THE FOURTH

VDF-TOKYO CONFERENCE
ON THE DEVELOPMENT OF VIETNAM


PROCEEDINGS

Saturday, 9 August 2008
Conference Room 3C
GRIPS, Tokyo


Organized by
Vietnam Development Forum (VDF) Tokyo
http://www.grips.ac.jp/vietnam/VDFTokyo/index.html


Supported by
Vietnam Development Forum (VDF) Hanoi
http://www.vdf.org.vn
National Graduate Institute for Policy Studies (GRIPS)
http://www.grips.ac.jp
GRIPS Development Forum
http://www.grips.ac.jp/forum/


Vietnam Development Forum (VDF)-TOKYO


CONTENTS
About VDF-Tokyo 1
List of Publications by VDF 3
Program of the Conference 6

MORNING SESSION <Keynote Presentations>

[1] Industrial Strategy for Vietnam's New Era: Policy Content and
Formulation Method
Dr. Kenichi Ohno
(Professor, GRIPS & Research Director, VDF)
8

[2] Finance and Enterprise Sector Reform in Vietnam
Mr. oritaka Akamatsu
(Finance & Private Sector Program Manager for Vietnam, Regional Advisor
for Capital Market Development, East Asia & Pacific, The World Bank)
23

AFTERNOON SESSION <Paper Presentations>

[1] Poverty, Inequality and Ethnic Minorities in Vietnam
Dr. Katsushi Imai
(Assistant Professor, University of Manchester, UK)
34

[2] Demographic Changes and the Pension Finances in Vietnam: A Long-term
Stochastic Actuarial Assessment
Dr. Wade D. Pfau
(Associate Professor, GRIPS, Japan)
60

[3] The Economics of Not Using the Health Insurance Card: A Case Study of
Vietnam
Mr. guyen Trong Ha
(Lecturer, NEU, Vietnam; PhD candidate, ANU, Australia)
85

[4] The Revolution of Vietnamese Distribution System
Dr. Le Viet Trung
(Postdoctoral Fellow, Kobe University, Japan)
113

ONE-PAGE PRESENTATIONS

(1) Efficiency Estimates for the Agricultural Sector in Vietnam: A Comparison of
Parametric and Non-parametric Approaches
(Nguyen Khac Minh & Giang Thanh Long)
127

(2) The Vulnerability of Elderly to Poverty: Determinants and Policy Implications for
Vietnam
(Giang Thanh Long & Wade Donald Pfau)
128

(3) An Inquiry into the Transformation Process of Village-based Industrial Clusters: The
Case of Iron and Steel Cluster in Northern Vietnam
(Vu Hoang Nam)
129

(4) Simulating the Potential Impacts and Costs of an Extended Social Pension in
Vietnam
(Giang Thanh Long & Wade Donald Pfau)
130

(5) Development of Corporate Bond Market in Vietnam
(Tran Thi Thanh Tu)
131

(6) The Interrelationship between Growth, Poverty and Inequality in Vietnam: An
Empirical Analysis
(Le Quoc Hoi)
132

(7) University-Industry Linkages for Improving Industrial Human Resource
(Nguyen Thi Xuan Thuy)
133

(8) Endogenous Development in the Era of Globalization: Analysis of interactions
between Local Governments and Local Enterprises in the Vietnamese Central Focal
Economic Zone
(Do My Hien)
134

(9) Is a Governmental Micro-Credit Program for the Poor Really Pro-Poor? Evidence
from Vietnam
(Nguyen Viet Cuong)
135

(10) Technical Efficiency and Productivity in Vietnam Electronic Industry: A Comparison
between Foreign Affiliated Firms and Domestic Firms with Application of Parametric
and Non-Parametric Analysis
(Tran Hoai Vu)
136

CONTRIBUTION PAPERS

(1) Modeling Economic Transformation in Vietnam Using Smooth Transaction Method
(Nguyen Khac Minh, Nguyen Manh Hung & Nguyen Viet Hung)
137

(2) Does the Reproductive Health Project in Nghe An Province, Vietnam Have Good
Impacts on the Community?
(Pham Thi Thu Huong)
153

(3) Using Vietnams I-O Table 2000 to Identify the Key Economic Sectors
(Dinh Hoang Yen & Hiroshi Osada)
168

(4) Tourism Destination Image: Vietnam as a Tourism Destination for Japanese
Travelers
(Le Tuan Anh)
180

(5) The Vietnamese Automobile Industry and Globalization
(Nguyen Bich Thuy)
197


1
About Vietnam Development Forum


Who Are We?
Vietnam Development Forum (VDF) was established in 2004 as a joint research
project between the National Graduate Institute for Policy Studies (GRIPS) in Tokyo
and the National Economics University (NEU) in Hanoi.
VDF-Tokyo is a branch office of VDF in Tokyo, which was established in October
2004, superseding the Japan-Vietnam Economist Club (JVEC) an informal
volunteer group set up in 2003 by Professor Kenichi Ohno (GRIPS), Professor Tran
Van Tho (Waseda Univ.) and Vietnamese graduate students studying in Japan in
order to promote research on Vietnamese economy.
VDF aims at innovating research methodology and dissemination for policy impact.
For this purpose, young talented Vietnamese people are mobilized.
VDF bridges the gaps between academic research and policy formulation. We build
open networks where researchers and policy makers can meet, discuss and help each
other.
VDF initiates and coordinates studies on Vietnams development. We also happily
publicize excellent research produced by scholars and organization outside the VDF.
VDF is an academic unit and not a consulting business. It is financially supported by
the Japanese government.

How Can We Help?
Policy Makers: We will listen to you and discuss with you concrete policy issues related to
Vietnams development, and explore ways to solve them together.
Researchers: We will offer opportunities to development and present your ideas, interact with
other researchers and policy makers, and publish your output (subject to review).
Everyone: We are happy to work with you and help you build an intellectual network.

Principal Project Areas
VDF is prepared to support any research topic relevant to Vietnams development and
conducted by capable researchers. Our research portfolio includes:
Formulation of industrial strategies and master plans under integration (with the
Ministry of Industry).
Sector studies including electronics, motorbike, automobile, steel, electricity and
supporting industries.
2
Social issues under market-orientation and globalization, such as social security,
street children, and drug addiction.
Trade and environmental analysis using input-output tables and other data.
Aid partnership and modality.

Activities
Continuous networking and cooperation with MOI, MOLISA, MPI, ADB, Japanese
government, domestic and foreign business communities and other research
organizations.
Research competition in which talented Vietnamese researchers are supported
academically and financially.
Logistic and financial supports for visits by Vietnamese and foreign researchers.
VDF-organized research trips to HCMC, Japan, Thailand, Malaysia, etc.
Hosting many conferences, including three international conferences (with ADB,
MOI, and VIES), a national conference with Hanois People Committee. So far,
VDF-Hanoi held 91 workshops, while VDF-Tokyo held 45 workshops.
VDF has published 12 books, 20 working/discussion papers, 2 policy notes, and
many issue papers and drafts.
Frequent appearance and contribution to newspapers, press and broadcasting media.

How to Contact?
VDF-Hanoi:
Suite 401, Melia Central Office Building
44B Ly Thuong Kiet Street, Hoan Kiem District, Hanoi, Vietnam
Tel: +84-4-936 2633 Fax: +84-4-936 2634
Website: http://www.vdf.org.vn

VDF-Tokyo:
Room E404, National Graduate Institute for Policy Studies (GRIPS)
7-22-1 Roppongi, Minato-ku, Tokyo 106-8677
Tel: +81-3-6439 6337 Fax: +81-3-6439 6010
Website: http://www.grips.ac.jp/vietnam/VDFTokyo/index.html




3
Vietnam Development Forum
List of Publications
Publication Title Author(s) Language Download *
No.1
Designing a comprehensive and
realistic industrial strategy
Kenichi Ohno E, V VDF
No.2
The pension scheme in Vietnam:
Current status and challenges in a
aging society
Giang Thanh Long E, V VDF
No.3
A comparative study on production
efficiency in manufacturing
industries of Hanoi and Ho Chi Minh
City
Nguyen Khac Minh E, V VDF
No.4
Cong nghiep Viet Nam: Danh gia co
cau, hoat dong, nhung co hoi va
thach thuc den nam 2020
Ministry of Industry V VDF
No.5
Xuat nhap khau Viet Nam: Danh gia
co cau, hoat dong, co hoi va thach
thuc
Ministry of Industry V VDF
No.6
Street children in Vietnam:
Interactions of old and new causes in
a growing economy
Duong Kim Hong
& Kenichi Ohno
E, V VDF
No.7
US technical barriers to trade and
Vietnamese seafood exports
Tran Van Nam E, V VDF
No.8
From material zone development to
rural industrialization
Pham Quang Dieu &
Nguyen Trung Kien
E VDF
D
i
s
c
u
s
s
i
o
n

P
a
p
e
r
s

No. 9
Iron and steel industry in Vietnam: A
new phase and policy shift
Nozomu Kawabata E, V VDF
No.1 Vietnam at the crossroads Kenichi Ohno E, V VDF
P
o
l
i
c
y

o
t
e
s

No.2
Supporting industries in Vietnam
from the perspective of Japanese
manufacturing firms
VDF E, J, V VDF Tokyo
Reconsidering Vietnam's exchange
rate mechanism: A preliminary
discussion
Kenichi Ohno &
Nguyen Chi Thanh
E, V VDF
Dollarization in Vietnam by
economic sectors
Nguyen Thi Hong E, V VDF
Rural finance in Vietnam Yoichi Izumida E, V VDF
Institutional arrangements for long-
term growth
Do Duc Dinh,
Kenichi Ohno &
Nguyen Thi
Thanh Huyen
E, V VDF
I
s
s
u
e

P
a
p
e
r
s

&

D
r
a
f
t
s

Is there a developmental threshold
for democracy? Endogenous factors
in the democratization of South
Korea
Nguyen Thi
Thanh Huyen
E, V VDF

4
Chinas steel market and implications
for Vietnam
Nozomi Kawabata E, V VDF
Electronics SCM Junichi Mori E, V VDF
The economic development of Japan:
The path traveled by Japan as a
developing country
Kenichi Ohno E, V
VDF/VDF
Tokyo
Improving industrial policy
formulation
Kenichi Ohno &
Nguyen Van Thuong
(eds.)
E, V VDF
Which institutions are critical to
sustain long-term growth in
Vietnam?
ADB&VDF
(eds.)
E, V VDF
Tang truong kinh te Viet Nam:
Nhung rao can can phai vuot qua
Nguyen Van Thuong
(ed.)
V n.a
Industrial policy formulation in
Thailand, Malaysia, and Japan
Kenichi Ohno (ed.) E, V VDF
Business Environment and Policies
in Hanoi
Kenichi Ohno &
Nguyen Van Thuong
(eds.)
E, V VDF
Lich su nhin tu quan diem sinh thai
hoc: Van minh Nhat Ban trong boi
canh the gioi
Nguyen Duc Thanh &
Bui Nguyen Anh Tuan
(translated)
V VDF
Technical efficiency and productivity
growth in Vietnam: Parametric and
non-parametric analyses
Nguyen Khac Minh &
Giang Thanh Long
(eds.)
E VDF
Building supporting industries in
Vietnam (Volume 1)
Kenichi Ohno (ed.) E, V
VDF/
VDF Tokyo
Social issues under economic
transformation and integration in
Vietnam (Volume 1)
Giang Thanh Long &
Duong Kim Hong
(eds.)
E, V
VDF/
VDF Tokyo
Social issues under economic
transformation and integration in
Vietnam (Volume 2)
Giang Thanh Long (ed.) E VDF
For sound development of the
motorbike industry in Vietnam
Motorbike Joint
Working Group
E, V VDF
Master plan of Vietnam motorcycle
industry for the period of 2006-2015
vision towards 2020
MOIT, ISP E, V VDF
B
o
o
k
s

Quan ly, day nghe va giao duc, phuc
hoi nhan cach cho nguoi sau cai
nghien: Van de va kinh nghiem o TP.
Ho Chi Minh
Tran Nhu &
Ho Ba Tham (eds.)
V VDF
No.1
VAR analysis of the monetary
transmission mechanism in Vietnam
Le Viet Hung &
Wade D. Pfau
E VDF
No.2
Abnormal returns after large stock
pice changes: Evidence from the
Vietnamese stock market
Pham Vu Thang Long E VDF
W
o
r
k
i
n
g


P
a
p
e
r
s

No.3 Do stock prices in Ho Chi Minh City
trading center have unit roots? A
discussion on power of ADF F test
Tran Viet Ha E VDF

5
with unexpected initial value
No.4
Empirical analysis of stock return
volatility with regime change using
GARCH model: The case of Vietnam
stock market
Vuong Thanh Long E VDF
No.5
Technology spillovers from foreign
direct investment in Vietnam:
Horizontal or vertical spillovers?
Le Quoc Hoi &
Richard Promfret
E VDF
No.6
The difference in development stages
and the costs of monetary union A
new open economy macroeconomics
model
Vu Tuan Khai E VDF
No.7
The vulnerability of the elderly
households to poverty: Determinants
and policy implications for Vietnam
Giang Thanh Long &
Wade D. Pfau
E VDF
No.8
Diversification of finance resources
for transport development in Vietnam
Luu Hoai Son E VDF
No.9
The effects of ODA in
infrastructure on FDI inflows
in provinces of Vietnam, 2002-2004
Pham Thu Hien E VDF
No.10
Industrial policy as determinant of
localization: The case of Vietnamese
automobile industry
Nguyen Bich Thuy E VDF
No.11
The impact of a common currency on
East Asian production networks and
China's export behavior
Mizanur Rahman E VDF
* VDF: http://www.vdf.org.vn/download.html
VDF-Tokyo: http://www.grips.ac.jp/vietnam/VDFTokyo/download.html

Note: E English; J Japan; V Vietnamese

6


The Fourth VDF-Tokyo Conference
on the Development of Vietnam


Tokyo, Saturday 9 August 2008


PROGRAM

Time Event Presenter Venue
09:00 - 09:45 Registration
09:45 -10:00
Opening and Introducing VDF and
VDF-Tokyo
Mr. Giang Thanh Long
(GRIPS/ VDF)

Room
3C

MORNING SESSION <KEYNOTE PRESENTATIONS>
10:00 - 11:00
Industrial Strategy for Vietnam's New
Era: Policy Content and Formulation
Method
Prof. Kenichi Ohno
(GRIPS/VDF)
11:00 - 12:00
Finance and Enterprise Sector Reform
in Vietnam
Mr. oritaka Akamatsu
(The World Bank, Vietnam)
Room
3C
12:00 - 13:15 Lunch Reception
1F
Canteen
AFTERNOON SESSION <PAPER PRESENTATIONS>
Chairperson: Prof. Kenichi Ohno (GRIPS/VDF)
13:30 - 14:15
Poverty, Inequality and Ethnic
Minorities in Vietnam
Dr. Katsushi Imai
(University of Manchester)
14:15 - 15:00
Demographic Changes and the Pension
Finances in Vietnam: A Long-term
Stochastic Actuarial Assessment
Prof. Wade D. Pfau
(GRIPS)


Room
3C

15:00 - 15:20 Tea break and one-page presentation display
3rd
floor
corridor

Vietnam Development Forum TOKYO



7
15:20 - 16:05
The Economics of Not Using the Health
Insurance Card: A Case Study of
Vietnam
Mr. guyen Trong Ha
(Australian National Univ.)
16:05 - 16:50
The Revolution of Vietnamese
Distribution System
Dr. Le Viet Trung
(Kobe University)
16:50 - 17:25

General Exchange of Ideas and Information

17:25 - 17:30 Closing Remarks
Prof. Kenichi Ohno
(GRIPS/VDF)
Room
3C













MORNING SESSION
<Keynote Presentations>
Industrial Strategy for Industrial Strategy for
Vietnam Vietnam s New Era: s New Era:
Policy Content & Formulation Method Policy Content & Formulation Method
Kenichi Ohno (VDF/GRIPS) Kenichi Ohno (VDF/GRIPS)
August 2008 August 2008
Topics
Vietnams challenge creating
internal value in manufacturing
A proposal for the Vietnam-Japan
strategic partnership for joint
monozukuri
A need for serious reform in policy
formulation
Vietnams New Challenge
Vietnam is entering a new era where
productivity breakthrough and value
creation are needed.
Opening up and receiving FDI can attain
middle income ($1,000+), but higher
income ($10,000+) requires good policy
and private dynamism.
Vietnams future depends on:
--Creation of internal source of growth
--Coping with new social problems
--New macroeconomic management
Per Capita GDP in 2004 ( PPP) World Bank data
0 5000 10000 15000 20000 25000 30000 35000
Hong Kong
Japan
Taiwan
Singapore
Brunei
S Korea
Malaysia
Thailand
China
Philippines
Indonesia
Vietnam
Cambodia
PNG
Mongolia
Laos
N Korea
Myanmar
East Timor
Green: participants in East
Asian production network
0
100
200
300
400
500
600
100 1000 10000 100000
Governance, WGI 2005 Governance, WGI 2005 Governance, WGI 2005 Governance, WGI 2005
Per capita income ($PPP2004, log scale) Per capita income ($PPP2004, log scale) Per capita income ($PPP2004, log scale) Per capita income ($PPP2004, log scale)
Sin Hkg
Jpn
Twn
S Kor
Bru
Mal
Thai
Mong
Phil
China
VN
E Timor
Indo Camb
PNG
Lao
N Kor
Mya
Diversity in Governance and
Economic Development
Sources: Compiled
from World Bank,
Worldwide Governance
Indicators, Sep. 2006;
and World Bank, World
Development
Indicators, 2006.
High correlation
(0.90) but
causality cannot
be argued from
this diagram
Only circled
economies
participate in
regional dynamism
Different Speed of Catching Up
0%
20%
40%
60%
80%
100%
1
9
5
0
1
9
5
5
1
9
6
0
1
9
6
5
1
9
7
0
1
9
7
5
1
9
8
0
1
9
8
5
1
9
9
0
1
9
9
5
2
0
0
0
2
0
0
5
Japan
Taiwan
S. Korea
Malaysia
Thailand
Indonesia
Philippines
Vietnam
Per capita real income relative to US
(Measured by the 1990 international Geary-Khamis dollars)
Sources: Angus Maddison, The World Economy: A Millennium Perspective, OECD Development Centre, 2001;
the Central Bank of the Republic of China; and IMF International Financial Statistics (for updating 1998-2006).
STAGE ONE
Simple
manufacturing
under foreign
guidance
STAGE TWO
Have supporting
industries, but still
under foreign
guidance
STAGE THREE
Technology &
management
mastered, can
produce high
quality goods
STAGE FOUR
Full capability in
innovation and
product design as
global leader
Vietnam
Thailand, Malaysia
Korea, Taiwan
Japan, US, EU
Agglomeration
Creativity
Glass ceiling for
ASEAN countries
(Middle income trap)
Catching Up Process
Initial FDI
absorption
Internalizing
parts and
components
Technology
absorption
Internalizing
skills and
technology
Internalizing
innovation
Lessons from
Thailand and Malaysia
Success Impressive industrialization and
growth led by FDI and reasonable policy
Failure Domestic private-sector capability
is still weak after many decades
Foreign dependencyinability to send
foreign managers home
Value and capability are not
internalizedmiddle income trap
Risk of wage pressure and FDI shift to
China/India/Vietnam
Vietnams Changing Sources of
Growth
Doimoi to mid 1990sliberalization effect
Mid 1990s to nowexternally driven
growth with large inflows of investment,
capital and aid
From nowcreation of internal value!
-2
-1
0
1
2
3
4
5
6
7
1
9
9
0
1
9
9
1
1
9
9
2
1
9
9
3
1
9
9
4
1
9
9
5
1
9
9
6
1
9
9
7
1
9
9
8
1
9
9
9
2
0
0
0
2
0
0
1
2
0
0
2
2
0
0
3
2
0
0
4
2
0
0
5
2
0
0
6
ICOR
TFP change (%)
Proposed Strategy for 2020
Building 3 pillars of industrial strength
-Supporting industries
-Industrial human resource
-Efficient logistics
Solving
social issues
Proper macro
management
Concrete targets
and action plans
Reform of industrial
policy formulation
Learning integral
manufacturing
Effective use of
capital inflows & ODA
Preconditions for industrialization
Internal Sources of
Manufacturing Strength
Supporting industries
SME promotion (HRD, technology, management,
finance), inviting FDI parts makers, industrial
standards, quality standards and testing, factory
doctors, FDI-local matching service, promoting SME
networking, trade fairs & reverse trade fairs, database
Industrial human resource
Vocational schools & training centers, Meister system,
curriculum reform, overseas education & training,
incentive for brain return
Efficient logistics
Transport & telecom infrastructure, efficient ports &
airports, trucking service, cargo handling centers,
trading companies, modern distribution network,
warehouse, customs and tariff systems
Malaysias Manufacturing++
Malaysias IMP3 (1996-2005) aimed at
raising and broadening the value chain
Leveling up of each
industrial cluster
-Core production
-Supporting industries
-Supporting services
-Human resources
-Logistics
-R&D
Endless pursuit of quality Quick results and flexibility Merits
Long-term relations,
building internal skills &
knowledge
Openness, quick decision
making, flexible
outsourcing
Institutional
requirement
s
It takes much energy and
time to achieve results
No differentiation, excess
entry, low profit, lack of
R&D
Demerits
Each product has unique
parts, specifically designed
Parts are common and can
be used for any model
Parts
interface
Integral
manufacturing
Modular
manufacturing
Performance
Time Time
Partnership Possibilities
From the Viewpoint of Business Architecture
Source: compiled from Prof.
Takahiro Fujimotos explanation
to the Joint VDF-MOI mission,
June 2005.
Japan=ASEAN (integral) and
USA=China (modular) can
form effective partnership
Vietnam-Japan Monozukuri Partnership
for Supporting Industries
For Leveling Up Vietnams Competitiveness in the Age of
Deepening Integration
A VDF Proposal
This proposal is based on VDFs past studies on specific
industrial sectors, supporting industries, and policy
formulation methodology.
It was drafted in June 2008 at the request of Japanese
Ambassador Mitsuo Sakaba (still confidential).
It is one of the ideas guiding the new bilateral program,
which is in the formation stage.
Vietnam-Japan Cooperation
in the Past
Japan has contributed to VNs industrialization
through trade, FDI, ODA and policy dialogue
Contribution of Japanese ODA
Infrastructure (esp. transport & power)
Skills and human resources
Institutional reforms
Improving business environment
Urban & regional planning
Removing negative impacts of growth
Poverty reduction
Action-oriented bilateral policy dialogue
Ishikawa Project (1995-2001)
New Miyazawa Plan (1999)
VN-Japan Joint Initiative Phase 1, 2, 3 (2003-09)
New Direction of
Vietnam-Japan Cooperation
Vietnam is no longer a low-income
transition country. Bilateral cooperation
should be more equal and strategic.
Cooperation should gradually shift from
removing negatives and providing basic
conditions to proactive creation of the
new source of competitiveness.
VDF proposes Vietnam-Japan strategic
partnership for internal value creation.
Sakichi Toyota
1867-1930
Konosuke Matsushita
1894-1989
Soichiro Honda
1906-1991
Akio Morita
(Sonys co-founder)
1921-1999
Monozukuri literally means making things in native
Japanese.
Pursuit of high quality and customer satisfaction as
the primary goal, with pride and dedication.
Many of Japans excellent companies were founded
by engineers full of monozukuri spirit.
The Concept of Monozukuri
Japanese Style Skill-based Manufacturing
Monozukuri cont.
Long-term relationship and skill & knowledge
building within companies and among companies
(assemblers-suppliers).
Strong demand for 5S, QCD (quality-cost-
delivery), kaizen, just-in-time, quality control,
and other efforts for constant improvement.
The Purpose
Establish bilateral partnership to jointly
produce high skill products through division
of labor:
Japan capital- & knowledge-intensive monozukuri
processes
Vietnam labor-intensive monozukuri processes
- Over time, as Vietnams skill improves, transfer
more knowledge from Japan to Vietnam
- The foreign partner (Vietnam) should work closely
with Japan, rather than just copying Japanese
products or infringing on IPR.
Vietnams Challenge
Large inflow of ASEAN products (esp.
Japanese brands) may destroy VNs
industrial base including FDI producers
(Daihatsu, Sony).
The risk of becoming a low-value, simple
product producer, or losing FDI to cheap
labor countries.
In the near future, VNs wage too high for
labor-intensive industries, but technology
too low for knowledge-intensive industries?
Vietnam needs a strategic alliance to level
up internal capability.
Japans Challenge
Japan has high technology, but faces high
wages and ageing population.
The 2007 Problem - Postwar baby boomers
(born 1947-49) with high skills began to retire
in 2007. Young workers are in shortage.
Situation will only get worse as time goes by.
Japan needs a young developing country as a
reliable partner in integral manufacturing
The inheritors of monozukuri tradition must be
found both domestically and abroad.
Proposed Features of
Monozukuri Partnership
Resetting the mindset Vietnamese managers
and workers must be more aggressive in learning
and marketing
Technology transfer and capacity building
Commitment to international standards
quality, safety, environment, intellectual property
Positioning in the global value chain jointly
specializing high skilled processes and outsourcing
other inputs
Reforming policy methodology
Non-exclusivity (win-win-win) other countries
can also enjoy the benefits of this partnership
Importance of
Supporting Industries
Supporting industry promotion is the first
important step to realize the vision of
monozukuri partnership.
Assembly-type manufactured products
large part cost (80-90%) vs. small
assembly labor cost (5-10%).
Without quick access to domestic suppliers
with QCD, assemblers cannot compete
(added transport cost + long lead time).
ASEAN4 have been promoting supporting
industries for a long time, but with limited
success in creating non-FDI suppliers.
The Concept and Scope of
Supporting Industries
MITI 1985 First official use of the term supporting
industries (susono sangyo)
Ichikawa 2005 Survey and scope of SI (included in
VDFs Industry Book)
VDF survey on VNs supporting industries (2006)
Thuy 2007 Concepts, scope, international experiences
(included in VDFs SI Book vol.1)
Sectoral vs. horizontal definition (by industry or by
process)?
VDFs working definition a group of industrial
activities which supply intermediate inputs (i.e., parts,
components, and tools to produce these parts and
components) for assembly-type or processing
industries (Thuy 2007, p.38E)
Mr. Kyoshiro Ichikawas Memo
(JICA Expert at MPI)
Definition - Supporting industries include (i)
production of parts for final products; (ii) processing
and treatment of such parts; and (iii) processing
and treatment of materials for producing such parts.
Promotion measures Supporting industry producers
Private SMEs
Educ. & training of managers;
tech. transfer (subsidies); SME
finance
SOEs Vietnam
ese
Taiwanese & others
FDI promotion, tax incentives,
industrial parks (for SI)
Japanese FDI
Human resource common issue for FDI and local
firms: (i) top management; (ii) middle managers; (iii)
workers
The Concept of Supporting Industries
Suggested Policy Menu
The following policy menu is constructed from:
- ASEAN4 experiences (MITIs New Aid Plan
1980s, Mizutani Report for Thailand 1999, Urata
Report for Indonesia 2000)
- Vietnams Supporting Industry M/P 2007,
Motorbike M/P 2007
- Opinions of Japanese and Vietnamese experts
This list is intended to be a reference for further
discussion only.
Policy Menu for Supporting Industry Promotion
(A tentative list for discussion)
Policy area Measures
1. Capacity building
(for specific firms)
- Shindanshi (enterprise evaluation) system
- TA for management and technology
- Large-scale mobilization of retired Japanese engineers
- Intensive support for limited sectors (e.g., die & mold)
- Awards, PR and intense support for excellent suppliers
2. Human resource
(general or
institutional)
- Management/technical centers and programs
- Large-scale mobilization of retired Japanese engineers
- Alliance between FDI firms and local universities/centers
- Monozukuri school (to be upgraded to university)
- Meister certification system
3. Finance - Credit guarantee
- SME finance institutions
- Two-step loans
4. Incentives - Exemption or reduction of taxes and custom duties
- Grants or loans for specified actions

Policy Menu (cont.)
5. Linkage - Database and matching service
- FDI-vendor linkage program
- Parts Industry Association and Business Study Meetings
- Trade fairs and reverse trade fairs
- Linkage with Taiwanese suppliers (motorcycles, electronics)
- Improving logistics between Hanoi and HCMC
6. FDI marketing - Creation of strategic industrial clusters
- Industrial parks and rental factories
- Efficient logistics and infrastructure
- FDI marketing targeted to specific sectors or companies
7. Policy framework - Supporting industry master plan
- SME law
- SME ministry
- Business associations and industry-specific institutes
- Quality standards and testing centers

Next Steps Proposed by the
Japanese Side
Japan (EoJ, JICA, JETRO, experts) and Vietnam (DPM Hai,
MOIT, VCCI) will set and implement concrete action
plans. VDF will support both sides.
Sep.2008 Kick-off Meeting for Supporting Industry
Action Plans (Hanoi, hosted by VCCI) - agreed
Jan.2009 Finalization of Action Plans and initiation
of implementation - agreed
Action Plan time scope:
End 2009 Short-term actions
2013 Medium-term actions
2020 Long-term actions
Action Plan also linked with:
- Vietnam-Japan Joint Initiative Phase 3 (2008-09)
- Bilateral cooperation after the conclusion of
Vietnam-Japan EPA (soon?)
Serious Reform in Policy
Formulation Needed
For continued industrialization for higher
income, Vietnam needs a fundamental
reform in policy formulation. Minor
changes of the current system are not
enough.
Government trap low efficiency, low
morale, low salary brain drain from the
public sector low-quality policies
Strong political will and risk-taking at the
top level (Prime Minister) is needed to
initiate such reform.
VNs Policy Process Is One and Only
K OhnoVietnamNet Interview (May 13&15, 2008)
VNs high growth owing to good location and good
workers; not because of good policy.
Industrial strategies are (i) not based on private
sector voices; and (ii) not implemented because
details are not given. These are unique
weaknesses in VN not seen in other countries.
The same people from planning years are still in
charge. I propose a technocrat group directly
under PM, as seen in many E Asian countries.
There are too many public-sector research
institutes which fail to produce useful policy
analyses. There should be competition among
policy-oriented research institutes to convince
policy makers and general public.
Vietnam: Traditional M/P Drafting Process
Prime
Minister
Minister
Drafting
Team
MPI & other
Ministries
Inter-
ministerial
review
Internal
review
Order
Submit
Review for
approval
Submit
Business Community
International
experts
Technical
assistance
(sometimes)
No permanent channel for continuous policy dialogue
(case-by-case, temporary, ad hoc)
Appeal letter to
Prime Minister when
problems arise
Contact
Ministry when
necessary
Interviews,
symposiums
(sometimes)
Government
MPI & other
Ministries
Data
Policy Formulation for Vietnam:
A Proposal
- Elite technocrat
group under strong
leadership of Prime
Minister
- Choose young,
well educated
officials and experts
- Simplify policy
authority and
procedure
Prime Minister
Technocrat Group
(Policy Maker)
Direction, full
authority for
policy making
Faithful
execution and
reporting
Policy,
guidance and
monitoring
Faithful
execution
and reporting
Ministries (Policy Implementers)
Experts
Donors
Multi-
layered
Model
Malaysia
Organization
for drafting
Industrial
Master Plan
Malaysia:
Drafting of Industrial Master Plan 2006-2020 (IMP3)
Business opinions reflected through TRGs and brainstorming
IPC: Industrial Planning Committee (headed by MITI Minister)
SC: Steering Committee (headed by MITI high official)
TRGs: Technical Resource Groups (headed by various experts)
Central Coordination Model:
Thailand under Thaksin 2001-06
Strong
Prime Minister
Policy direction
to be concretized
Relevant
Ministry
Experts
Private
Sector
Industry-specific
Institute
Direct
inputs
Industry-specific
Committees
Order
Master plans
Policy actions
Powerful & Intelligent Top Leader
Ethiopia under PM Meles Zenawi (1991-now)
Strong
Prime Minister
<Policy Structure>
Vision - ADLI 1991 (Agri.
Dev. Led Industrialization)
Agricultural Strategy 2002
Industrial Strategy 2003
Other sectoral strategies
Medium-term strategies
3-year budget frameworks
Leather M/P (UNIDO, Italy)
Garment M/P (UNIDO)
Flower M/P (Dutch support)
Annual budgets & reviews
Advising PM: EDRI,
Economic Ministers?
Ministries as
implementing bodies
Using Individual donors to
execute specific projects
UNIDO Italy Nether.
MoTI MARD
-Advice by Stiglitz
-Doing PhD in UK
-Writing books
?
Top-down
Decisions
MoFED
Under New Policy Structure
Vietnam Should Achieve:
Strong top-down visions from PM;
sufficient authority for technocrats to
execute orders
Action-oriented policy making with speed
and flexibility
Removal of bureaucratic formalism &
inaction, nepotism, corruption, secrecy
Public sector - competitive recruitment,
sufficient salary, internal promotion
mechanism, pride to serve the country
Summary of Policy Advice
Reform policy formulation greatly, by
creating a technocrat group under Prime
Minister.
Draft an overall industrial master plan
with concrete goals and action plans up to
2020.
Vietnam-Japan Monozukuri Partnership
should be an important part of this 2020
vision.
Financial and Enterprise
Reform in Vietnam
August 2008
Tokyo, Japan
Noritaka Akamatsu
The World Bank
Agenda
Landscape of Vietnams financial sector
Banking sector reform and development
Landscape of the banking sector
Enhanced banking environment
WTO commitments in the banking industry
Outstanding issues and challenges
Capital market development and SOE Equitization
Stock market development
SOE equitization and SCIC
Institutional investors and bond market
Recent macroeconomic turbulences, policy responses
and challenges ahead
Landscape of Vietnams financial
sector
Bank assets 150% of GDP (VND 1,872 trillion) of which
credits are 90~95%.
5 state-owned commercial banks (SOCBs) with 4 large ones
occupying 55% of the banking sector in credits;
36 joint stock banks (JSBs) with 30% market share,
5 joint venture banks, and
37 foreign bank branches.
12 finance companies and 13 leasing companies
Stock market capitalization 20% of GDP with 300 listed
companies. About 3,000 unlisted public companies.
124 securities companies (84 in operation) and 34 asset
management companies managing VND 51 trillion.
Market cap surpassed 40% of GDP in mid 2007 but
subsequently declined to the half despite continued listing.
22 lifes and 9 non-lifes with total assets VND 58 trillion.
Banking sector reform and
development
Snapshot of the banking sector in
Vietnam
5 SOCBs with 4 large ones including Vietcombank. The
asset size varies VND 170 ~ 320 trillion.
The asset quality and capital adequacy vary.
36 JSBs. The asset quality appeared good, and capital,
adequate. But since 2007, they grew unreasonably fast.
The asset size of leading JSBs up to VND 80 trillion.
Minimum cap: VND 1 trillion, and VND 3 trillion by end 2009.
15 new applications by major enterprise groups.
37 foreign bank branches and 5 J/V banks. 100% foreign-
owned subsidiary permitted but approved slowly.
30% credit growth till 2006. But from April 07, 63%, with
much flowing into the real estate. SBV is tightening the
liquidity and credit growth to cool the inflation.
Enhanced banking environment
Inter-bank payments system (IBPS)
Core banking systems adopted by 13 banks
Fast spreading ATMs (5,000), networks being integrated.
Smartlink and Banknetvn connection signed.
Card switching center/clearing house in preparation.
Gradual de-dollarization and increasing non-cash payments
Government trying to use non-cash payments.
Deposit insurance
Credit Information Center
Gradually leveled playing field for foreign and domestic banks
High growth of JSBs and gradual restructuring of SOCBs
WTO commitments in the Banking
industry
Leveling the playing field between foreign and domestic banks.
Already open for cross-border services and consumption.
Gradual liberalization of commercial presence / FDI:
A 100% foreign-owned subsidiary (from April 2007) or a J/V bank
can be created by foreign banks with more than $10 billion assets.
A branch can be created by foreign banks with more than $20
billion assets.
Foreign equity participation in a local bank (SOCB or JSB) is
permitted up to only 30%.
Foreign bank branch is now permitted to have ATMs and other
transaction points outside the branch itself ?
Foreign banks are allowed to issue credit cards.
Foreign bank branches are allowed to take VND deposits only
gradually (full national treatment from January 2011). 100%
foreign-owned subsidiaries will not be subject to this rule.
Outstanding issues and challenges
Still under-banked population (10%) particularly in rural areas.
Slow SOCBs equitization, NPL resolution and lack of
transparency.
Too high growth of JSBs with exposure to real estate market
and likely generation of new NPLs.
Still low rate of adoption of core banking systems.
Need to monitor the performance of individual loans to apply Article
7 of Decision 493 and Decision 457.
Under-developed and volatile money market, resulting skewed
allocation of liquidity and high service charges.
Financial conglomeration, and attempts to establish new banks
by major SOE groups.
Need consolidation (M&A) and growth to compete and survive.
Lack of skilled human resources.
SOCB equitization
NPL resolution & recapitalization
Vietcombank, 14.4% (2007); BIDV, 3.6% (2007); Agribank, 7~16%
(2006); and Vietinbank, ??
Clean up the balance sheet (sale of NPLs, provisioning, writing off,
etc.) and recapitalize them before equitization (with government
bonds, convertible bonds/subordinate debt, new equity, etc.).
The single foreign investor cap at 15% (20% with PMs approval).
Total foreign investment cap at 30%. The State will retain 51%.
Land use rights not included in the balance sheet of SOCBs while
buildings already depreciated (i.e., modest fixed assets).
Manageable fiscal cost, but dont allow moral hazard.
Restructuring & institutional capacity building
Enhance the corporate governance.
Centralize the credit decision making.
Upgrade MIS and train staff.
SOCB equitization continued
Vietcombank failed to attract strategic investors.
Too aggressive valuation.
BIDV, Incombank and VBARD have been given recap of VND
11 trillion while issuing subordinate debt and writing off bad
loans.
Issued sub-debt in 2006 and 2007 (Vietcombank issued
convertible debt in 2005).
BIDV: VND 3.4 trillion; Vietinbank: VND 3 trillion, in government
bonds/cash. VBARD will need more.
Land use rights not included in the value of SOCBs.
Except BIDV (and Vietcombank after IPO), other SOCBs have
not disclosed IFRS-audited financial statements.
Core banking system is needed to enable loan classification
according to Art. 7 of Decision 493. Otherwise, it will continue
to be difficult to value for equitization.
Incombank and VBARD are behind the schedule.
Strategy of some foreign banks
Different types of strategic foreign investors
Long-term partner vs. return-focused medium-term investor
Wholesale banking-focused (i.e., branch-based) vs. retail-focused
(i.e., investing in JSB or SOCB)
For equitization, should SOCB find a long-term partner or
return-focused medium-term investors? Need to avoid a loan
buyback provision, and to do so, clean up the balance sheet.
ANZ has a stake in Sacombank, HSBC in Techcombank,
Standard Chartered in ACB, Sumitomo Mitsui in Eximbank,
Deutsche in Habubank, Societe Generale in SeA Bank, OCBC
in VP Bank, UOB in Southern Bank, Maybank in An Binh Bank,
Many have applied for establishing a 100% subsidiary.
Many have two branches, in Hanoi and HCMC.
Are they also interested in investing in SOCBs? Seemingly, yes.
Why?
Valuable options.
JSB development
Two types:
Larger and competitive JSBs, 9 with foreign participation.
State controlled or small regional JSBs.
Can the former grow to survive by aggressive capital increase
and expansion?
Too aggressive credit growth though good capital.
What if their foreign partners want to establish a 100% owned
subsidiary instead?
ANZ, HSBC, Standard Chartered and more?
How will the latter survive?
Already in liquidity shortage amid SBVs tight monetary policy.
Consolidation through merger with larger banks? Partnership
with foreign banks?
SOE groups (EGs and GCs) are trying to establish banks.
Are more JSBs needed?
Bindings on SOEs creation of
banks
Bank governance rules (Decision 24, June 7, 2007)
20% maximum for a single institutional shareholder.
Minimum of 100 shareholders including three with at least VND 2
trillion in assets => a public company required to disclose audited
financial statements periodically.
Shareholders are not allowed to sell their stakes in the first three
years (founding shareholders, five years).
Prudential rules
Connected lending limit
Single borrower exposure limit
Competition law
Limit on excessive concentration of economic power.
Transparency requirement for affiliated party transactions and
transfer pricing.
SBV should be able to reject an application on the ground of
lack of necessity or inadequate business plan. But
SOE Equitization and Capital
markets
Capital markets and their
participants
Stock market capitalization jumped from 2% of GDP at 3Q
2006 to 40% in mid 2007 before it declined to 20% today.
Massive listing toward the end of 2006. Listing continued
throughout 2007 to reach nearly 300 now.
Equitization produced over 3,000 public companies.
The number of securities companies jumped from 14 in 3Q
2006 to 124 (84 of which are in operation) in 1Q 2008.
34 investment management companies managing 4 closed
end public funds, 16 membership funds and hundreds of
private customer funds (VND 51 trillion in total).
22 life insurance companies, 9 non-lifes and 1 reinsurance
company (and 8 insurance brokers) manages VND 58 trillion
assets.
Market infrastructure and
institutions
Securities Law issued in June 2006 and became effective in
January 2007 to replace Decree 144.
Re-established the State Securities Commission as part
of MOF.
Required all public companies to disclose audited
financial statements annually.
State Securities Commission (SSC)
The regulatory and supervisory authority.
Currently registers unlisted public companies but has not been
able to capture all (about 1,000 out of 3,000).
Ho Chi Minh Stock Exchange (HoSE) and Hanoi Stock
Trading Center (HaSTC).
HoSE for larger companies and HaSTC for smaller companies
and bonds. But HaSTC lists some important companies.
Does Vietnam need two stock markets??
Market infrastructure and
institutions continued
Vietnam Securities Depository (VSD)
Houses all publicly traded securities including both stocks and
bonds (incl. government bonds).
BIDV has been serving as the settlement bank.
Risky for both market participants and BIDV if BIDV has to take the
settlement risk or counterparty credit/liquidity risk.
Investors are required to submit securities / payments ahead of
trading. It eliminates the risk of default but at large cost /
inconvenience for investors.
Government bond market needs to move to real time gross
DVP settlement (preferably CPSS Model 1).
The money settlement function needs to be transferred from BIDV
to SBV, and VSD needs to be linked with Inter-Bank Payments
System (IBPS) for real time gross DVP.
The SBV needs to provide a flexible / inexpensive intraday
overdraft facility. => Need coordination with Monetary Policy.
Challenges ahead
Enforcement of Securities Law and regulation.
Information disclosure by public companies. The capacity and
expertise of the auditing industry cannot meet the demand of the
public companies while the latter lacks awareness of the need to
comply with this rule.
Transfer of public company registry from the SSC to VSD.
E-disclosure system to reduce the compliance cost for public
companies.
Consolidation of small securities companies into larger, well
capitalized ones capable of underwriting large scale public
offerings.
Need to permit securities companies to offer short-selling,
securities lending and margin account services but only after
consolidation.
Need strong corporate governance and Chinese Wall
Challenges ahead continued
Time to revisit the Securities Law and the status of the SSC.
Should the SSC be made an independent agency?
Should there be two exchanges?
Rudimentary market infrastructure needs fundamental
upgrading to cope with the increased volume and support
investment needs of market participants.
HoSE needs remote access and continuous auction market.
VSD needs to support securities lending, repo clearing, collateral
management with mark-to-market valuation.
Integrated supervisory infrastructure required (e.g., automated
market surveillance system, e-disclosure system for public
companies, e-registration system for licensed market
professionals, etc.)
New instruments, new trading rules, etc.:
Derivatives (index, FX, interest rate/Government bonds)
Short-selling, securities lending, margin account
Securitization
Equitization
Nearly 4,000 SOEs have been equitized, leaving about 1,500 in
the hands of the State.
Yet, the State assets privatized are only about 20%, leaving
about 80% in the hands of the State. I.e., mostly smaller SOEs
have so far been equitized (exceptions, Bao Minh,
Vietcombank)
In order for a business to be listed in the exchanges,
It must be a public company under Securities Law
It must have sold at least 20% shares to the public.
Equitization does not provide full clarity as to what assets will
be included in the balance sheet (particularly with respect to
real estate).
Causing a loss to the State through a sale of State assets is a
crime under Vietnamese law.
Vietcombank equitization
Failed to attract strategic investors ahead of IPO
Reportedly, the asking price was too high (VND 100,000?).
VAS-based financial statement made valuation difficult (e.g.,
NPLs based on Art. 6 of SBV Decision 493) while assets to be
included in the balance sheet were not made fully clear.
IPO set a very high price (VND 107,680). According to
HSBC, the price represented PER of 87.
Vietnams regulation requires that strategic sale after IPO not
be priced below IPO price.
Few strategic investors, if any, would be willing to pay.
Many JSBs, which eagerly awaited to follow after a
successful equitization of Vietcombank, postponed IPOs and
listing. Set an awkward precedent for other SOCBs.
Role of SCIC
State holding company having state stakes in equitized
enterprises.
Have State stakes of about 1,000 equitized enterprises and is
expected to have about 1,500.
Most of those are controlling stakes, which makes SCIC
different from a conventional portfolio investor.
Modeling after Temasek of Singapore or Khazanah of Malaysia
but also bearing characteristics of privatization funds in
Eastern Europe and former Soviet states.
Optimize the value of the strategic assets of the State.
Sell non-strategic assets, contract the portfolio size in terms of
number of enterprises and focus on building the value of
strategic ones
Participate in the corporate governance of enterprises,
promote sound corporate governance and restructuring.
A challenging task.
Equitization and challenges ahead
Vietnams investor base is thin.
Small institutional investors and contractual savings
institutions.
Offering of large SOEs would put a sell pressure on the
market, depressing the price and causing indigestion.
Need traffic control in IPOs.
Need to sell to strategic investors, and to do so, need to
create conditions friendly for them.
Strategic placement before IPO.
IFRS-based financial statement.
Negotiated sale, even if its price turns out to be lower than
subsequent IPO, should not be viewed as a loss made to the
Sate and, therefore, a crime.
Recent macroeconomic turbulences,
policy responses and challenges
ahead
Recent macroeconomic
turbulences
Accelerating inflation (27%) and suddenly large trade/
current account deficits (latter 10% of GDP).
The global inflation in commodities including oil, steel,
cement, food and agricultural products fueled by the
domestic demand and chronic excess liquidity in 2007.
Aggressive public sector investment through SOEs in
infrastructure fueled local demand amid deteriorating
investment efficiency (ICOR).
Q. Surprising??
SEDP aims to invest 41% of GDP while Vietnams gross
domestic savings have been around 32%.
Vietnam was by design expected to have a current account
deficit worth 9% of GDP despite expected strong
remittances. => The trade deficit was expected to be
substantially larger than the current account deficit.
Inflation and capital flows
Despite the massive trade account deficit, Vietnam has been
accumulating FX reserves due to massive capital inflows
since 2006 (due to the anticipation of WTO accession?).
Remittances (reducing the current account deficit).
FDI, FPI, ODA, global bond issue, cross border borrowings by
FDI firms and foreign banks.
Particularly FDI (USD 21 billion in 2007, USD 44.5 billion
through July 2008) for the economy of USD 70 billion GDP
(although disbursement is less and comes with a time lag).
VND exchange rate peg to USD created Impossible Trinity.
The strong expansionary pressure on VND liquidity fueled the
massive credit growth (63% in 1Q 2008).
SOCBs, 24%. JSBs, over 100% (crazy!)
The credits had flown into the real estate market.
Policy responses and challenges
ahead
Monetary tightening
Contractionary monetary policy since early 2007
Tightening of the prudential rules (SBV Decision 457) and
Directive 03 of 2007 to limit the credit flow into stock market.
Tight liquidity with interest rate cap (lending rate 21%).
Fiscal austerity measures
Cut inefficient projects while postponing not urgent ones.
But must avoid over-killing
Real estate market is said to have declined by 60% (Really?).
Likely under collateralization.
Critical infrastructure service providers, export sector
Banking sector making barely any money if not losing money
with the tight interest rate spread.
Interest cost could dent corporate earnings while causing NPLs
in the banking sector.
Policy responses and challenges
ahead continued
Currency crisis unlikely (NDF not a good indicator). But
some pressure on VND possible to continue due to the
inflation and strong imports.
In short-run, FDI inflows will be key.
In long-run,
The export sector must be strengthened while import
substitution needs to be promoted; and
The domestic long-term savings must be mobilized with
institutional investors and contractual savings.
Bond market (as well as the stock market) must be
developed.
Thank you.












AFTERNOON SESSION
<Paper Presentations>

34
Poverty, Inequality and Ethnic Minorities in Vietnam
*



Katsushi Imai
Department of Economics, School of Social Sciences,
University of Manchester, UK
E-mail: Katsushi.Imai@manchester.ac.uk


Raghav Gaiha
Faculty of Management Studies,
University of Delhi, India
E-mail: rgaiha@camail.harvard.edu


Corresponding Author:
Katsushi Imai (Dr)
Economics, School of Social Sciences
University of Manchester
Oxford Road, Manchester M13 9PL, UK
Phone: +44-(0)161-275-4827; Fax: +44-(0)161-275-4928;
E-mail: Katsushi.Imai@manchester.ac.uk




Abstract
The present study examines how and why ethnic minorities are poorer than ethnic
majorities in Vietnam using the VHLSS data for 2002 and 2004. First, the analysis
confirms that households belonging to the ethnic minority groups are not only poorer but
also more vulnerable to various shocks than those in the ethnic majority groups, namely
the Kinh and the Chinese. Second, household composition (e.g. dependency burden),
education, land holding, and location are important determinants of expenditure and
poverty, whilst there is some diversity among different ethnic groups. Finally, the
decomposition analyses reveal that the ethnic minorities are poorer not necessarily
because they have more disadvantaged household characteristics (e.g. educational
attainment or location), but, more importantly, because the returns to the characteristics
are much lower for ethnic minorities than for majorities. Government policies to reduce
structural differences between ethnic majorities and minorities are imperative to address
the disparities in returns to endowments between them.


*
This paper is based on the project authors carried out for the Asia and Pacific Division of the IFAD, under
the overall supervision of Dr. Ganesh Thapa, Regional Economist for Asia and the Pacific region. The first
author acknowledges the support from the University of Manchester and from Doshisha University to carry
out this research. However, only the authors are responsible for any errors.

35
1. Introduction
Vietnam is a multi-ethnic country with 54 ethnic groups, each has its own language,
lifestyle and cultural heritage. The most dominant group is called Viet or Kinh, which
accounts for 86 % of the population of about 84 million and is concentrated in inland deltas
and coastal areas. They are generally richer than minority groups, with easy access to
infrastructure, health services, and education. Hoa or Chinese is another relatively rich
group that also inhabits mainly in inland deltas and coastal areas. In the present study, we
define ethnic majorities as the Kinh and the Chinese, and ethnic minorities as the ethnic
groups other than these, following van de Walle and Gunewardena (2001). Many studies have
shown that ethnic minorities are concentrated in upland and mountain areas where access to
infrastructure or health and educational facilities is limited and they are much poorer than the
ethnic majorities (e.g. van de Walle and Gunewardena, 2001; ADB 2002; World Bank, 2004;
Gaiha and Thapa, 2006; Imai, Gaiha and Kang, 2007). Indeed, the issue of poverty in Vietnam
cannot be addressed without analysing the poverty of ethnic minorities, as their poverty
headcount ratio was 64.3% in 2002, almost three times larger than that of ethnic majorities
(22.3%) (see Table 1). The share of ethnic minorities among the poor in the whole nation rose
from 20% in 1993 to 30% in 2002 due to poverty reduction in ethnic majority groups and
poverty stagnation in the minority groups (World Bank, 2004). While the national poverty rate
fell from 58.1% in 1993 to 37.4% in 1998 and to 28.9% in 2002
1
, and Vietnam had already
achieved the Millennium Development Goal of halving income poverty by 1998, poverty
reduction of the ethnic minority groups is still an important policy concern.
Why disparities in well-being and in poverty rates persist between the ethnic majorities
and minorities is far from obvious. It may be asked, for example, whether ethnic minorities are
poorer simply because they are located in remote areas or because they do not have enough
human or physical capitals, such as education or land, or because of any structural constraints
(e.g. social exclusion)? To address this question, van de Walle and Gunewardena (2001)
applied the Oaxaca-Blinder decomposition of wage inequality into two components: one due
to differences in socio-economic characteristics and the other due to structural factors or
differences in the returns to these characteristics (Oaxaca, 1973; Blinder, 1973). Their analysis
was confined to expenditure of households mainly in Northern Vietnam using the Viet Nam
Living Standards Measurement Surveys (VNLSS) in 1992-1993. They show that, without
commune fixed effects, about one- half of the expenditure inequality between the ethnic
majorities and minorities is explained by the characteristic component and another half by the
structural component, whilst most of the expenditure inequality comes from the structural
factor once commune fixed effects are taken into account for selected communes where both
majority and minority groups are found
2
. We extend van de Walle and Gunewardena (2001)
using more recent and larger household data sets, namely, Vietnam Household Living
Standards Survey (VHLSS) for 2002 and 2004. First, given the possible diversity among the
ethnic minority groups, we will estimate the expenditure function separately for each ethnic
group. Second, we will use some recent decomposition methods applied to analyse persistent

1
Poverty rates used here are based on the international poverty line which was devised by the Vietnamese
General Statistics Office (GSO) to reflect food expenditure for an intake of 2100 calories a day and
corresponding non-food expenditure. The basket of food and non-food items is determined by the
consumption patterns of the third quintile of households in terms of per capita expenditure. The poverty
lines were VND 1.16 million per person per year in 1993, VND 1.79 million in 1998 and VND 1.92 million
in 2002. In the present study, we use the same national poverty line and adjust it for 2004, based on the
annual CPI. We have not used the poverty lines developed by the Ministry of Labour, Invalids and Social
Affairs (MOLISA), which reflect the regional disparity in rice consumption.
2
They selected those communes to avoid the problem of missing regressors. We do not take this approach
as we can find the same set of regressors in our larger data (i.e. VHLSS in 2002 and 2004) for communes
where either an ethnic majority group or a minority group is found

36
poverty among the scheduled caste and tribes in India (e.g. Borooah, 2005; Kijima, 2006;
Gang, Sen, and Yun, 2008; Gaiha et al., 2007). For example, we will carry out the
decomposition analysis not only for expenditure inequality but also for differences in poverty
levels of ethnic majorities and minorities. Moreover, we will disaggregate the decomposition
of expenditure inequality and poverty differences into the effects of each explanatory variable.
Our analysis is thus designed to throw additional light on persistent ethnic poverty and
inequality.
The rest of the paper is organised as follows. In the next section, we will first review the
data, and then comment on the estimates of poverty and vulnerability, and their correlates.
Section 3 discusses the methodology used to analyse poverty of ethnic minorities, including
the decomposition methods. Detailed econometric results are discussed in Section 4. The final
section offers some concluding remarks.

2. Data and Descriptive Statistics
2.1. Data
3

Most of the poverty assessments in Vietnam are based on Vietnam Living Standards
Surveys (VLSS) in 1992/3 and 1997/8, which covered 4,800 and 6,000 households,
respectively. Of these, about 4,300 households constitute a panel data set. The surveys were
designed to collect detailed data on households, communities, and market prices. While VLSS
were widely recognised as high quality, they required additional surveys, called Multi-Purpose
Household Surveys (MPHS), to provide estimates at provincial level due to the relatively
small sample size of VLSS. In 2002, VLSS and MPHS were merged into Vietnam Household
Living Standards Survey (VHLSS) to cover the larger sample of households with some
simplification of the questionnaires to minimize measurement errors. VHLSS is planned to be
carried out every two years until 2010.
VHLSS is supposed to have two modules: the core module includes topics which are
important and change rapidly over time, while the rotated module focuses on those that change
less often. However, VHLSS in 2002 contains only the core module. It covers a wide range of
data, including household composition and characteristics (e.g. education and health),
expenditures on food, non-food items, health and education, income by source (e.g. wage and
salary, farm or non-farm production), employment and labour force participation, housing,
ownership of assets and durable goods, local infrastructure and commune characteristics. The
sample size of VHLSS 2002 is 75,000 households, of which 30,000 households were
interviewed with all topics, and 45,000 with all topics except expenditure. Only the former is
used for the present study, as our focus is on income/expenditure poverty. Because of missing
observations for some variables, the final sample size is 28,806.
VHLSS in 2004 consists of the core module virtually identical to the 2002 survey, and
the rotated module on agricultural activities and non-agricultural household business, and
borrowing and lending activities. The total number of households is 45,000, of which 9,000
households were interviewed with all topics, and 36,000 households with all topics except
expenditure. We use only 9000 households interviewed on all topics. Due to missing
observations, the final size is 6,473. Given the larger sample size of the survey data in 2002,
we will mainly use the data in 2002 and supplement them by the data in 2004.


