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, �?rst immunization
EA Enumeration Area
EAP East Asia and Paci�?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 Of�?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 Of�?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 Of�?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 Of�?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 Of�?ce, Vietnam)
and Michaela Prokop (UNDP/Hanoi) on the Human Development Index and multi-dimensional
poverty indicators.

The report bene�?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 bene�?ted as well from
comments received at two seminars sponsored by the World Bank of�?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 �?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 �?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 �?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 �?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 �?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 �?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 Of�?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 Pro�?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 Diversi�?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 Diversi�?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 Comparable�?
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 do�? Perceptions of Inequality�? 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 Suf�?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 Of�?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 Coef�?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 Coef�?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 Of�?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 Speci�?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 Of�?cially Classi�?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 Of�?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 De�?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 Disadvantage�? 112
Box 5. 2 An Ede Coffee “Hotspot�? 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 needs�? 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 “international�? 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 needs�? 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 �?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 dif�?cult. Vietnam’s success
has created new challenges. The remaining poor are harder to reach; they face dif�?cult challenges—
of isolation, limited assets, low levels of education, poor health status—and poverty reduction has

1 The General Statistics Of�?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 bene�?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 �?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 Of�?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 pro�?le of poverty using data from the 2010 VHLSS and new
qualitative �?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 �?rst approach was developed by the Ministry of Labor, Invalids, and Social Affairs (MOLISA),
the agency identi�?ed by government in the early 1990s to have primary responsibility for Vietnam’s
poverty reduction programs and policies. MOLISA is tasked with proposing of�?cial urban and rural
poverty lines at the beginning of each �?ve-year Socio-Economic Development Plan (SEDP) and
setting the initial period poverty rate. Using the of�?cial lines, MOLISA is responsible for assessing
changes in poverty and updating the of�?cial list of poor households on an annual basis, using a
“bottom-up�? 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
of�?cial lines are not adjusted for inﬂation, but revised in real terms only every �?ve years. MOLISA
uses this approach to determine budget allocations and de�?ne eligibility for a number of targeted
poverty reduction programs (for example, the National Targeted Program for Sustainable Poverty
Reduction/NTP-SPR, Program 30a).

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 of�?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 of�?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 of�?cial poverty
lines and methodology are constrained by resource availability; they are revised every �?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 de�?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 poverty�? GSO-WB line is VND
435,000 per person per month (US $1.50, 2005 PPP). These compare to new of�?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 of�?cial poverty rate of
14.2 percent based on Vietnam’s of�?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 of�?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 of�?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 of�?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 of�?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 Of�?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 pro�?le of poverty based on the 2010
VHLSS, complemented by new information collected through Participatory Poverty Assessments
(PPAs) and qualitative �?eld studies. The poverty rate—de�?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 pro�?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 pro�?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, speci�?c disadvantages linked
to ethnic identity, and exposure to natural disasters and risks. Over the past decade, rising levels of
education and diversi�?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 diversi�?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 diversi�?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
bene�?ts to poor and at-risk individuals. In 2010, only half of the extreme poor reported that they
were eligible to receive bene�?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 “wealth�? 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 bene�?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 �?rst based on empirical analysis
of various rounds of the VHLSS and the second drawing on �?ndings from a new qualitative �?eld study
of “perceptions of inequality�? 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 identi�?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 �?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 diversi�?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 bene�?ts and the of�?cial
poverty lines used to target these bene�?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 speci�?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
bene�?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 pro�?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 �?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 diversi�?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 (�?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 dif�?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 dif�?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 �?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 �?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 de�?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 of�?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 of�?cials. The most important innovation introduced by the
Enterprise Law was the simpli�?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 con�?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 of�?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 �?rms, foreign �?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 �?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 identi�?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 classi�?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 �?rms are the source of half of Vietnam’s nonoil exports. In
terms of employment, however, these �?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 �?scal de�?cits and state-run �?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.�?

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 �?rst approach was developed and led by the Ministry of Labor, Invalids, and Social Affairs
(MOLISA), identi�?ed in the early 1990s as the primary government agency responsible for
poverty reduction programs and policies. MOLISA is tasked with proposing of�?cial urban and
rural poverty lines at the beginning of each �?ve-year Socio-Economic Development Plan (SEDP)
and setting the beginning period poverty rate. Using the of�?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-up�? 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 Of�?ce[GSO])
using a Cost-of-Basic-Needs (CBN) methodology similar to the second approach (see below) led
by GSO. The of�?cial lines are not adjusted for inﬂation, but are revised in real terms only every
�?ve years. MOLISA’s primary objective using this approach is to determine budget allocations
and de�?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 of�?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 �?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 of�?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 �?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 Of�?ce/World Bank
(GSO-WB) estimates and of�?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 of�?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

50
Heacount rate of poverty (HCR) (%)

GSOͲWB poverty HCR
Official MOLISA poverty HCR
40

20.7
20
14.2

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.�?

14
MOLISA poverty list, with a ceiling de�?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 of�?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 of�?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 of�?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 bene�?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 “budgeting�? or “planning�? lines, and
the process of agreeing on of�?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, �?nancial, and political
considerations. As described in chapter 2, other countries face similar challenges.

1.19 In September 2010, Vietnam announced a new of�?cial poverty rate of 14.2 percent (�?gure
1.2). The of�?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 of�?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 of�?cial “near-poor�? 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 of�?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 �?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 �?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 of�?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/Q�?-L�?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 scienti�?c
basis for bene�?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 (�?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 �?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 �?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 con�?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 pro�?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 “improved�? 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 de�?nitions of “clean�? and “improved�? 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 signi�?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
�?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 of�?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 �?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 �?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 �?nished in Vietnam, and indeed has
become more dif�?cult in many respects.

1.40 The task of eradicating poverty is not �?nished because:

�? 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.

�? 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 dif�?cult in other important respects. Growth
rates have fallen sharply compared to the �?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
diversi�?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.

�? 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.

�? 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 “socializing�? 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.

�? 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 bene�?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 of�?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)

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) “international�? 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 of�?cial poverty estimates (Chapter 2).
1.48 Changing norms and higher aspirations are also captured in a number of qualitative �?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 de�?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 �?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 de�?nitions of success. The most common response
was linked to suf�?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 of�?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).�?
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 of�?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 �?ght against poverty, given its rapid economic growth and vision of itself as a
modern industrial society—the new of�?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 �?nancial
crisis. The combination of large shocks and many small, often local shocks can be dif�?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 de�?ned near-poor poverty lines

23
that are 1.3 times the of�?cial poverty line. If a similar approach to de�?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, de�?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 speci�?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 �?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) identi�?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.�? 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 speci�?c (for example, bad weather) in affecting household living
conditions. They also document the emerging impacts of “macro�? risks such as inﬂation and global
economic crises. Even for the most affected groups, while macro risks worsened existing dif�?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 �?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 inequality�? 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
reduction�? 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 deviance�? study of poverty among ethnic
minorities analyzed a range of concerns speci�?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 of�?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 �?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

14
Kinh/Hoa
Ethnic Minorities
12

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 �?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 (�?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 identi�?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 “procedural�? 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 �?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 Paci�?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 inequality�? study documents “acceptable�?
and “unacceptable�? 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 bene�?ts of rapid development.

9 A growth incidence curve plots the annual rate of growth between two points in time for speci�?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 bene�?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 “socialization�? 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
(�?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 signi�?cant challenges of rising costs and economic instability
1.73 Vietnam has weathered the global economic storm following the �?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 �?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
dif�?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 dif�?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 �?xed incomes, which are not inﬂation-
indexed, such as pensioners, bene�?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 dif�?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 pro�?le of poverty using data from the 2010 VHLSS
and new qualitative �?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 �?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 deviance�? study on ethnic minority poverty

This �?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 “positive�? since they escape poverty despite facing the same challenges and obstacles as their
neighbors, and “deviants�? (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 of�?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
propositions�? for qualitative data collection that were explored through interviews and observation in
�?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 of�?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 �?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 Region�?, 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.�? 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.�? Joint publication of the UNDP and Vietnam General Statistics Of�?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.�? 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.�? Background paper prepared for the 2012 Vietnam Poverty Assessment, Hanoi.

Iyer, Lakshmi, and Quy-Toan Do. 2008. “Land Titling and Rural Transition in Vietnam.�? 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.�? 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.�? 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,�? Hanoi.

Oxfam GB/ActionAid. 2008. “Participatory Monitoring of Urban Poverty in Vietnam: Synthesis
Report 2008,�? Hanoi.

Ramalingam, B. 2011. “A Q&A on Positive Deviance, Innovation and Complexity.�? 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?�? World Bank Economic Review, 11(2): 357-382.

Ravallion, Martin and Shouhua Chen. 2007. “China’s Uneven Progress against Poverty.�? Journal of
Development Economics 82: 1-42

Ravallion, Martin, Shoahua Chen, and Prem Sangraula. 2008. “Dollar a Day Revisited.�? 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.�? The Professional Geographer.

Vietnam Academy of Social Sciences. 2009. “Participatory Poverty Assessment: 2008 Synthesis
Report,�? 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,�? 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.�? 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
Of�?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 diversi�?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 diversi�?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 de�?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 of�?cial estimates. Reasons for these differences are also discussed.

2.3 The chapter also describes a new methodology for estimating “subjective�? 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 de�?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 scienti�?c considerations. The enormous

10 The original GSO-WB poverty line was prepared as input to the 2000 “Poverty Assessment Attacking Poverty.�?
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 of�?cial poverty lines for the 2006–2010 Socio-economic
Development Plan and by MOLISA and GSO more recently to construct of�?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 of�?cial poverty lines are far too low (box 2.1).

Box 2.1 Do India’s New Of�?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 of�?cial poverty rate was �?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 de�?nition of poverty because they do not take into account malnutrition, sanitation,
drinking water, housing and health needs�? (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 bene�?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 of�?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 poverty�? (Banerjee 2011).

Sources: Banerjee 2011; Gill 2012.

2.6 Vietnam’s of�?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 of�?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 dif�?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
pro�?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
identi�?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 �?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 de�?ned as the proportion of the population in a speci�?c
period whose welfare (consumption per capita) falls below the poverty line (�?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.

De�?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 �?rst de�?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 suf�?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 �?rst de�?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-�?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 speci�?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 City�? 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 �?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
�?xed reference period (30 days) rather than a “typical month�? (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 �?xed reference period was triggered by dif�?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 �?xed reference period (prior month) rather than
being based on “typical�? 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 bene�?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 �?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 de�?ned on the basis of the welfare aggregate. Any changes in the de�?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 de�?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 dif�?cult to compare national poverty lines and poverty rates across
countries, even when converted into “internationally�? 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 comparable�?) 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 �?xed percentage of total nonfood consumption rather than
derived using conventional hedonic methods. This same �?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 diversi�?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 “comprehensive�? 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 �?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 �?xed share of the reported sales value of the dwelling unit. This
�?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 bene�?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 �?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 �?xed reference period in the expenditure
module, and a revised de�?nition of welfare, which make comparisons dif�?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-de�?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 pro�?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 �?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-speci�?c
prices and (sometimes) region-speci�?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 of�?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 speci�?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 speci�?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
dif�?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 �?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 de�?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.

De�?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.�? (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 De�?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 “real�? per capita expenditures) from least well-off to most well-off, and the poor were initially
de�?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 �?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 of�?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, de�?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 “acceptable�? 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-speci�?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 (�?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 diversi�?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 diversi�?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-quanti�?ed share of
consumption, that is, food reported under “other�? categories and meals eaten outside the household.
More than 95 percent of food consumption was recorded under quanti�?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 quanti�?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 de�?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 of�?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-quanti�?ed consumption) are
available for all food items in the VHLSS and, moreover, can be estimated speci�?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 speci�?ed to be used as prices. Even well-de�?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 of�?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 de�?ned “norm�? for nonfood expenditures in the way that caloric norms can
be used to de�?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 “austere�? 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 adequate�? 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௜
ൌ ߙ ൅ ߚଵ ቀ ௙ �? ൅ ߛ ᇱ ൫݀௧ െ ݀ҧ ൯ ൅ ‫݈ܽݑ݀݅�?�?ݎ‬௜
‫ݕ‬௜ ܾ

௙ሺ௬೔ ሻ ೔ ௬
where is the food budget share, α is a national intercept, ቀ௕೑ �? 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 adequate�? allowance for nonfood) as the new GSO-WB poverty line, which is thus

51
de�?ned as the food poverty line divided by Engel’s coef�?cient estimated from the regression (.525)16:
ܾ௙
ߙ ‫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 de�?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 Of�?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 of�?cial household-level poverty estimates for 2010,17 based on of�?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 of�?cial poverty
lines. Comparing the two estimates for 2010, of�?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 of�?cial estimates.

Table 2.5 Poverty Estimates for 2010: Comparing the GSO-WB Methodology
and Of�?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 de�?ned as α*= α+ β1 log(1/α*).
17 Of�?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, of�?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 of�?cial lines for the
following reasons:
�? Of�?cial lines were �?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.
�? Of�?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 modi�?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 method�? poverty rate
was 21.3 percent, slightly higher than the “new housing method�? poverty rate.
�? Although food poverty lines are similar in the of�?cial and GSO-WB approaches, a
decision was made to use a lower allocation for essential nonfood spending for the
of�?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:
�? Of�?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 pro�?le) than consumption-based estimates.
�? 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 of�?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 bene�?ciaries of social
assistance and social insurance policies and programs.

2.56 Of�?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 of�?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 suf�?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 �?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 speci�?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
suf�?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 insuf�?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 insuf�?cient consumption of foodstuffs, and the percentage was signi�?cantly higher in rural
than in urban areas—14 percent compared to 5 percent (�?gure 2.5). A surprisingly high percentage
of households (25 percent in rural areas) reported they were not able to consume suf�?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 Suf�?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 suf�?ciency also differed across regions. Households in poorer regions (for
example, Northern Mountains, Central Highlands) were more likely to report insuf�?cient levels of
consumption. Concerns about insuf�?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 suf�?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 suf�?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 speci�?cation of the regression
model. All estimates of subjective poverty lines were higher than Vietnam’s of�?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 Comparable�?
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 speci�?cally identi�?ed foods (including outdoor meals), and
another 19 specially identi�?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 speci�?cations (aided by photographs to
ensure standardization) and were instructed to take two observations on the price of the detailed
speci�?cation and to record whether that particular speci�?cation was the most common one in the
market. A particular size, and brand name (for packaged goods), was speci�?ed to avoid variation due
to either bulk discounting or quality discounting. In almost 80 percent of the market-item combinations,
the speci�?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 speci�?cation that was not the target speci�?cation. The price of the target speci�?cation was
regressed against the prices of the alternate speci�?cations (using brand name �?xed effects, or for
unbranded items, creating quasi-brands by dividing into intervals based on their unit prices) and a
set of regional �?xed effects. The regressions were used to impute the price of the target speci�?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. ୎
୩୨ ൅ ୧୨ ୧୨
ܶ ൌ �?‫݌ݔ‬ሾ�? ൬ ൰ �? ቆ ቇሿ
ʹ ୩୨
୨ୀଵ

where P denotes prices in each region and S is the budget shares.

The Törnqvist index speci�?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
�?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 “absolute�? poverty line that is held
�?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 speci�?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 identi�?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,�?, “bad,�?
and so forth, to “very good.�? A similar method, the Minimum Income Question (MIQ), asks about the
minimum income that respondents perceive to be necessary “to make ends meet�? (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.�? 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 “water�? “electricity�? “housing�? “clothing & footwear�?

Out of total respondents to the 2010 VHLSS consumption section, 440 reported insuf�?cient food
consumption, 8,218 reported just suf�?cient food, and 686 indicated that their food consumption
was more than suf�?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 suf�?cient
consumption, and 678 claimed their consumption was more than suf�?cient.

To calculate a subjective poverty line, we follow Pradhan and Ravallion (2000) in regressing perceived
suf�?ciency of consumption on household expenditure and household (head) characteristics, using
the suf�?ciency of foodstuff as the dependent variable. “Not Applicable�? 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 coef�?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
Coef�?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 suf�?ciency of foodstuff consumption�? with the following answer codes: 1 =
insuf�?cient, 2 = suf�?cient, and 3 = more than suf�?cient (“not applicable�? is recoded as missing). The results are from an
ordered probit regression. The natural logarithm is used for the log variables. Signi�?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.�? 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.�? American Economic Review, Papers and Proceedings 91 (2): 67–72.

Conti, G., and S. Pudney. 2011. “Survey Design and the Analysis of Satisfaction.�? Review of
Economics and Statistics 93 (3): 1087–1093.

Deaton, A., and O. Dupriez. 2011. “Spatial Price Differences within Large Countries.�? Princeton
University Working Papers, Princeton, NJ.

Deaton, A., and S. Zaidi. 2002. “A Guide to Aggregating Consumption Expenditures.�? Living Standards
Measurement Study Working Paper No. 135, World Bank, Washington, DC.

Deaton, Angus. 1988. “Quality, Quantity, and Spatial Variation in Price.�? 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.�? 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?�? 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.�? 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 Of�?ce.

Kapteyn, A. 1994. “The Measurement of Household Cost Functions: Revealed Preference versus
Subjective Measures.�? 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.�? European Economic Review 7: 313–335.

Kozel, Valerie, Ian Hinsdale, and Nguyen Phong. 2012. “Updated Methodologies for Poverty
Monitoring in Vietnam.�? 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.�? 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.�? Journal of Development Studies 42 (4): 559–591.

Marra, M. 2012. “Estimating Subjective Poverty Lines for Vietnam.�? Background note prepared for
the 2012 Vietnam Poverty Assessment, World Bank, Washington, DC.

Ministry of Health. 2006. “Proposed Nutrition Needs for the Vietnamese.�? Ministry of Health, Hanoi.

Phung, D. T. 2005. “Determination of a Consistent Poverty Line for Vietnam.�? Environmental and
Social Department, General Statistics Of�?ce, Government of Vietnam.

61
Pincus, J., and J. Sender. 2008. “Quantifying Poverty in Vietnam: Who Counts?�? Journal of Vietnamese
Studies 2 (1) (January): 108–150.

Pradhan, M., and M. Ravallion. 2000. “Measuring Poverty Using Qualitative Perceptions of
Consumption Adequacy.�? 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.�? The Review of Income and
Wealth, Series 47 (4): 473–487.

