Direct, indirect and total effect in spatial analysis of provincial FDI in vietnam

UNIVERSITY OF ECONOMICS ERAMUS UNIVERSITY ROTTEDAM HO CHI MINH CITY INSTITUTE OF SCOCIAL STUDIES VIETNAM THE NETHERLANDS VIETNAM – THE NETHERLANDS PROGRAMME FOR M.A IN DEVELOPMENT ECO

UNIVERSITY OF ECONOMICS ERAMUS UNIVERSITY ROTTEDAM

HO CHI MINH CITY INSTITUTE OF SCOCIAL STUDIES VIETNAM THE NETHERLANDS

VIETNAM – THE NETHERLANDS

PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS

DIRECT, INDIRECT AND TOTAL EFFECT IN SPATIAL ANALYSIS OF PROVINCIAL FDI IN

UNIVERSITY OF ECONOMICS ERAMUS UNIVERSITY ROTTEDAM

HO CHI MINH CITY INSTITUTE OF SCOCIAL STUDIES VIETNAM THE NETHERLANDS

VIETNAM - NETHERLANDS

PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS

DIRECT, INDIRECT AND TOTAL EFFECT IN SPATIAL ANALYSIS OF PROVINCIAL FDI IN

VIETNAM

BY

LE VAN THANG

ACADEMIC SUPERVISOR

DR NGUYEN LUU BAO DOAN

HO CHI MINH CITY, November 2016

DECLARATION

“This is to certify that this thesis entitled “Direct, Indirect and Total effect In Spatial Analysis of Provincial FDI in Vietnam”, which is submitted by me in fulfillment of the requirements for the degree of Master of Art in Development Economics to the Vietnam – The Netherlands Programme (VNP)

The thesis constitutes only my original work and due supervision and acknowledgement have been made in the text to all materials used.”

Le Van Thang

ACKNOWLEDGEMENT

This thesis could not be accomplished without the supporting and the motivation that I have received from many people It is a pleasure to convey my gratitude to them all in my humble acknowledgment

Foremost, I would like to express my sincere appreciation to Dr Nguyen Luu Bao Doan, my supervisor He gave me the greatest supporting, energetic assistance and valuable guidance as well

as an infinite patient to encourage me to complete my very first research Without Dr Nguyen Luu Bao Doan, this study would never finish

Besides, I also would like to give my gratitude to the Vietnam- Netherland Programme, especially

to all lecturers who provided me valuable knowledge, VNP staffs for their restless assistant for the time I have been studying in VNP as well as School of Economics

I would love to express my gratefulness to Prof Nguyen Trong Hoai and Dr Pham Khanh Nam for the first suggestion to encourage me to deal with a novelty field of my knowledge - spatial analysis Moreover, I would like to give my sincere thankfullness to Dr Pham Khanh Nam who has provided a valuable data source for me to complete this thesis

Besides that, I would like to thank all my friends, my fellows at the University of Economics, Ho Chi Minh City, my groups and all the classmates in K20-VNP All of them are always be my side encourage and support me to complete the thesis

Finally, I would like to send my gratefulness to my father, my mother, my two little brother, sister- Van Nam and Thuy Linh for their love, sacrifice, tremendous support for me not only to complete this thesis but also for my whole life

ABBREVIATION

AIC: Akaike Information Criteria

ESDA: Exploratory Spatial Data Analysis

EU: European Union

FDI: Foreign Direct Investment

GDP: Gross Domestic Product

GSO: General Statistical Official

GRP: Gross Regional Product

PCI: Provincial Competitiveness Index

LM test: Lagrange Multiplier test

MPI: Minister of Planning and Investment

MNE: Multinational Enterprises

SAR: Spatial Autoregressive Model

SDM: Spatial Durbin Model

SEM: Spatial Error Model

USAID: The United States Agency for International Development

VCCI: Chamber of commerce and industry

VIF: Variance Inflation Factor

ABSTRACT

This paper investigates the spatial pattern of Foreign Direct Investment (FDI) for all 63 provinces in Vietnam from 2011 to 2014 Empirical studies on locational determinants of FDI typically neglected the spatial interaction among observations which lead to inefficient and biased estimations Indeed, Moran’s I suggested by Moran, which is used to detect the spatial autocorrelation in data pattern of both dependent and independent variables, give hints of the necessity of spatial econometrics in analyzing the FDI determinants

Through General To Specific approach, the Spatial Durbin Model (SDM) has been chosen

as the most appropriate model, compared with other models like Non-spatial model, Spatial- Autoregressive Model (SAR) and Spatial Error Model (SEM) This study finds that the FDI flow into one province negatively spatially affects FDI inflow in remaining provinces Moreover, by applying SDM, this paper econometrically estimates the impact of host province’s determinants and its neighbor determinants on its FDI inflow