3
This sub-section draws upon Imai, Gaiha and Kang (2007).

37
2.2. Poverty among Ethnic Minorities
This sub-section focuses on socio-economic characteristics of ethnic majorities and
minorities as well as their sub-categories. Our comments are brief and selective.
First, 57% of the ethnic majority groups, the Kinh and the Chinese, live in Inland Delta,
while the corresponding figure for the ethnic minority groups is just 10%. The ethnic
minorities living in Inland Delta consists mainly of the Khmer, which is known as the Khmer
Krom, who inhabited the delta of the Mekong long before the arrival of the Vietnamese. The
ethnic majorities also live in low and high mountains (29%), coastal area (8%), and hills (7%).
Few ethnic minority groups live in coastal area and hills- 62% of them inhabit high mountains
and 27% are in low mountains. There are a few ethnic groups located mainly in low mountains,
such as the Muong, the Sandiu and the Stieng, but most of the minority groups are based
primarily in high mountain areas.
Then we compare the measures of disadvantage of ethnic majorities and minorities. It is
confirmed that (i) a majority of the ethnic minority groups live in remote areas (defined
subjectively by the survey); (ii) about 90% of the ethnic minority people live in rural areas;
(iii) the ethnic majority groups do not have easy access to the market, or medical care. While
the degree of disadvantage varies across different ethnic minority groups, all the minority
groups are geographically more disadvantaged than the majorities in terms of market access or
health services.
Table 1 summarises poverty and vulnerability measures for ethnic majority and minority
groups. Poverty head count ratio is based on the national poverty line, and two alternative
cases where 80% and 120% of the poverty line are considered to check the sensitivity of the
results. Vulnerability measure is defined as the probability of falling into poverty in the next
period, following Chaudhuri, Jalan, and Suryahadi (2002).
4
The results for VHLSS data for
2004 are given in Table 2.
5
Only key findings from these tables are summarized below.
(Table 1 and Table 2 to be inserted)
First, the poverty headcount ratio of the ethnic majority groups remained much lower
than that of ethnic minority groups in 2002 and 2004 (22.3% in 2002 and 20.2% in 2004 for
the majority
6
, and 64.3% in 2002 and 62.1% in 2004 for the minority). Second, the poverty
head count ratio varies among ethnic minority groups, ranging from 28.1% for the Khmer,
50% for the Nung, to 95.2% for the Bana. This suggests that policy efforts are necessary to
provide more intensive support for the poorest and the most disadvantaged ethnic groups.
Third, this pattern is largely unchanged if 80% or 120% of the poverty line is employed.
Finally, the difference in vulnerability of ethnic majority groups and minority groups is much
higher (7.3% versus 62.1%) than the difference in poverty (22.3% versus 64.3%). This implies
that ethnic minorities are much more vulnerable to various shocks (e.g. sudden weather
changes or illness of household members) than the majority groups.
7
This implies that
government policies designed to augment household incomes alone are not likely to be
effective in reducing poverty among ethnic minorities in the long run. More attention needs to
be given to social safety nets or insurance to protect the vulnerable ethnic minorities from
shocks.
Then we compare the average household characteristics of the ethnic majority and

4
See Appendix for computational details of the vulnerability measures.
5
As the sample size differs in the VHLSS for 2002 and 2004, some caution is required in comparisons of
the results.
6
A further disaggregation shows that poverty head- count ratio of the Kinh decreased from 23.2% to 20.2%,
while that of the Chinese increased from 14.6% to 20.2%. However, the latter should be interpreted
cautiously because of the small sample size of Chinese households (212 in 2002 and 63 in 2004).
7
See Imai, Gaiha, and Kang (2007) for more details.

38
minority groups in 2002. The findings are briefly summarized below. First, the average age of
household head is 5 years higher for the majority groups (48.2 years old) than for the minority
groups (43.2). However, the Khmer is an exception as the average age of the head is high
(50.3). Second, the average household size of the minority groups is larger (5.35) than that of
the majority groups (4.34). This reflects the heavier dependency burden among the ethnic
minority groups. 7.8% of the households of ethnic majority groups have members who
completed higher education, while the corresponding figure of ethnic minorities is only 2.5%.
The share of households with the educational level at upper secondary school is 18.3% for the
former and 8.8% of the latter. Third, ethnic minority groups have larger areas of land than do
ethnic majorities without any exception. Disaggregation of the total area of land into
subcategories shows that ethnic minorities hold larger areas for all the categories except for
aquacultural water (i.e. for agricultural land, syvilcultural land and unused land). Fourth,
despite some variation across different ethnic minority groups, they have generally similar
characteristics (e.g. large household size, low educational attainments). Recent anthropological
and other related studies focusing on the disparities among different ethnic groups or different
regions are generally consistent with the above findings.
8
For example, McElwee
(2006),focusing on the relationship between minority groups and the Kinh in the Annamite
uplands, reports that social and economic inequality worsened due to unequal access to
markets, government services and political representation. Scott and Chuyen (2004), on the
other hand, demonstrate that regional disparities stemmed from some regions limited access
to resources, information, and social infrastructure for entrepreneurial and other development
activities. In an important variation, Fforde (1998) draws attention to differences in capacities
to work in a process of adjustment of structure of household earnings to changing
circumstances. From a broader methodological perspective, he questions the homogeneity
assumption that underlies some recent contributions (Fforde, 2005). If, for example, attitudes
towards risks and insurance vary in different groups-as illustrated by Fforde (1998)- it is
necessary to go beyond physical and human capital endowments and market failures to reduce
vulnerability. Specifically, more careful attention must be given to correcting community
failures (e.g. in protecting the old, and orphans). We will address some of these issues using
econometric techniques.

3. Methodologies
3.1. Determinants of Consumption
First, we estimate
ijk
W ln or log of per capita expenditure of the i-th household in the
j-th ethnic group (e.g. majority, minority, the Kinh, or the Khmer) living in the k-th commune
taking into account a vector of socio-economic characteristics,
ijk
X , commune-level fixed
effects,
ij
q , and a random error term,
ijk
c , which is orthogonal to the explanatory
variables;
j
o is a constant error term.
n ..., , 1 j , K ,..., 1 k , N ..., , 1 i X W ln
ijk ij j ijk j ijk
= = = c + q + | + o = , (1)
ijk
X includes age of household head, the share of female household members, dependency
burden (the share of members whose age is below 15 years old or above 65 years old),
whether a household head is married, the maximum educational attainment of household
members, and areas of owned land and its square (for agricultural land, syvilcultural land,
aquacultural water, unused land).

8
This review draws upon Imai, Gaiha, and Kang (2007).

39
In an alternative specification, the commune fixed effects are dropped, as shown below:
ij j 2 j j 1 ij j ij
D X W ln c + | + | + o = , (2)
where a vector of dummy variables,
j
D , namely regional dummy variables (e.g. whether a
household is living in High Mountains, Coastal Area etc.) and dummy variables on religion
(e.g. whether the main religion for the commune is Buddhism) are added.

3.2. Determinants of Poverty
Here the same set of explanatory variables is used to analyse the determinants of poverty
based on the logit model. The dependent variable is whether a households expenditure is
below (=1) or above (=0) the national poverty line.
It is sometimes asserted that a model that analyses the probability of per capita income
falling within a given interval (say, below the poverty line) for a set of characteristics throws
away data that will be used in a model of income determination in which income per capita is
regressed on the characteristics in question. Two observations are pertinent. First, income
distribution data typically contain non-negligible errors -especially in the lower and upper tails.
These errors arise partly from imputation of prices to non-marketed inputs and outputs in the
rural areas and partly from under-reporting of high incomes. Given these errors, it is safer to
analyse the probability of per capita income falling within a specified interval. Second, in case
there are reasons to believe that per capita income distribution follows a non-normal
distribution and depends on the characteristics in question in more complex ways than a mere
location shift, it is conceivable that the logit or probit model used here may yield more robust
results than a direct regression of per capita income on these characteristics (Amemiya, 1981,
and Gaiha, 1988).
To take account of commune fixed effects, a conditional fixed-effects logistic model is
applied as follows.
9

( )
( )
( )
ikj ij
ikj ij
j ij ikj
X exp 1
X exp
, 1 y P
+ q' +
+ q'
= |' q' = , (3)
ij
q' is the commune fixed effects as above. Alternatively, a probit model is applied without
commune fixed effects, but with regional dummies (e.g. whether a household is located in
high mountains).
( ) ( )
}
'
|

| = |' =
j
dt t 1 y P
j ij
(4)
( )
ij j
X |' u =
The function ( ) . u denotes the standard normal distribution.

3.3. Decomposition of Expenditure and Poverty
A decomposition analysis of the differences in poverty (or of the expected probability of
poverty) and expenditure of ethnic majority and minorities is carried out. As noted earlier, this
relies on the Oaxaca-Blinder (1973) decomposition of the wage difference into two
components: the characteristic component, associated with the average levels of characteristics

9
A fixed-effects logistic model is chosen, as a fixed-effects probit model cannot be estimated by standard
commands available in Stata.

40
(e.g. education), and the structural component, related to returns to these characteristics.
Let us first consider the poverty decomposition. Denoting the average (predicted)
probability of being poor among the ethnic majority groups and ethnic minority groups as
maj
P and
min
P , respectively, the decomposition is obtained as:
( ) ( ) | | ( ) ( ) | |
min min maj min maj min maj maj min maj

, X p

, X p

, X p

, X p P P | | + | | = , (5)
Here the subscript, i, j or k, is suppressed to make the notation simpler. It is noted that the first
bracket contains the characteristics component and the second the structural component. In the
first component, the differences in characteristics are evaluated using the coefficient estimates
for the ethnic majority groups (
maj

| ) as the reference group. In the second, the characteristics

of the ethnic minority groups are evaluated taking into account the differences between
maj

| and
min

| . The second component is sometimes considered a measure of discrimination.

Along the lines of Kijima (2006), two observations are in order: (i) the first component, based
on differences in characteristics, could itself reflect discrimination over a period; and (ii) the
lower returns to land among the ethnic minorities, on the other hand, could be lower simply
because of locational disadvantages. This implies that this component could be non-zero even
if there is no discrimination in the sample year.
In order to disaggregate the characteristics and structural components, we take advantage
of a decomposition procedure proposed by Yun (2004). All that is needed is to disaggregate the
characteristics and structural components using two sets of weights.
( )
( ) maj min maj
maj
i
min
i
maj
i
i
X

X X

X X
|
|
= e
A
,
( )
( ) min maj min
min
i
maj
i
min
i
i

X

X
| |
| |
= e
| A
and

=
=
| A
=
=
A = e = e
K i
1 i
i
K i
1 i
X
i
1
These weights are defined for individual variables, i=1, 2,, K, and add up to 1.

4. Results
4.1. Regression Results
In this section, we will discuss the econometric results obtained from the model
specifications in Section 3. First, the key findings are summarised.
Table 3 contains the results on the determinants of log of per capita expenditure by each
ethnic group in 2002, with commune fixed effects.
(Table 3 to be inserted)
First, the pattern of the regression results is generally similar for ethnic majorit y and
minority groups. For example, in both cases, the coefficient of dependency burden is
negative and highly significant. That of whether a household head is married is also negative
and significant.
10
On the other hand, the coefficients are positive and significant for both

10
It is not clear why a household with a married household head has lower per capita consumption.
Because this dummy variable is negatively correlated with the heads age (with the correlation coefficient
-0.35) and positively correlated with household size (with the correlation coefficient 0.27), the sample of
married household heads is likely to include relatively young couples dependent on low income of the
husband.

41
educational attainments higher than primary schooling and land areas for all the different
categories. Square of land in each category is negative and significant, which suggests that the
effect of land on expenditure is non-linear.
Second, a few differences in the regression results across different ethnic minority
groups may be noted. For example, the coefficient of the share of female members is negative
and significant for the Khmer, the Muong, and the Nung, but positive and significant for the
Gietrieng, and not significant for others. Education is not a significant determinant for the
Khmer and some other minority groups.
11

Table 4 contains the results obtained from the data for 2004, with commune effects.
Note that these tables are not strictly comparable as the data for 2002 and 2004 are repeated
cross-sectional data, rather than panel data, and only a part (or about 16%) of the households
in the 2002 data was resurveyed in 2004. Short and selective comments are given below. Age
of the household head is not significant in most of the cases with a few exceptions in which
the coefficient is positive and significant (e.g. the Coho and the Sandiu in Table 4). The
coefficient of the dummy variable whether a household head is married has a significant and
positive (expected) sign for the ethnic majorities, but it is not significant for the ethnic
minorities. For 2004, the total land area and its square are used. As expected, the former is
positive and the latter negative. This implies that per capita household consumption increases
as does land area, but there is a non-linear effect, that is, the positive marginal effect of land
gets smaller as land area increases.
12

13

(Table 4 to be inserted)
The poverty regression results are given in Tables 5 and 6. Table 5 shows the results for
2002 of the conditional fixed-effects logistic model with commune fixed effects. Table 6,
based on the data for 2004, corresponds to Table 6 for 2002y. For each case, three sets of
poverty cut-off points, 100%, 80% and 120% of the poverty line, are used to test the
sensitivity of the results. Because we focus on poverty, rather than consumption, the signs of
most of the coefficient estimates are simply the opposite of those in the expenditure function.
A summary of the key findings is given below.
(Tables 5 and 6 to be inserted)
We note from Table 5 that significant determinants of poverty are: (i) age of household
head (negative, only for ethnic majorities and not for minorities), (ii) dependency burden (both
for majorities and minorities), (iii) whether a household head is married (negative (for 80% of
the poverty line) for majorities; positive for minorities), (iv) most of the education variables
(negative), and (v) most of the categories of level of land (negative).
Table 6 reports the results of poverty regressions for 2004. Most of the results are
expected in light of the earlier results for expenditure. The coefficients of dependency burden
are positive and significant; those for a married household head are positive and significant

11
We also tried the specification without commune effects, but with regional and religion dummies, using
the data for 2002. Disaggregation by each ethnic group is not carried out as the regional dummies cannot be
included in smaller samples. The results are not much different from those given in Table 3. As expected,
the coefficient estimates for the Buddhists and for those inhabiting Inland Delta or Coastal Area are positive
and significant. The coefficients of the dummies for high and low mountains are negative and significant.
These results are also used for the decomposition analysis.
12
Similar results are obtained for the 2002 data if the sub-categories of land are aggregated. We use
disaggregated data on land for 2002 in order to facilitate comparison of our results with those reported by
van de Walle and Gunewardena (2001).

13
In the cases without commune effects and with regional and religion dummies, we observe positive and
significant coefficients for the Buddhists and Catholics, and for those inhabiting Coastal Area. These results
are used for the decompositions.

42
only for the majority; those for higher levels of education are negative and significant; those of
land are negative and significant, while those of its square are positive and significant; and
those of living in mountain areas are positive and significant.

4.2. Decomposition Analysis
Poverty Decomposition
Since the predictions of conditional fixed-effects logistic regression are much lower than
the actual poverty, which would make the decomposition analysis difficult, we use the probit
model with religion and regional dummies as the base regression. In 2002, the share of the
structural component is much larger than the characteristic component, ranging from 72% to
82%. While the structural component includes not only the returns to household characteristics
but also the returns to (or the effects of) being located in a particular region, our analysis
suggests that the difference in poverty between the ethnic majority and ethnic minority groups
is largely structural. That does not necessarily rule out targeted interventions (e.g., to
concentrate resources for education on particular ethnic minorities). Rather, our results suggest
that, for example, efforts to increase returns to education by improving its quality would be
potentially more important.
The second panel of Table 7 for 2004 reinforces this view. The results suggest that more
than 90% of the poverty difference is structural.
14

(Table 7 to be inserted)
Expenditure Decomposition
Generally consistent with van de Walle and Gunewardena (2001), structural factors are
dominant when commune fixed effects are taken into account. The first and the last panels in
Table 8 show that 95%-97% of the difference of log per capita expenditure is associated with
the structural component.
(Table 8 to be inserted)
The case without commune fixed effects, based on the data for 2002, shows that the
characteristic component increases to 33%, while the structural component reduces to 67%,
consistent with van de Walle and Gunewardena (2001)
15
. However, the structural component is
considerably more dominant (96%) in 2004, without commune effects.
It is shown that the decomposition by each explanatory variable varies considerably
depending on the year or the specification used. However, if we focus only on the
decomposition of the structural factors dominant in all cases- with commune fixed effects,
household characteristics are the most important, followed by land, in both 2002 and 2004.
Without commune fixed effects, however, education is the most important in the structural
component, in both 2002 and 2004.
16



14
We also carried out the decomposition of the above results by each explanatory variable. The results vary
according to the poverty line chosen. If, for example, we choose the national poverty line, the structural
factors are mainly associated with location, religion, other household characteristics and education (only in
2002). The results will be furnished on request.
15
They report that, using the data for 1992/3, about one- half is explained by the characteristic component
and another half by the structural component, if commune fixed effects are ignored.
16
The results will be furnished on request.

43
5. Conclusion
The present study examined why ethnic minorities are poorer than ethnic majorities in
Vietnam, using the VHLSS data for 2002 and 2004. Some important findings are summarised
below from a broad policy perspective.
First, our analysis confirms that households belonging to ethnic minority groups are not
only poorer but also more vulnerable to various shocks than those in the majority groups,
namely the Kinh and the Chinese.
Second, household composition (e.g. dependency burden), education, land holdings, and
location are important determinants of expenditure and poverty, whilst there is some diversity
among different ethnic groups.
Finally, the decompositions reveal that ethnic minorities are poorer not necessarily
because they are more disadvantaged in terms of household characteristics (e.g. educational
attainment or location), but, more importantly, because the returns are much lower for them
than for majority groups. So it is imperative that government policies concentrate on not just
enhancing the endowments of the minorities but also seek to reduce the disparities in the
returns to such endowments between them and the majority. Policy priorites from this
perspective are to improve the quality of education that minorities receive, and build better
infrastructure in remote areas to promote easier market access and better livelihood prospects
for them.

References
ADB (Asian Development Bank) (2002). Indigenous Peoples / Ethnic Minorities and Poverty
Reduction - Viet Nam. Manila: ADB.
Amemiya, T. (1981). Qualitative Response Models: A Survey. Journal of Economic Literature,
19, 1483-536.
Blinder, A. S. (1973). Wage discrimination: reduced form and structural estimates. Journal of
Human Resources, 8, 436-455.
Borooah, V. K. (2005). Caste, Inequality and Poverty in India. Review of Development
Economics, 9, 399-414.
Chaudhuri, S., Jalan, J., Suryahadi, A. (2002). Assessing Household Vulnerability to Poverty
from Cross-Sectional Data: A Methodology and Estimates from Indonesia. Discussion
Paper, Columbia University.
Gaiha, R. (1988). On Measuring the Risk of Poverty in Rural India. in T. N. Srinivasan and P.
Bardhan (Eds.) Rural Poverty in South Asia, New York: Columbia University Press.
Imai, K., Gaiha, R., Kang, W. (2007). Poverty Dynamics and Vulnerability in Vietnam.
Economics Discussion Paper, 0708-2007, University of Manchester.
Kijima, Y. (2006). Caste and Tribe Inequality: Evidence from India, 1983-1999. Economic
Development and Cultural Change, 54, 369404.
Fforde, A. (1998). Vulnerable Groups in Rural Vietnam: Situation and Policy Response. A
Report based upon the Sample survey. A Draft, Retrieved June 15, 2008 from
http://www.aduki.com.au/Vulnerable%20Groups.pdf.
Fforde, A. (2005). Persuasion: Reflections on economics, data, and the 'homogeneity
assumption. Journal of Economic Methodology, 12(1), 63-91.

44
Gaiha, R., Thapa, G. (2006). Growth, Equity and Poverty Reduction in VietnamProspects
and Challenges, Draft, Asia and Pacific Division, the IFAD, Rome, December 2006.
Gaiha, R., Thapa, G., Imai, K., Kulkarni, V. S. (2007). Endowments, Discrimination and
Poverty in Rural India. Economics Discussion Paper, University of Manchester (0722,
2007).
Gang, I. N., Sen, K., Yun, M. (2008). Poverty in Rural India: Caste and Tribe. Review of
Income and Wealth, 54(2) 149-165.

McElwee, P. (2006). Blood Relatives or Uneasy Neighbors?: Kinh - Ethnic Minority
Interactions in the Annamite Mountains. A Draft Presented at the Conference Vietnam
Update 2006 at Research School of Pacific and Asian Studies, the Australian National
University.
Oaxaca, R. L. (1973). Male-female wage differentials in urban labor markets. International
Economic Review, 9, 693-709.
Scott, S., Chuyen, T. T. K. (2004). Behind the numbers: Social mobility, regional disparities
and new trajectories of development in rural Vietnam. in P. Taylor (Ed.), Social
Inequality in Vietnam: Challenges to Reform, Singapore: Institute of Southeast Asian
Studies (ISEAS), pp. 90-122.
van de Walle, D., Gunewardena, D. (2001). Sources of ethnic inequality in Viet Nam. Journal
of Development Economics, 65, 177-207.
World Bank (2004). Vietnam Development Report 2004: Poverty, Hanoi, World Bank.
Yun, M. (2004). Decomposing differences in the first moment. Economics Letters, 2004, 82
(2), 275-280.


45
Table 1: Poverty and Vulnerability by Ethnic Group in Viet Nam in 2002
Total Majority Kinh Chinese Minority Tay Thai Khmer Muong Nung
Real Consumption per capita 3480 3727 3699 5260 1939 2180 2180 2750 1735 2200
Poverty Headcount 80 0.155 0.106 0.113 0.071 0.461 0.389 0.437 0.135 0.542 0.330
(based on 80% of the povertyline)
Poverty Headcount 100 0.281 0.223 0.232 0.146 0.643 0.586 0.622 0.281 0.711 0.500
(based on 100% of the poverty line)
Poverty Headcount 120 0.406 0.349 0.358 0.212 0.761 0.710 0.751 0.496 0.807 0.683
(based on 120% of the poverty line)
Vulnerability Measure 80 0.057 0.010 0.005 0.033 0.351 0.301 0.449 0.027 0.490 0.193
(based on 80% of the poverty line)
Vulnerability Measure 100 0.149 0.073 0.068 0.123 0.621 0.637 0.745 0.160 0.809 0.517
(based on 100% of the poverty line)
Vulnerability Measure 120 0.339 0.268 0.266 0.250 0.782 0.799 0.848 0.389 0.927 0.756
(based on 120% of the poverty line)
No. of Observations 29530 24560 25108 212 4074 1045 1862 260 450 450
Dao Ngai Ede Coho Bana Sandiu Sedang Stieng Gietrieg Stingmun Other
Real Consumption per capita 1815 1348 1618 1580 1074 2861 1526 1513 1644 1637 2686
Poverty Headcount 80 0.478 0.730 0.486 0.556 0.827 0.241 0.519 0.571 0.414 0.286 0.305
(based on 80% of the poverty line)
Poverty Headcount 100 0.812 0.825 0.688 0.750 0.952 0.379 0.740 0.738 0.621 0.750 0.446
(based on 100% of the poverty line)
Poverty Headcount 120 0.862 0.905 0.817 0.815 0.971 0.471 0.818 0.857 0.862 0.929 0.567
(based on 120% of the poverty line)
Vulnerability Measure 80 0.440 0.479 0.099 0.476 0.712 0.163 0.641 0.085 0.712 0.819 0.436
(based on 80% of the poverty line)
Vulnerability Measure 100 0.753 0.788 0.401 0.773 0.932 0.419 0.842 0.473 0.990 0.998 0.731
(based on 100% of the poverty line)
Vulnerability Measure 120 0.911 0.884 0.693 0.830 0.973 0.648 0.924 0.817 1.000 1.000 0.874
(based on 120% of the poverty line)
No. of Observations 138 126 109 108 104 87 77 42 29 28 1640
*Poverty headcount ratios and vulnerability measures based on the standard (or 100% of) poverty line are shown in bold.

46
Table 2: Poverty and Vulnerability by Ethnic Group in Viet Nam in 2004
Total Majority Kinh Chinese Minority Tay
Real Consumption per capita 4824 5169 5160 6274 2799 3047
Poverty Headcount 80 0.146 0.093 0.092 0.158 0.463 0.336
(based on 80% of the poverty line)
Poverty Headcount 100 0.265 0.202 0.202 0.211 0.643 0.580
(based on 100% of the poverty line)
Poverty Headcount 120 0.397 0.337 0.337 0.263 0.759 0.706
(based on 120% of the poverty line)
Vulnerability Measure 80 0.049 0.006 0.006 0.000 0.311 0.271
(based on 80% of the poverty line)
Vulnerability Measure 100 0.144 0.066 0.066 0.111 0.622 0.667
(based on 100% of the poverty line)
Vulnerability Measure 120 0.332 0.261 0.260 0.370 0.768 0.829
(based on 120% of the poverty line)
No. of Observations 9188 7847 7787 63 1341 291

Dao Jrai Ede Coho Bana Sandiu Sedang Stieng
Real Consumption per capita 2389 1823 2351 2480 1779 2841 2490 3257
Poverty Headcount 80 0.538 0.842 0.625 0.400 1.000 0.200 0.182 1.000
(based on 80% of the poverty line)
Poverty Headcount 100 0.846 0.895 0.750 0.800 1.000 0.533 0.545 1.000
(based on 100% of the poverty line)
Poverty Headcount 120 0.923 0.947 0.750 0.800 1.000 0.600 0.727 1.000
(based on 120% of the poverty line)
Vulnerability Measure 80 0.247 0.415 0.017 0.434 0.676 0.421 0.350 0.250
(based on 80% of the poverty line)
Vulnerability Measure 100 0.508 0.604 0.624 0.600 1.000 0.667 0.577 0.607
(based on 100% of the poverty line)
Vulnerability Measure 120 0.845 0.749 0.688 0.600 1.000 0.799 0.818 0.875
(based on 120% of the poverty line)
No. of Observations 55 34 24 17 26 18 28 14


47
Table 3: Determinants of log of per capita Consumption by Ethnic Group in 2002 (with Commune Fixed
Total Majority Kinh Chinese Minority Tay Thai
Coef. Coef. Coef. Coef. Coef. Coef. Coef.
(t value) (t value) (t value) (t value) (t value) (t value) (t value
Age of Household head 0.001 0.001 0.001 -0.005 0.00 0.002 -0.003
(4.07)** (3.68)** (3.46)** (1.85) (0.05) (2.42)* (2.36)*
The Share of Female Members -0.022 -0.025 -0.029 0.059 -0.024 -0.115 -0.133
(1.85) (1.95) (2.20)* (0.35) (0.75) (1.78) (1.39)
Dependency Burden -0.319 -0.32 -0.31 -0.317 -0.325 -0.274 -0.429
(31.21)** (29.18)** (28.04)** (1.81) (11.45)** (4.75)** (4.75)**
Whether a head is married -0.024 -0.017 -0.02 -0.145 -0.066 -0.036 -0.069
(3.43)** (2.37)* (2.64)** (1.52) (3.48)** (0.90) (0.95)
Primary School -0.019 -0.047 -0.035 -0.039 0.035 0.02 0.125
(Max. attainment of members) (2.03)* (4.45)** (3.33)** (0.36) (2.15)* (0.43) (2.74)**
Lower Secondary School 0.07 0.03 0.05 (0.03) 0.13 0.11 0.21
(Max. attainment of members) (7.32)** (3.04)** (4.53)** (0.28) (6.73)** (2.37)* (4.04)**
Upper Secondary School 0.203 0.17 0.189 0.125 0.233 0.194 0.298
(Max. attainment of members) (19.15)** (14.36)** (15.93)** (1.03) (9.03)** (3.62)** (3.31)**
Technical School 0.344 0.312 0.337 0.454 0.354 0.332 0.306
(Max. attainment of members) (27.68)** (22.57)** (24.33)** (1.58) (11.83)** (5.71)** (4.10)**
Higher Education 0.51 0.473 0.494 0.723 0.576 0.618 0.511
(Max. attainment of members) (38.09)** (32.45)** (33.78)** (3.37)** (12.99)** (7.82)** (5.18)**
Agricultural Land 6.092 6.886 6.672 -17.111 8.166 2.647 1.303
(16.18)** (15.91)** (15.58)** (1.01) (7.76)** (0.44) (0.40)
[Agricultural land]
2
-6.471 -7.136 -6.909 589.267 -45.069 39.365 -7.995
(10.16)** (10.49)** (10.19)** (1.46) (6.85)** (0.27) (0.43)
Syvilcultural land 2.404 6.362 6.269 -271.415 1.915 0.664 -0.309
(4.31)** (4.13)** (4.27)** (1.87) (3.30)** (0.53) (0.05)
[Syvilcultural land]
2
-5.679 -46.526 -40.708 16,690.70 -4.099 -2.357 81.68
(2.81)** (3.03)** (3.03)** (1.60) (2.16)* (0.25) (0.85)
Aquacultural water 8.774 8.925 10.873 -55.323 88.384 58.511 309.895
(6.61)** (6.24)** (5.08)** (1.04) (3.77)** (1.34) (2.47)*
[Aquacultural water]2 -18.884 -19.146 -62.156 3,308.70 -3,778.56 -2,391.10 -23,362.43

48
(5.99)** (5.76)** (1.82) (1.25) (3.47)** (1.30) (0.30)
Unused land -1.088 0.525 -0.635 744.887 -20.075 -28.895 -28.317
(0.30) (0.11) (0.13) (1.56) (1.76) (1.03) (0.68)
[Unused land]2 19.982 -6.714 3.99 0 1,061.05 2,174.44 1,266.43
(0.37) (0.11) (0.07) (.) (1.84) (1.34) (0.54)
Constant 7.915 8.015 7.993 8.649 7.476 7.456 7.651
(472.93) (425.29) (423.77) (34.71) (199.78) (89.66) (64.05)
Observations 29530 25456 25108 212 4074 1045 510
R-squared 0.17 0.17 0.17 0.3 0.16 0.16 0.22
Joint Significance F Test
F(17,26621)
=313.69**

F(17,22780)
=271.11**
F(17,22623)
=271.50**
F(16,101)
=2.69**

F(17,3452)
=39.57**
F(17,933)
=10.69**
F(17,4
=7.23**
Prob. >F 0.0000 0.0000 0.0000 0.0014 0.0000 0.0000 0.0000
Number of Communes 2901 2659 2468 95 605 95 50
* significant at 5%; ** significant at 1%, Significant coefficients are shown in bold.


49
Table 3: Determinants of log of per capita Consumption by Ethnic Group in 2002 (with Commune Fixed Effects
Dao Ngai Ede Coho Bana Sandiu Sedang Stieng
Coef. Coef. Coef. Coef. Coef. Coef. Coef.
(t value) (t value) (t value) (t value) (t value) (t value) (t value) (
Age of Household head -0.001 0.001 0.001 0.002 0.002 -0.001 -0.001
(0.35) (0.54) (0.46) (0.68) (0.74) (0.33) (0.90)
The Share of Female Members 0.107 -0.001 0.007 0.433 -0.067 0.055 -0.186 -
(0.74) (0.01) (0.04) (1.99)* (0.40) (0.28) (1.24)
Dependency Burden -0.423 -0.366 -0.585 -0.291 -0.538 -0.541 -0.307
(3.73)** (2.42)* (3.32)** (1.61) (3.01)** (2.35)* (2.53)*
Whether a head is married 0.071 -0.02 0.166 -0.199 -0.044 -0.023 -0.245
(0.81) (0.24) (1.35) (1.68) (0.51) (0.16) (3.19)**
Primary School 0.127 -0.01 0.036 0.052 0.09 0.446 0.057 -
(Max. attainment of members) (2.42)* (0.19) (0.42) (0.50) (1.34) (2.74)** (1.06)
Lower Secondary School 0.19 0.10 0.07 0.10 0.32 0.33 0.13
(Max. attainment of members) (1.79) (1.10) (0.63) (0.81) (1.08) (2.10)* (1.11)
Upper Secondary School 0.42 0.018 0.025 0.141 0 0.543 0 -
(Max. attainment of members) (2.28)* (0.06) (0.16) (0.87) (.) (2.90)** (.) (3.37)**
Technical School 0.531 0.216 -0.327 0 -0.212 0.551 0
(Max. attainment of members) (2.15)* (1.40) (1.27) (.) (0.67) (2.91)** (.)
Higher Education 0 0.418 0 0.203 0 0.828 0
(Max. attainment of members) (.) (1.52) (.) (0.71) (.) (2.81)** (.)
Agricultural Land 8.574 13.729 21.071 18.149 51.136 -104.09 -8.258 12.113
(1.53) (2.06)* (4.20)** (1.14) (2.49)* (1.55) (0.62)
[Agricultural land]
2
-85.664 -107.424 -103.219 -231.594 -1,299.61 12,825.69 268.822 -
(1.54) (0.96) (2.61)* (0.51) (1.97) (2.29)* (0.69)
Syvilcultural land 3.182 -626.42 319.085 0 19.209 55.313 -49.263
(2.92)** (1.31) (1.45) (.) (0.61) (1.77) (1.16)
[Syvilcultural land]
2
-6.139 408,245.63 -45,164.58 0 -641.163 -1,993.34 705.019
(2.69)** (1.49) (1.11) (.) (0.35) (1.96) (1.14)
Aquacultural water -438.671 0 -207.3 0 0 -3,493.73 4,450.48

50
(0.59) (.) (0.14) (.) (.) (0.94) (1.34)
[Aquacultural water]2 508,590.80 0 981,358.63 0 0 17680441.72 -1.39E+07
(0.44) (.) (0.36) (.) (.) (1.47) (0.73)
Unused land -31.106 -30.981 0 3.28 0 -94,570.38 154.63
(1.32) (2.08)* (.) (0.20) (.) (1.99) (1.67)
[Unused land]2 2,493.47 1,660.39 0 0 0 1.18E+08 -17,836.46
(1.72) (2.47)* (.) (.) (.) (1.98) (1.94)
Constant 7.242 7.152 7.12 7.098 6.786 7.173 2,855.57
(45.87) (35.34) (34.51) (25.79) (33.21) (17.54) (0.73) (21.80)
Observations 138 126 109 108 104 87 77
R-squared 0.37 0.33 0.39 0.19 0.22 0.57 0.37
Joint Significance F Test
F(16,88)
=3.24
F( 15,91)
=2.98**
F( 14,76)
=3.44**
F(11,85 )
=1.84
F(11,82 )
=2.07*
F(17,50 )
=3.83**
F(14,48 )
=2.00**
F(9,25 )
=2.64*
Prob. >F 0.0002 0.0007 0.0002 0.0588 0.0315 0.0001 0.0386
Number of Communes 34 20 19 12 11 20 15
* significant at 5%; ** significant at 1%, Significant coefficients are shown in bold.

51
Table 4: Determinants of log of per capita Consumption by Ethnic Group in 2004 (with Commune Fixed Effects
Total Majority Kinh Minority Tay Thai Khmer
Coef. Coef. Coef. Coef. Coef. Coef.
(t value) (t value) (t value) (t value) (t value) (t value) (t value)
Age of Household head 0
0 0 -0.002
-0.007 -0.004 -
(1.05)
(1.07) (1.04) (1.63)
(2.96)** (1.26) (
The Share of Female Members -0.012
-0.023 -0.019 -0.024
-0.143 -0.143 0.125
(0.43)
(0.71) (0.58) (0.31)
(0.97) (0.66) (
Dependency Burden -0.294
-0.299 -0.301 -0.282
-0.356 -0.021 -
(13.34)**
(12.67)** (12.68)** (4.07)**
(2.42)* (0.11) (
Whether a household head is married 0.049
0.055 0.057 0.011
-0.036 0.122 0.133
(3.16)**
(3.29)** (3.42)** (0.22)
(0.37) (0.77) (
Primary School 0.003
-0.039 -0.037 0.054
0.01 0.133 0.256
(Max. attainment of members) (0.14)
(1.49) (1.42) (1.40)
(0.08) (1.33) (
Lower Secondary School 0.068
0.019 0.017 0.15
0.119 0.252 0.268
(Max. attainment of members) (3.08)**
(0.74) (0.64) (3.41)**
(1.05) (2.27)* (
Upper Secondary School 0.202
0.151 0.151 0.27
0.301 0.266 0.046
(Max. attainment of members) (8.61)**
(5.53)** (5.50)** (4.87)**
(2.49)* (1.39) (
Higher Education 0.443
0.399 0.398 0.41
0.197 0.914 -
(Max. attainment of members) (15.26)**
(12.36)** (12.27)** (4.16)**
(1.09) (1.87) (
Land 5.609
6.187 6.26 5.235
2.282 -1.771 28.518
(7.56)**
(5.86)** (5.90)** (3.91)**
(1.08) (0.15) (
Land2 -16.909
-19.969 -20.405 -15.644
-6.57 168.357 -113.825
(5.60)**
(2.54)* (2.59)** (3.70)**
(1.01) (0.67) (
Constant 8.093
8.217 8.212 7.707
8.175 7.522 7.846
(202.10)
(177.64) (176.79) (80.98)
(39.69) (25.55) (21.93)
Observations 8642
7330 7283 1312
291 186
R-squared 0.14
0.14 0.14 0.12
0.17 0.12
Joint Significance F Test
F( 10,5630)
=91.62**

F( 10,4675 )
=78.09**
F( 10,4635 )
=77.59**

F(10, 758 )
=10.44**
F( 10,184 )
=3.77**
F(10,114 )
=1.47
F( 10,35 )
=3.41**
Prob. >F 0.0000 0.0000 0.0000

0.0000 0.0001 0.1520 0.0033
Number of Communes 2996
2645 2638 544
97 62
* significant at 5%; ** significant at 1%, Significant coefficients are shown in bold.

52
Table 4 Determinants of log of per capita consumption by Ethnic Group in 2004 (with Commune Fixed Effects
Jrai Ede Coho Bana Sandiu
Coef. Coef. Coef. Coef. Coef.
(t value) (t value) (t value) (t value) (t value)
Age of Household head 0.001 0.009 0.061 -0.004 -0.021
(0.18) (1.08) (2.78)** (0.36) (7.13)**
The Share of Female Members -0.081 1.576 3.769 0.592 -2.232
(0.18) (1.19) (1.77) (1.04) (10.53)**
Dependency Burden 0.31 1.791 9.372 -0.628 -1.308
(0.73) (1.59) (3.05)** (1.05) (8.86)**
Whether a household head is married -0.326 -0.108 -4.408 -0.178 0
(1.13) (0.09) (3.41)** (0.54) (.)
Primary School 0.08 0.423 4.241 0.159 1.049
(Max. attainment of members) (0.50) (1.10) (2.97)** (0.90) (5.13)**
Lower Secondary School 0.44 1.421 7.258 0.184 1.112
(Max. attainment of members) (1.79) (2.02) (3.07)** (0.53) (5.10)**
Upper Secondary School 0.244 -0.596 0 0.034 0
(Max. attainment of members) (0.56) (1.23) (.) (0.08) (.)
Technical School 0 0 0 0 0
(Max. attainment of members) (.) (.) (.) (.) (.)
Higher Education 0.001 0.009 0.061 -0.004 -0.021
(Max. attainment of members) (0.18) (1.08) (2.78)** (0.36) (7.13)**
Land 42.312 -12.422 -346.316 8.659 296.497
(1.56) (0.48) (2.96)** (0.14) (8.17)**
Land2 -327.122 40.534 3,923.33 -395.719 -12,452.69
(1.00) (0.20) (2.78)** (0.22) (4.88)**
Constant 6.83 5.38 3.822 7.602 8.27
(11.64) (2.43) (1.86) (8.06) (38.86)
Observations 34 24 17 26 17
R-squared 0.71 0.85 0.97 0.42 1
Joint Significance F Test
F(9 ,12 )
=3.19*
F(9 ,4 )
=2.54
F( 8, 1 )
=3.54
F( 9, 7 )
=0.56
F(7 ,1 )
=2.40
Prob. >F 0.0323 0.1913 0.3903 0.7950 0.0759
Number of Communes 13 11 8 10 9

53
Table 5: Determinants of Poverty in Total Sample, and among Ethnic Majority and Ethnic Minority
(Conditional fixed-effects logistic regression, with Commune Fixed Effects
Poverty100 Poverty80 Poverty120 Poverty100 Poverty80 Poverty120
Total Total Total Majority Majority Minority
Coef. Coef. Coef. Coef. Coef. Coef.
(z value) (z value) (z value) (z value) (z value) (z value)
Age of Household head -0.008 -0.01 -0.008 -0.008 -0.011 -0.008
(5.70)** (6.08)** (6.12)** (5.58)** (5.68)** (5.75)**
The Share of Female Members 0.109 0.221 0.111 0.085 0.18 0.136
(1.17) (1.97)* (1.29) (0.86) (1.42) (1.49)
Dependency Burden 1.726 1.717 1.57 1.703 1.634 1.552
(21.86)** (17.46)** (21.77)** (20.08)** (14.78)** (20.45)**
Whether a household head is
married
-0.008 -0.084 0.027 -0.095 -0.186 -0.013
(0.14) (1.26) (0.56) (1.65) (2.52)* (0.24)
Primary School 0.061 0.011 0.095 0.198 0.164 0.234
(Max. attainment of members) (0.96) (0.16) (1.56) (2.76)** (1.85) (3.50)**
Lower Secondary School -0.365 -0.552 -0.305 -0.172 -0.335 -0.126
(Max. attainment of members) (5.44)** (7.07)** (4.77)** (2.28)* (3.55)** (1.81)
Upper Secondary School -1.029 -1.223 -0.992 -0.849 -1.035 -0.828
(Max. attainment of members) (12.92)** (12.32)** (13.53)** (9.60)** (8.85)** (10.48)**
Technical School -1.868 -2.025 -1.791 -1.852 -1.994 -1.682
(Max. attainment of members) (16.47)** (12.80)** (18.74)** (13.68)** (9.75)** (15.79)**
Higher Education -2.872 -2.974 -2.748 -2.867 -2.953 -2.686
(Max. attainment of members) (13.79)** (9.42)** (17.97)** (11.90)** (7.80)** (15.85)**
Agricultural Land -45.5 -44.075 -37.337 -57.76 -70.893 -44.957
(11.49)** (8.85)** (11.34)** (11.13)** (8.85)** (9.71)**
[Agricultural land]
2
49.125 47.902 38.845 58.758 73.071 40.194
(7.30)** (6.63)** (6.82)** (2.60)** (3.21)** (0.58)
Syvilcultural land -21.708 -18.544 -20.104 -38.927 -34.597 -41.948
(4.14)** (3.51)** (4.04)** (3.05)** (2.23)* (3.94)**
[Syvilcultural land]
2
59.46 56.578 59.555 230.356 294.45 307.445
(2.06)* (1.97)* (2.40)* (1.56) (1.95( (2.92)**
Aquacultural water -46.492 -80.228 -31.189 -41.663 -80.05 -31.097

54
(4.15)** (2.24)* (2.15)* (3.22)** (1.98)* (1.48)
[Aquacultural water]2 290.434 373.048 150.24 283.034 372.527 237.266
(.) -0.3 (0.69) (.) (0.26) (0.42)
Unused land 5.552 36.061 50.724 -49.731 -32.31 0.759
(0.12) (0.54) (1.11) (0.8) (0.22) (0.01)
[Unused land]2 44.643 -3,011.86 -1,935.80 2,078.54 -2,808.22 127.88
(0.02) (0.78) (1.07) (0.93) (0.23) (0.06)
Observations 28806 28806 28806 24836 24836 24836
Joint Significance F Test
LR Chi
2
(17)
=2117.27**
LR Chi
2
(17)
=1418.72**
LR Chi
2
(17)
=2437.97**
LR Chi
2
(17)
=1800.40**
LR Chi
2
(17)
=1093.31**
LR Chi
2
(17)
=2158.14**
Prob. >F 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
Number of Communes 1744 1245 1990 1496 967 1770
* significant at 5%; ** significant at 1%, Significant coefficients are shown in bold.
































55
Table 6: Determinants of Poverty in Total Sample, and among Ethnic Majority and Ethnic Minority
(Conditional fixed-effects logistic regression, with Commune Fixed Effects)
Poverty100 Poverty80 Poverty120 Poverty100 Poverty80 Poverty120 Poverty100
Total Total Total Majority Majority Minority
Coef. Coef. Coef. Coef. Coef. Coef.
(t value) (t value) (t value) (t value) (t value) (t value)
Total Total Total Majority Majority Majority
Age of Household head 0.002 0.001 0.004 0.004 -0.001 0.005
(0.74) (0.23) (1.46) (1.16) (0.20) (1.50)
The Share of Female Members 0.01 -0.009 -0.064 -0.173 0.458 -0.169
(0.04) (0.03) (0.29) (0.6) (1.09) (0.69)
Dependency Burden 1.688 1.837 1.762 1.746 1.826 1.913
(8.74)** (6.52)** (10.33)** (8.05)** (5.52)** (10.26)**
Whether a household head is
married
-0.302 -0.189 -0.338
-0.354 -0.423 -0.389
(2.34)* (1.07) (2.89)** (2.46)* (2.07)* (3.07)**
Primary School 0.037 -0.009 -0.089 0.26 0.283 0.014
(Max. attainment of members) (0.23) (0.04) (0.58) (1.27) (1.00) (0.08)
Lower Secondary School -0.306 -0.552 -0.386 -0.046 -0.144 -0.225
(Max. attainment of members) (1.79) (2.57)* (2.44)* (0.21) (0.50) (1.20)
Upper Secondary School -1.002 -1.379 -1.009 -0.72 -0.975 -0.8
(Max. attainment of members) (5.17)** (5.10)** (5.87)** (3.04)** (2.79)** (3.98)**
Higher Education -1.981 -2.572 -2.937 -2.197 -36.552 -3.06
(Max. attainment of members) (4.99)** (3.67)** (7.88)** (3.78)** (0.00) (6.59)**
Land -23.197 -27.46 -30.524 0.004 -0.001 0.005
(3.36)** (3.00)** (4.93)** (1.16) (0.20) (1.50)
Land2 56.678 78.181 91.212 539.346 -302.588 287.804
(2.49)* (2.86)** (3.56)** (2.87)** (0.27) (2.13)*
Observations 2911 1598 3853 2135 985 3122 588
Observations 6473 6473 6473 5304 5304 5304
Joint Significance F Test
LR Chi
2
(10)
=285.38**
LR Chi
2
(10)
=188.86**
LR Chi
2
(10)
=444.27**
LR Chi
2
(10)
=230.87**
LR Chi
2
(10)
=130.20**
LR Chi
2
(10)
=384.02**
LR Chi
Number of Communes 983 539 1307 729 337 1072 204
- significant at 5%; ** significant at 1%, Significant coefficients are shown in bold.

56
Table 7: Poverty Decomposition

X (Means of
Explanatory
Variables)
(Coeficient
Estimates)
Predicted
Poverty
Difference

Share


2002
Based on Probit Model (with Regional Dummies)
Predicted Poverty (100%) for Majority Majority Majority 0.22
(based on coefficient estimates for Majority) 0.10 23.5%
Characteristic
Component
Predicted Poverty (100%) for Minority Minority Majority 0.32
(based on coefficient estimates for Majority) 0.32 76.5%
Structural
Component
Predicted Poverty (100%) for Minority Minority Minority 0.64
(based on coefficient estimates for Minority)

Predicted Poverty (80%) for Majority Majority Majority 0.11
(based on coefficient estimates for Majority) 0.06 17.8%
Characteristic
Component
Predicted Poverty (80%) for Minority Minority Majority 0.17
(based on coefficient estimates for Majority) 0.29 82.2%
Structural
Component
Predicted Poverty (80%) for Minority Minority Minority 0.46
(based on coefficient estimates for Minority)

Predicted Poverty (120%) for Majority Majority Majority 0.35
(based on coefficient estimates for Majority) 0.12 28.1%
Characteristic
Component
Predicted Poverty (120%) for Minority Minority Majority 0.47
(based on coefficient estimates for Majority) 0.29 71.9%
Structural
Component
Predicted Poverty (120%) for Minority Minority Minority 0.76
(based on coefficient estimates for Minority)
2004
Based on Probit Model (with Regional Dummies)
Predicted Poverty (100%) for Majority Majority Majority 0.19
(based on coefficient estimates for Majority) 0.03 8.2%
Characteristic
Component
Predicted Poverty (100%) for Minority Minority Majority 0.21
(based on coefficient estimates for Majority) 0.31 91.8%
Structural
Component
Predicted Poverty (100%) for Minority Minority Minority 0.53
(based on coefficient estimates for Minority)

Predicted Poverty (80%) for Majority Majority Majority 0.08
(based on coefficient estimates for Majority) 0.02 7.8%
Characteristic
Component
Predicted Poverty (80%) for Minority Minority Majority 0.10
(based on coefficient estimates for Majority) 0.23 92.2%
Structural
Component
Predicted Poverty (80%) for Minority Minority Minority 0.33

57
(based on coefficient estimates for Minority)

Predicted Poverty (120%) for Majority Majority Majority 0.32
(based on coefficient estimates for Majority) 0.02 6.8%
Characteristic
Component
Predicted Poverty (120%) for Minority Minority Majority 0.35
(based on coefficient estimates for Majority) 0.33 93.2%
Structural
Component
Predicted Poverty (120%) for Minority Minority Minority 0.68
(based on coefficient estimates for Minority)


58
Table 8: Expenditure Inequality Decomposition

X (Means
of
Explanatory
Variables)
(Coeficient
Estimates)
Predicted
Poverty
Difference

Share


2002
Based on OLS with Commune Fixed Effects
Predicted log per capita consumption for
Majority
Majority Majority 8.03
(based on coefficient estimates for Majority) -0.03 4.6% Characteristic Component
Predicted log per capita consumption for
Minority
Minority Majority 8.00
(based on coefficient estimates for Majority) -0.57 95.4% Structural Component
Predicted log per capita consumption for
Minority
Minority Minority 7.43
(based on coefficient estimates for Minority)
Based on OLS without Commune Fixed Effects, with Regional Dummies
Predicted log per capita consumption for
Majority
Majority Majority 8.03
(based on coefficient estimates for Majority) -0.19 32.6% Characteristic Component
Predicted log per capita consumption for
Minority
Minority Majority 7.83
(based on coefficient estimates for Majority) -0.40 67.4% Structural Component
Predicted log per capita consumption for
Minority
Minority Minority 7.43
(based on coefficient estimates for Minority)

2004
Based on OLS with Commune Fixed Effects
Predicted log per capita consumption for
Majority
Majority Majority 8.24
(based on coefficient estimates for Majority) -0.02 2.8% Characteristic Component
Predicted log per capita consumption for
Minority
Minority Majority 8.22
(based on coefficient estimates for Majority) -0.52 97.2% Structural Component
Predicted log per capita consumption for
Minority
Minority Minority 7.70
(based on coefficient estimates for Minority)
Based on OLS without Commune Fixed Effects, with Regional Dummies
Predicted log per capita consumption for
Majority
Majority Majority 8.10
(based on coefficient estimates for Majority) -0.02 3.7% Characteristic Component
Predicted log per capita consumption for
Minority
Minority Majority 8.08
(based on coefficient estimates for Majority) -0.41 96.3% Structural Component
Predicted log per capita consumption for
Minority
Minority Minority 7.68
(based on coefficient estimates for Minority)





59
Appendix
17


Vulnerability is simply defined as the probability that a household will fall into poverty
in the future.
( ) z c Pr V VEP
1 t , i it it
s =
+
(A1)
where vulnerability of household i at time t,
it
V , is the probability that the i-th households
level of consumption at time t+1 ,
1 t , i
c
+
, will be below the poverty line, z. One of the
limitations of this definition of vulnerability is that it is sensitive to the choice of z.
This can be derived by estimating the consumption function:
i i i
e X c ln + = (A2)
where
i
c is per capita expenditure (i.e. food and non-food consumption expenditure) for the
i-th household,
i
X represents a bundle of observable household characteristics and other
determinants of consumption (e.g. age of household head, dependency burden, educational
attainments of household members, ethnic group, regional dummies, access to market,
infrastructure), is a vector of coefficients of household characteristics, and
i
e is a
mean-zero disturbance term that captures idiosyncratic shocks to per capita consumption. It is
assumed that the structure of the economy is relatively stable over time and, hence, future
consumption stems solely from the uncertainty about the idiosyncratic shocks,
i
e . It is also
assumed that the variance of the disturbance term depends on:
u = o
i
2
i , e
X (A3)
The estimates of and are obtained using a three-step feasible generalized least
squares (FGLS). Using the estimates

and

, we can compute the expected log

consumption and the variance of log consumption for each household as follows.
| =

X ] X C [ln E
i i i
(A4)
u =

X ] X C [ln V
i i i
(A5)
By assuming
h
c ln as normally distributed, the estimated probability that a household
will be poor in the future (say, at time t+1) is given by:
( )
|
|
.
|

\
|
u
|
u = < =

X z ln
X z ln c ln r P

v P E

V
i
i
i i i i
(A6)

17
See Imai, Gaiha and Kang (2007) for more details.
60
Demographic Changes and the Pension Finances in
Vietnam: A Long-term Stochastic Actuarial Assessment


Wade D. Pfau
Associate Professor,
National Graduate Institute for Policy Studies (GRIPS),
7-22-1 Roppongi, Minato-ku, Tokyo 106 8677, Japan
Email: wpfau@grips.ac.jp


Giang Thanh Long
Lecturer, National Economics University (NEU),
207 Giai Phong Street, Hai Ba Trung District, Hanoi 10000, Vietnam
Email: longgt@neu.edu.vn


Abstract
This paper aims to provide a long-term financial vision for the Vietnamese
pension scheme using stochastic modeling for key variables under an actuarial
framework. In particular, we project the pension fund balances in order to see
whether the scheme will be sustainable. The median values of the status-quo
projections show that the pension fund will be depleted in about 2052 with 90-
percent confidence interval range of about 8 years. The estimated results from our
sensitivity tests show that the retirement age, the indexation method for pension
benefits, and the contribution rate are all crucial determinants of the pension fund
balance in the long term. At the same time, some factors, including population
coverage rates, administrative costs, the long-term fertility rate, and the rate of
return on pension fund assets play less important roles in determining the funds
balance.