Ravallion, Martin. 1998. “Poverty Lines in Theory and Practice.�? 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.�?
World Bank Policy Research Working Paper 5968, World Bank, Washington, DC.

Ravallion, M., and B. Bidani. 1994. “How Robust Is a Poverty Pro�?le?�? The World Bank Economic
Review 8 (1) (January): 75–102.

Ravallion, M., and M., Lokshin. 2002. “Self-Rated Economic Welfare in Russia.�? 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.�? 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.�? 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 Pro�?le: Establishing the
Facts about Poverty and the Poor in
Vietnam

A new poverty pro�?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 pro�?le of the poor, using
the 2010 General Statistics Of�?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 �?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-poor�? 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 poor�? (per-capita expenditures below two-thirds of the poverty line) and “poor�? (per-
capita expenditures below the poverty line). The remainder of the population is analyzed on the
basis of per-capita expenditure quintiles and deciles. Speci�?cally:

�? Individuals are ranked by per-capita expenditures from least well-off to most well-off,
then divided into �?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.

�? Individuals are also categorized into expanded per-capita expenditure quintiles, where
the poor are classi�?ed into two groups (all poor and extreme poor) and the non-poor
are classi�?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-poor�? classi�?cation, which includes
households whose per-capita income lies between the poverty line and 1.3 times the poverty line.
If this de�?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 dif�?cult to reach than those
living close to the poverty line; they face a structural barriers and speci�?c constraints, and better
policies and programs are needed to address these speci�?c challenges. In many countries, including
Vietnam, the extreme and destitute poor are falling further behind. This chapter develops pro�?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 pro�?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 identi�?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 De�?ning Characteristics of Poor Households at the end of the 1990s

By the end of the 1990s, the key de�?ning characteristics of poor households included:
�? 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-�?fths of the poor were agriculture households.
�? 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.
�? 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.
�? Poor households were vulnerable to seasonal hardship and household-speci�?c and
communitywide shocks and some were socially and physically isolated.
�? Poverty among ethnic minority groups had declined, but not as rapidly as for the
majority population. Ethnic minorities faced many speci�?c disadvantages that could
best be addressed through an Ethnic Minority Development Program.
�? Migrants to urban areas who were poor and who had not secured permanent
registration faced dif�?culties accessing public services and some felt socially
marginalized. Further work was needed to identify the best way to help these
groups.
�? 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, speci�?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, diversi�?ed out of agriculture and into manufacturing and services, and reduced exposure
to seasonal hardships and shocks through income diversi�?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 (�?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 (�?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 De�?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 �?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 �?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 diversi�?ed into
off-farm activities. Progress in the 1990s was driven by on-farm diversi�?cation, for instance into cash
crops, livestock, and (in some parts of the country) �?sh and shrimp farming (World Bank 1999). But
progress in recent years has been driven by diversi�?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 �?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 “tail�? 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 (�?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 �?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 diversi�?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, “pushed�? by poverty and economic imperatives and “pulled�?
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 diversi�?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 diversi�?cation
into cash crops, and even more important, diversi�?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 pro�?ts, leaving
them without adequate means of livelihoods; others argued that land markets would promote greater
ef�?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-poor�? households in the Mekong and Southeast to diversify into
higher paid off-farm activities. However, off-farm diversi�?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 �?gures taken from the World Bank 2000 Vietnam Development Report, table 2.4. 2010 �?gures are
World Bank estimates from 2010 VHLSS.
Note: Swidden land is land cleared for cultivation by cutting and burning the vegetation.
“Land�? includes annual cropland, perennial cropland, forestry land, water surface, and shifting-cultivation farmland. It
excludes gardens, ponds, and land classi�?ed as “other.�?

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 �?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 cities�?
(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 “rural�? than larger cities; urban poverty is concentrated
in the more “rural-like�? 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 �?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
�?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 Speci�?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 �?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
“pyramids�? 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 �?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 “quality�? spending on children.

3.38 The long-run drivers study, with its two-decade reference period, also identi�?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 identi�?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 pro�?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 �?xed costs and less directly linked to household size. To adjust for economies of
scale, individual welfare is rede�?ned as

ܻ
‫ כݕ‬ൌ
ሺܰሻ�?

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-speci�?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 �?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-speci�?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 identi�?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 �?ve who are poor according to conventional income or expenditure criteria. (�?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 identi�?ed as poor according to
both methods, there is also a group that is only identi�?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 �?gures also provide evidence of a signi�?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 �?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 �?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 �?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 classi�?ed as poor—that is, whether they are on the of�?cial MOLISA poverty
list—and thus eligible for bene�?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 bene�?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 of�?cial poverty list. Thus, while the of�?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 classi�?ed as poor by commune authorities, and are thus on the of�?cial MOLISA
poverty list. 8 percent of individuals in the 2010 VHLSS are classi�?ed as extreme poor by the
updated GSO-WB poverty line. However, only 52 percent these households said they were on the
of�?cial poverty list. Similarly, 20.7 percent of individuals were classi�?ed as poor using the updated
GSO-WB poverty line, but only 36 percent of these households said they were on the of�?cial poverty
list. Thus coverage is low, but leakage of bene�?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 of�?cial poverty list.

Table 3.19 Percentage of Households Of�?cially Classi�?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
classi�?ed as poor using the GSO-WB poverty line. Only 11.5 percent of those of�?cially classi�?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 bene�?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 bene�?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 Of�?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 classi�?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 bene�?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, �?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. Speci�?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
dif�?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 re�?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 signi�?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
cat�?sh and shrimp and a variety of fruits.

91
References

GSO (General Statistics Of�?ce of Vietnam). 1998. “Decision to Approve the Orientations of the
Master Plan for the Development of Vietnam’s Urban Centers till 2020.�? Decree 10/1998/QD-TTg.
Hanoi, January 23, 1998.

GSO (General Statistics Of�?ce of Vietnam). 2011. “Report on Multidimensional Child Poverty in
Vietnam.�? Prepared jointly by UNICEF and GSO, Hanoi, September.

GSO (General Statistics Of�?ce of Vietnam), 2010. “Migration and Urbanization in Vietnam: Patterns,
Trends and Differntials.�? 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.�? Joint publication of the UNDP and Vietnam General Statistics Of�?ce, Hanoi.

Le, T. D., C. V. Nguyen, and T. D. Phung. 2012. “Natural Shocks, Vulnerability to Poverty in Vietnam.�?
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.�? 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
Vietnam�?. 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.�? 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 Asia�?.
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.�? 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 Of�?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,�? 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 �?ndings, the
mapping methodology is used to assess whether the 62 “poorest districts�? identi�?ed under Program
30A are indeed among the poorest in Vietnam. Initial �?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 �?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 signi�?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 (�?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 pro�?le presented in Chapter 3, results using the poverty mapping
approaches con�?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 con�?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
(�?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 “absolute�? inequality). Provincial results are presented in table
4.3. Provincial- and district-level estimates are presented in the �?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 �?nding is noteworthy in light of the common (often implicit) view
in Vietnam that everyone in poor communities is similarly poor. But the �?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
(�?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) (�?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 (�?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 diversi�?cation has been a strong force for
growth and poverty reduction. Much attention has been paid to diversi�?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 bene�?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 �?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
diversi�?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 signi�?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 bene�?t from an expanding nonfarm sector.
These �?ndings stand in contrast to Hoang et al. (2012), whose �?ndings suggest that the very poor
do not bene�?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 �?ndings is that the bene�?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 �?nding is that the bene�?ts from spatial targeting, at any level of geographic
disaggregation, are more clearly evident in 2009 than 1999. This �?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 �?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 �?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 �?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 speci�?c policy interventions (box 4.1). Mapping methods were used to see
whether the 62 poorest districts identi�?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 identi�?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 speci�?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 insuf�?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, �?rst, that in both 1999 and 2009, there are potentially
large gains in targeting performance from disaggregating to the local level. These bene�?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 “optimize�? our transfer
scheme with respect to this poverty measure. An important corollary of these �?ndings is that the
bene�?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 bene�?ts from spatial targeting, at any level of
disaggregation, are more clearly evident in 2009 than in 1999. This �?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 �?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 �?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) �?nd that �?ne geographic targeting offers signi�?cant bene�?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, “optimization�? is not well de�?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 bene�?ts from targeting will vary with the overall size of budget. In the limit, as the budget
goes to in�?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 “Optimal�? Geographic Targeting
Simulating the impact of the “optimal�? 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 part�? 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 – “Optimal�? 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 �?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 mapping�? 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 �?nd that n = 20 gives suf�?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). De�?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<g0:

Find k such that g k 1 d g g k ..

The per-capita transfers resulting in poverty gap g are

(A.6) gk g z
a( g ) ak .
g k g k 1 n

This scheme can be implemented using standard spreadsheet software.

Results
Table A4.1 presents the basic results from our simulations. Use of disaggregated data on poverty to
allocate transfers gives better results than a uniform lump-sum transfer across the entire population.
Targeting transfers to poor localities, in accordance with the optimization scheme outlined above,
yields lower values of the national FGT2 than when the budget is transferred as a uniform lump-sum
transfer to the entire population. Second, the more disaggregated the poverty map, the greater the
improvement over a uniform lump-sum transfer. Our simulations suggest that using estimates of
poverty at the province, district, and commune levels results in non-negligible improvements in the
FGT2 with a given budget. However while the general patterns we observe are similar across our
two poverty maps for 1999 and 2009, they are not identical. Notably, while commune-level targeting
in 1999 would reduce the FGT2 from a level of 0.0110, following a uniform transfer, to 0.0058 with
commune-level targeting (a 43 percentage point reduction), the improvement from commune-level
targeting in 2009 would be 66 percentage points—the FGT2 declining from 0.0166 to 0.0057 (table
A4.1). With district-level targeting rather than commune-level targeting, the gains are slightly less
marked but are still evident.31

Table A4.2 repeats the simulations presented in table A4.1 but focuses now on the headcount, or
FGT0, measure of poverty. As mentioned above, the optimization procedure outlined in Kanbur
(1987) applies to the squared poverty gap or FGT2 measure. There is no analogous optimization
algorithm for the FGT0 measure. We report in table A4.2, however, the resulting FGT0 estimates
from having applied the procedure to allocate our budget in such a way as to minimize the resulting
FGT2 measure. Table A4.2 indicates that the gains in terms of the FGT0 of geographic targeting are
far less marked than was observed when the FGT2 measure was our reference measure.

31 While targeting improves signi�?cantly as one is able to progressively disaggregate, for example, from the province, to
district, to commune level, it remains far from perfect. Simulating the impact of optimal targeting of our postulated budget
to individual households would result in a further decline in the FGT2 from 0.0057 in 2009 (table A4.1) to 0.0019. The
fact that commune-level targeting is unable to reproduce what would be achieved with perfect, household-level, targeting
con�?rms the �?ndings from earlier sections that inequality at the subnational level in Vietnam can be signi�?cant; even with
commune-level targeting, there would be signi�?cant leakage of resources to nonpoor households.

116
Table A4.1 Impact on FGT2 of Targeting at Different Levels of Geographic Disaggregation
Optimal Targeting Scheme

Budget = 5 percent of (Total Population * 25th Percentile Per Capita Expenditure)
Poverty Line = per capita expenditure de�?ning bottom quintile of population (pre-transfer)

1999 2009
Original FGT2 0.0159 0.0234
FGT2 after:
i) Uniform transfer 0.011 0.0166
ii) Province-level targeting 0.008 0.0096
(61/63 Provinces)
iii) District-level targeting 0.0066 0.0070
(614/685 Districts)
iv) Commune-level targeting 0.0058 0.0057
(10474/10896 communes)
Original FGT2 1.00 1.00
FGT2 after:
i) Uniform transfer 0.69 (1.00) 0.71 (1.00)
ii) Province-level targeting 0.50 (0.72) 0.41 (0.58)
(61/63 Provinces)
iii) District-level targeting 0.42 (0.61) 0.30 (0.42)
(614/685 Districts)
iv) Commune-level targeting 0.36 (0.57) 0.24 (0.34)
(10474/10896 communes)

Table A 4. 2 Impact on FGT0 of Targeting at Different Levels of Geographic Disaggregation
Optional Targeting Scheme

Budget = 5 percent of (Total Population * 25th Percentile Per Capita Expenditure)
Poverty Line = Per capita expenditure de�?ning bottom quintile of population (pre-transfer)

1999 2009
Original FGT0 0.2000 0.2000
FGT0 after:
i) Uniform transfer 0.1673 0.1724
ii) Province-level targeting 0.1522 0.1555
(61/63 Provinces)
iii) District-level targeting 0.1443 0.1465
(614/685 Districts)
iv) Commune-level targeting 0.1390 0.1372
(10474/10896 communes)
Original FGT0 1.00 1.00
FGT0 after:
i) Uniform transfer 0.84 (1.00) 0.86 (1.00)
ii) Province Level Targeting 0.76 (0.90) 0.78 (0.91)
(61/63 Provinces)
iii) District-level targeting 0.72 (0.86) 0.73 (0.85)
(614/685 Districts)
iv) Commune-level targeting 0.70 (0.83) 0.69 (0.80)
(10474/10896 communes)

117
Discussion
The stylized analysis presented in this section cannot be used to directly evaluate existing poverty
reduction programs in Vietnam. One possible exercise that could inform policy makers’ deliberations
is to compare the hypothetical “optimal�? provincial- and district-level budgetary distribution deriving
from an exercise as has been presented above with the actual provincial- and district-level distribution
that is currently in place. There is no presumption that these two should line up exactly. But follow-up
work would be justi�?ed if such an exercise were to reveal glaring inconsistencies.

There are important caveats attached to the �?ndings reported here. First, we assume that the
government is willing to accept that households with equal pre-transfer per-capita consumption levels
might enjoy different post-transfer consumption levels. Second, we have assumed that the budget
available for distribution is exogenously determined. We ignore the question of how the transfers are
�?nanced. Political economy considerations could inﬂuence options for resource mobilization (see,
for example, Gelbach and Pritchett, 2002). Third, we do not address the very real possibility that the
costs of administering a given transfer scheme might increase with the degree of disaggregation.
Fourth, we do not allow for behavioral responses in the population. Fifth, we do not address the
possibility that inequalities in power and inﬂuence that prevail in a community inﬂuence how transfers
are allocated. All these factors could result in an overestimation of the impact of spatial targeting on
poverty reduction.

The �?ndings presented here are illustrative only. At all times, the gains from targeting should be
juxtaposed against potential costs and political-economy considerations and should be scrutinized
against other possible policy objectives. In practice, a combination of geographic targeting among
villages and means-tested targeting within villages may be the best way forward. Policy makers in
Vietnam will need to assess such programs on a case-by-case basis to determine just how far to
rely on �?ne geographic targeting as a central element in their social protection and poverty reduction
strategies.

118
References

Baulch, B., and Vu Hoang Dat. 2012. “Exploring the Ethnic Dimensions of Poverty in Vietnam.�?
Background paper for the 2012 Poverty Assessment. World Bank, Washington, DC, May.

Bedi, T., A. Coudouel, and S. Simer. 2007. More than a Pretty Picture: Using Poverty Maps to Design
Better Policies and Interventions. Washington, DC: World Bank.

Bigman, D., and H. Fofack. 2000. “Geographic Targeting for Poverty Alleviation: Methodology and
Applications.�? Washington DC: World Bank Regional and Sectoral Studies, World Bank, Washington,
DC.

Elbers, C., J. Lanjouw, and P. Lanjouw. 2002. “Micro-Level Estimation of Welfare.�? Policy Research
Working Paper No. WPS 2911, World Bank, Washington, DC.

Elbers, C., J. Lanjouw, and P. Lanjouw. 2003. “Micro-level Estimation of Poverty and Inequality.
Econometrica 71 (1): 355–364.

Elbers, C., J. Lanjouw, and P. Lanjouw. 2003. “Micro-level Estimation of Poverty and Inequality.�?
Econometrica 71 (1): 355–364.

Elbers, C., T. Fujii, P. Lanjouw, B. Ozler, W. Yin. 2007. “Poverty Alleviation through Geographic
Targeting: How much does Disaggregation Help?�? Journal of Development Economics 83 (1):
198–213.

Ferreira, F., and P. Lanjouw. 2001. “Rural Nonfarm Activities and Poverty in the Brazilian Northeast.�?
World Development 29 (3): 509–528.

Foster, J., J. Greer, and E. Thorbecke. 1984. “A Class of Decomposable Poverty Measures.�?
Econometrics 52 (3): 761–66.

Gelbach, J., and L. Pritchett. 2002. “Is More for the Poor Less for the Poor? The Politics of Means-
Tested Targeting.�? Topics in Economic Analysis and Policy 2 (1) (July): 6.

GSO (General Statistics Of�?ce of Vietnam). 2009. Population and Housing Census Vietnam 2009.
General Statistics Of�?ce of Vietnam, Hanoi.

GSO (General Statistics Of�?ce of Vietnam), 2010. “Migration and Urbanization in Vietnam: Patterns,
Trends and Differntials.�? Monograph prepared with support from UNFPA based on 2009 Housing
and Population Census, Hanoi.

Hoang, Xuan Thanh, Nguyen Thu Phuong, Vu Van Ngoc, Do Thi Quyen, Nguyen Thi Hoa, Dang
Thanh Hoa, and Nguyen Tam Giang. 2012. “Inequality Perception Study in Vietnam.�? Background
paper for the 2012 Vietnam Poverty Assessment. Ageless Consultants, Hanoi.

Kanbur, R. 1987. “Measurement and Alleviation of Poverty.�? Staff Papers – International Monetary
Fund 34 (1) (March): 60–85.

Lanjouw, J. O., and P. Lanjouw. 2000. “The Rural Non-farm Sector: Issues and Evidence from
Developing Countries.�? Agricultural Economics 26 (1): 1–23.

Lanjouw, P., and M. Marra. 2013. “Rural Poverty Reduction, Non-farm Employment, and Proximity
to Cities.�? Background paper prepared for the 2012 Vietnam Poverty Assessment, Washington DC.

Minot, N., B. Baulch, and M. Epprecht. 2003. “Poverty and Inequality in Vietnam: Spatial Patterns and
Geographic Determinants.�? Final report of project Poverty Mapping and Market Access in Vietnam,
conducted by IFPRI and IDS. International Food Policy Research Institute, Washington, DC.