Keywords: Foreign Direct Investment, Moran’s I, Spatial analysis

CONTENTS

DECLARATION i

ACKNOWLEDGEMENT ii

ABBREVIATION iii

ABSTRACT iv

LIST OF FIGURE vii

LIST OF TALBE viii

CHAPTER 1: INTRODUCTION 1

1.1 Problem statement 1

1.2 Research objective 3

1.3 Research questions 3

1.4 Scope of the study 3

1.5 Thesis structure 4

CHAPTER 2: OVERVIEW OF FDI IN VIETNAM 6

2.1 Stages of foreign direct investment in Vietnam 6

2.2 Distribution of foreign direct investment among provinces 8

2.3 Country of origin 10

2.4 Sectors of foreign direct investment 11

CHAPTER 3: LITERATURE REVIEW 14

3.1 Theories about location choices of foreign direct investment 14

3.1.1 The eclectic paradigm OLI 14

3.1.2 Agglomeration and foreign direct investment 16

3.2 The inter-dependence of FDI between locations 17

3.2.1 MNE choice theory 17

3.2.2 Agglomeration effect 20

3.3 Empirical studies 21

3.3.1 Empirical studies of FDI determinants in spatial analysis 21

3.3.2 Empirical studies of FDI determinants in Vietnam 23

3.3.3 Fundamental FDI determinants 26

CHAPTER 4: DATA AND METHODOLOGY 31

4.1 Data sources 32

4.1.1 Dependent variable 32

4.1.2 Explanatory variables 33

4.1.3 Descriptive statistics 36

4.2 Spatial econometric model 37

4.2.1 Spatial Autoregressive Model 38

4.2.2 Spatial Error Model 38

4.2.3 Spatial Durbin Model 38

4.2.4 Marginal effect in Spatial Durbin Model 39

4.2.5 Model selection 40

4.3 Pre-test for spatial existent with Moran’s I 43

4.4 Spatial weight matrix 44

4.5 Comparisons of models 46

CHAPTER 5: EMPIRICAL RESULT 47

5.1 Direct effect 51

5.2 Indirect effect 54

5.3 Total effect 55

CHAPTER 6: CONCLUSION 56

6.1 Main finding 56

6.2 Policy implication 57

6.3 Limitation and future research 57

REFERENCES 59

APPENDICES 64

LIST OF FIGURE

Figure 2.1: Registered, implement FDI (million USD) and Number of FDI projects 7

Figure 2.2: The distribution of FDI in Vietnam from 1988 to 2014 9

Figure 2.3: The sector distribution of FDI 12

Figure 3.1: Analytical framework of FDI and determinants 31

Figure 4.1: General to Specific strategy 42

Figure 5.1: The Local Moran’s I of FDI inflow Vietnam in 2011-2012-2014 49

LIST OF TALBE

Table 2.1: Sharing of FDI in Vietnam from 1988 to 2014 7

Table 2.2: Top ten countries of origin of FDI in Vietnam 11

Table 3.1: Multinational Enterprise Motivation 18

Table 4.1: The variable descriptive 35

Table 4.2: The summary statistics of variables 36

Table 5.1: The Moran’s I coefficient of FDI 47

Table 5.2: The Moran’s I coefficient of explanatory variables 48

Table 5.3: The AIC value 50

Table 5.4: The Marginal effect of Spatial Durbin Model 52

CHAPTER 1: INTRODUCTION

1.1 Problem statement

Foreign Direct Investment (FDI) plays a major role in the countries’s growth, especially in developing countries, thank to its benefits, including technological transferring, management skill, job creations, and other positive externalities Moreover, the FDI is considered as one of the essential elements for economic development (Cave, 1996; Nwaogu, 2012)

Aware of these positive effects, nations have implemented several manners to promote the FDI inflow such as issuing supportive law and policies, opening their market, enhancing the human capital or improving the infrastructure capability Particular to Vietnam, since the “Doi Moi” in

1986, the economy system was reconstructed from planned economy into the market economy The foreign sector is accepted as a component of the economy Vietnam started its new policies to attract the FDI inflow and become an attractive destination for investment from abroad

Due to the growth of FDI activities, researchers has paid considerable attention into finding the FDI determinants in recent years Blanc-Brude et al (2014) have reviewed hundred previous studies on FDI determinants with varying scales: countries within a region or sub-national in a country For sub-national level, there are some remarkable studies such as Cheng and Kwan (2000), Sun, Tong and Yu (2002), Kang and Lee (2007) for China, Crozet et al (2005) for France, Guimarães, Figguieredo and Woodward (2000) for Portugal Regarding to Vietnam, the examining

on FDI’s determinants at the provincial level are relatively inadequate There is only few papers

in this field, like Pham (2002), Meyer and Nguyen (2005), Anwar and Nguyen (2010)

Nonetheless, the similarities of above studies is that they have assumed each region is isolated and have no impact on the others With this assumption, the amount of FDI inflow to each region are functioned by its characteristic only and therefore, these researchers just explored the disparity of

FDI in term of locational determinants However, according to the Tobler’s law (1970):

“Everything is related to everything else, but near things are more related than distant things”

To illustrate for the Tolber’s Law, Neumayer and Plumper (2010) gave an example of a person in attempt to avoid the traffic jam to arrive the destination as quick as possible One conclusion might

be obtained from this example is that the time travel for this person to reach the destination would

be a function of the vehicle used, the velocity, the road route utilized Also, the amount of time

depends on the time of others to arrive their destination Besides, his travelling time also depends

on the other’s options such as their vehicle, their velocity, and their road route

The Tolber’s law is also applied in examining the FDI determinants For instance, if Ha Noi attracts more FDI would possibly boost or deteriorate the FDI inflow of its nearby neighbors, or if Ha Noi holds a good infrastructure or a high level of human capital, then it would not only assist to attract more FDI inflow itself but also possibly make positive externalities on nearby provinces such as Ninh Binh or Hai Duong More comprehensive, this implies that the level of FDI inflow in one province not only depends on its determinants but also influenced by the FDI inflow of other nearby provinces as well as their determinants Alternatively, the geographical proximity between provinces in Vietnam also contributes a particular effect on the level of FDI inflow and the closer proximity-the stronger effect The impact caused by the proximity between regions is known as spatial effect

Therefore, due to the existence of spatial effect, the reliability of previous studies on FDI determinants with assumption that regions are distinct, is in doubt According to Anselin (1988), the omission of spatial effect lead to biased, inconsistent or inefficient parameter estimates As a result, these spatial effects should be controlled to yield a more accurate estimation Nevertheless, previous works on FDI determinants of Vietnam provided a useful suggestion for selecting potential determinants Currently, there are only two empirical studies of Hoang and Goujon (2014), Esiyok and Ugur (2015), which embraced the spatial effect By applying two different models, they stated that the FDI inflow to provinces in Vietnam has impact on one another with different signs However, the restraint of using two basic spatial models does not allow them to distinguish the real impact of characteristics from nearby provinces on the host province

Thereby, this study is expected to partially fulfill the drawbacks in previous studies by accounting for the spatial interaction between provinces in investigating the FDI determinants, which might offer more precise results Especially, by applying recent spatial econometric techniques, this study aims to reveal not only the spatial dependence of FDI in Vietnam but also the effect of alternative provinces determinants on the FDI flow into one province

1.2 Research objective

As discussed above, the FDI inflow to each province does not simply rely on its determinants but also be mutually affected by the FDI inflow to other provinces and their determinants through the spatial interaction Followed by that, this study is designed to analyze the FDI determinants at provincial level and examine the spatial interaction between provinces in attracting FDI

1.3 Research questions

In order to reach the research goal, this study focuses on answering two main research questions:

(1) Does the spatial dependence of FDI inflow between provinces in Vietnam exist?

(2) Which determinants would affect the FDI inflow at the provincial level?

1.4 Scope of the study

This study adopts the panel data at the provincial level for all 63 provinces in Vietnam from 2010

to 2014 The amount of registered FDI in US dollars as the dependent variable is collected from Vietnam Statistical Year Book by the General Statistical Official of Vietnam (GSO) and the Provinces Statistical Yearbook from 2011 to 2014

The Gross Domestic Product (GDP) proxy for the market size, the sum of export and import over the GDP proxy for the degree of openness and the proportion of employment in foreign firms over the total employment proxy for the agglomeration These data are collected from the Provinces Statistical Yearbook from 2010 to 2013 Additionally, the rate of trained labor force over 15 years

of age is employed to proxy for the labor quality and the monthly average income of wage workers over 15 years of age proxy for the labor cost The data of labor quality and labor cost are gathered from Report on Labor Force Survey from 2010 to 2013 In order to account for the infrastructure, the dummy variable of national marine port is employed The data of national marine port in Vietnam are retrieved from the Vietnam Government Website Finally, the Provincial Competitiveness Index (PCI) is used to proxy for the institution variable The PCI data is collected from VCCI reports from 2010 to 2013

