Funding commitments for the different components are made in three-year tranches. According to MOLISA, state budget funding for 2009–11 was VND 8.5 trillion. For 2012–15, funding is VND 7.2 trillion. A substantial portion of the funding has gone toward boosting incomes by paying citizens to protect speciï¬?ed areas of forest. However, as with Program 135-II, the vast majority of funds are invested in infrastructure. Thus far, no attempt has been made to evaluate the impact of this program. The 62 districts selected under Program 30A do not receive support directly only through 30A. Their designation as particularly needy districts also makes them eligible for other targeted programs. For example, in order to improve cadre quality, Program 30A is linked to the 600 Deputy Chairman Program, which is run by the Ho Chi Minh Youth League and the Ministry of Home Affairs. This program, initiated in 2011, targeted 600 communes in the 62 districts an additional (trained) person to support the People’s Committee. Figure 4.7 District Poverty: MOLISA compared to Poverty Map Estimates 112 Annex 4. 1 The Spatial Distribution of Poverty and the Gains from Spatial Targeting Chapter 4 documents changing patterns in the spatial distribution of poverty between 1999 and 2009. But what do these patterns imply for the design of policy? A series of simulations was carried out to assess how much the spatial disaggregation provided by poverty maps can help to improve area- based targeting schemes in Vietnam.26 We consider here the distribution of a hypothetical budget to the population of Vietnam. We assume that we have no information about the poverty status of this population other than the geographic location of residence and the level of poverty in each location. As a benchmark case, we make the extreme assumption of no knowledge whatsoever about the spatial distribution of poverty, in which case our given budget is distributed uniformly to the entire population. We set up a series of comparisons to this benchmark, where we assume knowledge about poverty levels in progressively smaller subpopulations. For a given level of disaggregation, we ask how knowledge about poverty outcomes across localities can be incorporated into the design of a transfer scheme so as to improve the overall targeting performance relative to the benchmark case. In light of the observations made above, concerning the evolving spatial distribution of poverty in Vietnam, we ask whether and how our conclusions differ between 1999 and 2009. We consider a transfer scheme that makes use of our knowledge of the spatial distribution of poverty in such a way that poverty is minimized at the national level. We consider the gains from spatial targeting at alternative levels of disaggregation. We focus on the squared poverty gap, a measure of poverty that is particularly sensitive to the distance between a poor person’s income level and the poverty line.27 We specify a poverty line that accords with a poverty rate of around 20 percent nationally, in each respective year, and we consider a modest hypothetical budget that would be insufï¬?cient, in and of itself, to eliminate all poverty, even if it were perfectly targeted at the household level. The results from this exercise show clearly, ï¬?rst, that in both 1999 and 2009, there are potentially large gains in targeting performance from disaggregating to the local level. These beneï¬?ts are clearly seen when we examine the squared poverty gap as our poverty measure of choice. The impact on the headcount rate is, unsurprisingly, more muted, given that we do not “optimizeâ€? our transfer scheme with respect to this poverty measure. An important corollary of these ï¬?ndings is that the beneï¬?ts from spatial targeting become increasingly evident as more and more disaggregated data on poverty are used. We show that a given impact on poverty can be achieved at considerably less expense with detailed spatial targeting than with a uniform transfer. The results from this exercise also show that the beneï¬?ts from spatial targeting, at any level of disaggregation, are more clearly evident in 2009 than in 1999. This ï¬?nding follows directly from the evidence presented in the earlier section on the changing spatial distribution of poverty in Vietnam over time. As Vietnam has prospered, moderately poor households living in relatively well-off areas in 1999 were able to traverse the poverty line, so that by 2009, such relatively well-off areas no longer contributed as much to overall poverty levels. Poverty has become more spatially concentrated. For policy makers, this is an important ï¬?nding, because it indicates that there may be an even stronger rationale for spatial targeting of resources today than was the case a decade earlier. 26 We build on an earlier analysis in Ravallion (1993), who ï¬?nds that spatial disaggregation to the broad regional level in Indonesia, the lowest level at which household survey data provide reliable estimates of poverty, improves targeting, but only to a modest extent. In contrast, Elbers et al. (2007) ï¬?nd that ï¬?ne geographic targeting offers signiï¬?cant beneï¬?ts over broad targeting. 27 We focus on the squared poverty gap because of its appealing properties from both a conceptual and technical point of view. The basic approach explored here would also work for other poverty measures, particularly Foster-Greer-Thorbecke measures with values of parameter α greater than 1. However, with the headcount measure (the FGT measure with α=0) welfare, “optimizationâ€? is not well deï¬?ned and the approach taken here is thus less obviously applicable (see, for example, Ray [1998, 254–55]). 113 Transfer Scheme We postulate that the government has a budget, S, available for distribution and wishes to transfer this budget in such a way as to reduce poverty. We specify a baseline case in which the government is assumed to have no knowledge of who the poor are or where they are located. It is therefore unable to distribute its budget in any manner other than a lump-sum transfer to the entire population of size N. We thus calculate the impact of transferring S/N to the entire population. Kanbur (1987) shows that to minimize poverty summarized by the Foster-Greer-Thorbecke (FGT) class of poverty measures with parameter value α>1, the group with the highest FGT(α-1) should be targeted on the margin.28 Hence, to minimize the squared poverty gap (equal to a poverty measure from the FGT class with α=2), target populations should be ranked by the poverty gap (FGT with α=1) and lump-sum transfers made until the poverty gap of the poorest locality becomes equal to that in the next poorest one, and so on, until the budget is exhausted. Budget and Poverty Lines We assume that the budget available for distribution has been exogenously set. As is intuitively clear, the potential beneï¬?ts from targeting will vary with the overall size of budget. In the limit, as the budget goes to inï¬?nity, there is no need for targeting, as even a uniform transfer will eliminate poverty. As a benchmark, we identify the per-capita consumption value of the 25th percentile of the consumption distribution.29 We scale this consumption value by the total population. Our benchmark budget is set to equal 5 percent of this total value. Gains from targeting also vary with the choice of poverty line. The higher the poverty line, the less need for targeting, as leakage to the nonpoor diminishes to zero. In this study, we select as the benchmark a poverty line that yields a poverty rate of exactly 20 percent in both 1999 and 2009. Simulating the Impact of Uniform Transfers Our policy simulation in the case of uniform transfers is calculated in a very straightforward manner. Budget S is divided by total populationp N. p The resulting transfer a is added to each predicted (r ) ych expenditure in our database, to yield +a. . For each replication r we estimate post-transfer national poverty. The average across the r replications of the estimated posttransfer poverty rates yields our expected poverty rate associated with the benchmark, untargeted lump-sum transfer scheme. This new estimated poverty rate can be compared to the original national-level poverty estimate from the poverty map to gauge the impact of the transfer. Simulating the Impact of “Optimalâ€? Geographic Targeting Simulating the impact of the “optimalâ€? targeting scheme is slightly more complicated. Following Kanbur (1987), we want to equalize the following expression across the poorest locations of a country: z ³ (z  y  a )  (7) Gc ( a c ) c dFc ( y ) , 0 28 Following Foster, Greer and Thorbecke (1984), the FGT class of poverty measures takes the following form: 1 FGT (D ) ( )¦ wi (1  ( xi / z ))D ¦w i where xi is per capita expenditure for those individuals with weight wi who are below the poverty line and zero for those above, z is the poverty line and ¦wi is total population size. takes a value of 0 for the Headcount Index, 1 for the Poverty Gap and 2 for the Squared Poverty Gap. For further discussion, see Ravallion (1994). 