119
Nguyen, Viet Cuong. 2009. “Updating Poverty Maps without Panel Data: Evidence from Vietnam�?.
Asian Economic Journal, Vol. 253(4): 397-418.

Nguyen, Viet Cuong, Tran Ngoc Truong, and Roy Van Der Weide. 2010. “Poverty and Inequality
Maps in Rural Vietnam: An Application of Small Area Estimation,�? Asian Economic Journal, East
Asian Economic Association, Vol. 24(4): 355-390.

Nguyen, Cuong Viet and Linh Vu. 2012. “Poverty Targeting and Social Protection Strategies in
Vietnam�?. Background paper prepared for the 2012 Vietnam Poverty Assessment, World Bank,
Hanoi.

Nguyen, Viet Cuong, Peter Lanjouw, and Marleen Marra. 2012. “Vietnam’s Poverty Mapping using
the 2009 Housing Population Census and 2010 Vietnam Living Standards Survey.�? Background
paper prepared for the 2012 Poverty Assessment. World Bank, Hanoi.

Oseni, G., and P. Winters. 2009. “Rural Nonfarm Activities and Agricultural Crop Production in
Nigeria.�? Agricultural Economics 40 (2) (March): 189–201.

Ravallion, M. 1993. “Poverty Alleviation Through Regional Targeting: A Case Study of Indonesia.�? In
The Economics of Rural Organization: Theory, Practice and Policy, ed. A. Braverman, K. Hoff, and J.
Stiglitz. New York: World Bank and Oxford University Press.

Ravallion, M. 1994. Poverty Comparisons. Chur, Switzerland: Harwood Academic Press.

Ravallion, M., and K. Chao. 1988. “Targeted Policies for Poverty Alleviation under Imperfect
Information: Algorithms and Applications.�? Journal of Policy Modeling 11 (2): 213–224.

Ravallion, M., S. Chen, and P. Sangraula. 2007. “New Evidence on the Urbanization of Global
Poverty.�? World Bank Policy Research Working Paper 4199, World Bank, Washington, DC.

Ray, Debraj. 1998. Development Economics. Princeton: Princeton University Press.

120
Chapter 5
Reducing Poverty among Ethnic
Minorities

Data on poverty levels of ethnic minority groups in Vietnam is
analyzed using multiple dimensions of well-being, such as access to
education, water and sanitation, and public utilities. A combination
of qualitative and quantitative methods shows the diversity of ethnic
experiences, encompassing rural entrepreneurship, vulnerability
to shocks, and ongoing stigma and disadvantage. While ethnic
minorities’ welfare has increased overall, poverty reduction has
been uneven among ethnic groups and regions, resulting in a wider
poverty gap between most ethnic minorities and the Kinh majority

121
A. Introduction
5.1 Ethnic minority poverty presents a particular and persistent challenge for Vietnam. Although
households belonging to Vietnam’s 53 ethnic minority groups have experienced rising living standards
since 1998, they have not progressed as rapidly as the Kinh majority. As noted in Chapter 1, per-capita
consumption grew at an annual rate of 7.4 percent for minorities between 1998 and 2010 compared
to 9.4 percent over the same period for the Kinh. At the same time, ethnic minority households have
become increasingly linked to the commercial market, while continuing some elements of traditional
noncash livelihoods such as semi-subsistence agriculture and livestock raising (McElwee 2011;
Turner and Michaud 2011).

5.2 Ethnic minority poverty rates have fallen as a result of rising incomes and expenditures. From
a rate of 75.2 percent in 1998, the level of ethnic poverty (excluding the Hoa Chinese) fell to 50.3
percent by 2008, using the original General Statistics Of�?ce-World Bank (GSO-WB) poverty lines
and methodology. This rate remains much higher than among the Kinh majority, however. The pro�?le
of ethnic minority poverty in Chapter 3 based on the new 2010 poverty lines suggests that disparities
have risen; 47 percent of the poor in Vietnam are ethnic minorities, and the ethnic minority poverty
rate is 66.3 percent. Although the well-being of minorities has increased in income and consumption
terms, for many households these improvements have not been enough to put them over the
poverty line. Yet, these same data also show that almost a quarter (24.9 percent) of ethnic minority
households have escaped poverty since 1998.

5.3 The gap in reported poverty levels between Kinh and ethnic minorities increased rapidly during
the earlier years of Vietnam’s period of high economic growth and rapid poverty reduction. In 1993,
a member of an ethnic minority group was only 1.6 times more likely to be poor than a Kinh person
(see table 1.7). By 1998, this had risen to 2.4 times more likely, and by 2004, 4.5 times. By 2010,
minorities were on average 5.1 times more likely to be poor than the Kinh and, as documented in
Chapter 4, substantial gaps are evident throughout Vietnam.

5.4 The causes of persistent ethnic minority poverty have been researched in depth (ADB 2003;
DFID and UNDP 2003; Oxfam and ActionAid 2009; World Bank 2009). The World Bank’s 2009 “Country
Social Analysis: Ethnicity and Development�? found that minorities face disadvantages in access to
education, mobility, credit, land, linkages to markets, and ethnic stereotyping by the Kinh majority
(box 5.1). The reasons why some ethnic minorities have escaped poverty despite these barriers have
received less attention, yet may offer suggestions of positive practices that can be incorporated into
better-targeted and more innovative poverty reduction programs (Wells-Dang 2012).

Box 5.1 Six “ Pillars of Disadvantage�?

The 2009 World Bank “Country Social Analysis: Ethnicity and Development�? (World Bank
2009) identi�?ed three trends that account for different economic outcomes in minority and Kinh
communities: differences in assets, differences in capacity, and differences in voice. Within each
broad trend, there are numerous speci�?c causal factors for ethnic minority poverty, summarized
as six “pillars of disadvantage�?:

1. Lower levels of education
2. Less mobility
3. Less access to �?nancial services
4. Less productive, lower-quality land
5. Limited market access
6. Stereotyping and other cultural barriers.

There is no single factor that explains the difference in outcomes among ethnic minorities and
Kinh, even among those who live in the same areas. Instead, differences in these six areas
combine in a “vicious cycle�? to inﬂuence ethnic minority livelihood outcomes and lead both directly
and indirectly to persistent poverty. The Country Social Analysis concludes that poverty reduction
depends on comprehensive approaches to remove each of these pillars of disadvantage that
minorities face.

122
5.5 The gap in living standards between minorities and Kinh can be explained through reference
to the structural disadvantages faced by minorities (box 1.1). Research shows that although minority
household assets have improved over time—in particular, higher levels of education and better access
to basic infrastructure and services such as roads, clean water and sanitation, and electricity—there
is still a substantial gap in returns to assets between minorities and the Kinh (Baulch and Vu 2012;
Imai and Gaiha 2007; Kang 2009). A contributing factor to the ethnic poverty gap is the fact that
minorities continue to work primarily in agriculture (Chapter 3), which has grown more slowly than
other sectors of the economy. The gap, however, may be overstated due to measurement errors,
subjective linking of minorities and poverty by researchers and of�?cials, and the likelihood that some
minorities have unreported and noncash income sources that are not captured in the statistics.

5.6 This chapter draws on the broad framing of ethnic minority poverty in Chapters 3 and 4, with
the aim of looking in greater depth at the situation and challenges faced by diverse ethnic minority
groups, and examples of successful development for speci�?c groups and in various regions.

B. Ethnic Minority Poverty Reduction Varies across Regions, among and
within Ethnic Groups

5.7 Results from poverty mapping (Chapter 4, also Nguyen, Lanjouw, and Marra 2012) demonstrate
that ethnic minorities are not a homogeneous group. Figure 5.1 disaggregates changes in living
standards among four broad categories of ethnic groups that share certain cultural, geographic,
and social similarities. Among these four categories, the Khmer and Cham have seen the largest
increases in incomes and have the lowest overall poverty rates. From 1998 to 2008, poverty fell
steadily for all groups except Central Highland minorities; however, there are some indications that
progress is slowing. In 1998, minorities in the Central Highlands had the highest poverty and lowest
expenditures, but by 2010, this distinction had passed to groups in the Other Northern Uplands
category, including the Hmong and Dao and many smaller ethnicities.

Figure 5. 1 Changes in Welfare Levels (per-capita consumption) for different Ethnic Groups
in Vietnam,1998-2010

Source: World Bank estimates from various rounds of the Vietnam Household Living Standard Survey
(VHLSS): comparable per-capita consumption during 1998 and 2002; comprehensive per-capita
consumption during 2004–10.

123
5.8 Table 5.1 shows the predicted poverty headcount, poverty gap, and mean per capita
expenditures in 2010 for the 20 largest ethnic groups in Vietnam (listed in order of population size),
using the poverty mapping methodology presented in Chapter 4.32 Attention is con�?ned to rural areas
since this is where the vast majority of ethnic minority people live (84.3 percent, according to the
2009 Census). Of the largest ethnic minority groups, the Tay and Khmer have relatively low poverty
rates and high per-capita expenditures, while the �?gures for the Hoa (Chinese) are higher than for the
Kinh majority. Poverty rates can vary signi�?cantly among ethnic groups residing in the same region,
as shown in the differences between the historically more prosperous Tay, Nung, Thai, Muong, and
other northern minorities such as the Hmong and Dao. These groups, and many Central Highlands
minorities, have poverty rates over 75 percent and poverty gaps of over 25 percent. Compared to
the 1990s, however, the difference between Central Highland minorities and others has gradually
decreased, continuing a trend that was noted in earlier VHLSS surveys (Baulch, Pham, and Reilly
2007).

Table 5.1 Poverty and Median Expenditures of Major Ethnic Groups in Rural Areas, 2009

Ethnic Group Poverty Poverty MeanPer Capita Primary region
Headcount Gap Expenditures
1 Kinh 17.0 3.6 12,145,000 —
2 Tay 46.5 13.0 9,918,800 N. Mountains
3 Thai 69.1 22.6 7,210,600 N. Mountains
4 Muong 56.3 16.8 8,603,800 N. Mountains
5 Khmera 43.2 11.6 9,976,300 Mekong Delta
6 Hoa 13.4 3.1 19,727,500 Mekong Delta
7 Nung 56.0 17.5 8,611,600 N. Mountains
8 Hmong 93.3 45.3 4,455,100 N. Mountains
9 Dao 75.6 27.9 6,456,900 N. Mountains
10 Gia Rai 81.9 32.2 5,754,600 C. Highlands
11 Ede 75.1 27.6 6,460,100 C. Highlands
12 Ba Na 86.2 36.6 5,311,400 C. Highlands
13 San Chay 57.2 17.0 8,263,300 N. Mountains
14 Cham 57.2 17.0 8,504,100 South-Central
15 Co Ho 76.2 28.1 6,329,300 C. Highlands
16 Xo-Dang 91.1 42.4 4,760,600 C. Highlands
17 San Diu 37.5 10.2 11,132,400 N. Mountains
18 Hre 79.1 26.2 6,294,400 C. Highlands
19 Ra Glai 84.9 31.1 5,716,200 South-Central
20 Mnong 80.9 32.9 5,828,000 C. Highlands

Source: Estimates based on poverty mapping methods described in Chapter 4 using 2010 VHLSS and 2009 Housing and
Population Census.
Note: a. In Vietnamese, Khơ me. The H’Mông and Ê �?ê are also listed here by their common English names.

5.9 Figure 5.2 shows the distribution of per-capita expenditures in 2006 and 2010 (based on
the VHLSS) for the �?ve ethnic minority groupings. Both the mean and distribution of expenditures
improved for all groups from 2006 to 2010, resulting in declining poverty rates. The peak of the
distribution curve for Kinh and Hoa is now far past the 2010 GSO-WB poverty line. For the Tay, Thai,
Muong, and Nung, and for the Khmer and Cham, the curve peaks near the poverty line. But for the
Other Northern and Central Highlands minorities, the vast majority of households still live well below
the poverty line, despite improvements in the upper and middle ends of the expenditure distribution
between 2006 and 2010.

32 The sample size in the VHLSS is too small to permit disaggregation by speci�?c minority groups; hence, we use mapping
methods based on the 2009 Housing and Population Census.

124
Figure 5.2 Real Per-capita Expenditures for Five Ethnic Categories, 2006-10

Proportion of Population
Kinh-Hoa Legend
0 .1 .2 .3
2006 2010

dashed line shows the 2010 GSO-WB poverty line

best per capita expenditures, rural areas only
0 10 20 30 40 50
expenditure per person (real 2010 VND million)
Proportion of Population

Proportion of Population
Tay-Thai-Muong-Nung Other Northern Minorities
0 .1 .2 .3

0 .1 .2 .3
0 10 20 30 40 50 0 10 20 30 40 50
expenditure per person (real 2010 VND millions) expenditure per person (real 2010 VND millions)
Proportion of Population

Proportion of Population
Khmer & Cham Central Highland Minorities
0 .1 .2 .3

0 .1 .2 .3

0 10 20 30 40 50 0 10 20 30 40 50
expenditure per person (real 2010 VND millions) expenditure per person (real 2010 VND millions)

Source: 2010 and 2006 VHLSS.

5.10 Focusing in further on speci�?c ethnic groups in distinct locations increases the diversity of
results. In Lao Cai province, for example, the Ministry of Labour, Invalids and Social Affairs reports
an overall poverty rate of 43 percent. The Hmong (the most populous ethnicity in the province) have
a reported rate of 83 percent, Nung 75 percent, and Dao 72 percent (Lao Cai DOLISA 2012). One
of the smaller ethnic groups, the Phu La, have the highest reported poverty rate, at 84 percent. But
not all very small groups are equally disadvantaged. The Tu Di, a subgroup of Bo Y, are involved
in intercommunion and cross-border trade and have high reported educational attainment (Baulch
and Vu 2012; Wells-Dang 2012). Central Highlands provinces such as Dak Nong are characterized
by “complex patterns of inter-penetration between ethnic groups�?; Kinh make up a majority of the
population, have a 20 percent poverty rate, but make up 41 percent of poor people in the province.
In-migrating northern ethnic minorities (Thai, Tay, Nung, Dao, Muong, and Hmong) comprise 20
percent of the population and 37 percent of poor people, with a poverty rate of 56.8 percent, and
indigenous minorities (Ede, Mnong, Ma, and others) make up only 11 percent of the population and
21 percent of poor people, but their poverty rate is 63.8 percent (Shanks et al. 2012, 22–4).

5.11 Comparisons of 1999 and 2009 poverty maps (Chapter 4) indicate that the fastest poverty
reduction has taken place among ethnic minorities in the Central Highlands. Of districts with at least
40 percent ethnic minority populations, seven of the 10 with the highest rates of poverty reduction are
located in this region (three in Dak Lak and two each in Gia Lai and Lam Dong). Two of the others,
in Quang Nam and Binh Dinh provinces, border the Central Highlands. All of these districts started
from a very low income level in 1999 and have now reached a low to moderate level.

125
Map 5. 1 Regional Patters of Poverty and Wealth for Ethnic Minorities

Poor ethnic minorities live primarily in The wealthiest minorities live in the Mekong
mountainous regions in the north of Vietnam, Delta and Southeast regions, with some
and in the Central Highlands also in cities and towns in the Northeast
Mountains

Source: Lanjouw, Marra, and Nguyen 2012.

5.12 As described earlier, poor ethnic minority households are still concentrated in mountainous
and upland areas in the north of Vietnam and the Central Highlands. In contrast, the wealthiest
ethnic minorities (de�?ned as ethnic minorities with per-capita expenditures in the top 15 percent of
the national expenditure distribution) in Vietnam primarily (57 percent) live in the Mekong Delta and
Southeast regions. A third area with a concentration of wealthier minorities is in cities and towns in
the northeast mountains. The lowest reported welfare levels for ethnic minorities are found in the
northwest mountains and central coast areas (Quang Binh and Quang Tri). In the Central Highlands,
Dak Lak and Lam Dong report average income levels, while other provinces are below average (map
5.1).

5.13 Among rural districts with more than 5,000 ethnic minority residents surveyed in the 2009
Census, nine of the top 10 are located in the Mekong Delta, and all have predominantly Khmer
and Cham inhabitants. This includes four districts in Tra Vinh province and three in Soc Trang.
Expanding the subsample to include urban districts, higher expenditure levels are found among
ethnic minorities in Cao Bang and Lang Son cities, and in two peri-urban districts of Ho Chi Minh City
(Hoc Mon and Binh Chanh), home to many migrant workers. Ethnic minority residents of these areas
are predominantly Tay/Nung and Khmer, respectively.

126
C. Disparities in Access to Education, Infrastructure, and Public Services
Accompany and Reinforce Ethnic Minorities’ Poverty Reduction Outcomes

5.14 Including noneconomic indicators of well-being adds further complexity to the picture of
differential development outcomes among ethnic minorities. For instance, relative gaps between
Kinh and ethnic minorities have reduced access to education, due to increased numbers of schools,
improved roads, and higher incomes among minority households (Hoang et al. 2012). Particularly
at the primary and lower secondary level, ethnic minorities have greater levels of public school
enrolment than in the late 1990s (�?gure 5.3). Primary school enrolments for ethnic minority groups
are only a little lower than for Kinh but fall as children move through the school system. By the time
they reach upper secondary school, majority pupils are more than twice as likely to attend school
as minority pupils. This is in part a question of access, because most upper secondary schools are
located far from rural villages, and in part one of formal and informal costs of secondary education.
A focus group in Son La described these limitations:

“Education [in our community] is good, drop-out rates at primary and lower secondary
levels are low. We try to bring our children to school up the 12th grade. At upper
secondary level the children have to go to school in the district town, renting rooms,
bringing rice and vegetables from home, attending extra classes. Room rental is 150
thousand dong, pocket money 200–300 thousand dong per month at the lowest. But
many households cannot afford such costs, their children have to drop out.�? (Hoang et
al. 2012: 25)

5.15 As a result of increased access to public education, and to television and roads, the Vietnamese
language capabilities of many young minorities are greater than in the past. Without upper secondary
diplomas, however, employment options may remain limited for many young people, due to both
location and discrimination. Data show that Khmer and Cham have relatively high incomes and
better than average nutritional outcomes for children, but low secondary school completion rates
in public Vietnamese-language schools affect subsequent job opportunities (Baulch et al. 2010). In
the Central Highlands, local enterprises require upper secondary diplomas for most industrial jobs,
leading to the exclusion of indigenous minorities from a wide range of possibilities (Truong 2011).