Firstly, this study applies the Moran’s I to detect the spatial autocorrelation in the data pattern of the dependent variable and explanatory variables If the spatial autocorrelation exists, then the using spatial econometric models are necessary Through the General to Specific approach, the 3 spatial models include the Spatial Durbin Model (SDM), Spatial Autoregressive Model (SAR) and

the Spatial Error Model (SEM) will be estimated with the binary contiguity weight matrix Besides the binary contiguity weight matrix, other spatial weight matrixes include the inverse distance weight matrix, cut-off km weight matrix and the k-nearest weight matrix will also be estimated to give the AIC value which is used as criteria for choosing the most efficient weight matrix By using the Wald Test, the SDM should be pointed out as the best model describes the data over SAR and SEM In addition, a non-spatial model will also be estimated to give the information for comparing the results between non-spatial model and spatial model Last, the Hausman test will

be employed to find out if the time-fixed effect is more favorable than the random effect

1.5 Thesis structure

This study is organized as following: Chapter 1 presents the necessity of using spatial techniques

in investigating the FDI determinants as problems statement Chapter 2 includes the stages of FDI, locational distribution of FDI, sector of FDI and country of origin Chapter 3 introduces some theories of FDI that relate to the location choice of Multinational Enterprises (MNE), an extraordinary concept of agglomeration and its effect on FDI This chapter also explains how FDI interdependence in the aspect of agglomeration effect and MNE motivations Moreover, the empirical study part will review some outstanding studies that examined spatial patterns of FDI from the world Last, empirical studies on the FDI determinants of Vietnam with and without spatial interaction will be also discussed

The data part introduces the measurement of dependent and explanatory variable as well as their sources In addition, the methodology introduces the three main spatial econometric models includes SAR, SEM and SDM, their properties, disadvantages/advantages of each model and their relationship In addition, this part also demonstrates the technical to obtain the marginal effect in SDM Besides, this part also provides readers the detail of Moran’s I test which could carry out the spatial autocorrelation of dependent variable as well as explanatory variables As a key component of spatial econometrics, this part will also introduce the construction of a spatial weight matrix which stands for the spatial linkage between provinces In this part, four spatial weight matrix include the binary contiguity weight matrix, inverse distance weight matrix, binary contiguity weight matrix with the cut-off km criteria and the k-nearest spatial weight matrix will also be presented Chapter 5 is the result which embraces the Moran’s I result and estimation

spatial regression result of SDM include direct effect, indirect effect and total effect The last part

of this study is the conclusion, policy implication and limitation

CHAPTER 2: OVERVIEW OF FDI IN VIETNAM

It had been nearly 30 years since the economy reform from planned economy into the market economy in 1986, the Vietnam’s economy has integrated intensity with the rest of world The Vietnam has become one among the most rapid economies growth around the world Undoubtedly, the FDI is an essential factor to the impressive growth of Vietnam As reported by Nguyen and Nguyen (2007), the FDI capital contributes a critical role in the development of economy growth

as well as the progress of poverty reduction and improve the standard of living The following part will introduce details of FDI in Vietnam in term of stages of FDI process, provincial distribution

of FDI, country of origin of foreign investment and sectors of foreign investment

2.1 Stages of foreign direct investment in Vietnam

Parallel with the opening market and economy reforming, Vietnam had established the legal framework by legislating the first investment law (FIL) in 1987 This is considered the earliest effort of Vietnam government in calling the foreign capital inflow With the advantage of emerging market, Vietnam has become a new attractive destination for the foreign capital from around the world In general, the trend of FDI inflow Vietnam could be separated into five stages from 1988

to 2014 as figure 2.1 This part will introduce the details of the five stages and some changing that take significant impact on the FDI in Vietnam

The initial stage is from 1988 to 1990 with the legislation of the foreign investment law in 1987 During this period, the overall amount of registered FDI capital into Vietnam had reached 1.8 million dollars However, the FDI inflow in this time remains quite low which is caused by several constraints such as the high inflation rate, the infrastructure’s abilities in adapting to the requirement of foreign investors Moreover, according to Thuy Le (2007), foreign investors were still so anxious and doubted on the Vietnam government commitment on economy reforming to guarantee their benefit

The second stage is from 1991 to 1997 This stage is marked by the amending of FIL in 1992,

1996 and the ceasing of U.S embargo policy in 1995 The amending of FIL in 1992 had adjusted two basic regulations include the Build-Operate-Transfer, which is considered a form of FDI and the duration of FDI projects could be extended up to 70 years Especially, the amending of FIL in

1996 which permits the provincial authorities to issue investment licenses with certain

requirements on their own As a result, the registered FDI inflow Vietnam for this period had accrued rapidly and reached 9.63 billion dollars in 1996

Figure 2.1: Registered, implement FDI (million USD) and Number of FDI projects

Source: GSO

The next stage is from 1997 to 2000 that marked by the Asian financial crisis During this period, the amount of FDI inflow dropped sharply to the level even lower than 1993 The amount of registered FDI in 1997 with 5.95 billion dollars declined to 2.82 billion dollars in 1999 and 2.76 billion dollars in 2000 According to Nguyen and Nguyen (2007), even though the Vietnam economy had not suffered too many the negative impact from the crisis, but almost the foreign investors originate from the countries that had to face the crisis As a result, foreign investors are forced to abandon or slowdown FDI projects in Vietnam This leads to the declining of FDI inflow for this period

The fourth stage is from 2000 to 2008 and this stage could be separated into two small periods from 2000 to 2006 and from 2006 to 2008 The amendment of Foreign Investment Law in 2000 has triggered a new expansion phase of the FDI in Vietnam However, the amount of FDI inflow

in this period is not too noteworthy and remain relatively low In 2006, Vietnam became a member

of WTO that required the equality treatment between foreign enterprises and domestic enterprises

as well as cut-off the tariff barriers As an upshot, this had encouraged more foreign investors to

0 200 400 600 800 1000 1200 1400 1600 1800 2000

Registered FDI and implement FDI (million

USD) and Number of FDI projects

Vietnam The registered FDI inflow in this period increased dramatically and peaked in 2008 with more than 70 million US dollars

The last stage is considered as a downward trend of FDI, which caused the global financial crisis

in 2008 The FDI inflow in 2009 had dropped sharply and reached 23.1 billion dollars 2009, which

is just about 32% of the total FDI inflow in 2008 The FDI inflow for this period has fallen and reached the bottom at 15.6 million US dollars in 2011 Recently, for 2012 and 2013, the amount

of FDI inflow just be recovered but not too remarkable

It is worth noting that there is co-movement between the amount of FDI registered and the number

of FDI projects as figure 2.1, especially in the period from 2007 to 2014 Meanwhile, the amount

of registered FDI declined, but the number of FDI projects tends to increase during this period In

2008, the average scale of FDI was about 61.2 million US dollars per project when compared with 19.12 million US dollars per project in 2009 and 11.89 in 2014 Despite the growth of FDI projects,

as in the report on 25 years in FDI attraction of MPI, almost FDI projects are small and medium scale