29 The consumption distribution is constructed on the basis of the average, across r replications, of household-level predicted per-capita consumption in the population census. 114 which is z times the poverty gap in location c, after every person in the location has received a transfer ac. Fc(y) is the average of the R simulated expenditure distributions of c. The function (x)+ gives the “positive partâ€? of its argument, that is, (x)+=x, if x is positive, otherwise 0. Transfers ac (which must be nonnegative) add up to a given budget S: (8) ¦N a c c c S, where Nc is the population size of location c. After transfers, there is a group of locations all sharing the same (maximum) poverty gap rate in the country. These are the only locations receiving transfers. We describe below how this problem is solved given that we are working with a database of incomes for every household in the 15 percent sample population census. Solving the Problem – “Optimalâ€? Geographic Targeting As described in Elbers et al. (2007), given our interest in minimizing the FGT2, optimal geographic targeting implies that after transfers there is a group of locations all sharing the same (maximum) poverty gap in the country. We determine the level of transfers going to each location by ï¬?rst solving a different problem. Following the notation introduced above, consider the minimum budget S(G) needed to bring down all locations’ poverty gaps to at most level G/z. This amounts to transferring an amount ac (G) to locations with before-transfer poverty gaps above G/z, such that Gc (ac (G )) G. Once we know how to compute S(G), we simply adjust G until S(G) equals the originally given budget for transfers S. To implement this scheme, we must solve the following equation for ac: z ³ (z  y  a )  (A.1) . G c dFc ( y) 0 . In what follows we drop the location index c for ease of notation. Using integration by parts it can be shown that z z a ³ ( z  y  a) dF ( y) ³  (A.2) G (a) F ( y )dy. 0 0 In other words, we need to compute the surface under the expenditure distribution between expenditure levels y=0 and y=z-t, for values of t up to z. Instead of computing G(t) exactly, we use a simple approximation. For this to work we split the interval [0,z] in n equal segments and assume that the “poverty mappingâ€? software has generated expected headcounts for poverty lines z k/n, where k=0, …,n. In other words we have a table of F(z k/n). Using the table we approximate F(y) by linear interpolation for y between table values. With the approximated expenditure distribution, it is easy to solve for transfers as a function of G (see below). In practice, we ï¬?nd that n = 20 gives sufï¬?ciently precise results. The computational set-up is as follows (note that the numbering we adopt means going from z in the direction of 0 rather than the other way around). Deï¬?ne b0=0, and for k=1,...,n, bk as the surface under the (approximated) expenditure distribution between z-kz/n and z-(k-1)z/n, divided by z:30 (A.3) bk 1 F ( z  kz / n)  F ( z  (k  1) z / n) 2n . Let g0 be the original poverty gap, or in terms of the discussion above, g0=G(0)/z. Fork=1,...n, put (A.4) gk g k 1  bk . . 30 Other interpolation schemes are possible. For instance, if the poverty gap is given at table values zk/n, an even simpler computation presents itself. Often, the poverty mapping software will give percentiles of the expenditure distribution. These can also be used for interpolation, but the formulas are more cumbersome, since the percentiles are not equally spaced. 115 The gk are the poverty gaps of the approximated expenditure distribution for successively lower poverty lines z-kz/n. Let ak be the per-capita transfer needed to bring down the poverty line to z-kz/n: (A.5) . ak kz / n . We can now solve for per-capita transfers as a function of the intended poverty gap g 1 0.75 Inequality< 0.5 0.25 0 Ͳ0.25 Source: World Bank estimates from a Shorrock’s decomposition by income source. Note: A relative concentration coefï¬?cient greater than 1 suggests that the income source is inequality increasing, and a value less than 1 suggests that it is inequality decreasing (that is, it is not disproportionately concentrated among richer households). 6.51 Income from the agricultural sector, notably income from crop activities, agricultural wage labor, and livestock and aquaculture, is inequality decreasing. Agricultural wage labor and cropping activities are among the most equalizing income components.52 A rise in the relative concentration coefï¬?cient of agriculture between 2004 and 2010 implies that the extent to which agriculture was equalizing declined over time. Relative to its share of income, however, the contribution of the agricultural sector to overall inequality is low; the agricultural sector (including agricultural wages) contributed approximately 29 percent of total income but accounted for only 15 percent of inequality. In rural areas, agricultural sideline activities were a relatively equalizing source of income in 2004; in 2010 they had become mildly disequalizing, a change that reflects the faster growth in these sources of income among richer rural households. 6.52 The distribution of remittance incomes has become more equalizing over time in both rural and urban areas. In 2004, the share from remittances in the richest quintile was double that in the poorest quintile; by 2010 the shares of remittances were similar. The change in the distributional impact on remittances appears to be predominantly driven by changes in migration patterns among richer households. The quantitative and perceptions studies both suggest a declining role for higher-paid international migration among richer households; the share of remittances coming from international migration has declined from 35 percent of remittances to 30 percent over time. Income from remittances dropped in absolute terms in the top quintile, and the share of international remittances declined from 47 percent of remittance income to 42 percent among the richest 20 percent of the population. 52 Agricultural sidelines activity, notably livestock, aquaculture, and agricultural services, are the least equalizing of all agricultural sources and contribute more to income inequality than crop income. This is corroborated when examining the structure of incomes across income quintiles; sideline activities continue to be an important source of income for both rich and poor households. 160 Figure 6.12 Contribution of different Income Sources to the Gini, 2010 0.45 ShareofGiniCoefficientofinequality Manufacturing Services Pensions 0.4 0.35 0.3 0.25 0.2 0.15 0.1 0.05 0 2004 2010 2004 2010 2004 2010 2004 2010 2004 2010 2004 2010 2004 2010 2004 2010 Agriculture Livestockand Non Agricultural NonͲ Remittances Scholarships Other Acquculture Agricultural Wages Agricultural Businesses Wages Source: 2010 VHLSS. 6.53 Households working in the nonagricultural sector earn more than those working in the agricultural sector, and their incomes have grown at a faster pace. Figure 6.13 shows per-capita incomes conditional upon the sector of employment of the household head. Incomes of households with a household head employed in white-collar occupations in the nonagricultural sector are highest in both urban and rural areas, followed by the incomes of self-employed nonagricultural workers. In rural areas, households whose head works in agriculture have the lowest incomes in both periods and the lowest average growth. Note that the difference between these households and agricultural households was relatively small in 2004 but has grown over time. Figure 6.13 Per-capita Income per Year by Occupation of the Household Head in Rural and Urban Areas, 2004 and 2010 40,000 35,000 ThousandJan.2010VND 30,000 25,000 20,000 15,000 2004 10,000 2010 5,000 0 WhiteͲCollar BlueͲCollar SelfͲEmployed Agriculture WhiteͲCollar BlueͲCollar SelfͲEmployed Employee Employee Employee Employee Rural Urban OccupationofHouseholdHead Source: 2004, 2010 VHLSS. 