Figure 5.3 Changes in Net School Enrolment Rates for Kinh and Ethnic Minorities
in Rural Areas, 1998-2010

100
2010
90
1998

0
Ethnic minority Kinh Ethnic minority Kinh Ethnic minority Kinh

Primary Lower Secondary Upper Secondary

Sources: 1999 VLSS; 2010 VHLSS.

127
5.16 Analysis of school enrolment rates from the 2009 Population Census shows that certain ethnic
groups, including the Hoa, Nung, and Tay, have primary and lower secondary school net enrolment
rates that are equal to or slightly higher than those of the Kinh (�?gure 5.4). In contrast, other ethnic
groups have net primary enrolment rates of less than 85 percent and lower secondary rates under
50 percent, notably the Hmong, whose primary enrolment rate of 69.6 percent is nevertheless
substantially higher than the 41.5 percent recorded in 1999. Primary school enrolment in the Central
Highlands has also increased signi�?cantly since 1999. By the upper secondary level, only the Kinh,
Hoa, and Tay have net enrolment rates greater than 50 percent, with 21 groups enrolling less than
10 percent of children in upper secondary school (Baulch and Vu 2012).

Figure 5.4 Net School Enrolment of Selected Ethnic Minority Groups, 2009
100%

90%

80%

70%
Kinh
Net Enrolment Rate (%)

60% Tay
Hmong
50%
Ede
Khmer
40%
Majority, rural
Minori es, rural
30%

20%

10%

0%
Primary Lower Secondary Upper Secondary

Source: 2009 Housing and Population Census.

5.17 Patterns of improved services and remaining inequalities may also be observed in access
to public utilities. Coverage has improved since 2004 for both majority and minority groups in rural
areas, but access for ethnic minorities is still unequal in terms of access to improved water, improved
sanitation facilities, and electricity. Differential access is particularly stark for sanitation, where in 2010
around seven out of 10 majority households had access to improved sanitation facilities compared to
fewer than two out of 10 for minorities. In contrast, more than two-thirds of ethnic minority households
had access to an improved water source in 2010, with the Khmer and Cham having better access
than the majority. This dramatic increase in access to improved water by minorities since 2004 may
be partly attributed to Program 134, which, along with distributing land and building houses for ethnic
minority households, had a clean-water component.

5.18 Improvements in access to improved water and sanitation have contributed to better nutrition
among children. Drawing on anthropometric data from the 1998 VLSS, the 2006 VHLSS, and the
2010 Multiple Indicator Cluster survey (GSO, UNICEF, and UNFPA 2011), stunting (low height for
age) has fallen rapidly and consistently among the rural Kinh from 49.5 percent of children aged
0–5 in 1998 to 23.3 percent in 2010. Meanwhile, stunting among minority children has fallen from

128
essentially the same level (48.7 percent) as the Kinh in 1998 to 42.3 percent in 2010, with a slight
rise in 2006 (�?gure 5.5).33

Figure 5.5 Stunting among Children under Age 5 in Rural Areas, 1998-2010

Source: 1998 VLSS, 2006 VHLSS, 2010 MICS.

5.19 Wasting (low weight for height)34 is a short-term measure of nutritional status that is often
seasonally dependent. It also records a decline from 1998 to 2010, although with very small changes
between 1998 and 2006. Children under 5 years of age from both the majority and minority started
with similar levels of wasting (just under 12 percent) in 1998, with wasting declining to 3.9 percent
among majority children compared to 5.5 percent among minority children by 2010. The stunting and
wasting statistics provide evidence of a widening gap in nutrition of majority and minority children.

5.20 Investment in rural electri�?cation during the 2000s has improved access to grid electricity to
near universal levels for the majority, but over a quarter of ethnic minorities rely on other sources of
power for their main source of lighting (table 5.2). Access to electri�?cation in the Central Highlands is
greater than in the northern mountains, even though both are upland areas with signi�?cant hydropower
resources.

33 Due to sample size considerations and less detailed ethnic codes in the 2010 MICS, it is not possible to disaggregate
these nutritional results into the �?ve broad ethnic categories used earlier. However, stunting (and wasting) is generally
lower among the (better off) Tay-Thai-Muong-Nung category.
34 De�?ned as weight for height z-scores less than two standard deviations from the 2006 World Health Organization child
growth standards.

129
Table 5.2 Access to Public Utilities by Ethnicity in Rural Areas, 2004-10

Percent of Households with Access to:

Improved Water Improved Sanitation
Electricity Grid
Facilities
Ethnic Category 2004 2010 2004 2010 2004 2010

Kinh and Hoa 89.1 90.9 46.8 69.2 94.5 98.9
All Ethnic Minoritiesa 53.3 69.6 9.9 18.4 72.5 83.2
Khmer-Cham 85.9 93.6 5.5 13.8 69.0 84.2
Tay-Thai-Muong-Nung 52.0 68.8 13.4 23.6 74.0 87.4
Other Northern Mountains 37.1 64.2 8.2 12.0 56.0 61.5
Central Highlands 51.3 67.0 4.6 13.7 80.5 91.9

Source: 2004 and 2010 VHLSS.
Note: a. Excluding Hoa.

5.21 In addition to intergroup and geographic differences, ethnicities are also internally
heterogeneous. Hmong in one district of Lao Cai employ different livelihood strategies and cultural
practices from Hmong in another, and the range of practices among Hmong within a single district
overlaps with practices of other ethnic groups. Even within a single commune, there are frequently
signi�?cant differences in poverty rates among villages. In light of this diversity, poverty reduction
and development programs that target “extremely dif�?cult�? geographic areas, or all ethnic minorities
as an undifferentiated group, will inevitably bene�?t some populations more than others. Findings
from the 2010 VHLSS indicate that this may be taking place. The mean ethnic expenditure gap is
increasing at all levels of income except the highest sixth, where it has decreased slightly since 2004.
Although some of the disparities are explainable by commune characteristics, much of the difference
in returns to endowments faced by ethnic minorities depends on unobserved factors such as the
quality of education or land combined with discrimination in access to employment and markets
(Baulch and Vu 2012).

5.22 New research on “perceptions of inequality�? carried out for this report suggests that ethnic
inequality is one component of broader income and social inequalities (Hoang et al. 2012). Focus
groups of ethnic minority youth, senior citizens, and local leaders emphasized livelihood-related
modalities of inequality in terms of access to market, credit, and agricultural services. In rural areas
such as Chieng Khoa commune, Son La, there was perceived to be little inequality within ethnic
minority communities, since agricultural production remains the key source of livelihood. However,
the transition to a commodity-based economy is seen as a source of growing inequality.

5.23 Ethnic minority focus groups identi�?ed inequalities of opportunity when comparing their
communities with better-off towns nearby. The disparities noted link to the six “pillars of disadvantage�?
(box 5.1) and are perceived as linked; that is, poor infrastructure leads to poor education, poor
employment, and then poor access to markets and services. Although some of these structural
disadvantages can be corrected by policy measures, they continue to play an important role in
keeping many ethnic minorities from earning a better living.

5.24 Agricultural land disparity is perceived as very important in determining outcome inequality
in the rural mountainous ethnic minority areas of Son La and Quang Nam provinces, where off-
farm employment and migration are negligible (Hoang et al. 2012). In Son La, rice paddy land was
equally allocated among Thai households in the early 1990s. Better-off households expanded their
rice �?elds by reclaiming vacant land, but such land is no longer available. The more important source
of land disparity is in sloping land for maize and tea farming. Well-established households have

130
large holdings, while newly separated households and newcomers have little land and are often
considered poor. In Quang Nam, by contrast, the Ve people (a branch of the Gie Trieng ethnic group)
do not see land disparity as a key driver of increased wealth disparity, which results instead from
livestock ownership and access to public sector employment. Ve households can still expand their
cultivated area based on the availability of labor.

5.25 The perceptions of inequality study found little concern about interethnic inequalities. Thai
people in Son La admit that they are more advantaged than Hmong people in terms of access to
infrastructure, education, and markets, but feel disadvantaged in terms of land quality and quantity.
These differences appear to be decreasing over time, in part due to government investment in
infrastructure. Similarly, commune of�?cials in Quang Nam draw comparisons between the Ve people
and the larger Co Tu group, who live in more central parts of the district with better access to markets
and employment.

5.26 However, many minority respondents raised concerns about unfair behavior of the Kinh
toward ethnic minorities. Such behavior and related prejudice was widely perceived to have serious
implications for social unity and ethnic cohesion. Minority youth living nearer provincial towns and
cities experience ethnic discrimination in their schooling, employment, and social relations, as in this
example of a young Nung woman in Lao Cai:

“As we [people from the ethnic minorities] can be recognized by clothing, the way
medical staff treat people from the ethnic minorities is different from the way they treat
Kinh people. They [doctors] don’t treat us well … In the market, Kinh people who are
cleverer usually get good bargains … In a bus, their (Kinh) prejudice towards us is
demonstrated through language and intonation, shouting with disrespectful words.�?

5.27 Kinh focus groups in the inequality perceptions study, by contrast, denied that they discriminate
against ethnic minority groups, and many believe that minorities receive special bene�?ts. A Kinh
student in Quang Nam stated:

“We don’t think we are superior to the ethnic [minority] classmates. They are receiving
preferential treatments such as subsidies and scoring incentives. Perhaps they
themselves feel inferiority; there is no discrimination from us.�?

D. The Experiences of Ethnic Households that have already Escaped
Poverty Offer Lessons and an Innovative Orientation for Future Policies and
Programs

5.28 The Vietnamese government, with World Bank and donor support, has implemented an array
of economic policies since the 1990s, such as land reforms, infrastructure investments, education
and vocational training projects, and agricultural commercialization efforts. These policies have
brought many Vietnamese into the growth process and have succeeded in reducing poverty among
the Kinh more than twice as rapidly as among ethnic minorities (Pham 2009). The remaining poor are
thought to be less easy to help (DFID and UNDP 2003; Oxfam and ActionAid 2008). This situation
has led to pessimism about the likely effectiveness of future development programs, mixed with
reinforcement of stereotyping of ethnic minorities as culturally “backward�? (lac hau), uneducated,
and unwilling to develop themselves. Meanwhile, anthropologists and other external observers have
criticized the Vietnamese government and donor agencies for perceived assimilationist policies
leading to a decline in cultural identity among ethnic minority groups (McElwee 2004; Taylor 2004).
Although government of�?cials, donors, and academics may reach divergent conclusions, they share
a common constraint-based approach to analysis, looking for what is wrong with a situation and how
it can be �?xed.

5.29 In background research for this Poverty Assessment, Wells-Dang and Nguyen (2012) adopted a
contrasting approach of identifying communities that are succeeding where others are not, and sought

131
to understand the reasons behind their success. This approach, which bears some similarities to
methodologies of “positive deviance�? applied worldwide in health and business management sectors,
aims to build con�?dence and social interactions among participants and points toward effective future
project and policy interventions, something that a constraint approach is unlikely to do (Marsh et al.
2004; Ramalingam 2011). The research presumes that ethnic people are actively engaged in their
own development as “innovative constructive agents … not as resistance to domination, but as a
logical or obvious response to new opportunities�? (Sowerwine 2011).

5.30 Based on an analysis of census data on ethnic minority poverty reduction and expenditures,
the research team selected �?eld visit sites in Dak Lak, Lao Cai, and Tra Vinh provinces and sought to
identify villages, or ethnic groups within a commune, that show uncommonly positive results in ethnic
minority development and poverty reduction. All three provinces have been included in previous
studies of ethnic minority poverty; Dak Lak was one of four provinces visited in the “Country Social
Analysis�? (World Bank 2009). Tra Vinh and Lao Cai were both included in the 1999 Participatory
Poverty Assessments conducted by the World Bank and a group of international nongovernmental
organizations (NGOs) (Oxfam 1999; World Bank 1999). It is also remarkable that both Lao Cai
(ranked second of 63 provinces in 2010) and Tra Vinh (ranked fourth) score highly on the Provincial
Competitiveness Index of business and investment criteria (USAID 2011).

E. Ethnic Minority Poverty Reduction begins with an Agricultural
Transformation from Semi subsistence to Commercial Production

5.31 Agriculture is the primary livelihood activity for ethnic minority communities in all three sites, as
well as generally across Vietnam (World Bank 2009). In most communes visited for this study, 80 to
90 percent of households were involved in agriculture. Thus, any program of ethnic minority poverty
reduction must include a strong agricultural component. Ethnic minority farmers have land holdings
equivalent to or even higher than the average land holdings of Kinh, but their land is of variable
quality (World Bank 2009, 113). In the Central Highlands, a coffee farmer with as little as 0.25 hectare
of high-quality land can earn above the poverty line for a family of �?ve. Vegetable and fruit growers in
other provinces require approximately double this amount of land to reach the same income level.

5.32 Farmers with suf�?cient, quality land have multiple options to escape poverty. Those with less
land can only do so through high-value cash crops, the opportunities for which depend on local soil
and weather conditions. A third group of households, found mainly in the Mekong Delta, lost their
land through indebtedness or sale. These families have mostly migrated or shifted to nonagricultural
work, although some continue as agricultural wage laborers. Landlessness is no longer viewed as the
crisis it was in the 1990s, given the increased availability of nonagricultural work and the possibility of
migration.

5.33 Cash crop farmers are highly dependent on local and world market prices for their commodities.
In this sense, they are already connected to the global economy, not at all “remote�? (vung sau vung
xa), as perceived by many urban Vietnamese (Taylor 2007). Coffee and other commodity farmers
sell their crops to dealers (who are mostly Kinh), who then resell to export processing facilities in
provincial cities. Ethnic minority farmers do not know where their crops are exported, but they do
follow market prices, which are broadcast on television and radio, printed in newspapers, and posted
at local of�?ces. Cash crop farmers in border areas export their products directly or via ethnic and Kinh
middlemen (box 5.2).

5.34 Since previous research on ethnic minority development (ADB 2003; Oxfam 1999; Oxfam and
ActionAid 2008; World Bank 2009), certain key features of the agricultural economy have improved.
One of these aspects is price information, mentioned above. Another is better access to credit, via
the Social Policy Bank (Ngan hang Chinh sach) and the Vietnam Bank for Agriculture (Ngan hang
Nong nghiep). According to data from the 2010 Vietnam Household Living Survey, 32.6 percent of all
rural ethnic minority households and 52 percent of poor ethnic minority households have access to

132
preferential loans from the Vietnam Social Policy Bank and other sources compared to 10.4 percent
of all rural Kinh and 35.2 percent of poor Kinh. In communes visited during background research for
the Poverty Assessment, access to loans for ethnic households reached up to 80 percent. Loans are
often channeled through local mass organizations; loan amounts have increased to a maximum of
VND 30 million (US$1,500) compared with 3 million VND to 5 million VND noted in the Country Social
Analysis (World Bank 2009, 148).

5.35 Most respondents report using loans for purchasing seeds, raising animals, or small business
activities, such as purchasing goods for a market stall. Borrowers through mass organizations
receive some instruction and support for their stated use of the loan, such as agricultural extension or
animal raising. Formal and informal farmers’ groups play a signi�?cant role in agricultural production,
particularly among Khmer in Tra Vinh. These cooperative groups (to hop tac) exchange price and
technical information, produce cash crops cooperatively for �?xed-price contracts, and link poor and
better-off farmers in a community.

5.36 Ethnic minority farmers are skilled at producing crops, raising animals, and other agricultural
activities. However, their relative position in the market has weakened over time; many of the
bene�?ts of economic growth have accrued to better off households and those working in industrial
and commercial activities. (Chapter 6) Few ethnic minorities are represented in these groups. The
lower relative returns to agriculture are in part a result of policy decisions that have a disproportionate
effect on ethnic minorities. Future growth in agricultural livelihoods is also threatened by risks and
vulnerabilities such as changes in commodity market prices, natural disasters, climate change, and
environmental degradation.

Box 5. 2 An Ede Coffee “Hotspot�?

Ede are the largest indigenous ethnic group in Dak Lak, although they make up less than 20
percent of the total population. Before waves of migration after the Vietnam War, Ede were the
only residents of Ea Khal commune, extending 20 kilometers westward from the provincial town
of Ea Drang. Now there are 16 villages in the commune, of which only two are indigenous Ede.
One of these is Buon Dung, about 2 kilometers from the commune center, an Ede village with
high incomes from coffee and other crops. According to commune statistics, overall poverty rates
in 2011 were 23 percent for Ede, 34 percent for other ethnic minority in-migrants, and 16 percent
for Kinh. In Dak Lak province overall, 50 percent of ethnic minorities are considered poor. Thus,
Ede in Ea Khal are less than half as poor as average ethnic communities in the province.

Young Ede coffee farming families in Buon Dung have between 1 and 4 hectares of good-quality
�?elds and are accessing large, high-interest loans from the Vietnam Bank for Agriculture. They
have taken part in technical training on coffee production organized by agricultural extension
services or the Farmer’s Union. Cognizant of the risks in coffee production, they monitor prices
carefully to get the best return for their crop. The village also has storage and drying facilities, so
farmers can wait until prices are high before selling.

After several years of good harvests, families are investing their pro�?ts in additional land
purchases in neighboring villages and in construction of new houses in a mixture of traditional
Ede and Kinh styles. The reasons for their relative prosperity include access to land, social
cohesion, and preferential treatment of indigenous minorities by local authorities.

133
F. Successful Ethnic Farmers are Beginning to Diversify into Nonagricultural
Employment, Particularly in Areas with Access to Major Cities or International
Markets
5.37 Diversi�?cation is a key, though not universal, feature of ethnic minority livelihood strategies,
moving from subsistence production to a multiplicity of activities and income sources (Minot et al.
2006; Shanks et al. 2012, 51). Agricultural work remains the norm for the majority of ethnic minority
families, but most respondents plant multiple crops—grain in the wet season and vegetables in the
dry season, a combination of hybrid and traditional rice and maize seeds, or a mixture of export
cash crops. Almost all ethnic households raise some animals for household use or sale. Of families
pursuing nonagricultural livelihoods, such as factory work, trading, or tourism, nearly 100 percent
maintain some tie to agriculture, at a minimum through usufruct rights of leased land. With the
exception of a few large export dealers, ethnic minorities view handicrafts, tourism, trading, and
other service employment as a complement to agriculture. This strategy of “selective diversi�?cation�?
simultaneously allows for cultural preservation and higher incomes (Turner and Michaud 2011).