2.2 Distribution of foreign direct investment among provinces

The distribution of FDI inflow in Vietnam through years is displayed as figure 2.2 The map of FDI distribution is drawn from the data of registered FDI from 1988 to 2014 All 63 provinces have received the FDI capital, but it is effortless to recognize that there is an uneven distribution

of FDI across these provinces The more dark area implies the higher FDI inflow province and vice versa, the lighter area implies the lower FDI inflow province The FDI mainly concentrates

in the Red River Delta and the South-East Vietnam On the other hand, the North-West, Central highland and Mekong Delta exhibit a lower of FDI inflow Alternatively, the FDI inflow is clustered across provinces in Vietnam More specifically, provinces with high FDI inflow tends to

be closed in distance with each other and vice versa For example, in the South East Vietnam, Ho Chi Minh City with the highest FDI is surrounded by provinces with high FDI such as Dong Nai, Binh Duong and Ba Ria- Vung Tau In the Red River Delta, figure 2.2 also exhibits that Ha Noi with high FDI inflow is also surrounded by its neighbor provinces such as Thai Nguyen and Bac Ninh In contrast to the South East and Red River Delta, provinces in the Mekong Delta or provinces in North-West exhibit a low FDI with each other In addition, the FDI inflow is clustered

in provinces belong to the KEZ

Figure 2.2: The distribution of FDI in Vietnam from 1988 to 2014 1

Source: Author’s calculation

As seen in table 2.1, the FDI inflow tends to concentrate in provinces belong to the Southern KEZ2

and the Northern KEZ These provinces are characterized by their proximity geographical and the

similarity in the development of economy than the other 24 provinces in Key Economic Zone

received nearly three- quarters of total FDI inflow in Vietnam with 73.5% The Northern Key

Economic Zone takes account of 28.2% and Southern Economic Zone takes account of 35.55%

the total FDI, respectively Meanwhile, the Mekong Delta KEZ and the Central KEZ just capture

a small proportion of FDI inflow as 1.7% and 8.7%, respectively The rest 39 provinces of Vietnam

just take a share of 26.5% total FDI This could be caused by the advantage of being one of the

KEZ where provinces benefit from the effort of Vietnam government in developing the

1 The FDI distribution across Vietnam is drawn by Geoda Software

2 Vietnam is decided into 4 Key Economic Zone with 24 provinces include Northern, Central,

Southern and Mekong Delta

infrastructure capabilities, enhancing the level of education, training, science researching and modern technology transformation

Table 2.1 Sharing of FDI in Vietnam from 1988 to 2014

Source: Author Calculation

2.3 Country of origin

In terms country of origin, according to the Ministry of Planning and Investment report, until 2014,

101 nations and territories have made investment in Vietnam Table 2.2 presents the top ten investors in Vietnam with the cumulative amount of FDI inflow from 1988 to 2014, this group accounts for 81.7% total amount of FDI inflow Except for British Virgin Island and US, the rest

8 countries are all originated from Asia countries with more than 70% total FDI inflow South Korea is the biggest source with 4190 projects and account for 14.92% total FDI Japan and Singapore also contributed a high share of FDI in Vietnam with 14.77% and 13.03%, respectively

In South East Asia area, Singapore, Malaysia and Thailand also dominated approximately 20% total the FDI inflow Vietnam The US as the latecomer also contributed 4.34% of total FDI with

725 projects

Country Number of FDI

2.4 Sectors of foreign direct investment

In terms of sectors of investment, the FDI inflow is relatively diversified with various sectors As seen in figure 2.3, the foreign investors tend to be favored in the manufacturing sector and real estate when each of them accounts for 55.9% and 19.1%, respectively The construction sector and

the accommodation and food service take a small share of FDI capital with 4.51% and 4.42%, respectively All other sectors such as mining and quarrying, energy, retail, education, communication, et cetera are summed up and they take only 16.02% of total FDI Detail of sectors

of foreign investment is provided in appendices part

Figure 2.3 The sector distribution of FDI

Source: Author calculation

According to the report on 25 years in FDI attraction of Ministry of Planning and Investment, FDI has played a crucial role in Vietnam’s economy Vietnam has achieved numerous benefit in the aspect of socioeconomic such as economy growth, economy structure transformation, job creation, technological transfer, enhance the competitive ability for at all the national, enterprise and production level and contribute its effect to the progress in world economy integration Nguyen and Nguyen (2007) proposed seven reasons for explaining the successes for FDI attracting in Vietnam include: the strategy location of Vietnam, the stable political system and economy,

Manufacturing 56%

Real Estate 19%

THE SECTOR DISTRIBUTION OF FDI IN VIETNAM

abundant natural resources, young and quality labor force, large market size, a potential platform for exporting to EU and USA and its commitment to economic reform

However, besides the benefit caused by FDI, there still are many restraints such as high portion FDI enterprises are operated with the medium-low technological, low living standard of workers, pollution, energy resource wasted, et cetera

CHAPTER 3: LITERATURE REVIEW

The following part introduces some theories associated with the location choices of FDI with the paradigm OLI as the main theory In addition, the linkage between the agglomeration economies and FDI also be presented Besides, this part also explains the spatial interdependence of FDI inflow between regions by the agglomeration effect and MNE motivation Further, this chapter also provides some empirical studies on FDI determinants with and without spatial dimension

3.1 Theories about location choices of foreign direct investment

Before making investment decisions in given country/regions, MNE have to agitate several considerations such as why they should invest, how should they be organized or managed or where should they invest Parallel with the development of FDI around the world, several theories have been developed to explain its activities Some common theories are typically used as the Market Perfection, Market Imperfection theory, Internalization theory, Transaction Cost theory, Product Life Cycle theory, Macro Economic theory, Internationalization theory, the Eclectic Paradigm OLI And as a consequence, this also leads to problematic to search for only one theory, which can explain for all issues related to the FDI However, regarding to location choices of FDI, some theories could be listed as the Production Life Cycle theory (Veron, 1966), the Macroeconomic theory of FDI (Kyoshi Kojima, 1973) and the Paradigm OLI (Dunning, 1979, 1981) Among theories above, the eclectic OLI paradigm (Dunning, 1979, 1981) is the most typical theory that is employed to explain the FDI location choice

3.1.1 The eclectic paradigm OLI

The eclectic OLI paradigm (Dunning, 1981) combines three components include the Ownership, the L-Location and the I-Internalization The three components of the Paradigm OLI will be presented as below:

O-The O- Owner advantage

When foreign firms determine to invest abroad, they have to face with many disadvantages once compare to local firms By operating abroad, firms would have borne an oversea cost This means that foreign firms would have to possess specific advantages that make them more competitive These advantages could be the trademark, superior production techniques, management skills,

production skills, marketing system or the advantage of scale And these advantages would be strong enough to compensate the cost and gain a higher return

The I- Internalization

By owning its advantage, foreign firms may have several choices for production abroad include the collaboration with other firms in the host countries They might sell the license or joint venture with others local firms However, foreign firms would have an incentive to believe that, advantages

by production and operation internally might earn more benefit rather than the collaboration with others

The L- Location

In the scope of this study, assume that the component “I” and the “O” are given, the component

“L” in OLI should be the most appropriate theory for explaining the investing based on the location advantages The L-Location advantages include many factors such as:

 The abundant material resource

 The market size

 The cost factors

 The labor-capital

 The business environment

 The physical infrastructure

 The political system

 The social/cultural

Apparently, as mentioned above, the location choice of FDI firms could be explained by the Production Cycle Theory and the Macro Economic Theory However, the using of these theories shows some shortcoming when comparing with the L-Location in the paradigm OLI