Keywords: aging, stochastic modeling, pension scheme sustainability, Vietnam
JEL classification: H55, J11, J14

61
1. Introduction
Demographic changes have significant impacts on the social and economic performances
of countries, regions, and the whole world. Recently, the most observable demographic change
involves a rapidly aging population, meaning that the proportion of elderly people in the total
population is sharply increasing. In the developed and high-performing developing economies,
this situation results from important demographic trends: declining fertility rates, declining
mortality rates, and increasing life expectancies. Population aging requires extensive public
expenditures for the aged on pensions and health care, and therefore influences pension funds,
government budgets, and eventually the long-term fiscal sustainability of governments.
The aging population problem becomes more serious if it is associated with a pay-as-
you-go (PAYG) pension scheme. A number of empirical analyses indicate that these two
issues are potential threats to the financial stability of pension funds and government budgets
in almost all economies in the world. For OECD countries, studies on aging, such as
Hagemann and Nicoletti (1989), Dang et al. (2001), and OECD (2005), find that most of these
countries will face the problems of declining investment, saving, economic growth, and
rapidly increasing public expenditures under aging societies and PAYG pension schemes.
Similar issues are also challenging transitional and developing economies, especially in the
East-Asian region, in which a huge population is aging as a result of the economic miracle and
social improvements. Many studies, including Friedman et al. (1996), Heller (1997, 1998),
and UNESCAP (2005), also indicate that the PAYG pension schemes in these countries will
also be facing fiscal imbalances in the near future.
The medium-variant population projections by United Nations (2007a) show that
Vietnam is still a young economy with about 7.6 percent of its total population aged 60 and
over in 2005, but it will face the same issues of aging as other countries in the coming
decades. Recent demographic changes show that life expectancy at birth increased from 68 in
1995 to 71 in 2007, and fertility rates declined from 2.8 children per woman in 1995 to 1.9 in
2007 (International DatabaseIDB, 2008). The aforementioned population projections also
indicate that the elderly population will increase significantly, reaching 26.1 percent in 2050,
and the total dependency ratio will be mostly driven by the elderly dependency ratio.
Furthermore, previous studies on the current publicly-managed PAYG scheme in Vietnam
also show that the scheme will face financial instability and generational inequity if there are
no systematic policy reforms. For instance, Nguyen (2004) shows that on average the amount
contributed by a laborer over 30 years will only be able to pay for retirement benefits for 6-8
years, while the average life expectancy at retirement is about 15 years. Giang (2004, 2008a)
and Nguyen (2006) find that the pension fund will be financially depleted in about 40 years
under the current regulations.
The number of studies on the long-term status of the Vietnamese pension scheme has
increased over the past decade, but these studies have limitations due to the deterministic
estimation techniques. The common drawbacks of such techniques are presented in a number
of studies, such as Foster (1994), Bongaarts and Bulatao (2000), and the American Academy
of Actuaries (2005). This paper will mitigate the limitations of the previous studies by using
stochastic estimates of key variables under an actuarial framework. In particular, the paper will
provide estimates showing the long-term financial status for the Vietnamese pension scheme.
We will also present the results after modifying a number of parameters to show the critical
factors of the schemes financial stability.
The remainder of the paper is organized as follows. Section 2 makes an overview of the
Vietnamese pension scheme with an emphasis on key regulations for contributions and
benefits. Section 3 reviews some previous studies about long-term forecasts for social security
funds with a focus on stochastic simulation methods. The analytical framework, data, and
62
assumptions are presented in Section 4, while Section 5 discusses our findings and policy
implications. Some remarks in the last section will conclude the paper.
In general, the median values of the baseline projections show that the pension fund will
be depleted in about 2052 with a 90-percent confidence interval range of 8 years. The
estimated results from our sensitivity tests show that the retirement age, the indexation method
for pension benefits, and the contribution rate are all crucial determinants of the pension fund
balance in the long term. At the same time, some factors, including population coverage rates,
administrative costs, the long-term fertility rate, and the rate of return on pension fund assets
play less important roles in determining the funds balance.

2. The Vietnamese Pension Scheme at a Glance
The pension scheme is a component of the social insurance system, which has been in
operation since the early 1960s. Prior to 1995, the pension scheme, which provided a non-
contributory defined benefit, covered only the employees in the state sector, and it was
managed by different public agencies under the supervision of the government. In that scheme,
the benefit levels for retirement were defined by the number of working years and the
reference wage (usually, it was the wage of the last year prior to retirement). The benefits were
paid by the social insurance fund, which was funded by contributions from employers and
government subsidies. The fund was managed and guaranteed by the government. For about
thirty years, the scheme significantly contributed to the income and living standards of the
insured people.
However, dramatic changes in the economy and society with the growing private sector
resulting from the Doi moi (renovation) programs in 1986 forced the government to reform the
scheme. The reform led to the establishment of a publicly-managed pay-as-you-go defined-
benefit (PAYG DB) scheme in 1995. Vietnam Social Security (VSS) is responsible for
administration of this scheme under the guarantees of the government. Due to this reform, the
current pension scheme has two kinds of beneficiaries, i.e. pre-1995 and post-1995 pensioners.
The formers benefits are paid directly from the government budget, while the latters benefits
are entitled to and paid by the VSS fund. All current contributors are entitled with the VSS.
Though participation in the scheme is mandatory for all employees of both the state and
non-state sectors in Vietnam under some certain conditions, the actual coverage rate of the
scheme is very low. According to MOLISA (2006), the coverage rate of the scheme was only
about 15 percent of labor force in 2005 (or about 9 percent of the whole population in
Vietnam), in which 60 percent were from the state sector (which represented only 10 percent
of the labor force), and the remaining 40 percent were from the non-state sector (which
accounted for 90 percent of the labor force). The average compliance rate was only 72 percent
in 2004 (Nguyen, 2006).
For the pension and survivorship, an employee contributes 5 percent of monthly salary,
while an employer contributes 11 percent of payroll. From 2010, the contribution rates of both
employees and employers will increase by 1 percentage point every two years until they reach
8 percent and 14 percent, respectively. In 2005, the average contribution was VND 258,000
(about $US 17) per capita per month, based on the minimum wage of VND 450,000, and the
average length of service of the current contributors was about 13.5 years (MOLISA, 2006).
A retirement pension is normally paid to men and women at the age of 60 and 55,
respectively, who have at least 20 years of contributions. Pension benefit formulas vary
depending on the number of years of contributions. Up to the 15
th
year of contributions, the
pension accrual rate is 3 percent, and is then 2 percent for males and 3 percent for females for
each additional year. Conversely, the pension accrual rate will be reduced by 1 percent for
63
each year of early retirement. If contributions have been paid for in excess of 30 years for men
and 25 years for women, the retiree can receive an additional lump sum payment of half a
months average salary for each year of the excess. The maximum replacement rate is 75
percent. If the total number of contribution years is less than 20, retirees can only receive a
lump sum payment of 1.5 times the average monthly salary for each year of contributions.
There is no floor or ceiling for pensionable earnings. The reference earnings for estimating
pension benefits is calculated as the average wage of the last 10 years prior to retirement for
contributors who have been lifetime employees of the state sector, or the average wage of the
whole working and contributing time for other contributors. Recently, the average replacement
rate was 69 percent, and the average retirement age for males and females was 57 and 52,
respectively (VSS, 2004). The dependency ratio of the scheme was about 5.8 percent in 2005,
meaning that 1 pensioner was covered by 17 contributors.
Most of the accumulated pension fund assets have been lent to or invested in
government-guaranteed sources, including state-owned commercial banks, government budget,
treasury bonds, or the Development Assistance Fund (DAF). These assets usually earn lower
average rates of return than the average market rates for bank deposit (Nguyen, 2006; Giang,
2008b).

3. A Review of Relevant Studies
Since the early 1990s, instead of using traditional deterministic (or scenario-based)
modeling, researchers have been interested in using stochastic modeling to evaluate the long-
term financial capacity of social security schemes, particularly in OECD countries. Ronald
Lee, Shirad Tuljapurkar and others have produced a series of studies to discuss the impacts of
demographic changes on the long-term finances of the U.S. social security system. In Lee and
Tujapurkar (1998a), they make a population forecast for the period 1994-2070 by creating
stochastic forecasts for mortality and fertility, and using a deterministic forecast for net
immigration. For the economic variables, they obtain most of the assumptions from the Social
Security Administration (SSA) deterministic forecasts, but they do make stochastic forecasts
for the productivity growth rates and interest rates, which follow independent AR(1) processes.
In another paper (Lee and Tuljapurkar, 1998b), they also use stochastic forecasts for these
variables, but the productivity growth rate follows an ARMA constrained mean (ARMA-CM)
model, while the interest rate follows an AR(1). In these two papers, the authors also examine
how much these variables contribute to the uncertainty of financial forecasts by alternatively
allowing these variables to be deterministic or stochastic.
Also, to see only the impacts of demographic changes on the U.S. Old Age, Survivor,
and Disability Insurance (OASDI) Trust Funds, Tuljapukar and Lee (2000) make stochastic
population forecasts with stochastic mortality and fertility, and deterministic net immigration,
while taking values of economic variables, such as the inflation rate, real interest rate,
productivity growth rate, and labor force participation rate from the Trustees 1995 report.
They find substantial differences in possible future trajectories of the fund, and then suggest
using stochastic forecasts with supplementary methods to examine uncertainty. In a similar
work for stochastic projections of the Social Security Trust Fund, Lee et al. (2003) also make
stochastic forecasts for real wage growth, the interest rate, and equity returns using a vector
autoregression. They then examine several plans for achieving long-term solvency by raising
the normal retirement age, increasing payroll taxes, and investing some portion of the fund in
the stock market. Further examining the impacts of net immigration in the social security fund,
Lee et al. (2004) make stochastic forecasts for this variable, and the results show that
stochastic net immigration makes only little difference to the probability distribution of the
old-age dependency ratio, and thus a small difference in trust fund projections.
64
In developing their own long-term actuarial model to evaluate the Social Security
Administration (SSA)s 75-year projections of the Trust Funds, the Congressional Budget
Office (2001) makes stochastic forecasts for demographic variables (including mortality,
fertility, and immigration) and economic variables (including real wage growth,
unemployment, inflation, real interest rate, disability incidence rate, and disability termination
rate). For instance, they estimate immigration and fertility with an ARMA(4,1) model, and real
wage growth and mortality improvements with AR(1) processes. To examine the simulated
results, they use both Monte Carlo techniques and bootstrap simulations in order to generate a
probability distribution for future outcomes.
Among a handful of papers discussing the long-term pension fund balance with
stochastic methods in other OECD countries, Fehr and Habermann (2004) discuss the
sustainability of the German pension system under demographic uncertainty during
20012050. The paper makes stochastic forecasts for fertility and mortality, but uses
deterministic net immigration, to obtain stochastic population forecasts, while economic
variables are estimated by using an overlapping generation model with three sectors, i.e.
household, production, and government. They find that demographic changes have significant
impacts on the benefit levels in the long-term. The paper then looks for the influences of
changes in contribution rates and replacements rates on the financial balance of the system, in
which generational equity is also in focus. Brlum (2004) uses the same methodology, but
takes net immigration to be stochastic, and finds also that the demographic uncertainty will
have substantial influences on the long-term benefit levels of different participating cohorts in
the Danish PAYG DB pension system during the 20002050 simulation period.
The paper by Kitamura et al. (2005) uses the Asset-Liability Model (ALM) with
stochastic simulation to analyze the Japanese Public Employees Pension Scheme. Among
various variables in the model, they make stochastic forecasts for the growth rate of prices,
growth rate of wages, and rate of investment return of the pension reserve fund in order to
evaluate the appropriateness of an indexation rule, which affects future benefit levels and
financial conditions of the scheme.

4. Analytical Framework, Data, and Assumptions
4.1. Projections for Pension Fund
Figure 1. Projection Methodology for the Pension Fund Balance











Source: Own modifications using International Labor Office ILO (1998).
Number of active
contributors
Investment Returns
and other incomes
Contribution
rates
Total contribution
Number of pensioners
(retirement,
survivorship)
Average
benefits
Total benefit payments Other costs
(administration, etc)
Total revenues (+)
Total outlays (-)
Fund balance
Demographic factors, economic factors, and pension scheme factors
65
To assess the long-term financial capacity of the pension fund, we need to make
projections for the related indicators of the scheme, including demographic factors (such as
population and active labor force), macroeconomic factors (such as inflation and wage growth),
and pension scheme indicators (such as active contributors and pensioners). The general
projection flows are illustrated in Figure 1.
In this paper, we will apply stochastic estimates for some key variables in order to
provide estimates for the pension fund balances in the period 2005-2105.
4.2. Projections for Population and Labor Market
Figure 2. Projection Methodology for Population and Active Labor Force











Note: S: stochastic; D: deterministic.
Source: Own modifications using ILO (1998).
Our projections for population and the labor market are illustrated in Figure 2. The
development of population projections depends on three major factors: fertility rates, mortality
rates, and net immigration rates. The population projections are based on an iterative
procedure with an initial year. To project the population in subsequent years, we combine the
population of the previous year with the projected rates of mortality, fertility, and net
immigration for each age. Population projections can be expressed as follows:
t t t t t
POP MIG POP L POP + =
+1
, (1)
where POP is an n x 1 vector of gender-specific population totals calculated by age for n ages;
MIG is an n x 1 vector of net immigration rates by age; t represents the projection year; and L
is the n x n Leslie Matrix:

=
t n
t
t
t x
t
m
m
m
f
L
,
, 1
, 0
,
1 0 ... ... 0 0
... ... ... ... ... ...
0 ... 0 0 1 0
0 ... ... 0 0 1
0 ... ... 0 0
, (2)
where f(x,t) and m(x,t) respectively represent the gender-specific fertility rates and central
death rates for people age x during the year t.
Initial general population by age & gender
Projected
general
population by
age & gender
Initial labor force participation rates
by age & gender
Projected
working age
population by
age & gender
Historical
mortality rates
S
S
D
Historical
fertility rates
Projected
fertility rates
Projected
mortality rates
Historical net
immigration rates
Projected net
immigration rates
Projected labor force participation rates
by age & gender
P
r
o
j
e
c
t
e
d

a
c
t
i
v
e

l
a
b
o
r

f
o
r
c
e

D
66
Figure 3. Distribution of Vietnamese Population, 2005
1200000 900000 600000 300000 0 300000 600000 900000 1200000
0
5
10
15
20
25
30
35
40
45
50
55
60
65
70
75
80
85
90
95
100+
Male Female

Source: Own calculations using IDB (2008) and VHLSS 2004.
Our base-year for population projections is 2005, and the data are obtained from IDB
(2008). The data are available by gender and five-year age groups between 0 and 99, and for
the sum of individuals aged 100 and over. For projections, we expand the population data in
2005 from 5-year age groups to 1-year age groups using the population pattern by age and
gender from the Vietnam Household Living Standards Survey (VHLSS) in 2004. This survey
provides representative population data for all regions and areas of Vietnam. Figure 3 presents
the population pyramid by age and gender for Vietnam in 2005.
In the case of migration, as the data for net international migration by age and gender in
Vietnam are rare and not available to date, we use the Japanese populations net immigration
structure as proxy for Vietnam. The data show that net immigration has not had significant
impacts on the size of Japans population, which also has been the case for Vietnam since the
1980s to date. The data are obtained from the Statistics Bureau of Japans Ministry of Public
Management, Home Affairs, Posts and Telecommunications (Okita et al., 2008).
Both mortality and fertility rates are projected using the Lee-Carter method, which has
been described in a number of studies (see, for instance, Lee and Carter, 1992; Lee, 1993,
2000; Lee and Tuljapukar, 1998; and Lee and Miller, 2001). The method for mortality uses the
historical age-sex breakdown of central death rates by single year of age, and uses the notion
that mortality can be decomposed into age-specific and time-specific components. In general,
the model for mortality forecasts can be presented as follows:
log[m(x,t)] =a(x)+k(t)b(x)+(x,t), (3)
where m(x,t) is the central death rate of a person aged x in the year t; a(x) represents time-
invariant general age pattern across the mortality schedule; b(x) reflects the time-invariant
relative speed that various ages respond to the time-varying trend in mortality, which is
presented by k(t); and (x,t) is the error in this approximation to the actual age schedule.
67
In estimating these models, standard regression techniques cannot be used, because none
of these terms are observable. Instead, a singular value decomposition (SVD) technique
1
is
used, and it leads to setting a(x) equal to the time average of log[m(x,t)] over the historical
data, and using the matrix of deviations, log[m(x,t)]-a(x), in the SVD routine. The estimates
are derived from constraints that the sum of k(t) terms over t equals 0 and the sum of b(x)
terms over x equals 1. In other words, SVD can help us to transform the tasks of forecasting
log[m(x,t)] into forecasting k(t) and the error (x,t).
Similarly, the model for estimating fertility can also be expressed as follows:
log[f(x,t)]=c(x)+l(t)d(x)+e(x,t), (4)
where f(x,t) is fertility of women aged x in the year t; c(x) is an addictive age-specific constant,
which reflects the general age pattern of fertility; l(t) is a period-specific index of the general
level of fertility; d(x) reflects the responsiveness of fertility at age x to the variations in the
general level; and e(x,t) represents any remaining residual of this approximation.
The key to forecast these age-sex variables is to assume that the age-specific components
of the model will remain constant in the future. After fitting an appropriate model, 1000 Monte
Carlo simulations are produced using the specified model, in order to obtain a probability
distribution over mortality and fertility.
In this paper, we use yearly data of mortality rates for the Vietnamese population by
five-year age groups, in which data for the year 1999 are obtained from Lopez et al. (2000),
while those for the period 2000-2005 are obtained from WHO (2008). The former provides
mortality rates for all ages between 0 and 84, and the sum of individuals aged 85 and over,
while the latter provides mortality rates for all ages between 0 and 99, and the sum of
individuals aged 100 and over. We use the Coale and Guo (1989) method to extend data for
the year 1999 to get consistent data with those of WHO. Then, for the sake of simplicity, we
assume that all individual ages have the same mortality rates as the average rate of their
respective age groups, in order to get historical mortality rates by age.
For fertility projections, we also use annual data for the period 1990-2005, which are
obtained from the IDB. This dataset provides historical fertility rates of all the females aged
15-49 by five-year age groups. Also for the sake of simplicity, we assume that all individual
ages have the same fertility rates as the average rate of their respective age groups, in order to
get historical fertility rates by age. The data show that the total fertility rate (TFR) in Vietnam
has experienced a decreasing trend over the past two decades, from 3.65 in 1990 to 1.94 in
2005. We assume that the long-term total fertility rate will fluctuate around 1.85, which is the
same as the assumed rate for the medium-variant projections of the United Nations (2007a).
Also, to estimate the number of newly-born children by gender, we assume that the average
historical birth rate, which was 1.06 boys per girl during 1990-2005, will remain for the whole
projection period.
With the projected total population described above, we can get the projected working-
age population, i.e. the 15 to 59 year-old population. We assume that the total labor force
participation rates by age and gender in 2005 will remain the same for the whole projection
period. The projected rates of labor force participation and the projected working-age
population will produce the projected active laborers by age and gender, which in turn are
important for estimating the number of contributors for the pension scheme. Labor force
participation rates are available from yearly surveys on labor and employment during 1996-
2005 by the Ministry of Labor, War Invalids, and Social Affairs, Vietnam (MOLISA).

1
The SVD theorem states that any n x p matrix (namely, matrix A) can be factored as A
nxp
=U
nxn
S
nxp
V
T
pxp

where U
nxn
is an n x n orthogonal matrix, and V
pxp
is a p x p orthogonal matrix, and S
nxp
is an n x p matrix
with singular values of A on the main diagonal, and zeros elsewhere.
68
4.3. Projections for Active Contributors and Pensioners
Current and ew Active Contributors
The projected number of active contributors will be calculated using the projected active
labor force. Although active contributors from state and non-state sectors still have different
treatments under the current regulations, we will not consider any differences between these
people in our projections. This is a reasonable assumption in the long term, as the Vietnamese
government is currently undertaking a variety of policy measures to close the gaps,
particularly for wages, between the two sectors.
Figure 4. Projection Methodology for Active Contributors








Source: Own modifications using ILO (1998).
Figure 4 illustrates our projection methodology for the number of active contributors.
We will project the number of active contributors by age and gender using the historical data
on the percentages of active laborers participating in the pension scheme by age and gender.
Data on the current active contributors, provided by MOLISA (2006), are presented by gender
in 5-year age groups. For the projection purposes, we use the representative data from VHLSS
2004 to adjust the current active contributors by single age and gender.
Figure 5. Distribution of Active Contributors by Age and Gender, 2005
0
30000
60000
90000
120000
150000
180000
15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59
Age
P
e
r
s
o
n
Male Female

Source: Own calculations using MOLISA (2006) and VHLSS 2004.
Figure 5 shows the age distribution for male and female active contributors in 2005.
Such a distribution was inherited from the fact that many pre-1995 schemes contributors
transferred their status to the post-1995 scheme when the VSS was established.
Projected
number of active male
contributors by age
Projected active male
laborers
by age

N
u
m
b
e
r

o
f

a
c
t
i
v
e

c
o
n
t
r
i
b
u
t
o
r
s

b
y

a
g
e

a
n
d

g
e
n
d
e
r


Active male contributors as
percentage of male laborers

Active female contributors as
percentage of male laborers
Projected active
female laborers
by age

Projected
number of active female
contributors by age
P
r
o
j
e
c
t
e
d

c
o
v
e
r
a
g
e

r
a
t
e

69
Figure 6. Historical and Projected Coverage Rates, 1996-2105
0
7
14
21
28
35
42
49
56
63
70
1996 2006 2016 2026 2036 2046 2056 2066 2076 2086 2096
%

A
c
t
i
v
e

L
a
b
o
r

F
o
r
c
e

Source: Own calculations using VSS (2004), MOLISA (2006), and United Nations (2007b).
Over the past decade, the coverage rate of the pension scheme, which is measured by
percentage of the labor force participating in the scheme, gradually increased from 8.4 percent
in 1996 to about 15 percent in 2005. Under the current regulations of the Social Insurance Law,
the coverage rate is expected to increase further. Moreover, coverage should increase as the
urban population share grows (Sin, 2005). Therefore, we will use the urbanization projections
by United Nations (2007b) to estimate the growth of coverage rates for the Vietnamese
pension scheme. Since the data are presented in 5-year intervals from 2005 to 2050, we
recalculate to get yearly growth rates, and assume that the urbanization growth rates for the
period 2050-2105 will be the same as the average rate of 2045-2050.
Figure 6 shows the historical and projected coverage rates of the Vietnamese pension
scheme in 1996-2105. We assume the coverage rate will be about 66 percent of the projected
active labor force at the end of the projection period.
Current and ew Pensioners
For the current pensioners, we will use stochastically projected mortality rates for the
general population to estimate the number of people remaining over time from the 2005 base.
Even though it is commonly recognized that pension recipients may have lower mortality rates
and higher life expectancy than the average population because they usually have better
healthcare treatment and other social services, our assumption is necessary because historical
data on mortality rates for the pensioners by age and gender are not available. We only
consider pensioners belonging to the VSS (or post-1995 scheme). We will not consider pre-
1995 pensioners, as they belong to the government system, and their benefits are paid by the
government budget, which is separate from the VSS pension fund.
Figure 7 shows the age distribution of the current male and female pensioners in 2005.
The distribution shape results from the situation that some people contributed in both the pre-
1995 and post-1995 periods. Most of the pensioners were at the normal retirement ages, but
there were also a number of early pensioners.
70
Figure 7. Distribution of the Current Pensioners by Age and Gender, 2005
0
4000
8000
12000
16000
20000
24000
28000
32000
36000
40000
44000
48000
52000
56000
60000
64000
43 45 47 49 51 53 55 57 59 61 63 65 67 69 71 73 75 77 79 81 83
Age
Persons
Male Female

Source: Own calculations using MOLISA (2006) and VHLSS 2004
For projections of new pensioners, it would not be reasonable to assume that individuals
will contribute fully from the time they enter the labor force to the time they reach the full
retirement ages, and thus we assume that male and female contributors on average will retire
at age 57 and 51, respectively, based on the data for current pensioners. Such retirement ages
are hereafter referred to as the assumed retirement ages. We estimate the number of new
pensioners using the projected number of active contributors over time. As historical data for
retirement rates by age and gender are not available to date, we will estimate the number of
new pensioners at the assumed retirement ages as percentages of the active contributors in the
previous year at the pre-retirement ages, and assume that these percentages will remain over
the projection period. In addition, the existing data show that the average length of service of
male and female participants were 33.75 and 31.25 years, respectively, which in turn provides
respective replacement rates of 70.5 percent and 68.5 percent.
4.4. Projections for Macroeconomic Variables
There are many relevant macroeconomic variables that need to be considered in the
long-term financial projections for the pension scheme. In this paper, we take the following
variables to be stochastic: (i) inflation rate, (ii) real interest rate, and (iii) real wage growth.
We assume that these variables are independently and identically distributed (i.i.d) with log-
normal distributions, in which means and variances are estimated from their respective
historical patterns as shown in Table 1.
Table 1. Historical Values for Time-series Data
Variable Time Period Arithmetic Mean Standard Deviation Source
Inflation (headline CPI) 1994 - 2005 2.64% 3.44%
IMF International Financial
Statistics (various years)
Real investment return for
pension fund assets
1996 - 2005 3.35% 3.73%
MOLISA (2006) and IMF
International Financial
Statistics (various years)
Real wage growth 1992 - 2005 5.20% 1.20%
IMF Country Statistics for
Vietnam (various years)


71
Figure 8. Average Contribution Wage of the Current Contributors by Age, 2005
0
20
40
60
80
100
120
140
160
180
200
15 18 21 24 27 30 33 36 39 42 45 48 51 54 57
Age
E
a
r
n
i
n
g
s

(
a
s

%

o
f

2
0

y
e
a
r
-
o
l
d
)
Male Female

Source: Own calculations using MOLISA (2006) and VHLSS 2004
Figure 9. Average Benefit of the Current Pensioners by Age, 2005
0
20
40
60
80
100
120
140
160
180
43 45 47 49 51 53 55 57 59 61 63 65 67 69 71 73 75 77 79 81 83
Age
P
e
n
s
i
o
n

b
e
n
e
f
i
t
s

(
n
o
r
m
a
l
i
z
e
d

t
o

a
s
s
u
m
e
d

r
e
t
i
r
e
m
e
n
t

a
g
e
,

%
)
Male Female

Source: Own calculations using MOLISA (2006) and VHLSS 2004.
For the projections, we assume that both contributors wages and pensioners benefits
will be adjusted by the projected nominal wage growth rates, which are the sum of the
projected real wage growth rates and the projected inflation rates. Figure 8 presents the by-age
wage pattern of the current contributors as percentages of respective male and female wages at
age 20, while Figure 9 shows the normalized benefits of the current pensioners as percentages
of respective male and female pensioners benefits at their assumed retirement ages.
4.5. Projections for Pension Fund Balance
As presented in Figure 1, the pension fund balance depends on total revenues and total
expenditures. In this paper, we assume that total revenues only include revenues from
contributions and investment returns, while total expenditures only include expenditures for
retirement, survivorship, and administration.
Revenues
Revenues in year t (Rev
t
) can be defined as follows:
Rev
t
= contributions
t
+ investment returns
t
(5)
= number of contributors
t
* average wage
t
* contribution rate
t
+
72
+ fund
t-1
* (1+r
t
),
in which r
t
is the projected rate of return on investments in year t.
Expenditures
Expenditures in year t (Exp
t
) can be defined as follows:
Exp
t
= Retirement
t
+ survivorship
t
+ administrative costs
t
= (6)
= number of retirees
t
* average benefit
t
+ survivorship
t
+administrative costs
t

= number of retirees
t
* average reference wage
t
* average replacement rate +
+ survivorship
t
+administrative costs
t

As mentioned above, the benefit formula is varied for beneficiaries from different
economic sectors, since it refers to different reference wages, which are used to estimate
pension benefits. The dataset, however, does not include detailed information on the
relationship between average compensation and the average reference wage of the contributors
by age and gender, and thus it would be reasonable to assume that average reference wages for
males and females are proportional to their respective average compensation at the time of
retirement. In practice, the actual difference between these indicators gradually reduced from
about 15 percent in 1998 to 5.7 percent in 2002,
2
and we therefore assume that the average
reference wage is about 90 percent of the average compensation at the time of retirement
throughout the projection period. In addition, the average replacement rates for male and
female contributors in 2005 will be used, and they are also assumed to be unchanged during
the projection period.
A further issue is that the benefits of dependants or survivors (e.g. children, spouses, or
parents) will entail lump-sum or monthly payments. It is infeasible, however, to estimate these
benefits with existing data. For the present purposes, these payments are assumed to represent
about 20 percent of total retirement benefits in each year, which is in line with the historical
situation.
As regulated, the total administration costs are equal to 4 percent of the total
contributions each year. We assume that this rate will remain until 2020, and then decrease to
2 percent afterwards, as this represents the governments current intentions.
4.6. Sensitivity Analysis
In order to examine how the pension fund balance will be changed when relevant
variables change, we will conduct sensitivity analyses as well. Table 2 summarizes our
sensitivity tests.
As inflation and real wage growth play key roles in adjusting pension benefits and
contributions, in the first and second tests, we will consider how the pension fund balance will
be influenced when stochastic forecast values of each variable are reduced by 1 percentage
point.
In the third test, the assumption that the stochastic forecast values for the pension fund
returns are 1 percentage point less will also be examined to see more clearly about the role of
investment returns.
Coverage rates are usually considered as an important factor in balancing the pension
fund. Thus, in the fourth test, we will compare our projected coverage rate increases to a
coverage rate that stays at the current level (15 percent of the labor force).


2
These figures are respectively from Holzmann et al. (2000), and own estimation from VSS (2004).
73
Table 2. Sensitivity Tests
Test Variable Baseline Value Modified Value
1 Inflation Mean value is 2.64% Stochastic forecast values are reduced
by 1 percentage point
2 Real wage growth Mean value is 5.20% Stochastic forecast values are reduced
by 1 percentage point
3 Real returns on pension fund
assets
Mean value is 3.35% Stochastic forecast values are reduced
by 1 percentage point
4 Coverage rates Continuous rise in coverage rates up
to 66% of labor force in 2105
Coverage Rates remain at their 2005
value of 15% throughout the projection
period
5 Retirement ages Men retire at 57, Women at 51 (1) Men retire at 58, Women at 52
(Higher Retirement Age)
(2) Men retire at 56, Women at 50
(Lower Retirement Age)
6 Administrative costs 4% of contributions until 2020, then
2% thereafter
Always 4% of total contributions
7 Cost-of-Living-Adjustment
(COLA)
COLA for benefit payments is nominal
wage growth
(1) COLA for benefit payments is
inflation rate.
(2) COLA for benefit payments is
inflation rate plus one half of real wage
growth.
(3) COLA for benefit payments is
nominal wage minus 1 percentage
point.
8 Contribution rates Currently 16%, but adjust upward to
22% by 2014 and thereafter
Always stay at 16%
9 Long-term total fertility rates Expected to be 1.85 Assume to be at replacement rate of
2.1

There have also been a number of debates about the current normal retirement ages. As
the assumed retirement ages remain for the whole projection period, we will conduct two
analyses in the fifth test. The first scenario, namely higher retirement ages, assumes that male
and female contributors will retire at 58 and 52 years old, respectively. Conversely, the second
scenario, namely lower retirement ages, assumes that male and female contributors will retire
at 56 and 50 years old, respectively.
In the sixth test, we consider the role of administrative costs by comparing a scenario in
which administrative costs always stay at 4 percent of each years total contributions.
Regarding indexation of pension benefits, we will examine in the seventh test how the
pension fund balance will respond to a scenario in which pension benefits are indexed to
inflation, instead of nominal wage growth.
In the eighth test, we will also examine the impacts of contribution rates on the pension
fund balance. In our test, we assume that the total contribution rate will stay at 16 percent for
the whole projection period.
Lastly, as fertility rates play an important role in demographic changes, we will compare
our estimates with an assumption that the long-term total fertility rate fluctuates around 2.1,
which is higher than our baseline value of 1.85.
74
5. Analysis of Findings
5.1. Projected Population and Labor Force
Figure 10. Projected Population, 2050 and 2105
Source: Own calculations.
Figure 10 presents results from the stochastic population projection for Vietnam in 2050
and 2105. The median forecasts are surrounded by a 90 percent confidence interval. The aging
process can be seen with these two population pyramids, as the number of children will
experience dramatic declines, while the number of elderly people, especially oldest people
(aged 80 and over), will increase substantially.
Figure 11. Projected Dependency Ratios, 2005-2105
2020 2040 2060 2080 2100
0
0.2
0.4
0.6
0.8
1
1.2
Youth Dependency Ratio
2020 2040 2060 2080 2100
0
0.2
0.4
0.6
0.8
1
1.2
Old-Age Dependency Ratio
2020 2040 2060 2080 2100
0
0.2
0.4
0.6
0.8
1
1.2
Population Dependency Ratio
Source: Own calculations.
-5 -4.5 -4 -3.5 -3 -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
0
5
10
15
20
25
30
35
40
45
50
55
60
65
70
75
80
85
90
95
100
Stochastic Forecast of Vietnam's Population Pyramid in 2050
Median Forecast Surrounded by 90% Confidence Interval
# Females # Males
Population Size (Millions of People)
Age
-5 -4.5 -4 -3.5 -3 -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
0
5
10
15
20
25
30
35
40
45
50
55
60
65
70
75
80
85
90
95
100
Stochastic Forecast of Vietnam's Population Pyramid in 2105
Median Forecast Surrounded by 90% Confidence Interval
# Females # Males
Population Size (Millions of People)
Age
75
The aging population can also be observed with three important dependency ratios: (i)
the youth dependency ratio, which is estimated by the number of children (aged 0-14) divided
by the number of working age people (aged 15-59); (ii) the elderly dependency ratio, which is
measured the number of elderly (aged 60 and over) divided by the number of working-age
people, and (iii) the population dependency ratio, which is measured by the sum of the
aforementioned dependency ratios.
Figure 11, which shows the median projections surrounded by a 90 percent confidence
interval, indicates that the population dependency ratio will be mostly driven by the elderly
dependency ratio. In other words, the growth rate of youth will be dominated by the growth
rate of elderly population in the coming decades, especially from 2020 onwards. The youth
dependency ratio can actually be expected to decrease throughout the projection period. This
has important implications for any social and economic policies related to demography.
5.2. Projection Results for the Pension Fund Balance: Baseline Scenario
Figure 12. Projected Pension Fund, 2005-2105: Baseline Case
2010 2020 2030 2040 2050 2060 2070 2080 2090 2100
0
1
2
3
4
5
6
7
x 10
12
Pension Fund - Baseline Case

Source: Own calculations.
Figure 12 presents the baseline projected pension fund with median values, which are
surrounded by a 90 percent confidence interval. The median values show that the pension fund
size will peak at about VND 2.5 trillion in 2046, and then decrease rapidly to zero in 2052. In
other words, the pension fund will be depleted in 2052 under current regulations. Within the
bound of the 90 percent confidence interval, we can see that such depletion can happen as
early as in 2048, and as late as in 2056. This finding is also consistent with some previous
deterministic studies on the long-term pension fund balance in Vietnam, such as Giang (2004),
Castel and Rama (2005), and Nguyen (2006). It means that, without further reform
considerations, the Vietnamese pension scheme will face a variety of financial difficulties,
which in turn result in generational imbalances between participating cohorts.
76
Figure 13. Projected Investment Income
as a Percentage of the Projected Revenues, 2005-2105: Baseline Case
2010 2020 2030 2040 2050 2060 2070 2080 2090 2100
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Investment Income as % of Total Revenues - Baseline Case

Source: Own calculations.
Figure 13, which shows the median projections surrounded by a 90 percent confidence
interval, indicates that investment income will play a small role in the total revenues of the
pension fund, as its peak will be about 24 percent of the total revenues in the early 2030s. In
other words, participant contributions will be the main driving force for pension fund revenues
during the projection period.
5.3. Projection Results for the Pension Fund Balance: Sensitivity Scenarios
Figure 14. Pension Fund under Lower Inflation and Real Wage Growth, 2005-2105
2010 2020 2030 2040 2050 2060 2070 2080 2090 2100
0
0.5
1
1.5
2
2.5
3
x 10
12
Pension Fund - Inflation is One Percentage Point Less
2010 2020 2030 2040 2050 2060 2070 2080 2090 2100
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
x 10
12
Pension Fund - Real Wage Growth is One Percentage Point Less
Source: Own calculations.
Figure 14 presents the stochastic projections for the pension fund balance surrounded by
a 90-percent confidence interval in the case that the stochastic forecast values of the inflation
rate and real wage growth are reduced by 1 percentage point. The median values imply that
that the pension fund status will not be significantly improved in either case. Lower inflation
has almost no effect, presumably because there will be proportional reductions in contributions
and benefits. Meanwhile, lower real wage growth extends the life of the fund about two years.
Essentially, both inflation and real wage growth affect contributions and pension benefits in
similar ways, such that the net effect is small.
77
Figure 15. Pension Fund under Lower Rates of Returns, 2005-2105
2010 2020 2030 2040 2050 2060 2070 2080 2090 2100
0
0.5
1
1.5
2
2.5
3
3.5
4
x 10
12
Pension Fund - Investment Returns are One Percentage Point Less

Source: Own calculations.
When the stochastic forecast values of rate of returns are 1 percentage point less than
those in the baseline projections, the balance of fund will not be changed substantially, as
shown in Figure 15. The median values of projections imply that the fund will be zero in 2051,
which is only one year earlier than the baseline projections. The estimates within a 90-percent
confidence interval also indicate that the fund depletion will happen in as early as 2047, and
2055 at the latest. The pension fund is not particularly sensitive to investment returns because
contributions are the main source of revenues.
Figure 16. Pension Fund under Constant Coverage Rates, 2005-2105
2010 2020 2030 2040 2050 2060 2070 2080 2090 2100
0
0.5
1
1.5
2
x 10
12
Pension Fund - Aggregate Coverage Rates Stay at 15%

Source: Own calculations.
Our estimates under the assumption that coverage rates will remain constant at 15
percent also provide similar results, which are presented in Figure 16. The median forecast
values show that the pension fund depletion will occur in 2050, which is slightly sooner than
the baseline projections. Higher coverage rates can help to slightly extend the duration of the
pension fund as more contributions flow into the pension fund, but eventually this will mean
more benefit payments as well, such that the effect is short lived. Nonetheless, this result is
important because the effect is even more short lived than may be generally thought.
78
Figure 17. Pension Fund under Lower and Higher Retirement Ages, 2005-2105
2010 2020 2030 2040 2050 2060 2070 2080 2090 2100
0
2
4
6
8
10
12
x 10
11
Pension Fund - Lower Retirement Ages (Men=56, Women=50)
2010 2020 2030 2040 2050 2060 2070 2080 2090 2100
0
0.5
1
1.5
2
2.5
x 10
13
Pension Fund - Higher Retirement Ages (Men=58, Women=52)
Source: Own calculations.
Figure 17 presents our estimated results under 1-year lower/higher retirement ages in
comparison with the assumed retirement ages. The left panel shows the estimated results from
a scenario of lower retirement ages (56 and 50 for males and females, respectively), while the
right panel shows those from a scenario of higher retirement ages (58 and 52 for males and
females, respectively). In comparison with the baseline projections, the median values of these
projections show that the pension fund balance will get worse under the former scenario, while
it will be significantly improved under the latter scenario. Indeed, we are finding the first case
of a parameter that plays a particularly important role in the solvency of the pension fund.
Under lower retirement ages, the median forecast values indicate that the pension fund
balance will be depleted in about 2043, which is about 9 years earlier than the baseline
scenario. Estimates with a 90-percent confidence interval for this scenario show that the fund
will become zero in as early as 2040, and 2046 at the latest. Conversely, the right panel shows
that higher retirement ages will help to maintain the pension fund solvency for about 13 more
years in comparison with the baseline case. The 90-percent confidence interval values now
cover a wider range, from 2060 to 2072. All these estimates imply that the dominance of early
retirement will become a great challenge for maintaining the long-term financial balance of the
Vietnamese pension scheme.
Figure 18. Pension Fund under Constant Administrative Costs, 2005-2105
2010 2020 2030 2040 2050 2060 2070 2080 2090 2100
0
0.5
1
1.5
2
2.5
3
3.5
x 10
12
Pension Fund - Administrative Costs Stay at 4 Percent of Contributions

Source: Own calculations.
79
Our estimated results in Figure 18 are found in the case that administrative costs will
stay at 4 percent for the whole projection period. The median forecast values of the projections
indicate that the pension fund will be depleted in 2051, which is just a year earlier than the
baseline case. This means that, given other factors, the current administrative costs will not be
determinant factor of the pension fund balance.
Figure 19. Changes in Pension Fund Indexation, 2005-2105
2010 2020 2030 2040 2050 2060 2070 2080 2090 2100
0
1
2
3
4
5
6
x 10
13
Pension Fund - COLA is inflation rate

2010 2020 2030 2040 2050 2060 2070 2080 2090 2100
0
2
4
6
8
10
12
14
16
x 10
12
Pension Fund - COLA is inflation rate plus a half of real wage growth

2010 2020 2030 2040 2050 2060 2070 2080 2090 2100
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
x 10
12
Pension Fund - COLA is nominal wage growth minus 1 percentage point

Source: Own calculations.
80
Figure 19 presents the results for three different cost-of-living adjustment (COLA)
changes. The most upper figure shows stochastic projections for the pension fund balance
interval when the COLA for pension benefits are indexed to inflation, instead of nominal wage
growth. In comparison with the baseline estimates, the median forecast values in this case
show that the pension fund will stay solvent for about 29 more years, i.e., it will be depleted in
2081. The 90-percent confidence interval values now range from 2074 to 2091. The middle
figure shows the case when the COLA is equal to inflation rate plus one half of real wage
growth. The pension fund would stay solvent for about 15 more years, i.e. it would be depleted
in 2067. The 90-percent confidence interval values now range from 2058 to 2073. The last
figure shows the results when changing the COLA to 1 percentage point less than the nominal
wage growth. The median forecast values show that the pension fund will stay solvent until
2054, which is only two more years than will the fund in the baseline scenario. Modifying the
COLA downward can substantially improve the solvency of the pension fund. With such a
large impact, it is worthwhile to note that for the case of the United States, Pfau (2006) finds
that among a variety of different ways to reduce pension benefits, a decrease in cost-of-living
adjustments does provide an effective path to reform with minimal poverty increases for the
elderly.
Figure 20. Pension Fund under Constant Contribution Rates, 2005-2105
2010 2020 2030 2040 2050 2060 2070 2080 2090 2100
0
1
2
3
4
5
6
7
x 10
11
Pension Fund - Contribution Rates Stay at 16%

Source: Own calculations.
We also find an important role for the contribution rate in maintaining the pension fund
balance. Figure 20 shows the estimated results for the pension fund balance when the
contribution rate is kept constant at 16 percent. The median forecast values indicate that the
pension fund will be depleted in about 2038, which is 14 years sooner than in the baseline
projections. The 90-percent confidence interval values show a more narrow range, from 2036
to 2041. The estimates imply that contribution rates will become an important factor to
maintain the pension fund balance. It does not mean, however, that the government can
continue increasing the contribution rates, as we can expect disincentive effects for working
and increased contribution evasion when social security taxes are too high.
5.4. Pension Fund Balance under Higher Long-term Total Fertility Rate
In order to see how demographic changes will influence the long-term pension fund
balance, we assume that the long-term total fertility rate (TFR) for the Vietnamese population
will fluctuate around 2.1, while other factors will be kept the same as those in aforementioned
population projections.
81
Figure 21. Projected Pension Fund and Investment Income
under Higher Long-term TFR, 2005-2105
2010 2020 2030 2040 2050 2060 2070 2080 2090 2100
0
2
4
6
8
10
12
14
16
x 10
12
Pension Fund with higher long-term TFR - Baseline Case

Source: Own calculations.
Figure 21 presents the estimated results for the pension fund balance when Vietnam has
a higher long-term TFR. The median forecast values indicate that the pension fund will be
depleted in about 2055, which is 3 years later than in the baseline projections under the lower
long-term TFR assumption. The 90-percent confidence interval values show a wider range
(about 13 years, from 2050 to 2063) than that of the baseline projections. Though the
improvements are slight, higher fertility does extend the solvency of the pension fund by
providing a larger contribution base.
5.5. Projection Results for the Pension Fund Balance: Reform Proposals
Figure 22. Reform Proposals, 2005-2105
2010 2020 2030 2040 2050 2060 2070 2080 2090 2100
0
0.5
1
1.5
2
2.5
3
3.5
x 10
14
Pension Fund - Reform Proposal 1
2010 2020 2030 2040 2050 2060 2070 2080 2090 2100
0
0.5
1
1.5
2
2.5
3
3.5
4
x 10
15
Pension Fund - Reform Proposal 2
Source: Own calculations.
The estimated results from the baseline and sensitivity scenarios show that contribution
rates, retirement ages, and indexation methods for pension benefits will be very important
factors determining the long-term balance of the pension fund, while the other factors will not
play a significant role. Under assumptions that Vietnam continues the current pension scheme
and contribution rates will remain as regulated in the current Social Insurance Law, we will
provide two proposals for maintaining the pension fund balance over a longer term: (i) the
assumed retirement ages for males and females will increase to the normal retirement ages (60
82
and 55 for males and females, respectively), and (ii) the assumed retirement ages will increase
as in (i) and the cost-of-living adjustments for pension benefits will be indexed to inflation.
Our results for the two reform proposals are presented in Figure 22. For the first proposal,
the median forecast values show that the pension fund balance will be substantially improved
in comparison with the baseline case. For the median forecast, the fund will be depleted in
2088, which is 36 more years than in the baseline scenario. Under the second proposal, the
pension fund will be sustainable and even accumulate large surpluses over the projection
period.

6. Concluding Remarks
In this paper, we applied stochastic estimation techniques to some key variables related
to population and macroeconomy in order to provide a long-term vision for the financial status
of the Vietnamese pension scheme. The estimated results show that the pension fund will be
depleted in about 2052 with a 90-percent confidence interval range of 8 years. Sensitivity tests
indicate that the long-term pension fund balance will depend significantly on contribution
rates, retirement ages, and indexation methods for pension benefits, while it will not depend
much on coverage rates, administrative costs, the rate of return on investments, and fertility
rates. Our reform proposals imply that increases in retirement ages and inflation indexation for
the pension benefits will help to substantially improve the pension fund balance over the long
run. With these results, we recommend that Vietnam reform the current pension scheme as
soon as possible in order to maintain its long-term financial stability.
Although this paper could mitigate some limitations of the previous studies which used
deterministic estimation techniques, it cannot avoid some limitations. First, due to the short
length of Vietnamese historical data, the projections in this paper depend on a variety of strong
assumptions. Second, the paper did not estimate financial gains and losses between
participating generations of the scheme, which are useful for indicating whether the current
pension scheme can maintain a generational balance under a stochastic environment. We hope
to address these limitations in subsequent research.

References
American Academy of Actuaries (AAA). 2005. A Guide to the Use of Stochastic Models in
Analyzing Social Security, Issue Brief (October 2005). Washington D. C: American
Academy of Actuaries.
Bongaarts, J., and R. A. Bulatao (eds.). 2000. Beyond Six Billion: Forecasting the Worlds
Population. Washington D. C: National Academy Press.
Castel, P. and M. Rama. 2005. Comments on the New Social Insurance Law, mimeo. Hanoi:
World Bank Vietnam.
Coale, A., and G. Guo. 1989. Revise Regional Life Tables at Very Low Level of Mortality,
Population Index, Vol. 55, No. 4. (Winter, 1989): 613-643.
Dang, T., T., P. Antolin, and H. Oxley. 2001. Fiscal Implications of Ageing: Projections of
Age-related Spending. Economic Department Working Papers ECO/WKP31,
Organization for Economic Co-operation and Development (OECD). Paris: OECD.
Giang, T. L. 2004. The Pension Scheme in Vietnam: Current Status and Challenges in an
Aging Society, Vietnam Development Forum (VDF) Discussion Paper No.2 (E). Hanoi:
Vietnam Development Forum.
83
Giang, T. L. 2006. Pension Liabilities and Generational Relations: The Case of Vietnam,
Oxford Institute of Ageing Working Paper No. 106. Oxford: OIA, the University of
Oxford.
Giang, T. L. 2007. The Pension Scheme in Vietnam: Structure, Operation, and Challenges.
Presentation at Conference on Alternative Investments in Asia, March 14-15, 2007, Hong
Kong (SAR, China).
Giang, T. L. 2008a. Aging and the Public Pension Scheme in Vietnam: A Long-term
Financial Assessment, East & West Studies, Vol. 20, Issue 1, June 2008: 171-193.
Giang, T. L. 2008b. Quy bao hiem xa hoi: Nhan to tac dong, thach thuc, va lua chon chinh
sach cho Vietnam (Social Insurance Fund: Determinant Factors, Challenges, and Policy
Options for Vietnam) (unpublished manuscript). Hanoi: Department of Social Insurance,
Ministry of Labor, War Invalids, and Social Affairs, Vietnam (MOLISA).
Giang, T. L, and W. D. Pfau. 2007. The Elderly Population in Vietnam during Economic
Transformation: An Overview. In Giang Thanh Long and Duong Kim Hong (eds.)
Social Issues under Economic Transformation and Integration in Vietnam: 185-210.
Hanoi: Vietnam Development Forum.
Hagemann, R., and G. Nicolleti. 1989. Ageing Populations: Economic Effects and
Implications for Public Finance. Department of Economics and Statistics, Organization
for Economic Co-operation and Development (OECD). Paris: OECD.
Heller, P. 1997. Aging in Asian Tiger Economies: Challenges for Fiscal Policy. IMF
Working Paper No. 97/143. Washington D. C: IMF.
Heller, P. 1998. Aging in Asian Tiger Economies, Finance and Development, June 1998:
26-29.
Holmer, M., R. 2003. Methods of Stochastic Trust Fund Projection, Policy Simulation
Group, Social Security Administration (SSA), mimeo.
Holzmann, R., I. W. M. Arthur, and Y. Sinn. 2000. Pension Systems in East Asia and the
Pacific: Challenges and Opportunities, Social Protection Discussion Paper Series
No.0014, the World Bank.
IDB (International Database, U.S Census Bureau). 2008. Country Summary: Vietnam.
Retrieved from http://www.census.gov/ipc/www/idb/country/vmportal.html on March 1,
2008.
ILO (International Labor Organization). 1998. Vietnam: Actuarial Valuation of the Vietnam
Social Security Scheme, Project VIE24V3.598. Geneva: ILO.
Kunieda, S., 2002. Japanese Pension Reform: Can We Get out of Intergenerational
Exploitation? in Proceedings of the International Symposium on Pension Reforms in
Asian Countries: 23-63. Tokyo: Hitotsubashi University.
Lee, R., D. 1993. Modeling and Forecasting the Time Series of US Fertility: Age Distribution,
Range, and Ultimate Level, International Journal of Forecasting, 9 (1993): 187-202.
Lee, R., D. 2000. The Lee-Carter Method for Forecasting Mortality, with Various Extensions
and Applications, orth American Actuarial Journal, Vol. 4, No. 1: 80-92.
Lee, R., D, and L. R. Carter. 1992. Modeling and Forecasting U.S. Mortality, Journal of
American Statistical Association, Vol. 87, No. 419 (Sep., 1992): 659-671.
Lee, R., and T. Miller. 2001. Evaluating the Performance of the Lee-Carter Method for
Forecasting Mortality, Demography, Vol. 38, No. 4 (Nov., 2001): 537-549.
84
Lee, R., and S. Tuljapurkar. 1994. Stochastic Population Forecasts for the United States:
Beyond High, Medium, and Low, Journal of the American Statistical Association, Vol.
89, No. 428 (Dec., 1994): 1175-1189.
Lee, R., and S. Tuljapurkar. 1998. Uncertain Demographic Futures and Social Security
Finances, AEA Papers and Proceedings, Vol. 88, No. 2 (May 1998): 237-241.
Lopez, A, D., J. Salomon, O. Ahmad, C. J .L. Murray, and D. Mafat. 2000. Life Table for
191 Countries: Data, Methods, and Results. The World Health Organization.
MOLISA (Ministry of Labour, War Invalids, and Social Affairs, Vietnam). 2006. Tong hop
so lieu thong ke Bao hiem Xa hoi (Statistical Summary of Social Insurance)
(unpublished reports). Hanoi: MOLISA.
MOLISA. 2006. Statistical Data of Employment and Unemployment in Vietnam 1996-2005.
Hanoi: The Publishing House of Social Labor.
Nguyen, T. H. T. 2004. Thuyet trinh cua Uy ban cac van de xa hoi ve khac phuc mot so bat
cap ve bao hiem xa hoi gop phan bao dam quyen loi chinh dang cua nguoi lao dong
(Debates of the National Assemblys Committee for Social Affairs on Solutions for the
Difficulties of Social Insurance Performance in order to Guarantee the Rights of the
Workers). Hanoi: The National Assembly.
Nguyen, T. T. A. 2006. Du bao tai chinh quy bao hiem xa hoi Vietnam (Projection of the
Vietnamese Social Insurance Fund). Hanoi: Social Insurance Department, MOLISA.
OECD (Organization for Economic Co-operation and Development). 2005. Ageing and
Pension Reforms: Implications for Financial Markets and Economic Policies,
November 2005 Supplement. Paris: OECD.
Okita, Y., W. D. Pfau, and T. L. Giang. 2008. Stochastic Forecast for Japans Population,
2006-2100 (mimeo). Tokyo: National Graduate Institute for Policy Studies (GRIPS).
Pfau, W. D. 2006. Comparing the Impacts of Social Security Benefit Reductions on the
Income Distribution of the Elderly, ational Tax Journal. Vol. LIX, No. 2 (June 2006):
195-210.
Sin, Y. 2005. China: Pension Liabilities and Reform Options for Old Age Insurance. World
Bank Working Paper 2005-1. Washington D.C: World Bank.
United Nations. 2007a. World Population Prospects: The 2006 Revision Population Database,
United Nations Population Division. New York: United Nations. Retrieved on March 1,
2007 from http://esa.un.org/unpp/index.asp?panel=3.
United Nations. 2007b. World Urbanization Prospects: The 2007 Revision Population
Database, United Nations Population Division. New York: United Nations. Retrieved on
March 1, 2007 from http://esa.un.org/unup/index.asp?panel=3.
UNESCAP (United Nations Economic and Social Commission for Asia and the Pacific). 2005.
Economic and Social Survey of Asia and the Pacific 2005: Dealing with Shocks. New
York: UNESCAP.
Van der Noord, P., and R. Herd. 1993. Pension Liabilities in Seven Major Economies.
Economics Department Working Paper No.142, Organization for Economic Co-
operation and Development (OECD). Paris: OECD.
VSS (Vietnam Social Security). 2004. Du bao quy bao hiem xa hoi Vietnam (Projections of
Social Insurance Fund in Vietnam) (unpublished report). Hanoi: VSS.
WHO (World Health Organization). 2008. Life Tables for WHO Member States. Retrieved
from http://www.who.int/whosis/database/life_tables/life_tables.cfm on March 1, 2008.

85
The Economics of Not Using the Health Insurance Card:
A Case Study of Vietnam

Trong-Ha Nguyen
Crawford School of Economics and Government,
Australian National University
Email: ha.trong@anu.edu.au



Abstract
In developing countries, many factors may prevent the insured from the full
utilization of the health insurance card. In this paper we propose a theoretical
framework of health insurance utilization determination. The empirical
investigation focuses on the role of income, health status, information, types of
health insurance and care providers on health insurance utilization. The results
show that using health insurance to pay for outpatient treatment is perceived as
an inferior good. Income, however, has no impact on the use pattern of the health
insurance card in inpatient treatment. The insured with more information use the
health insurance card more often in outpatient but not in inpatient treatment. The
ill have higher utilization rates than the healthy. The utilization rates vary by
types of insurance schemes and care providers.

Keywords: health insurance, utilization, developing countries, Vietnam
JEL Classification: I11, D12, O12


86
1. Introduction
Two leading issues of the economics of health insurance and health care are the
determinants of choices of insurance and the impact of the choice of insurance level on the
demand for health care. Several theoretical models and empirical evidence show that attitude
toward risk, health status, income, insurance premium, user fees and trust in providers quality
of care are important factors determining health insurance demand
1
PT(Arrow 1963; Cameron
and Trivedi 1991; Cameron et al. 1988; Manning and Marquis 1996; Marquis and Holmer
1996; Mechanic 1994). Most studies on the impact of insurance on care utilization show that
having health insurance significantly increases health care utilization for the insured in
developed (Buchmueller et al. 2004; Cameron et al. 1988; Manning et al. 1987) and
developing countries (Hofter 2006; Hugh 1999; Pagn et al. 2007; Pradhan et al. 2007).
The question why the insured are reluctant to use their cards seems irrelevant in the
context where health insurance has been found to be overwhelmingly beneficial to the bearers.
This paper argues that in developing countries, there are a number of factors that may drive the
insured to partially use or, in some extreme cases, not use their health insurance cards. For
example, using a qualitative research method, Sinha et al. (2006) suggest that illiteracy among
members, lack of health care infrastructure and cumbersome claim rules are the main barriers
faced by members of a community-base insurance scheme in accessing the scheme benefits in
India. In addition, Kamuzora and Gilson (2007) show that low quality of care or lack of trust
in the health fund has led to under-utilization of health care cards and hence low enrollment in
the community health fund in Tanzania. These examples suggest that assumptions made in
health economics such as the insured being fully aware of their benefits or there being no
constraint on the supply of health care are often less applicable to the developing world. The
difference in health care sectors between developed and developing countries thus calls for in-
depth studies about the functioning of the health insurance sector in the latter (Ensor and
Witter 2001).
This study attempts to provide insights into the functioning of the health insurance
sector in the developing world by examining the factors underpinning the insureds decision
not to use their cards to claim for treatment. In particular, we explore two sub-questions: first,
if there is evidence of under-utilization of health insurance benefits, Second and more
important, what factors impede the insured from using their cards. We do so by proposing a
theoretical model of factors that may influence the insureds decision to use their cards to
receive treatment. Then we empirically test what determines the insureds level of health
insurance card utilization by using the data from the Vietnam Household Living Standard
Survey 2006 (VHLSS06).
This study distinguishes itself as the first to formally model and empirically test the
determinants of health insurance utilization in the context of a developing country. Related
literature in the context of developed countries includes studies on the impact of waiting time
on the insureds behavior in which the waiting time in the public hospital may create
incentives for the insured (under some form of insurance) to move to the private hospital at
full cost payment or to purchase additional health insurance (Cotton and Feigenbaum 1984;
Hoel and Sther 2003).
Vietnam is an interesting case to study health insurance utilization. Over the past two
decades, the economy has grown rapidly. In line with economic growth, there are concerns
about increasing economic inequalities. Under this pressure, the government intervenes by
introducing various health insurance policies to provide necessary health services. Despite the
beneficial impacts these insurance policies can bring to the insured, there is anecdotal

1
See Schneider (2004) and Jowett (2004) for recent reviews of theoretical models for demand for health
insurance in the context of developing countries

87
evidence that the insured are reluctant to use their cards for treatment. The answer for
Vietnams low coverage problem, therefore, lies in an understanding of the reasons why the
insured are reluctant to use their health insurance cards. The experience of Vietnam can then
be shared with other developing countries that may also be struggling to increase their health
insurance coverage.
The remainder of the paper is structured as follows. In the next section, we review the
health insurance system in Vietnam. Section 3 presents a theoretical framework to study the
insureds choice in using the health insurance card. Section 4 introduces a formula to measure
the level of health insurance utilization. Section 5 describes the data set and discusses its
relative strengths for our study. Section 6 describes the evidence of health insurance under-
utilization using our defined ratio of insurance utilization. Section 7 introduces the empirical
model and econometric strategy used to investigate the determinants of health insurance
utilization. Empirical results are introduced in Section 8. Finally, we summarize and discuss
policy implications in Section 9.