5.38 The involvement of ethnic minorities in nonagricultural work varies from very little in Dak Lak
and modest in Lao Cai to signi�?cant in Tra Vinh, where Khmer are involved in all kinds of trading,
services, and industrial jobs. Factory work has become available in Tra Vinh since 2007 and now
employs 30,000 workers province-wide, primarily women under age 35. Base salaries in such
factories are substantially lower than in Ho Chi Minh City, but living costs are also lower by a factor
of a third or more. For some Khmer families, industrial work offers a stable income and a way out of
poverty even for a family with little (or no) land. Respondents said they preferred to stay in their own
communities rather than migrate for industrial work, even though local salaries are lower.

5.39 Local ethnic minority traders in Muong Khuong, Lao Cai, use their comparative advantages
of a location on the Chinese border, relationships with relatives and others of the same ethnic group
across the border, and knowledge of the regional Chinese dialect Quan Hoa. One young Hmong man
who had spent several years as a laborer in China is now trading mineral ore and other products
across the border, earning enough to purchase a private car. A Phu La-Nung couple in another
village began by trading rice and corn in local markets, then took advantage of available loan capital
and switched to pineapple growing in 2009 (box 5.3). In these cases, ethnic minorities are no longer
only clients of Kinh private traders, as was the case a decade ago (DFID and UNDP 2003). Their
involvement in business contributes to a leveling of opportunities and information, as shown by
a decrease in complaints by ethnic minorities about being cheated or unfairly treated in market
transactions with the Kinh. Near border areas, ethnic minorities may have more trading connections
than Kinh do. Ethnic business owners are also more likely to employ minority staff, adding to the
limited job opportunities in the local private sector.

5.40 Figure 5.6 describes the sources of income of Kinh and minorities in rural areas based on the
2010 VHLSS. Apart from the substantial difference in overall household incomes, the �?gure reveals
three outstanding factors (Baulch and Vu 2012). First, nonagricultural wages make up a much smaller
part of ethnic minority income compared to Kinh. This was true even controlling for income; poor Kinh
have more diversi�?ed earnings and income portfolios than poor minorities (Chapter 3). Second,
minority households earn very little from nonfarm enterprises, consistent with the dominance of Kinh
traders found especially in the Northern Mountains (Wells-Dang 2012; World Bank 2009). Finally,
income transfers, including private remittances and public programs, are considerably lower among
minority households, a result of lower domestic migration and access to public services (Baulch et
al. 2010).

134
Box 5.3 Pineapples along the Border

Na Loc, a cluster of seven villages in Ban Lau commune, Muong Khuong district, Lao Cai, extends
through a narrow valley on one side of a small stream: the Chinese border. Hmong farmers in
Na Loc have long had close links to the Chinese market. In the 1990s, three men crossed into
China to work as wage laborers and brought back techniques of pineapple cultivation that they
introduced to other villagers. One of the �?rst pineapple growers later became a village chief.

Na Loc villagers have gained high pro�?ts from pineapple for over 15 years, earning incomes of 150
million VND (US$7,500) per year or more. Since around 2005, cash crop production has spread
from Na Loc to other villages in Ban Lau commune. Almost all land in the commune, including
steep hillsides, has been converted to pineapple, banana, and tea production. Returns were
high until 2012, when Chinese buyers suddenly stopped purchasing pineapple and Vietnamese
market prices plunged to as low as 1,000 VND (US$0.05) per kilogram. Farmers in Na Loc are
now struggling to break even, but most are suf�?ciently diversi�?ed and have accumulated enough
savings that they believe they can ride out the downturn.

This experience, like that of coffee in the Central Highlands, shows that long-term poverty
reduction cannot depend on a single commodity.

Figure 5.6 Sources of Income for Majority and Minority Households in Rural Areas, 2010

Kinh & Hoa Minorities
60
Household income (VND millions)
40
20
0

Crops Livestock Aquaculture
Forestry Ag Wages Non-Ag Wages
Non-Farm Ent Transfers Other

Source: 2010 VHLSS.

135
5.41 Income sources vary across the distribution for minority households (�?gure 5.7). Crop incomes
almost double from the poorest to the richest quintile, while nonagricultural wages increase by a factor
of 10. Income from forestry, aquaculture, and agricultural wages remains roughly constant across
quintiles and does not contribute signi�?cantly to income gains. Income from nonfarm enterprises is
negligible for quintiles 1 and 2, and then expands rapidly in the top three quintiles. These patterns are
broadly consistent with the patterns of diversi�?cation identi�?ed in qualitative research, showing that
rural households generate a surplus from agriculture �?rst before investing in a nonfarm enterprise. For
the richest quintile, transfers (in particular remittances) are also important, since households at this
income level may have family members working in cities, government jobs, or other nonagricultural
positions.

Figure 5.7 Sources of Income by Quintile for Minority Households in Rural Areas, 2010
60
Household income (VND millions)
20 40
0

1 2 3 4 5
Crops Livestock Aquaculture
Forestry Ag Wages Non-Ag Wages
Non-Farm Ent Transfers Other

Source: 2010 VHLSS.

5.42 The data on sources of income and diversi�?cation suggest that minority households generally
earn a relatively small share of their incomes from nonagricultural wage employment. This is
principally because ethnic minority workers �?nd it more dif�?cult to obtain wage jobs than the majority,
but differences in wage rates also play a role. In 2010, 28.8 percent of ethnic minority households
had wage workers compared to 60.5 percent for the majority. Ethnic minority workers in rural areas
earn on average 13.8 percent less than Kinh workers, and gaps remain even after controlling for
education and sector of employment. While some of this differential may be attributable to differences
in education and experience, wage differentials are also substantial for workers with secondary
education or university quali�?cations.

G. Most Ethnic Minorities Continue to Live in their Communities of Origin

5.43 In the Central Highlands and Northern Mountains, there are few cases of young indigenous
minorities migrating to cities for industrial work. Migration from the north to the Central Highlands
has also slowed. Provincial of�?cials stated that a majority of ethnic migrants who had gone to work
in urban factories in the past �?ve years have returned home for a combination of economic and
cultural reasons. In most instances, the wages available are relatively low. Ethnic minority informants,

136
including some returned migrants, stated that they preferred to stay in their communities and do not
feel con�?dent or comfortable in large cities. The reasons given for the low levels of out-migration
are that agricultural work is available locally, net returns from work in cities are not much higher,
and living far away from home is not culturally comfortable. If more industrial and service jobs were
available locally, informants indicated that they would be willing to work in these sectors.

5.44 Out-migration of ethnic minorities is a signi�?cant pattern only in the Mekong Delta. According
to Tra Vinh of�?cials, there are now 80,000 workers from the province in and around Ho Chi Minh City,
about half of them Khmer. Both poorer and better-off Khmer practice migration as a strategy, but for
different purposes. Those with large land holdings (or established nonagricultural businesses) send
their children to urban areas for education and subsequent entry into white-collar professions such
as teaching, business management, and public sector employment. The land-poor and landless, by
contrast, migrate for employment and survival, acquiring skills and knowledge in the process that raise
their incomes over the poverty line, but at a social cost of distance from their home communities.

5.45 Many poor and landless young people, especially women, move to the city to look for work
when they reach adulthood. The pace of migration has remained relatively constant in recent years,
with few migrants returning to the Delta permanently (Oxfam and ActionAid 2009). Given the high
cost of living in the city, few workers are able to send much money back to their families. Migration
is thus more an employment strategy than a source of remittances. Without the safety valve of
migration, land holdings would be divided into smaller pieces and it would become more dif�?cult to
�?nd nonagricultural work nearby. Local of�?cials do not view migration as a problem, but rather as one
of a number of livelihood strategies practiced by local households.

H. Ethnic Minority Poverty Reduction Strategies Follow a Series of Steps
from Agricultural Specialization to Diversi�?cation and Accumulation of
Financial, Social, and Cultural Capital

5.46 Despite regional and cultural diversity, ethnic minorities in Vietnam share certain important
characteristics in common. They all reside in the same nation-state, with the same national policies
and structures; they all largely practice agriculture; and all must de�?ne and maintain their identities
in relation to a much larger ethnic majority group that controls most important political, economic,
and social institutions. To escape poverty in these conditions, ethnic minorities �?rst shift from semi-
subsistence agriculture to a market orientation, then make efforts to maintain their cultural identity
while building �?nancial and social capital. This process, outlined in �?gure 5.8, has four main steps
toward success, with agricultural and nonagricultural branches.

Figure 5.8 Paths to Successful Ethnic Minority Development

Step 3a.
Agricultural
diversification
Step 1. Step 2. Step 4.
Cash crop production Intensive agriculture Consolidation,
investment in
education
Step 3b.
Trading and services

Source: Wells-Dang 2012.

5.47 In step 1, poor households with average land holdings and land quality shift part of their
available land (or one planting season) away from semi subsistence grain production and begin
planting a cash crop. In Dak Lak, this is usually coffee or sometimes pepper; in other locations,

137
vegetables and fruit are common cash crops. The key requirements for cash crop production are
capital to purchase fertilizers, water for irrigation, and technical knowledge to achieve a decent yield.
Many households meet part of the initial capital outlay through a loan from the Social Policy Bank,
supplemented by no-interest loans from relatives and community members, as well as support from
other government programs. However, ﬂuctuating prices and climate conditions pose serious risks to
getting started in cash crop production. Many families who are no longer classi�?ed as poor are still
not yet con�?dent of staying out of poverty in future years. According to a Jarai village chief in Ea H’leo
district, Dak Lak, it takes a family about �?ve years of small-scale cash crop production to achieve this
con�?dence.

5.48 Once households amass some savings and experience in cash crop production, they next
take the greater risk of concentrating their effort on a single product. This step requires a quantum
leap into a fully marketized economy. These farmers have bought or leased small amounts of
additional land where possible, even if far from their homes. Using this land as collateral, they begin
to access higher-interest loans from the Vietnam Bank for Agriculture, although some continue to
renew loans from the Social Policy Bank (some of which are open to ethnic minority borrowers
regardless of poverty). They take part in technical training organized by agricultural extension
services or the Farmer’s Union. Compared to the farmers at step 1, they monitor prices carefully to
get the best return for their crops and are highly conscious of price risks; the cost of failure would
be extremely high.

5.49 In the agricultural variation of step 3, farmers who have achieved higher incomes from cash
crop production—around VND 100 million per year for a family of �?ve, or a per-capita income near
the national average of US$1,000—then take steps to reduce risk by diversifying into other crops
or into larger-scale animal raising. Aquaculture, forestry, or tree plantations such as rubber are
additional options for diversi�?cation in some areas for those with enough capital to purchase larger
tracts of land and the ability to wait �?ve or more years for returns. Households at this level have
above-average landholdings and are eligible for larger loans from the Bank for Agriculture, although
some have enough savings to avoid the need for loans. As experienced, successful farmers, they are
well-known and respected members of their communities with good connections with commune- and
district-level authorities.

5.50 Relatively few ethnic minorities have pursued step 3b in the diversi�?cation strategy model, to
move into trading and services; those with signi�?cant nonagricultural income are typically located
in the top income quintile (�?gure 5.7). Of ethnic minority households that do select nonagricultural
diversi�?cation strategies, most are already successful commercial farmers �?rst. They begin off-farm
business activities by selling their own or neighbors’ agricultural products at markets, then investing
in a truck or small shop. After gaining experience and con�?dence, some traders and shop owners
drop their involvement in agriculture entirely and concentrate fully on their new business. Others
continue to be involved in both sectors. Once trading or service business becomes the primary
livelihood of the household, �?elds are typically leased out or workers are hired to grow rice or corn,
rather than more intensive cash-cropping. Families at this level receive (and require) little support
from government programs.

5.51 The small number of ethnic minority households that reach step 4 in �?gure 5.8 have resources
and savings above the national average. As their children approach adulthood, older farmers
consolidate their status and further reduce risk by sending children for secondary and higher education
in provincial cities or beyond. After graduation, children are then expected to get nonagricultural jobs
to contribute to the family income. In most observed cases, children had not yet begun sending any
funds back to their parents, but the presence of nonagricultural work balances the risk of the family
farm or small business. Even among the most prosperous minorities, the researchers did not see

138
strong evidence of cultural assimilation at the village level; ethnic minority communities remain as
distinct villages, with local languages spoken and social structures persisting. These results concur
with �?ndings from research in the northern mountains that identify “some models of development
based on local knowledge that have reduced poverty and even made some people rich, while still
preserving the value of traditional culture and the local environment�? (Mai, Le, and Le 2011, 55–6).
However, an unanswered question is how the lives of youth who access education in the cities will
change in the future—whether it will be toward absorption into mainstream Kinh society or toward a
renewed sense of ethnic identity.

5.52 Government programs are particularly important for households below or slightly above the
poverty line, as a source of capital and livelihood inputs. No single program has been most effective
at poverty reduction; instead, ethnic minority respondents point to the combination and interplay of
several programs providing low-interest credit, infrastructure, housing, and cash transfers, and the
role of farmers’ cooperative groups. Existing credit and extension services are mainly targeted to
households with agricultural land; animal-raising training is an important exception. Land is held as
collateral for interest-bearing loans. Most households that have bene�?ted from Decree 167, which
allocates land to the landless, have received residential land only; very few have received scarce
agricultural land. Many of the changes brought about by these programs have taken effect since
2006, due to improved targeting of programs, greater availability of funds, and the bene�?ts of higher
market prices for agricultural products, among other possible factors.

5.53 Other government programs, including forestry, labor export, and vocational training, were
assessed by interview respondents as contributing less to ethnic minority development and poverty
reduction. The vocational training courses available from the local government are not yet well
matched with market demand; as many as half of trainees have dif�?culty using skills after completing
training. Training in local languages is available in only a few locations, such as the Women’s Union in
Bac Ha district, Lao Cai, which uses Hmong staff in majority Hmong areas to reach its membership.

5.54 When asked about dreams for their children’s careers, parents across all ethnic minority
groups said that they hoped their children would get an upper secondary or higher education and
then a job in the state sector as a teacher or public of�?cial. No one expressed a desire for children to
work in industry or business, with the exception of Khmer families already involved in trading in Tra
Vinh. In Dak Lak and Lao Cai, some industrial jobs are available near the provincial cities, but few
minorities work in these companies. In part this is because many do not meet the required educational
quali�?cations, but even if they do, they may be labeled as “lacking knowledge,�? part of the vicious
circle of ethnic disadvantage. Since there are few private sector jobs in many mountainous areas,
the thinking that “jobs are public jobs�? persists. However, the number of government jobs available is
also limited, so few young ethnic minorities who have completed secondary or higher education can
be assigned to government positions. According to a youth focus group of Ve people in Dak Pree
commune, Quang Nam,

“We have many graduates, but few of them �?nd jobs. I have seen many students who
had no choice but came back to farming work. The year 2011 alone saw eight graduates
from pedagogic schools, but only one of them could work on a �?xed-term contract basis
at the commune. The remaining seven students came back to farming work. It is not
possible to apply for jobs in other districts, as they also have enough staff.�? (Hoang et
al. 2012, 30)

139
I. Prevailing Narratives of Ethnic Minority Livelihoods, Cultures, and
Gender Relations are Shifting along with Diversi�?ed Development,
although some Stereotypes Persist

5.55 Interview respondents, both community members and local of�?cials, spoke of changing
attitudes toward ethnic minority capacities and cultures. In this narrative, Ede, Khmer, Hmong, and
other ethnic minorities are hard-working and serious, with high levels of intra-village cooperation.
In some cases, having a critical mass of a minority population, including adequate representation
in local leadership, was seen to promote greater equity (box 5.4). In Dak Lak and Tra Vinh, Kinh
of�?cials at the district and commune levels perceive a shift in ethnic minority work, savings habits,
and lifestyles over the past decade (although these characteristics might have been true previously).
Ethnic stereotyping was rarely heard of, and then most often in the past tense, sometimes from
ethnic minorities themselves, as in “we used to be backward.�? In Tra Vinh, for example, respondents
said that previously Khmer planted only rice and did not work in the dry season, but when more
opportunities became available, they adapted to cash crops and nonagricultural work. The local
explanations offered for this change were the opportunity to become better off through cash crop
production and the positive inﬂuence of education. The younger generation is becoming more literate
in Vietnamese than their parents. Yet the question remains, with prevailing cultural stereotypes,
whether or not formal education will lead to more employment opportunities in the future.

Box 5.4 Equity in the Khmer Heartland

Luong Hoa A commune in Chau Thanh district, Tra Vinh, is a majority Khmer community with
poverty levels that are average overall, but relatively equal between the two main ethnic groups.
Both Kinh and Khmer of�?cials spoke of equality, respect, and tolerance among ethnic groups. At
the provincial and district levels, this came across as the party line, but in the three communes,
relative equality is backed up by observations and data. In Luong Hoa A and other Khmer majority
communes, Khmer appear to be doing as well as Kinh, even though this is not true district- and
province-wide.

Among the factors leading to this success is, �?rst of all, a cohesive Khmer majority population that
is well represented in local leadership. In other words, the difference between Kinh and ethnic
minorities is smaller in areas with a greater concentration of ethnic minority residents. If poverty
is considered an “ethnic problem,�? then this is a counterintuitive �?nding. Conversely, Khmer are
relatively worse off in areas where Kinh are the majority. Where it is “normal�? to be Khmer, then
Khmer and Kinh appear to have relatively equal access to information and leadership positions.

5.56 A shift in gender patterns has accompanied the perceived cultural shift in work habits.
Families that have transitioned to market-based livelihoods appear to have adopted a more
equitable working style between husbands and wives. Women in trading families play important
roles in managing �?nances and interacting with customers. Men used to be the primary participants
in agricultural extension training and community meetings, but of�?cials and NGOs now report
greater participation of women; only when women are actively involved do livelihood habits change.
Women’s Union representatives mentioned positive impacts of credit and savings programs in
fostering participation, and a model of better-off women in a village cooperating to help one or
more poor women out of poverty.

5.57 The shift in ethnic minority livelihood patterns captured in the process of diversi�?cation and
consolidation has cultural and economic aspects. Embodied in the leap from semi subsistence to
commercial agriculture, this transformation is a consequence of the marketization and commodi�?cation
of upland products, land, and labor in a capitalist direction (Sikor 2011, 19). At the same time, it reﬂects

140
a conscious attempt by ethnic minority people to reimagine themselves as modern individuals, in
charge of their destinies and not conforming to old stereotypes.