Firstly, in the Life Cycle Theory of Veron (1986), the investment of firms only occurs at the last stage of production, and the incentive of moving out is caused by the cost factors only Second, in the Macro Economic theory of Kyoshi Kojima (1973), the motivation of investing abroad comes from the cost advantage and the abundant resource In comparison with the Production Cycle Theory and the Macro Economic theory, the “L” factor in OLI theory can explain the incentive of FDI inflow better and more sufficient

However, the component L-Location in the OLI paradigm framework is able to explain the FDI attraction based on the differences in location characteristic The locations with the more comparative advantage would receive more FDI Assume that if all the OLI advantages are similar

to two locations, but there still exists a disparity in their FDI inflow between them This is also the weakness of the L component of the OLI paradigm to explain such situation

One theory that may give the solution for this issue named “agglomeration theory” which could explain the FDI in the aspect of the clustering of economic activities The agglomeration theory and its linkage to the FDI attraction will be presented as below

3.1.2 Agglomeration and foreign direct investment

The very first idea about the concept of agglomeration was defined by Alfred Marshall in the 1890s with the “The principle of Economics” and until now, the agglomeration continues to be the object for researching of many fields such as urban development, productivity, labor and also the foreign investment Essentially, the agglomeration is described as the phenomenon that the locality concentration of particular characteristic For example, the concentration of high-tech companies

in the Silicon Valley in the US could be considered as the agglomeration of firms or the concentration of FDI in Vietnam as illustrated in figure 2.2 also be the signal of agglomeration So why agglomeration does matter with Foreign Direct Investment? The answer comes from the scale effects of the agglomeration which contribute the positive externalities on FDI According to Fujita and Thisse (2002), these externalities include the localized knowledge spillover, the abundant labors and the intermediate input for manufacturing and complementary services

Firstly, in the aspect of localized knowledge spillover, the proximity of geography permits employees easily contact with other employees in other firms or move between firms The mobility

of employees contributes to the spreading of knowledge, ideal as well as the innovation between firms in the area As Bronzini (2004), these spillovers would facilitate firms to enhance their productivity as a consequence

Second, once firms concentrate in specific areas that lead to the specialized pooling labor market

in that area One hand, foreign firms would tend to move to areas where possesses the abundant specialized labor that reduces the risk of labor lacking On the other hands, labor would tend to concentrate in agglomeration area with the incentive of job available and be able to choose firms with a higher salary

The last positive externality of agglomeration is the intermediate input for manufacturing and complementary services By locating in the concentration area, firms would benefit from the backward linkage and the forward linkage The difference between backward linkage and forward linkage is the approach of supply side and demand side In aspect to the backward linkage, firms are buyers and they would move to the areas where a closed range with input suppliers Contrast with the backward linkage, the forward linkage indicated that firms as suppliers and would move

to the areas where buyers locate their production The proximity help firm to benefit the specialized input and reduce the transaction cost as well Therefore, most of the empirical studies have pressed the importance of agglomeration economies which contributes the positive impact on the FDI

However, the agglomeration not only contributes the positive externalities but also negative externalities on FDI According to Borowiecki (2013), the agglomeration may cause the diseconomies of scale Besides that, K Head et al (1995) stated that the concentration of firms is not always caused by the agglomeration externalities, but the endowment factor driven Bronzini (2004) also stated that without concerning to the factor endowment, there would be an erroneous conclusion about the relationship between FDI and agglomeration

3.2 The inter-dependence of FDI between locations

Followed the Tolber’s Law (1970), everything is related and impact on each other through space Also, according to Baltagi et al (2007), the relatively small of any pair of countries when comparing with the rest of the world, then the third country effect does matter Particular to the FDI, when the MNE make investment in the host country, the effect of other countries to the host country in attracting FDI is clearly existed But, how regions mutually impact to the FDI inflow and which way?

According to Coughlin and Segev (2000), the spatial interdependent of FDI could be the results of agglomeration effect or the similarity of topography characteristic between regions Besides, the spatial interdependence of FDI could be the results of MNE strategies as Ekholm et al (2007) and Blonigel et al (2007) The following part will explain the spatial interdependence of FDI through the MNE motivation and the agglomeration effect

3.2.1 MNE choice theory

The following part presents the spatial interdependence of FDI, which relies on the MNE motivation theory The core content of MNE motivation illustrates the spatial interdependent of

FDI through 2 forms include: the FDI inflow host region is influenced by the FDI inflow of nearby regions and the FDI inflow host country is influenced by the potential market size of all other nearby regions

According to Blonigen et al (2007), motivation of MNE could be summarized as below:

Motives of FDI Sign of neighbor FDI inflow Sign of surrounding market

Table 3.1: MNE motivation

Source: Blonigen et al (2007)

Markusen (1984) had developed the horizontal FDI, which indicates the motive of foreign investors into the host country because of market access only Firstly, before making a decision on investment, MNE have to face the trade-off between the production in their country and then export goods to the host countries or they should settle their affiliate in the host country One hand, if they prefer to keep operating in their country and export goods, they would have to face with the trade cost, custom duties, legally issues of the host country On the other hand, if they choose to move out, they would have to face with the fixed cost by setting up another production affiliate By comparing the cost and the benefit, they would select the best strategies The purpose of moving out is serving the host demand only There is no impact of the third countries on the investment decision at all According to Blonigen et al (2007), the sufficient condition of horizontal motive

is that the host country would have a trade restriction high enough that make foreign investors only produce goods for the host country only, but not for exporting

Unlikely the horizontal motive, the vertical motive by Helpman (1984) indicated the incentive of investors for moving abroad is caused by the low input factor cost When MNE realize that they might produce goods to serve their customers with a lower price in another country rather than in

their country So MNE would establish their affiliates in the host country and then export goods back to their country Therefore, regarding cost, which country has the cost advantage for MNE then it would receive the FDI According to Blonigen (2007), this form of MNE would lead to the competition between countries in attracting the FDI inflow As a consequence, this leads to the negative sign of nearby countries on the FDI inflow of host country

Contrast with the horizontal motive, if trade restriction of the host country is low, then foreign investors may use this country as a platform to export goods to other countries as the export-platform motivation of Ekholm et al (2007) If any countries have a lower input cost as well as the lower fixed cost of setting affiliate then it would make it become more attractive and hence make a negative impact on the FDI inflow of others Besides, because the incentive of investors to host country for exporting goods, then the market size of the third country would take a positive impact on the FDI inflow of the host country However, this study uses the provincial data therefore the MNE would product to serve for nearby provinces Therefore, follow Hoang and Goujon (2014) and Sharma et al (2014), the export-platform is replaced by the regional trade platform

The last form of FDI is named the complex vertical which indicates that each phase production of MNE would operate in different countries as a chain production to benefit the cost differences among these countries As a consequence, the FDI inflow host province would also increase the FDI inflow of neighbor provinces Besides, the complement in FDI inflow between countries may

be caused by the agglomeration of immobile resources as endowment factors driven Foreign investors would parallel exploit the abundant resource in nearby countries, but still keep their production in the host countries by advantage comparative So, this form of FDI would create a positive spatial effect on neighboring countries However, the effect of market potential is still ambiguous

In sum, the MNE motivation suggests that, the cost factors caused the spatial dependence of FDI between regions According to Coughlin and Segev (2000), if one region has an increasing in cost then, it would make the cost structure of nearby regions become more attractive and hence, it leads

to the spatial dependence of FDI as well

3.2.2 Agglomeration effect

The spatial dependence of FDI between regions is also explained by the agglomeration effect By the concentration of FDI activities, then it would generate the positive externalities within regions But what if these externalities are strong and spill-over the space?