161 6.54 Education is an important determinant of whether an individual works in the agricultural or nonagricultural sector, and the type of nonagricultural work conducted. The relationship between education and employment type can be readily seen for more recent labor market entrants who have completed their schooling. Figure 6.14 shows the structure of employment for workers aged 25 to 30 in 1998 and 2010. Having an upper secondary education or above is a signiï¬?cant determinant of having nonagricultural employment, and those with a college education are the most likely to be found in more attractive, higher-skilled employment.53 Figure 6.14 Workers Aged 25-30 by Education Level and Job Type 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 1998 2010 1998 2010 1998 2010 1998 2010 PrimaryͲ Rural PrimaryͲ Urban UpperSecondary UpperSecondary andAboveͲ Rural andAboveͲ Urban ManualWork LowerͲSkilled HighSkilled Source: 2010 VHLSS. Note: High-skilled workers are professional/ofï¬?ce workers. These positions are usually classiï¬?ed as white-collar work. Lower-skilled workers are workers in the service sector, sales, machine operators, and skilled manual/handicraft workers. Manual workers include agricultural laborers and unskilled manual workers. 6.55 Returns to education have increased over the 2000s, with substantially larger increases for workers in urban areas (ï¬?gure 6.15). Empirical work carried out for this report ï¬?nds evidence of rising returns to education in the wage labor market during the 2000s; for non-agricultural jobs, the hourly wage return to a year of schooling increased from 5.3 percent in 2004 to 5.8 percent in 2010. The labor income return to education (based on total earnings) is greater than the wage return (based on 53 Those with upper secondary education and above are still likely to be found doing unskilled work in rural areas, either in the agricultural sector or as an unskilled manual laborer in the nonagricultural sector. In the qualitative assessment, focus groups in rural areas discussed instances where individuals who had obtained higher education were unable to ï¬?nd skilled work (either lower- or higher-skilled work), and hence returned to farming. They attributed this worrying observation to differences in the quality of education between urban and rural areas, and to students choosing ï¬?elds of study, such as pedagogy, for which labor market demand is limited. 162 hourly earnings) to education, since more-educated individuals work longer hours in the wage labor market than less-educated individuals. An additional year of education is estimated to have raised labor incomes by 9.7 percent in 2010 compared to a labor income return of 8.9 percent in 2004. Returns to education are higher for workers in urban areas than in rural areas and have risen faster over time. In urban areas, an additional year of schooling was associated with a 7.6 percent increase in hourly wages, while in rural areas it was associated with a 4.1 percent increase. Within rural areas, returns to education among ethnic minorities are lower than those accrued by the majority, and appeared to decline between 2004 and 2010. The lower returns for ethnic minority workers reflect the fact that minorities tend to work in lower-paid occupations, including wage employment in the agriculture sector. Figure 6.15 Hourly Wage and Labor Income Returns to Schooling 12% g 10% 8% 6% y 4% 2% 0% 2004 2010 2004 2010 2004 2010 2004 2010 2004 2010 Ͳ2% Ͳ4% National Urban Rural Rural,Minority Rural,Majority WageReturns LaborIncomeReturns Source: 2004, 2010 VHLSS. 6.56 The increase in returns over time has increased the gap between the wages and incomes of individuals with higher and lower levels of education (World Bank staff estimates).54 Since education is unequally distributed across the working-age population and adjusts only slowly over time, some people will beneï¬?t more from nonagricultural growth and higher returns to education than others. Therefore, nonagricultural growth and rising returns to education are associated with rising inequality in income. 6.57 The link between education and rising income inequality can be explored through examining the relative gap between the incomes of more and less educated households, which rose between 2004 and 2010. In 2004, households with at least one working-age individual with a college education earned 1.3 times more income than those with an upper-secondary-educated individual, and 2.5 times more than households with no education. By 2010, the college-educated households earned 54 There has been a substantial rise in the returns to education over time, although the majority of this rise has been driven by urban areas. Assessments of the average wage earned by individuals with different levels of education ï¬?nd low rates of return in the early 1990s. In 1993, the return to education using a basic Mincerian earnings equation was found to be approximately 4 percent (Gallup 2002; Glewwe and Patrios 1999). Returns in the 1990s were low by international standards, although they were similar to rates of returns found in China in the early 1990s (Psacharopoulos 1994). 163 1.7 and 3 times more, respectively. Figure 6.16 shows income in urban and rural households, by education level. More educated households earn more than less educated households, and the incomes of the most educated households grew faster than all other education categories between 2004 and 2010 in both rural and urban areas. Although urban households continued to earn more in every education category in 2010, as they did in 2004, the ratio of incomes of rural households to urban households at education levels above lower secondary has fallen over time. This suggests that the decline in mean incomes between rural and urban areas is due to the relatively richer, more educated individuals in rural areas catching up to their urban peers, rather than to catch-up at the bottom end of the income distributions. Figure 6.16 Per-capita Income per Year by Education of most Educated Working-age Household Member, Urban and Rural Households, 2004 and 2010 35,000 PerͲcapitaIncomeinThousandJan.2010VND 30,000 25,000 20,000 15,000 10,000 5,000 0 Noeducation Primary Lowersecondary Uppersecondary Collegeor vocationalTraining Urban2004 Urban2010 Rural2004 Rural2010 Source: 2010 VHLSS. E. Inequalities in Opportunities that Perpetuate Income Differences across Generations 6.58 The analysis of opportunities is predominantly focused on education. This choice of focus was driven in part by the perceptions study; education and employment were central concerns in many focus groups. This focus was also motivated by the empirics, which suggest an increasingly important role of education as a determinant of income inequality. It is recognized that the focus on education comes at the exclusion of other important opportunities that drive inequality, however, in particular access to health care and basic public services. 55 6.59 Growth in the demand for educated labor and increases in the return to education in urban areas imply that education is an increasingly important—and dividing—asset in Vietnam. Education levels in the labor market and in households are rising as more educated younger cohorts join the labor market and less educated older cohorts retire. However, the stock of education among the 55 For an excellent discussion on inequalities in these other important dimensions, see the background paper for the 2008– 2010 Vietnam Poverty Assessment by Hoang et al. (2010). 164 working-age population changes slowly in response to changing returns; therefore, initial differences in education endowments can translate into large differences in incomes as returns to education rise and the demand for skilled labor in the nonagricultural sector grows. 6.60 Whether income inequality and disparities will perpetuate across generations depends on whether investments in human capital among younger generations are responding to changes in income generation opportunities, or whether they reflect inequalities in opportunities linked to their circumstances of birth, such as where a child was born, the characteristics of their parents, or ethnicity. The evidence suggests that inequalities in education are likely to be transmitted to future generations, implying that deprivations continue to be perpetuated across generations and require decisive action. 6.61 The transmission of deprivations across generations was reflected in multiple focus group discussions, where groups highlighted that children born to poorer households were likely to drop out of school earlier than those born to richer households, and to work in less-skilled occupations. Many participants recognized that gaps in education enrolment have narrowed between better-off and worse-off households at lower levels of education, but suggest that gaps remain at higher levels of education, and quality gaps arise at all ages, implying that poverty perpetuates across generations. As one member of a lower-educated migrant group expressed it, “Education is an important cause of inequality. Without education, I work as an unskilled worker and send my children to lower-quality schools. With a good education and income, I could send my children to good schools. It is a vicious cycle.