5.58 Ethnic minority experience in poverty reduction is not fundamentally different from that of
Kinh in certain respects. Kinh have also entered into market relations and international markets,
although without some of the additional barriers and obstacles facing ethnic minorities. The fact that
minority group encounters with commodity markets and transnational social identities are occurring
in distinct places at different times means that outcomes of their transformations will be distinct,
not merely repetitions of Kinh experience. No single ethnic group (in Vietnam or elsewhere) has a
monopoly on particular livelihood strategies. To suggest that minorities who engage in trading or other
nonagricultural businesses are “acting like Kinh�? or “following a Kinh path to development�? is simply
another form of ethnocentrist prejudice. Although pressures for cultural and linguistic assimilation
are real, perhaps especially for some of the smallest minority groups, processes of poverty reduction
and development show that some ethnic minority communities have begun to prosper without losing
their identities. In fact, cohesive communities of people who are not poor have better chances of
maintaining their language, religions, and other cultural traditions than those who are struggling to
make a living.

5.59 This chapter has presented a mixed picture of ethnic minority development and poverty
reduction. Expenditure and income gaps between Kinh and minorities continue to increase, as do
gaps in important noneconomic measurements of welfare such as child nutrition. Yet, evidence also
indicates that some of the “pillars of disadvantage�? identi�?ed in the 2009 “Country Social Analysis�?
may be shrinking. Ethnic minorities have increasing access to education, credit, mobility, and markets,
which may take time to translate into higher incomes. Although it is beyond the scope of this report
to evaluate speci�?c Vietnamese government and donor-funded programs, it is clear that without
investments in schools, rural infrastructure, and �?nancial services, some of these changes would not
have been possible. At the same time, �?ndings discussed in previous chapters suggest that better
targeting and, more important, better coverage of poverty reduction policies and programs, would go
further to reduce the Kinh/ethnic minority poverty gap. Design is important as well. Effective programs
for ethnic minority poverty reduction must be targeted to address speci�?c factors of marginality and
build on positive examples of what ethnic households are already doing to improve their lives. Box
5.5 presents policy recommendations on reducing poverty among ethnic minorities.

141
Box 5.5 Emerging Policy Recommendations: Ethnic Minority Poverty

Recent research on ethnic minority development and poverty reduction in Vietnam, including
background papers for this Poverty Assessment, stresses the need for nuanced and targeted
policies, programs, and projects that address speci�?c needs of ethnic communities. Rather
than a standardized national approach to poverty reduction that may have been appropriate in
the past, current recommendations favor provincial or regional foci with components aimed at
disadvantaged groups in the population, such as youth, migrants, older women, or members
of one or more particular ethnicity. Activities should be based on evidence of success in one or
more ethnic minority area.

As important as the content of these interventions is their methodology. Policies and programs
should respect cultural norms while also seeking integration of ethnic minority communities with
local governance and social programs. Activities should be conducted bilingually where possible
and include local ethnic minorities as trainers and facilitators, as well as bene�?ciaries.

Among the concepts proposed for future initiatives are the following:
�? Business training for ethnic women (and men), such as Start and Improve Your
Business training
�? Expanded vocational training for youth, with an emphasis on skills with an identi�?ed
local market in the agricultural and nonagricultural sectors
�? Provision of credit, agricultural extension training, and market information to formal
and informal farmers’ groups, on a demand basis that responds to locally identi�?ed
needs
�? Scaling up of mother-tongue-based bilingual education in larger ethnic minority
languages, following the pilot conducted by Ministry of Education and Training and
UNICEF in Lao Cai, Gia Lai, and Tra Vinh
�? Incentives for responsible industrial development and local enterprise investment in
ethnic minority areas, providing diversi�?ed employment options without the social
costs of migration
�? Recruitment and capacity development of local ethnic leadership, in both formal
governance structures such as commune and district People’s Committees and
traditional village leaders
�? Greater involvement of local and international NGO projects in cooperation with
government and the private sector, such as through provincial innovation funds for
local social projects.

Sources: Shanks et al. 2012; Wells-Dang 2012; World Bank 2009.

142
References
ADB (Asian Development Bank). 2003. “Participatory Poverty and Governance Assessment: Central
Coast and Highlands Region.�? Asian Development Bank, Manila, October.
Baulch, B., and Vu Hoang Dat. 2012. “Exploring the Ethnic Dimensions of Poverty in Vietnam.�?
Background paper for the 2012 Poverty Assessment. World Bank, Washington, DC, May.
Baulch, B., H. Nguyen, P. Phuong, and H. Pham. 2010. “Ethnic Poverty in Vietnam.�? Chronic Poverty
Research Centre Working Paper No. 169, Manchester, UK, February.
Baulch, B., T. P. Pham, and B. Reilly. 2007, “Ethnicity and Household Welfare in Vietnam: Empirical
Evidence from 1993 to 2004.�? Mimeo, Institute of Development Studies, University of Sussex.
DFID and UNDP (Department for International Development and United Nations Development
Programme). 2003. “Poverty Reduction in the Northern Mountains: A Synthesis of Participatory
Poverty Assessments in Lao Cai and Ha Giang Province and Regional VHLSS Data.�? Department
for International Development and United Nations Development Programme, Hanoi, September.
GSO, UNICEF, and UNFPA (General Statistics Of�?ce of Vietnam, UNICEF, and United Nations
Population Fund). 2011. “Vietnam: Multiple Indicator Cluster Survey, 2010–2011: Final Report.�?
General Statistics Of�?ce of Vietnam, UNICEF, and United Nations Population Fund, Hanoi,
December.
Hoang, Xuan Thanh, Nguyen Thu Phuong, Vu Van Ngoc, Do Thi Quyen, Nguyen Thi Hoa, Dang
Thanh Hoa, and Nguyen Tam Giang. 2012. “Inequality Perception Study in Vietnam.�? Background
paper for the 2012 Vietnam Poverty Assessment. Ageless Consultants, Hanoi, May.
Imai, K., and R. Gaiha. 2007. “Poverty, Inequality, and Ethnic Minorities in Vietnam.�? Discussion
Paper Series EDP-0708, University of Manchester.
Kang, W. 2009. “Pro-poor Growth, Poverty, and Inequality in Rural Vietnam: The Welfare Gap between
the Ethnic Majority and Minority.�? Discussion Paper Series EDP-0906, University of Manchester.
Lanjouw, Marra, and Cuong Viet Nguyen. 2012. “Spatial Poverty and its Evolution in Vietnam: Insights
and Lessons for Policy from the 1999 and 2009 Vietnam Poverty Maps.�? Background paper for the
2012 Poverty Assessment, World Bank, Washington, DC, June.
Loa Cai DOLISA (Lao Cai Department of Labor, Invalids and Social Affairs). 2012. “Tinh hinh giam
ngheo doi voi nguoi dan toc thieu so�? (“Situation of Poverty Reduction for Ethnic Minorities�?). Report
prepared for World Bank delegation visit. Hanoi, February.
Mai Thanh Son, Le Dinh Phung, and Le Duc Thinh. 2011. “Bien doi khi hau: Tac dong, Kha nang ung
pho va mot so van de ve chinh sach (nghien cuu truong hop Dong bao cac dan toc thieu so vung
nui phia bac)�? (“Climate Change: Effects, Response Capacity and Some Policy Issues [Research on
Ethnic Minorities in the Northern Mountains]).�? Climate Change Working Group and Ethnic Minority
Working Group, Vietnam Union of Friendship Organizations-NGO Resource Center, Hanoi. http://
www.ngocentre.org.vn/content/comingo-vufo-and-paccom.
Marsh, D., D. Schroeder, K. Dearden, J. Sternin, and M. Sternin. 2004. “The Power of Positive
Deviance.�? British Medical Journal 329 (7475): 1177–1179.
McElwee, P. 2004. “Becoming Socialist or Becoming Kinh? Government Policies for Ethnic Minorities
in the Socialist Republic of Vietnam.�? In Civilizing the Margins: Southeast Asian Government Policies
for the Development of Minorities, ed. C. Duncan. Ithaca, NY: Cornell University Press, pp. 182–213.

McElwee, P. 2011. “‘Blood Relatives’ or Uneasy Neighbors? Kinh Migrant and Ethnic Minority
Relations in the Truong Son Mountains.�? In Minorities at Large; New Approaches to Minority Ethnicity
in Vietnam, ed. P. Taylor. Singapore: Institute of Southeast Asian Studies, pp. 81–116.

Minot, N., M. Epprecht, Tran Thi Tram Anh, and Le Quang Trung. 2006. “Income Diversi�?cation and

143
Poverty in the Northern Uplands of Vietnam.�? International Food Policy Research Institute Research
Report Abstract 145, Washington, DC.

Nguyen Viet Cuong, Peter Lanjouw, and Marleen Marra. 2012. “Vietnam’s Evolving Poverty Map:
Patterns and Implications for Policy.�? Background paper prepared for the 2012 Poverty Assessment,
Hanoi.

Oxfam Great Britain. 1999. “Participatory Poverty Assessment in Tra Vinh Province.�? Background
paper for the 2000 Vietnam Poverty Assessment, Hanoi.

Oxfam and ActionAid (Oxfam Great Britain and ActionAid Vietnam). 2009. “The Impacts of the Global
Financial Crisis on Socio-economic Groups in Vietnam.�? Monitoring report, Oxfam Great Britain and
ActionAid Vietnam. Hanoi, August.

Pham Anh Tuan. 2009. “Viet Nam Country Case Study: Background Paper for the Chronic Poverty
Report 2008–09.�? Manchester, UK: Chronic Poverty Research Centre.

Shanks, E., Duong Quoc Hung, Dao Ngoc Nga, Cao Thi Ly, and Bao Huy. 2012. “Central Highlands
of Viet Nam: Ethnic Minority Livelihoods, Local Governance Context, and Lesson-learning Study.�?
Report prepared for the World Bank. Mandala Consulting, Hanoi, April.

Sikor, T. 2011. “Introduction: Opening Boundaries.�? In Upland Transformations in Vietnam, ed. T. Sikor,
Nghiem Phuong Tuyen, J. Sowerwine, and J. Romm. Singapore: National University of Singapore
Press, 1–24.

Sowerwine, J. 2011. “The Politics of Highland Landscapes in Vietnamese Statecraft: (Re)Framing the
Dominant Environmental Imaginary.�? In Upland Transformations in Vietnam, ed. T. Sikor, Nghiem Phuong
Tuyen, J. Sowerwine, and J. Romm. Singapore: National University of Singapore Press, 51–72.

Taylor, P. 2004. “Introduction: Social Inequality in a Socialist State.�? In Social Inequality in Vietnam
and the Challenges to Reform, ed. P. Taylor. Singapore: Institute of Southeast Asian Studies, 1–40.

Taylor, P. 2007. “Poor Policies, Wealthy Peasants: Alternative Trajectories of Rural Development in
Vietnam.�? Journal of Vietnamese Studies 2 (2): 3–56.

Truong Huyen Chi. 2011. “‘They Think We Don’t Value Schooling’: Paradoxes of Education in the
Multi-Ethnic Central Highlands of Vietnam.�? In Education in Vietnam, ed. J. London. Singapore:
Institute of Southeast Asian Studies, 171–211.

Turner, S., and J. Michaud. 2011. “Imaginative and Adaptive Economic Strategies for Hmong Livelihoods
in Lao Cai Province, Northern Vietnam.�? In Minorities at Large; New Approaches to Minority Ethnicity
in Vietnam, ed. P. Taylor. Singapore: Institute of Southeast Asian Studies, 158–90.

USAID (U.S. Agency for International Development). 2011. “The 2010 Vietnam Provincial
Competitiveness Index: Promoting Economic Governance and Sustainable Investment.�? USAID/
VNCI Policy Paper #15, Hanoi.

Wells-Dang, Andrew. 2012. “Ethnic Minority Development in Vietnam: What Leads to Success?�?
Background paper prepared for the 2012 Poverty Assessment, Hanoi, April.

World Bank. 1999. “A Synthesis of Participatory Poverty Assessments from Four Sites in Viet Nam:
Lao Cai, Ha Tinh, Tra Vinh & Ho Chi Minh City.�? Hanoi, July.

World Bank. 2009. “Country Social Analysis: Ethnicity and Development in Vietnam.�? World Bank,
Washington, DC.

144
Chapter 6
Is Inequality Rising in Vietnam?
Perceptions and Empirics

Inequality is examined through two lenses - a qualitative study of
perceptions of inequality and a quantitative analysis. The chapter
documents widespread concerns across the population about
rising inequality. The qualitative study draws upon 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. 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.
Rising inequality is linked to growth processes in the service sector
and industry that have left some groups and regions behind.

145
A. Introduction

6.1 Over the last two decades, Vietnam has undergone rapid growth, substantial poverty reduction,
and economic transformation. Unlike other fast-growing economies, such as China and Indonesia,
past empirical work suggests that Vietnam’s extraordinary economic transformation has been one of
growth without an appreciable rise in inequality, a path similar to that of the Republic of Korea and
Taiwan during their early stages of development (ADB 2012; Hoang et al. 2010; McCaig, Benjamin,
and Brandt 2009; VASS 2011; World Bank 2009). Commonly used measures of inequality suggest
that inequality rose modestly during the 1990s and stabilized during the 2000s (Hoang et al. 2010;
VASS 2011).

6.2 Recent studies, including a major report on poverty prepared in 2010 by the Vietnamese
Academy of Social Sciences, note that relatively modest changes in empirical measures of inequality
based on household surveys stand in sharp contrast to widely shared perceptions among Vietnamese
people that inequality in incomes and wealth is rising (VASS 2011). The perception of rising inequality
is also notable in the press, among policy makers, and among academics in Vietnam.

6.3 This chapter examines inequality through two lenses: a qualitative study of “perceptions of
inequality,�? and a quantitative analysis that builds on lessons from the qualitative assessment.35
Examining inequality using both quantitative and qualitative tools gives a richer picture of the
inequalities in outcomes, opportunities, and social and political capital among Vietnamese people.
Inequality in outcomes refers to inequalities in income, consumption, and wealth, while inequality in
opportunities refers to differences in human capital driven by circumstances such as gender, ethnicity,
location, or parental characteristics. Inequality in social and political capital refers to differences
among individuals measured in terms of connections, voice, and inﬂuence.

6.4 The perceptions study helps to identify which types of inequalities are tolerated and which are
viewed as unacceptable in the eyes of Vietnamese people, and also captures inequalities that are
dif�?cult to measure in quantitative analysis, such as inequalities in connections, voice, and inﬂuence.
The quantitative assessment focuses on measuring changes in the distribution of outcomes and
opportunities over time, and on understanding the drivers of these changes using data from household
surveys, including the various rounds of the Vietnam Household Living Standards Survey (VHLSS).

6.5 The perceptions study suggests that Vietnamese people from all backgrounds—rural and
urban, rich and poor—think that inequality has risen substantially over the last �?ve years. Focus
group participants rarely discussed income or expenditure inequality in isolation, but instead linked
it to determinants—notably inequalities in education, access to good employment opportunities,
inequalities in access to land, and in connections, power, and inﬂuence. As such, inequality in
access to employment is often explained as a consequence of inequality in access to education,
and inequality in employment is then linked to inequalities in income, expenditures, and wealth.
Inequalities in power and connections are perceived as increasingly important in determining access
to jobs (transforming education into employment) or maintaining land rights. Despite perceived
rising inequalities in income and wealth, the majority of respondents consider inequality in outcomes
acceptable as long as it is generated through fair and legitimate means. The tolerance for income
inequality demonstrated by respondents in the perceptions study is a major shift in public attitudes
toward inequality compared to Vietnam’s pre-reform period.

6.6 Empirical evidence on inequality from the 2010 round of the VHLSS suggests a modest rise in
income inequality, driven primarily by growth in rural areas, where income from higher-value sideline

35 The perceptions study was led by a team from the Vietnamese Academy of Social Sciences and Ageless Consulting.

146
activities and nonagricultural income sources has been rising among better-off households. The
rise in income inequality is partly a reﬂection of growth processes that have altered the relative
return to assets such as education and productive capital in the economy. As such, the empirical
analysis suggests that growth has interacted with existing inequalities in opportunities—inequalities
in education, patterns of social exclusion between ethnic minorities and the majority, access to good
jobs, geographic disparities—to increase income inequality and income gaps between rich and poor
households.

B. A Step Back: Why are we Concerned about Inequality?

6.7 Should policy makers be concerned about rising inequality in income or expenditures? Whether
inequality in outcomes is likely to be a concern depends, in part, on the drivers and processes
that generate the inequality. It is useful to distinguish between “good�? and “bad�? processes and the
subsequent inequality created. “Good�? processes and inequalities are those that reward effort and
hard work, that reﬂect incentives to innovate, that stimulate entrepreneurship, and that provide the
impetus for economic growth.36 “Bad�? processes and inequalities can be considered to be those
that prevent certain segments of the population from bene�?ting from growth processes and from
transitioning out of poverty and low-income-generating activities.37 These inequalities often reﬂect
unequal opportunities for children born into certain circumstance, such as ethnicity, location, the
income or education level of the parents, or gender (Roemer 2011). They also reﬂect inequities in
connections, voice, and inﬂuence, where people from different backgrounds face different chances
of getting into a good university, acquiring a well-paying job, or of converting land because of their
backgrounds and circumstances. It is these second drivers of inequality—linked to inequalities in
opportunity and unequal process—that are most likely to be damaging for growth, social inclusion,
and societal tolerance for inequality in income and wealth (World Bank 2006).

6.8 The evidence suggests that the rise in income inequality seen in Vietnam since the mid-2000s
is the result of both “good�? and “bad�? processes. While a substantial fraction of the population has
contributed to the growth processes and has bene�?ted from growth, inequalities in opportunities
continue to repeat themselves across generations, and there is an increasing sense of unfairness
in processes such as access to public services, how jobs are acquired in the public sector, and how
land conversions occur.

6.9 The perceptions study provides us with a unique depiction of “good�? and “bad�? types of
inequality as seen through the eyes of Vietnamese people from a variety of backgrounds, including
young and old, rural migrants and long-term urban residents, workers in the informal sector and
the higher paid, formal sector employees, and minority populations and poorer individuals more
generally, particularly living in rural areas. In the perceptions study, focus group participants were
asked to categorize which forms of inequality were seen as more or less acceptable, and to explain
their views.