Firstly, one needs to be distinguished is the concept of agglomeration and spatial dependence According to Nwaogu (2012), the spatial dependence and the agglomeration are different but interrelated with each other What makes agglomeration differ from spatial dependence is the

“distance decay” The agglomeration effect would lose its power as the distance increase, meanwhile, the spatial dependence could capture the effect of agglomeration through distance According to Couglin and Segev (2000), Orr (2008), Nwaogu (2012), the dependence of FDI between regions is the result of its agglomeration externalities whenever these externalities are strong enough to cross the border between regions

One hand, the agglomeration with the spillover of technical, specialized labor or intermediate input would bring a positive effect on neighbor FDI On the contrary, as Coughlin and Segev (2000), if these externalities limit themselves within regions as “distance decay” then it would decrease the FDI inflow of neighboring regions

3.3 Empirical studies

In this part, the reviewing of empirical studies will include 2 main sections: the spatial analysis

of FDI determinants on the world and the reviewing of FDI determinants analysis in Vietnam

3.3.1 Empirical studies of FDI determinants in spatial analysis

The traditional approach of studies on FDI determinants relies on the bilateral gravity model, which is grounded by the horizontal motive of Markusen (1980) and also the pure vertical motive

of Helpman (1984) Followed by that, the FDI inflows to the host country is caused by its determinants and the home country determinants However, as presented, this approach is unable

to explain the effect from the third country such as MNE motivation and agglomeration As a consequence, this requests researchers to find a more suitable approach As Blonigen et al (2007) stated that the spatial analysis is a powerful mechanism once concern about the spatial interactions within regions in country or nations in areas To explore for the interdependence of FDI, empirical works relied on 2 main forms include the spatial dependent lag and the spatial error The spatial dependent lag which indicates that the FDI of one region depends on the FDI of the nearby regions, this form is named Spatial Autoregressive Model The latter means the dependence of FDI on unobserved factors of the nearby regions This form is named the Spatial Error Model

Until the present, spatial works on FDI determinants is developed through 2 directions The first path uses the spatial econometric to investigate the spatial interdependence of FDI between regions purely This trend could be found in studies of Coughlin and Segev (2000), Kayam et al (2013)

or Blanc-Brude et al (2014) Coughlin and Segev (2000) are the first one who employed the spatial econometric to analyze the FDI determinants They have based on the agglomeration effect to explain for the spatial interdependence of FDI between provinces in China By applying the Spatial Error Model with binary contiguity weight matrix, Coughlin and Segev (2000) discovered that, the FDI inflows to the host provinces which take a positive effect to surrounding provinces in China Also, Coughlin and Segev (2000) found out that, except for the infrastructure, other determinants include market size, labor cost, labor productivity, education and the dummy variable for coastal provinces have an effect on the FDI Blanc-Brude et al (2014) also examined the FDI determinants in China by both SAR and SEM, separately The same as Coughlin and Segev (2000), they have revealed that FDI inflow one region in China complement to the FDI inflow of other regions for both models Besides, they found that the FDI inflow China is caused by the market

size, degree of openness, labor wage, the quality of human capital, the government spending and also the distance to the coastal regions Especially, they found that the foreign agglomeration has

a strong and positive effect on FDI Kayam et al (2013) have employed both SAR and SEM in examining the FDI determinants in Russia They found a strong effect of market size, unemployment rate, education, infrastructure and the material resource on the FDI inflow However, no spatial interdependence of FDI inflow Russia is detected in their study

The second approach uses FDI spatial analysis to investigate the MNE behaviors This approach could be considered as an expanding of the first direction What makes the second differ from the first approach is the market potential variables By adding this variable, researchers would make the conclusion regarding the motive of FDI by combining the sign of spatial lag coefficient or spatial error coefficient and the sign of the market potential variable as figure 3.1 Blonigen et al (2007) have tested the FDI spatial interaction through the cross data of US FDI outbound to 35 countries in Europe from 1983 to 1998 By applying the SAR model, they have found the positive spatial interdependence of FDI in these countries Besides that, they also detected the significant effect of traditional FDI’s determinants such as the population, trade cost, labor skill, investment cost and distance from the US to the host country Garretsen and Peeters (2009) have used both models SAR and the SEM to examine the FDI from Dutch to 18 host countries from 1984 to 2004 Their result reflects a spatial interaction in the FDI of host countries from Dutch with different signs over the sample In addition, they also confirm a substantial impact of the market size as GDP, degree of openness, the tax rate on the FDI Ledyavea (2009) also employed the SAR model

to examine the spatial distribution of FDI in Russia By separating the sample into 3 distinct periods, pre-crisis 1996-1998, post-crisis 1999-2002 and the period with high shock FDI inflow 2003-2005 The author has found a negative effect and significant of FDI inflow between regions

in Russia in 2 periods 1999-2002 and 2003-2005, but not significant in the period 1996-1998 In addition, the market potential variables are all positive and significant for 3 periods Escobar Gamboa (2013) also used the same approach as Blonigen et al (2007) to examine the US FDI inflow of Mexico However, before employing the SAR model, he has used the Moran’s I and Geary’s C coefficient to prove that the FDI in Mexico is not randomly distributed but clustered instead Therefore, he concluded that the using of spatial econometric is necessary He has found

a complement in FDI inflow between states with the positive sign spatial lag coefficient Moreover, his result has shown the effect infrastructure, education and the delinquency rate impact on the

FDI However, the significant of market size proxy by GDP and the agglomeration proxy by the industrial unit are wiped out when controlling for state fixed effect

The similarity among empirical studies above is that, they have used the SAR and SEM separately

to investigate the dependence of FDI between regions A mixed approach is applied by combining both of them into one model, which is known as the SAC model Blanc-Brude et al (2014) used the SAC in examining the spatial effect of FDI in China They found a complex spatial interaction

of FDI between provinces in China, the positive sign of spatial lag coefficient, but the negative coefficient of spatial error coefficient

Recently, a brand new dimension of spatial analysis has been developed which encompasses the spatial dependence of explanatory variables According to Kayam et al (2013), the FDI inflow the host country does not only depend upon its determinants and neighbor FDI but also the neighbor determinants They claimed that the impact of the neighbor determinants on the host FDI inflow would take an opposite sign when compared with the sign from host determinants to host FDI inflow In terms of spatial econometric, the effect of the neighbor on the host regions is called