â€? (lower-educated migrant group, Ho Chi Minh City) 6.62 Substantial progress has been made in equalizing enrolments and completion rates at the primary level. Between 1998 and 2010, differences in enrolments at the primary and secondary level have narrowed across the rich and the poor and in rural and urban areas, as can be seen in ï¬?gure 6.17. At the primary level, educational enrolment is close to universal for all groups, although important differences remain between ethnic minorities and the majority, and across minority groups, as discussed in Chapter 5. Figure 6.17 Ratio of Enrolments in Primary, Lower Secondary, and Upper Secondary School by Various Groups, 1998 and 2010 Source: 1998 VLSS, 2010 VHLSS. 165 6.63 Educational investment continues to be unequally distributed at higher levels, an inequality that will feed into inequalities in outcomes later in life. Gaps in enrolment at an upper secondary level continued to be high in 2010, and a child’s background plays a large role in determining their educational attainment at a higher level. Upper secondary enrolment for children in rural areas is still only 70 percent of enrolment rates for children in urban areas, and ethnic minority enrolment is only half that of the majority. Only four poor students are enrolled in upper secondary school for every 10 richer students enrolled. Since many of those richer students will continue on to college or university, the ï¬?nal education difference between students residing in the top and bottom income quintiles will be wider than it is for upper secondary education. 6.64 The characteristics of a child’s parents and household wealth continue to be signiï¬?cant predictors of whether a child is enrolled in lower secondary or upper secondary school, although their impact on enrolment diminished between 1998 and 2010. Educational enrolment at the secondary level is affected by income, which can be considered a short-term liquidity constraint, and is linked to longer-term, or permanent, factors such as parental education (World Bank 2011).56 The evidence also suggests that the impact of income on education decisions is twice as large for ethnic minorities as for the Kinh/Hoa majority (World Bank 2011). 6.65 Beyond family background, the quality of schooling is an important factor that influences the skills that a child acquires in school. At the primary level, the characteristics of teachers, schools, and classrooms are statistically signiï¬?cantly related to student achievement in math and science, and these inputs have been found to be unequally distributed across schools in Vietnam (World Bank 2011). 6.66 Evidence from the Young Lives data suggests that children from poorer households perform worse on math tests prior to entering primary school, and continue to perform worse than children from richer households throughout primary and lower secondary school. Figure 6.18 shows the average rank of children in math tests at ages 5, 8, 12, and 15 by household wealth quantile. At age 5, prior to entering school, the average math scores of children increase with wealth quantiles, so that children from the poorest 25 percent of households have lower scores, on average, than children from other wealth quantiles. 6.67 Most worrisome, the circumstances that a child is born into appear to be a more important determinant of success than a child’s potential when entering school. Figure 6.19 shows the score trajectories of children who had math scores in the top and bottom 20 percent at age 5. Trajectories are divided by the wealth status of their households at age 8. We can see that high-scoring children from poor households perform poorly relative to their high-scoring peers from rich households. Similarly low-scoring children from rich households make more substantial gains in their scores over time than low-scoring children from poorer households. 6.68 The perceptions study indicates that parents perceive signiï¬?cant variation in the quality of education across rural and urban areas at all levels of education. A frequently raised concern is that teachers in rural areas at higher levels appeared to be less qualiï¬?ed than those in urban areas, and that the poor were unable to afford to send their children to the same quality schools as rich children. 56 Income is also likely to be related to unobserved correlates such as local returns to education, which are also likely to positively influence education decisions. Furthermore, income is unlikely to reflect a true liquidity constraint since households also have access to savings and formal and informal credit institutions. 166 Figure 6.18 Average Rank in Math Test, by Wealth Quantile, at Ages 5, 8, and 15 Years 0.7 AverageRankinMathTest 0.65 0.6 0.55 BottomWealthQuantile 0.5 0.45 2ndQuantile 0.4 3rdQuantile 0.35 TopWealthQuantile 0.3 5 8 12 15 Age Source: World Bank staff estimates using Young Lives data. Figure 6.19 Average Rank in Math Test, by Initial Test Score and Wealth 1 BetterOffChildrenwith 0.9 AverageRankinMathTest HighTestScores(n=126) 0.8 0.7 0.6 PoorerChildrenwith 0.5 HighTestScores(n=44) 0.4 0.3 BetterOffChildrenwith 0.2 LowTestScores(n=44) 0.1 0 PoorerChildrenwith 5 8 LowTestScores(n=108) Age Source: World Bank staff estimates using Young Lives data. 6.69 A striking perceived inequality in education quality is found between richer and poorer households in urban areas, where the rich children can go to high-quality schools, attend extra classes, and pay private tuition, including for English and computer courses. Meanwhile, poor children attend average schools with few extra classes. In the past, there was little differentiation in the quality of education services, but now such differentiation in the cities in Vietnam is perceived to be very big, and the rich are viewed as having the capability to invest in better-quality education for their children. For example, a student from Ward 26 in Ho Chi Minh City reports that: “As early as the child is still in preschool, the rich families will start to seek their way into good primary schools, the poorer families just want their children to be literate, so they don’t care about which school their children are going to. Previously, there was a small number of international schools for the rich families to choose from, both rich and poor students would attend the same school, now there are more schools providing a wider range of services, the rich-poor gap also gets widened.â€? 167 6.70 Unequal education quality is perceived to start from an early age, with children from poorer households sending their children to lower-quality kindergartens. Some poorer households in An San ward, Tam Ky city, Quang Nam, reported not being able to afford to send their children to kindergarten. Others who were able to do so expressed concerns about quality differences between the preschools attended by their children and those attended by children from wealthier backgrounds: “The disparity can be found right from the preschool level. The poor households, who try their best, can send their kids to school[s] that cost 500,000 VND per month. The better-off households, on the contrary, send their kids to key schools that ask for fees of 700,000 to 900,000 VND per month. The diet and care services among these schools are different.â€? 6.71 Although empirical evidence on quality differences at higher levels of education is limited, looking at the composition of education expenditures across households can give insight into why quality differences may emerge. As noted in Chapter 1, spending on inputs like extra courses is substantially higher among richer and urban households at the lower and upper secondary level, and the amount spent on these courses has increased over time among the richest households. These trends are strongest in urban areas, but can also be seen in rural areas. If children from richer households are able to beneï¬?t from extracurricular activities and additional training through tutoring and foreign language studies, they are likely to receive a higher-quality and more rounded education than children from poorer households. 