6.10 Rising income inequality was largely viewed as acceptable among study respondents if rising
disparities in incomes are associated with market-orientated growth-generating processes that
reward education, skills, hard work, and talent. The acceptability of inequality of incomes generated

36 We may also be concerned about rising inequality if there is a causal relationship between inequality and growth. While
there are many theoretical models that postulate a negative (and positive) relationship between inequality and growth, a
comprehensive assessment of the empirical literature suggests that the empirical evidence is inconclusive (Banerjee and
Duﬂo 2003; Bourguignon 2004; World Bank 2006).
37 These inequalities are linked to “pockets of poverty�? whereby certain groups in the population continue to remain in
poverty and poverty continues to perpetuate across generations, despite high average growth rates in the economy
(VASS 2008).

147
through legitimate means across all demographic and socioeconomic groups was considered by the
research team as a major shift in public attitudes with respect to inequality, away from the previous
focus on egalitarianism toward market-based mechanisms and incentives. As explained by two
interviewees,

“Disparity and competition is natural in a market-orientated economy. If you are talented,
you can be rich.�? (group of elder persons, Me Tri commune, Hanoi)

“Those who have talent and luck are conditioned to succeed. Those who have none
just suffer. I heard no complaint about inequalities. Such is reasonable.�? (village of�?cials
group, Cam Hung commune, Hai Duong)

6.11 The empirical evidence also suggests that inequalities generated due to reforms and structural
transformation partly reﬂect “good�? processes that are associated with economic momentum and
enhanced economic incentives. Since the Doi Moi reforms began in 1986, Vietnam has witnessed
a rapid economic transformation that has harnessed the power of market incentives to foster rapid
economic growth alongside strong poverty reduction. The rise in income inequality partly reﬂects
the process of structural transformation that has occurred since the reforms, which have shifted
labor away from agriculture and into the manufacturing and service sectors where value added per
worker is higher.38,39 The inequalities generated through these growth processes can be considered
inevitable in the sense that they are associated with a positive momentum in the economy and are
likely to encourage growth.

6.12 However, not all of the forces driving income inequality are perceived as “fair,�? and inequalities
in connections, voice, and inﬂuence are perceived by some to be rising. 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. Inequality in outcomes were widely accepted
by study participants, if the income or expenditure was generated through processes or sources that
were considered to be fair, while inequalities generated through illegitimate practices were seen as
unacceptable. For example, inequalities arising from differences in education, capital, hard work,
honest business practices, and luck were seen as acceptable in many focus groups, while those
generated through the illegitimate use of power or inﬂuence were unacceptable. As explained,

“There are types of illegitimate richness, and we do not accept these types, we see them
as being an injustice. For example, some traders sell seedlings to us at extremely high
prices. And corruption happens at all levels.�? (youth group, Chieng Khoa commune,
Son La)

“Without [unfair] power and connections the directors just differ from the workers by
some coef�?cient of basic salary. Because they have power and information, holding
important positions, doing businesses, they have used this to become much richer.�?
(long-term migrant group, An Son ward, Tam Ky city, Quang Nam)

6.13 Inequalities in opportunities imply that current differences in incomes will be perpetuated in
future generations unless the intergenerational links are broken. Therefore, the inequalities currently
seen in labor markets are likely to replicate themselves in the children of those who are unable to
take advantage of growth processes, and may result in groups that are already disproportionately

38 Vietnamese poverty reduction in the 1990s and early 2000s was driven in part by strong growth in the agricultural sector,
linked to the opening of agricultural markets from 1993 onward. The equitable distribution of land across the population
meant that this period of growth was broad based, and one that was accompanied by a substantial rise in income in poor
rural areas (Benjamin and Brandt 2002; Ravallion and van de Walle 2008).
39 In 2010, value added per worker in the manufacturing and service sectors was �?ve times higher than in agriculture (World
Bank calculations based on data from General Statistics Of�?ce statistical yearbooks).

148
poor falling even further behind. Although inequalities in educational attainment have narrowed
in recent years, particularly at the primary level, the educational attainment of children from poor
rural households remains low, particularly in some regions of the country (Chapter 3) and the
characteristics of the family a child is born into continue to be a strong predictor of whether a child
acquires secondary education and beyond. Therefore, the inequalities currently seen in income and
wealth are likely to replicate themselves in the children of those who are unable to take advantage of
growth processes, resulting in the intergenerational transmission of poverty and well-being.

C. Is Inequality of Outcomes Rising in Vietnam?

6.14 Past empirical work suggests that the fast growth seen in Vietnam over the last two decades
has not been accompanied by an appreciable rise in inequality. The Gini Coef�?cient of income
inequality remained fairly stable in the early 2000s (McCaig et al. 2009), and expenditure inequality
showed no appreciable rise on a national level (VASS 2011). According to a 2010 study led by a team
from the Vietnamese Academy of Social Sciences, the Gini Coef�?cient of expenditure inequality
increased from 0.33 to 0.35 between 1993 and 2002, but remained fairly stable between 2002 and
2008 (Hoang et al. 2010).

6.15 Empirical work done for this study suggests that income inequality has risen modestly since
2004, while inequality in expenditures remained stable between 2004 and 2010, according to several
commonly used measures of inequality. Findings from the perceptions study are, however, somewhat
at odds with the empirical picture of inequality emerging from the 2010 VHLSS. The perceptions
study �?nds that inequality in outcomes is widely perceived to have risen over the last �?ve years and
to have risen in both urban and rural areas. This section looks brieﬂy at the source of some of these
discrepancies.

6.16 Focus group respondents in both urban and rural areas reported that they perceived inequality
in outcomes—typically de�?ned using incomes, but also including spending on consumer durables
and assets—to have risen, and to have risen signi�?cantly in urban areas since 2005.

6.17 Perceptions of inequality were often but not always rooted in direct life experiences, and
varied across groups according to socioeconomic characteristics. Individuals tended to �?rst compare
themselves within their communities and then go one step further to compare themselves with slightly
better-off individuals or places. For example, low-skilled workers in Hai Duong and Ho Chi Minh City
compared themselves to higher-skilled workers, and individuals living in peri-urban areas in Hai
Duong, Hanoi, and Da Nang compared themselves with people living in inner-city areas. Those living
in urban environments tended to have broader frames of reference, and in these areas disparities
relating to conspicuous consumption (automobiles, high-end cell phones, large houses) were noted,
in particular.

6.18 Some focus groups in more remote and dif�?cult rural areas were less comfortable discussing
inequality of outcomes within their own communities than inequality within society, potentially due to
unease in singling out differences in closely knit communities with common disadvantages in location,
agricultural livelihoods, social and political capital, and other ethnic speci�?cations. Participants of
these focus groups appeared to be more at ease, however, when discussing inequalities beyond
their communities, and in particular inequalities in connections, voice, and inﬂuence.

6.19 Focus groups consisting of less educated individuals from poorer households saw disparities
related to substantially wealthier groups as being less important for their lives and showed limited
interest in comparing their situation with others in more favorable circumstances. For example, one
member of the migrant group in Da Nang city stated that,

“I feel it okay. I do not spend much and my earning is suf�?cient for my living. My life
might not be as good as theirs, but I spend to my liking and do not want to compare
myself with others.�?

149
6.20 The empirical evidence suggests that income inequality has been rising at a national level
in Vietnam, albeit modestly. Figure 6.1 shows the ratio of mean per-capita incomes of the top and
bottom quintile, decile, and vigntile of the income distribution. Although all groups saw substantial
growth between 2004 and 2010, the unevenness of growth implies that the ratio of mean per-capita
income of the top 20 percent relative to the bottom 20 percent (referred to recently by General
Statistics Of�?ce as the “rich/poor gap�?) has increased from just over 7 to 8.5. Similar tendencies are
seen across other income quintiles, and the increase in disparities grows as one narrows in on the
very poorest and very richest households.

6.21 It is clear that ethnic minorities are becoming increasingly left behind in these growth
processes. The last three columns of �?gure 6.1 show that average incomes and growth among
the bottom 20 percent and 10 percent of the ethnic minority distribution have been lower than that
among the majority population. Comparing average incomes among those in the bottom 20 percent
of households in the ethnic minority population with those in the top 20 percent of the majority
population, we see that the top 20 percent of majorities earned 11.4 times what was earned by the
bottom 20 percent of minorities in 2004, and 17.5 times what was earned in 2010. In comparison,
when we look at the entire population, the ratio of incomes among the top to bottom 20 percent rises
from 7.2 in 2004 to 8.4 in 2010. This suggests that ethnic minorities are increasingly overrepresented
among the poor, as has been discussed earlier. The gaps between minorities and the rest of the
population are rising. The ratio of incomes earned by the bottom 20 percent of minorities relative to
the bottom 20 percent of majorities has also increased, from 1.4 to 2.1. This may reﬂect, in part, the
predominance of agriculture as a major source of income among minorities and poorer households
(see Chapter 5).

Figure 6.1 Ratio of Mean Per-capita Income by Percentile, 2004-2010

20
Ratio

0
Top20%/BottomTop10%/Bottom Top5%/Bottom Bottom20% Top20%Majority Top10%Majority
20% 10% 5% Majority/Bottom /Bottom20% /Bottom10%
20%minority Minority Minority

2004 2010

Source: 2004, 2010 VHLSS.

6.22 The rural sector has been the driving force behind the recent rise in income inequality. Figure
6.2 shows the growth incidence curves for income by per-capita income decile in rural areas. Growth
in rural areas has been far higher among richer households than among poorer households; growth
in the poorest 10 percent of households was less than half that seen in the richest 10 percent of
households, and the ratio of income consumed by the top income decile to that consumed by the
bottom income decile increased by 25 percent between 2004 and 2010. For the �?rst time since

150
VHLSS data started being collected, the Gini Coef�?cient of income inequality is now of a similar
magnitude in urban and rural areas. The Gini Coef�?cient of income inequality in rural areas rose from
0.365 in 2004 to 0.413 in 2010, while in urban areas the Gini remained stable over the same period,
at approximately 0.381. 40

Figure 6.2 Mean Per-capita Rural Income Figure 6.3 Theil Decomposition of the
per Year by Rural Income Decile, Level and Changes in Income Inequality,
2004-10 2004 to 2010

45,000 9 0.450 Differencesinmean
incomesbetweeenrural
40,000 8 0.400 andurbanareas

AnnualizedGrowthRate(percent)
35,000 7 0.350
ThousandJan.2010VND

Inequalitywithinruralareas
30,000 6 withineachregion
0.300
25,000 5
0.250
Differencesinmean
20,000 4
0.200 incomesbetweenrural
15,000 3 areasofdifferentregions
0.150
10,000 2 Inequalitywithinurban
0.100 areaswithineachregion
5,000 1
0 0 0.050

1 2 3 4 5 6 7 8 9 10 Differencesinmean
0.000 incomesbetweenurban
2004
RuralIncomeDecile 2004 2010 areasofdifferentregions
2010

Source: 2004, 2010 VHLSS.

6.23 The contribution of differences in mean incomes between rural and urban areas and between
provinces to explaining overall inequality has declined over time. The Theil index of inequality can be
decomposed into �?ve components: (a) differences in mean incomes between rural and urban areas
nationally, (b) differences in mean incomes between rural areas of different provinces, (c) inequality
within rural areas within each province, (d) differences in mean incomes between urban areas of
different provinces, and (e) inequality within urban areas within each province.41

6.24 Figure 6.3 shows the fraction of income inequality attributable to these various components
in 2004 and 2010. Between 2004 and 2010, the fraction of income and expenditure inequality
attributable to differences in income between rural and urban areas declined. This is a reﬂection of
the faster average rate of growth in rural areas, with the result that mean incomes and expenditures
in rural areas have been catching up with those in urban areas. The ratio of income in urban areas
to income in rural areas declined from 1.87 in 2004 to 1.70 in 2010.42 Similar patterns were seen in
consumption; the ratio of mean consumption in urban areas to rural areas declined from 2.26 in 2004
to 2.01 in 2010. This appears to be driven by the top end of the rural income distribution; households
in the top 40 percent of incomes in rural areas have seen faster growth than households in the top
40 percent of incomes in urban areas, while the bottom 20 percent of rural households have seen
slower growth than their urban counterparts. The decline in rural-urban welfare differences over time

40 Trimming for measurement error and removing the bottom and top 1 percent of the income distribution reduces the
magnitude of the Gini Coef�?cients, but the trends over time remain the same; the Gini Coef�?cient of inequality in urban
areas remains fairly stable, while the Gini Coef�?cient of inequality in rural areas rises above the urban Gini.
41 Since the fraction of the population in urban and rural areas, and by region, is changing over time, changes in the between
component of inequality may also be attributable to changes in the relative share of the population living in urban areas.
42 These �?gures reﬂect spatially deﬂated income and consumption aggregates. The patterns for nonspatially adjusted
�?gures reﬂect a similar decline, from a ratio of 2.15 to 1.98 for income and from 2.72 to 2.57 for consumption. The higher
nonspatially adjusted ratio reﬂects price differences between urban and rural areas.

151
in Vietnam is in contrast to the development patterns of China, where a rapid expansion of the rural-
urban gap has been an important source and driver of inequality (World Bank 2009).43,44

6.25 Despite rising income inequality, inequality in consumption at a national level has not been
increasing. The difference between income and consumption inequality patterns warrants further
analysis. Income is a ﬂow measure while consumption (as de�?ned for this report) has been smoothed
over time; for example, it also includes imputations for housing and durables. In addition, the way that
consumption was measured changed in 2010, which has raised issues of comparability with earlier
rounds. Therefore, income was deemed a more suitable candidate for over-time comparisons of
inequality.

6.26 Perceptions of inequality as captured in the qualitative study appear to capture different
concepts than are reﬂected in empirical measures of inequality, and as such provide a different albeit
complementary facet of inequality. For example, the perception of rising inequality in urban and rural
areas is at odds with the empirical evidence, which suggests that the rise in income inequality at the
national level is driven mostly by rising inequality in rural areas. Furthermore, at the national level
inequality in expenditures appears to have remained stable in the 2000s, in contrast to perceptions
that it has been rising. The annex to this chapter discusses how to reconcile differences between
empirical measures of inequality and perceptions of inequality.

D. Why has Income Inequality Increased in Vietnam?

6.27 Disparities in incomes across Vietnam and the rise in income inequality can be attributed to
multiple and interrelated factors.45 First, and as discussed elsewhere in this report, ethnic minority
groups have progressed less rapidly than the Kinh majority. Second, and closely related, geographic
variations in growth patterns are likely to contribute to the rise in inequality; that is, differences in drivers
of agricultural and nonagricultural growth across regions contribute to differences in growth rates.
Third, the rise in income inequality reﬂects changes in the pattern of production away from agriculture
into the nonagricultural sector, and from low-skill to higher-skill work outside the agriculture sector.
The changes in production vary in their scope across region and interact with existing disparities in
human and physical capital to change the distribution of incomes in Vietnam over time. Finally, the
misuse of power, corruption, and connections are also likely to be linked to inequality, although it is
not clear to what degree these factors have contributed to the rise in income inequality.

6.28 The �?rst three explanations for rising income inequality are examined in this section; inequality
in power, corruption, and connections are discussed in the next section. Other factors such as
changes in land-holding patterns and regional variation in agricultural productivity are also likely to
play an important role, and are left for future exploration.

6.29 The rise in income inequality reﬂects the increasing economic polarization of many ethnic
minority groups. The evidence suggests that differences in growth rates between ethnic minorities

43 The rural-urban income gap and trends in the gap vary substantially between provinces, and more recent analyses of the
gap �?nd that it has declined, in part due to rural-to-urban migration.
44 Between-group inequality consists of three factors: differences between groups in mean incomes, the number of groups,
and their relative sizes. Therefore, changes in the underlying population structure can cause dif�?culties with comparisons
of decompositions over time. We therefore compare the standard measures of between inequality to the maximum
possible between inequality for groups of the same size and number using the method of Elbers et al. (2008). We �?nd that
the conventional measure of inequality between regions accounts for a declining share of maximum between inequality
between 2004 and 2010. However, although declining, inequality attributable to differences between rural and urban
areas, and between regions, continues to be an important characteristic correlated with inequality.
45 The factors discussed most detail in the text are those that were considered to be key factors related to rising inequality,
as identi�?ed through empirical analysis and also emerging from the qualitative study.

152
and the majority have particularly contributed to rising inequality within rural areas. Since ethnic
minorities have lower education outcomes and lower access to productive capital, differences in
these other assets contribute to and substantially reinforce differences in incomes across ethnicities.
As the non-agricultural sector has grown in Vietnam and more educated individuals have been able
to pro�?t from this growth, the predominance of minorities in the slower growing agricultural sector has
resulted in a widening gap, on average, between minorities and the Kinh majority.

6.30 Figures 6.4 and 6.5 show growth by income source among ethnic minorities and the majority,
by quintile, between 2004 and 2010. The majority of income growth among poorer ethnic minority
households has come from agriculture and side-line activities. Incomes among all minority quintiles
apart from the richest are growing at a slower rate than those of the majority, and even the fastest-
growing minority households experience lower income growth than the average majority households.
The divergence in growth rates is strongly related to the income-generating activities of households.
The fraction of income and growth from wage income and nonagricultural sources rises as one moves
up the income quintiles. Only the richest 20 percent of minority households experience substantial
growth in incomes arising nonagricultural business activities.

6.31 The fraction of inequality attributable to differences in mean incomes between the majority
and minority has risen over time, from 9 percent of total inequality to 14 percent, and approximately
one-quarter of the rise in income inequality over time in rural areas can be attributed to differences
between the majority and ethnic minorities. Therefore, differences in growth rates between minorities
and the majority have contributed to the rise in inequality over time, particularly in rural areas where
ethnic minorities are concentrated.

6.32 Alongside an increase in mean income differences between minorities and the majority, the
uneven patterns of growth across income quintiles suggests that inequality has risen within the
majority group and within minority groups. The income data suggest that incomes among the poorest
20 percent of minorities grew at an average annual rate of only 2 percent, substantially slower than
the growth rate for the wealthiest 20 percent of minorities.
6.33 The percentage rise in the Gini Coef�?cient of income inequality among the Kinh majority (in
urban and rural areas) is greater than that seen in the combined sample, suggesting that the overall
rise in income inequality is additionally driven by other factors.