2. Overview of Vietnams Health Insurance
Vietnams health insurance can be classified into three main schemes: compulsory
health insurance (CHI), health care for the poor (HCFP) and voluntary health insurance
(VHI)TP
2
PT. First, the compulsory part of the health insurance system includes social health
insurance (SHI) and free health care for children below six years of age. The compulsory
scheme mainly covers public servants, employees in State Owned Enterprises (SOEs) and
those in the private formal sector. The contribution to the scheme is set at 3 per cent of salary,
of which the employer contributes 2 per cent and the rest is paid by the employee. The
mandatory scheme is also applied to pensioners and people of merit. In 2005, the government
decided that children age under six were to be provided free health services at public health
facilities or private providers who have contract with the authorities.
Second, Health Care Fund for the Poor (its predecessor was the Free Health Care
Certificate) program was established in 2003 to provide free access to services and financial
protection to those who cannot afford health care charges. The beneficiaries of the HCFP are:
households classified as poor according to the official standard; all households in especially
disadvantaged communes (defined and covered by Decision 135 of the Government to provide
support for those communes); and ethnic minorities living in six mountainous northeast and
five highland provinces.
Finally, the voluntary part of the health insurance system was initiated in 1994 to
provide the rest of the population not covered by CHI and HCFP. In 2006, enrolment in VHI
was restricted to specific groups of the population and largely group based
3
PT.
Table 1 reports that more than half, an increase from 38 per cent in 2004, of Vietnamese
were covered by at least one type of health insurance in 2006. The three main types of health
insurance are the free health care card, required (both state and non-state) and student
voluntary schemes. Over the 2002- 04 period, the number of people enrolled in the free health
care card, health care for the poor, required and voluntary health insurance schemes increased.
Among these schemes, voluntary, other than student, health insurance has tripled (from a very
low base of 1.4 per cent of the population who were coved in 2004).
[Table 1 about here]

2
For other reviews of Vietnam health insurance, see Ekman at el. (2008) or World Bank (2007)
3
Group enrollment requirements were dropped in 2008 via a joint circular no. 14/2007/TTLT-BYT-BTC
dated 10/12/2007

88
Health Insurance Benefits
The benefit package is quite generous and includes outpatient and inpatient treatment at
all levels of the health care system
4
PT. In particular, the health insurance covers consultations;
diagnostic and treatment; X-ray and laboratory tests, diagnostic imaging procedures and drugs
listed as reimbursable by the Ministry of Health. The health insurance fund also covers some
types of expensive high-tech treatment (including Magnetic Resonance Imagery (MRI) or
open-heart surgery). The benefit package is uniform across all schemes. Some exceptions are
observed. For example, the insured under pension or merits basis are entitled to 100 per cent
of expensive high-tech treatment without limit while others have to pay the amount over the
limit.
The insured can receive care delivered by public health care facilities. Recently, the
insured can acquire health services by private health care providers who have signed a contract
with Vietnam Social Security (VSS). Regulations on co-payment for treatment change from
time to time. From 2003 to September 2005, 20 per cent co-payment rate for treatment was
shared by the insured (joint circular no. 77/2003/TTLT-BYT-BTC dated 07/08/2003). The
benefits for the insured were more generous from September 2005 to April 2007 as all
expenditure under VND 7 million per treatment was covered by the insurer. For treatment
above VND 7 million, 40 per cent co-payment is required for the insured (joint circular no.
22/2005/TTLT-BYT-BTC dated 24/08/2005 and joint circular no. 21/2005/TTLT-BYT-BTC
dated 27/07/2005). The co-payment requirement was reintroduced in 2007 where the insured
has to pay 20 per cent of costs incurred during treatment (joint circular no. 06/2007/TTLT-
BYT-BTC dated 30/03/2007). VHI requires a waiting period to be put into effect. The
voluntary health insurance card can be used 30 days after the day the premium is paid for first
time participants. In contrast, the CHI has immediate effect.
Impact of Health Insurance on Health Care Utilization
Results from recent studies on Vietnam show that the insured (mostly under compulsory
and health care for the poor schemes) have benefited from the schemes in terms of health care
utilization increase, out-of-pocket (OOP) expenditure decrease or health status improvement.
For instance, Sepehri et al. (2006a) find a positive influence of health insurance on hospital
admission and length of stay. In terms of financial protection, health insurance is found to
reduce average out-of-pocket expenditure by from 20 per cent Sepehri et al. (2006b) to 200
per cent (Jowett et al. 2003). Health insurance is also found to have a positive impact on the
nutritional status of the insured (Wagstaff 2007a). Other studies focusing on health care for the
poor also find the beneficial impacts of this scheme in increase health care utilization, reduce
household OOP expenditure or reduce the risk of catastrophic health expenditure (Wagstaff
2007b). The impact of health insurance on the behavior of the insured in choosing care
providers has also been studied. For example, insured patients are found to be more likely to
use outpatient facilities and public providers than the uninsured (Jowett et al. 2004). In
addition, health insurance has been reported to decrease self-treatment among the insured
(Chang and Trivedi 2003).
Given widespread empirical evidence on the beneficial impacts of health insurance on
the insureds wellbeing, there is anecdotal evidence and reasons for under-utilization of health
insurance cards in Vietnam. For example, in their group based study, Axelson et al. (2005)
report that some insured do not use their HCFP cards because they do not know of their
existence or how to use them. In a recent report, the World Bank (2007) also shows that the
insured are not fully aware of the benefits they are entitled to. It also points out that insured
patients, under a health care system where it is common practice to make informal payments

4
Vietnam health care system has four levels: commune health center, district hospital, provincial hospital
and central hospital.

89
to get better and faster treatment, perceive that they receive less attention and lower quality of
treatment than patients paying on a private basis (Sepehri et al. 2005). Therefore, the lack of
information, perceived low quality of treatment and cumbersome procedures explain why the
insured in many cases do not use their cards when receiving treatment (World Bank 2007).
Our data from the VHLSS06 confirm this. The reasons for not using health insurance
cards are given in Figure 1. Figure 1 shows that 23 per cent of the insured do not use their
health insurance cards to claim for treatment. Among the unused, only 16 per cent say they do
not use the card because of their ineligibility, such as the insurance card is not available (7 per
cent) or health insurance is not reimbursed at the health center (9 per cent)). Interestingly, 34
per cent of the unused say that poor quality (11 per cent) and cumbersome procedures (22 per
cent) when using the health insurance cards are the main reasons why they do not use the
cards. Three per cent of the unused say that they do not know how to use the cards. About half
(48 per cent) of the unused give other reasons for not using the health insurance cards. Most
patients report not using the cards for reasons other than ineligibility which gives further
support to our attempt to examine which economic factors drive them not to use the cards.
[Figure 1 about here]

3. Economic Framework
Our theoretical framework for this discussion is based on consumer theory. A typical
consumer (now a patient) seeks to maximize their utility which depends on private
consumption and health status. The patient is endowed with an income and a health insurance
card (taken for granted). The insured patient goes to a care provider to seek medical treatment.
The benefit from consuming medical care is the improvement in health and the cost of medical
care is the decrease in the expenditure on other private goods. At the care facility, the insured
must choose between two payment methods: the health insurance card or their own money.
Each alternative offers an expected quality for a given price, where the price includes both
monetary payment and access costs such as the opportunity cost of waiting time. With the
income constraint at hand, the patient will take all possible costs and benefits associated with
each payment method into consideration before choosing the alternative that yields the highest
utility.
Formally, let the expected utility, conditional on receiving care by payment method j, be
given by:
) , (
j j j
C H U U = (1)
in which
j
H is expected health status after receiving care services from payment method j ,
j
C is private consumption net of the expenditure on care services from payment method j .
The method of payment j takes two values: 0 = j if the insured use the health insurance card
to pay for their treatment and 1 = j if they use their own money.
The treatment purchased on payment basis j is invested in health. The expected quality
of care purchased with own money is defined as the expected improvement in health over the
health the patient would expect if they used the health insurance card to pay. Let
0
H be
expected health under the health insurance payment basis. Then the perceived quality of care
received on the monetary basis is
0 1
H H Q = .
The total price for treatment includes both the direct outlays to the care provider and the
indirect costs of access (such as the opportunity cost of waiting time or information seeking
time). We assume that the patient can use the health insurance card to obtain health services

90
free of charge ) 0 (
0
= P . Using this method of payment, however, entails some indirect costs
for the patient. First, the patient has to wait to receive treatment. The waiting time reduces the
time the patient can spend on work in the market place to earn income. Second, in order to
claim for the health insurance benefits the patient needs some understanding of how the health
insurance system works. The time spent on seeking information on how to use the health
insurance card also costs the insured in terms of forgone earnings. In total, the opportunity cost
of waiting and information seeking time is w , where is the time spent on waiting and
seeking information and w is the wage rate. These indirect costs, however, are not borne by
insured individuals who do not use their health insurance cards to pay for treatment. The direct
cost for them is
1
P , which is assumed to be positive and propositional to the clinical severity
level ) ( . In particular, we assume that 0 ) ( / ) ( and 0 ) (
1 1 1
> > = P P P .
Then, budget constraints associated with the two payment methods are:
card insurance health by payment for
0
Y w C = + (2)
and
money own by payment for ) (
1 1
Y P C = + (3)
In equations (2) and (3), Y is the total income that can be earned with all available
time
5
PT. The unit price of private consumption is normalized to the unity and
j
C is assumed to
be non negative. Substituting (2), (3) and
0 1
H Q H + = into (1) we get the indirect utility
function for each alternative:
) 4 ( card insurance health by payment for ) , (
0 0
w Y H U U =
and
money own by payment for )] ( , [
1 0 1
P Y H Q U U + = (5)
We are now ready to specify the patients utility maximization problem. The patient
chooses the method of payment that yields higher utility. In particular, the insured patient
chooses to purchase care with their money ) 1 ( = j if
) , ( )] ( , [ or
0 1 0 0 1
w Y H U P Y H Q U U U + (6)
and chooses to pay by the health insurance card ) 0 ( = j otherwise. At the optimum of the
patients utility maximization problem, the patients decision whether to use the health
insurance card will be a function of various factors such as income, health status, monetary
price for treatment, wage rate, waiting and information seeking time, and quality of treatment
purchased with private money
6
PT.


5
We assume that the total available time is distributed between working time and treatment time. The time
spent on housework or leisure is not included to keep our model simple.
6
In the empirical framework, the existence of initial health ) ( 0 H and income (Y) in the insurance
utilization equation depends on the use of the utility functional forms. For example, if we assume the
patients utility function is of the form
j j j
C H U + = , both Y H and 0 will not be present in the
utilization equation. If we, however, assume
2
j j j j
C C H U + + = , only income (Y) will appear in the
utilization equation. In this study, we use the latter example.

91
4. Rate of Health Insurance Card Utilization
Our next task is to find a variable to describe the insured patients choice between the
two methods of payment. One possible candidate is a binary variable. Accordingly, for each
treatment which qualifies for insurance benefits, the value of one is assigned to the binary
variable if the insured patient chooses to pay with the health insurance card and zero
otherwise. This approach, however, is only appropriate for the case where the patient has
exactly one treatment over the study period. If the patient has more than one treatment in the
study period then this approach is inappropriate.
In this paper, we propose to use the rate of health insurance utilization to describe the
insureds choice between the two payment methods. The health insurance utilization rate is
calculated as the ratio of the number of times the insured use their health insurance card to
receive treatment ) (t over the number of treatments which are eligible for insurance
benefits ) (n . In mathematical form, the formula for the health insurance utilization rate for
each individual is:
n
t
r = (7)
Equation (7) is used to calculate the rate of utilization for the insured who receive at
least one treatment during the study period. From equation 7, the utilization rate is bounded
between zero and one. The higher r indicates the higher proportion of treatment in which
health insurance cards are used to pay for health services. We apply equation (7) to calculate
the rates of utilization in outpatient and inpatient treatment. For descriptive analysis, we also
calculate the health insurance utilization rates by types of care providers.

5. Data and Sample
Data
For our empirical analysis, we use the latest Vietnam Household Living Standard
Survey undertaken in 2006 (VHLSS06) by the Vietnamese General Statistical Office (GSO)
with technical assistance from the World Bank. The VHLSS06 is nationally representative and
consists of the most detailed questions about health insurance. In particular, the VLHSS06
provides information on how and when an individual starts having their health insurance
cards. This information enables us to distinguish among various types of health insurance
schemes and also calculate the length of stay in the schemes for each individual. Another
attractive feature of the VHLSS06 is that it provides information on health care utilization (as
measured by inpatient and outpatient treatment) for various types of care providers. Then for
each type of treatment (inpatient or outpatient) and each type of care provider, the number of
times the individual uses their health insurance cards for treatment is also listed. The
availability of information on the number of treatments and number of times the patient uses
their health insurance card for treatment thus enables us to construct a utilization ratio as our
main dependent variable. Most importantly, the purpose of the visit to the health facility is also
reported for each type of care facility. The information on the purpose of visits is critical to
our analysis. We use this information in combination with information on why the insured did
not use their health insurance cards to pay for their treatment to determine whether a visit is
entitled for claim (we discuss this further in the eligibility section).
In addition to the detailed information on health insurance and health care utilization,
the VHLSS06, like other living standards surveys, provides demographic, social, economic
and residential information, which are important in testing our model.

92
Eligibility
It is crucial in our study to identify which visit is eligible for insurance benefits. The
VHLSS06 allows us to identify whether a visit to a health facility is qualified to be covered by
the insurer in two ways. First the VHLSS06 provides the purpose of the visit for each type of
provider. Four main reasons are listed: vaccination; pregnancy test, birth control, abortion or
delivery; checkup and consulting; and treatment. Under the current health insurance policy,
the first three kinds of visits are not covered by the schemes, therefore, we exclude those visits
from our calculation of the number of visits
7
PT. As a result, our calculation of the number of
visits to health facilities includes only visits for treatment which are qualified to be covered by
the health insurer. Second, the VHLSS06 asks why the insured did not use their health
insurance cards to claim for treatment (see Figure 1). Some individuals report that they did not
use health insurance cards because the health insurance is not reimbursed at the health facility
or the health insurance card is not available
8
. We also exclude those individuals from our
sample to make sure that all treatments included in the total number of treatments are eligible
for insurance benefits.
Sample
To test the models of health insurance utilization, we restrict our sample to individuals
with at least one type of health insurance card. Children aged under seven are excluded from
the sample because they are exempted from hospital fees even if they are not holding
insurance cards
9
PT. As we are interested in why the insured do not want to use their health
insurance cards to receive treatment, we restrict our sample further by excluding those who
have not had any (inpatient or outpatient) treatment in the 12 months prior to the survey. As
discussed in the eligibility section, a number of the insured report that they do not use their
cards because they are not eligible to. To make sure that all treatments included in our
calculation of utilization rate qualify for insurance benefits, we exclude them from the sample.
Finally, we have a sample of insured patients who are older than six years of age and who are
entitled to insurance benefits for all their treatments. With these restrictions, we have different
sample sizes to study the utilization rates for outpatient and inpatient treatments. In particular,
the respective sample sizes for utilization models of outpatient and inpatient treatment are
5,105 and 1,322.

6. Prevalence of Under-Utilization
Using our indicator of health insurance utilization, we see that the insured do not use
their health care cards to pay for all treatment as the utilization rates are not always equal to
unity (see Table 2). On average, patients use their health insurance cards about six out of ten
times they visit health facilities for outpatient treatment. The utilization rate is higher for
inpatient treatment as the insured present the card eight out of ten times they visit their care
provider. Some 32 and 15 per cent of patients respectively do not use their health insurance

7
Expenditures related to pregnancy tests and deliveries are covered for the insured while those for birth
control and abortions are not. As the VHLSS06 does not enable us to distinguish between these two types of
visits, we exclude those visits from our calculation.
8
However, for those who report not using health insurance cards because of their ineligibility, their
utilization rates are not zero. Their utilization rates for outpatient and inpatient treatment are 4 and 5 per
cent, respectively. It is likely that only a fraction of their all visits is ineligible for reimbursement.
9
By law, children aged under seven are provided a compulsory health insurance card by the government.
However, our data show that 16 per cent of children aged under seven do not have any kind of health
insurance cards. According to a regulation by the Ministry of Health, they can still receive health services
for free at public facilities on the condition that they presents equivalent papers such as birth certificates
(World Bank, 2007)

93
cards for outpatient and inpatient treatment at all. In contrast, respective 56 and 82 per cent of
patients use their health insurance cards every time they receive outpatient and inpatient
treatment.
[Table 2 about here]
Regarding the utilization rate across types of care providers, Table 2 indicates that the
insured have the highest utilization rates at district hospitals, then commune health centers and
provincial hospitals. Of all health care facilities (including public and private), the lowest
utilization rate is observed at private clinics/hospitals. This is expected as at the time of the
survey, the participation of the private sector to provide health services for insured patients
was still limited.
Of the public health facilities, central hospitals have the lowest utilization rate. This can
be explained by the highest skip rate for this level of the health care system. The central
hospitals are supposed to provide services for patients with clinically serious conditions and
are perceived as of high quality. Insured patients, regardless of their level of clinical condition,
may be willing to skip their primary care providers to receive better treatment in the central
hospitals (World Bank 2007). In these cases, the insured are not entitled to use their insurance
cards
10
PT. The utilization rate thus is lowest at the central hospitals of all public health care
facilities. It is also interesting to observe that while the difference in utilization rates between
central hospitals and other public care facilities in inpatient treatment is low (ranging from 0.5
to .11), and that for outpatient treatment is much larger (ranging from 0.25 to 0.36). This
difference indicates that at central hospitals outpatients have higher skip rates than inpatients.
By type of treatment, the insured have higher utilization rates in inpatient treatment
(0.83) than in outpatient treatment (0.59). The higher rates of utilization in inpatient treatment
is also predicted by our model as these kinds of treatment are usually required for patients with
more clinically serious conditions and are more expensive than outpatient treatment.
We now look at the utilization rates across various health insurance schemes. As
discussed in the review section, the benefits seem to be the same among all health insurance
schemes. However, Figure 2 shows that the utilization rates are not homogenous across
schemes. The highest utilization rate in inpatient treatment (0.91) is observed for patients with
required health insurance acquired from working for the state sector. The insured under the
policy beneficiaries and non-state sector schemes have the second highest utilization rates
(0.89 and 0.87, respectively). Interestingly, the lowest utilization rates in both outpatient and
inpatient treatment are attributed to the insured under the voluntary basis (student and non-
student). The low utilization rate for insured students is understandable as they are young and
healthy. The low utilization rate for patients with voluntary insurance, however, raises concern
over a claim that they were a source of deficit of the Social Insurance Fund in late 2006 and
early 2007 (World Bank 2007). In our case, the low utilization rate for this group can be partly
explained by the time of the survey. The VHLSS06 was carried out during May and
September 2006 so collected information referring to the last 12 months prior to the survey
time was in late 2005 and early 2006. In late 2005 and early 2006, the Fund was still in
surplus. This time period may also be not long enough for the insured under the voluntary
scheme to fully utilize the insurance benefits.
[Figure 2 about here]

10
The insured are required to register a health facility, usually at a low level, as their original care provider.
Most of the insured (around 85 per cent) choose health facilities other than provincial or central hospitals as
their primary care providers. Unless in case of emergency, they are required to start at their primary care
centers before being transferred to health facilities at higher level.

94
Figure 3 represents the rates of utilization across income (as represented by expenditure)
groups. The insured in the poorest group have the highest utilization rate as they use their
insurance cards for 71 and 86 per cent of their outpatient and inpatient treatment, respectively.
In contrast, the lowest utilization rate is observed for the insured in the richest group who use
their card to pay for 57 and 81 per cent of their outpatient and inpatient treatment,
respectively. Taking the near-poor group as an exception, it seems that the utilization rate
decreases as income increases. However, this negative relationship between income and
utilization rate seems to be more visible for outpatient rather than inpatient treatment.
[Figure 3 about here]

7. Empirical Model
Econometric Method
In our model, the dependent variable is fractional and bounded between zero and one.
The quasi-likelihood method is chosen over the Ordinary Least Squared (OLS) or log-odds
ratio models
11
PT. The shortcoming of the OLS regression in dealing with a fractional dependent
variable as the predicted values from this method can lie outside the unit interval. Another
common model named log-odds ratio is also inappropriate for our case. First, the log-odds
ratio model cannot be applied to the case where the dependent variable takes on the value 0 or
1 with positive possibility. Second, even if the log-odds ratio model is well defined, we still
have a problem with recovering the expected value of our original variable which is our
primary interest. In this paper we use a Bernoulli (binomial) quasi-maximum likelihood
method proposed by Papke and Wooldridge (1996) to estimate the model of the factional
dependent variable. The Bernoulli quasi-maximum likelihood model is the most appropriate
for the characteristics of our dependent variable as it ensures that the expected values will not
be out of the unit range. In addition, it supports the case where the dependent variable takes
values zero and one with positive possibility.
Other econometric issues involve sampling weights and clustering. First, since the
VHLSS06 was over-sampled in certain areas, it requires the use of sampling weights to
calculate representative statistics
12
PT. Second, the VHLSS06 uses a three stage stratified cluster
sampling method with commune as the primary sampling unit. Clustering raises the possibility
of intra-commune correlation and hence results in smaller estimated variance of the
coefficients. In this study, to deal with the effects of clustering we calculate the robust
standard errors of parameter estimates adjusted for clustering. We also calculate and report the
marginal effects after regressions. The model is estimated by the Maximum Likelihood
method.
Empirical specifications
In this paper, the dependent variable is the health insurance utilization rate. The
explanatory variables, as suggested by the theoretical framework, include income, health
status, monetary price for treatment, wage rate, waiting and information seeking time, and
quality of treatment purchased with private money.
In particular, the income (represented by the household expenditure levels) variable is
included in the regressions in a spline form (five quintiles with the first quintile the poorest

11
The log-odds ratio model transforms the original dependent variable (for example, r) into )] 1 /( log[ r r
and then apply a linear model (such as the OLS) to estimate this log-odds ratios as a dependent variable (see
Maddala (1993) for more detail).
12
Unless otherwise indicated, all figures calculated in this study are adjusted for sampling weights.

95
group as the base) to gauge for possible heterogeneity in income elasticity among income
groups.
To control for health status of patients we use a dummy variable indicating whether the
individual had any illness in the last 12 months prior to the survey. The level of clinical
severity is presented by the number of days stay in bed due to illness in the last 12 months. We
also use a dummy variable to indicate if the individual has any chronic disease, to measure the
impact of long-term disease on insurance utilization. For female patients, a dummy variable
indicating pregnancy status is also included. The variables representing health status also
proxy for the price of health care as more clinically serious conditions require more expensive
treatments
13
PT. The price of treatment is also measured by type of health insurance schemes. The
insured, under different health insurance schemes, may be entitled to different benefits. The
insured, under a more generous insurance scheme, may find it more costly to use their own
money to pay for care services. In addition to heterogeneity in the defined benefits among
different schemes, the insured may have dissimilar reasons to enroll in those schemes.
Therefore, we expect the utilization rates to be varied between schemes. Accordingly, six
types of health insurance schemes are included (with the free health care card defined as the
base group)TP
14
PT
In the theoretical framework we assume that the insured have to pay some indirect
access costs (such as information seeking or waiting time) when using the health insurance
card. However, both information seeking and waiting time is not directly observed in our data.
To measure the impact of information seeking time on the insureds choice of the payment
method we propose to use the patients endowed information about the insurance scheme. We
assume that the insured, with more endowed information about the scheme, can receive more
benefits or the same benefits but at a lower cost (such as information seeking) than less
informed ones. To measure the importance of information in health insurance utilization we
use the length of stay in the scheme, acquired education levels and whether the insured receive
any instruction on how to use the card, in the regressions. Household assets such as television
or radio that may help to provide the insured more information about the use of the card are
also included. We believe that these information variables are also appropriate proxy for
waiting time since insured patients with more information about the insurance scheme are
expected to complete procedures more efficiently.
In the indirect utility function (5), the quality of treatment is not observed. Following
Dor et al. (1987) and Gertler et al. (1987), we assume that quantity of treatment purchased on
a monetary basis is a function of its observable determinants. The observable factors include
provider characteristics and individual characteristics.

13
Another possible proxy for price of health care is the average out-of-pocket (OOP) expenditure for each
inpatient or outpatient treatment. As using health insurance may have a resultant impact on reducing the
amount of OOP expenditures, we do not include these variables in the regression. The information on
whether the individual has to pay additional fees when using health insurance for diagnosis/treatment is also
available. Unfortunately, this information is not available for the insured who report not using their cards for
treatments. As a result, we do not include this variable in the regressions either.
14
In the VHLSS06, there are 121 individual reporting that they have both required and voluntary health care
insurance. Most of them (80 per cent) received their compulsory health insurance from working for the
public sector, private/foreign invested enterprises or health care for the poor Fund (where they are defined
as poor or people with merit). In addition, a majority of those with required health insurance (72 per cent of
121 persons) acquired their voluntary health insurance under the association, commune or family (of those
with compulsory insurance ) basis. For these people, we experimented by adding a dummy variable (=1 if
have additional voluntary health insurance cover and = zero if otherwise) to capture any impact of the
additional cover. As the estimates for this variable are insignificant in both outpatient and inpatient
regressions, we exclude it from our final regressions.

96
As the utilization rates may be different across care providers, we propose to use the
ratios of visit for various health care providers to measure this impact. For each patient, the
ratio of visit to a particular care provider ) (
k
v is calculated as the proportion of the number of
visits to that care provider ) (
k
n over the total number of visits ) (n . The mathematical formula
for the ratio of visits is:
n
n
v
k
k
= (8)
where, k denotes type of care provider. The ratio of visits is calculated for individuals with at
least one treatment (inpatient or outpatient) over the study period. The ratio of visits thus
represents the patients choice over various care providers. Like the ratio of utilization, the
ratio of visits is also bounded between zero and one. In addition, the sum of the ratios of visits
of all care providers equals unity.
We also include gender, age (and its square), marital status, educational levels to
measure the impact of individual characteristics on quality of treatment purchased on a
monetary basis. Family background (household size and ethnicity) and location (rural/urban or
eight regions) are also included to examine the impact of the living environment on utilization
rates
15
PT.
Definitions of the variables are presented in Table 3. Sample statistics are provided in
Table 4. Table 4 shows that most of the insured are under the student, required state, free
health care card, health care for the poor and voluntary schemes
16
PT. On average, the insured
have three or four years staying in their schemes for outpatient and inpatient treatment sample,
respectively. Each patient, conditional on positive treatment, spends 3.63 and 1.43 times in
outpatient and inpatient treatment, respectively. Each inpatient treatment costs the insured
around one million VND (equivalent to USD 62), nine times as much as the cost of one
outpatient treatment. For outpatient treatment, patients most often go to commune health
centers (31 per cent of their visits), district hospitals (26 per cent) and private hospitals (20 per
cent). In contrast, for inpatient treatment, the most often visited care providers are district
hospitals (44 per cent), provincial hospitals (37 per cent) and commune health centers (16 per
cent). Interestingly, while private care providers are one of the most popular destinations for
outpatient treatment, they are least preferred for patients seeking inpatient treatment as only 2
per cent of inpatient treatments are carried out here.
[Tables 3 and 4 about here]

8. Empirical Results
The regression results for the main variables are reported in Table 5. The results are
reported separately for outpatient and inpatient treatment. In this section, however, we
concentrate on the impact of type of insurance, type of care providers, income, information
and health status on the utilization rates for both types of treatment.


15
We experimented with the inclusion of household living environment that may have impact on the health
insurance utilization such as types of the dwelling, main sources of cooking or drinking water, types of
toilets in the utilization equations. These variables turn out insignificant. Behavior affecting health as
measured by ever smoking is also included but insignificant. We, therefore, drop them from our final
regressions
16
In our sample the ratio of the insured with voluntary scheme is higher than average because our sample
only includes those with at least one treatment. As insured under the voluntary scheme may consume more
care services than the insured in other schemes, the former is likely over-represented in our sample.

97
[Table 5 about here]
Regarding the type of insurance, the regression results for outpatient treatment show that
the insured under the student insurance scheme have the lowest utilization rate of all. Other
things being equal, the utilization rate of the insured under the student voluntary scheme is 14
percentage points lower than that of the base group (the insured under the free health care card
scheme). As mentioned in the descriptive analysis, the lowest rate of utilization for insured
under the student health insurance scheme is expected, as they are young and healthy. In
contrast, we find no significant difference in the utilization rates in outpatient treatment among
other insured groups and the base group.
For inpatient treatment, the insured under policy or required state health insurance
schemes have the highest utilization rates as their respective utilization rates are 8 or 9
percentage points higher than that of the base group (which is also found to have the same
utilization rate as those of the remaining groups). The higher rate of utilization of the insured
under the policy or required state health insurance schemes can be attributed to the more
generous benefits associated with these schemes. It is also possible that the insured under these
schemes have better relations with their care provider (which are mainly in the public sector)
so they can receive good services even if they use health insurance cards to pay for the
services.
With respect to type of care provider, Table 5 shows that district hospitals have the
highest utilization rate in outpatient treatment. A positive estimate for district hospitals means
that if patients go to the district hospitals more often, they are likely to use their cards more
often. In particular, if patients increase their rate of visits to the district hospital by 1 per cent
(or 0.01) their utilization rates are expected to increase by 12 percentage points. In contrast,
the lowest utilization rate is observed in private care facilities as the utilization rate is expected
to decrease by 96 percentage points for a 1 per cent increase in the rate of visits to this care
provider. Private care providers also have the lowest utilization rate in inpatient treatment as a
1 per cent increase in the rate of visit to these care providers leads to a drop of 23 percentage
points in the utilization rate.
We now analyze how income affects the insureds behavior in using the health
insurance card for treatment. Taking the insured in the average expenditure group as an
exception, the estimation results suggest that the wealthier tend to use the health insurance
cards less often for outpatient treatment. In particular, the insured in the near poor or better-off
groups use health insurance cards 8 or 9 percentage points fewer than the ones in the poorest
group. In addition, the insured in the richest group have the lowest utilization rate where their
utilization rate is 14 percentage points lower than that of the insured in the poorest group. This
negative relationship between income and the use of the health insurance card suggests that
using the health insurance card to pay for outpatient treatment is considered as an inferior
good. Therefore, this finding is consistent the insureds perception that treatment paid with the
health insurance card is of low quality (as discussed in Table 1 or in the World Banks (2007)
study). We find no difference in the utilization rates across income groups in inpatient
treatment. Since inpatient treatment is more expensive than outpatient treatment, the
insignificant impact of income on utilization rates in inpatient treatment suggests that price
effect may outweigh income effect. As a result, like the poor, the wealthy have to rely on
health insurance to pay for their inpatient treatment as well.
Regarding the role of information, our estimates place the importance of information on
the utilization of health insurance. For example, one more year staying in the scheme is found
to increase the insureds utilization rate by 1 percentage point. The importance of information
is also found as the utilization rates of the insured who receive instruction on how to use the
card are 9 percentage points higher than that of the ones who do not. Having broadcasting
devices such as television or radio in the household also increases the utilization rate of

98
household members. However, these variables are never statistically significant. Similarly, all
variables representing information in our model are found to positively but insignificantly
influence utilization rates in inpatient treatment. To the insignificant impact of information on
the utilization rate in inpatient treatment, it can be inferred that information is more important
for the use of the health insurance card in outpatient treatment than in inpatient treatment.
The impact of health status on utilization rates is as expected: patients with poor health
use their cards more often. In particular, people who had any illness during the last 12 months
use their cards 4 and 5 percentage points more often than healthy ones in inpatient and
outpatient treatment, respectively. Additionally, the utilization rates of individuals with
chronic diseases are 4 percentage points higher than that of those without. This holds for both
inpatient and outpatient treatment. Females who are currently pregnant or have been in the last
12 months are also found to have higher utilization rates (7 percentage points higher) than
other individuals.
The level of education is found to have a positive impact on health insurance utilization.
For example, in outpatient treatment, university graduates are found to use their card 14
percentage points more often than illiterate individuals. However, in inpatient treatment, the
insured with upper secondary degrees have 6 percentage points in the utilization rates than
their illiterate counterparts. The higher utilization rate for insured with higher education levels
can be attributed to their higher capacity to exploit the insurance system.
[Table 6 about here]
Other variables have their expected signs (Table 6). In outpatient treatment, Kinh or
Chinese ethnic groups are found to use their cards less often (8 percentage points lower) than
other minority ethnic groups. The lower utilization rate among Kinh and Chinese ethnic group
may be linked to their higher income. Similarly, in inpatient treatment, the insured who are
currently at school have the higher utilization rate (9 percentage points higher) than their
counterparts not at school. The higher rates of utilization for school students again can be
explained by their low income and thus low opportunity cost of waiting for treatment. Males
are found to have 3 percentage points of utilization rate higher than females in outpatient
treatment. Our finding of a gender difference in the utilization rate is consistent with the
observation that males have more negotiation power than females. This is also in line with the
study of Doerpinghaus et al. (2003) who found females to be more risk-averse and less
aggressive negotiators than males. The location of residence is also found to have some impact
on utilization rates. For instance, insured patients living in urban areas have 13 percentage
points higher utilization rates in outpatient treatment than their counterparts in rural areas. This
significant urban/rural difference in utilization rate in favor of urban residents can be
explained by the better access to health care and health insurance of urban residents than rural
ones
17
PT. Other demographic variables of the insured such as age and marital status, however,
are not found to have any significant impact on their utilization rates.

9. Conclusion and Policy Implications
This paper examines factors determining the insureds choice in not using their health
insurance cards for treatment. The theoretical model predicts that income, health status,
monetary price for treatment, access costs such as opportunity costs of waiting or information
seeking time and quality of treatment are the main determinants of health insurance utilization.

17
We also test for the impact of the supply side on the utilization of health insurance card by including the
distance from commune to the nearest district or provincial hospital. Data for distance are taken from
commune questionnaire and only available for communes in rural areas. These variables are not statistically
significant. The results , therefore, are not reported here for brevity

99
The empirical analysis shows that under-utilization in health insurance does exit. Barriers such
as low quality of service, lack of necessary information, and mistrust of the health care system
prevent people from using the cards. These reasons are closely related to our models
implications.
The regression results also support our models hypotheses. In particular, utilization
rates vary across types of health insurance schemes. The insured under the student scheme
have the lowest utilization rate in outpatient treatment. In inpatient treatment, the insured
under policy and required state schemes have the highest utilization rates. The utilization rates
are also varied by type of care provider. In outpatient treatment, the highest and lowest
utilization rates are found in district hospitals and private hospitals, respectively. Private
hospitals also have the lowest utilization rate in inpatient treatment.
There is good and bad news from our estimations on the impact of income and health
status on health insurance utilization. The good news is the poor and ill outpatients use health
insurance cards more often than the rich and healthy ones. This suggests that health insurance
is more beneficial for the poor and the ill than the rich and the healthy. In addition, there is no
difference in utilization rates among income groups in inpatient treatment. This indicates that
wealthy people still rely on health insurance for more expensive treatment. The bad news is
that using health insurance to pay for outpatient treatment is considered as an inferior good.
This finding supports the claim of a bias in treatment against the insured. This paper argues
that this bias in treatment is an important explanation of why the insured find it optimal to stay
away or drop out of the scheme. Therefore, if the purpose of the policy is to increase health
insurance coverage on the voluntary basis, the quality of health services, especially for the
insured, should be improved. This policy implication seems to be more important for Vietnam
as currently the countrys health insurance system is facing two problems: low coverage and
large deficit (World Bank 2007). An improvement in health care services would help to
increase the coverage and more importantly to motivate the healthy and wealthy population to
join the program. By doing so, the risk-sharing basis, an important requirement of the
sustainability of the health insurance system, would be achieved.
Our estimation results also highlight the importance of offering information to the
insured. In the developing world, where most of the insured enroll on an involuntary basis,
their understanding about the scheme is limited, so the role of information is even more
important.

REFERENCES

Arrow, K.J., 1963, 'Uncertainty and the Welfare Economics of Medical Care', The American
Economic Review, 53(5): 941-973.
Axelson, H., Cuong, D.V., Phuong, N.K., Oanh, T.T.M., Luong, D.H. and Tuan, K.A., 2005,
'The Impact of the Health Care Fund for the Poor on Poor Households in Two Provinces
in Viet Nam', paper presented to the Global Forum for Health Research 9, Mumbai,
India, 12-16 September 2005.
Buchmueller, T.C., Couffinhal, A., Grignon, M. and Perronnin, M., 2004, 'Access to physician
services: does supplemental insurance matter? Evidence from France', Health
Economics, 13(7): 669-687.
Cameron, A.C. and Trivedi, P.K., 1991, 'The role of income and health risk in the choice of
health insurance: Evidence from Australia', Journal of Public Economics, 45(1): 1-28.

100
Cameron, A.C., Trivedi, P.K., Milne, F. and Piggott, J., 1988, 'A Microeconometric Model of
the Demand for Health Care and Health Insurance in Australia', The Review of
Economic Studies, 55(1): 85-106.
Chang, F.-R. and Trivedi, P.K., 2003, 'Economics of self-medication: theory and evidence',
Health Economics, 12(9): 721-739.
Cotton, M.L. and Feigenbaum, B., 1984, 'Rationing by Waiting Lists', The American
Economic Review, 74(3): 404-417.
Doerpinghaus, H., Schmit, J. and Jason Jia-Hsing, Y., 2003, 'Personal Bias in Automobile
Claims Settlement', The Journal of Risk and Insurance, 70(2): 185-205.
Dor, A., Gertler, P. and van der Gaag, J., 1987, 'Non-price rationing and the choice of medical
care providers in rural Cote d'Ivoire', Journal of Health Economics, 6(4): 291-304.
Ekman, B., Liem, N.T., Duc, H.A. and Axelson, H., 2008, 'Health insurance reform in
Vietnam: a review of recent developments and future challenges', Health Policy and
Planning, forthcoming.
Ensor, T. and Witter, S., 2001, 'Health economics in low income countries: adapting to the
reality of the unofficial economy', Health Policy, 57(1): 1-13.
Gertler, P., Locay, L. and Sanderson, W., 1987, 'Are user fees regressive? : The welfare
implications of health care financing proposals in Peru', Journal of Econometrics, 36(1-
2): 67-88.
Hoel, M. and Sther, E.M., 2003, 'Public health care with waiting time: the role of
supplementary private health care', Journal of Health Economics, 22(4): 599-616.
Hofter, R.H., 2006, 'Private health insurance and utilization of health services in Chile',
Applied Economics, 38(4): 423-439.
Hugh, R.W., 1999, 'Measuring the impact of health insurance with a correction for selection
bias - a case study of Ecuador', Health Economics, 8(5): 473-483.
Jowett, M., 2004, Theoretical Insights into the Development of Health Insurance in Low-
income Countries, Centre for Health Economics, University of York Discussion paper #
188, <HTUhttp://ideas.repec.org/p/chy/respap/188chedp.html>UTH.
Jowett, M., Contoyannis, P. and Vinh, N.D., 2003, 'The impact of public voluntary health
insurance on private health expenditures in Vietnam', Social Science & Medicine, 56(2):
333-342.
Jowett, M., Deolalikar, A. and Martinsson, P., 2004, 'Health insurance and treatment seeking
behaviour: evidence from a low-income country', Health Economics, 13(9): 845-857.
Kamuzora, P. and Gilson, L., 2007, 'Factors Influencing Implementation of the Community
Health Fund in Tanzania', Health Policy and Planning, 22(2): 95-102.
Maddala, G.S., 1983, Limited-dependent and Qualitative Variables in Econometrics,
Cambridge University Press, Cambridge.
Manning, W.G. and Marquis, M.S., 1996, 'Health insurance: The tradeoff between risk
pooling and moral hazard', Journal of Health Economics, 15(5): 609-639.
Manning, W.G., Newhouse, J.P., Duan, N., Keeler, E.B., Leibowitz, A. and Marquis, M.S.,
1987, 'Health Insurance and the Demand for Medical Care: Evidence from a
Randomized Experiment', The American Economic Review, 77(3): 251-277.
Marquis, M.S. and Holmer, M.R., 1996, 'Alternative Models of Choice Under Uncertainty and
Demand for Health Insurance', The Review of Economics and Statistics, 78(3): 421-427.

101
Mechanic, D., 1994, 'Trust and Informed Consent to Rationing', The Milbank Quarterly, 72(2):
217-223.
Pagn, J.A., Puig, A. and Soldo, B.J., 2007, 'Health insurance coverage and the use of
preventive services by Mexican adults', Health Economics, 16(12): 1359-1369.
Papke, L.E. and Wooldridge, J.M., 1996, 'Econometric Methods for Fractional Response
Variables With an Application to 401 (K) Plan Participation Rates', Journal of Applied
Econometrics, 11(6): 619-632.
Pradhan, M., Saadah, F. and Sparrow, R., 2007, 'Did the Health Card Program Ensure Access
to Medical Care for the Poor during Indonesia's Economic Crisis?' World Bank
Economic Review, 21(1): 125-150.
Schneider, P., 2004, 'Why should the poor insure? Theories of decision-making in the context
of health insurance', Health Policy Plan., 19(6): 349-355.
Sepehri, A., Chernomas, R. and Akram-Loghi, H., 2005, 'Penalizing patients and rewarding
providers: user charges and health care utilization in Vietnam', Health Policy and
Planning, 20(2): 90-99.
Sepehri, A., Simpson, W. and Sarma, S., 2006a, 'The Influence of Health Insurance on
Hospital Admission and Length of Stay - the Case of Vietnam', Social Science &
Medicine, 63(7): 1757-1770.
Sepehri, A., Sarma, S. and Simpson, W., 2006b, 'Does non-profit health insurance reduce
financial burden? Evidence from the Vietnam living standards survey panel', Health
Economics, 15(6): 603-616.
Sinha, T., Ranson, M.K., Chatterjee, M., Acharya, A. and Mills, A.J., 2006, 'Barriers to
accessing benefits in a community-based insurance scheme: lessons learnt from SEWA
Insurance, Gujarat', Health Policy and Planning, 21(2): 132-142.
Wagstaff, A., 2007a, 'The economic consequences of health shocks: Evidence from Vietnam',
Journal of Health Economics, 26(1): 82-100.
Wagstaff, A., 2007b, Health Insurance for the Poor: Initial Impacts of Vietnam's Health Care
Fund for the Poor, World Bank Policy Research Working Paper number 4134, World
Bank, Washington DC.
World Bank, 2007, Vietnam Development Report 2008: Social Protection, Joint Donor Report
to the Vietnam Consultative Group Meeting, Hanoi, December 6-7, 2007.



104
Table 1: Percentage of people with health insurance by household expenditure quintile
2006 2004
Expenditure quintile
Types of health
insurance
Poor
Near
poor
Averag
e
Better-
off
Rich
Total

Total
Free 31.7 13.8 7.4 6.0 4.3 12.6 8.2
Child 5.3 3.5 2.9 2.5 1.7 3.2
Poor 15.6 6.7 3.7 1.8 0.5 5.7 4.2
Policy 3.0 2.8 2.5 3.1 3.1 2.9 3.0
State-required 3.1 3.4 4.9 9.2 20.9 8.3
Non-state required 0.2 0.4 1.2 1.7 3.2 1.3
5.8P
(a)P
Student 6.5 14.3 16.6 19.3 18.3 15.0 15.0
Voluntary 1.0 2.9 4.2 5.9 7.2 4.2 1.4
Total 66.3 47.9 43.4 49.4 59.2 53.3 37.5
otes: - Population means weighted to reflect sampling weights.

(a)
required health insurance (including state and non-state)
Source: Own-calculation from the VLHSS 04 and 06.



105
Figure 1: Why not use the health insurance card (percent of 18640 insured)
No diagnosis/t reat ment
43%
Dont know how t o use
1%
Healt h insurance card
not available t hen
2%
Healt h insurance not
reimburse at healt h
facilit y
2%
Lower qualit y when
using healt h insurance
card
3%
Ot hers
11%
Cumbersome procedures
5%
Not use
23%
Use
33%

Source: Own-calculation from the VHLSS06.


106
Table 2: Health insurance utilization rates by types of treatments and health care providers
Proportion (%) with
Utilization rate Mean
Number of
observations
(a)P


R = 0 R = 1
Out-patient treatments (all providers) 0.59 5105 32 56
Commune health center 0.76 1892
District hospital 0.83 1581
Provincial hospital 0.72 878
Central hospital 0.47 152
Private clinic/hospital 0.03 1170
Others 0.16 290

In-patient treatments (all providers) 0.83 1322 15 82
Commune health center 0.83 121
District hospital 0.87 631
Provincial hospital 0.81 507
Central hospital 0.76 116
Private clinic/hospital 0.48 15
Others 0.62 21
otes: (a) number of insured individuals with at least one treatment in the corresponding care provider.
- Means are adjusted for sampling weights.
Source: Own-calculation from the VHLSS06.

107
Figure 2: Health insurance utilization rates by health insurance schemes
0.69
0.62 0.62
0.66
0.51
0.47
0.58
0.59
0.83 0.83
0.89
0.91
0.87
0.73
0.75
0.83
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Free Poor Policy State-
required
Non-state
required
Student Voluntary All
U
t
i
l
i
z
a
t
i
o
n

r
a
t
e
Out-patient In-patient

Source: Own-calculation from the VHLSS06.


108
Figure 3: Health insurance utilization rates by expenditure groups
0.71
0.57
0.62
0.57
0.52
0.59
0.86
0.82
0.83
0.84
0.81
0.83
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Poorest Near poor Average Better-of f Richest All
U
t
i
l
i
z
a
t
i
o
n

r
a
t
e
Out-patient In-patient

Source: Own-calculation from the VHLSS06.


1
0
9

T
a
b
l
e

3
:

V
a
r
i
a
b
l
e

D
e
f
i
n
i
t
i
o
n
s

V
a
r
i
a
b
l
e

n
a
m
e

V
a
r
i
a
b
l
e

d
e
f
i
n
i
t
i
o
n
s

U
t
i
l
i
z
a
t
i
o
n

r
a
t
e

R
a
t
e

o
f

h
e
a
l
t
h

i
n
s
u
r
a
n
c
e

c
a
r
d

u
t
i
l
i
z
a
t
i
o
n

f
o
r

i
n
p
a
t
i
e
n
t

o
r

o
u
t
p
a
t
i
e
n
t

t
r
e
a
t
m
e
n
t

T
y
p
e
s

o
f

h
e
a
l
t
h

i
n
s
u
r
a
n
c
e

T
y
p
e
s

o
f

h
e
a
l
t
h

i
n
s
u
r
a
n
c
e
:

f
r
e
e

h
e
a
l
t
h

c
a
r
e

c
a
r
d

(
t
h
e

b
a
s
e

g
r
o
u
p
)
;

h
e
a
l
t
h

i
n
s
u
r
a
n
c
e

f
o
r

t
h
e

p
o
o
r
;

h
e
a
l
t
h

i
n
s
u
r
a
n
c
e

f
o
r

p
o
l
i
c
y

b
e
n
e
f
i
c
i
a
r
i
e
s
;

r
e
q
u
i
r
e
d

h
e
a
l
t
h

i
n
s
u
r
a
n
c
e

p
r
o
v
i
d
e
d

b
y

t
h
e

s
t
a
t
e
;

r
e
q
u
i
r
e
d

h
e
a
l
t
h

i
n
s
u
r
a
n
c
e

p
r
o
v
i
d
e
d

b
y

n
o
n
-
s
t
a
t
e

o
r
g
a
n
i
z
a
t
i
o
n
s
;

s
t
u
d
e
n
t

v
o
l
u
n
t
a
r
y

h
e
a
l
t
h

i
n
s
u
r
a
n
c
e
,

v
o
l
u
n
t
a
r
y

o
t
h
e
r

t
h
a
n

s
t
u
d
e
n
t

h
e
a
l
t
h

i
n
s
u
r
a
n
c
e

T
i
m
e

N
u
m
b
e
r

o
f

y
e
a
r
s

s
t
a
y
i
n
g

i
n

t
h
e

h
e
a
l
t
h

i
n
s
u
r
a
n
c
e

s
c
h
e
m
e

(
i
n

y
e
a
r
s
)

H
e
a
l
t
h

f
a
c
i
l
i
t
y

u
s
e
d

P
r
o
p
o
r
t
i
o
n

o
f

u
s
i
n
g

c
o
m
m
u
n
e

h
e
a
l
t
h

c
e
n
t
r
e

(
t
h
e

b
a
s
e

g
r
o
u
p
)
,

d
i
s
t
r
i
c
t

h
o
s
p
i
t
a
l
,

p
r
o
v
i
n
c
i
a
l

h
o
s
p
i
t
a
l
,

c
e
n
t
r
a
l

h
o
s
p
i
t
a
l
,

p
r
i
v
a
t
e

h
o
s
p
i
t
a
l

a
n
d

o
t
h
e
r

h
e
a
l
t
h

f
a
c
i
l
i
t
i
e
s

f
o
r

i
n
p
a
t
i
e
n
t

o
r

o
u
t
p
a
t
i
e
n
t

t
r
e
a
t
m
e
n
t
s

M
a
l
e

D
u
m
m
y

=

1

i
f

m
a
l
e
,

=

0

i
f

f
e
m
a
l
e

(
t
h
e

b
a
s
e

g
r
o
u
p
)

A
g
e

C
u
r
r
e
n
t

a
g
e

(
i
n

y
e
a
r
s
)

A
g
e

s
q
u
a
r
e
d

A
g
e

s
q
u
a
r
e
d

(
i
n

y
e
a
r
s

s
q
u
a
r
e
d
)

M
a
r
r
i
e
d

D
u
m
m
y

=

1

i
f

m
a
r
r
i
e
d
,

w
i
d
o
w
e
d
,

d
i
v
o
r
c
e
d

o
r

s
e
p
a
r
a
t
e
d
;

=
0

i
f

o
t
h
e
r
w
i
s
e

(
t
h
e

b
a
s
e

g
r
o
u
p
)

K
i
n
h

E
t
h
n
i
c
i
t
y

D
u
m
m
y

=

1

i
f

K
i
n
h

o
r

C
h
i
n
e
s
e
;

=

0

i
f

o
t
h
e
r
w
i
s
e

(
t
h
e

b
a
s
e

g
r
o
u
p
)

E
d
u
c
a
t
i
o
n

A
c
h
i
e
v
e
d

l
e
v
e
l
s

o
f

e
d
u
c
a
t
i
o
n
:

n
o

e
d
u
c
a
t
i
o
n

(
t
h
e

b
a
s
e

g
r
o
u
p
)
,

p
r
i
m
a
r
y
,

l
o
w
e
r

s
e
c
o
n
d
a
r
y
,

u
p
p
e
r

s
e
c
o
n
d
a
r
y
,

c
o
l
l
e
g
e
,

u
n
i
v
e
r
s
i
t
y

a
n
d

m
o
r
e

T
r
a
i
n
i
n
g

=

1

i
f

o
b
t
a
i
n
e
d

l
o
n
g
-
t
e
r
m

v
o
c
a
t
i
o
n
a
l

t
r
a
i
n
i
n
g

o
r

p
r
o
f
e
s
s
i
o
n
a
l

h
i
g
h

s
c
h
o
o
l
,

=

0

i
f

o
t
h
e
r
w
i
s
e

(
t
h
e

b
a
s
e

g
r
o
u
p
)

I
n
s
t
r
u
c
t
i
o
n

=

1

i
f

w
a
s

g
i
v
e
n

i
n
s
t
r
u
c
t
i
o
n

o
n

h
o
w

t
o

u
s
e

H
I

c
a
r
d
s
;

=

0

o
t
h
e
r
w
i
s
e

(
t
h
e

b
a
s
e

g
r
o
u
p
)

I
l
l

D
u
m
m
y

=

1

i
f

h
a
v
e

a
n
y

i
l
l
n
e
s
s

i
n

t
h
e

l
a
s
t

1
2

m
o
n
t
h
s
;

=

0

i
f

o
t
h
e
r
w
i
s
e

(
t
h
e

b
a
s
e

g
r
o
u
p
)

D
a
y
s

i
n

b
e
d
s

N
u
m
b
e
r

o
f

d
a
y
s

s
t
a
y

i
n

b
e
d

d
u
e

t
o

i
l
l
n
e
s
s

i
n

t
h
e

l
a
s
t

1
2

m
o
n
t
h
s

C
h
r
o
n
i
c

=

1

i
f

h
a
s

a
n
y

c
h
r
o
n
i
c

d
i
s
e
a
s
e
,

e
.
g
.

d
i
a
b
e
t
e
s
,

h
e
p
a
t
i
t
i
s
,

=

0

i
f

o
t
h
e
r
w
i
s
e

(
t
h
e

b
a
s
e

g
r
o
u
p
)

P
r
e
g
n
a
n
t

D
u
m
m
y

=

1

i
f

c
u
r
r
e
n
t
l
y

p
r
e
g
n
a
n
t

o
r

h
a
s

b
e
e
n

i
n

l
a
s
t

1
2

m
o
n
t
h
s
;

=

0

i
f

o
t
h
e
r
w
i
s
e

(
t
h
e

b
a
s
e

g
r
o
u
p
)

W
o
r
k

T
y
p
e
s

o
f

w
o
r
k
:

n
o
t

w
o
r
k
i
n
g

(
t
h
e

b
a
s
e

g
r
o
u
p
)
,

w
a
g
e

e
a
r
n
e
r
,

f
a
r
m
e
r
,

n
o
n
-
f
a
r
m

s
e
l
f

e
m
p
l
o
y
e
d

A
t

s
c
h
o
o
l

=

1

i
f

c
u
r
r
e
n
t
l
y

a
t

s
c
h
o
o
l

o
r

o
n

v
a
c
a
t
i
o
n
,

=

0

i
f

o
t
h
e
r
w
i
s
e

(
t
h
e

b
a
s
e

g
r
o
u
p
)

E
x
p
e
n
d
i
t
u
r
e

g
r
o
u
p

F
i
v
e

h
o
u
s
e
h
o
l
d

e
x
p
e
n
d
i
t
u
r
e

q
u
i
n
t
i
l
e
s
:

p
o
o
r

(
t
h
e

b
a
s
e

g
r
o
u
p
)
,

n
e
a
r

p
o
o
r
,

a
v
e
r
a
g
e
,

b
e
t
t
e
r
-
o
f
f

a
n
d

r
i
c
h

H
o
u
s
e
h
o
l
d

s
i
z
e

N
u
m
b
e
r

o
f

h
o
u
s
e
h
o
l
d

m
e
m
b
e
r
s

T
e
l
e
v
i
s
i
o
n

=

1

i
f

t
h
e

H
H

h
a
s

T
V
,

=

0

i
f

o
t
h
e
r
w
i
s
e

(
t
h
e

b
a
s
e

g
r
o
u
p
)

R
a
d
i
o

=

1

i
f

t
h
e

H
H

h
a
s

a
n
y

r
a
d
i
o
s
,

v
i
d
e
o

c
a
s
s
e
t
t
e

p
l
a
y
e
r
,

m
u
l
t
i
-
t
i
e
r

s
t
e
r
e
o
s
,

d
i
s
k

p
l
a
y
e
r
;

=

0

i
f

o
t
h
e
r
w
i
s
e

(
t
h
e

b
a
s
e

g
r
o
u
p
)

R
e
g
i
o
n

E
i
g
h
t

r
e
s
i
d
e
n
t
i
a
l

r
e
g
i
o
n
s
:

N
o
r
t
h
e
a
s
t

(
t
h
e

b
a
s
e

g
r
o
u
p
)
,

R
e
d

R
i
v
e
r

D
e
l
t
a
,

N
o
r
t
h
w
e
s
t
,

N
o
r
t
h

C
e
n
t
r
a
l

C
o
a
s
t
,

S
o
u
t
h

C
e
n
t
r
a
l

C
o
a
s
t
,

C
e
n
t
r
a
l

H
i
g
h
l
a
n
d
s
,

S
o
u
t
h
e
a
s
t

a
n
d

M
e
k
o
n
g

R
i
v
e
r

D
e
l
t
a

U
r
b
a
n

D
u
m
m
y

=

1

i
f

r
e
s
i
d
e
n
t
i
a
l

a
r
e
a

i
s

u
r
b
a
n
,

=

0

i
f

r
u
r
a
l

(
t
h
e

b
a
s
e

g
r
o
u
p
)


110
Table 4. Summary statistics for variables used in the regressions
Outpatient treatment Inpatient treatment
Variables Mean Std. Dev

Mean Std. Dev
Utilization rate 0.59 0.46 0.83 0.37
Number of visits 3.62 5.08 1.43 1.10
Average out-of-pocket payment (mil
VND)

113 695 1055 2514
Free health care card 0.17 0.38 0.20 0.40
Health care for the poor 0.14 0.35 0.14 0.34
Policy 0.09 0.29 0.14 0.35
Required state 0.18 0.39 0.21 0.40
Required non-state 0.02 0.16 0.01 0.12
Student health insurance 0.25 0.43 0.16 0.37
Voluntary 0.13 0.34 0.14 0.35
Time 3.30 5.59 4.07 6.12
Commune health center 0.31 0.45 0.16 0.34
District hospital 0.26 0.42 0.44 0.49
Provincial hospital 0.15 0.34 0.37 0.47
Central hospital 0.03 0.15 0.08 0.26
Private hospital 0.20 0.38 0.01 0.10
Other health facilities 0.05 0.20 0.01 0.11
Male 0.46 0.50 0.50 0.50
Age 37.39 22.62 44.25 22.70
Age squared 1910 1922 2473 2035
Married 0.60 0.49 0.71 0.45
Kinh 0.85 0.36 0.80 0.40
Primary education 0.26 0.44 0.24 0.43
Lower secondary 0.23 0.42 0.24 0.42
Upper secondary 0.12 0.32 0.12 0.33
Upper secondary 0.02 0.13 0.01 0.11
University 0.04 0.19 0.05 0.21
Training 0.08 0.27 0.09 0.28
Instruction 0.86 0.35 0.87 0.33
Ill 0.56 0.50 0.55 0.50
Number of days in beds 4.80 25.36 14.61 33.81
Chronic 0.21 0.40 0.30 0.46
Pregnant 0.01 0.12 0.01 0.12
Wage-earner 0.25 0.43 0.22 0.41
Farmer 0.34 0.47 0.36 0.48
Non-farm self employed 0.11 0.31 0.09 0.29
At school 0.32 0.47 0.21 0.41
Near poor 0.17 0.38 0.17 0.38
Average 0.18 0.38 0.19 0.39
Better-off 0.20 0.40 0.22 0.41
Rich 0.25 0.43 0.24 0.43
Household size 4.58 1.78 4.46 1.82
Television 0.83 0.37 0.81 0.40
Radio 0.57 0.50 0.51 0.50
Red River Delta 0.17 0.37 0.20 0.40
Northwest 0.02 0.15 0.06 0.24
North central coast 0.11 0.31 0.16 0.37
South central coast 0.11 0.31 0.11 0.31
Central highlands 0.09 0.29 0.07 0.26
Southeast 0.18 0.38 0.12 0.32
Mekong River Delta 0.22 0.41 0.14 0.35
Urban 0.30 0.46 0.30 0.46
Number of observations 5105 1322
ote: - Population means weighted to reflect sampling weights.
Source: Own-calculation from the VHLSS 06.