6.72 There is evidence of inequality of opportunities in Vietnam beyond education, and that circumstances beyond the control of an individual contribute substantially to these inequalities in access to basic services. Attitudes toward inequality, and whether it is perceived as unjust, unnecessary, and undesirable, depend on the processes that form it. An important factor is whether inequalities are perceived to be driven by differences in factors for which the individual can be held accountable (“effortsâ€?) or are due to circumstances that fall beyond an individual’s responsibility (“circumstancesâ€?) (Roemer 1998). Factors beyond an individual’s control that lead them to have different levels of well-being can thus be considered inequalities of opportunity (Paes de Barros et al. 2009). 6.73 The Human Opportunity Index (HOI), developed by Paes de Barros et al. (2009), captures inequality of opportunity by examining the extent to which the circumstances that children are born into, such as gender, parental education, and ethnicity, affect the likelihood of their access to basic building blocks of human capital, such as education and health services. The index captures two moments of access to basic services. It captures absolute levels of access, and then calculates how different the access rate is across gender, location, parental background, income, and other indicators capturing circumstances. The degree of inequality is measured by the D-index, which captures the dissimilarity in access rates due to differences in circumstance. Differences is the degree of inequality of opportunity and can be interpreted as the fraction of a given inequality that needs to be redistributed in order to achieve equality. The D-index measure of inequality of opportunity is used to scale down the average national access rate of a service to the given HOI. 6.74 The HOI in Vietnam was examined between 2004 and 2010 in a background paper for the poverty assessment led by researchers from the Vietnamese Academy of Social Sciences, with inputs from the World Bank (VASS 2012). Opportunities for access to basic building blocks were examined in three domains—education, health, and housing infrastructure—and the paper investigates whether access to these basic foundational blocks is evenly spread across children in the population or circumscribed by inherent characteristics beyond an individual’s control. The circumstances examined include a number of individual and household characteristics, including gender, parental education and well-being (expenditures), location, and ethnicity. 168 6.75 In international comparisons with countries in Africa and Latin America and the Caribbean, Vietnam fares well on some dimensions, such as access to electricity and school attendance, and poorer on others, such as access to piped water and flush toilets. Speciï¬?cally, the HOI for school attendance is higher than that of most African countries and several countries in the Latin America and the Caribbean region, while the HOI for access to electricity is higher than all African countries and only slightly lower than most Latin American and Caribbean countries. The international comparison is, however, less favorable in other dimensions. Vietnam’s HOI for access to piped water is higher than only some African countries, and it is lower than all Latin American and Caribbean countries. The HOI for flush toilets is in the middle of the whole range of African and Latin American and Caribbean countries. However, Vietnam falls considerably behind top-performing countries in both of these basic services. 6.76 Although equality of access is high for education “quantityâ€? in 2010, the HOI suggests that the quality of education is more divergent across the population. Among 7-to-11 year-olds, both the coverage rate and HOI are high, suggesting that there are low inequalities in accessing primary education, and access overall is high. At the lower secondary level, however, although the coverage rate is high, the evidence suggests that there are some inequalities in access. The education of the household head is the most important characteristic determining whether a child attends lower school between ages 12 and 15, followed by household well-being (expenditure). These two circumstances account for more than 50 percent of the dissimilarity. Although ethnic minorities have lower education outcomes, ethnicity alone plays a smaller role than well-being and education of the household head, a ï¬?nding that suggests that differences in other circumstances contribute substantially to and reinforce inequalities across ethnicities. 6.77 The quality of schooling received by a child is measured by his or her ability to advance independently to lower secondary school without help when he or she is in the last grade of primary school. Only 62 percent of pupils in grade 5 would be able to advance to the lower secondary school without help. The considerable difference between the HOI of the quantity and quality dimensions of education suggests that a greater emphasis needs to be placed on raising quality in the education system, in general, and primary school, in particular. Household well-being and education are the two most important circumstances determining the quality of education received. 6.78 Although the HOI for access to electricity and improved water sources is high, the coverage of access to improved sanitation facilities is lower and less evenly distributed than the other infrastructure measures. Although there was signiï¬?cant progress during 2002–08, and further improvement in 2010, the coverage rate was approximately 64 percent in 2010, suggesting that more can be done to improve access to this basic service.57 Furthermore, a substantial gap between the coverage rate and HOI indicates a remarkable inequality in access to this service. The region a household is located in plays the biggest role in determining access to clean water and sanitation, followed by a household’s well-being, ethnicity, and the education of the household head. 6.79 The HOI is high for some indicators of health and low for others. Notably, the index suggests that Vietnam is doing well on the fraction of women receiving prenatal care, assistance at delivery, and child immunization against measles; 92 percent of children aged 1 to 5 were vaccinated against measles in 2010. Immunization against polio, however, displays a lower coverage rate. 57 Due to changes in the sampling frame between 2008 and 2010, it is not possible to compare the progress achieved between 2002 and 2008 to that achieved between 2008 and 2010. Therefore, access to improved sanitation facilities is analyzed separately in 2010. 169 6.80 Household well-being is a leading determinant for opportunities in the health domain. Figure 6.20 shows the relative importance of circumstances for key health indicators in 2010, decomposed into the fraction attributable to different circumstances. Ethnicity is the most important circumstance for access to care for mothers, and accounted for one-quarter of dissimilarities in receiving prenatal care and assistance at delivery. Among children, household well-being, region of residence, and the education of the household head account for 65 percent or more of the dissimilarity in opportunities. Figure 6.20 Relative Importance of Circumstances for Health Opportunities 100 23 26 27 31 24 80 18 19 Percent 60 23 21 30 16 12 40 21 17 19 25 27 20 15 12 12 12 9 9 8 8 0 Receivingantenatal Skilledattendantat Immunization Complete Completion carebyskilled delivery againstMeasles Immunization Immunization person Opportunity againstPolio Childgender HHcomposition Location Ethnicity Source: VASS 2012. 