6.34 The evidence from the VHLSS of growing disparities between ethnic minorities and the
majority population is corroborated in a study tracking rural households over time using the Vietnam
Access to Resources Household Survey (McKay and Tarp 2011). This study �?nds that incomes
for ethnic minorities grew more slowly, on average, between 2006 and 2010 than the rest of the
rural population, and this was the case even among minority and majority households with similar
observable productive assets and education. Interestingly, the study documents substantially higher
growth rates for ethnic minority households with high levels of education compared to other minority
households.

6.35 A second explanation for rising inequality is geographic variation in growth patterns that might
have caused an increase in inequality between regions, provinces, and districts (see Chapter 4 for
a detailed discussion of regional variation in growth). Regional variation in growth patterns does,
however, prompt the question: Why are certain regions growing faster than others, and what is
driving these differences in growth?

6.36 The evidence suggests that regional variation in growth patterns contributes to the explanation
of the rise in inequality, but appears to play a more limited role than differences across households
within regions. There is substantial evidence of variation in growth across regions, with some poorer
regions such as the North East, North Central Coast, and North West growing substantially more
slowly than the Red River Delta and the Central Highlands. Figure 6.6 shows mean incomes and
growth between 2004 and 2010 by region. Growth has been uneven across regions; income growth

153
Figure 6.4 Growth by Income Source, 2004-2010, Ethnic Minorities

10.00%
9.00%
8.00%
7.00%
6.00%
5.00%
4.00%
3.00%
2.00%
1.00%
0.00%
1 2 3 4 5

IncomeQuintile(Minority)
Agriculture Sidelines
Business(NonͲAgriculturalSector) WageIncome
Remittances Scholarships/Grants

Figure 6.5 Growth by Income Source, 2004-2010, Ethnic Majority

10.00%
9.00%
8.00%
7.00%
6.00%
5.00%
4.00%
3.00%
2.00%
1.00%
0.00%
1 2 3 4 5

IncomeQuintile(Majority)
Agriculture Sidelines
Business(NonͲAgriculturalSector) WageIncome
Remittances Scholarships/Grants
Other

Source: 2004, 2010 VHLSS.

154
in the North East has been lower than in other parts of the country, while income growth in the Red
River Delta and in the Central Highlands has been substantially greater than average growth rates of
8 percent. The South East region remains the region with the higher income per capita. These growth
patterns differ somewhat from patterns in the 1990s; between 1993 and 1998, the Northern Uplands
and Central Highlands experienced the lowest rates of growth, while the South East experienced the
highest rates of growth (World Bank 2000).

Figure 6.6 Mean Annual Per-capita Rural Income per Year by Region, 2004-2010

30,000 14

AnnualizedGrowthRate(percent)
25,000 12

10
ThousandJan.2010VND

20,000
8
15,000
6
10,000
4

5,000 2

0 0

2004 2010 Growth

Source: 2004, 2010 VHLSS.

6.37 The fraction of income variation attributable to differences across regions and provinces
has risen over time in rural areas, in part due to uneven growth in agriculture and in part due to
geographic variation in opportunities in the nonagricultural sector. In contrast, China saw a reduction
in the variation in incomes attributable to location over the 1990s and early 2000s (Benjamin et al.
2004; Benjamin et al. 2007). An important caveat is that migration and remittances are likely to play a
mediating role in reducing variation in incomes and growth across regions, and the extent of this role
is not fully captured in the data.46 This is an area that deserves further attention in future analysis of
inequality.
6.38 Uneven growth across regions and provinces in Vietnam has contributed to rising inequality,
although the majority of the rise in income inequality is attributable to rising inequality within regions
rather than rising inequality between regions. Approximately 8 to 10 percent of the rise in the Theil

46 Four percent of rural households declare having a household member who has stayed away from home for more than six
months over the last 12 months. This number appears low relative to evidence from the census (GSO 2009) and misses
shorter-term, longer-term, and household migration patterns.

155
index can be attributable to an increase in inequality between regions, with the remainder due to
inequality within regions. These trends are in marked contrast to patterns seen in Vietnam in the
1990s. Between 1993 and 1998, 83 percent of the increases in inequality was attributed to an
increase in inequality between regions (World Bank 2000).

6.39 Differences in incomes and expenditures are increasingly related to differences in household
characteristics rather than to where households live, although location continues to be an important
correlate of household welfare. Education appears to be one of the most important characteristics
for explaining differences in income and expenditure across households in 2010. Controlling for
the average education of working-age adults explains more of the variation in household incomes
in rural areas than taking into account the region of residence. The fraction of variation in income
explained by education increased between 2004 and 2010, suggesting that education is becoming an
increasingly important correlate of income. The amount of variation in household income attributable
to differences between regions of residence has also increased over time, although it has done so
from a lower base. Of the total increase in the Theil-L between 2004 and 2010, 65 percent of the
increase can be attributed to an increase in inequality between household education levels, where
household education is de�?ned using the education of the household head.

6.40 The third explanation for rising inequality relates to shifts in the pattern of production away
from agriculture into the nonagricultural sector. Nonagricultural opportunities and employment were
strongly identi�?ed in the perceptions study as contributory factors to the rise in inequality. Factors
discussed included a move away from agricultural production toward greater nonagricultural wage
and business opportunities, rising returns to education, disparities in education across households,
and differences in initial capital endowments. In urban areas, discussions centered around access
to good employment opportunities and land conversion, while in rural areas higher value-added
agricultural and sideline activities and access to nonagricultural employment opportunities are cited
as prime candidates for rising inequalities. Respondents noted increasing dif�?culties in access to
good jobs, particularly with respect to public sector employment.

6.41 The composition of household income and employment has moved away from agriculture
toward manufacturing and services. Figure 6.7 shows the share of workers in the primary, secondary,
and tertiary sector and indicates the fraction of workers in each sector in rural and urban areas.
Between 1998 and 2010, the share of the working population employed in agriculture declined from
68 percent to 45 percent while the share employed in manufacturing increased from 12 to 24 percent
and that in services increased from 20 to 31 percent. In both rural and urban areas, wage incomes
have seen fast and above-average growth over the period, while incomes from agricultural and allied
activities have grown relatively slowly. Although agricultural and allied activities continue to be an
important source of income for rural households, their contribution has declined from an estimated
55 percent of rural income in 1998 (McCaig et al. 2009) to only 35 percent of rural income in 2010.

6.42 There is substantial regional variation in both the speed at which economic activity has moved
away from agriculture in rural areas, and in the intensity with which nonagricultural activities are
conducted at a household level. In rural areas, diversi�?cation into nonagricultural employment has
occurred at both the household and individual level, and it has been a powerful force for poverty
reduction over the past decade. Variation in the speed at which this is occurring is likely to be related
to variation in living standards and in growth rates across regions.

6.43 The expansion of nonagricultural wage and salaried work in urban and rural areas continues
a trend seen in the 1990s. In rural areas in 1998, wages and salaried work contributed only 14
percent of total income overall (McCaig et al. 2009). Wages have become a more signi�?cant source
of income throughout the 2000s; by 2010, wages accounted for 32 percent and 52 percent of income

156
in rural and urban areas, up from 26 percent and 44 percent, respectively in 2004.47 Although 19
percent of individuals receiving wages in rural areas in 2010 worked for wages in the agricultural
sector, the vast majority of rural wage work is outside of agriculture.48

6.44 Employment patterns in the nonagricultural sector are very different in rural and urban areas.
In rural areas, the move out of agriculture has been accompanied by a sharp rise in employment in
manufacturing and construction. In 2010, nearly 70 percent of individuals in the secondary sector
were found in rural areas, and this sector accounted for nearly 20 percent of overall employment in
these areas. By contrast, urban areas have seen a decline in the fraction of individuals employed in
the manufacturing sector and a corresponding expansion in services.

6.45 Occupations in the nonagricultural sector differ in their demand for skill, and the composition of
nonagricultural growth by occupation type has differed across rural and urban areas. Figure 6.8 shows
the split of workers between agriculture and lower- and higher-skilled nonagricultural work (blue- and
white-collar work) in rural and urban areas. Although the fraction of workers conducting high-skilled
(white-collar) work has risen over time, the majority of the increase has been seen in urban areas.49
By contrast, rural areas have seen growth in lower skilled, blue-collar nonagricultural employment,
which partly reﬂects a substantial increase in manufacturing work in rural areas over time.

Figure 6.7 Sector of Employment for Working- Figure 6.8 Type of occupation for working-
age Individuals in 1998, 2004 and 2010 age individuals in 1998, 2004 and 2010
Fraction of the working population aged 18-65
80
80
Fraction of working individuals aged 18-65 by sector

70
Rural Urban 70
Rural Urban
60
60

50 50

40 40

30 30

20 20

10 10

0 0
1998 2004 2010 1998 2004 2010 1998 2004 2010 1998 2004 2010 1998 2004 2010 1998 2004 2010

Agriculture and Manufacturing and Service Sector Agriculture Blue Collar White Collar
Mining Construction

Source: 1998 VLSS, 2004 VHLSS, 2010 VHLSS.
Note: Classi�?cations are based on occupation codes. Agriculture includes high and low-skilled agricultural work. Non-
agricultural occupations are separated into lower and higher-skilled work: higher-skilled work consists of all professional
and of�?ce based categories, lower-skilled work includes machine operators, service and sales and unskilled work. The
blunt classi�?cation is due to changes in occupation codes over time.

47 Wages are likely to include income remitted from members of the household who work in another region. Since many
migrants go from rural to urban areas, the fraction of rural incomes coming from wages is likely to overstate the amount
of wage work actually being conducted in rural areas.
48 There is substantial regional variation in the prevalence of agricultural wage work in rural areas. In the North, only 8
percent of individuals working for wages in rural areas are found in the agricultural sector. In the South, nearly 29 percent
of wage workers in rural areas can be found in agriculture.
49 High-skilled work has become disproportionately urbanized over time. In 1998, 56 percent of professional jobs was found
in urban areas compared to approximately 20 percent of the population; by 2010, 64 percent of professional jobs and 30
percent of the population were in urban areas.

157
6.46 The pattern of nonagricultural growth—greater manufacturing and blue-collar employment
growth in rural areas and greater service sector and white-collar employment growth in urban
areas—is perceived as a source of disparity among focus group respondents in rural areas and
in small urban towns. For example, in rural areas with industrial parks, such as Hai Duong, factory
employment is the primary source of labor demand in the nonagricultural sector. While it is possible
to �?nd low-skilled and relatively low-paid work in a factory, it is perceived that there are far fewer
higher-skilled and higher-paid employment opportunities than in big cities such as Hanoi.

6.47 Figures 6.9 and 6.10 show the composition of income and growth across income quintiles in
urban and rural areas, respectively. The share of income from agriculture and allied activities has
declined over time but continues to be the major source of income for the poorest 40 percent of
the rural population. The share of income coming from sideline activities related to agriculture has
remained substantial among poorer households and has grown as a share of income for the poorest
quintiles since 1993 (McCaig et al. 2009). The majority of income from this component across all
income quintiles comes from livestock farming and aquaculture.

6.48 The rising share of wage incomes across the income distribution can be readily seen in the
�?gures. In urban areas, wages are the most important source of income across all income groups
and account for over half of incomes. This is in stark contrast to the income pro�?le in 1993, when
the majority of incomes from the top half of the income distribution came from business income.50
In rural areas, all groups earned a greater share of income from agriculture and sideline activities
than from wages in 1993 and 2004. By 2010, wage incomes had overtaken agricultural incomes for
the third and fourth quintiles. Although the share of wages increased for the richest quintile, they
continue to earn more from business and agriculture. The fraction of working individuals receiving
wages as either their primary or secondary employment has also risen over time, from approximately
17 percent of the workforce aged 18 to 65 in 1998 to 40 percent in 2010, and from 13 percent to 37
percent of the workforce in rural areas.51

6.49 To more formally explore the contribution of different income sources to income inequality, we
decompose the Gini Coef�?cient into its source components (Adams 1999; Stark et al. 1986). The
Gini Coef�?cient of total income can be written as the sum of the contributions of each income source.
The effect of a source on total income can then be broken down into three components: (a) the share
of income component in total income; (b) the inequality within the sample of income from a given
source; and (c) the correlation between a given source of income and total income. The larger the
product of these three components, the greater the contribution of income from the source to overall
income inequality.

50 The income structure of the richest quintile of the urban population has converged on the structure of the poorer groups
over time. In 1993 and 2004, the income composition of the top 20 percent appeared to be quite different from the rest
of the population; business incomes were a much larger share of income for the top quintile, and they had the smallest
share of income from wage sources. By 2010, the top quintile looked more similar to the other groups; their share of wage
income rose to 49 percent of income in 2010, from 38 percent, while the share of income from business sources declined
from 37 percent to 28 percent. These trends continue patterns seen in the 1990s; in 1993, the upper quartile of the income
distribution earned nearly 60 percent of their income from a housbusiness compared to 10 percent from wages sources
(McCaig et al. 2009).
51 Note that labor market participation has also changed over this period. In 1998, 90 percent of individuals between 18 and
65 reported working compared to 84 percent in 2010, while the fraction of the population that is of working age has risen
over time, from 54 percent to 64 percent between 1998 and 2010 (calculated from the 1998 and 2010 VHLSS).

158
Figure 6.9 Composition of Income in Urban Areas, 2010
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
1 2 3 4 5

Quintile

Agriculture Sidelines Business Wages Remittances Scholarships/Grants Other

Source: 2010 VHLSS.

Figure 6.10 Composition of Income in Rural Areas, 2010

100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
1 2 3 4 5

Quintile

Agriculture Sidelines Business Wages Remittances Scholarships/Grants Other

Source: 2010 VHLSS.

6.50 Figure 6.11 presents relative concentration coef�?cients which indicate whether an income
source is inequality increasing or inequality decreasing. If the relative concentration coef�?cient is
greater than 1, then the source is inequality increasing, while if it takes on a value less than 1, the
source of income is inequality decreasing. Figure 6.12 shows the contribution of the different sources
of income to the Gini Coef�?cient of inequality, including their share of total income.

159
Figure 6.11 Relative Concentration Coef�?cients of Different Sources of Income, 2010
1.75
Relativeconcentration

1.5 2004 2010
1.25
coefficient

Inequality >
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 coef�?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
coef�?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 signi�?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/of�?ce workers. These positions are usually classi�?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 (�?gure 6.15). Empirical work carried out for this report �?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 �?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 �?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%

6%
y

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 bene�?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 �?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.�? (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
�?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 �?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 signi�?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 signi�?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 signi�?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 quali�?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.�?

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.�?

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 bene�?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 (“efforts�?) or are due to circumstances that fall beyond an individual’s responsibility
(“circumstances�?) (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. Speci�?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 “quantity�? 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
�?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 signi�?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 identi�?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 magni�?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 �?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 �?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 of�?cials.59 These concerns were widespread and
expressed by individuals from all backgrounds, including commune of�?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.�? (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.�?
(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.�? (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 pro�?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.�?

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 politics�? 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 bene�?ts from public social assistance
programs targeted at the poor.

6.89 It is perceived that those who have been of�?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.�? (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 bene�?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 ful�?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 ef�?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 do�? Perceptions of Inequality�? 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
Coef�?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 Coef�?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 dif�?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 �?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.�? Site Report from Phuc Xa Ward, Hanoi
(better-off residents)

“The students claimed that the rich-poor gap over the past �?ve years has been
increasingly widened due to the increasing relative gap: the rich develop faster than the
poor.�? 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.�? 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 Del�?na, and Tuyet Thi Phung. 2009. “Vietnam
Development Report 2010: Modern Institutions.�? World Bank, Washington DC.
Asian Development Bank. 2012. “Outlook 2012: Confronting Rising Inequality in Asia.�? Asian
Development Bank, Manila.
Banerjee, Abhijit V. and Esther Duﬂo. 2003. “Inequality And Growth: What Can The Data Say?�?
Journal of Economic Growth 8 (3) (September): 267–99.
Benjamin, Dwayne, and Loren Brandt. 2002. “Property Rights, Labour Markets, and Ef�?ciency in a
Transition Economy: The Case of Rural China.�? 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.�? University of Toronto, Toronto.
Benjamin, Dwayne, Loren Brandy, and John Giles. 2005. “The Evolution of Income Inequality in
Rural China.�? 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.�? 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.�? World Bank, Washington, DC.
CECODES, FR, CPP, and UNDP. 2012. “The Viet Nam Governance and Public Administration
Performance Index (PAPI): Measuring Citizens’ Experiences.�? 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.�? Journal of Economic Inequality Springer 6 (3) (September): 231–45.
Gallup, J. 2002. “The Wage Labour Market and Inequality in Vietnam in the 1990s.�? 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.�? World Development 27 (5): 887–902.
GSO (General Statistics Of�?ce of Vietnam). 2009. “Vietnam Population and Housing Census 2009:
Migration and Urbanization in Vietnam: Patterns, Trends and Differentials.�? Ministry of Planning and
Investment, General Statistics Of�?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.�? 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.�? 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.�? 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.�? University of Toronto, Toronto.
McKay, Andy, and Finn Tarp. No date. “Welfare Dynamics in Rural Vietnam, 2006 to 2010.�? 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.�? World Bank, Washington, DC.
Psacharopoulos, G. 1994. “Returns to Investment in Education: A Global Update further Update.�?
World Development 22 (9): 1325–1343.
Ravallion, Martin. 2004. “Competing Concepts of Inequality in the Globalization Debate.�? 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?�? Journal of Development Economics, Elsevier 87 (2)
(October): 191–209.
Ravallion, Martin, and Shouhua Chen. 2007. “China’s Uneven Progress against Poverty.�? 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.�? 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.�? Department of Political
Science Discussion Paper, Yale University, New Haven.
Stark, O., J. E. Taylor, and S. Yitzhaki. 1986. “Remittances and Inequality.�? Economic Journal 96
(383): 722–740.
VASS (Vietnamese Academy of Social Sciences). 2008. “Participatory Poverty Assessment 2008.�?
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.�?
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?�?
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.�? 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.�? National Political Publishing House, Hanoi.

177
Publishing licence number No:742/Q�?LK-L�? and �?KKHXB CXB No: 279-2012/CXB/21-69/L�? 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