111
Table 5: Regression results for main variables of the utilization models
Variables Out-patient Inpatient
Types of health insurance




Health care for the poor -0.03 0.02
[-0.855] [0.611]
Policy -0.03 0.08**
[-0.622] [2.217]
Required state -0.02 0.09**
[-0.633] [2.234]
Required non-state -0.04 0.09
[-0.447] [1.441]
Student health insurance -0.14*** -0.05
[-2.800] [-0.754]
Voluntary -0.05 -0.02
[-1.152] [-0.419]
Types of care providers
District hospital 0.12*** 0.04
[2.673] [0.807]
Provincial hospital -0.05 -0.01
[-1.318] [-0.162]
Central hospital -0.28*** -0.11*
[-4.221] [-1.773]
Private hospital -0.96*** -0.23***
[-22.59] [-3.352]
Other health facilities -0.65*** -0.07
[-11.39] [-0.581]
Income groups
Near poor -0.08** -0.04
[-2.189] [-0.929]
Average -0.04 -0.04
[-0.845] [-0.645]
Better-off -0.09*** -0.03
[-2.677] [-0.524]
Rich -0.14*** -0.05
[-3.186] [-0.645]
Information about health insurance
Time 0.01** 0
[2.310] [0.647]
Instruction 0.09** 0.03
[2.288] [0.817]
Television 0.02 0.02
[0.489] [0.393]
Radio 0.01 0
[0.569] [0.182]
Health status
Ill 0.05** 0.04*
[2.016] [1.748]
Number of days in beds 0 0
[0.322] [1.635]
Chronic 0.04*** 0.04*
[2.752] [1.731]
Pregnant -0.07 0.07*
[-1.320] [1.655]
otes: Robust asymptotic t statistics in brackets; Regression results are weighted for sampling.
Marginal effects are calculated at mean; for dummy variables: the marginal effect is for discrete change
from 0 to 1; ***, ** and * denote significance at 1, 5 and 10 per cent, respectively.
Source: Own-calculation from the VHLSS06.

112
Table 6: Regression results for remaining variables of the utilization models
Variables Out-patient Inpatient
Male 0.03* -0.02
[1.800] [-1.167]
Age 0 0
[1.354] [0.982]
Age squared 0 0
[-1.395] [-0.470]
Married 0.03 -0.02
[0.708] [-0.361]
Kinh -0.08** -0.04
[-1.982] [-1.392]
Primary education -0.03 -0.04
[-1.589] [-1.231]
Lower secondary -0.04* -0.02
[-1.672] [-0.642]
Upper secondary -0.01 0.06**
[-0.345] [2.405]
Upper secondary -0.1 -0.04
[-1.603] [-0.236]
University 0.14*** 0.08
[2.641] [1.379]
Training 0 -0.12
[-0.106] [-1.389]
Wage-earner -0.01 0.04
[-0.408] [1.489]
Farmer -0.02 0.02
[-0.914] [1.087]
Non-farm self employed 0.01 0.05*
[0.445] [1.960]
At school 0.09 0.09***
[1.528] [3.003]
Household size 0.01* 0
[1.726] [-0.0317]
Red River Delta -0.16*** 0.03
[-3.214] [0.877]
Northwest -0.09 0
[-1.126] [0.0451]
North central coast -0.1 0.03
[-1.503] [0.920]
South central coast 0.07 0.03
[1.352] [0.744]
Central highlands 0.02 -0.04
[0.443] [-0.978]
Southeast -0.02 -0.06
[-0.332] [-1.331]
Mekong River Delta 0.05 -0.05
[1.025] [-0.858]
Urban 0.13*** 0
[6.060] [0.0398]

Log likelihood -4509479 -1061196
Number of observations 5105 1322
Predicted utilization rates 0.58 0.86
otes: Robust asymptotic t statistics in brackets; Regression results are weighted for sampling. Marginal effects
are calculated at mean; for dummy variables: the marginal effect is for discrete change from 0 to 1; ***,
** and * denote significance at 1, 5 and 10 per cent, respectively.
Source: Own-calculation from the VHLSS06.

113
The Revolution of Vietnamese Distribution System


Masayoshi Maruyama
Graduate School of Business Administration
Kobe University, Japan


Le Viet Trung
*

Graduate School of Business Administration
Kobe University, Japan
Email: hvtkobe@yahoo.com


Abstract
This paper attempts to shed light on the Vietnamese consumer preferences and
behavior based on some of our previous work. We link between consumer behavior
(preferences) and the development of a particular kind of markets. We verified the
continued dominance of the traditional markets in terms of proximity, freshness and
price as the main barrier to the supermarket development. Our analysis provides a
basis for understanding the strengths and weaknesses of the old and the new retail
outlet formats, and it will be helpful for discussing the revolution in Vietnamese
retail system.

Keywords: Organized bazaar, supermarket, consumer behavior, Probit analysis, Vietnam


*
Corresponding author.

114
1. Introduction
Since the implementation of Doi Moi from 1986, a term meaning new thinking or
renovation, a lot of changes have been taking place in the economic structure of Vietnam, one
of which is structure change of it distribution system. Before 1986, under planning system, all
food distribution was carried out by means of a rationing system or through state-owned stores.
With the liberalization of commerce and price, business activities have continuously increased.
From around 1987-1988, markets emerged for most products. Traditional bazaar system (wet
markets, mom and pop stores) became the central hub for all trade activities.
Following with the rapid economic development, Vietnamese struggle to make sense of
the multitude of products and brands that have flooded into their market, the glut of
advertising, retail outlet proliferation, and the confusion concerning consumer choices. Before
1995, most Vietnamese had no choice except to carry out all of their purchases at traditional
markets. Today, instead of shopping at the traditional outlets, consumers can choose to shop at
one-stop, air-conditioned markets. Although, the modern markets (supermarkets) were just
introduced in Vietnam just ten years ago, the rapid development of these modern markets has
changed the face of the Vietnamese distribution system. While more than 90 percent of all
products continue to be distributed by traditional channels, it must be noted that the traditional
bazaar system is now experiencing tremendous pressure from mushrooming supermarkets.
Supermarket is attracting increasing number of shopper away from traditional markets (VIR,
September 8, 2004). Vietnam is at the early state of the rapid revolution in its retailing
modernization process.
Surprisingly, however, despite the prevalence of the change that occurring, so far,
Vietnamese distribution continues to be studied very little by academics (see for example,
Vernard, 1996; Speece and Huong, 2002; Cadilhon et al., 2003; 2005; 2006). Overall, there are
only few published academic studies from the consumers perspective related to traditional
bazaars and modern retail outlets in Vietnam. The previous studies mainly collected
observations and employed qualitative, descriptive approaches to interpret the data. Only very
rarely have studies involved the use of survey data and the application quantitative and
statistical analysis methods. In addition, only a few studies have focused on the links between
consumer behavior and preferences and the development of a particular kind of markets as
well as the matter of consumer shopping behavior that involves barriers to the advancement of
markets (traditional markets and modern markets). Further more, many food retail
modernization studies and policy-makers always assumed supermarkets to provide superior
outputs to consumers. Consequently, they did not believe in traditional retailers ability to
change and develop and they strongly advocated a replacement strategy. However, some
present empirical studies provide evidences for the coexistence of both traditional and modern
food retail formats (Goldman et al., 2002; Goldman and Hino, 2005). The purpose of this
study is to attempt to undertake the first analysis related to traditional retail stores versus
supermarkets in Vietnam by identifying the factors which influence consumer decision-making
related to the selection of particular retail outlets. By analyzing the characteristics of market
shoppers we provide a basis for understanding the evolution of the old and the new retail
outlet forms.

2. Study Method
Facing the shortage of statistic data and the paucity of previous academic researches of
the retail sector in Vietnam, this study is based on a number of research methods. A systematic
review of published reports related to supermarkets and traditional retail system was
conducted. To focus on the matter of consumer shopping behavior that involves barriers to the
advancement of markets (traditional markets and modern markets), we conducted a detailed

115
consumer survey between March and April 2006 in Hanoi. Hanoi, the capital city of Vietnam
where provided a convenient location for survey was chosen for this study because it is the
second largest city, the one of the most important economic and commercial centers and its
retail system is one of the most advanced in Vietnam.
We sent out questionnaires to 2,000 inner city Hanoi households. A total of 570
questionnaires were returned. After eliminating incomplete survey questionnaires, the
remaining 413 questionnaires were used as the sample. We asked that the questionnaire be
answered by the family member who is most often responsible for shopping. We collected the
socioeconomic and demographic information of consumers. We also asked about travel
distances to stores, frequency of shopping, and perceptions of different store formats.
The analysis of the raw data from the surveys indicates that 85 percent of the
respondents were females and 15 percent were males. Even though women are still mainly
responsible for household shopping, they are also joining the workforce. The respondents
covered a broad age but mainly distributed between 25-40 year-olds. This data is
representative of the population of active consumers in Vietnam. Current statistics show that
more than 60 percent of the Vietnamese population is under the age of thirty. In terms of the
average income per family member (income divided by family size), 45 percent of the
respondents had income levels of between 500,000 and 1.5 million Vietnam dong (VND), 37
percent between 1.5 and 2.5 million VND, and 18 percent had income of more than 2.5 million
VND. Nearly one hundred percent of respondent households had a television, 99 percent had a
motorbike, 92 percent had a refrigerator, and less than10 percent had a car. These income
levels are slightly biased towards higher income respondents in comparison with the general
population characteristics of Hanoi. But if one only considers the urban areas, the sample
closely resembles the general population characteristics in the Hanoi urban area.

3. Background: Food Retail System
Rapidly Changing Situations
The Soviet-style centralized planning model was imposed in the late 1950s upon the
North Vietnam, and the South in the late 1970s after the official reunification of the country
in 1975. According to socialist doctrine, a socialist economy is one in which there exists
virtually only two forms of ownership: state ownership and collective ownership. Domestic
trade was squeezed and in many areas simply banned (Ffrorde, 1993). The government
ordered that all trade and business operations of bourgeois tradesmen be abolished; only
small merchants retailing those goods not controlled by the state could exist (Charles and
Hoa, 1996). Under the government planning system, the state assigned regular production
targets for various state-owned plants, in quantity terms, and in order for them to meet these
targets they were provided with capital resources and current inputs directly by the state. The
output produced by plants was then supplied directly to the state. The state determined the
prices and the distribution flow of the commodities by means of a rationing system or
through the state-owned stores. The application of this target system not only resulted in
great deficiencies, but also destroyed all incentive for people to work.
With heavy loss in production and distribution process and the low productivity induced
by the lack of incentives, the state could only meet one tenth of peoples demand. Therefore,
many people had to find ways to save their life in free markets. In general, people could sell
only goods that were not distributed by the state. Goods like rice, meat, soap, clothes, etc.,
therefore, were considered as illicit goods and everyone who engaged in transaction of these
goods could be considered as an illicit dealer. All the states effort was to try to prohibit
market. However, people tried on any way of breaking out, such as wearing rice-contained
double-trousers or tying up around their body with meat covered under their clothes, etc.

116
By the late 1980s the country was suffering from an acute staples crisis. The failure of
the centrally planned economy has made Vietnamese governments introducing renovation
policy (Doi moi) in 1986. In the reformed system, each farmer has a long-term lease of a plot
of land. The farmers only have to deliver an annual quota of produce to the state as a rental
payment for the use of the land. They may sell their output in markets. As a result, the farmer
faces full market incentives, in the sense that any increased effort translates directly into
increased income. These led to an immediate jump in farm output such that after facing near
famine conditions in some parts of the country in 1988, Vietnam emerged as the worlds third
largest rice exporter one year later (Charles and Hoa, 1996). As agricultural output boomed,
rural marketplaces developed rapidly. In the towns and cities, many new markets were created.
From around 1987-1988, markets emerged for most products (Charles and Hoa, 1996).
Already in 1988, almost every house in the central streets of Hanoi was using its frontage as a
retailing outlet again (Waibel, 2004). Traditional bazaar network is again established and
continuously developed in the whole country. The number of markets is more than doubled
from 4000 in 1993 to 8300 in 2003 (VNS, June 12, 2004). Hanoi alone has over 120 markets,
with total area of over 550 thousand square meters.
With the liberalization of commerce as well as price liberalization, business activities
have continuously increased. The service sector expanded from 38.6 percent of GDP in 1990
to 41.6 percent in 1994. Trade mark-ups contributed around 33 percent of the service sectors
total contribution to the GDP. The development of commercial activities was due to the rapid
growth of the private sector. The percentage of private trade in total trade increased from a
low of 42 percent in 1986 up to 71 percent in 1991 and around 80 percent in 2003. In 1985,
the state sector and the co-operative sector represented 60 percent of the business in the
Vietnam dongs and 40 percent for the private sector. In 1988, the breakdown was reversed
with the private sector accounting for a little more than 50 percent to 80 percent in 1993
(Venard, 1996). The number of household trade establishments was also growing sharply,
from 630 thousand in 1991, to 1.1 million in 2000, and more than 1.6 million in 2002 (Table
1).
Table 1:. Some Indicators of Economics and Trade Development in Vietnam
Year 1991 1993 1995 1997 1999 2000 2002
Real GDP growth rate 5.81 8.08 9.54 5.15 4.77 6.75 7.04
Number of household trade (Thous.) 630 743 1160 949 1089 1100 1669
Numbers of persons working in trade
(Thous.)
1473.9 1689.7 1936.6 2216 2537.5 2714.4 3106
Total retail of trade (Bil. VND) 29183 58424 94863 131770 166989 183864 221569
Growth rate of retail of trade (%) 74.3 30.5 28.0 12.1 8.6 10.1 10.8
Source: General Statistics Office
The country has experienced strong economic growth in recent years. GDP growth rate
was over 7.5 percent from 2001 to 2005. Furthermore, cities account for 70 percent of the
national GDP due to industrial and trading activities. This brings annual per capita GDP to
US$1,395 in the urban center of the country. Although, Vietnam still relies on an
agriculture-based economy and remains a developing country with an annual per capita
income of US$425, if one only considers the cities, economic indicators are more
representative of the middle-income group of countries in transition (Cadilhon et al., 2003).
Household revenues have increased and changes in food consumption are appearing.
Families with growing incomes are buying more food rather than producing it themselves.
Furthermore, Vietnamese government has definite plans to encourage economic

117
development by liberalizing domestic retail markets and international trade. The government
is keen to develop a modern distribution network throughout the country by encouraging
investors from all economic sectors to construct more modern types like supermarkets (VNS,
March 12, 2005). In parallel, in its plan, the government still uses the existing bazaar network
as the basis of the national market network (VNS, June 12, 2004). In its effort to modernize
the existing bazaar system, the government has announced a market development program
extending to the year 2010 in Decision No. 559/QD-TTg on May 31, 2004. Remarkably, this
decision formally puts the traditional bazaar system in the project list-up that this system will
receive preferential treatment. With support from the government, the traditional bazaars are
also on the way to being updated and also newly built.
Retail System in Hanoi
The retail system in Hanoi is defined by the new emerging retail formats and older,
more traditional retail formats. The distribution system is represented in Figure 1. The
distribution system for processed food runs from importers and domestic producers to
wholesale sector (i.e., wholesalers, trading companies, and wholesale agents) to retail sector
(i.e., mom and pop stores and agent, supermarkets and organized bazaars). The marketing
channels of agricultural products (fresh produces) can be classified into two major groups, i.e.
traditional and modern distribution systems. The traditional system runs from producers to
wholesale sector (assemblers, transporters, wholesalers) to retail sector (i.e., organized bazaar
retailers, frog market retailers and hawker). The modern distribution system runs from
producers to wholesalers and assemblers to supermarkets to consumers. The traditional
marketing system still dominates in Hanoi. Less than 10 percent of product distribution takes
place using a modern distribution network. The older retail forms such as the traditional
organized bazaars, traditional mom and pop stores, and frog markets (informal markets)
account for more than 90 percent of the remaining demand (VNS, July 1, 2005).
Figure 1. Food Distribution Channels

Note: Figure 1 represents only vertical links between stakeholders. Solid line means the marketing channel for
food. Dotted line means not a main channel.

Domestic
Producers
Importers Local Agricultural producers
Assemblers, wholesalers
Wholesalers, trading
companies
Mon & Pop stores,
Agents
Organized Bazaars Supermarkets Frog Markets
Consumers
Wholesale markets
Agents
Processed food, imported food

Fresh produce

118
Traditional Retail Outlets
Organized Bazaars
A formal bazaar is defined as a market established and formally approved by either the
provincial or district level authorities. This kind of bazaar is a market in which many, tiny and
highly specialized retailers operate. The environment is chaotic, crowded, dirty, and noisy.
Food is typically sold from open stalls. In general, space is organized in locations where each
specializing in one of food line.
Hanoi alone has over 127 organized bazaars of which 60 markets are located in the
inner city. They are classified into three types of markets. Type 1 is built with concrete,
modern and well managed, and employed more than 400 traders. Type 2 is built with concrete
or semi-concrete and well managed with more than 200 traders. Type 3 consists of temporary
facilities with less than 200 traders. A market management board is selected and operated
under the control of the local authority. They have limited, low-grade facilities, such as for
water supply and liquid and solid waste treatment systems. They lack storage facilities and
refrigeration (Gia et al., 2003). The high commercial attractiveness of Hanoi is triggering an
increasing demand for more business activity sites. The floor space per trader at these
markets in the inner city is very small and decreasing.
Informal Bazaars
In addition to the economic development, big cities like Hanoi have also become prime
areas for the booming and spontaneous development of informal bazaars, called Cho Coc
(frog markets). This kind of market is not approved by local governments. A retailer in these
markets is broadly defined as a person who offers goods for sale to the public without having
a permanent built-up structure from which to sell. The vendors (street vendors or hawkers) in
these bazaars may be stationary, in the sense that they continuously occupy a given space on
the pavement or other public space or a private place. Otherwise, they may be mobile, in
the sense that they move from place to place by transporting their wares on push-carts or in
baskets. Gia et al. (2003) confirmed that Hanoi has more than 600 informal bazaars. Many of
the retailers are farmers and the poor from the surrounding countryside or laid-off people or
retirees. Sellers operations are tiny in scale. Because of the small scale of such operations,
they can easily move about or flee the police. Sellers at these markets usually have no storage
facilities. These frog markets have existed for a decade even though they are increasingly
harassed by the police. City authorities planned to dismantle all these markets in 2005.
Nevertheless, local authorities have been unable suppress this kind of selling activity (BAI,
August 22, 2005).
Traditional Mom & Pop Stores
The traditional mom & pop stores are family-owned retailers that sell a limited variety
of processed food, dry goods, drinks, toys, and household supplies, etc. Specialized stores
have emerged as well, often for such products as shoes, clothing and furniture, hi-fi
equipment, air-conditioning equipment, modern refrigerators, washing machines, and other
appliances. Many streets often specialize in a single category of products. In this paper, these
kinds of streets are referred to as street trade bazaars. There are often 200-300 identical mom
& pop stores (occasionally, as many as 400-500 stores) along a few hundred meters of a
single street. Most clustered mom & pop stores have similar inventories. They have the
characteristic of being the only sales points to offer competitive brands for the same product
category. In general, merchandising with these small businesses is rudimentary. The exterior
and interior features of these stores are very simple. The presentation of products is
simplified, often with no shop windows and no decent sales racks. The sales area is nearly
always very small, due to the high cost of land use.

119
Supermarkets
After the first supermarket appeared in Vietnam in 1993, about 70 supermarkets more
were quickly established in HCMC and another 20 in Hanoi. About two-thirds of the original
outlets failed, but some survived and have reaped the benefits of strong growth in the modern
retail sector. By 2000, it was estimated that there were about 40 mini supermarkets in HCMC
and about 20 in Hanoi (Speece and Huong, 2002). The number of supermarkets and shopping
malls has been increasing rapidly from 10 in 1995 to 210 at the end of 2004 (Nhieu, et al.,
2005). Hanoi and HCMC had 155 supermarkets. Besides the rise in the number of
supermarkets, the dimensions of these outlets are also on the increase. In the early stages, the
average floor space of the typical supermarket store ranged from 500 to 800 square meters,
but by 2000, a few supermarkets had begun to appear with larger floor spaces of around 2000
square meters or more (Speece and Huong, 2002; KHTDC, 2001). Vietnam has five domestic
supermarket chains vying for customers and two foreign-owned chains. Big C operates four
superstores and Metro has six.
In 1994, 95 percent of goods displayed at the city supermarkets were imports. At
present, domestic goods account for 70 percent of total stock. Between 85 to 90 percent of
goods at big supermarkets, including Co-op, Big C and Metro, are locally made products
(VNS, May 5, 2004). Supermarkets initially focused on dry items, packaged and processed
food lines and on non-food products. Today, many supermarkets carry fresh food, but offer a
relatively weak range. Variety is relatively low. They have made little progress in fresh food
lines where their share accounts for a very small percentage (Loc, 2003; Cadilhon et al.,
2006).
Some foreign retailers have taken a foothold in Vietnam. The French Bourbon Group
opened its first hypermarket in Vietnam in late 1998. Metro Cash & Carry has opened two
centers in HCMC (in 2002 and 2003), one in Hanoi (in 2003), and the other three centers in
other cities. Malaysian retailer, Parkson, also entered the market in June 2005 (VIR, May 28,
2006). Japans Seiyu, which has been operating the Hanoi Seiyu for five years, is working on
obtaining licenses to extend its retail network to other provinces. In addition, a leading
pan-Asian retailer, Dairy Farm (Hong Kong) has negotiated an agreement in principle to
launch a supermarket chain, and could enter into a partnership with Citimart. Tesco from the
UK, US giant Wal-Mart and the French group Carrefour are also trying to obtain licenses
enter the Vietnamese market (VNS, July 1, 2005; AFP, May 21, 2006; VIR, May 28, 2006).
Initial supermarket successes have prompted both domestic and foreign investors to expand
their businesses in Vietnam.
4. Results
Adoption of supermarket shopping
A number of indicators attest that the supermarkets popularity among Vietnamese
consumers and their growing importance. Our 2006 survey reported that nearly all of the
respondents indicated that they had shopped in a supermarket before. Thirty five percent of
shoppers reported visiting supermarkets at least once a week, up from 28 percent in 1997
survey by Speece and Huong (2002). In the 1997 survey done by Speece and Huong (2002),
consumers indicated that the most important reason to go into supermarket was quick
examination of prices but this was one of the lowest ranked reasons in the recent survey.
Curiosity also used to be a key reason, but this is no longer an important reason for infrequent
shoppers. Our survey show that self service, guarantee of quality, fixed prices, safe and clean
goods, the ability to search for something unique, and a one-stop service were the major
reasons for shopping at supermarkets identified by respondents (Table 2).


120
Table 2: Reasons for going to supermarkets

Resent survey
(2006 survey)
Speece and Huong (2002)
(1997 survey)
Reasons
Self service 80.7 -
Guarantee of quality 76.3 -
Fixed price 75.6 -
Search for something unique 61.6 46.0
One-stop service 60.1 -
Enjoyable and relaxing 40.6 33.5
To buy everyday necessities 38.4 50.6
Quick examination of prices 22.5 65.3
Curiosity 21.8 48.9
Prestige image 2.7 5.7
Fashionable 2.4 5.1
Note: Reported figures are the percentages of respondents who selected the reasons in the checklist.
Hanoi supermarkets are mostly used for toiletries and household amenities. Shoppers
buy a medium amount of processed food, frozen food, confectionery and drinks. Very few
consumers buy fresh food, personal care products, cloths, consumer durables, and footwear.
This shopping pattern has not changed since the 1997 survey of Speece and Huong (2002)
(Table 3).
In general, Vietnamese consumers are still accustomed to small and less frequent
shopping trips to supermarkets. More than 65 percent of the respondents shopped less than
one a week and the amount of money they spent per visit was also small, more than 63
percent of the respondents spent less than 300 thousand VND per visit (approximately
US$18).
Table 3: Kinds of goods consumers bought at supermarkets

Resent survey
(2006 survey)
(*)
Speece and Huong (2002)
(1997 survey)
(**)

Mean Std. Dev. Percent
Kinds of goods
Toiletries 3.59 0.81 -
Household amenities 3.48 0.86 82.6
Confectionery 3.13 0.91 36.4 (and drink)
Frozen food 3.09 0.95 -
Drinks 3.05 0.96 -
Processed food 3.04 0.93 -
Fresh food 2.75 1.13 -
Personal care products 2.65 1.09 48.5
Ready-to-eat food 2.56 1.00 53.0
Clothes 2.38 0.95 18.2
Consumer durables 2.00 0.99 16.7
Footwear 1.99 0.95 25.8
Note:
*
In our 2006 survey we used 1 to 5 scale; 1 = never, 2 = rarely, 3 = sometimes, 4 = often, 5 = always. For
vertical comparison, if the difference between any two sample means is greater than 0.2, then the difference
between these two normal population means is significant at the 5 percent level.
**
Reported numbers are the
percentage of respondents


121
Although, a majority of the respondents still do most of their shopping for food at
traditional retail outlets, 64 percent of the respondents indicated that they preferred
supermarkets to traditional bazaars. A recent survey in the South of Vietnam also found that
up to 85 percent of urban dwellers were also fond of shopping at supermarkets (Thanh Nien,
2004). However, 63 percent of our respondents revealed that they are not loyal to any specific
supermarkets. This suggests that supermarkets can increase market share by improving
consumer satisfaction and royalty.
A Probit Analysis of Consumer Choice for the traditional bazaar versus the supermarket
We found that a major tendency among consumers is to split their goods purchase. With
respect to the purchasing behavior for fresh food, 73 percent of the respondents indicated that
they do most of their food shopping at traditional organized markets. Buying fresh food at
frog markets accounted for 20 percent of purchases. Only 7 percent of the respondents
purchased fresh food regularly from supermarkets. For the processed food and drinks, over 50
percent of respondents indicated that they regularly purchased such items at mom and pop
stores. The percentage of respondents that regularly purchased such items at supermarkets has
increased significantly to nearly 30 percent. For non-food items, the majority of the
respondents, 78 percent, indicated that they do most of their shopping at mom and pop
specialty stores. The data suggest that traditional retail formats continue to serve as the main
outlets for both food and non-food products, while traditional indoor markets dominate fresh
food and mom and pop stores dominate non-food products.
In order to provide a detailed analysis of the Vietnamese consumer behavior and
perceptions with respect to different kinds of retail outlets, the respondents were asked to
grade the importance from levels 1 to 5 for factors (reasons) that influence their decisions in
selecting the type of retail outlet at which they do most of their shopping. Maruyama and
Trung (2007b) used these data applying to the Probit model for binary choice to analyze the
question of why shoppers choose to do most of their shopping for fresh food at organized
markets (choice between organized markets and supermarkets). Binary choice for retail outlet
by each consumer was treated as an individual observation, Y
i
taking the value 1 if a consumer
chose to do most of their shopping at formal markets and 0 if a consumer chose to do most
of their shopping at supermarkets. It was also used to analyze why shoppers choose to do
most of their shopping for processed food and non-food products at mom and pop stores
(choice between mom and pop stores and supermarkets). Explanatory variables used in the
models included importance ratings of the factors of freshness (FRESH), safety (SAFETY) or
quality (QUALITY) concerning items purchased, cheap price (CHEAP), convenience for
shopping (COVEIECE), and factor of income (ICOME). Some other variables that may
influence consumer choice of retail outlets are also added, i.e. importance ratings of the
factors of good service of sellers, maintaining close relationships with sellers
(USUALSELLER), means of transportation (TRASPORT) which shoppers use to go shopping,
and the important rating of the absence of bargaining (OBARGAI). We also included some
other demographic variables in the model such as gender (SEX), age (AGE), or whether there
are children or not (CHILD). The results are reported in Table 4.
The empirical results show that freshness (FRESH), cheap price (CHEAP) and
proximity (COVEIECE) are the key significant factors attracting consumers to traditional
organized markets for fresh food. An increase in income or higher concern about safety of
fresh food increases probability of choosing a supermarket.

122
Table 4: Results from the Probit binary choice model
Fresh food Processed food & drinks on-food products
Variable Coefficient
Marginal
probability
Coefficient
Marginal
probability
Coefficient
Marginal
probability
FRESH 0.45
**
0.0012
NEWGOODS -0.004 -0.0013
QUALITY -0.24
**
-0.0431
SAFETY -0.46
*
-0.0012 -0.15
**
-0.0543
CHEAP 0.30
**
0.0008 0.22
***
0.0789 0.19
**
0.0339
USUALSELLER 0.76
***
0.0020 0.32
***
0.1171 0.13 0.0239
CONVENIENCE 0.32
**
0.0008 0.05 0.0182 0.09 0.0167
SERVICE 0.18 0.0005 -0.05 -0.0173 0.09 0.0167
NOBARGAIN -0.42
***
-0.0011 -0.18
***
-0.0641 -0.33
***
-0.0584
SEX 0.23 0.0005 -0.16 -0.0587 -0.45
**
-0.0969
CHILD 0.55 0.0013 0.20 0.0735 -0.18 -0.0333
AGE 0.06 0.0002 0.06 0.0212 0.13 0.0234
INCOME -0.44
**
-0.0011 0.04 0.0148 0.03 0.0055
TRANSPORT -1.31
***
-0.0052 -0.44
***
-0.1519 -0.68 -0.0815
CONSTANT 0.54 0.43 2.35
***

Log likelihood -35.52 -173.08 -130.21
Number of
observations
330 327 381
LR chi2 (12) 115.87 82.64 64.63
Prob > chi2 0.0000 0.0000 0.0000
Pseudo R
2
0.62 0.19 0.20
Note: ***and ** and * indicate coefficients are significant at the 0.01, 0.05 and 0.1 levels, respectively.
For processed food & drinks and non-food products, price (CHEAP) and safety
(SAFETY) or quality (QUALITY) of goods appeared to be important in shaping consumer
choice of shopping outlet type. These suggest that shoppers who consider price as an
important factor in their purchasing decisions, is likely to chose shopping at mom and pop
stores. Shoppers who place more important rating on safety would be more likely to choose to
buy at supermarkets. The impact of gender (SEX) was not found in the model for fresh food
and processed food & drink, but significant for non-food products. This suggests that it is
highly probability that a man likely to choose to shop at supermarkets. Income had significant
effect in shaping consumer selection of shopping outlet for fresh food, but did not have a
significant effect in choosing shopping outlet for processed food & drinks and non-food
products. OBARGAI shows a negative and significant for three models, suggesting that the
higher concern for not bargaining, the higher the probability of choosing to shop at
supermarkets for three kinds of goods.
These results suggest that traditional bazaar still very competitive in term of price,
freshness and convenience. The preference for purchasing fresh produce items in traditional
outlets is strong. In fact, pperceiving meals prepared from natural and fresh ingredients as
more delicious, healthier and less risky most Vietnamese consumers do their shopping at a
high frequency (more than 60.2 percent of the respondents in our sample shop at least five or
six times a week or everyday). They mainly go to traditional indoor bazaars or frog one in the
morning or the late afternoon (more than 65 percent of the sample), the time the freshness is
highest as evaluated by the respondents. They usually do not buy fresh food in bulk or use
refrigerators to store fresh food. This shopping habit has taken root over a long time and it is
not likely to be changed in the short-term. Maruyama and Trung (forthcoming) found that
freshness, convenience (distance) had a significant effect on the frequency of shopping. In
other words, the higher emphasis on freshness, the less distance from home to the traditional

123
markets, the higher is the frequency of shopping. In contrast, factors such as owning a
refrigerator as well as income did not have a significant effect in shaping consumer shopping
behavior. The tendency to purchase fresh produces in traditional outlets combined with the
proximity (convenience) and cheap price these markets serve were found to be the biggest
barriers to supermarket development.
Evaluation of Market Performance
Table 5: Evaluations of market performance
Evaluation of
Supermarkets
Evaluation of
Organized
Markets
Evaluation of
Mom & Pop
Stores
409 413 413

Sample size
Mean
Std.
Dev.
Mean
Std.
Dev.
Mean
Std.
Dev.
Elements
Variety of product lines 3.70 0.62 3.65 0.78 3.35 0.77
Air conditioning 3.70 0.61 - - - -
Cold storage (infrastructure at organized) 3.67 0.60 2.70 0.84 3.16 0.84
Variety in each kind of product 3.60 0.65 3.48 0.78 3.48 0.75
Quality of merchandise 3.56 0.54 3.33 0.58 3.42 0.67
Product display 3.51 0.61 2.89 0.76 3.35 0.78
Location 3.42 0.64 3.19 0.70 3.27 0.78
Scale 3.40 0.62 3.18 0.68 3.07 0.74
Parking lot 3.29 0.79 2.81 0.84 2.68 0.86
Services of salesperson 3.25 0.68 3.11 0.70 3.39 0.75
Advertisement and promotion 3.23 0.71 2.62 0.88 3.01 0.89
Checkout 3.12 0.75 - - - -
Development of reputation (honest of
sellers)
3.07 0.70 2.86 0.77 3.18 0.81
Distance (density of supermarkets) 2.99 0.70 3.34 0.70 3.44 0.80
Price level 2.94 0.50 3.20 0.60 3.14 0.58
Freshness 2.70 1.04 3.68 0.69 3.45 0.68
Return and adjustment policy 2.63 0.79 2.39 0.90 2.88 0.90
Note: (1) 1 to 5 scale, 1 = very bad, 2 = bad, 3 = normal, 4 = good, 5 = very good.
(2) For vertical comparison, if the difference between any two sample means is greater than 0.1, then the
difference between these two normal population means is significant at the 5 percent level. Comparing the mean
values of evaluation on each element of supermarkets and traditional indoor bazaars (horizontal comparison), the
same results hold in case of horizontal comparison.

Table 5 summarizes the consumer evaluations for the forms of retail outlets:
supermarkets, traditional organized bazaars and mom and pop stores. In general, supermarket
performance is perceived to be rather poor in a number of respects. All of the elements rated
lower than a good level of performance. Supermarkets were evaluated highly in some areas
such as quality of merchandise, variety of product lines, product variety within a single
category of product, cold storage, and air conditioning. Price level, checkout, service of
salesperson, density of supermarkets, parking lot, advertisement and promotion, development
of reputation, and return and adjustment policy factor were also rated at fairly low levels.
By comparing the evaluations for each element of supermarkets and that of traditional
organized bazaars, we found that consumers evaluated the traditional bazaars at significantly
higher levels than supermarkets in terms of freshness, price levels, and convenience (distance).
There was no statistically significant difference between organized bazaars and supermarkets
in terms of variety of product lines, but consumers evaluated supermarkets significantly higher

124
for the rest of the listed elements. This information again suggests that freshness, price levels
and convenience (distance) are the key factors attracting consumers to the traditional
organized bazaars. By comparing the evaluations for each element of supermarkets and that of
traditional mom and pop stores, we found that consumers evaluated the mom and pop stores at
significantly higher levels than supermarkets in terms of price level, service of salesperson,
freshness, return and adjustment policy, and convenience (distance). Consumers evaluated the
other criteria such as variety of product lines, quality of merchandise, product display,
advertisement and promotion significantly higher for supermarkets than those for mom and
pop stores.
Traditional organized bazaars provide higher levels of freshness but do not offer
promotions or provide a good environment (for example, clean, bright windows, and
air-conditioning, etc.). However, such environment and lack of promotions may have less of a
negative effect on Vietnamese consumers. The above analysis shows that shoppers go to
traditional organized bazaars for freshness, cheaper prices and convenience and to protect
themselves from being cheated by establishing relationships with sellers. Shoppers that do
most of their shopping at supermarkets are different from shoppers at traditional markets,
because they are attracted by superior safety and quality, and because they do not need to
bargain. Cheap prices are less significantly important in their shopping decisions.

5. Conclusions
This study provides a better understanding about Vietnamese consumer shopping
behavior and their consumption environment. Our analysis shows that at present, the
traditional markets remain the major distribution channels for both food products and
non-food. For fresh food, we verified the continued dominance of the traditional organized
markets in terms of proximity, freshness and price as the main barrier to the supermarket
development. Traditional markets offered more advantages than supermarkets in terms of fresh
food, proximity, and price.
The initial results show evidences for the coexistence of both traditional and modern
food retail formats, where organized markets provide one-stop shopping for a large variety of
fresh food, and supermarkets provide the same advantage for packaged and processed foods or
non-food products. Thus, organized markets and supermarkets should be viewed as
complementing one another and as being equally important. These results are important and
cast a doubt on the common belief that traditional retail system implies backwardness and
inefficiencies. Many food retail modernization studies and policy-makers always assumed
supermarkets to provide superior outputs to consumers. Consequently, they did not believe in
traditional retailers ability to change and develop and they strongly advocated a replacement
strategy.
Traditional organized markets could be changed in the way that distribution of fresh food
tends to be much greater than processed food items, while supermarkets will become the main
distribution route for processed food and non-food products. For processed food & drinks,
price appeared to be important in shaping consumer choice of shopping outlet type. Traditional
mom and pop stores will suffer from losing market share for processed food and non-food
items to supermarkets because of price competition.
Our findings in this study will have major implications for the future strategies of each of
the players in the retailing area. Supermarket executives need to develop formats that will
cater better the present supermarket formats to the particular needs of the consumers. The
findings suggest that supermarkets that provide various types of products and continue to add
new, high quality products at reasonable prices, and are built near consumer homes, have
opportunities to build store loyalty and increase market share.

125

References
Cadilhon, J. J., Fearne, A. P., Moustier, P. and Poole, N. D. (2003), Modeling Vegetable
Marketing Systems in South East Asia: Phenomenological insights from Vietnam,
Supply Chain Management: An International Journal 8(5), 427-441.
Cadilhon, J. J., Fearne, A. P., Tam, P. T. G., Moustier, P. and Poole, N. D. (2005),
Collaborative Commerce or Just Common Sense? Insights from Vegetable Supply
Chain in Ho Chi Minh City, Supply Chain Management: An International Journal
10(3), 147-149.
Cadilhon, J. J., Moustier, P., Poole, N. D., Tam, P. T. G. and Fearne, A. P. (2006), Traditional
vs. Modern Food Systems? Insights from Vegetable Supply Chains to Ho Chi Minh
City (Vietnam), Development Policy Review 21(1), 31-49.
Charles, H. and Hoa T. V. (1996), Vietnams Reforms and Economic Growth, Macmillan Press
Ltd.
Fforde, A. (1993), Vietnamese Commerce: the Tiger on a Bicycle Syndrome, Columbia
Journal of World Business 28(4), 48-53.
Gia, B.T., Cuong, T.H., Tru, N.A., Huong, T.T.T. (2003), Vegetable retail marketing in Hanoi
province, VEGSYS: Research Project Report No. PR12.
Goldman, A., Ramaswami S. and Krider R. E. (2002), Barriers to the Advancement of
Modern Food Retail Formats: Theory and Measurement, Journal of Retailing 78 (4),
pp. 281-295.
Goldman, A. and Hino H. (2005), Supermarkets vs. Traditional Retail Stores: Diagnosing
the Barriers to Supermarkets Market Share Growth in an Ethnic Minority Community,
Journal of Retailing and Consumer Services 12 (4), 273-284.
Hong Kong Trade Development Council (HKTDC) (2001) Supermarkets Replace Wet
Markets for Vietnams Wealthy, available at tdctrade.com
(http://www.tdctrade.com/mne/food/food010802.htm).
Loc, N. T. T. (2003), The Development of Shops and Supermarkets in Fresh Vegetable
Commodity Markets in Hanoi and HCM City, Research Institute on Fruits and
Vegetables of Vietnam (RIFAV: FSP Project 2000-56).
Maruyama, M. and Trung L. V. (Forthcoming), High Shopping Frequency of Vietnamese
Consumers: Theory and Measurement, Applied Economics Letters. Forthcoming.
Masayoshi Maruyama and Le Viet Trung (2008), The Revolution of Distribution Systems in
Transition Economies: Probit Analysis of Consumer Behavior in Hanoi, Vietnam,
Proceedings of The 37 European Marketing Academy Conference 2008, May 2008,
Brighton, UK.
Maruyama, M. and Trung, L. V. (2007a), Supermarkets in Vietnam: opportunities and
obstacles, Asian Economic Journal 21(1), 19-46.
Maruyama, M. and Trung, L. V. (2007b), Traditional bazaar or supermarkets: A Probit
Analysis of Affluent Consumer Perceptions in Hanoi, International Review of Retail,
Distribution and Consumer Research 17(3), 233-252.
Masayoshi Maruyama and Le Viet Trung (2006), A Probit Analysis of Consumer Shopping
Behavior, Proceedings of Australian and ew Zealand Marketing Academy Conference
2006 (CD-ROM).ANZMAC 2006.

126
Nhieu, N. T., Hung, N. V., Cai., P. T., Thin P. H. and Hien, T. T. T. (2005), Real State and
Methods for Developing Supermarket System in Vietnam, Ministry of Trade, in
Vietnamese.
Speece, M.W. and Huong, L.T.T. (2002), Attitudes of Mini-Supermarket Shoppers in Hanoi,
Vietnam: A Case Study in the Early Development of Modern Retailing, Journal of
Korean Academy of Marketing Science 10(11), 187-212.
Venard, B. (1996) Vietnamese Distribution Channels, International Journal of Retail &
Distribution Management, 24(4), pp. 29-40.
Waibel, M. (2004) The Ancient Quarter of Hanoi - A Reflection of Urban Transition
Processes, ASIE, 92(S), pp. 30-48.
AFP (2006), Vietnam, Rise of Supermarkets Sparks Global Competition, Agence France
Presse, May 21.
Business and Advertisement Information (BAI) (2005), How to Do with a Hard Problem of
Bazaar, Business and Advertisement Information, No. 478, August 22 (in Vietnamese).
Thanhnien News (TNN) (2004), Foreign Investors Eyes on Vietnams Supermarket Industry,
Thanhnien ews, July 26.
Vietnam News (VNS) (2004), Domestic Goods Sold at Supermarkets Up, Vietnam
ews,May 5.
Vietnam News (VNS) (2004), Vietnam Rethinks its Market Strategy, Vietnam ews, June
12.
Vietnam News (VNS) (2005), VN to back home-grown retailers, Vietnam ews, March 12.
Vietnam News (VNS) (2005), Supermarkets Expect Fierce Competition, Vietnam ews,
July 1.
Vietnam Investment Review (VIR) (2004), Shoppings Big Guns Ride into Town, Vietnam
Investment Review, September 8.
Vietnam Investment Review (VIR) (2006), The Price is Right Competition, Vietnam
Investment Review, May 28.













ONE-PAGE
PRESENTATIONS
Efficiency Estimates for the Agriculture Sector
Efficiency Estimates for the Agriculture Sector
in Vietnam: A Comparison of Parametric
in Vietnam: A Comparison of Parametric
and Non
and Non
-
-
parametric Approaches
parametric Approaches
This is a forthcoming article in Agriculture Economics Review. The paper uses parametric
approach (based on stochastic frontier production functionSFPF) and non-parametric
approach (based on data envelopment analysisDEA) to estimate technical, allocative, and
economic efficiency measures for the agricultural production in sixty provinces in Vietnam
during 1995-2005. We use provincial data of agricultural inputs and outputs for these
research purposes.
Under the specification of variable returns to scale (VRS), the mean technical,
allocative and economic efficiency indices of the sample provinces were 52.3%; 80.5%;
and 42.1%, respectively, for the SFPF; and 82.1%; 81.5%; and 67.2%, respectively, for
the DEA.
Under the specification of constant returns to scale (CRS), they were 58.5%; 71.9%;
and 42.1%, respectively, for the SFPF; and 79.3%; 80.9%; and 64.4% for the DEA.
Although the estimated mean technical, allocative, and economic efficiency measures
obtained from the DEA are higher than those from the SFPF in both VRS and CRS
models, efficiency rankings of the sample provinces based on these two approaches are
highly correlated, in which the highest correlation is achieved for the technical
efficiency rankings under VRS and CRS.
Prof. Nguyen Khac Minh (NEU) khacminh@gmail.com
Giang Thanh Long (GRIPS & VDF) gtlong_grips@yahoo.com
The Vulnerability of the Elderly to Poverty:
The Vulnerability of the Elderly to Poverty:
Determinants and Policy Implications for
Determinants and Policy Implications for
Vietnam
Vietnam
This paper seeks to identify the determinants of the poverty incidence for the elderly and their
households in Vietnam by examining various possible individual and household characteristics.
This paper uses the Vietnam Household Living Standard Survey in 2004, which provides
representative data about elderly and their households in Vietnam.
Using Chow tests, we find that urban and rural elderly are substantially different, and thus
they should be analyzed separately.
The results for urban elderly generally show that higher ages, marital status, residential
regions, and working status have significant impacts on the likelihood of poverty.
In rural areas, higher ages, gender, marital status, ethnicity, residential regions, household
composition, and household size are determinant factors of the likelihood of poverty for the
elderly.
Moreover, remittances and social security benefits appear to be important for reducing
poverty of the elderly households, particularly in the rural areas.
We also found some factors which are less important for both areas, including
characteristics of household heads
Based on these findings, we formulate policy priorities, including reducing regional
disparities, promoting the rural economy, and reforming the social security system.
Associate Prof. Wade Pfau (GRIPS) wpfau@grips.ac.jp
Giang Thanh Long (GRIPS & VDF) gtlong@vdf.org.vn
AN INQUIRY INTO THE TRANSFORMATION PROCESS OF
VILLAGE-BASED INDUSTRIAL CLUSTERS: THE CASE OF AN IRON
AND STEEL CLUSTER IN NORTHERN VIETNAM
This paper investigates the determinants of the transformation of village-based industrial
clusters, taking an iron and steel cluster in Dahoi village in northern Vietnam as an example.
The empirical analyses based on data collected from a survey of 204 enterprises that produce
various iron and steel products show that the proprietor's human capital acquired through
formal education and experiences in marketing and management and social capital measured
by family ties with blood siblings play an important role in the transformation process of the
cluster.
Using the Heckman-two step estimation method to estimate the determinants of the innovations in
products, production organization, and marketing channels and the performance of the enterprises, the
paper finds that:
1) The proprietor's formal education is important to produce more modern and higher-quality products,
to increase the direct transactions with the outside traders, to improve the management of labor, and to
achieve better performance.
2) The proprietor's experience in marketing contributes to the increase in the product quality, the direct
transactions with the outside traders, and the operation size of the enterprise while his experience in
management is essential to improve the management of labor and the productivity of the enterprise.
3) The proprietor's parents and parents-in-law influence his decision on the production of more modern
products while his blood siblings play an important role in the multifaceted innovations and the
improvement in the performance of the enterprise.
Vu Hoang Nam (FASID/GRIPS & Foreign Trade University, Hanoi) hoangnamftu@yahoo.com
Simulating the Potential Impacts and Costs
Simulating the Potential Impacts and Costs
of an Extended Social Pension in Vietnam
of an Extended Social Pension in Vietnam
This paper uses the Vietnam Household Living
Standard Survey in 2004, and applies micro-simulation
techniques to calculate how the poverty measures
would have been changed in the counterfactual
situation that such a scheme had been introduced to
Vietnam in the past. Different categorical targeting
groups of elderly people are considered.
We find that, depending on the characteristics of the
social pension, there would be beneficial poverty
reductions, but also large leakages to the non-poor
people. For a variety of measures, our results suggest
that targeting the elderly in rural areas might be the
most effective use of limited resources.
Both static and stochastic long-term financial
projections show that total cost of the scheme will be
about 2-3 percent of GDP.
Associate Prof. Wade Pfau (GRIPS) wpfau@grips.ac.jp
Giang Thanh Long (GRIPS & VDF) gtlong@vdf.org.vn
The paper aims to quantify the potential impacts and costs of
an extended social pension scheme for reducing elderly
poverty in Vietnam.
60 65 70 75 80 85 90
0
0.2
0.4
0.6
0.8
1
1.2
1.4
Total Cost as Percentage of 2004 GDP
60 65 70 75 80 85 90
4
6
8
10
12
14
16
18
Percent of Benefits That Reduce Poverty Gap
Recipients
All Elderly
Total Population
60 65 70 75 80 85 90
-50
-40
-30
-20
-10
0
Percent Change in Poverty Rates
Starting Eligible Age
60 65 70 75 80 85 90
-60
-50
-40
-30
-20
-10
0
Percent Change in Poverty Gap
Starting Eligible Age
2005 2010 2015 2020 2025 2030 2035 2040 2045 2050
0
0.5
1
1.5
2
2.5
3
3.5
4
ALL64-55
ALL 64-55
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
2004 2010 2015 2020 2025 2030 2035 2040 2045 2050
MediumCost Low Cost High Cost
Static Simulations
Stochastic Simulations
Development of Corporate Bond Market in Vietnam
Tran Thi Thanh Tu, NEU lecturer VDF researcher, email: tutt@neu.edu.vn
Weaknesses
Only a small portion of VN firms
fund financed by issuing bonds
Low liquidity of bond market
Unattractive bond interest rate, much
lower than deposit interest rate
Types of bonds is undiversified
Lack of market maker
No credit rating agency
Solutions
- Issue convertible bond, bond with warrant,
bonds insured by gold or hard foreign
currency to protect investors profit from
high inflation
- Encourage private agency providing credit
rating services, such as model of potential
shareholders for credit rating agency
- Push up operating of market makers such
as banks, insurance companies
- Develop bond investments through
investment and mutual funds
Potential
shareholders
Multilateral agencies
International rating
agencies
Domestic institutional
Investors (e.g. pension funds)
Stock exchanges
Other market players (e.g. domestic
rating agencies, mutual funds, insurance
Companies, brokerage house, etc.)
Bank and financial
Institution operating
In the region
Potential shareholders for regional credit rating agency
Bond Trading Volume - Value
0
100
200
300
400
500
600
700
Q
1
+
2
/0
8
Q
3
+
4
/0
7
Q
1
+
2
/0
7
Q
3
+
4
/0
6
Q
1
+
2
/0
6
Q
3
+
4
/0
5
Period
V
o
l
u
m
e

(

M
i
l
.
)
0
10000000
20000000
30000000
40000000
50000000
60000000
70000000
V
a
l
u
e

(

M
i
l
.