6.81 An analysis of the HOI at the region level suggests that there is substantial heterogeneity across regions with regard to access to improved sanitation facilities in both the initial year examined, 2002, and in improvements between 2002 and 2008, and in 2010. The South East shows the largest and most stable increase, while the North West had a very low HOI in 2002, which improved in a slow and unstable manner. F. Inequalities in Connections, Voice, and Influence 6.82 Qualitative and quantitative evidence suggests that inequality in Vietnam reflects processes that may be more socially and economically damaging, such as inequalities in social and political capital, which manifest themselves through inequalities driven by influence, connections, and uneven voice. Inequalities of these forms were raised in many focus groups, urban and rural, rich and poor alike, as an important driver of inequality, and were identiï¬?ed as having risen in recent years.58 6.83 Corruption has been recognized in previous work as a systemic problem in Vietnam, and the qualitative evidence reflects many of the issues raised in previous analyses of corruption and transparency in the country (Anderson et al. 2009; Cecodes, FR, CPP, and UNDP 2012; World Bank 2010; World Bank, Embassy of Sweden, and Embassy of Denmark 2011), but does so through the lens of rich-poor differences and inequality, therefore shedding light on how inequalities in socioeconomic outcomes interact with, are magniï¬?ed by, and are perpetuated by inequalities in power and connections. Inequality of treatment by public authorities was raised with respect to a 58 Quantitative evidence suggests mixed trends in reported corruption, as would be expected (World Bank 2010). Surveys of ï¬?rms suggest that corruption is less of an obstacle for their operations, but the same surveys show that the magnitude of bribes, as a percentage of revenues, has not declined. Individual reports from household surveys suggest that, while citizens do not ï¬?nd that corruption has worsened, they do not report that the situation has improved (World Bank 2010). 170 number of things, from land conversion practices that favor investors over landholders to the uneven quality of public service delivery in hospitals and public notaries that led to frustration among poorer and less-well-connected individuals. 6.84 Rural respondents were concerned about increasing disparities in employment opportunities in the public sector, and cited the need to pay bribes or have connections to obtain jobs as teachers, doctors, in state-owned enterprises, and as public ofï¬?cials.59 These concerns were widespread and expressed by individuals from all backgrounds, including commune ofï¬?cials. Evidence from the nationally representative Provincial Administrative Procedural Index study suggests that 29 percent of individuals agree that bribes are required to obtain jobs in the public sector, and nearly half of all respondents believe that connections are important in obtaining various types of state employment (Cecodes, FR, CPP, and UNDP 2012). Moreover, these views are shared in urban and rural areas. 6.85 Unfair recruitment mechanisms in the public sector are linked to concerns about youth unemployment following substantial investment in higher levels of education. Focus groups of youth, in particular, voiced frustration with perceived procedural inequalities that affected their ability to translate their education into good jobs, such as the unfair roles of power and relationships to get public sector employment. In their words: “Money is not enough. Money without connections can’t get you a job in the public sector. I know some cases where the workers quit their job in pursuit of higher education but after graduation, they returned to work in the previous position as if they had never attended such courses.â€? (better-off group, Cam Hung commune, Hai Duong) “In my place, there are some guys who have to work as simple workers after completing university just because their families do not have 50 billion VND to 70 billion VND to bribe their way into an agency just to work as an administrative assistant. Many with poor academic performance somehow passed university entrance exams and were placed [in] a job after graduation. This is irrational but unlikely to abate in the future.â€? (senior citizen, Cam Hung commune, Hai Duong) 6.86 In peri-urban areas undergoing conversion of agricultural land into nonagricultural land for industrial zones, inequalities in outcomes related to land were seen as an unfair source of disparities, whereby people with connections and information gain from land speculation while those without are unable to convert their land into income. Focus group participants perceived that the current land conversion policies and processes favored commercial investors, and that local landowners did not secure their rights to proper compensation and resettlement, effective vocational training, occupation replacement, and employment generation. As one group expressed it: “Many owners of bogus projects have exploited loopholes under Decree 64 to appropriate land from local farmers with false claims of using it [the land] for public utilities.â€? (poor group, Me Tri, Ha Noi) 6.87 Focus group participants raised concerns suggesting that corruption in land management is regressive since it involves a transfer of land at lower-than-market prices from poorer households to relatively well-off investors. People with connections and access to information were reported 59 In 2010, the public sector (including state-owned enterprises and civil servants) accounted for only 4 percent of nonagricultural work and 15 percent of wage or salaried jobs, but for 52 percent of high-skilled jobs in rural areas. In urban areas, the data suggest that public sector jobs account for 9 percent of all nonagricultural work, 28 percent of wage or salaried jobs, and 42 percent of high-skilled jobs. Ho Chi Minh City stands out as having the highest private sector opportunities in the nonagricultural sector, while the North West mountains regions have the lowest private sector opportunities for highly skilled wage or salaried work. 171 to have made substantial proï¬?ts from land speculation and trade, while those who lost land in the process have to struggle for their basic necessities after land conversion. A key concern here is speculative behavior, wherein land was bought at a low price and resold shortly after at a higher price, as reported by youth in Me Tri, Ha Noi: “People in [the] land sector they know in advance the information so that they can advise others to buy land when the price is low and then sell it out at much higher prices.â€? 6.88 Unequal access to public services was another major source of concern across focus groups, with differences in treatment noted between those who “do politicsâ€? and ordinary people. Concerns about access to quality public services are widespread and cover multiple forms of public services, from lengthy administrative procedures such as registering a marriage to the length of wait and quality of treatment given by doctors and hospital staff in public hospitals. In addition, concerns were raised in multiple settings regarding who receives the beneï¬?ts from public social assistance programs targeted at the poor. 6.89 It is perceived that those who have been ofï¬?cials of government agencies are often given priority when they go through administrative procedures. In particular, a commonly voiced concern was that richer people use bribes to better access education or health care services. Participants expressed concern over the predominance of valuing money over traditional ethical values on the part of employees in public services as outcome inequalities widen. As one person put it: “For example, when it comes to doing paperwork at the ward people’s committee, if you had been with the state before you retired, you will still be given priority over other ordinary people. Even if you have to queue up, you will still be quicker to have the paperwork done than the others. Likewise in hospital, if you are an average person, you will not get the same treatment as the privileged.â€? (youth group, Ho Chi Minh City) 6.90 The use of power, connections, and corrupt means to get ahead in life and acquire better public services and employment opportunities was seen as unacceptable by many focus group participants, and was a key source of frustration. The evidence suggests that whether inequality in outcomes is viewed as acceptable or not appears to depend more on the process by which the inequality is generated than on the level of disparity. A key concern among focus group participants in both urban and rural areas was whether existing inequalities in outcomes were generated through fair or unfair means, such as corruption, misuse of power, and dishonest business practices. Unfair use of political capital and corruption were perceived to have affected well-being through multiple routes, from employment opportunities and land conversion to the ability to access high-quality public services and education. 