“indirect effect” In addition, by combining the SAR model or the SEM model with the spatial dependence of explanatory variables will create new models named Spatial Durbin Error Model (SDEM) or Spatial Durbin Model (SDM) Nwaogu (2012) appear to be the first one that applied that SDM in order to investigate the relationship between FDI inflow countries in Africa, Latin America and Caribbean (LAC) and explanatory variables of nearby regions He found that the FDI inflow of these countries is influenced by several determinants of nearby countries include the population, market size, infrastructure, trade cost Kayam et al (2013) have used both models SDEM and SDM for investigating the spatial dependence of FDI in Russia Their result has shown that the FDI inflow each state in Russia also depends on the transportation cost, infrastructure and the natural resource of nearby states However, from the result of Nwaogu (2012) and Kayam et

al (2013), one conclusion could be obtained is that, the sign of impact from neighbor determinants

to host FDI inflow is ambiguous

3.3.2 Empirical studies of FDI determinants in Vietnam

It is undeniable that the FDI inflow takes a crucial role in the Vietnam’s economy However, the actual fact that, caused by the incomplete and constraint of data, so there are not many researches

on the provincial FDI determinants in Vietnam Some empirical studies could be listed as Pham

(2002), Meyer and Nguyen (2005) and Anwar and Nguyen (2010) Moreover, almost the studies only focus exam the FDI inflow on the aspect of provinces characteristic and totally ignore the presence of spatial interaction of FDI between provinces This part introduces some outstanding empirical studies about the Vietnam FDI determinants

Pham (2002), by using the data of 53 provinces to analyze the regional FDI determinants for the period from 1988 to 1998 with OLS model The 2 types of FDI inflow include the commitment FDI for 1988 to 1998 and the implement FDI from 1991 to 1998 is used as the dependent variables His result indicated that the committed FDI depended on the market size, education and infrastructure, but not for the institution with proxy by the tax ratio In the case of the implement FDI, all of the explanatory variables are significant Meyer and Nguyen (2005) by employing the binomial regression model re-confirm traditional determinants of FDI They used both the type of data FDI includes the cumulative FDI as well the new FDI Their result has shown that the FDI is affected by the GDP growth, the population, education and the transportation However, there is

no impact of labor cost on FDI As an attempt in searching the linking between economic growth and FDI, Anwar and Nguyen (2010) have used the GMM to examine the FDI determinant in Vietnam from 1996 to 2005 at the national level They found that the FDI inflow Vietnam is a function of the economic growth, market size, domestic investment, labor quality, labor cost, infrastructure and the exchange rate

So far, there are only two studies of Hoang and Goujon (2014) and Esiyok and Ugur (2015) have concerned the spatial dependence of FDI in Vietnam with 2 different approaches By adapting the SEM model with cross data for 2 periods 2001-2006 and 2007-2010, Hoang and Goujon (2014) have found out a strong spatial dependence of FDI with 2 basic spatial weight matrix include the binary contiguity weight matrix and inverse distance weight matrix The FDI inflow to one province has a negative effect on neighboring provinces for both periods Their success in linking the impact of many variables includes the human capital, dummy variable key economic area, labor cost, market size, productivity to the FDI in Vietnam In addition, they have added the agglomeration variable include of the domestic agglomeration and foreign agglomeration into the SEM and reveal a very strong impact of them on the FDI

Unlike Hoang and Goujon (2014), Esiyok and Ugur (2015) have used the SAR to exam the FDI spatial interdependence instead of SEM By examining the spatial distribution of FDI in Vietnam

for the period from 2006 to 2009 with various types of spatial weight matrix includes the cut-off

km weight matrix (186km and 350km), k nearest neighbor weight matrix (k=1 and k=3) They have found the positive effect of FDI inflow between provinces Besides that, they also used traditional determinants as market size, labor cost, labor quality, domestic investment, the degree

of openness, institution and confirm their significant impact on the FDI

In summarizing, until the present, almost empirical studies of both non-spatial and spatial analysis has been discussed above are quite successful in finding the FDI determinants as well as its spatial interdependence However, their studies still have some drawback as discussed below:

Firstly, the lacking of spatial factors is a weakness that leads to omission and biased estimation of non-spatial method However, if, the suspecting of spatial factors does matter, then how to identify it? Almost studies on spatial FDI determinants skipped this step and utilized spatial econometric model directly and sometimes, they failed in finding any form spatial interdependence of FDI such

as Kayam et al (2013) Fortunately, some tests have been developed to detect the autocorrelation in the data pattern such as Moran’s I or the LM test by Anselin (1988)

spatial-Second, with the variety of spatial models that lead researchers to use them arbitrarily As presented, many models have been used to investigate the spatial interdependence of FDI such as SAR, SEM, SAC, SDM or SDEM So, in the aspect of spatial econometrics, how to identify the

“right” model should be used? The right model should depend on the characteristic of data If the spatial correlation in data pattern does exist, then is it spatial lag dependence or spatial error dependence or both? Therefore, a benchmark approach need to be utilized to avoid mismatching spatial model or omitting spatial interaction

Third, almost empirical studies typically use 2 basic spatial weight matrix include the binary contiguity weight matrix and inverse weight matrix or some kinds of distance weight matrix So what is the best spatial weight matrix that could describe the best data pattern? This suggests the necessity of tools that could point out the best spatial weight matrix Blanc-Brude et al (2014) is

a good example in identifying the best matrix among many types of the matrix by using the AIC value

Particular to Vietnam, indeed, the result of Hoang and Goujon (2014) and Esiyok and Ugur (2015) pointed out the necessity of using spatial analysis in investigating the FDI determinants However, their studies also have some disadvantages By using the SEM does not allow to give the indirect

effect of explanatory variables and just ability to interpret the variable coefficient as the direct effect The using SAR could give a specific interpretation through direct effect, indirect effect and total effect However, because of the properties of SAR that make the inference of indirect effect inaccurately

Through the SDM, this study also aims at testing the spatial dependence of FDI in Vietnam which could give more advantage information through the marginal effect include the direct, indirect and total effect To the author’s best knowledge, this study expected to be the first one to analyze the FDI determinants of Vietnam with the Spatial Durbin Model

3.3.3 Fundamental FDI determinants

This part presents the construct variables are employed in this study There variables are expected

to take effect on the FDI inflow of province in Vietnam These variables include: Market size, Infrastructure, Agglomeration, Labor Cost, Labor Quality, Degree of Openness and the Institution