V
N
D
)
Trading Volume
Trading Value
1) There is a robust negative relationship between initial poverty rate and
subsequent economic growth. However, there is no link between initial inequality
and subsequent economic growth.
2) Lower inequality leads to lower poverty rate, and poverty reduction could help
to reduce inequality. Other determinants of inequality and poverty reduction
include human capital, investment, GDP growth rate, and trade openness.
Given the empirical findings, the paper points out some policy
implications:
(i) Reducing poverty, apart from helping the most vulnerable segments of society,
is beneficial from the perspective of improving the future distribution of income
and promoting economic growth.
(ii) Concentrating on poverty elimination will help us to build a more equitable
society without sacrificing economic growth.
The Interrelationship between Growth, Poverty and
Inequality in Vietnam: An Empirical Analysis
Le Quoc Hoi (VDF Researcher and NEU Lecturer)
hoilq@vdf.org.vn
This paper examines the interrelationship between growth, poverty, and inequality in
Vietnam from 1996 to 2004 by using the provincial data and data from the household living
standard surveys which were implemented by the General Statistics Office of Vietnam (GSO)
in 1998, 2002, and 2004. The empirical results show that:
UniversityIndustry Linkages for
Improving Industrial Human Resource
Nguyen Thi Xuan Thuy (VDF Researcher) thuy@vdf.org.vn
Vietnam has abundant young laborers, but they usually have low
skills and poor experiences. As such, their unemployment rates are
among the highest.
Creating linkages between universities and industries is one of the
key factors for improving quality of university/college graduates.
This paper will take some successful cases from other countries,
such as Japan and ASEAN, to provide lessons for Vietnam in
strengthening such linkages, and improving its industrial human
resource (IHR).
Population Pyramid, 2005
6000 4000 2000 0 2000 4000 6000
0-4
10-14
20-24
30-34
40-44
50-54
60-64
70-74
80-84
90-94
100+
Male Female
Labor structure by qualifications, 1996 - 2005
0%
20%
40%
60%
80%
100%
96 97 98 99 00 01 02 03 04 05
Colleges & Univ.
TVET
High school
Secondary school
Primary school
Pre-primary school
Not going to school
Unemployment rate by age group, 1996 - 2005
0
2
4
6
8
96 97 98 99 00 01 02 03 04 05
Year
%
15-24
25-34
34-44
45-54
55-64
65+
Endogenous Devel opment in t he Era of Gl obali zat i on
~ Analysis of int eract ions bet ween Local Government s and Local Ent erprises in
t he Viet namese Cent ral Focal Economic Zone ~
Endogenous Devel opment in t he Era of Gl obali zat i on
~ Analysis of int eract ions bet ween Local Government s and Local Ent erprises in
t he Viet namese Cent ral Focal Economic Zone ~
I n my on-going research, I am using t he approach of Endogenous Development t o
analyze social, economic, and polit ical int eract ions bet ween Local Government s and Local
Ent erprises in t he Cent ral Focal Economic Zone, Viet nam.
For t his purpose, I chose Local Ent erprises in food processing indust ry as my research
t arget s. Because, in Viet nam, food sect or is labor int ensive and providing employment
opport unit ies as well as nut rit ious food t o t he local. As analyzing t heir dynamic int eract ions in
t he communit y, I t hink, CSR ( Corporat e Social Responsibilit y) , BOP ( Bot t om of Pyramid) , and
PPP ( Public Privat e Part nership) are import ant concept ual t ools. They are not only bet t er off t he
welt er of local people, but also become t he resource of power t o connect all st akeholders in t he
localit ies, including Local Ent erprises and Local Government s, t o bring int o sust ainable
development societ y.
I n my on-going research, I am using t he approach of Endogenous Development t o
analyze social, economic, and polit ical int eract ions bet ween Local Government s and Local
Ent erprises in t he Cent ral Focal Economic Zone, Viet nam.
For t his purpose, I chose Local Ent erprises in food processing indust ry as my research
t arget s. Because, in Viet nam, food sect or is labor int ensive and providing employment
opport unit ies as well as nut rit ious food t o t he local. As analyzing t heir dynamic int eract ions in
t he communit y, I t hink, CSR ( Corporat e Social Responsibilit y) , BOP ( Bot t om of Pyramid) , and
PPP ( Public Privat e Part nership) are import ant concept ual t ools. They are not only bet t er off t he
welt er of local people, but also become t he resource of power t o connect all st akeholders in t he
localit ies, including Local Ent erprises and Local Government s, t o bring int o sust ainable
development societ y.
PPP
DO My Hien
Graduat e School of I nt ernat ional Development , Nagoya Universit y
m070105d@mbox. nagoya- u. ac. j p
DO My Hien
Graduat e School of I nt ernat ional Development , Nagoya Universit y
m070105d@mbox. nagoya- u. ac. j p
Local Government s
Policies, Support s,
Training
Local Ent erprises
CSR, BOP
Endogenous
Development
Globalization
Is a Governmental Micro Is a Governmental Micro- -Credit Program Credit Program
for the Poor Really Pro for the Poor Really Pro- -Poor? Evidence from Vietnam Poor? Evidence from Vietnam
Nguyen Viet Cuong
(Faculty of Trade, National Economics University, Hanoi, Vietnam.
Email: c_nguyenviet@yahoo.com ; hanhhuong@fpt.vn
The paper examines the poverty targeting and impact of the
preferential credit program for the poor from Vietnam Bank for
Social Policies using data from Vietnam Household Living Standard
Surveys 2002 and 2004. The empirical results are:
The program targeting is not very pro-poor. Only 12% of the poor
households in rural areas participated in the program in 2004.
Meanwhile, the program covered 6.4% of the non-poor households.
Up to 67.1% of the participants were found as non-poor households.
The program has positive and statistically significant effect on
poverty reduction. The program decreases the poverty incidence for
its participants by almost 4 percentage points. The program also
reduces the poverty-gap and poverty-severity indexes by around 20
percent.
Technical Efficiency and Productivity in Vietnam Electronic Industry
A Comparison between foreign affiliated firms and domestic firms
Application of Parametric and Non-Parametric Analysis
The purpose of this research is to find out the performance between FDI foreign affiliated firms and domestic firms in
electronic sector. Based on parametric approach (Stochastic Frontier Production Function-SFPF) and non-parametric approach
(Input-based Data Envelopment Analysis-DEA), author investigates the technical efficiency, its determinants, productivity
growth. Author uses cross-section data of 658 firms to study of technical efficiency and that of determinants and uses panel
data of 38 firms in order to confirm the productivity growth (Malmquist Productivity Index) through out six years.
Tran Hoai Vu (Waseda University, Graduate School of Asia Pacific Studies)
tranhoaivu@gmail.com
The estimated technical efficiency on average for foreign affiliated firms and domestic firms were
60.58%; 56.2%, respectively, under the specification of SFPF model, and were 40.2%; 36.01%,
respectively, under the input-based DEA model. Most of firms in two sectors were having inefficiency in
term of optimal return to scale. The implication is that those of firms were better to reduce the production
cost by increasing return to scale.
According to two different models, the foreign affiliated firms perform higher technical efficiency than
domestic ones. Despites domestic firms show more rapidly increasing in investing capitals than in scale of
labor, the labor production of domestic firms still low in comparison with foreign affiliated firms.
Nevertheless, both of the foreign and domestic sector employed labor-intensive method of production.
In term of determinant factor affecting technical efficiency, beside of strong effect of capital-labor
ratios and labor productivity, the wage and experiences of firms have not so much impact. Especially, the
type of firm in aspect of foreign direct investment firms and domestic firm show very little influence to
technical efficiency performance.
In term of productivity growth, two sectors declined in TFP change, but the foreign sector improve its
productivity better than domestic sector.
During 2000-2006, domestic sector is catching up to foreign sector in aspect of improvement of
performance and productivity growth.












CONTRIBUTION
PAPERS

137
Modeling Economic Transformation in Vietnam
Using Smooth Transition Method


Nguyen Khac Minh
Vietnam-Netherlands Center for Development Economics and Public Policy (CDEPP),
National Economics University (NEU), Hanoi, Vietnam.
Email: khacminh@gmail.com


Nguyen Manh Hung
Toulouse School of Economics, LERNA-INRA, France.
Email: nguyen@toulouse.inra.fr


Nguyen Viet Hung
Faculty of Economics, National Economics University (NEU), Hanoi, Vietnam.
Email: hungnv@neu.edu.vn


Abstract
This study employs a Logistic Smooth Transition Regression to model the impact
of economic reforms in Vietnam from 1985 to 2006 in three economic sectors, i.e.
the agricultural sector, the industrial sector, and the services sector. We examine
the response of both GDP and labor productivity to the economic reform in each
of these sectors. While our results suggest that allowing for structural breaks in
GDP and labor productivity is preferred to either a trend stationary or a unit root
process, they also provide evidence of a positive impact of the reform on GDP
growth. The estimated results also indicate that the speed of the transition of the
agricultural sector between the initial and final states was faster than the others.
This implies that reforms in the agricultural sector succeeded in encouraging
people to work and also in quickly stabilizing the macroeconomy. Finally, we find
that the different reform policies undertaken had different impacts on the sectoral
growth in the Vietnamese economy.



138
1. Introduction
The transition from a planned economy to an open market economy began in 1986 after
the important reform decisions, known as Doi moi. A comprehensive restructuring and
renovation process is considered the main point of departure for Vietnams economic reforms.
The reforms introduced economic adjustments, including allowing rapid development of
private markets for agricultural goods, reforms in land allocation and the land-use rights to
formalize the fundamental changes in the agricultural sector, stabilization policies, entry
regulations and privatization, state enterprise restructuring, and tax and banking reforms.
The introduction of reform policies differed widely in scope, speed, and sequencing
across the industrial, agricultural and services sectors. The reforms adopted have given rise to
a number of questions, many of which relate to the impacts of the Land Law and recognition
of the long-term land-use rights in the agricultural sector, combined with price and trade
reforms which have contributed to sustained growth in agriculture. Furthermore, the issuance
and amendments of laws relating to government budgets, state and non-state enterprises, credit
and banking, the expansion of trade and financial relations within the international
community, and the new Enterprise Law all have a significant impact on the production and
business activities of private enterprises.
A related issue concerns how quickly these economic sectors recovered from the
implementation of the reforms in the economy. Furthermore, it is considered whether growth
has been higher following the policy reforms or prior to them, and whether any differences
across sectors in post-transition performance are related to the speed at which reforms were
undertaken. The main focus of this paper concerns this final aforementioned issue. To address
this we model GDP and labor productivity using a logistic Smooth Transition Regression,
LSTR and refer to LSTAR, models on a sample of three sectors during the period from 1985
to 2006.
The paper is organized as follows. Section 2 provides a literature review regarding the
methods and empirical studies. Section 3 presents the analytical framework, in which
econometric modeling is discussed in detail. A review on the performance of the Vietnamese
economy is presented in Section 4. We will present an analysis of the empirical findings in
Section 5, and then some concluding remarks are provided in the last Section.

2. A Review of Relevant Studies
There have been two major theoretical issues of importance in time series econometrics
over the past two decades. The first is modeling nonlinearity and structural changes in
economic series. The second considers the question of whether economic series are best
characterized as being stationary processes around a deterministic trend or as having a unit
root, a different stationary process. There are many reasons for nonlinearity in time series such
as changing policy regimes and technological changes. There have been many tests of
structural change designed to detect discrete shifts in the model parameters. Some of them
have considered the possibility of allowing any structural break to occur gradually. In this
study, we apply structural change using a smooth transition regression and refer to it as a
smooth transition autoregression regression (STAR). Following is a brief summary of the
relevant studies that used such a method.
Luukkonen et al. (1988) examined a general univariate smooth transition autoregressive
model. This was a special case of the threshold autoregressive (SETAR) model. They
presented three tests to examine linearity against STAR models and discussed their properties.
They showed that the power of the tests in small samples would be investigated by simulation
when the alternative was the logistic STAR model. They also concluded that the tests were

139
more powerful than the CUSUM test proposed for testing linearity against the SETAR model.
Similarly, McMillan (2005) considered whether a series of spot and forward exchange rates
exhibited smooth transition non-linear error-correction dynamic behavior, and their results
supported the STAR model.
The application of two families of nonlinear autoregressive models, i.e. the logistic
(LSTAR) and exponential (ESTAR) autoregressive, were considered in Teravirta (1994). He
studied the specification of the model based on simple statistical tests, including linearity
testing against smooth transition autoregression, determining the delay parameter and
choosing between LSTAR and ESTAR models. He also considered estimation by nonlinear
least squares and evaluating the properties of the estimated model. In another study, Teravirta
et al. (2005) examined the forecast accuracy of linear autoregressive, smooth transition
autoregressive (STAR), and new network (NN) time series models for 47 monthly
macroeconomic variables of the G7 economies. The authors used a single but dynamic
specification. They found that the results for neural network models were mixed in the sense
that at long forecast horizons, an NN model obtained using Bayesian regularization procedures
produced more accurate forecasts than a corresponding model specified using the specific to
general approach.
Eitrheim et al. (1996) used smooth transition autoregressive models for modeling
economic data. This paper contributed to the evaluation stage of a proposed specification,
estimation, and evaluation cycle of the models by introducing a Lagrange multiplier (LM) test
for the hypothesis of no error autocorrelation, and LM-type tests for the hypothesis of no
remaining nonlinearity and parameter constancy. Forster et al. (2007) employed a Logistic
Smooth Transition Regression to model economic reforms in CEECs. They allowed for
transitions in trends and levels, and then examined the response of both GDP and labor
productivity to reforms. They suggested that allowing for structural breaks in most CEECs
was preferred to either a trend stationary or a unit root process. They also suggested little
evidence of a positive impact of reforms on GDP growth. Examining the response of labor
productivity to reforms, they found that labor productivity provided stronger support for an
impact from reforms. They suggested that labor supply adjustments have been important in
increasing overall efficiency, and that differences in the speed in which reforms were
undertaken had an impact on the depth and the length of the transitional recession.
Stephen et al. (1999) considered the disinflation experiences of Australia, Canada, and
New Zealand. Their approach allowed the price data to determine the speed and timing of the
reforms using smooth transition analysis. They showed that the process of transition is related
to two factors: central bank independence and the general slowdown in economic activity
which occurred during the study period in all OECD countries. They found that other reforms
to the labor market and fiscal policy had been less influenced. Similarly, Greenaway et al.
(2000) tested whether GDPs in 12 European Union countries were integrated or stationary
around a deterministic component that might change gradually and smoothly between two
regimes over time. They found that in two-thirds of cases there appeared to be a role for
modeling with deterministic functions that allowed smooth transitions, in some cases standing
alone and in others in conjunction with additional integrated regressors. Their findings
constituted a challenge to traditional approaches of modeling trend-breaking behavior in GDP.
Traditional approaches typically suggested that breaks, when present have to occur
instantaneously.
Koster (2005) addressed the problem of non-linearity by applying smooth transition
autoregressive (STAR) specifications to existing simultaneous macroeconomic model of the
South African economy. The results supported the view that non-linear models provide better
forecasts than linear specifications of equations.

140
Following the study by Forster et al. (2007), we try to model the GDP and productivity
series for our sample of the industrial, agricultural and services sectors in the Vietnamese
economy in transition using the Smooth Transition Regression and the Smooth Transition
Autoregressive Models, in order to find out whether support to a trend stationary or a unit root
process. The following section gives the model specifications.

3. The Model Specifications
Switching regression model can be generalized as models in which the transition from
one regime to the other is not discrete but smooth, so that there can be a continuum of states
between extreme regimes. The speed of adjustment can be the type of nonlinear process. Thus,
Smooth Transition Regression Modeling can be illustrated follows.
{ }
t t t t t
t t t t
u z s c G z z
T t u s c G z z y
+ + =
= + + =
' ) , , (
,...., 1 ; ) , , ( ' '


(1)
where ( )
' '
,
t t t
x w z = is a vector of explanatory variables )' ,...., , 1 (
1
'
p t t t
y y w

= ,
and )' ,...., (
1
'
kt t t
x x x = , which is a vector of exogenous variables. Furthermore
)' ,...., , (
1 0 m
= and )' ,...., , (
1 0 m
= are ((m+1) x1) parameter vectors and
t
u iid (0,
2
).
The transition function ) , , (
t
s c G is a bounded function of the continuous transition variable s
t

, continuous everywhere in the parameter space for any value of s
t
, is the slope parameter,
and c (c
1
,c
2
,...,c
K
)' is a vector of location parameters, c
1
c
2
......c
K.
This study assumes that the model allows for a deterministic trend, with u
t
a zero mean I
(0) process. The transition function is a general logistic function, based on sample size T,
which is given as:
1
1
) ( exp 1 ) , , (

=
|
|

\
|
)
`

+ =

K
k
t t t
c s s c G , > 0. (2)
where s
t
is the transition variable, which in this study will be time, t, and c
t
is the location
parameter.
The most common choices for K are K=1 and K=2. For K=1, the parameter
{ } ) , , (
t t t
s c G z z + change monotonically as a function of s
t
from to +. For K=2, they
change symmetrically around the midpoint (c
1
+c
2
)/2, where this logistic function attains its
minimum value. The minimum lies between zero and 1/2. It reaches zero when and
equal 1/2 when c
1
=c
2
and <. Slope parameter controls the slope and c
1
and c
2
the location
of the transition function.
The LSTR model with K=1 (LSTR1 model) is capable of characterizing asymmetric
behavior. On the other hand, the LSTR2 model (K=2) is appropriate for situations in which the
local dynamic behavior of the process is similar at both large and small values of s
t
and
different in the middle.
When =0, the transition function ) , , (
t
s c G =1/2, and thus the STR model (1) - the
linear model. At the other end, when in the LSTR2 model, the result is another
switching regression model with three regimes such that the outer regimes are identical and
the mid-regime is different from the other two.

141
Shift in Intercept and Trend
The LSTR model allows for gradual changes in both the level and trend of the series of
interest. In this study, we allow for shifts in the intercept and trend growth rate of GDP, in
both cases allowing the shift to occur over time rather than instantaneously. The model is
illustrated as follows:
t t t t t t
s G s G t y Ln + + + + = ) , ; ( ) , ; ( ) (
1 1 0 0
, t=1, 2, T (1')
where the model allows for a deterministic trend, with
t
a zero mean I (0) process and
) , ; (
t t
s G the logistic smooth transition, based on sample size T, which is written as:
{ }
1
( exp( 1 ) , ; (

+ = T s s G
t t t
(2')
where s
t
is the transition variable, which in our case will be time, t. Under this formulation and
assuming that > 0, the model transition occurs smoothly between the initial state
+ + = t v t y Ln
t
, ) (
0 0
, and the final state
+ + + + = t v t y Ln
t
, ) ( ) ( ) (
1 0 1 0
corresponding to 0 =

G and 1 =

G ,
respectively. Hence the growth rate of Ln(y
t
). The coefficient on the trend variable changes
from
1 0 0
+ to over time. The model also allows the level to change from
1 0 0
+ to .
Smooth Transition Autoregressive Model (STAR)
Smooth transition autoregressive (STAR) models permit the autoregressive parameters
to transform slowly. Consider the following special nonlinear autoregressive (NLAR) model
t t t t t
c y G y y y + + + =

) , , (
1 1 1 1 1 0
. (3)
According to Enders (2004), if G () is a smooth continuous function, the autoregressive
coefficient (
1
+
1
) will change smoothly along with the value of y
t-1
. There are various forms
of the STAR model that allow for a changeable degree of autoregressive decay.
In the present study, we employ the logistic transition function, which has the general
form:
| |
1
1
) ( exp( 1

+ = c y G
t
(3')
Equations (3) and (3') jointly define the LSTAR model, where is the smoothness
parameter. As approaches zero or infinite, the value of is constant and the LSTAR model
becomes an AR (p) model. For other values of the extent of autoregressive decay depends on
the value of y
t-1
. Behavior of y
t-1
for intermediate values of y in the LSTAR model can be
illustrated as follows:
In the initial state
t p t p t t
y y y + + + + =

...
1 1 0
, y
t-1
- , and in the final state
t t t
y y + + + + + =

... ) ( ) (
1 1 1 0 0
, y
t-1
+ , corresponding to G 0 and G 1,
respectively. The intercept and the autoregressive coefficients change smoothly between two
extremes as the value of y
t-1
alters.
The LSTAR models are capable of generating asymmetric realization, which make them
an interesting tool for modeling macroeconomic time series, exhibiting, such as changes in
their dynamic properties over the business cycles
1
.

1
We will not analyze the results of LSTAR model here. See Appendix 3 for the estimated results from LSTAR
model.


142
4. A Review of the Performance of the Vietnamese Economy
In this paper we use the macroeconomic data collected by Vietnams General Statistics
Office (GSO) over the period of 1985-2006. The reforms included the rapid development of
private markets for agricultural goods; adjustments in land allocation, the land use rights
system, stabilization policies, entry regulations, and privatization; state enterprise
restructuring, and tax and banking reforms. The reform process in Vietnam has been
inherently experimental and gradual given the extremity of the changes. Doi moi in Vietnam
should be considered a learning process in which the leadership has continually responded to
the outcome of economic policies. Since Vietnam experienced many different economic
fluctuations and policy changes during the period of 1985-2006, we can deconstruct the study
period into four sub-periods, namely 1985-1988, 1989-1996, 1997-1999, and 2000-2006.
The first sub-period (1985-1988) was marked by the initial adjustments which aimed at
reducing macroeconomic instability and creating new economic incentives. Some of the major
reforms included abolishment of internal check points for the free movement of goods,
adjustment of prices towards unofficial levels and reducing rationings (such that the
Vietnamese Dong (VND) was evaluated in line with the parallel market rates), the approval of
the Land Law and recognition of long-term land use rights which encouraged farmers to work
(the Communist Party Resolution No.10, issued in April 1988). For example, farmers could no
longer be coerced into joining cooperatives and thus also were allowed to sell their products
on the open market. The launch of economic renovation created the good growth performance
of the economy. The economy started with a low growth rate of 2.8% in 1986 and then
increased in the following two years. The growth rate of the economy reached 6% in 1988. In
particular, the positive trend of GDP growth also strengthened the willingness of the
government to undertake further reforms.
In the second sub-period (1990-1996), the economy was on a high growth track, peaking
in 1995. Fast growth in this phase can be attributed to the effects of several reforms. Some of
the major reforms included the issuance and amendments of laws relating to government
budgets, state and non-state enterprises, credit and banking. Also important was the
encouragement of domestic and foreign investments, and expansion of trade. Take for
example the trade agreement signed with the European Union in 1992 for quota allocated
garment exports to the European Union and granting tariff preferences on selected imports.
There were also increased financial relations with the international community via
negotiations and further liberalizations. In particular, Vietnam joined the Association of
Southeast Asian Nations (ASEAN) in 1995 and commits to AFTA.
In the third sub-period (1997-1999), the major challenge to Vietnams young market
economy was apparent. The major reforms during this period included temporary prohibitions
on imports of a wide range of consumer good and approval of certain foreign investment
projects to be decentralized to selected provincial people's committees and industrial zones in
1997. However, the Asian financial crisis led to trade and investment disruptions. The
Vietnamese economy was not directly hit by this crisis due to strong capital controls.
However, a reduction in foreign direct investment (FDI) and intensified competition in export
markets brought about real blows to the economy. The economic growth rate declined
sharply from 8.2% in 1997 to 5.8 % and 4.8% in 1998 and 1999, respectively (CIEM, 2002).
In the fourth sub-period (2000-2006), when the financial crisis was dying down, the
economy resumed its growth momentum. After laying down the fundamental framework in
the previous sub-periods, the reform agenda focused on other structural reforms, including the
promotion of non-state sector and equitization of the state-owned enterprises (SOEs). The new
Enterprise Law enacted in 2000 has helped promote the private sector with more facilitation of
business activities for private enterprises . FDI law was amended to streamline procedures

143
clarify land -use right provisions in 2000. The Labor code was amended in 2002 to provide
more labor market flexibility. There was rapid growth in the number of newly established
enterprises, most of which were private ones.
Table 1: Growth Rates of GDP during the period 1985-2006
Period 1985-1988 1989-1996 1997-1999 2000-2006
P D G
&

0.04 0.08 0.06 0.08
P D G
&
I
0.1025 0.1126 0.0955 0.1024
P D G
&
S
0.0463 0.0802 0.0482 0.0692
P D G
&
A
0.0017 0.034 0.0436 0.0388
Notes: P D G
&
is growth rate of GDP during the period of studying; subscript I, S and A denote industrial sector,
service sector and agricultural sector, respectively.
Source: Authors estimates using GSO data.
Table 1 presents our estimates for the growth rates of GDP for the entire country and for
the industrial ( P D G
&
I
), service ( P D G
&
S
) and agricultural sectors ( P D G
&
A
), respectively. The
average annual economic growth in the study period was about 7 percent.
Table 2: Basic Statistics for Output
Output (VD billions) Share of Output (%)
GDP GDP
A
GDP
I
GDP
S
SGDP
A
SGDP
I
SGDP
S

Mean 221716 55078 74201 92438 27.3 30.8 41.9
Median 204700 52448 62783 89469 25.7 30.6 42.0
Maximum 425088 79488 174210 171390 38.4 41.0 43.8
Minimum 106176 40792 21351 44033 18.7 20.1 40.3
Std. Dev. 98689 13007 46806 39049 6.1 6.5 1.3
Observations 22 22 22 22 22 22 22
Note: GDP, GDP
A
, GDP
I
, and GDP
S
are real GDP of the whole economy, the agricultural sector, the industrial
sector, and the services sector, respectively. SGDP
A
, SGDP
I
, and SGDP
S
are the percentage of the agricultural
sector, the industrial sector, and the services sector, respectively, in the GDP.
Source: Authors estimates using GSO data.
Table 2 shows the basic statistics for the output of each of the three economic sectors and
the economy as a whole. In general, the contribution of the agricultural sector to the economy
gradually decreased over time, while those of industry and services increased significantly.
While the output of the services sector increased over time, its contribution as percentage of
GDP remained stable over the study period. In recent years, the industrial sector has become
the driving force of the economys growth.
So far we have concentrated on the dynamics of the level of GDP in each sector, which
is determined by macroeconomic forces such as investment behavior, government
consumption, export performance, and monetary policies. However, we may also expect an
impact on productivity following economic reforms. The effects on the overall efficiency of
each sector may be even stronger and take place faster due to underlying firm-level
adjustments like scrapping of inefficient technologies, improving production factors and other
general restructuring that has taken place in these sectors.
One way of examining the impact of reforms on productivity is by using the standard
accounting framework, including measures of the labor force and capital stock as independent
variables. By accounting for the contribution of capital and labor to GDP in each sector this
approach would allow one to isolate the impact of reforms on Total Factor Productivity, TFP.
However there are a number of problems with this approach. Most importantly, reliable data
on the capital stock in each sector in Vietnamese economy are not available. Moreover,
constructing capital stocks using the perpetual inventory method assumes a constant

144
depreciation rate is less valid in the case of each sector that was undergoing massive structural
changes involving the destruction of capital. As a result we use an alternative measure of
productivity defined as GDP to the active labor force. This measure of labor productivity
seems to be a useful measure to capture the path of overall efficiency.
Figure 2. Labor productivity of three sector overtime (1985=1) .
Labor productivity defined as GDP to the active labor force.










Source: Authors estimates using GSO data.
In Figure 2, we plot an index of productivity of labor for our sample of the three sectors.
The resulting productivity dynamics of each look rather different. The performance of the
industrial sector in terms of productivity has been better and more highly fluctuated than that
of other sectors. In the case of the agricultural sector, the productivity of labor experienced a
decline in the formative years of transformation but increased over time.
5. Empirical Findings
Testing for Unit Roots in GDP and TFP
Table 3: Values of unit root test
(natural log of real GDP of the economy and each sector)
ADF test KPSS test
p p
LnDP
I
-3.48* 4 0.0908 2
LnGDP
S
-3.62* 3 0.0954 3
LnGDP
A
-3.52* 2 0.1757 2
LnTFP
I
-1.912 3
LnTFP
S
-1.706 3
LNTFP
A
-2.787 2
Note: The first two columns of this table report the trend ADF tests on the log of real GDP. (2) Column (p) refers
to the number of lags included in ADF test. The number of lags was determined following the common approach
of testing down to find the appropriate number of lags, using t-statistic on the last coefficient as a guide. (3) The
final two columns report the KPSS test statistics.. Once again number of lags included in the Dickey-Fuller type
regression was determined using the t-statistic on the last lag. */**/*** Rejection of the unit-root hypothesis in
favor of a trend-stationary or a stationary process around a smooth transition in intercept and trend at the 10% ,
5% and 1% levels, respectively.
Source: Authors estimates.
We begin by considering the natural log of real GDP as the dependent variable. For GDP
series of the three sectors, the results from the ADF test indicate that we can reject the unit
root hypothesis in favor of trend stationary in all cases. For the productivity series of three
sectors, the results from the ADF test indicate that we cannot reject the unit root hypothesis in
Labor Productivity
0
0.5
1
1.5
2
2.5
3
3.5
4
1 3 5 7 9 11 13 15 17 19 21
Year
I
n
d
e
x

o
f

L
a
b
o
r

P
r
o
d
u
c
t
i
v
i
t
y
TFPA
TFPI
TFPS

145
favor of trend stationary in all cases. In these cases therefore we find little support for
modeling GDP using the LSTR model and as such we have chosen not to report the LSTR
results for this series. Table 3 presents results of the ADF test of stationary versus a unit root
along with the ADF test.
We applied the test presented above to the natural logarithm of the real GDP series,
measured at 1994 prices, to each of the three sectors. Each series contained 22 observations
from the period of 1985 to 2006. We also compute the standard with - trend ADF test, here
denoted by t*, to test the same null hypothesis as above, against the alternative that the series
is stationary around a fixed intercept and trend. The results are presented in Table 3. For each
series, the necessary degree of augmentation is given under p, this being selected by
conventional significant tests using 0.05 level critical values from a t distribution. A rejection
at the 0.10/0.05/0.01 test size is denoted by */**/***. Even using p-correction critical values
for the standard unit root test t*, the unit root hypothesis is rejected in favor of stationary
around a fixed trend.
In the case of the GDP of the economy and three sectors, The ADF and KPSS results
suggest that we can model GDP using a simple model with trend and intercept. While
considering the impact of economic reforms on the level and growth rate of the GDP in each
sector is important, a further question of interest is the extent to which reforms impacted
productivity. The way to proceed is analogous to that described above, beginning by
calculating the ADF and KPSS tests using the productivity data previously discussed. These
results are reported in Table 3, using labor productivity as our dependent variable. Using these
tests we can reject the null hypotheses of a unit root for all cases.
In the case of unit root series the modeling strategy is much different than the LSTR
model and requires us to first differencing the series and then building a model on it. We have
chosen not to estimate such a model since it adds little to the discussion of the impact of
reforms on economic performance. Nevertheless, the finding of a unit root for these series may
provide valuable information. From an economic point of view the presence of a unit root
implies that the impact of shocks on TFP have permanent effects.
Testing linearity and choosing the type of the model.
Before testing linearity, we fitted a linear model to our data and have the following
results:
For the agricultural sector:
) 001 . 0 ( ) 017 . 0 (
~
035 . 0 49 . 10 + + = t LnGDP
t
, T=22.
For the industrial sector:
) 001 . 0 ( ) 017 . 0 (
~
035 . 0 487 . 10 + + = t LnGDP
t
, T=22,
For the service sector:
) 001 . 0 ( ) 015 . 0 (
~
067 . 0 575 . 10 + + = t LnGDP
t
, T=22,
The ensuing identification problem in testing linearity can, in the STR context, be
circumvented by approximating the transition (2) in (1) by a Taylor expansion around the null
hypothesis =0. We assume K=1 in (2) and use the third-order Taylor approximation. The
resulting test has power against both the LSTR1 (K=1) and LSTR2 (K=2) models.
Assume that the transition variable s
t
is an element in z
t
and let )"
~
, 1 (
'
t t
z z = , where
t
z
~
is
an (mx1) vector. The approximation yields, after merging terms and reparameterizing, the
following auxiliary regression:
*
3
1
' '
0
~
t
j
t
j
t j t t
u s z z y + + =

=
, t=1,2....,T. (4)

146
When linearity has been rejected and a transition variable is subsequently selected, the
next step is to choose the model type. The variable choices are K=1 and K=2 in (2). When we
choose the LSTR1 model, the parameters change monotonically as a function of the transition
variable. The choice between these two types of models can be based on the auxiliary
regression (4). The coefficient vectors j
,
j=1,2,3 , in (4) are functions of the parameters in (1).
In the special case c=0, it can be shown that
2
=0 when the model is an LSTR1 model,
whereas
1
=
3
=0 when the model 1s and LSTR2 or ESTR2 model. Even when c# 0,
2
is
closer to the null vector than
1
or
3
when the model is an LSTR1 model, and vice versa for
the LSTR2 model. These suggest the following short test sequence:
1. Test the null hypothesis H
04
:
3
=0
2. Test H
03
:
2
=0|
3
=0.
3.Test H
02
:
1
=0|
2
=
3
=0.
If the test of H
03
yields the strongest rejection measured in the p-value, choose the
LSTR2 or ESTR model. Otherwise, select the LSTR1 model. All three hypotheses can
simultaneously be rejected at a conventional significance level such as 0.05 or 0.01 which is
why the strongest rejection counts.
In choosing the STR model type, either one of the two test sequences has proven useful
in practice. It is also possible to fit both an LSTR1 and an LSTR2 (or ESTR) model to the data
and make the choice between them at the evaluation stage. In practice, this is a sensible way of
proceeding if the test sequence does not provide a clear-cut choice between the two
alternatives if the p-values of the test of H
03
and of H
02
or H
04
are close to each other.
Table 4: Choosing model type
Sector Transition
Variable
F F4 F3 F2 Suggestion
t* 6.7113e-07 6.1250e-05 1.6252e-03 2.5140e-02 LSTR1
I t* 5.7266e-03 8.1528e-02 2.2070e-03 9.2517e-02 LSTR2
S t* 2.1884e-06 8.5290e-07 1.5211e-03 2.0316e-02 LSTR1
A t* 5.7477e-11 3.9339e-02 4.3596e-03 2.7184e-02 LSTR2
ote: F-statistics of null hypothesis H
04
, H
03
,H
02
are denoted by F4, F3,F2. I- industrial sector, S- service sector,
A- Agricultural sector.
Source: Authors estimates.
On the other hand finding support for the LSTR model suggests that shocks have no
permanent impact on series, with the exception of the shock captured by the smooth transition.
Where support for the LSTR model was found we proceeded to estimate the single LSTR
model. The results for GDP series of the three sectors are reported in Table 4. The F-statistic
testing of the null hypothesis of linearity was carried out using the LM-type test described in
Table 4. We suggest that one can reject the null of linearity in all cases at the 1 percent level.
Estimated Results from Modeling
Estimated results from LSTR1 model
The estimate results from LSTR1 model for the three series of GDP are reported in
Table 5. The coefficient
0
, according to the definition of the model is an estimate of the
initial logged values of GDP. The coefficient
1
is an estimate of the change in the intercept
following the structural break. As estimated, results of the coefficients on this variable is
negative in the case of the agricultural sector, indicating a drop in the level of GDP following
reforms in this area. In the two remaining sectors, the results of the coefficients on these
variables are positive, indicating an increase in the level of GDP following reforms in these
areas.

147
Table 5: Single transition results (natural log of real GDP) from LSTR model

0

1


Obs R
2
LnGDP
I
9.89 0.0918 0.085 0.0084 7.76 0.445 22 0.99
SE (0.118)*** (0.029)*** (0.0028)*** (0.003)***
LnGDP
S
10.61 0.110 0.052 0.0069 5.964 0.374 22 0.98
SE (0.052)*** (0.047)** (0.0097)*** 0.0095
GDP
A
10.61 -0.2005 0.0065 0.037 168.5 0.318 22 0.99
SE (0.006)*** (0.0091)*** (0.0016)*** (0.0017)*** 22
ote: (i) Standard-errors are reported in brackets; (ii) Obs. refers to the number of observations for
each sector; ** /*** Statistical significance at 5% , 1% levels, respectively..
Source: Authors estimates.
In terms of size of the coefficient on
1
we find that there is a great deal of variation
across sectors. Meanwhile, the agricultural sector suffered the greatest decline in GDP
(estimated coefficient of
1
was -0.2005 in the Table 5).
The coefficient on
0
provide an estimate of the trend growth rate prior to transition,
while the coefficient on
1
is an estimate of the change in trend growth associated with the
transition. Compared to Forsters study applying smooth transition regression to the data set of
Bulgaria, Czech Republic, Hungary, Latvia, Romania, and Slovakia, the estimated
coefficients of
0
here are more consistent with the expectation that the trend growth rate prior
to the transition would be positive in the three sectors. This suggests a positive growth rate
trend prior to transition. However, the main coefficient of interest for us regarding the impact
of reforms on the long-run growth rate trend is that of
1
. The coefficient of
1
is found to be
positive in all models. The results reported here generally suggest that there have been positive
impacts on the growth rate trend in all three sectors following reforms undertaken in Vietnam.
The reason may be the impact of the initial adjustments on creating new economic incentives
for the economy in general and farmers in particular.
The final two coefficients of interest are those of

and , indicating the position of the
mid-point and the speed of transition in the three sectors from 1985 to 2006, respectively.
This gives a clearer idea of when the estimated transition began in each sector and the speed at
which they moved between the initial and the final states. The coefficient provides an
estimate of the speed of transition, with larger coefficients signaling a faster speed of
transition. The results indicate that the speed of transition was particularly quick in the case of
agricultural sector (168.5) and much slower in the service sector (5.964). The faster speed of
transition in the agricultural sector could be explained by the impacts of the adjustments like
the Party Resolution No.10, passed in 1988. It provided farmers with property rights, price and
trade reforms, thus encouraging farmers (who comprise more than 70 percent of the countrys
population) to work on their farms. These not only contributed to sustained growth but explain
the faster speed of transition in agricultural sector.
The coefficients of indicates the positioning of the mid-point of transition in the three
sectors. The mid-points of the process of transition in each sector are calculated as 1993-1994,
1992-1993, and 1991 for the industrial sector, service sector and agricultural sector,
respectively. These results suggest that our model tends to capture the impact of
transformational reforms on GDP and that there are significant differences in the speed at
which sectors moved between economic regimes.

148
Table 6: Optimal transition results (natural log of real GDP) from LSTR model

0

1

1

2
Obs R
2
LnGDP
I
9.476 0.397 0.137 -0.037 5.29 0.289 0.602 22 0.99
SE (0.082)*** (0.083)*** (0.008)*** (0.008)***
Note: (i) Standard-errors are reported in brackets; (ii) Obs. refers to the number of observations for each sector;
**/*** Statistical significance at 5%, 1% levels, respectively.
Source: Authors estimates.
For the three series of GDP for which the LSTR model was supported we proceeded to
test for the optimal number of regimes, by testing for any remaining nonlinearities. In practice
this involves following the method suggested by Terasvirta (1996) and others who adapted the
LM-type linearity test. Although there has only been one fundamental change in regime from a
planned economy to a market economy, the Vietnamese government introduced important
policies quite late in the transition process, which may reflected by a second transition. The
results for the optimal number of transitions are reported in Table 6. For all case the data
suggests that there exists only one transition.
For the economy as a whole and the service sector specially, the data suggest that there is
only one transition, and hence the results are identical to those reported in Table 5. For the
remaining two sectors we found support for a double LSTR model. However, in the case of
the agricultural sector c
1
= c
2
and < , the result could not be reported in this table.
In Table 6 the coefficients that correspond closest to the period of transition are
presented. We can see that by allowing for a second threshold in the industrial sector the
results indicate a significant increase in trend growth following the transition, represented by
a negative and significant coefficient of
1
and a significant increase in the level of GDP (as
expected), which is signified by the positive and significant coefficient of
1
. However, the
positive effect of the transition on growth (in Table 5) is partially offset by a second structural
break, the mid-points of which are calculated as 1990-1991 and 1997-1998. The main reason
for this may be the impact of the Asian financial crisis in Asian countries (1997) that led to
trade and investment disruptions.

6. Concluding Remarks
In this study we used the LSTR model to examine the impact of reforms on the level and
growth rate of GDP and productivity of three sectors, namely the industrial, service and
agricultural. This model has an advantage over others because it allows for the impact of
reforms on economic performance through a smooth transition rather than as a discrete jump.
Using this model we considered whether and in what direction the movement towards the
market economy affected the growth rate of both GDP and labor productivity in the three
sectors of the economy. Given information about the speed in which reforms were
implemented, we were also able to consider the impact of the reform strategy on the speed of
transformation within each sector. Results indicate that in all cases for the GDP of each sector
the LSTR model was preferred over a trend stationary one. In the case of the TFP of each
sector we found evidence that supported the unit root hypothesis. From an economic point of
view the presence of a unit root in TFP series implies that the impact of shocks on TFP have
permanent effects.
The results of estimated smooth transition models for the GDP of each sector indicated
that there is evidence of one structural break the service and agricultural sectors but two
structural breaks in the industrial sector. Results from the LSTR model suggested that there
has been impact from reforms of the early 1990s on the trend growth rate of GDP.

149
It was very interesting to find that the speed of transition was particularly quick in the
case of the agricultural sector (168.5) and much slower in the service sector (5.964) .The mid-
points of the process of transition of each sector are calculated as 1993-1994, 1992-1993 and
1991 for industrial, service and agricultural sectors respectively. This shows that reforms in
agriculture succeeded in encouraging farmers to work on their farms and also succeeded in
quickly stabilizing the macro economy. Our results also suggested that the government's
policies in economic reform should bring benefit for people in the country in the long-term.

References
Andrew, Donald W.K (1993). Test for Parameter Instability and Structural Change with
Unknown Change Point," Econometrica 61, 821-856.
Eitrheim and et al (1996). " Testing the adequacy of smooth transition autoregressive models,"
Journal of Econometrics, 74, 59-75.
Enders, W (2004). Applied Econometric Time Series Wiley: USA
Forster, N and Robert Stehrer (2007). Modeling Transformation in CEECs using Smooth
Transitions," Journal of Comparative Economics 35, 57-86.
Granger, C.W.J and svirta a Ter & & , T (1996) Modeling onlinear Economic Relationship, Oxford
University Press: New York.
Greenaway, David, Leybourne, Steve, Sapsford, David (1997)." Modeling Growth (and
Liberalization) using Smooth Transition Analysis," Economic Inquiry 35, 798-814
Greenaway, David, Leybourne, Steve, Sapsford, David (2000)." Smooth Transition and GDP
Growth in the European Union," Manchester School 68, 145-165.
Koster, M (2005) " A macroeconomic model for South Africa: A non-linear economic
modeling approach" .
Leybourne, S and et al (1996) "Unit roots and smooth transition," Journal of time Series
Analysis, 19, 83-97.
Luukkonen, R and et al (1988) Testing linearity against smooth transition autoregressive
models," Biometrika 75, 491-499.
McMillan, D.G (2005) " Smooth-transition error -correction in exchange rates ," orth
American Journal of Economics and Finance , 16, 217-232.
Stephen, J and et al (1999) " Understanding the disinflation in Australia, Canada and New
Zealand using evidence from smooth transition analysis," Journal of International
Money and Finance 18, 799-816.
svirta a Ter & & , T (1994) Specification, estimation, and evaluation of smooth transition
autoregressive models," Journal of American Statistical Association, 89, 208-218.
svirta a Ter & & , T. et al (2005) " Linear models, smooth transition autoregressions, and neural
networks for forecasting macroeconomic time series: A re-examination," International
Journal of Forecasting 21,755-774.
T svirta a Ter , & & (2007) Smooth Transition Regression Modeling," in Applied Time Series
Econometrics Edited by Helmut tkepohl u L & & and Markus tzig a Kr & & , Cambridge University
Press.



150
Appendix 1: PKSS Test

The integration properties of a series y
t
may also be investigated by testing
t
y H :
0
I(0) and
t
y H :
1
I(1), that is, the null hypothesis that the DGP is stationary is tested against a unit
root. Kwiatkowski et al. (1992) derived a test for this pair of hypotheses. In the case that there
is no trend linear term, it can start from a DGP
t t t
z x y + = , where x
t
is random walk, x
t
=x
t-
1
+v
t
,
t
v ) , 0 (
2
v
IID , and z
t
is a stationary process. In this framework the foregoing pair of
hypotheses is equivalent to the pair: 0 :
2
0
=
v
H and 0 :
2
0
>
v
H .
The following test statistics was proposed:

=

=
T
t
t
S
T
KPSS
1
2
2
2

, where

=
=
T
j
j t
w S
1
with y y w
t t
= and
2

is an estimator of
|

\
|
=

=

T
t
t
T
z T Lim
1
1 2
var ,
That is,
2

is an estimator of the long-run variance of the process z

t
. Kwiatkowski et al.
(1992) proposed a nonparametric estimator for this quantity based on a Bartlett window with a
lag truncation parameter
4 / 1
) 100 / (T q l
q
=
|
|

\
|
+ =

+ =

= =

T
j t
j t t
l
j
j
T
t
t
w w
T
w w
T
q
1 1 1
2 2

1
2
1
, where
1
1
+
=
q
j
l
j
w . The critical values may be
found in Kwiatkowski et al. (1992). The null hypothesis of stationary is rejected for large
values of KPSS.
If a deterministic trend is suspected, the point of departure is a DGP
t t t
z x t y + + =
1
,
and the
t
w are residuals from a regression
t t
w t y + + =
1 0
. With these quantities the test
statistic is computed in the same way as before. Its limiting distribution under H
0
is different
from the case without trend term, however. Critical values for the case with trend are available
from them.






151
Appendix 2: Estimated Models for Three Sectors

LSTR1 models
{ }
9968 . 0 , 0135 . 0 , 22
) 445 . 0 )( / 76 . 7 ( exp 1 ) 0084 . 0 092 . 0 ( 085 . 0 899 . 9
2
1
1
) 88 . 1 ( ) 003 . 0 ( ) 029 . 0 ( ) 003 . 0 ( ) 0118 . 0 (
= = =
|

\
|
+ + =

R T
t t t LnGDP
t C

{ }
9935 . 0 , 0131 . 0 , 22
) 374 . 0 )( / 18 . 6 ( exp 1 ) 024 . 0 065 . 0 ( 043 . 0 52 . 11
2
1
1
) 91 . 1 ( ) 004 . 0 ( ) 027 . 0 ( ) 004 . 0 ( ) 0138 . 0 (
= = =
|

\
|
+ + =

R T
t t t LnGDP
t DV

{ }
9932 . 0 , 0069 . 0 , 22
) 318 . 0 )( / 51 . 68 1 ( exp 1 ) 0337 . 0 2004 . 0 ( 0065 . 0 61 . 10
2
1
1
) 91 . 1 ( ) 004 . 0 ( ) 009 . 0 ( ) 002 . 0 ( ) 0064 . 0 (
= = =
|

\
|
+ + =

R T
t t t LnGDP
t

Source: Own calculation.

152
Appendix 3: Estimated Results for LSTAR Model

0

1

c Obs. R
2
LnGDP
I
0.415 0.976 0.048 107.4 6.89 22
SE (0.0108)*** (0.0106)*** (0.0013)***
LnGDP
S
-1.113 1.11 0.00808 9.137 11.58 22
SE (0.327)** (0.03)*** (0.0024)***
GDP
A
0.265 0.976 0.0397 173 6.01 22 0.99
SE (0.173) (0.016)*** (0.0007)*** 22
Note: Standard-errors are reported in brackets
Source: Own calculation.


153
Does the Reproductive Health Project in Nghe An Province,
Vietnam Have Good Impacts on the Community?


Pham Thi Thu Huong
PhD candidate, Ritsumeikan Asia Pacific University
Email: huong72apu@yahoo.co.uk


1. Executive Summary
The research is a quasi-experimental survey on the JICA reproductive health project in
Nghe An province, Viet Nam. Research question is: Is the quality of reproductive health
services in project area better than the quality of reproductive health services in non project
area? The result of the analysis shows that the situation of facilities in project area is better
than the situation of facilities in non project area. The project has good impact on reproductive
health situation in Nghe An province.

2. Introduction
We provide background understanding of the JICA reproductive health project, where
the experimental survey was conducted.
Nghe An province locates in the north of Vietnam, about 300 km from Hanoi with the
capital of Vinh City. The area of the province is 16,487 km2 and the population is about 3
million, living in about 600000 households. The territory includes mountains, plains, and
costal lines. It is mainly agricultural province with 17 districts combining 466 communes.
Main agricultural products are rice, potato, peanuts, oranges, etc. The Reproductive Health
Project is the first Japans GO-NGO partnership project in Vietnam, handled by JOICFP a
Japan based international NGO on population and RH and JICA.
The project purpose is to improve the reproductive health status of women in
reproductive age with special focus on commune women in Nghe An province. The project
outputs include upgrading commune health centers, providing equipment to commune health
centers, and providing training to 244 midwives. Project are consists of 244 (among 264)
communes of 9 (among 19 districts of Nghe An province). The project period is from 1995 to
2000. Those districts are Nghia Dan, Con Cuong, Yen Thanh, Dien Chau, Anh Son, Do
Luong, Thanh Chuong, Nghi Loc, Nam Dan.
JICA team conducted the monitoring and evaluation work, in which the author was a
member. The quasi experimental survey was conducted in 2001. The Population Council,
particularly, Doctor Nhan and Doctor Mai designed the questionnaire and made inputs to the
raw data file. The author utilized the raw data file to evaluate the impact of the reproductive
health project in Nghe An province. It is different from the baseline survey of Doctor Mai. In
2001 the baseline survey on the reproductive health services provided by the public sector in
Nghe An province was conducted by Doctor Mai. The products of this survey included the
raw data 2001, the database 2001 and the report Baseline survey on the reproductive health
services provided by the public sector in Nghe An province of Doctor Vu Quy Nhan, Doctor
Le Thi Phuong Mai, and Nguyen Thi Thom. The result of the survey was much utilized by the
author to evaluate the performance of the project as a literature reviews item. In 2001 the

154
survey was also conducted in a non project area with the same variables. Those districts are
the remaining districts, both plain and mountainous, in Nghe an province. They have similar
economic and social conditions like the project districts. The difference is that there is no
project in those districts in 2001. Those districts are Vinh City, Cua Lo town, Que Phong, Quy
Chau, Ky Son, Quy Hop, Tuong Duong, Quynh Luu, Tan Ky, Anh Son, Hung Nguyen. The
purpose was to observe the quality of reproductive health services in the reproductive health
facilities of project area and to observe the quality of reproductive health services in the
reproductive health facilities of non project area. The comparison may say something. The
variables were taken from the output design of the project and were utilized to fill the
completion report in all items of the project framework: goal, objective, output, activities, and
verifiable indicators. The survey is financed by JICA fund.

3. Objective of the Quasi Experimental Survey
It is to evaluate if there is difference in reproductive health care services based on the
observation at facilities in project area and non project area.

4. Methodology of the Quasi Experimental Survey
To evaluate the impact of projects, quasi experimental survey method is used to compare
the situation of the project area and the situation of non-project area providing those areas
have similar economic and social conditions.
SPSS software with analyze, crosstab tool is utilized to count the percentage of yes
answer. The comparing mean of two independent samples is the best instrument for a quasi
experimental survey; however, it is not used in this research because the variables here are all
characteristic variables, not quantity variables. In order to compare mean for testing statistical
significance, we need to create new quantity variables from characteristic variables, which
seems so difficult and not effective. Comparing percentage of two independent variables help
us avoid the testing mean even though it takes more time. N is small (71 for project area and
66 for non project area), so that the comparing average value is used. Regression cannot be
used because of the small n.

5. Result of the Quasi Experimental Survey on Reproductive
Health in Nghe An Province
Chart 1: Telephone service
Telephone available
0 10 20 30
Frequent
None
%
Non-JICA
JICA

Source: Author made

155
The first pair of variables we compare is the observation at the reproductive health
facilities. Good telephone operation rate is 23.9% of total facilities in project are and 19.7% in
non-project area. The percentage of facilities without telephone is 18.2% in non-project area
and 4.2% in project area. That means telephone is more available in project area than that in
non-project area, as shown in the chart.
Frequent electricity rate is 84.5% in project area and 63.6% in non-project area. The
percentage of facilities without electricity is 21.2% in non-project area and 15.2% in project
area. Situation of electricity is better in project area, meaning electricity in non-project area is
not as available as that in project area.
Chart 2: Electricity available
Electricity available
0 50 100
Frequent
None
%
JICA(%)
NonJICA(%)

Source: Author made
The percentage of facilities with water from river, canal, and pond as primary source of
water is 25.8% in non-project area and 4.2% in project area. The percentage of facilities with
water from pipe, which is used to get water from streamline is 0% in project area while it is
9.1% in non-project area. The percentage of facilities with water from deep well is 16.9% in
project area and 7.6% in non-project area. Regarding primary source of water, situation of
project area is better than that in non-project area.
Chart 3: Water source of facilities
Water source
0 10 20 30
Pipe
Deep well
River
%
JICA
NonJICA

Source: Author made
Second pair of indicators for comparison is the sign or poster advertising the availability
of services. The percentage of facilities without any sign or poster for advertising the services
is 33.3% in non-project area and 5.6% in project area. The percentage of facilities with
advertising sign or poster inside the facility is 33.3% in non-project area and 63.4% in project
area. The percentage of facilities with the sign or poster both inside and outside the facility is
15.2% in non-project area and 18.3% in project area. It means the situation of project area is
better than the situation of non-project area.

156

Chart 4: Sign or poster
Sign or poster
0 20 40 60 80
INSIDE
BOTH OUTSIDE AND
INSIDE
THERE IS NO
%
JICA
NonJICA

Source: Author made
If we look at the percentage of facilities with obstetric, gynecological specialty included
in the signboard, we can see that the readiness of services is more available in project area.
The percentage of facilities, which have obstetric, gynecological specialty included in the
signboard, advertised inside the facility is 52.3% in non-project area and 68.7% in project
area. The percentage of facilities, which have obstetric, gynecological specialty included in the
signboard, advertised both outside and inside the facility is 9.1% in non-project area and
14.9% in project area. In contrast, the percentage of facilities, which has not ob/gyn specialties
in the signboard is 18.2% in non-project area and 9.0% in project area.
Chart 5: Ob/gyn poster
Ob/gyn poster
0 20 40 60 80
INSIDE
BOTH OUTSIDE AND
INSIDE
THERE IS NO
%
JICA
NonJICA

Source: Author made
The showing of midwifery specialty in the signboard says that the service in project area
is more available than that in non-project area. The percentage of facilities, which have
midwifery specialty included in the signboard, advertised inside the facility is 38.6% in non-
project area and 59.7% in project area. The percentage of facilities, which have midwifery
specialty included in the signboard, advertised both outside and inside the facility is 0% in

157
non-project area and 4.5% in project area. In contrast, the percentage of facilities, which have
no midwifery specialty in the signboard is 50% in non-project area and 32.8% in project area.
Chart 6: Midwifery service poster
Midwifery service poster
0 20 40 60 80
INSIDE
THERE IS NO
JICA
NonJICA

Source: Author made
Observing waiting seat for clients shows that the quality of services in project area is
better than that in non-project area. The percentage of facilities that have no waiting seats for
clients is 39.4% in non-project area and 28.2% in project area. In contrast, the percentage of
facilities that have enough waiting seats for clients is 27.3% in non-project area and 33.8% in
project area.
Chart 7: Waiting seat for clients
Waiting seat
0 10 20 30 40 50
No
Enough
%
JICA
NonJICA

Source: Author made
Let us observe the toilet availability. The percentage of facilities with working toilet is
63.6% in non-project area and 95.8% in project area. In contrast, the percentage of facilities
that have toilet but the toilet is not in use is 9.1% in non-project area and 4.2% in project area.
The percentage of facilities with no toilet is 27.3% in non-project area and 0% in project area.
That means toilet is more available in project area.