172 6.91 If left uncurbed, inequalities in voice and connections that manifest themselves in a myriad of forms, from uneven land conversion practices to poor public service delivery, are likely to be damaging for social cohesion, economic progress, and growth. In the perceptions study, these inequalities provoked the most concern and frustration among participants, and were the focus of lengthy discussions. Inequalities in voice and connections are likely to play a role in determining whether individuals tolerate rising inequality in the future, directly through a sense of injustice and indirectly through their revised expectations of growth. There are suggestions that this may already be occurring via a reduction in the perceived return to education in rural areas, where focus group participants have suggested that their inability to translate education into employment opportunities, in part due to a lack of transparent recruitment mechanisms, has diminished their perception of the value of education for future generations. Box 6.1 discusses policy recommendations for dealing with inequality. Box 6.1 Emerging Policy Recommendations: Inequality Three key messages emerge for policy makers in Vietnam from the analysis of inequality. First, income inequality has risen in Vietnam, indicating that growth processes have been less favorable to poorer households and that poorer households are being left behind. Ethnic minority households have experienced slower growth on average than Kinh majority households, although there is substantial variation among minority households depending on endowments and sources of income. There is evidence of regional variation in growth rates, which has contributed to the rise in inequality. In addition, households characterized by lower average education levels are less likely to beneï¬?t from growth processes and to transition into the nonagricultural sector than more educated households. These patterns suggest an active role for policy to help households overcome the structural constraints facing poorer households that limit their growth potential. Second, inequality of outcomes affects the opportunity of children to fulï¬?ll their potential, and circumstances overtake potential early in life in Vietnam. Evidence presented in this chapter suggests that children who show promise at age 5 are unable to sustain that promise by age 8 to the same degree as their peers from better-off households. Inequality in opportunities of this form are likely to dampen growth and progress in Vietnam, since they imply that the full potential and talent of Vietnamese children are not being fully achieved. Moreover, it contributes to social tensions. Closing the gap in early childhood development and education quality in Vietnam is, therefore, desirable in terms of both equity and efï¬?ciency. Finally, there is widespread concern that inequality in connections, influence, and voice is affecting many aspects of Vietnamese peoples’ lives, from the ability of individual’s to attain public sector employment to obtaining access to good-quality public services and administration. These inequalities in political and social capital are not acceptable to Vietnamese citizens from all backgrounds, and inequality in income and spending that is due to unfair processes is less tolerated than inequality that arises through talent and hard work. Promoting transparent processes in Vietnam is necessary to ensure equitable growth—growth that is viewed as fair by its population. 173 Chapter Annexes Annex 6. 1 Why doâ€? Perceptions of Inequalityâ€? Diverge from Empirical Measures of Inequality? The empirical measurement of inequality includes four components (Cowell 2011). Perceptions of inequality may differ from empirical measures of inequality due to the following considerations: (a) the factor examined, (b) the unit of analysis, that is, whether a household or individual; (c) the reference group, that is, the universe of comparison, such as inequality at the national, regional, rural, or urban level; and (d) the inequality thermometer, or the tool used to capture changes in inequality, such as the Gini or Theil index. This section examines why perceptions may vary from empirical measures of inequality. First, it may be that our measures of inequality focus disproportionately on easily measured dimensions of inequality, such as outcomes, while Vietnamese people focus on other dimensions of inequality, such as the quality of education they receive or whether there is perceived unfairness in society. This chapter discussed modalities of inequality as they were perceived through the eyes of Vietnamese people. Not all modalities of inequality were discussed in each focus group, and the emphasis on different modalities of inequality varied substantially by group. For example, young working people often discussed employment inequalities in greater detail; ethnic minorities paid more attention to livelihood-related modalities of inequality in terms of access to market, credit, and technical services; and students and senior groups talked more about education and the unfair roles of power and connections in employment.60 Second, perceptions may differ from empirical measures because the frames of reference used in empirical analysis differ from that used by individuals when thinking about inequality. In contrast to most empirical measures of inequality, which capture inequalities at the national, regional, rural, or urban level, perceptions of inequality are often rooted in direct life experiences and have a narrower focus. Groups often discussed disparities within their communities, and then conceptualized a step up from their income levels to compare themselves with people in more favorable places or higher positions. For example, in contrast to the decline in inequality attributable to differences between rural and urban areas, rural respondents perceive inequality between rural and urban areas to have risen. However, in contrast to the empirical measure of inequality that compares the average level of welfare within urban areas to the average levels of welfare within rural areas, participants in the focus groups compared their own rural communities to nearby urban centers in the region. Since the empirical measures of inequality and perceptions of inequality are taking place at different levels of aggregation, it may be that, at a more local level, perceptions of inequality and measures of inequality converge.61 An empirical examination of inequality at a lower level of aggregation than normally used in a quantitative assessment may help to bridge the gap between empirical measures and perceptions of inequality. Figure 6A.1 shows inequality at a district level in 1999 and 2009, where a district is a lower 60 Another concern is that the incomes or expenditures of the rich are underreported and undercaptured in household surveys. Therefore, empirical measures of inequality may be downward biased (Cowell 2011; VASS 2011). 61 It may also be that people do not compare mean levels of welfare, but instead compare the richest people in urban areas with the richest, or poorest, in rural areas. 62 District-level inequality was computed using small area estimation techniques. See Benjamin et al. (2009) for more details. 174 unit of analysis than normally used when empirically examining inequality.62 District-level inequality rose in previously low-inequality districts and fell in higher-inequality districts. While this gets closer to the unit of analysis used by our focus group respondents, since the frames of reference used appear to vary substantially across individuals, it remains an approximation. Figure 6A.1 District-level Expenditure Figure 6A.2 District-level Expenditure Inequality, 1999 and 2009 Inequality, 1999 and 2009 Absolute Gini Coefï¬?cients 0.45 0.8 0.7 0.4 A b so lu te expen ditu re gini 0.6 District level gini, 2009 0.5 0.35 0.4 0.3 0.3 0.2 0.25 0.1 0 0.2 1998 2004 2006 2008 2010 0.2 0.25 0.3 0.35 0.4 0.45 District level gini, 1999 Income Consump on The most commonly used measures of inequality—the Gini Coefï¬?cient, the class of generalized entropy measures including the Theil index, and ratios of outcomes of people at different percentiles of the outcome distribution—capture inequality in relative terms. However, individuals may view inequality in absolute terms (Amiel and Cowell 1999; Ravallion 2004). For example, if everyone’s income rises by 7 percent, then relative measures of inequality will not register a rise in inequality even though the absolute gap has grown. Evidence from a developed country setting suggests that approximately 40 percent of individuals in a study on concepts of inequality thought of inequality in absolute terms rather than relative terms (Amiel and Cowell 1999). There is evidence in Vietnam that absolute inequality has been rising. Figure 6A.2 shows that the absolute Gini has risen in Vietnam since 1998. Whether individuals view inequality in relative or absolute terms is very difï¬?cult to capture, and there are only hints of this in the qualitative assessment. The suggestive evidence indicates that, in Vietnam, there are likely to be some individuals who also think about inequality in an absolute sense, and others who think of it in a relative sense. Therefore, even if relative measures of inequality remain constant, they may perceive inequality to be rising. For example, the ï¬?rst comment below suggests one focus group was discussing inequality in absolute terms, while the second comment suggests that another focus group was discussing inequality in relative terms. Whether Vietnamese people conceptualize inequality in absolute or relative terms will be examined further in follow-up work underway. “The group claimed that the government’s move to increase the salary base at times of inflation only broadened the income gap between the better-off and the poor. Justifying the irrationality of raising the salary base in percentage terms, they cited an example where the increase is 20 percent and the poor with the lower salary will get just some dozens of thousand VND while the better-incomed with the often higher salary base will receive additional millions of VND to their pay.