 Market size

Under the “L-Location” component in OLI paradigm, the market size is a vital factor that attracts FDI in aspect to the demand side with positive effect, the larger market size the higher FDI Chakparati (2001) stated that the market size has become the most broad acquired for the FDI determinants in almost studies The importance of market size on FDI is clearly undeniable Foreign firms would be more favorable to invest in regions with greater market size where they can potentially earn a higher profit Moreover, when MNE invest in large market regions, they also benefit a lower cost caused by the economy of scale In addition, as Castellani (2016) also claimed that the foreign investors, one hand tends to be in favorable of the greater market of the host region, on the other hand, they would also extend to the nearby region for exporting as the market potential The literature on FDI determinants used many indicators to proxy for the market size Previous studies such as Blonigel et al (2007), Cassidy (2006) or Poelhekke and Ploeg (2009), Hoang and Goujon (2014) have used the GDP to proxy the market size Some studies used different indicators such as the GDP per capita as Anwar and Nguyen (2010), GRP as Kayam et

al (2013), Sun et al (2002) used vary indicators such GDP, GDP per capita, retail sale and retail sale per capita Dermihan and Masca (2008) stated that the using of GDP to proxy for market size could be inefficient if the host’s market size is too small and foreign investors would hence expect the growth instead They used the growth of GDP per capita instead Le (2007) had used a novelty

indicator named the retail sale to proxy for the market size However, her result shown that retail sale is a poor proxy for market size with the insignificant coefficient

 Infrastructure

Infrastructure is another indicator of the FDI determinants in the OLI framework The infrastructure illustrates the ability of a country/region with the physical aspect Regions with good quality and high ability of infrastructure would facilitate the economic activities which helps minimize the transportation cost and maximize the profit In addition, Dermihan and Masca (2008) claimed that the development of infrastructure in the host country may boost the productivity of MNE Hence, foreign investors would be more favorable to invest in locations with the better infrastructure The infrastructure may contain many aspects as the road capability, the presence of port, the telecommunication system as well as the railway

Coughlin and Segev (2000), by using the length of the paved road as a proxy for the infrastructure and pointed out the positive relationship between infrastructure and FDI Ladyeva (2009) employed the number of seaport as the proxy for the infrastructure Similar, Hoang and Goujon (2014) also used the dummy variable of provinces which have the marine port for the infrastructure Besides that, Hoang and Goujon (2014) used 2 other variables are the number of fixed and postpaid phone subscribers per 1000 inhabitants and the percentage of paved road Kayam et al (2013) also used the same indicators as Hoang and Goujon (2014) to proxy for the infrastructure Others studies about the FDI determinants in Vietnam also used the freight volume traffic or the passenger volume traffic (Meyer and Nguyen, 2005) but insignificant and not suitable

 Agglomeration

As mentioned above, even the crucial role of agglomeration in generating the positive externalities, but almost previous studies overlook it as a determinant of FDI Recently, economists started to link the impact of agglomeration on FDI Many types of agglomeration have been employed in previous studies Guiramaes et al (2000) have found a strong positive effect of 3 out 4 types of agglomeration include the total manufacturing, industry-specific and service agglomeration on foreign firms in Portugal Basile et al (2008) have used the number of local firms and the number

of foreign firms in the same industry to exam the effect of agglomeration on foreign affiliates in 8 Euro countries Their result has shown that the both types of agglomeration do impact on these affiliates with a positive sign Besides, Blanc-Brude et al (2014) also reveal the impaction of

foreign agglomeration on FDI in China Regarding Vietnam, Le (2007) and Hoang and Goujon (2014) found out that the foreign agglomeration and local economy agglomeration take a positive impact on the FDI inflow Vietnam

 Labor Cost

It is commonly believed that the labor cost as a determinant and take a negative effect on FDI and this has been highlighted by many previous studies such as Couglin and Segev (2002), Anwar and Nguyen (2005), Meyer and Nguyen (2005) Foreign investors are driven to locate their affiliate in provinces with low labor cost as a strategy to reduce the cost and gain more profit, special with the resource seeking FDI This is a specific characteristic of FDI in developing countries Provinces with low labor cost would also be an advantage comparative to attract more FDI when compared with the other provinces Couglin and Segev (2002) found a negative relationship between the labor cost and FDI in China Broadman and Sun (1997) also give the same conclusion about the relationship between labor cost and FDI However, Castellani (2016) also noted that the high labor cost also is a signal to a higher level of labor quality or high productivity and therefore the high labor cost may positive impact on the FDI, especially in developed countries

Pham (2002) has considered the labor cost into the study of FDI in Vietnam and be proxy as the income per capita However, this approach seems to go wrong since the income per capita would

be a better proxy for the market size Meyer and Nguyen (2005) also take the labor cost by employing the monthly average income of state employees However, Hoang and Goujon (2014) argue that this approach also mismatches the required of labor cost data when it only captures the effect of labor cost in state sector that is not relevant to FDI firms They have used the average annual income per worker in firm sector instead

 Degree of Openness

As presented in the component “L” of OLI framework, the openness of the economy also is a factor that takes effect on the FDI The openness of provinces could have an effect on location choice of MNE through 2 directions First, if MNE move abroad with the horizontal motive to exploit the market size of the host country Then, according to Sun et al (2002), the openness also takes a negative effect on FDI since the higher degree of openness, the more competition that reduces the FDI inflow Then, the degree of openness should take an adverse effect on FDI Seconds, under the vertical motive, when MNE move out for production and then export the goods

out of the host country Foreign investors would tend to locate their affiliate where has the high openness which could be a suggestion for the support of currently linkage of trade as well as the less of trade restriction Therefore, the degree of openness should be a relevant factor to the FDI inflow However, the effect of degree of openness on FDI is ambiguous

The degree of openness could be proxy by many indicators Anwar and Nguyen (2005) used the export per capita as a proxy for the openness, Sun et al (2002), Blanc-Brude et al (2014), Blonigel and Piger (2014) also input the openness through the ratio of trade to GDP into their model and show a strong positive effect on the FDI

 Labor Quality

An economy with the high level and abundant labor force would attract more FDI MNE may raise their productivity by using the quality workers On the other hand, in term of financial benefit, by using the high level of labor also lead to increase the cost of operation However, by being an emerging country with the advantages of abundant and cheap labor, the labor quality be used in this study would be expected a positive effect on FDI

Previous studies have used many indicators to proxy for the labor quality Pham (2002) has used the number of middle and secondary student per capita, the same approach such as Cheng and Kwan (2000) Meyer and Nguyen (2002) used another indicator as the number of university lecture per thousand people All of them have found a positive impact of labor quality on the FDI inflow

In addition, Hoang and Goujon (2014) have used the percentage of skilled labor over the active population and they stated that the skilled labor could move across regions easily and investors could take account of skilled labor in neighboring provinces

 Institution

The importance of the institution to the FDI attraction is widely recognized by almost empirical studies According to Blonigen (2005), in particular to less developed countries, the institution is considered the key factors impact to the decision of foreign investors First, the institution regarding the asset protections, a poor institution of host countries would lead to a high probability

of asset expropriation In the aspect of the market, a poor institution which would cause the corruption that makes a more extra cost of doing business And finally, a poor institution could be

a signal to the weakness in public goods as the infrastructure Hence, the poor institution would lead to the diminution in FDI inflow

In FDI literature, the institution variable might be proxy by many indicators Hoang and Goujon (2014) used the dummy of provinces belong to the KEZ and revealed the significantly positive of this variable on the FDI inflow Esiyok and Ugur (2015) has found the positive correlation between the PCI and the FDI inflow

An analytical framework would be presented as below to demonstrate for the FDI determinants as well as its spatial interaction in the scope of this study