158
Chart 8: Working toilet
Working toilet
0 50 100 150
THERE IS
THERE IS NO
%
JICA
NonJICA

Source: Author made
Observation at the arrangement of facilities shows that facilities in project area are better
arranged than those in non-project area.
The percentage of facilities with a designed area for examination is 45.1% in project area
while it is only 27.3% in non-project area. Percentage of facilities with designed area for
counseling is 47% in project area while it is 39.4% in non-project area. Percentage of facilities
that have private area where people cannot see available for examination is 93% in project
area while it is less, 86.4% in non-project area. Percentage of facilities, which have private
area where people cannot see available for counseling is 40.8% in project area while it is only
33.3% in non-project area. Percentage of facilities where people cannot hear available for
examination is 87.3% in project area while it is 78.8% in non-project area. Percentage of
facilities with private area where people cannot hear available for counseling is 38% in project
area while it is only 30.3% in non-project area.
Chart 9: Working place
Working place
0 50 100
DESIGNED AREA FOR
EXAMINATION
DESIGNED AREA FOR COUNSELING
AREA PEOPLE CAN NOT SEE FOR
EXAMINATION
AREA PEOPLE CAN NOT SEE FOR
EXAMINATION
AREA PEOPLE CAN NOT HEAR FOR
EXAMINATION
AREA PEOPLE CAN NOT HEAR FOR
COUNSELING
%
JICA
NonJICA

Source: Author made

159
The percentage of facilities with clean examination area is 98.6% in project area while
it is 90.9% in non-project area. The percentage of facilities with adequate source of light in the
examination area is 95.8% in project area while it is less, 80.3% in non-project area.
Percentage of facilities with adequate amount of safe water available in the examination area
is 97.2% in project area while it is only 59.1% in non-project area.
Chart 10: Examination area
Examination area
0 50 100 150
CLEAN
EXAMINATION AREA
ADEQUATE
SOURCE OF LIGHT
IN THE
EXAMINATION AREA
SAFE WATER
JICA
NonJICA

Source: Author made
Percentage of facilities that have recovery room is 90.1% in project area while it is
75.8% in non-project area. The percentage of facilities, which have clean recovery room is
93.8% in project area while it is only 86% in non-project area. Percentage of facilities, which
have well ventilated recovery room is 97.2% in project area while it is 86.4% in non-project
area. Percentage of facilities, which have no dirt coming through is 91.5% in project area
while it is only 66.7% in non-project area.
Chart 11: Recovery room
Recovery room
0 50 100 150
THERE IS RECOVERY
ROOM
THE RECOVERY ROOM IS
CLEAN
IT IS WELL VENTILATED
NO DIRT COME THROUGH
%
JICA
NonJICA

Source: Author made
Percentage of facilities, windows of which have screen or curtain is 90.1% in project
area while it is only 54.5% in non-project area. Percentage of facilities, ceiling lamp of which

160
is working is 83.1% in project area while it is 71.2% in non-project area. Percentage of
facilities, spot lights of which are working is 83.1% in project area while it is only 34.8% in
non-project area. Percentage of facilities that have lamp/lantern to use when the current is out
is 54.9% in project area while it is 36.4%.
Chart 12: Facility
Facility
0 20 40 60 80 100
W
I
N
D
O
W
S

H
A
V
E

A
S
C
R
E
E
N
/
C
U
R
T
A
I
N
S
P
O
T

L
I
G
H
T
S
W
O
R
K
JICA
NonJICA

Source: Author made
Percentage of facilities that have container of already boiled water is 81.7% while it is
69.7% in non-project area. Percentage of facilities, which have a container of already boiled
water with a wooden hand longer than 20cm is 81.7% in project area while it is only 42.4% in
non-project area. Percentage of facilities, floor of which is clean is 94.4% in project area while
it is only 74.2% in non-project area. Percentage of facilities, which have cabinet for equipment
is 97.2% in project area while it is only 60.6 in non-project area.
Chart 13: Facility
Facility
0 20 40 60 80 100 120
CONTAINER OF
ALREADY BOILED
WATER
WOODEN HAND
LONGER THAN 20CM
THE FLOOR IS CLEAN
THERE IS A CABINET
FOR EQUIPMENT
JICA
NonJICA

Source: Author made

161
Observing equipment and materials at facilities we also see that readiness of
reproductive health services in project area is better than that in non-project area.
Percentage of facilities that have oxytocin or ergometrin is 100% in project area is 100%
in project area while it is 83.3% in non-project area. Percentage of facilities, in which there is
a kit for neonate resuscitation (catheter with 4-5 mm in diameter 20 ml bulb seringe for
suctioning infant, sterile compresses, two resuscitating balloon) is 85.9% in project area while
it is only 31.8% in non-project area. Percentage of facilities, in which resuscitation box is
stored in a separate sterile tray is 73.2% in project area while it is only 18.2% in non-project
area. Percentage of facilities, which have clean sterile cloth for covering instrument is 64.8%
in project area while it is only 36.4% in non-project area. Percentage of facilities, which store
boiled soap in closed box is 90.1% in project area while it is only 39.4% in non-project area.
Percentage of facilities, in which used water is drained into a closed tank outside labor room is
39.4% in project area while it is only 21.2% in non-project area.
Chart 14: Equipment
Equipment
0 20 40 60 80 100 120
OXYTOCIN OR
ERGOMETRIN
NEONATE
RESUSCITATION KIT
STERILE TRAY
CLOTH FOR
COVERING
INSTRUMENT
BOILED SOAP IN
CLOSED BOX
TANK FOR USED
WATER
JICA
NonJICA

Source: Author made
Observing floor of delivery room shows that the readiness of reproductive health in
project area is more available that that in non-project area. Percentage of facilities with
delivery room floor, paved with glazed brick in project area is 81.3% while it is only 42.4% in
non-project area. In contrast, percentage of facilities with delivery room floor, paved with
cement in project area is 16.9% while it is as much as 57.6% in non-project area. Glazed brick
is cleaner than cement.
The delivery room wall also tells us that the readiness of reproductive health services in
non-project area is not as available as it is in project area. Percentage of facilities with
delivery room wall paved with white whash is 8.5% in project area while it is as much as
60.6% in non-project area. In contrast, percentage of facilities with delivery room wall, paved

162
with glazed brick at 1.2m high is 91.5% in project area while it is 37.9% in non-project area.
Glazed brick is better than white whash in ensuring sanitary.
Delivery room ceiling also shows the sign of better readiness of services in project area.
Percentage of facilities with brick delivery room ceiling is 18.2% in non-project area while it
is 0% in project area. Percentage of facilities with bamboo mat, plywood, rain proof cloth
delivery room ceiling is 57.6% in non-project area while it is 43.7% in project area. In
contrast, percentage of facilities with cement ceiling in project area is 56.3% while it is 24.2%
in non-project area. Cement ceiling is better than bamboo mat ceiling and brick ceiling,
meaning the facility in project area is better than facility in non-project area.
Chart 15: Facility
Facility
0 50 100
CEMENT FLOOR
GLAZED BRICK FLOOR
WHITE WHASH WALL
GLAZED BRICK WALL
BRICK ROOF
BAMBOOMAT CEILING
CEMENT CEILING
JICA
NonJICA

Source: Author made
Chart 16: IEC material
IEC material
0 50 100 150
ANTENATAL CARE/DELIVERY
POST-PARTUM CARE
FAMILY PLANNING
CHILDWELFARE
MOTHER NUTRITION
RTI/STD PREVENTION
AIDS PREVENTION
JICA
NonJICA

Source: Author made
IEC (information, education, communication) material availability also shows that
services in project area is more available. Percentage of facilities, which have IEC material on
antenatal care and delivery care in project area is 95.8% while it is 92.4% in non-project area.
Percentage of facilities, which have IEC material on family planning is 97.2% in project area
while it is 89.4% in non-project area. Percentage of facilities, which have IEC material on
child welfare (immunization, child growth monitoring) is 94.4% in project area while it is

163
83.3% in non-project area. Percentage of facilities, which have IEC material on mother
nutrition in project area is 87.3% while it is 83.3% in non-project area. Percentage of facilities,
which have IEC material on RTI/STD prevention (reproductive track infection and abortion) is
80.3% in project area while it is 92.4% in non-project area. Percentage of facilities, which
have IEC material on AIDS prevention in project area is 95.8% while it is 92.4% in non-
project area. All indicators of project area show higher percentage than indicators of non-
project area.
Percentage of facilities, receiving IEC materials from JICA is 97.2% in project area
while this rate is only 36.4% in non-project area. Percentage of facilities, which have enough
IEC materials in project area is 42.3% while this rate is only 9.1% in non-project area. In
contrast, percentage of facilities, which have not enough IEC materials in non-project area is
75.8% while it is 47.9% in project area. Percentage of facilities, which have no IEC materials
is 15.2% in non-project area while this rate is 9.9% in project area. It is sure that in terms of
IEC materials, project area takes much advantage in comparison with non-project area.
Chart 17: IEC
IEC
0 50 100 150
IEC materials f rom JICA
ENOUGH IEC MATERIALS
NOT ENOUGH IEC
MATERIALS
NO IEC MATERIALS
JICA
NonJICA

Source: Author made
Observing public services, available at the facilities, we also can know that the readiness
of the services in project area is better than that in non-project area.
Percentage of facilities, which have permanent telephone service is 29.6% in project area
while this rate is 27.3% in non-project area. Percentage of facilities, which have telephone
service, but only sometimes because of many reasons, such as the telephone line is broken due
to typhoon or flood, is 1.4% in project area while this rate is 3% in non-project area.
Percentage of facilities without telephone service is 69.7% in non-project area while it is 69%
in project area. It shows that the situation of project area is better than the situation of non-
project area.
The same result comes out when we compare the situation of electricity power.
Percentage of facilities with frequently available electricity power is 95.8% in project area
while it is 78.8% in non-project area. Percentage of facilities with infrequently available
electricity power is 4.2% in project area while it is 7.6% in non-project area. Percentage of
facilities with no electricity power is 13.6% in project area while it is 0% in non-project area.
If we compare the situation of water source as natural water (river, canal, and pond) and
underground water (well), we also can see a better situation in project area. Percentage of
facilities with water source from deep well is 28.2% in project area while it is only 12.1% in
non-project area. In contrast, percentage of facilities with water source from river, canal, and

164
pond is 0% in project area while it is as high as 25.8% in non-project area. Readiness for
services is better in project area.
Chart 18: Facility
FACILITY
0 20 40 60 80 100 120
TELEPHONE
NOT REGULAR TELEPHONE
NO TELEPHONE
REGULAR ELECTRICITY
NOT REGULAR ELECTRICITY
NO ELECTRICITY
WELL WATER
NATURAL WATER
JICA
NonJICA

Source: Author made
Laboratory availability in the facilities shows better readiness of reproductive health
services in project area. Percentage of facilities with lab on vaginal discharge smear (wet
mount) is 11.3% in project area while it is only 6.1% in non-project area. Percentage of
facilities with lab on gram stain is 8.5% in project area while it is only 3.0% in non-project
area. Percentage of facilities with pregnancy test is 85.9% in project area while it is only
56.1% in non-project area. Percentage of facilities with urine analysis is 32.4% in project area
while it is only 22.7% in non-project area. Reproductive health services are more available in
project area.
Chart 19: Lab offered
LAB offered
0 20 40 60 80 100
VAGINAL DISCHARGE SMEAR
GRAM STAIN
PREGNANCY TEST
COMMON BLOOD TESTS
URINE ANALYSIS
JICA
NonJICA

Source: Author made

165
Comparing services offered at the facilities, we also discover that the readiness of
reproductive health services is more available in project area. Percentage of facilities, offering
maternity services all the time, including at night and at weekends is 98.6% in project area
while it is 97% in non-project area. Percentage of facilities, offering family planning
counseling services is 98.6% in project area while it is 97% in non-project area. Percentage of
facilities, offering menstrual regulation, abortion including management of abortion
complication services is 22.5% in project area while it is 18.2% in non-project area.
Percentage of facilities, offering antenatal care services is 100% in project area while it is
98.5% in non-project area. Percentage of facilities, offering postpartum care services is 100%
in project area while it is 95.5% in non-project area. Percentage of facilities, offering
gynecological examination care services is 98.6% in project area while it is 97% in non-
project area. Percentage of facilities, offering obstetric complication care services is 77.5% in
project area while it is 68.2% in non-project area. Percentage of facilities, offering RTI/STD
diagnosis and treatment services is 88.7% in project area while it is 86.4% in non-project area.
Percentage of facilities, offering HIV/AIDS testing services is 7.0% in project area while it is
6.1% in non-project area. Percentage of facilities, offering HIV/AIDS counseling services is
73.2% in project area while it is 71.2% in non-project area. Percentage of facilities, offering
child care services is 97.2% in project area while it is 93.9% in non-project area.
Chart 20: Services offered
Services offered
0 20 40 60 80 100 120
D33-MATERNITY SERVICES ANY TIME
FP COUNSELING
ABORTION
ANTENATAL CARE
POSTPARTUM CARE
GYN
OBSTETRIC COMPLICATION
STD/RTI
HIV/AIDS TESTING
HIV/AIDS COUNSELING
CHILD CARE
JICA
NonJICA

Source: Author made
Comparing the means of transportation the facilities use for emergency referral, we
know that readiness of reproductive services is more available in project area. Percentage of
facilities, using nothing for emergency referral, meaning the clients family members take care
by themselves is 81.3% in non-project area while this rate is 62.5% in project area. Percentage
of facilities, using bicycle for emergency referral is 6.3% in non-project area while it is 0% in
project area. In contrast, percentage of facilities, using car for emergency referral is 50% in
project area while this rate is 25% in non-project area. Percentage of facilities, using
motorbike for emergency referral is 31.3% in project area while this rate is 18.8% in non-
project area.


166
Chart 21: Means for emergency referral
Means for emergency referral
0 20 40 60 80 100
NOTHING
BY CAR
BY MOTORBIKE
BY BIKE
JICA
NonJICA

Source: author made
Chart 22: Medicine available
Medi cine
0 20 40 60 80 100 120
Diazepam
Magnesium sulfate
Dopegyt
Hypothiazid
Lidocain/xylocain
Spasmaverine
Papaverin
Epinephrin/Adrenaline
Promethazine
Agyrol eye drop
NitratAg eye drop
Clorua Natri 9%o
Glucoza
Antibiotic
Nyst atin
JICA
NonJICA

Source: Author made
Observing medicine availability, we see that medicine is more available in project area.
Percentage of facilities, which have Diazepam is 53% in non project area and 69% in project
area. Percentage of facilities, which have Magnesium sulfate is 27.3% in non project area and
56.3% in project area. Percentage of facilities, which have Dopegyt is 0% in non project area
and 9.9% in project area. Percentage of facilities, which have Hypothiazid is 60.6% in non
project area and 80.3% in project area. Percentage of facilities, which have Lidocain/xylocain
is 71.2% in non project area and 85.9% in project area. Percentage of facilities, which have
Spasmaverine is 6.1% in non project area and 19.7% in project area. Percentage of facilities,

167
which have Papaverin is 93.9% in non project area and 100% in project area. Percentage of
facilities, which have Epinephrin/Adrenaline is 78.8% in non project area and 98.6% in project
area. Percentage of facilities, which have Promethazine is 51.5% in non project area and
83.1% in project area. Percentage of facilities, which have Agyrol eye drop is 15.2% in non
project area and 28.2% in project area. Percentage of facilities, which have NitratAg eye
drop is 1.5% in non project area and 2.8% in project area. Percentage of facilities, which have
Clorua Natri 9%o is 78.8% in non project area and 83.1% in project area. Percentage of
facilities, which have Glucoza is 93.9% in non project area and 94.4% in project area.
Percentage of facilities, which have Antibiotic is 100% in non project area and 100% in
project area. Percentage of facilities, which have Nystatin is 72.7% in non project area and
88.7% in project area.

6. Conclusion
The conclusion is that there is difference in all indicators of the reproductive health
facilities in the project area and non project area. Indicators of project area show bigger
number than those on nonproject area. Quality of services in project area is better than the
quality of services on non project area. From this cross analysis we can conclude that the
project has good impact on the quality of reproductive health in Nghe An province.

Reference
Database of the JICA Reproductive Health Survey, Nghe An province
Vu Quy Nhan, Le Thi Phuong Mai, Final Assessment on the Reproductive health Project in
Nghe An province, Hanoi, 2005.


168
Using Vietnams I-O Table 2000
to Identify the Key Economic Sectors


DINH Thi Hoang Yen and OSADA Hiroshi
Graduate School of International Development, Nagoya University
Furo-cho, Chikusa-ku, Nagoya-shi 464-8601, Japan
Email: dinhhoangyen@yahoo.com


Abstract
The Vietnams industrial policy is aiming at structural adjustment, renovation of
production relationship, and modernization of industrial management. The paper
applies several analytical methods for input-output table (IO2000), namely
Chenery-Watanabe, Rasmussen, total linkage, and pure linkage to define the key
sectors in Vietnams economy. If the values of both backward linkage and
forward linkage of a sector are above the corresponding averages, the sector is
called key sector. However, different methods results in dissimilar groups of key
sectors, , The key sectors identified by Chenery-Watanabe and Rasmussen
methods are intuitively heavy industries while by total and pure linkage methods
are prone to light industries. Basic sectoral indicators from IO2000 are also
scrutinized to support the key sector analysis results.

169
1. Introduction
The input output table (IO table
1
), also called the inter-industrial transaction table, was
first created by Wassily Leontief to show the demand for industry by adding up all industries
demand and produce supply according to that aggregate demand as the idea of Soviet social
planning (Sadoulet, 1995). The purposes of constructing and analyzing the IO table are for
both practical planning task and academic evaluation task and up to now the theoretical
background for the IO table remains its predilections. The prominent advantage of quantitative
method using the IO table is that it can capture the influence between sectors in the economy
thought using the static input-output coefficients and fixed price unit of all accounts.
The present work tries to utilize the latest available data of the IO table of Vietnam in
2000 to identify the key sectors which have both strong backward and forward linkages in the
economy. The similar approach has been employed by Andreosso-OCallaghan and Yue
(2004) to identify the intersectoral linkages and key sectors in China for 10 years (with the IO
table of 1987, 1990, 1992, 1995, and 1997 and 28 aggregated sectors). The strength of
backward and forward linkages changed according to different IO tables, therefore the final
conclusion of the key sectors for the whole 10 years is considered by the common result of the
four methods. The core sectors comprise agriculture, traditional manufacturing (textiles,
chemicals, building materials, primary metals, and machinery), commerce and other services.
Another interesting conclusion is that the degree of intersectoral interference had increased
along with the economic growth in China over the past 10 years to evocate an analogous
implication for Vietnam.

2. Analytical Methods
Assumption of the IO table model is that the amount of sector is output required for the
production of sector js output Xij is proportional to sector js output Xj. Input coefficient aij
(and output coefficient bij) shows the share of various input for one unit of production Xj and
is calculated by:
(1) aij = Xij/Xj
(2) bij=Xij/Xi
So, write in matrix form, the IO model is as follows:
(3) AX +Y = X (demand-driven model)
(4) BX+V=X (supply-driven model)
where
X is row vector output, Xi, i=1, , n,
170
X is column vector input, Xj, j=1,.m,
A is the matrix of input coefficient aij, i, j=1, ., n,
B is the matrix of output coefficient bij, i, j=1,..m,
Y is the column vector of final demand, yij, i, j=1,n,
V is the row vector of value added, vij, i, j=1.m,
I is the unit matrix, which has ones on its diagonal and zeros everywhere else,
G=(I-A)
-1
is the Leontief inverse matrix, multipliers which can be used to calculate
overall changes in sectoral outputs resulted from changes in final demand,
Z=(I-B)
-1
is the Ghosh inverse matrix, multipliers which is developed to compute overall
changes in sectoral inputs resulted from changes in output.
Input output multipliers are used to calculate the linkages in the economy based on the
backward and forward linkage concept of Hirshman (1958). Backward linkage is the
proportion of an industrys inputs that is purchased from other industries. Backward linkages
2

are measured by:
(5)

=
= =
=
= =
n
i
ij
R
j
n
i
n
i
ij
j
ij C
j
g BL
a
x
x
L
1
1 1
B

Forward linkages
3
are computed by using a matrix of supply coefficients obtained by
dividing the row of intermediate inputs supply by the total supply of a sector.
(6)

=
= =
=
= =
n
j
ij
R
i
n
i
n
j
ij
i
ij C
i
g FL
b
x
x
L
1
1 1
F

But total linkages cannot be defined as the sum of the backward and forward linkages
since they rely on some inconsistent theories. Alternatively, the total contribution of a sector to
the economy has been measured with the extraction method which compares the observed
total output with the output that would be obtained with the same vector of final demand. The
difference between these two output levels measures the total linkage effect of each sector.

1
Vietnams IO table at producer price was published by the General Statistic Office in 2000. The original IO
table 2000 consisted of 112 sectors and was aggregated to 22 sectors in the Appendix 1.
2
BL of Chenery-Watanabe refers to the magnitude of sector is output used as production by sector j while that
of Rasmussen measures the extent to which an unit change in the demand for the product of sector j causes
production increase in all sectors.
3
FL of Chenery-Watanabe indicates the ratio of intermediate delivers sold by sector i and its output whereas that
of Rasmussen assesses the extent to which sector i is affected by an expansion of 1 unit in all sectors.
171
Suppose that all sectors of an economy are partitioned into two groups: (1) consist of the
sectors that are to be extracted from the economy and (2) encompasses all the remaining
sectors of the economy. Given the existence of two sectors, the basic equation (2) of Leontief
model may be rewritten as:
(7)

2
1
2
1
22 21
12 11
2
1
y
y
x
x
A A
A A
x
x

Or
(8)

+
=

2
1
22 12 21 22 21 22
22 12
2
1
( y
y
G HA A I G H A G
G HA H
x
x

where
1
22 22
1
21 22 12 11
) (
) (

=
=
A I G
A G A A I H

Submatrices A
12
and A
21
indicate the relationships between sector 1 and sector 2 in
production.
Hypothetically extract sector 1 entirely from the economic system, that is, A
11
= A
12
=
A
21
= 0, equation (7) and (8) then can be rewritten as equation (9) and (10) as follows:
(9)

2
1
2
1
22
2
1
0
0 0
y
y
x
x
A
x
x
Or
(10)

2
1
1
22
2
1
) ( 0
0
y
y
A I
I
x
x
The difference between the outputs before the extraction (x) and the outputs after the
extraction (

x ) can be expressed as:
(11)

2
1
22 12 21 22 21 22
22 12
2 2
1 1
y
y
G HA A G H A G
G HA I H
x x
x x

Thus the total backward linkage effect (TBL) can be defined as:
(12) [ ] [ ]
22 22 12 21 22
'
2 22 12
'
1 1 21 22
'
2
'
1
'
) ( ) ( y G HA A G e G HA e y H A G e I H e x x e TBL + + + = =


Where e is summation of column vector, e
1
and e
2
are column summation vectors for sector 1
and vector 2, respectively.
Pure backward linkage effect (PBL) is identified by:
172
(13)
[ ]
1 21 22
'
2
x A G e PBL=

Or
(14)
[ ] [ ]
2 22 12 21 22
'
2 1 21 22
'
2
y G HA A G e y H A G e PBL + =

Because from (8) when extract sector 1, we have
(15)
2 22 12 1 1
y G HA Hy x + =

By the same arguments, the equations for total forward linkage (TFL) and pure forward
linkage (PFL) are derived as:
(16)

+ +

+ = =

2 22 12 21 22 1 21 22 2 2 22 12 1 1
) ( ) ( e Z B H B Z e H B Z v e Z B H e I H v e x x TBL

(17)
2 22 12 21 22 2 2 22 12 1
e Z B H B Z v e Z B H v PBL

=


where
1
22 22
1
21 22 12 11
) (
) (

=
=
B I Z
B Z B B I H


3. Results
3.1. Key sectors by Different Methods
3.1.1 Backward Linkages
Applying Chenery-Watanabe and Rasmussen methods, industries having strong
backward linkages are leather, textiles, and metals. In other words, the effects of an increase in
final demand of these sectors on overall output are the greatest. While by total linkage and
pure linkage methods, industries have strong backward linkages are food processing, other
manufactures, textiles. This means the impacts of these sectors on the production of other
sectors are the strongest (further details in Appendix 2).
3.1.2 Forward Linkages
Using Chenery-Watanabe and Rasmussen methods, industries having strong forward
linkages are metals, other manufactures, rubber and plastics, and chemicals. An increase in
output of all sectors will mostly affect those sectors. All these industries belong to the primary
and heavy industry sectors. As the results, final demand grew faster than intermediate demand
for these sectors. Whereas, the total and pure linkage methods indicate that the strong
backward linkage industries are food processing, trade, agriculture, and other services. So the
impacts of production of other sectors on these sectors are biggest if those sectors were
hypothetically extracted
(Appendix 3)
173
3.1.3 Identification of Key Sectors
Key sectors are termed as industries whose values of both backward and forward
linkages of a sector are above the corresponding averages. Therefore, key sectors by Chenery-
Watanabe and Rasmussen are quite resemblance: metals, other manufactures, electrical and
telecom equipments, transport means, other machinery. Likewise, the key sector which shares
the consensus of total and pure linkage is food processing. It is likely that hypothetical
extraction methods expose the more realistic result than the traditional methods.
3.2. Sectoral Contrast
Agriculture, leather, and trade sectors generate the most output to the economy (17%,
14%, and 11% respectively). However, in terms of the share of value added over total output,
crude oil, electricity, gas and water, and other services sectors have the highest share (74%,
71%, and 70% respectively). Concerning the contribution of value added to gross value added,
agriculture, trade, hotel, and tourism, and crude oil have the biggest share in gross value added
of the economy (26%, 14%, and 9% respectively).
In general, labor income takes more than half of the sectors value added, especially the
sector which are highly labor-intensity have the greatest share of labor income are public
services (82%), agriculture ( 76%), and other mining (58%). Taxes on production are collected
in most sectors using raw materials such as food processing (33%), leather (21%), and wood
and paper (20%). Profits, however, are earned mainly in sectors enjoying the subsidy from the
government such as electricity, gas and water (42%), crude oil (41%), and wood and paper
(36%).
(Appendix 4)
3.3. Import Export Dependency
Import plays a crucial role for many sectors, notably machinery, electrical and
telecommunication equipments, and other manufactures with the lion share of imports out of
total domestic supply (464.3%, 271.2%, and 141.5% respectively). These sectors also take the
large share of total imports (8.3%, 16.9%, and 11% correspondingly) while other primary
sectors produce the most export volume, namely crude oil (20.5%), food processing (15.3%),
and textiles and clothes (12.6%).
Regarding to import tariff, sectors which have the highest effective rate of protection are
also the highest contributors to the import tariff revenue; e.g. other manufactures (31.3% and
17.2%), crude oil (25% and 19.1%), and food processing (55.6% and 17.1%).
Intra industry trade occurs mostly in textiles and clothes, other mining, and wood and
paper (92.3%, 91.3%, and 90.4% accordingly).
(Appendix 5)

174
4. Discussion
It is not surprised that food processing is the key sector in Vietnam because of its old-
age agrarian economy. But this also implies the severe problem of rural development which
could lead to large impact on both backward and forward linkages of food processing industry.
As one can see from the sectoral analysis, food industry ranks among the top output
production and tax revenue. However, labor income and profit of this sector were still low,
partly due to the low-priced labor cost and food price in Vietnam. Although protection for
food processing exports has been kept high, the intra trade indicator (IIT) suggests that
Vietnam currently imports input for food processing.
The results of key sectors by traditional methods should also be mentioned because they
are all capital intensive and industrial-biased sectors, notably the electricity, gas and water
sector serving as public goods in Vietnam. Compared to food processing industry, on average
metals, other manufactures, electrical and telecom equipment, transport means industries
contributed lower output and tax, but had higher value added, labor income, and profit. In
general, they tend to inward rather than outward with smaller exports but bigger imports,
implying excess domestic demand for those industries.
Along the path of Vietnamese economic development, the interlinkages between
industries will be more and more important. This paper is just a small attempt to provide some
implications for Vietnams industrial policy and hope to draw attention of relating studies.

5. References
[1] Andreosso-OCallaghan B. and Yue G., Intersectoral linkages and key sectors in China,
1987-1997, Asian Economic Journal, Vol. 18, No.2, pp. 165-183, 2004.
[2] Hirschman, A. O., The Strategy of Economic Development, Yale University Press, New
York, 1958.
[3] Sadoulet, Janvry, Quantitative development policy analysis, The Johns Hopkins University
Press, Baltimore and London, 1995.

175
Appendix 1. Sectoral Mapping of Vietnam IO Table 2000
Agriculture
1 Agriculture
Mining
2 Crude oil
3 Other mining
Manufactures
4 Food processing
5 Wood and paper
6 Non-metallic minerals
7 Chemicals
8 Rubber & plastics
9 Transport means
10
Electrical &telecom.
Equipments
11 Other machinery
12 Metal
13 Textiles & clothes
14 Leather
15 Other manufactures
Public utilities
16 Electricity and gas & water
17 Construction
Services
18 Trade, hotel & tourism
19 Transport
20 Finance
21 Public services
22 Other services


176
Appendix 2. Vietnam IO2000s Backward Linkages
Resulting from Various Methods
Sector Chenery-Watanabe Rasmussen Total linkage Pure linkage
1 0.3666 2.1026 24.3471 0.0722
2 0.2580 1.6533 1.6555 0.2620
3 0.4596 2.4070 3.1904 0.2904
4 0.8141 2.9953 119.8027 17.5213
5 0.9196 3.6243 29.8642 3.8424
6 0.7412 3.3557 16.0045 0.9405
7 0.6731 3.1342 32.5075 0.7561
8 0.6822 3.0938 26.9458 0.6583
9 0.7264 3.7657 36.1370 0.6128
10 0.7901 4.0578 33.6277 0.7797
11 0.7061 3.6868 10.9127 0.5636
12 0.8294 4.3459 46.9341 0.6421
13 0.8144 4.1341 87.6249 1.5899
14 0.8358 4.3496 39.0656 0.9325
15 0.8145 3.7245 51.2030 3.6268
16 0.2937 2.0054 3.0764 0.3353
17 0.7323 3.6002 9.4655 1.6352
18 0.4683 2.3988 5.0426 0.7473
19 0.4845 2.6086 3.7247 0.5804
20 0.3257 1.8195 3.1652 0.2494
21 0.3967 2.1915 8.9070 0.9046
22 0.3813 2.1485 6.6210 0.6914
Average 0.6143 3.0547 27.2648 1.7379

177
Appendix 3. Vietnam IO2000s Forward Linkages
Resulting from Various Methods
Sector Chenery-Watanabe Rasmussen Total linkage Pure linkage
1 0.5501 4.9678 13.8634 0.1980
2 0.0544 1.1302 0.0665 0.0525
3 0.9453 1.8963 0.1165 0.0047
4 0.2420 3.3544 4.3422 3.6390
5 0.7939 2.6051 0.3026 0.1371
6 1.1056 2.7457 0.4993 0.0394
7 1.6313 4.6556 0.6119 0.0141
8 1.6321 4.9647 0.4489 0.0105
9 1.0432 3.5057 0.8240 0.0121
10 1.3515 4.2871 0.4381 0.0098
11 1.8276 2.0753 0.0915 0.0033
12 1.6171 7.8904 0.5711 0.0087
13 0.6041 3.7105 1.6177 0.0599
14 0.5339 2.4754 0.6247 0.0254
15 2.5204 6.3317 0.3874 0.0653
16 0.7657 2.4886 0.1795 0.0182
17 0.0000 1.0000 0.1981 0.1981
18 0.2748 1.2892 1.5131 0.6631
19 0.2833 1.1293 0.0451 0.0275
20 0.6901 1.4154 0.1961 0.0077
21 0.1058 1.2032 0.7058 0.1826
22 0.5483 2.0818 0.8987 0.1977
Average 0.8691 3.0547 1.2974 0.2534

178
Appendix 4. Sectoral Contrasts in the Generation of Income
and the Role of Sectors in Domestic Production
Sector Output/output VA/output VA/VA Labor income/VA Tax/VA Profit/VA
1 17% 67% 26% 76% 8% 10%
2 5% 74% 9% 39% 5% 41%
3 1% 54% 1% 53% 10% 28%
4 14% 21% 6% 34% 33% 24%
5 2% 25% 1% 31% 20% 36%
6 3% 29% 2% 34% 16% 28%
7 2% 31% 2% 52% 9% 28%
8 1% 32% 1% 51% 8% 27%
9 2% 27% 1% 53% 11% 26%
10 2% 24% 1% 49% 10% 27%
11 0% 29% 0% 58% 10% 23%
12 2% 17% 1% 48% 14% 17%
13 5% 19% 2% 39% 20% 25%
14 3% 17% 1% 42% 21% 22%
15 1% 27% 1% 47% 7% 26%
16 2% 71% 3% 49% 7% 42%
17 9% 27% 6% 47% 11% 29%
18 11% 54% 14% 50% 13% 30%
19 2% 51% 3% 54% 9% 23%
20 1% 68% 2% 49% 15% 30%
21 7% 61% 9% 82% 2% 6%
22 6% 70% 9% 53% 12% 22%

179
Appendix 5. Import Dependency, the Structure of External Trade,
Import Tariff Rates, Effective Rate of Protection, Intra-industry Trade
Sector M/domestic supply M/M E/E Tariff/M ERPj=(tj-aij*ti)/(1-aij) IIT
1 2.31% 1.53% 10.92% 4.20% 3.73% 26.2%
2 0.41% 0.08% 20.45% 19.14% 24.97% 0.7%
3 26.34% 0.82% 0.76% 1.14% -1.80% 91.3%
4 6.70% 3.67% 15.34% 17.10% 55.59% 37.3%
5 27.50% 2.26% 2.76% 11.45% 24.89% 90.4%
6 15.49% 1.72% 0.44% 2.96% -2.45% 38.3%
7 106.43% 10.70% 0.63% 2.58% -2.82% 10.1%
8 80.70% 4.80% 0.39% 4.42% 4.02% 13.9%
9 88.36% 8.48% 0.58% 16.69% 26.55% 13.4%
10 141.48% 11.06% 3.66% 4.10% 2.46% 50.7%
11 464.26% 8.34% 3.37% 2.90% -2.99% 49.7%
12 87.27% 7.43% 1.00% 2.80% -4.59% 22.0%
13 50.53% 9.96% 12.57% 3.66% 1.51% 92.3%
14 18.26% 2.80% 9.74% 4.07% 2.87% 47.3%
15 271.19% 16.93% 3.26% 17.18% 31.25% 33.5%
16 2.79% 0.20% 0.00% 0.00% -3.66% 0.0%
17 0.00% 0.00% 0.00% - - -
18 18.01% 1.80% 5.82% 0.00% -6.77% 50.3%
19 32.99% 1.63% 3.16% 0.00% -9.95% 75.6%
20 58.71% 2.59% 2.06% 0.00% -1.76% 83.2%
21 12.02% 2.78% 1.10% 0.00% -3.58% 52.3%
22 2.16% 0.43% 1.98% 0.00% -2.52% 38.2%


180
Tourism Destination Image:
Vietnam as a Tourism Destination for Japanese Travelers


LE Tuan Anh
*

E-mail: anhtle07@apu.ac.jp; anhtle77@gmail.com


Abstract
This study identifies and measures the images of Vietnam as a tourism destination
in the Japanese mind. The results show that the presentation of Vietnam in Japan is
similar to that of almost any other destinations covering natural landscapes,
historical places, typical architectural monuments, cultural activities, and local life.
The most unique images representing Vietnam are Vietnamese women wearing
coconut hats, traditional Ao dai (Long dress), and some traditional food dishes.
Vietnam is highly appreciated in low prices, delicious cuisine, rich culture, new
experience opportunities, and friendliness. However, the country is depreciated in
personal safety, political stability, quality of service, healthcare system,
infrastructure, and cleanliness. Words of mouth, travel guidebooks,
pamphlets/brochures, television, and internet are the most influential information
sources on the formation of the Vietnamese images in Japan. Some implications
for tourism marketing policy are also included.

Key words: Tourism, destination, image, Vietnam, Japanese, travelers
JEL Code: L83

*
LE Tuan Anh is now a doctoral student at the Graduate School of Asia Pacific Studies, Ritsumeikan Asia Pacific
University (APU), Japan. He worked for the Vietnam National Administration of Tourism (VNAT) from 2001 to
2005. His interest of study is about national tourism destination marketing and Japanese tourists.

181
1. Introduction
In the world tourism industry, Japan has always been a major tourism generating
market sending tourists in large numbers as visitors to countries around the world
(UNWTO, 2005a). Over the past more than ten years, there has been a high average of
approximately 16 million Japanese people annually traveling overseas, and this number is in
an increasing trend (JNTO, 2005 & JTM, 2006) (see Figure 1). The Economist Intelligence
Unit (EIU) forecast that the volume of Japanese outbound tourists might surpass 20 million by
2010 onward (EIU, 2007). Moreover, Japanese tourists belong to the group of the highest
spending capacity (UNWTO, 2005b).
In Vietnam, Japanese tourists are regarded one of the target markets (Vietnam, 2001),
and Vietnams tourism industry has achieved initial success in attracting Japanese tourists. The
number of Japanese tourists to Vietnam rapidly increased in 2000, 2001, and 2002 (VNAT,
n.d.), hitting a peak of nearly 280,000 in 2002, then declining to 210,000 in 2003 due to SARS
and Avian Flu, but immediately increasing with the speed of 20 % in 2004 and 2005 to reach
records of 320,605 and 383,896 in 2005 and 2006 respectively (see Figure 2). In 2007,
Vietnam achieved a new record of 411,557 Japanese visitors, increasing by 7.20% against
2006 (VNAT, n.d.).
Also according to VNAT (n.d.), Japan has been for a long time listed among the five
countries and territories that contribute the highest number of travelers, together with China,
the U.S, the Republic of Korea, and Taiwan. In 2007, the number of Japanese tourists to
Vietnam accounts for 9.87 % (411,557/4,171,564) of the total volume of foreign tourists.
However, the number of Japanese travelers to Vietnam is still very small in comparison with
the total number of Japanese outbound tourists (411,557 out of 17,298,000 million in 2007, or
accounting for 2.38 % of the total).
In order to attract more Japanese travelers to Vietnam, Vietnam should understand the
tastes of Japanese travelers and their position in the Japanese mind. There have been few
studies taking the destination image approach in/about Vietnam as a tourism destination. The
most similar work is Vietnam Sketch (2007) conducting a survey on Japanese visitors in 2007
JATA exhibition using multiple-choice questions based on different attributes. However, it is
far from being enough to position Vietnam in the Japanese mind. In such a given context, this
research attempts to identify and measure the images of Vietnam as a tourism destination for
Japanese travelers by answering three questions:
1. What are the images of Vietnam as a tourism destination for Japanese travelers?
2. How positive/negative are the images of Vietnam as a tourism destination for
Japanese travelers? In this question, the importance-performance analysis is also
used to position Vietnam in the Japanese mind.
3. How strongly do different information sources affect the formation of Vietnamese
images in the Japanese mind?
The next section is to briefly review the literature concerning tourism destination image
measurement.

2. Tourism Destination Image Measurement
The pioneer studies focusing on methods of measuring destination image emerged in
the early 1990s. Lew (1994) is one of the fist works providing a comprehensive framework for
studying the attractiveness of tourism destinations. It asserts that most studies on tourist
attractions and tourist attractiveness of places can be classified into one or more of three
general perspectives: the ideographic listing, the organization, and the tourist cognition of

182
attractions. Each of these perspectives shares a distinct set of questions concerning the nature
of the attractions, as expressed through the typologies used in their evaluation. At the same
time, all three perspectives make comparisons based on the historical, locational, and various
valuational aspects of attractions. More specifically, Echtner & Ritchie (1991 & 1993)
recommends a framework to completely measure destination image. In that framework,
components that must be captured include attribute-based images, holistic impressions, and
functional, psychological, unique and common characteristics. It is proposed that a
combination of structured and unstructured methodologies is necessary to measure destination
image as envisaged in the conceptual framework. A series of open-ended questions and scale
items are developed to successfully capture all of the components of destination image.
Referring to the detailed measures to identify destination image, Jenkins (1999) and
Echtner & Ritchie (2003) are the most comprehensive works especially concentrating on the
issue. Jenkins (1999) reviews different techniques for the measurement of a tourists
destination images after Echtner & Ritchie (1991), compares the advantages and
disadvantages of structured and unstructured methods (see Table 1), and highlights the
dominance of structured, word-based approach. Moreover, it argues that to provide valid
image research, a preliminary phase of qualitative research is important in order to distil the
constructs relevant to the population being studied. Construct elicitation techniques, such as
free-elicitation, interactive interview and focus group interview, are discussed along with new
techniques that include the visual aspect of image, such as photo elicitation.
In a focused research, OLeary & Deegan (2005) introduces Martilla & James (1977)
presenting importance-performance analysis (IPA) as a technique for evaluating the elements
of a marketing program that uses a three-step process either to develop a new marketing
strategy or to evaluate an existing strategy: First, a set of product attributes or features is
identified through techniques such as literature review, focus group interviews, and so on.
Second, consumers are asked two questions about each attribute: How important is it? and
How well did the product or service perform? Third, importance and performance scores for
each attribute are calculated. These values provide x and y coordinates that are then placed on
a two-dimensional plot called an IPA grid. The importance and performance scores are plotted
on the vertical and horizontal axes respectively. By plotting the numerical results in this way,
the components are effectively sorted into a 4-cell typology. This typology categorizes
importance and performance on a scale of high or low, so four combinations are possible as in
Figure 3.
It can be said that the methods for measuring tourism destination image have been
rather systematically developed and can be applied for tourism destinations in one way to
another according to specific purposes. The next section is to present the methods used for this
research.

3. Methodology
Research model
This research adopts the recommendation by Jenkins (1999) to form the research model
as in Figure 4. Accordingly, it begins with a preliminary qualitative phase to identify the
general images of Vietnam in the Japanese mind. Then, it continues with an empirical phase to
measure the positiveness/negativeness of the identified images by deeply interviewing and
surveying Japanese people based on a set of attributes withdrawn from the qualitative phase.
In the qualitative phase, the images of Vietnam are captured by content analysis of
Japanese travel guidebooks and brochures, and comments of Japanese officials/politicians and
managers. The result is a list of images serving as the attributes to be measured in the

183
empirical phase.
In the empirical phase, firstly, to identify the importance of the attributes withdrawn
from the results of the qualitative phase, Japanese people are asked about the importance of
each attribute when they consider traveling abroad. The answer is based on a Likert-scale from
1 to 5 representing the level from very unimportant to very important.
Secondly, to identify the performance of Vietnam about the attributes, a set of positive
statements about Vietnam (see Table 3) are made based on the set of attributes identified in the
preliminary qualitative phase. The respondents are asked to express their reaction to the
statements by choosing one of the Likert-scale numbers from 1 to 5 representing the level
from strongly disagree to strongly agree.
The importance-performance analysis (IPA) suggested by Martilla & James (1977) with
some adaptations is then used to position Vietnam in the Japanese mind and recommend the
directions for a marketing strategy of Vietnam to exploit the Japanese tourist market.
Accordingly, the importance and performance scores are plotted on the horizontal x and
vertical y axes respectively. This typology categorizes importance and performance on a scale
of high or low, so significant combinations are identified.
To examine the effect of different information sources to the formation of the Vietnamese
images in the Japanese mind, respondents are asked the following questions: How much do the
following information sources influence your image of Vietnam as a tourism destination? The
answer is also Likert scale based numbers from one (1) to five (5) representing the level
from Not influential at all to Strongly influential. Comparing means is the main technique
for weighting the affect of the information sources.
Research Implementation
Regarding preliminary the qualitative phase, secondary data were intensively collected
during September 2007 March 2008, from the two main sources: (1) Japanese travel
guidebooks and brochures from prestigious Japanese travel companies, such as JTB (Japan
Travel Bureau), KNT (Kinki Nippon Tourist), NTA (Nippon Travel Agency), ANA, ANA
Hallo Tour, TABIX World, and H.I.S; and (2) various comments on Vietnam from JETRO
(Japan External Trade Organization), Japanese politicians, professors, photographers, and the
mass media.
As for the empirical phase, which was during April July 2008, direct contacts with
Japanese people in Vietnam and Japan and online survey were the two main strategies to
approach Japanese people. The author made full use of his relational network throughout
Japan and Vietnam to approach the respondents.
In brief, the research uses both qualitative and quantitative methods to identify and
measure the images of Vietnam as a tourism destination for Japanese travelers as well as
examine the effect of different information sources to the formation of the Vietnamese images.
The next section is to present the results obtained from using the above methodology.

4. Results and Discussion
Images of Vietnam in the Japanese mind
The images of Vietnam from Japanese travel guidebooks and brochures
Geographically, Vietnam as a tourism destination is introduced in three patterns: (1)
Vietnam in connection with Cambodia (Angkor Wat), typically in JTB and ANA and TABIX
Worlds brochures; (2) Vietnam in connections with other Asian destinations such as
Cambodia (Angkor Wat), India, Myanmar, Thailand, Malaysia, Indonesia, etc., typically in

184
NTA and KNTs brochures; and (3) only Vietnam in focus in travel guidebooks and some
brochures by H.I.S. The first and second patterns are more popular than the other. The
southern (Ho Chi Minh City, MeKong Delta) and central (Hue Former Capital, Dang Nang,
Hoi An Ancient Town) parts are more offered to Japanese travelers than the northern part.
Regarding the physical images of Vietnam, the most frequently presented images
include: Ha Long Bay, French architectural buildings (opera houses and churches in Ho Chi
Minh City and Hanoi, Ho Chi Minh City Hall and Ben Thanh Market in Ho Chi Minh City),
the three world cultural heritages of Hoi An Ancient Town , Hue Former Capital, and My Son
Sanctuary, and the ancient living quarter in Hanoi.
Although not so highlighted as the places mentioned above, some historical places are
also rather frequently presented. These include Ho Chi Minh Mausoleum in Hanoi,
Reunification Hall in Ho Chi Minh City, and Cu Chi Tunnels. Beaches and sea resorts are not
commonly highlighted by all the travel agencies, except for H.I.S that shows all the most
beautiful sea resorts in Vietnam, including those in Da Nang, Hoi An, Hue, Nha Trang, Phan
Thiet, and Phu Quoc.
Referring to the cultural physical images of Vietnam, many of the brochures show
traditional characteristics through pictures of Vietnamese girls wearing traditional Ao dai
(Long dress) in a traditional background in Hoi An Ancient Town (KNT, H.I.S) or French
architectural quarters in Hanoi or Ho Chi Minh City (ANA Hallo Tour, H.I.S). Also, food is
very frequently highlighted. The most popular dishes are Pho (Vietnamese noodle), Goi
cuon (Fresh spring rolls), Cha gio (Fried spring rolls), and different kinds of fruits.
Vietnamese coffee is also introduced as a speciality by H.I.S. Moreover, the pictures of
everyday life in the Me Kong delta area, Ho Chi Minh City, and Hanoi are also usually
repeated. Water puppet shows are the most popular traditional performance introduced in small
pictures.
As for products and services, among traditional handicraft products (bags, purses, small
boxes, laces, pretty small items, etc.), things made from silk and ceramics are most presented.
Some traditional shopping corners in the ancient quarter of Hanoi or Ben Thanh Market in Ho
Chi Minh City are also introduced, but specific products are not focused. Also, traditional spa
and massage services are offered by most of the travel agencies. Luxury accommodation of
four or five star standard is usually offered to Japanese tourists.
According to the analysis of Japanese travel guidebooks and brochures, it is inferred that
the images of Vietnam are still not very highlighted in Japan. The inference is based on the
fact that Vietnam is mostly introduced in connection with Cambodia (Angkor Wat) and other
Asian destinations, which indicates that the country has not had strong-enough attractors or
favorable conditions to attract Japanese travelers as a separate destination. The southern and
central regions of Vietnam are more introduced in the guidebooks and brochures may be
explained by the more open and dynamic economic environment and the intensity of tourism
attractions in these regions.
Also, it is recognized that the physical description of Vietnam in Japanese travel
guidebooks and brochures are similar to that of almost any other destinations covering natural
landscapes, historical places, typical architectural monuments, cultural activities, and local life.
Perhaps, the most unique images representing Vietnam are Vietnamese women wearing
coconut hats, traditional Ao dai (Long dress), and some traditional dishes such as Pho,
Nem cuon, and Cha gio.
The images of Vietnam from Japanese comments
Regarding positive images, many of the comments, especially from JETRO, Japanese
politicians, and the media, highlight safety, political stability, economic growth and prospect,

185
and central geographical position as advantages of the country (VietnamNet, 2006, VNA/Asia
Pulse, 2006, VNA, 2007b, Ha, 2007, Shimizu, 2007, Quang, 2005, Japanese investors, 2007,
Reuters, 2007). Additionally, beautiful natural landscapes and delicious cuisine are the two
strong images of Vietnam stressed by Japanese managers and the mass media. Also, Japanese
managers in Vietnam highly appreciate world heritage sites, peaceful/relaxing environment,
original fine arts, handicraft products, friendliness, and nice weather (VNJFD, 2006, The SGT
Daily, 2005). Besides, the media highlights Vietnam as a new destination with famous war
history, young people, and strong spiritual life (VNA, 2007a, Takenouchi, 2007, Swinnerton,
2003).
Beside positive images, there are also some opinions pointing out the negative images of
Vietnam that include poor infrastructure, poor health care system, and lack of tourism product
diversity for shopping (VietnamNet, 2006, VNA/Asia Pulse, 2006, VNA, 2007b). In addition,
there are some conflicting opinions about the images of Vietnam related to the Vietnamese war
history. According to Ms. Yoshino Oishi, a photographer who made about twenty (20) photo-
taking trips to Vietnam between 1981 and 2000, Vietnam still signifies war, rather than tourism
and she admits that she has difficulties relating to the country's phenomenal popularity as a
tourist destination, especially among young Japanese women. She is quoted as saying: "I
suppose that young Japanese women who are bored with shopping for designer labels consider
Vietnam a novelty". However, Mr. Hiromi Tanaka, a tour leader of Sinh Cafe Tours in
Vietnam says that Vietnam for many Japanese has become a country of fresh spring rolls and
variety stores, rather than a country with painful war memories" (Hasegawa, 2002).
One of the most prominent features from the comments from JETRO, Japanese
politicians, and the media is that they focus on Vietnams social and economic macro-
environment of safety, political stability, and bright development prospect. In this regard, the
poor condition of infrastructure, health care system, and tourism product diversity are likely to
be analyzed in parallel. Also, Vietnamese people are highly appreciated in terms of young
people, friendliness, and strong spiritual life The conflicting opinions about whether Vietnam
signifies war or tourism is understandable when Vietnam experienced a along period of wars
in the last century and has only recently emerged as a developing economy.
Positive and negative images of Vietnam in the Japanese mind
The participants of the survey include 257 Japanese people: 45.14 % male and 54.86 %
female; aged from 18 to 75 (mean = 35.87); over 60.55 % have been to Vietnam.
Table 2 presents the importance means of the image attributes. It shows that Japanese
people prefers destinations offering them beautiful natural landscapes, new experiences, rich
culture, and personal safety. Surprisingly, although shopping has been considered a major part
of a Japanese journey, good places for shopping, souvenir products, and crafts/small lovely
items are not the attractors to Japanese travelers. Moreover, sea-related attractions are not
attractive to Japanese people when beautiful beaches and sea resorts are not highly rated.
Table 3 shows the evaluation of the performance of Vietnam about the attributes.
Accordingly, Vietnam is highly appreciated in the attributes of low prices, delicious cuisine,
rich culture, new experience opportunities, and friendliness. Also, economic prospect and
natural landscapes of Vietnam are also highly rated. It is surprising that Japanese people are
unlikely to agree that Vietnam has a political stability and personal safety which the
Vietnamese side is very much confident about. Regarding negative images, cleanliness,
infrastructure, health care system, and service quality are the worst attributes of Vietnam.
If the mean value of 3.50 is set as the point differentiating low and high
importance/performance, following which the mean value under 3.50 is considered low and
the mean value from 3.50 and above is regarded high, the importance performance analysis
grid is indicated as in Figure 7. Accordingly, the importance and performance scores are

186
scattered in the horizontal and vertical axes and the attributes are classified into four groups.
As shown in the importance performance analysis grid, Vietnam should keep up the
good work related to the high importance high performance attributes, which reveals a new
destination of rich culture, friendly people, beautiful natural landscapes, and low prices.
Among the specific attributes, Vietnamese cuisine is strongly appreciated, which is described
and demonstrated by all the sources of information including travel guide books and brochures,
Japanese comments, and respondents. Especially, prices in Vietnam are considered very
competitive.
It is necessary that Vietnam highly concentrate on changing or correcting its negative
images related to political stability, personal safety, architectural buildings, health care system,
and cleanliness. These negative images indicate that Vietnam is generally believed to not have
enough fundamental conditions for well-off Japanese travelers, especially when the mentioned
attributes are highly important for Japanese people.
Noticeably, there is a conflicting result between the Japanese comments and the various
surveyed respondents about the evaluation of personal safety and political stability in Vietnam.
While the comments highly appreciate the two attributes, the various surveyed respondents
depreciate them. This may be because of the comments are mostly from business travelers
who have been to Vietnam and enjoyed high and standard services and the favorable economic
condition in Vietnam, which are generally safe and stable, while the various surveyed
respondents are from different groups of people who can not always do that.
Another feature from the importance performance analysis is that Vietnam may have
some strengths but they are not very significant in attracting Japanese travelers because of
their low importance to Japanese people when they consider traveling abroad. Those strengths
include bright economic prospect, crafts/small lovely items, unique souvenir products,
many/various tourists attractions, good places for shopping, exciting traditional performance,
and opportunities for adventure.
It also noticeable that there are some attributes very commonly regarded the strengths of
Vietnams tourism, but they are not very attractive to Japanese travelers and Japanese people
also do not appreciate these attributes of Vietnam. These include beautiful beaches, sea resorts,
and world heritage sites.
Surprisingly, good quality of service and good infrastructure are not highly considered by
Japanese people. It may not automatically mean that Japanese travelers do not care about these
aspects when traveling abroad. Rather, they may take it for granted that service and
infrastructure should be always qualified enough.
Effect of different information sources to the images of Vietnam
The effect of different information sources to the formation of Vietnamese images in the
Japanese mind is presented in Table 4. Real experience has the strongest influence on the
formation of Vietnamese images, which demonstrates the major participation of Japanese
people who have been to Vietnam in the sample. Together with real experience, relatives and
friends (words of mouth), travel guidebooks, pamphlets/brochures, television and internet are
the major information sources affecting the images of Vietnam.
The other popular sources of information such as books, magazines, travel agents,
newspapers, education, and world travel fairs do not have significant influence. Tourism
promotion of Vietnam is also not significant in influencing the images of Vietnam in the
Japanese mind.


187
5. Conclusion
Firstly, regarding the images of Vietnam as a tourism destination for Japanese travelers,
the presentation of Vietnam in Japanese travel guidebooks and brochures is similar to that of
almost any other destinations covering natural landscapes, historical places, typical
architectural monuments, cultural activities, and local life. Perhaps, the most unique images
representing Vietnam are Vietnamese women wearing coconut hats, traditional Ao dai (Long
dress), and some traditional dishes such as Pho, Nem cuon, and Cha gio. Generally, the
images of Vietnam are still not very highlighted in Japan, and Vietnam has not had strong-
enough attractors or favorable conditions to attract Japanese travelers as a separate destination.
Moreover, some Japanese people may still think of Vietnam as a destination of wars rather
than of tourism.
Secondly, referring to the positiveness/negativeness of the images of Vietnam in the
Japanese mind, Vietnam is highly appreciated in the attributes of low prices, delicious cuisine,
rich culture, new experience opportunities, and friendliness. Also, economic prospect and
natural landscapes of Vietnam are also highly rated. Noticeably, there is a conflicting result
between the Japanese comments and the various surveyed respondents about the evaluation of
personal safety and political stability in Vietnam. While the comments highly appreciate the
two attributes, the various surveyed respondents depreciate them. The worst performance of
Vietnam is about the attributes of cleanliness, infrastructure, health care system, and service
quality.
As for the importance of the image attributes to Japanese people, the most important ones
are beautiful natural landscapes, new experiences, rich culture, and personal safety. Shopping
may be a major part of a Japanese journey, but not a strong attractor; sea-related attractions are
not attractive to Japanese people.
According to the importance performance analysis, it is important to consider the
suggestions about marketing Vietnam to Japanese travelers as follows:
Vietnam should promote itself as a new destination of rich culture, friendly people,
beautiful natural landscapes, delicious cuisine and low prices.
Vietnam should highly concentrate on changing and correcting its negative images
related to political stability, personal safety, architectural buildings, health care
system, and cleanliness.
Vietnam do not need to overemphasize its possible strengths of bright economic
prospect, beautiful beaches and sea resorts, because of the low importance of these
attributes to Japanese travelers when they consider traveling abroad.
Thirdly, relatives and friends (words of mouth), travel guidebooks, pamphlets/brochures,
television, and internet are the most influential information sources on the formation of
Vietnamese images. Therefore, Vietnam should actively promote the cooperation partnerships
between travel agencies of Japan and Vietnam to further present Vietnam in Japanese
guidebooks and brochures, and increase the introductions of Vietnamese images in the internet
and television programs in Japan.
Overall, this research has identified and measured the images of Vietnam as a tourism
destination for Japanese travelers and examined the effect of different information sources to
the Vietnamese images in the Japanese mind. Also, it has suggested some ideas about
marketing Vietnam to Japan. Although it has some contributions to the research of Vietnam as
a tourism destination to Japanese travelers, further research should be supplemented to shed
more light on the issue. Specifically, (1) the sample should be more enlarged do other analyses,
such as Chi-Square tests to identify the significant differences among the different segments of

188
Japanese people; and (2) more interviews of tourism experts should be conducted to analyze
the important facts and their causes.

6. Tables
Table 1. Methods used in destination image research: structure versus unstructured
(after Echtner & Ritchie, 1991)
Structured Unstructured
Description Various common image attributes
are specified and incorporated into a
standardized instrument and the
respondent rates each destination on
each of the attributes, resulting in an
image profile
The respondent is allowed to freely describe his
or her impressions of the destination. Data are
gathered from a number of respondents. Sorting
and categorization techniques are then used to
determine the image dimensions
Techniques Usually a set of semantic
differential or Likert type scales
Focus group, open-ended survey questions,
content analysis, repertory grid.
Advantages Easy to administer
Simple to code
Results easy to analyze using
sophisticated statistical
techniques
Facilitates comparisons between
destinations
Conducive to measuring the holistic
components of destination image
Reduces interviewer bias
Reduces likelihood of missing important
image dimensions or components
Disadvantages Does not incorporate holistic
aspects of image
Attribute focused that is, it
forces the respondent to think
about the product image in terms
of the attributes specified
The completeness of structured
methods can be variable it is
possible to miss dimensions.
Level of detail provided by respondents is
highly variable
Statistical analyses of the results are limited
Comparative analyses are not facilitated.
Source: Jenkins (1999, p. 6).