â€? Site Report from Phuc Xa Ward, Hanoi (better-off residents) “The students claimed that the rich-poor gap over the past ï¬?ve years has been increasingly widened due to the increasing relative gap: the rich develop faster than the poor.â€? Site report from Linh Xuan Ward, Ho Chi Minh City (student group). 175 References Adams, Richard H. 1994. “Non-farm Income and Inequality in Rural Pakistan: A Decomposition Analysis.â€? The Journal of Development Studies 31 (1): 110–33. Amiel, Yoran, and Frank Cowell. 1999. Thinking about Inequality. Cambridge UK: Cambridge University Press. Anderson, James H., Alcaide Garrido, Maria Delï¬?na, and Tuyet Thi Phung. 2009. “Vietnam Development Report 2010: Modern Institutions.â€? World Bank, Washington DC. Asian Development Bank. 2012. “Outlook 2012: Confronting Rising Inequality in Asia.â€? Asian Development Bank, Manila. Banerjee, Abhijit V. and Esther Duflo. 2003. “Inequality And Growth: What Can The Data Say?â€? Journal of Economic Growth 8 (3) (September): 267–99. Benjamin, Dwayne, and Loren Brandt. 2002. “Property Rights, Labour Markets, and Efï¬?ciency in a Transition Economy: The Case of Rural China.â€? Canadian Journal of Economics 35 (4) (November): 689–716. Benjamin, Dwayne, Loren Brandt, and Brian McCaig. 2009. “The Evolution of Income Inequality in Vietnam between 1993 and 2006.â€? University of Toronto, Toronto. Benjamin, Dwayne, Loren Brandy, and John Giles. 2005. “The Evolution of Income Inequality in Rural China.â€? Economic Development and Cultural Change 53 (4) (July): 769–824. Benjamin, Dwayne, Loren Brandt, John Giles, and Sangui Wang. 2007. “Inequality and Poverty in China during Reform.â€? Poverty Monitoring, Measurement and Analysis Working Paper 2007, No 7. Partnership for Economic Policy–Poverty Monitoring, Measurement and Analysis. Bourguignon, F. 2004. “The Poverty-Growth-Inequality Triangle.â€? World Bank, Washington, DC. CECODES, FR, CPP, and UNDP. 2012. “The Viet Nam Governance and Public Administration Performance Index (PAPI): Measuring Citizens’ Experiences.â€? A Joint Policy Research Paper by the Centre for Community Support and Development Studies (CECODES), The Front Review of the Central Committee for the Viet Nam Fatherland Front (FR), the Commission on People’s Petitions of the Standing Committee for the National Assembly of Viet Nam (CPP), and the United Nations Development Programme (UNDP), Hanoi. Cowell, F. A. 2011. Measuring Inequality (third edition) . Oxford: Oxford University Press. Elbers, Chris, Peter Lanjouw, Johan Mistiaen, and Berk Özler. 2008. “Reinterpreting Between-group Inequality.â€? Journal of Economic Inequality Springer 6 (3) (September): 231–45. Gallup, J. 2002. “The Wage Labour Market and Inequality in Vietnam in the 1990s.â€? World Bank, Washington, DC. Glewwe, P., and H. Patrinos. 1999. “The Role of the Private Sector in Education in Vietnam: Evidence from the Vietnam Living Standards Survey.â€? World Development 27 (5): 887–902. GSO (General Statistics Ofï¬?ce of Vietnam). 2009. “Vietnam Population and Housing Census 2009: Migration and Urbanization in Vietnam: Patterns, Trends and Differentials.â€? Ministry of Planning and Investment, General Statistics Ofï¬?ce, Government of Vietnam, Hanoi. Hirschman, Albert O., and Michael Rothschild. 1973. “The Changing Tolerance for Income Inequality in the Course of Economic Development; with a Mathematical Appendix.â€? Quarterly Journal of Economics 87 (4): 544–66. Hoang, Thanh Houng, Le Dang Trung, Pham Thi Anh Tuyet, Pham Thai Hung, and To Trung Thanh. 2010. “Preserving Equitable Growth in Vietnam.â€? Background paper for the 2008–2010 Vietnam Poverty Assessment. Vietnamese Academy of Social Sciences, Hanoi. Hoang, Xuan Thanh, Nguyen Thu Phuong, Vu Van Ngoc, Do Thi Quyen, Nguyen Thi Hoa, Dang 176 Thanh Hoa, and Nguyen Tam Giang. 2012. “Perceptions of Inequality in Vietnam: a Qualitative Study.â€? Background paper for the 2012 Vietnam Poverty Assessment. Hanoi. McCaig, Brian, Dwayne Benjamin, and Loren Brandt. 2009. “The Evolution of Income Inequality in Vietnam between 1993 and 2006.â€? University of Toronto, Toronto. McKay, Andy, and Finn Tarp. No date. “Welfare Dynamics in Rural Vietnam, 2006 to 2010.â€? Policy Brief No. 3 of 2012, Central Institute for Economic Management, Hanoi, Vietnam. Paes de Barros, R., F. Ferreira, J. Molinas Vega, and J. Saavedra Chanduvi. 2009. “Measuring Inequality of Opportunities in Latin America and the Caribbean.â€? World Bank, Washington, DC. Psacharopoulos, G. 1994. “Returns to Investment in Education: A Global Update further Update.â€? World Development 22 (9): 1325–1343. Ravallion, Martin. 2004. “Competing Concepts of Inequality in the Globalization Debate.â€? In Brookings Trade Forum 2004, ed. S. Collins and C. Graham. Washington, DC: Brookings Institution, 1–38. Ravallion, Martin, and Dominique van de Walle. 2008. “Does Rising Landlessness Signal Success or Failure for Vietnam’s Agrarian Transition?â€? Journal of Development Economics, Elsevier 87 (2) (October): 191–209. Ravallion, Martin, and Shouhua Chen. 2007. “China’s Uneven Progress against Poverty.â€? Journal of Development Economics 82 (1): 1–42. Roemer, John 1998. Equality of Opportunity. Cambridge, MA: Harvard University Press. Roemer, John E. 2006. “Economic Development as Opportunity Equalization.â€? Cowles Foundation Discussion Papers 1583, Cowles Foundation for Research in Economics, Yale University, New Haven, Connecticut. Roemer, John. 2011. “Equality of Opportunity as Opportunity Equalization.â€? Department of Political Science Discussion Paper, Yale University, New Haven. Stark, O., J. E. Taylor, and S. Yitzhaki. 1986. “Remittances and Inequality.â€? Economic Journal 96 (383): 722–740. VASS (Vietnamese Academy of Social Sciences). 2008. “Participatory Poverty Assessment 2008.â€? Viet Nam Academy of Social Sciences, Hanoi. VASS (Vietnamese Academy of Social Sciences). 2011. Poverty Reduction in Viet Nam: Achievements and Challenges. Hanoi: The World Publisher. VASS (Vietnamese Academy of Social Sciences). 2012. “Opportunities for Children in Vietnam.â€? Background paper for the 2012 Programmatic Poverty Assessment, World Bank, Washington, DC. Wells-Dang, Andrew. 2012. “Ethnic Minority Development in Vietnam: What Leads to Success?â€? Background paper for the 2012 Poverty Assessment, World Bank, Washington, DC. World Bank. 1999. Vietnam Development Report 2000: Attacking Poverty. Washington DC: World Bank. World Bank. 2004. Vietnam Development Report 2003: Poverty. Washington DC: World Bank. World Bank. 2006. World Development Report: Equity. Washington DC: World Bank. World Bank. 2009. From Poor Areas to Poor People: China’s Evolving Poverty Reduction Agenda – an Assessment of Inequality and Poverty. Washington, DC: World Bank. World Bank. 2010. “Assessing and Monitoring Governance in the Land Sector: The Land Governance Assessment Framework.â€? World Bank, Washington DC. World Bank. 2011. Vietnam – High-quality Education for All by 2020. World Bank, Washington DC. World Bank, Embassy of Denmark, and Embassy of Sweden. 2011. “Recognizing and Reducing Corruption Risks in Land Management in Vietnam.â€? National Political Publishing House, Hanoi. 177 Publishing licence number No:742/QÄ?LK-LÄ? and Ä?KKHXB CXB No: 279-2012/CXB/21-69/LÄ? Issued on 28 December 2012 Design and printed by Golden Sky Co., Ltd. Well Begun, Not Yet Done: Vietnam’s Remarkable Progress on Poverty Reduction and the Emerging Challenges The World Bank in Vietnam 8th Floor, 63 Ly Thai To Hanoi, Vietnam Tel: 84 4 39346600 Fax: 84 4 39346597 Website: www.worldbank.org/vn WORLD BANK