189
Table 2. Importance scores of the attributes
Attributes Mean Std. Deviation
Beautiful natural landscapes
257 4.11 1.056
New experience
257 4.04 1.032
Personal safety
257 4.00 1.090
Rich culture
257 3.98 1.113
Delicious cuisine
257 3.95 1.097
Low prices
257 3.91 .992
Relaxing atmosphere
254 3.87 .917
National characters
255 3.86 1.002
Political stability
257 3.85 1.043
Opportunities to increase knowledge
255 3.71 1.046
Famous and interesting history
255 3.64 1.137
Friendliness toward Japanese people
257 3.60 1.008
Cleanliness
257 3.59 1.196
Good traditional fine arts
257 3.59 1.125
Good health care system
257 3.59 .981
Beautiful architectural buildings
257 3.58 1.126
Interesting local ways of life
257 3.56 1.117
Good quality of service
254 3.45 1.119
Beautiful beaches
257 3.37 1.136
Exciting traditional performance
257 3.37 1.015
Nice climate
253 3.32 1.078
Good infrastructure
257 3.26 1.123
Opportunities for adventure
257 3.25 1.244
Beautiful sea resorts
257 3.24 1.180
Good places for shopping
253 3.11 1.073
Many/various tourist attractions
253 3.09 1.052
Unique souvenir products
257 3.07 1.167
Easy to travel to other tourism
destinations
257 2.94 1.012
Crafts/small lovely items
252 2.85 1.286
World heritage sites
254 2.81 1.017
Bright economic prospect
257 2.80 1.015
Opportunities to improve career/business
253 2.72 1.153
Good massage and spa services
256 2.62 1.250
Valid N (listwise)
223


190
Table 3. Performance means about the attributes of Vietnam
Statements Mean
Std.
Deviation
Prices in Vietnam are low
257 4.13 .951
Cuisine in Vietnam is delicious
257 4.09 1.029
Vietnam is rich in culture
257 4.03 .964
You can have many new experiences in Vietnam
253 3.99 .930
Vietnamese people are friendly
255 3.98 .968
Vietnam has a famous and interesting history
256 3.92 1.026
You can gain much more knowledge by visiting Vietnam
256 3.91 .912
There are many crafts/small lovely items in Vietnam
255 3.89 1.033
Vietnam has a bright economic prospect
256 3.86 .988
Vietnamese national characters are impressive
255 3.85 .986
Vietnam has many beautiful natural landscapes
257 3.81 1.049
There is a relaxing atmosphere in Vietnam
253 3.78 .890
There are many souvenir products in Vietnam
257 3.77 .999
Vietnamese ways of life are interesting
256 3.75 .953
Vietnam has good traditional fine arts
257 3.63 .992
Vietnam has various tourist attractions
257 3.58 1.024
There are many good places for shopping in Vietnam
252 3.56 .919
Vietnamese traditional performance is exciting
254 3.52 1.016
You can adventure a lot when visiting Vietnam
255 3.52 .995
Massage and spa services are good in Vietnam
250 3.44 .877
Vietnam is politically stable
256 3.41 .990
There are many beautiful architectural buildings in
Vietnam
257 3.40 .960
It is easy to travel to other tourism destinations from
Vietnam
254 3.40 .992
Vietnam has many beautiful beaches
257 3.36 1.063
The climate in Vietnam is pleasant
252 3.36 .944
Vietnam has many beautiful sea resorts
255 3.33 .869
You are safe in Vietnam
257 3.28 .930
You can improve your career/business by visiting Vietnam
253 3.23 .986
There are many world heritage sites in Vietnam
253 3.05 .935
Service quality is good in Vietnam
252 2.91 .890
Vietnam has a good health care system
253 2.84 .832
Infrastructure is good in Vietnam
257 2.79 .887
It is clean in Vietnam
256 2.71 .897
Valid N (listwise)
235


191
Table 4. The effect of information sources to the images of Vietnam
Sources of information Mean Std. Deviation
Real experience 251 4.08 1.150
Relatives and friends
257 4.04 1.020
Travel guidebooks
257 3.99 1.015
Television
255 3.74 1.082
Pamphlets/brochures
257 3.70 1.074
Internet
257 3.63 1.146
Books
257 3.45 1.064
Magazines
247 3.28 1.063
Tourism promotion of Vietnam
248 3.20 1.163
Travel agents
257 3.19 1.190
Newspapers
253 3.05 1.162
Education
257 3.00 1.141
World travel fairs
242 2.84 1.177
Valid N (listwise)
236

7. Figures
Figure 1. Japanese outbound tourists: actual and EIU forecasted numbers
1
5
.
3
0
1
6
.
6
9
1
6
.
8
0
1
5
.
8
1
1
6
.
3
6
1
7
.
8
2 1
6
.
2
2
1
6
.
5
2
1
3
.
3
0
1
6
.
8
3
1
7
.
4
0
1
7
.
5
4
1
7
.
7
3
1
8
.
1
1
1
8
.
9
1
1
9
.
5
8
2
0
.
0
7
2
0
.
5
6
0
5
10
15
20
25
1
9
9
5
1
9
9
6
1
9
9
7
1
9
9
8
1
9
9
9
2
0
0
0
2
0
0
1
2
0
0
2
2
0
0
3
2
0
0
4
2
0
0
5
2
0
0
6
2
0
0
7
2
0
0
8
2
0
0
9
2
0
1
0
2
0
1
1
Year
J
a
p
a
n
e
s
e

o
u
t
b
o
u
n
d

t
o
u
r
i
s
t
s

(
m
i
l
l
i
o
n
s
)
Actual number EIU forecasted number

Source: Adapted from JNTO (2005), JTM (2006), and EIU (2007)


192
Figure 2. Japanese tourists to Vietnam since 1995
119540 118310
124862
95258
113514
152755 152755
279769
209730
267210
320605
383896
411557
0
5
0
,
0
0
0
1
0
0
,
0
0
0
1
5
0
,
0
0
0
2
0
0
,
0
0
0
2
5
0
,
0
0
0
3
0
0
,
0
0
0
3
5
0
,
0
0
0
4
0
0
,
0
0
0
4
5
0
,
0
0
0
1
9
9
5
1
9
9
6
1
9
9
7
1
9
9
8
1
9
9
9
2
0
0
0
2
0
0
1
2
0
0
2
2
0
0
3
2
0
0
4
2
0
0
5
2
0
0
6
2
0
0
7
Year
J
a
p
a
n
e
s
e

t
o
u
r
i
s
t
s

t
o

V
i
e
t
n
a
m

Source: VNAT (n.d.)

Figure 3. Importance performance analysis grid

Source: OLeary & Deegan (2005, p. 250)

193
Figure 4. The research model


Qualitative Phase

Purpose: To identify the general images of
Vietnam in the Japanese mind

Methods: Content analysis of travel
guidebooks and brochures,
and comments
Empirical Phase
(Quantitative and Qualitative)

Purpose: To measure the images of
Vietnam in the Japanese mind

Methods: Survey and interviews based on
Likert-scale questions
Relevant image attributes

1
9
4

F
i
g
u
r
e

5
.

I
m
p
o
r
t
a
n
c
e

p
e
r
f
o
r
m
a
n
c
e

a
n
a
l
y
s
i
s

g
r
i
d

4
.
0
0
3
.
5
0
3
.
0
0
I
m
p
o
r
t
a
n
c
e
4
.
0
0
3
.
5
0
3
.
0
0
D
e
r
i
v
e
d

f
r
o
m

I
m
p
o
r
t
a
n
c
e
4
.
2
0
3
.
9
0
3
.
6
0
3
.
3
0
3
.
0
0
2
.
7
0
P e r f o r m a n c e
4
.
2
0
3
.
9
0
3
.
6
0
3
.
3
0
3
.
0
0
2
.
7
0
Derived from Performance
G
o
o
d

m
a
s
s
a
g
e

a
n
d

s
p
a

s
e
r
v
i
c
e
s
O
p
p
o
r
t
u
n
i
t
i
e
s

t
o

i
m
p
r
o
v
e

c
a
r
e
e
r
/
b
u
s
i
n
e
s
s
B
r
i
g
h
t

e
c
o
n
o
m
i
c

p
r
o
s
p
e
c
t
W
o
r
l
d

h
e
r
i
t
a
g
e

s
i
t
e
s
C
r
a
f
t
s
/
s
m
a
l
l

l
o
v
e
l
y

i
t
e
m
s
E
a
s
y

t
o

t
r
a
v
e
l

t
o

o
t
h
e
r

t
o
u
r
i
s
m

d
e
s
t
i
n
a
t
i
o
n
s
U
n
i
q
u
e

s
o
u
v
e
n
i
r

p
r
o
d
u
c
t
s
M
a
n
y
/
v
a
r
i
o
u
s

t
o
u
r
i
s
t

a
t
t
r
a
c
t
i
o
n
s
G
o
o
d

p
l
a
c
e
s

f
o
r

s
h
o
p
p
i
n
g
B
e
a
u
t
i
f
u
l

s
e
a

r
e
s
o
r
t
s
O
p
p
o
r
t
u
n
i
t
i
e
s

f
o
r

a
d
v
e
n
t
u
r
e
G
o
o
d

i
n
f
r
a
s
t
r
u
c
t
u
r
e
N
i
c
e

c
l
i
m
a
t
e
E
x
c
i
t
i
n
g

t
r
a
d
i
t
i
o
n
a
l

p
e
r
f
o
r
m
a
n
c
e
B
e
a
u
t
i
f
u
l

b
e
a
c
h
e
s
G
o
o
d

q
u
a
l
i
t
y

o
f

s
e
r
v
i
c
e
I
n
t
e
r
e
s
t
i
n
g

l
o
c
a
l

w
a
y
s

o
f

l
i
f
e
B
e
a
u
t
i
f
u
l

a
r
c
h
i
t
e
c
t
u
r
a
l

b
u
i
l
d
i
n
g
s
G
o
o
d

h
e
a
l
t
h

c
a
r
e

s
y
s
t
e
m
G
o
o
d

t
r
a
d
i
t
i
o
n
a
l

f
i
n
e

a
r
t
s
C
l
e
a
n
l
i
n
e
s
s
F
r
i
e
n
d
l
i
n
e
s
s

t
o
w
a
r
d

J
a
p
a
n
e
s
e

p
e
o
p
l
e
F
a
m
o
u
s

a
n
d

i
n
t
e
r
e
s
t
i
n
g

h
i
s
t
o
r
y
O
p
p
o
r
t
u
n
i
t
i
e
s

t
o

i
n
c
r
e
a
s
e

k
n
o
w
l
e
d
g
e
P
o
l
i
t
i
c
a
l

s
t
a
b
i
l
i
t
y
N
a
t
i
o
n
a
l

c
h
a
r
a
c
t
e
r
s
R
e
l
a
x
i
n
g

a
t
m
o
s
p
h
e
r
e
L
o
w

p
r
i
c
e
s
D
e
l
i
c
i
o
u
s

c
u
i
s
i
n
e
R
i
c
h

c
u
l
t
u
r
e
P
e
r
s
o
n
a
l

s
a
f
e
t
y
N
e
w

e
x
p
e
r
i
e
n
c
e
B
e
a
u
t
i
f
u
l

n
a
t
u
r
a
l

l
a
n
d
s
c
a
p
e
s
G
o
o
d

w
o
r
k

(
H
i
g
h

i
m
p
o
r
t
a
n
c
e

-

h
i
g
h

p
e
r
f
o
r
m
a
n
c
e
)
N
e
c
e
s
s
a
r
y

c
o
n
c
e
n
t
r
a
t
i
o
n

(
H
i
g
h

i
m
p
o
r
t
a
n
c
e

-

l
o
w

p
e
r
f
o
r
m
a
n
c
e
)
L
o
w

p
r
i
o
r
i
t
y

(
L
o
w

i
m
p
o
r
t
a
n
c
e

-

l
o
w

p
e
r
f
o
r
m
a
n
c
e
)
P
o
s
s
i
b
l
e

o
v
e
r
k
i
l
l

(
L
o
w

i
m
p
o
r
t
a
n
c
e

-

h
i
g
h

p
e
r
f
o
r
m
a
n
c
e
)


195
References
EIU (The Economist Intelligence Unit, Ltd.) (2007). EIU market research: travel and tourism
forecast Vietnam 2006. The Economist Intelligence Unit, Ltd.
Echtner, C. M., & Ritchie, J. R. B. (1991). The meaning and measurement of destination
image. Journal of Travel Studies, 2(2), 2-12.
Echtner, C. M., & Ritchie, J. R. (1993). The measurement of destination image: an empirical
assessment. Journal of Travel Research, 31(4), 3-13.
Echtner, C. M., & Ritchie, J. R. B. (2003). The meaning and measurement of destination
image. Journal of Tourism Studies, 14(1), 37-48.
Ha, C. (2007). Vietnam la diem den hap dan nhat chau A [Vietnam the most attractive
destination in Asia]. Retrieved October 3, 2007, from
http://www.tuoitre.com.vn/Tianyon/Index.aspx?ArticleID=188967&ChannelID=11.
Hasegawa, Y. W. (2002, August 13). Tour leader opens eyes to harsh realities of Vietnam. The
Japan Times.
Jenkins, O. H. (1999). Understanding and measuring tourist destination images. The
International Journal of Tourism Research, 1, 1-15.
JNTO (2005). Statistics on tourism for Japan. Tokyo.
JTM (Japan Tourism Marketing Co.) (2006). Tourism statistics. Retrieved May 5, 2007, from
http://www.tourism.jp/english/statistics/.
Japanese investors seeking rough diamonds in Vietnamese stocks. (2007, March 8). Jiji Press
Ticker Service.
Lew, A. A. (1994). A framework of tourist attraction research. In J. R. B. Ritchie, & C. R.
Goeldner (Eds.), Travel, tourism, and hospitality research: a handbook for managers and
researchers (pp. 291-304). New York: John Wiley & Sons, Inc.
Martilla, J. A., & James, J. C. (1977). Importance-performance analysis. Journal of Marketing,
41, 77-79.
OLeary, S., & Deegan, J. (2005). Irelands image as a tourism destination in France: attribute
importance and performance. Journal of Travel Research, 43(3), 47-256.
Quang M. (2005). Hien ke cho du lich phat trien [Initiatives for tourism development].
Retrieved April 29, 2007, from http://vietnamnet.vn/kinhte/2005/05/420191/.
Reuters (2007). Du lich chung khoan Vietnam sot o hat [Fever of tourism for
Vietnams stock market securities in Japan]. Retrieved October 3, 2007, from
http://vietnamnet.vn/kinhte/2007/05/692772/.
Shimizu, K. (2007, May 2). Japanese firms flock to booming Vietnam. The Japan Times.
Swinnerton, B. (2003, July 11). Vietnamese cooking puts a spring in your step. The Japanese
Times.
Takenouchi, T. (2007, September 28). Vietnamese food finding new popularity. The Daily
Yomiuri (Tokyo), p. 16.
The SGT Daily (2005). Big Japanese tourist group comes to Ho Chi Minh City. Retrieved
October 3, 2007, from
http://www.saigontourist.hochiminhcity.gov.vn/news/detail_en.asp?id=1261.

196
UNWTO (2005a). East Asia and the Pacific: the Most Dynamic Tourism Region. Retrieved
December 8, 2005 from http://www.world-
tourism.org/newsroom/Releases/2005/June/CAP.htm
UNWTO (2005b). Worlds top tourism spenders. Retrieved Jan 1, 2006 from
http://www.world-tourism.org/facts/eng/pdf/indicators/topSpenders.pdf.
Vietnam (2001). Chien luoc phat trien du lich Vietnam [The strategy for tourism development
in Vietnam for the 2001 2010 period]. Hanoi: Vietnam National Administration of
Tourism (VNAT).
VietnamNet (2006). Viet am dung thu 5 trong mat nha dau tu hat [Vietnam stands fifth in
the eyes of Japanese investors]. Retrieved October 3, 2007, from
http://www.tienphongonline.com.vn/Tianyon/Index.aspx?ArticleID=48822&ChannelID=
3.
Vietnam Sketch (2007). Report on the participation of Vietnam in JATA - Japan 2007. Hanoi:
VNAT.
VNA (2007a). Bao hat Ban ca ngoi ve dep cua dong bang song Cuu Long [A Japanese
newspaper appreciates the beauty of Cuu Long River delta]. Retrieved October 3, 2007,
from http://www.tuoitre.com.vn/Tianyon/Index.aspx?ArticleID=190883&ChannelID=100.
VNA (2007b). Viet am co moi truong dau tu tot nhat [Vietnam has the best investment
environment]. Retrieved October 3, 2007, from
http://vietnamnet.vn/chinhtri/2007/05/701028/.
VNA/Asia Pulse (2006). Japanese prefer Vietnam. Retrieved October 3, 2007, from
http://www.atimes.com/atimes/Southeast_Asia/HE12Ae03.html.
VNAT (n.d.). Tourist statistics. Retrieved April 18, 2008, from
http://www.vietnamtourism.gov.vn/english/index.php?option=content&task=section&id=
6&Itemid=45
VJFD (V one of Japanese 20 favorite destinations) (2006). Retrieved May 5, 2007, from
http://www.footprintsvietnam.com/vietnam_news/VN%20one%20of%20Japanese%2020
%20favourite%20destinations.htm



197
The Vietnamese Automobile Industry and Globalization



Nguyen Bich Thuy
PhD Candidate, Waseda University
Email: nnthuywaseda@yahoo.com


Abstract
Although Vietnamese automobile industry has more than two decades of
development, it so far just limits at simple automobile assembling. The
industry is currently in the grips of a vicious circle, in which small size of
the market prevents automotive makers from reaping the mass-production
merits, with the result that market growth is hindered by high production
cost. The globalization waves even deepen the difficulties faced by this
infant industry due to strong competition from overseas. How to integrate
into the global production network and utilize globalization will be the most
decisive condition for the industrys survival. And to realize it needs
governments effective incentives to the automotive makers basing on good
understanding of the domestic production situation and desires from the
automotive makers and also enthusiasm of automotive makers to find out
how to promote localization.

Keywords: Industrial policy, automobile industry, localization, local content, local
content ratio, globalization, marketing.
JEL Code: F15


198
1. Introduction
Promotion of the automobile industry has become a target of Vietnam for its
industrialization. In fact, Vietnam is a new emerging and potential market and
production base for automobile products due to its dramatic economic growth and its
big population of more than 84 million with low car ownership rate and possibility to
grow as manufacturer with good and big labor forces. However, as a late comer in the
worlds automobile market, the Vietnamese automobile industry is still behind of
well-developed automobile industries in the region (ASEAN-Association of Southeast
Asian Nations) and around the world in the context of free trade also. Support from the
government is indispensable to strengthen the infant industry and enter into the global
production network. However, the current policy is not only too general and unclear but
also has negative effect on automotive production. Learning from other countries about
history of development of their automobile industries may suggest Vietnam many hints
to encourage the automobile industry.
This study differentiates from other studies at that it examines concrete responses
of the automakers to the governments policies and to the industrys situation as well as
to their desires of governments supports, especially, how to make the industry adapt
with globalization tendency.
1.1. Scale demerit as obstacle for localization
The automobile industry needs to have very big initial investment capital for
equipments, factories, technology, R&D and so on. Thus, it is difficult for them to
reduce production cost, make profit, and reinvest in production if sales volume is
limited. Usually, an automobile industry can only take off when sales volume exceeds
300,000 units a year (Takayasu, 1998, p.22), otherwise localization is extremely costly,
making price of car even higher. The automakers said that their current sales and
production volume were too small for autoparts localization. And more than tens of
thousands of autoparts to a model of automobile could not be produced domestically
just for such a tiny sales volume of around 80,000 vehicles that were shared among 49
assemblers (Report on investment in automotive industry, Vietnamese Ministry of
Planning and Investment, 2007) in 2007.
1.2. Globalization
Integration into the regional economy and the worlds economy is indispensable in
trade liberalization process of Vietnam. And joining in and taking advantage from
globalization brings about benefits of market expansion, technology transfer, labor
division, etc for Vietnam. However, the Vietnamese automobile industry seems to quite
focus on the domestic market while exporting and participating in the regional and
global production networks are not accelerated. The automakers asserted that Vietnam
should actively exchange with the regional and international automotive makers for
mass production and invite global autoparts makers rather than independently
self-establish its own automobile industry who just serves limited Vietnamese people.
1.3. Utilization of industrial policies to promote localization
Industrial policy is usually used to adjust and direct investment and production in a
country. However, how to utilize this useful economic management tool to promote
localization is not a simple question. The automakers said that they could not localize
more if the government had no change in its policies to adapt with demand from the
automakers. Many requirements of the automakers were refused and many policies were
approved just basing on subjective valuation of the policy makers. Moreover, the
current policies seem to limit operation of the automotive manufacturers because the

199
policies miss positioning the Vietnamese automobile industry in the regional and global
production network and fail to encourage its exchanging with the outside automotive
manufacturers.

2. Overview of Vietnamese Automobile Industry
The Vietnamese automobile industry has experienced nearly two decades of
development and achieved a certain progress. However, it is still in the early stage of an
infant industry.
2.1. Development history
Prior to 1991, central planning economy was applied in Vietnam, so the
automobile industrys production was operated according to the governments plan and
no policy for encouragement of the industry was utilized. Thus, it resulted in
underdevelopment of the industry. At that time, Vietnam imported cars mainly from the
socialist countries while some state-owned enterprises worked basically on repairing
imported vehicles and built buses of 46-50 seats from imported chassis. There were no
domestic companies that invested in synchronous car assembly lines.
Since 1991, the Vietnamese automobile industry has begun to dramatically change
and gradually take shape owing to the open policy and various incentives of the
Vietnamese government. Foreign automakers started investing in Vietnam and joining
with some prominent state-owned automobile enterprises to establish their production
bases in Vietnam.
In this period, many worlds leading automakers have been allowed to invest in
because the government wants to create competition in the domestic market, which has
been highly protected. So by the end of 1990s, there were up to 11 JVs in Vietnam and
currently there are 14 JVs (Plan to develop the automobile industry, Vietnamese
Ministry of Industry, 2007).
Since 2003 pure Vietnamese automakers (PVCs) have particularly blossomed
when the Vietnamese government approved the strategy of Vietnamese automobile
industry development to the year 2010, vision 2020 in which the government
encouraged all economic sectors participating in the automobile industry and eagerly
accelerated state-owned automakers. So far 35 PVCs have been built up. However, only
some of them have invested seriously in production and the other ones remain at simple
assembling level. In this period, globalization has triggered controversy of the future of
the Vietnamese automobile industry because the schedules of tariff reduction following
to free trade agreements such as AFTA (ASEAN free trade agreement), ACFTA
(ASEAN-China free trade agreement) and WTO (World Trade Organization) were
settled down.
2.2. Operation of automakers
Since the early years of the twenty first century, the Vietnamese automobile
industry has started speeding up. However, the products are mainly focused on the
domestic market, though there are some hundreds of cars that were exported under the
state agreements
1
.
In 1990s, products of JVs accounted for majority of car domestic market share,
and those of PVCs only accounted a tiny rate just around 10%. However, since 2003, the
PVCs have had a big jump and their sales volume accounted for 43% of the total sales in
2007(Report on investment in automotive industry, 2007).

200
Generally, JVs focus on passenger car production, which serves high-income
people or organizations in Vietnam, and 100% passenger cars are produced by JVs. In
the meanwhile, PVCs mainly concentrate on assembling busses, light trucks, minivans
and their products account for majority of domestic market share of 80% of buses and
85% of trucks (Plan to develop the automobile industry, 2007) owing to their reasonable
prices.

3. Vietnamese Governments Policy
Though the Vietnamese government has maintained protection policy in quite a
long time in order to establish an automobile industry, since 2004 many adjustments in
the policy have been issued due to its disappointment of the localization progress and
regulations according to the free trade agreements.
3.1. Foreign direct investment policy
As regulation, 100% foreign-invested autoparts makers are allowed in Vietnam,
but 100% foreign-invested automakers had not been allowed, so they must make joint
venture with Vietnamese firms. Although since participation in WTO, Vietnam has
allowed 100% foreign investment automakers, it keeps asking investment license. The
automakers are received incentives for starting production such as exemption from
income tax for the period of 4 years, reduction of 50% of income tax for 8 years since
making profits, preferential income tax of 15% to 20% depending on their location
(income tax is normally 25%). However, the current foreign direct investment policy is
criticized because it just encourages automobile assembling and does not stimulate
autoparts production. Moreover, the government seems to want to raise state-owned
enterprises instead of private enterprises and foreign investment is not got much
attention from the government.
3.2. Regulation on automotive products import
Taxation policy is the main tool of policy makers in management of the automobile
industry. It has direct effect on decision of business and production of the automotive
makers. The taxation policy is shown basing on products as follow:
Regulation on CBU import
For quite a long time, the imported CBU automobiles have been restricted from the
domestic market. They were levied extremely high tax and tariff including 100% import
tariff rate, 100% special consumption tax rate (SCT) and 10% value-added tax rate
(VAT). Thus, imported cars hardly had chance to enter into Vietnam.
Since 2006 the import tariff rate for brand new cars has been consecutively
reduced for 4 times to 60% from 100% because the policy makers want to press the
automakers decreasing price of cars by increasing of number of imported cars. However,
the import tariff rate was adjusted to increase to 70% on March 3rd 2008 because the
local automakers had no change in prices of their products and number of imported cars
increased quickly.
Since 2006, used under-16-seat cars have been officially allowed to be imported to
Vietnam. The purpose is to push competition in the domestic market up. Used cars were
put the import tariff rate of 150% (more than one and a half time in comparison with
brand new imported car) ( Decree 12/2006/N-CP). Up to now, tariffs on used cars
remain at high level in absolute value of cash rather than percentage as before.


201
The automakers complained that the current policy promoted imported CBU
automobiles, restraining their sales. And they wanted the government limited CBU
import, especially used cars.
Regulation on CKD and IKD autoparts import
Before 2007, CKD
2
(complete knock down) kits were levied import tariff rates of
from 10% to 25% depending on the autoparts deletion ratios, while IKD
3
(incomplete
knock down) kits had much lower tariff rates of from 3% to 5%. This regulation is to
encourage automakers to establish their production base in Vietnam and to improve
localization. However, due to slow progress in localization, since September 2007
import tariff on CKD and IKD sets of autoparts has been abolished in Vietnam.
The automakers complained that this tariff did not make them further localize. And
it narrowed down the domestic market because of its addition to the price of automobile
products.
Regulation on autoparts import
Since 2007, tariff on CKD and IKD set has been replaced by tariff on separated
autoparts and components (Decision No. 57/2005/QD-BTC). The change aims at
supporting the improvement of localization basing on 3 criteria: the first is to ensure
that the total tariff on all imported autoparts for an automobile is not higher than the
former tariff on a CKD kit; the second is that the average tariff of autoparts and
components is not changed in comparison with the former tariff regulation for most
types of automobiles; the third is to protect the autoparts and components that are
encouraged to produce in Vietnam.
However, the automakers said that this regulation had many inappropriate points.
Firstly, many products that currently could not be produced locally must be suffered
high tariff rates. Secondly, it made costly for packaging. Thirdly, it made complicated
for custom procedures. And fourthly, it made higher tariff on imported autoparts of
truck.
Regulation on components import
Imported components are levied lower tariff rates than the autoparts that are
assembled from those components. As regulation, each autopart is detailed into various
components or semi-assembled parts. For example, engines are levied tariff rates of
from 20% to 40% depending on types of cars, and engine components are levied tariff
rates of from 15% to 20% (carburetor, piston, piston ring, piston ring stop and cylinder
have tariff rate of 15%, meanwhile engine block, cylinder barrel and cylinder head are
taxed at 20%).
However, the automakers said that tariff rates on imported components were not
lower enough than those on imported autoparts to encourage local production of them.
And there is no incentive for components imported for production of autoparts.
Regulation on materials import
Up to now, there is no specific regulation on imported materials for the automobile
industry. The automakers expected a concrete regulation for materials import to
encourage local production of components.
3.3. Incentives for foreign autoparts makers
Although there were some suggestions on promotion of autoparts industry in
Vietnam following arguments of failure of Vietnamese automobile industry, so far there
is no specific incentives on it. Autoparts industry is offered the same incentives

202
provided to mechanical industry such as incentive investment loans, incentive
investment, export incentives, freely overseas profits remittance and localization
incentives (Circular No.176/1998/TTLT-BTC-BCN-TCHD, Circular
No.120/2000/TTLT/BTC-BCT).
However, the business side believed that the autoparts makers needed more
incentives for investment in Vietnam. Especially, they needed bigger demand for their
products to start their production.

4. Localization: Real State and Essential Factors
Over the past years, even under high protection, localization in the Vietnamese
automobile industry remains in low progress and there exist many problems in the
industry. However, the governments policy seems not to show its positive role in
promotion of localization.
4.1. Real state of localization in Vietnam
Table 1 shows local content ratios of 18 automakers belonging to VAMA
(Vietnam automobile manufacturers association). It is realized that JVs achieved lower
local content ratios than PVCs. However, in general, the autoparts localization just
remains at labor-intensive and low-technology autoparts.
Low local content ratios
Toyota Vietnam and Honda Vietnam are the best among JVs with local content
ratios in some models of products of 20% and over. Toyota Vietnam achieved local
content ratio of as high as 25% in Vios just launched on September 2007 and 23% in
Innova
4
launched on October 2006. And sedan of Honda Vietnam labeled as Civic
achieved local content ratio of 23%. In the meanwhile, the other JVs are mostly very
unmindful with localization. Mekong and VMC
5
are the two first batches of automakers
which have operated since the early 1990s but VMC only reached 12 percent local
content ratio, Mekong only got 4.6 percent and Isuzu Vietnam reached local content
ratio of 12 percent.
PVCs achieve much better local content ratios than JVs. Their local content ratios
were 35% by Vinacomin, while majority reached 40%. Vinaxuki (a private automaker)
even achieved highest local content ratio of 50%.
(Table 1)
Localization is mainly from assembling rather than autoparts
The automakers have little progress in localization. They mostly import autoparts
and components under CKD kits for assembling. Their local content ratios are added
basing on bodywork welding, stamping, painting, assembling, and testing rather than
using locally-produced autoparts and components.
Toyota Vietnam sources a limited local labor-intensive autoparts such as seat sets,
door console box, quarter trim, wire harness, battery, and exhaust pipe. Besides inviting
Denso and Harada, 100% foreign-invested autoparts makers, Toyota Vietnam opened
the first ever stamping shop and autoparts production center in Vietnam. And Honda
Vietnam has its own well-established autoparts suppliers owing to its available parts
network for motorbike products
6
. And most of locally sourced autoparts of Mitsubishi
Vietnam are assembled in its factory. In the meanwhile, the other companies almost do
not source any autoparts and components domestically. Although activities of
experience exchange, initiative contribution among workers are still promoted, there

203
is no R&D and design operation in all automakers in Vietnam.
Even though PVCs achieve higher local content ratios than JVs, their local
autoparts sourcing is limited mainly at labor-intensive and low-technology autoparts and
components, and their further localization depends on technology transfer from overseas.
In all PVCs, assembling, testing, welding, painting, stamping are implemented.
However, only two private automakers; Truong Hai and Vinaxuki equip electrostatic
painting shops. PVCs autoparts localization is mainly interiors (such as glasses, seats,
belts, carpets, so on), tires, tubes, batteries, air-cons, shock absorbers, antennas, leaf
springs, coil springs, lamps, wire harness, brakes, mufflers, bearings, technical rubber
parts, plastic parts. Vinamotor produces and assembles driving axles, engines. And in
this year, Vinaxuki also plans to build up factory for engine assembling with the
technology transfer from a German company. Currently, localization in PVCs seems to
slow down. Some plans were drawn out but they had not yet been implemented.
4.2. Essential localization factors
Though market, autoparts industry, globalization, foreign investment and cooperation
between government and business side are considered as the most important factors that
have decisive effect on localization in the Vietnamese automobile industry, they received
much complaint from the automakers for obstacles of localization.
4.2.1. Market
Although automobile production requires big demand in order to cover the
production cost, the current domestic market is so small, making localization extremely
costly. In fact, sales volume of 49 automakers just got 80,392 units in 2007. And most
automakers in Vietnam only have annual sales of some hundreds to some thousands
units a year. Toyota Vietnam topped the list of the automakers with over 20,000
domestically-assembled vehicles sold, accounting for 25% of total sales of 2007 (Sale
volume record of VAMA 2007). Meanwhile, Hino Vietnam with sales volume of 613
units (Sale volume record of VAMA 2006) in 2006 got local content ratio of only 2%;
and Daihatsu Vietnam, which just closed their production in May 2007 due to its limited
sales, had reached only 4% local content ratio (Sale volume record of VAMA 2006).
Moreover, the governments policy limits the market for the automobile industry.
Up to now, the products of local automakers are mostly to serve the domestic market,
while there are a limited export number of automobiles according to the state
agreements.
4.2.2. Autoparts industry
Despite the importance of autoparts production in development of the automobile
industry, it is said to be underdevelopment and unable to supply enough products to the
automakers. The reasons are come from the shortage of locally-manufactured autoparts
and the lack of autoparts production unification.
Firstly, there is very small number of autoparts makers. So far, there are totally
more than 70 autoparts makers (Report on Investment in Vietnamese automotive
industry, 2005) including pure Vietnamese ones and foreign invested ones.
Moreover, the local autoparts products is just limited at some labor-intensive ones
such as tires, tubes, seat sets, door console boxes, quarter trims, wire harness, tool sets,
alpine radios, manual antennas, batteries and exhaust pipes.
Additionally, there is a big difference in quality, price and delivery between pure
Vietnamese autoparts makers and foreign autoparts makers. Generally, products of pure
Vietnamese autoparts makers are supplied mostly to PVCs and unable to be supplied to

204
JVs because their quality and delivery do not meet with requirement of JVs. And
products of foreign autopart makers are just supplied to JVs and exported while scarcely
sourced by PVCs because of their high prices. In fact, JVs just accept batteries, tool sets
from pure Vietnamese autoparts makers while PVCs source only paint, tires or tubes
from foreign autoparts makers.
4.2.3. Globalization
The Vietnamese automobile industry has not yet participated in the global production
network. It only focuses into the domestic market and the locally-assembled automobiles
are sold domestically. And importation of automotive products is far bigger than their
exportation.
Moreover, governments policy remains central planned. It keeps the domestic
market isolated from the global production network. The policy makers usually require
business side to implement their targets rather than compromise in order to realize the
targets.
4.2.4. Invitation of foreign investment
The Vietnamese automobile industry has been built up from the zero with the flooded
investment of the world leading automakers and the PVCs just started development of
assembly lines imported from overseas. However, the number of foreign autoparts makers
remains extremely small and they have not yet formed a pyramid structure of automotive
suppliers as discussed above. Vietnam needs more investment to produce thousands of
autoparts, especially high technology autoparts which are not available in Vietnam and
incapable to be produced by pure Vietnamese autoparts makers.
However, the government still wants to raise state-owned automakers by granting
investment capital to some approved projects of automobile assembling and autoparts
production. It results in unfair competition among automotive makers.
4.2.5. Cooperation between government and business side
So far the government and automotive makers seem to be unable to unify their
opinions. The government independently gives out targets which lack of essential supports
conforming to the automotive makers demand.
In fact, there are a limited number of mutual consultation meetings between
government and business side and they usually result in no conclusion. Both sides can not
agree with each other on various matters. The result is that various targets of localization
are drawn by the policy makers without agreements of the automotive makers. And policy
adjustments often happen without well-informing to the automotive makers. Especially,
recent taxation changes are strongly opposed by the automakers and have no positive
effect as policy makers expectation.

5. Analysis
It is clearly that the automotive makers are extremely unsatisfied with the current
situation. And they have negative responses to the governments policies and
localization factors.
5.1. Import regulation on automotive products
The current regulation is not enough to encourage autoparts production. Although
taxes on automotive products are so regulated as tariff on imported components is less
than tariff on imported autoparts which is far less than tariff on CBU import in order to

205
give more incentives for autoparts production, there are only a modest number of
autoparts makers investing in. And the automakers still want to import autoparts for
automobile assembling rather than local production of autoparts because it is much
cheaper. Moreover, there is no regulation on materials imported for locally made
components. And there is no difference in tariff between autoparts, components and
materials imported for production and those for services. Furthermore, there is no
special incentive for local production of autoparts and components. And there is no
more incentive for components production than autoparts production. They get the same
incentives of being free from tariff on imported components for exported autoparts
assembling and free from tariffs on imported materials for exported components
production. Additionally, tariffs on autoparts and components have yet been fully
appropriate. Many products have not yet been produced locally but are applied high
tariff rates such as engine and transmission. In addition, there is no encouragement of
sourcing locally-made components for autoparts assembling.
Furthermore, these ineffective incentives for local autoparts and components
production make tariffs on them be added to prices of automobiles, so narrowing down
the market. And SCT deepens this burden because it is also plus into prices of the
products.
5.2. Regulation to limit the numbers of automakers
Out of 49 assemblers, majority are equipped with just simple assembly lines
without painting shop or/and stamping shop. They produce low quality products on
assembly lines chiefly imported from China. This kind of automakers might be excluded
from the automobile industry. However, so far regulations of factory standard and
product standard are not suitable enough to stop their operation.
5.3. Regulation on foreign investment
Foreign investors play very important role because they possess not only
technology to produce autoparts meeting with requirement of JVs, but also management
of technology. They also can do marketing the products domestically and internationally.
Thai automobile industry is successful owing to its active invitation of foreign
automakers and autoparts makers and creation of a fair competition in the domestic
market. Although there are no special incentives for pure Thai companies individually,
pure Thai autoparts makers are gradually developed to support the higher tiers of
autoparts makers (Sakai, 2005, pp. 137-169).
For the foreign investment regulation, general speaking, all automotive makers
enjoy the same policy. However, the state-owned enterprises receive incentive of the
government by cash grants or products orders, so creating unfair competition among
automotive makers. Moreover, these grants are not useful because the fact proves that
state-owned automakers are less effectively operated than the private automakers.
5.4. Market expansion policy
The recorded automobile sales volume achieved in 2007 was just 80,392 units.
This number is just equivalent to the Thai automobile industrys production volume of
82,106 units in 1985 (For-in's World Automobile Statistic Year Book, 2007) when Thai
automobile industry reached local content ratio of 50% and sourced mainly
labor-intensive autoparts and components (Ishizaki, 1994, p.23).
The small market is caused by many reasons. Firstly, Vietnams infrastructure is in
such poor condition that it is already reaching the limits of transportation. Parking lots
are very limited and many small streets are not suitable for motorization. Secondly,
imported CBU vehicles including second-hand ones add more pressure of competition

206
to domestically-produced vehicles. The imported vehicles in 2006 were 12,619 units,
accounting for 20% of total newly-registered vehicles (Report on Investment in
automotive industry, 2006). Thirdly, taxes are not suitable with the industrial facts,
making higher price of locally-assembled automobiles, so narrowing down the domestic
market and limiting their exports. Fourthly, the local automakers provide all types of
automobiles including sedan, SUV, bus, heavy truck, light truck, minivan, etc. In each
segment, several models of cars are supplied in a tiny volume and no type of automobile
or model are promoted for mass production.
To encourage mass production, Thai automakers paid big efforts to promote
market-selected one-ton pick up trucks for 60 million Thai people. Market share of
which was 60% with 170,059 units in 1990 (For-in's World Automobile Statistic Year
Book, 2007). However, Thailand achieved local content ratio of 85% in 1998 that
included body panel stamping and engine parts casting and forging when its sales
volume reached the scale of 525.000 (Sakai,2005.pp.137-169). Thus it soon acquired
world competitiveness in cost and quality.
5.5. Regulation on encouragement of globalization
In the current wave of globalization, labor division is happening around the world
and in spite of visible or invisible obstacle, automotive products are exported and
imported, so Vietnam cannot stand outside of global production network. Most foreign
autoparts makers export products to broaden their market. And technology is transferred
for local production. However, Vietnam has not yet fully participated in the global
production network. Although concepts of autoparts mutual sourcing and automobile
joint production as well as countrys specialization of autoparts production have been
arisen for a long time (Guiheux and Lecler, 2000, pp. 218,219), Vietnam has not yet
been taken them into consideration. Though the neighbors of Vietnam as Thailand,
Malaysia, Indonesia and Philippines (ASEAN-4) aggressively compete to invite foreign
investments, Vietnam seems to take no such kind of action. Even though ASEAN-4 is
considered by the worlds leading automakers in their plans of location of their regional
production network, Vietnam seems not to be in their consideration (Ishizaki, 1996,
pp.2-48).
Moreover, the Vietnamese government tends to consider globalization as an
endanger rather than an opportunity, so taking advantage of globalization does not
receive serious consideration. Export of automotive products to overseas markets and
exchange of automotive products with other countries have not withdrawn attention of
the policy makers. In the meanwhile, other countries are very eager to create common
market for mass production, such as ASEAN autoparts complementation scheme
(Guiheux and Lecler, 2000. pp. 213,214).
5.6. Autoparts maker invitation policy
Ordinarily, a carmaker needs at least 20 autoparts suppliers (Sturgen, 1998). But
up to now, Vietnam has 14 JVs and 35 PVCs with about 70 autoparts makers, so each
automobile company on average has only one or two domestic autoparts suppliers. In
comparison with Thailand (Sakai,2005.pp.145-150), it has more than 300 autoparts
makers that supply to 13 local automakers, so on average each automakers has more
than 20 autoparts makers.
Moreover, the autoparts makers in Vietnam just provide some labor intensive
autoparts while an automobile needs high-tech and capital-intensive autoparts. They
establish a pyramid of automotive makers, in which automakers are on the top and
autoparts makers are on the second, third and lower tiers. In the Thai automobile
industry (Sakai, 2005,pp.145-150), there are up to more than 300 autoparts makers with

207
110 tier-1 suppliers and other 200 in tiers - 2 and 3. Among tier-1 suppliers, 50 are pure
Thai makers while 60 are Japanese and American ones. The establishment of autoparts
industry is owing to active invitation from automakers in Thailand for localization of
their automobiles with supports from the government.
Furthermore, each automaker in Vietnam has their own product standard, so they
can not source from the same pool of autoparts makers. Even one autoparts maker can
only supply to a limited number of automakers. For example, both Toyota Vietnam and
Honda Vietnam produce sedan but their autoparts suppliers are different. This
difference are even enlarged when comparing between JVs and PVCs because pure
Vietnamese autoparts makers that mostly are subsidiaries of the PVCs, usually provide
autoparts of low price meeting with requirement of PVCs, while foreign autoparts
makers produce autoparts of high quality which meets with requirement of JVs. This
phenomenon even deepens the shortage of autoparts. So the best way to narrow down
this gap is to invite foreign autoparts makers. However, the current incentives for
autoparts production are not good enough to convince autoparts makers to invest in.
5.7. egotiation with automakerss and among automotive makers
Unconcluded meetings between the policy makers and automotive makers usually
happen because of their viewpoint disagreement. While Vietnamese policy makers tend
to be friendly with planning targets without proper evaluation of industrial facts or
global trends, the automakers consider competitive factors of cost, quality and delivery
as most important in their business operation. So the policy makers think that the
automakers are not making enough effort in localization and technology transfer and
that they usually overstate the difficulties they face to entice more favorable condition.
The automakers in turn think that the government is making unreasonable demands
without knowing industrial facts or global trends. This results in unexpected
consequence to both sides, that the government can not achieve their targets, while the
automakers cope with difficulties and damages.
However, Thai policy makers seem to be wise in keeping close relation with
automobile companies. They find out a way to reconcile benefits of both sides. For
example, at the end of 1980s, policy makers decided to promote local production of
engine, so meetings were hold between policymakers and automakers. The government
representatives made clear what they wanted and automakers pointed out their
difficulties. Finally, the government agreed to grant more land for automakers to locate
near Bangkok and gave tax exemptions to supplementary engine components imported
for assembling engines main parts of which were manufactured locally.

6. Recommendation
It could be said that the governments policy is not effective in encouraging the
Vietnamese automobile industry. Thus much change in the policies basing on review of
viewpoints of policy makers is essential to create a better future of the industry.
6.1. Incentives to invite foreign autoparts makers
To encourage autoparts makers to localize their products in Vietnam, the
government should actively invite them with special incentives as well as facilitate
automakers in expansion of market.
The current policy just promotes automakers via the local content requirement.
Actually main customers for domestic autoparts makers are domestic automakers.
However, domestic automakers purchase orders are not enough to encourage autoparts

208
makers to localize autoparts in Vietnam. So they need more incentives to export also in
order to cover the short production volume of domestic automakers. But for domestic
autoparts makers, overseas marketing is not easy. So they are advised to engage in
global production network.
At the same time, it is suggested that good foreign autoparts makers are invited
positively to invest in Vietnam. The Vietnamese government needs to be active in
drawing autoparts makers. Firstly, autoparts makers need to be facilitated by guarantees
of long-time land renting permission, factory permission and working visa for foreign
engineers. Secondly, they need to be provided a stable business environment by
guarantee of not nationalization and of a transparent and stable policy. Thirdly, they
need an encouragement of export by reduction of sale tax. Fourthly, they need to be
stimulated to further invest or reinvest their profits in production enlargement.
6.2. Taxation
Taxation must be regulated to encourage localization, so incentives must be
increased from the lowest encouraged one for CBU automobile import to the highest
encouraged one for materials import for production of components.
CBU automobile import should be kept high tariff, while used car import should be
banned in order to protect domestic market for locally made products.
Locally assembled automobiles should be considered as common products for
transportation rather than luxury products. The government should abolish special
consumption tax to reduce their price, so encouraging their consumption domestically
and internationally. Moreover, special consumption taxes, sales taxes for strategic
vehicle as suggested below should be cheaper as this is promoted for mass sales and
accordingly mass production. Additionally, taxation discount for export (for example,
exemption from export tariff and sales tax) should be regulated to encourage export.
Autoparts import tariff for locally-produced automobiles should be totally
decreased in order to reduce production cost. Meanwhile higher tariff for import of such
autoparts to be produced locally is kept, the high technology ones should be imposed
low tariffs at least in this period. However, tariff regulations can be adjusted later on;
for example, tariff on high technology autoparts can be increased to promote local
production. Tariff on labor-intensive and low technology autoparts import can be kept
high for protection of locally made ones.
Locally assembled autoparts should be decreased of sales tax. And rewarded
taxation discount for export and taxation discount for localization should be introduced.
Components import for locally assembled autoparts must be free from import tariff.
But imported components for service should be levied higher tariff. Localization should
be encouraged by tax discount.
Local production of components needs to be accelerated by exemption from sales
tax, rewarding of taxation discount for export and tax discount for localized
components.
Materials import for production of components needs to be given much incentive
including tariff free.
6.3. Promotion of a certain kind of automobile
For such small current market, positive promotion of a selected kind of vehicle
may help increasing people demand, and thus it may attract automakers and
accordingly autoparts makers to enjoy economy of scale for localization. This measure

209
is especially appropriate for such tiny domestic market that is shared by many
automakers and models. However, to realize it needs enthuses of automakers and good
supports of the government.
The automakers in Vietnam can choose minivans or light trucks as a strategic
vehicle which their lower price can fit to majority of Vietnamese people as cargo
transportation as well as passenger carrier for low and middle-income people. These
types of automobile have an increasing demand owing to the development of
agricultural sector as well as commodity transportation demand in the cities. Estimation
of the Ministry of Industry shows that light truck is anticipated to have the biggest
demand of 100,000 units by 2010 (Report on Strategy to develop the Vietnamese
automobile industry, 2003). The success of one-ton pickup in Thailand for securing big
demand enabling mass production and export through localization should be studied.
6.4. Infrastructure Improvement
Good infrastructure is significant element in encouraging citizens to buy cars. In fact,
there is a big demand of cars for transportation among people in big cities, but many
people are hesitated buying cars because of poor condition of Vietnams infrastructure.
Vietnam needs to push up motorization, including roads, bridges, highways, petrol
distribution, vehicle sales and service, parking lots, local government apparatus to regulate
the use of vehicles (e.g. traffic police, vehicle registration), and driver training programs.
A long-run plan in infrastructure improvement also releases policy makers from the matter
of traffic jam in big cities, so they can freely promote domestic consumption of cars.
6.5. Encouraging export and limiting CBU import
The government should give more incentives on export of automotive products,
especially autoparts, through for example, rewarding more points on localization or tax
and tariff reduction for enterprises which are going to export their autoparts.
When the ban of CBU import does not conform to WTO free trade agreement, the
technical measures of product standard, technology standard could be set to limit
low-technology and used automobile import.
6.6. Limitation of the number of automakers
When the limitation of investment does not conform to free trade agreement with
WTO, the technical measures of product standard, technology standard and factory
standard could be set to limit short-term investment, while encourage long-term
investment, thus help to narrow down number of assemblers and exclude badly-invested
assemblers. At the same time, investment conditions such as localization plan,
investment capital and technology capability could be required.
6.7. Promotion of local participation in ASEAs production network
ASEAN is a common market, in which each country plays a specific role and
contributes to the overall production scheme in the area. Vietnam should not develop an
independent and isolated automobile industry by production of every item and self
supply to its domestic market. It should utilize its dynamic comparative advantages to
supplement the other countries. To realize it, firstly Vietnam should engage in ASEAN
autoparts complementation scheme (according to it, import tariff for autoparts and
components would be half if they are originated from ASEAN countries). By taking
advantage from common market and mutual exchange with ASEAN makers,
Vietnamese autoparts makers may become strong enough to supply locally and
overseas.


210
6.8. Mutual consultation of policy
First of all, the viewpoints of the policy makers must be changed to conform to
free trade, free competition, and globalization.
Secondly, encouragement of cooperation between government and business side is
necessary. The government must consult with business side on localization steps,
regarding what autoparts or components should be locally produced, what support is
necessary to the automotive makers, how clear of taxation schedule conforming to the
governments policy, free trade agreements, etc.
Thirdly, how to realize the agreed targets should be proposed by business side.
They may consult with each other to accelerate production cooperation of autoparts.
They should decide how to promote autoparts production or how to invite autoparts
makers from abroad.

7. Conclusion
Amid the strong economic globalization, the infant Vietnamese automobile
industry might not be supported merely by domestic market protection and
governments administrative requirements. It needs to be built up by competitive
companies that are supported by the government to expand market and enter into the
global production network.
7.1 Government's role
Government must create an environment that support rising productivity. This
limits intervention of the government in some areas such as trade barriers and pricing
and stimulates its role in ensuring vigorous competition, providing high education and
so on. Any subjective requirement from the government that goes against companies
benefits and economic rules has no positive effect on the development of the automobile
industry. A shifting environment of policy adds uncertainty to an already volatile
business setting. Furthermore, the government must admit globalization as
indispensable tendency and encourage local automakers and autoparts makers to join in
global production network.
7.2 Automobile and parts makers' responsibilities
Short-term investments for utilizing current incentives from the government are
not welcome. The companies should be active in localization with long-term vision,
encouraging their suppliers to invest locally or from abroad, and giving invitation to
their suppliers, giving suggestion to the policy makers to create good business
environment and good policy.
7.3 Globalization and local participation
Each country needs to take advantage of their comparative merits, to locate its own
position in the global economy. Vietnam also needs to admit that automotive production
is globalized now and it should utilize worlds leading automakers and autoparts makers
who may have big effect on Vietnamese automobile industry.
Traditional protection measures are no longer to conform to tendency of trade
liberation and globalization. The industrial policies need to be more actively changed to
be suitable with the current situation of global production.
7.4. Marketing of suitable type of automobile
It is very necessary for mass-production and participation in global network to

211
specialize in a product. The business side should promote their role in marketing
suitable products and the government should play the role as supporting business side.

212
Footnotes
1. Vinamotor exported 100 buses to Dominica. Daewoo Vietnam exported 300 buses to
Iraq.
2. CKD1 means all parts imported, painting done in Vietnam.
3. CKD2 means all parts imported, body welding and painting done in Vietnam.
4. IKD means greater than 10% of locally-sourced par
5. Innova is a MPV car of Toyota Vietnam, which is best sold in Vietnamese market in
2007.
6. In fact, VMC currently stops assembling cars in Vietnam; they mainly import cars for
sale from Germany. Mekong also operates perfunctorily with small production volume.
7. Before 2005 Honda Vietnam only focused on motorbike manufacture and was very
successful in Vietnamese market. However, since 2005, Honda Vietnam started its
automobile production base with first output of brand name Civic.

213
References
1. Takayasu, Ken-ichi (1998). Developing the Vietnamese automobile industry. RIM
Pacific Business and Industries, 40 (2), 22.
2. Tng kt cui nm v u t trong ngnh t, 2007- Report on investment in
automotive industry, Vietnamese Ministry of Planning and Investment, 2007.
3
.
Quy hoch pht trin ngnh t, Plan to develop the automobile industry, Vietnamese
Ministry of Industry, 2007.
4. Report on investment in automotive industry, 2007.
5
.
Plan to develop the automobile industry, 2007.
6. Ngh nh 12/2006/N-CP. Decree 12/2006/N-CP. Jan 23rd 2006
7. Quyt nh s 57/2005/QD-BTC v vic sa i thu u i xut nhp khu ph
tng t- Decision No. 57/2005/QD-BTC on amendment of incentive import
and export tariffs levied on autoparts, Minister of Finance, October 8th 2005.
8. Circular No.176/1998/TTLT-BTC-BCN-TCHD, Ministry of Finance, Dec 25
th
1998
9. Circular No.120/2000/TTLT/BTC-BCT, Ministry of Finance, Dec 25
th
2000.
10. Sale volume record of VAMA 2007.
11. Sale volume record of VAMA 2006.
12
.
Tng kt cui nm v u t trong ngnh t Report on Investment in Vietnamese
automotive industry, Vietnamese Ministry of Planning and Investment, 2005.
13. Sakai, Hiroyuki (2005). Accumulation of autoparts industry in Thailand, pp.
137-169 in Kobayashi, Hideo, Global cooperation in autoparts industries in East
Asia, Bunshindou.147-150.
14. For-in's World Automobile Statistic Year Book, 2007.
15. Ishizaki, Yukiko (1994), The automobile industries of ASEAN countries: Toward
greater regional cooperation and competitiveness. RIM Pacific Business and
Industries, Vol.I, 23.
16. Report on Investment in automotive industry, 2006
1

17. Guiheux, Gilles and Lecler, Yveline (2000). Global Strategies and local realities.
Great Britain: Macmilan Press Ltd. 218,219.
18. Ishizaki,Yukiko and Takayasu, Kenichi (1996). The imminent advent of the age of
global competition for the automobile industry in Southeast Asia. RIM Pacific
Business and Industries.Vol.III.2-48.
19. Sturgen, Timothy J. (1998). The automotive industry in Vietnam: Prospects for
development in a globalizing economy, Report prepared for Development strategy
Institute- Vietnamese Ministry of Planning and Investment and Medium-term
Industrial Strategy Project- United Nations Industrial Development Organization,
July 1998.
20. Tng kt cui nm v u t trong ngnh t, 2006- Report on Investment in
automotive industry, Vietnamese Ministry of Planning and Investment, 2006.

214
Table 1. Local content ratios of automakers in Vietnamese automobile industry
Source: Field research in March 2007


Automakers Types of cars
Local content
ratio (%)
Toyota Vietnam Passenger cars, minibus 15-25
Ford Vietnam Passenger cars, pick-up, minibus 6,45
Mitsubishi Vietnam Passenger cars, truck 10-14
Isuzu Vietnam Passenger cars, truck, pick-up 12
Vietnam Suzuki Passenger cars, truck, van 10
Daewoo Vietnam Passenger cars, bus 8
Mercedes-Benz Vietnam Passenger cars, minibus, bus 1.5
Honda Vietnam Passenger cars, 23
VMC Passenger cars, truck 12
Hino Vietnam Truck 2,06
Daihatsu Vietnam Passenger cars, truck, van 4
JVs
Mekong Passenger cars, pickup, truck, minibus 4.6
Samco Truck, bus 40
Vinamotor Truck, bus 40
Veam Truck, bus 40
Vinacomin Bus 35
Vinaxuki Truck, bus 40-50
PVCs
Truong Hai Truck, bus 40