Tài liệu Researching the determinants of the attraction of fdi flows into vietnam

--1-- INTRODUCTION 1. Research proposal The flow of foreign direct investment (FDI) into Vietnam begun in 1988. This turning-point was considered an achievement of the economic transformation from central planning to socialism market-based (Kokko et al., 2003). After economic reform was implemented, Vietnam’s annual FDI increased dramatically, reaching $ 22.352,2 mil with annual growth rate higher than 30% (General statistics directorate, 2014). The height of this flow was in 2008 with a total of $ 71.000 mil, but there was a steep decline from 2009 to the present (from registered $23.107,3 mil in 2009 to $16.348 mil in 2012, equivalent of 29.3%). Though the decrease was not against the global trend under the impact of economic crisis, when compared with ASEAN and China, the flow showed an increment during the past four years. This begs for questions: What determines the attraction of Vietnam’s FDI? What are the factors effecting the FDI distribution across Vietnam’s regions? Given the importance of FDI to economic growth, the answers for these are necessary for bringing in appropriate suggestions to further improve the encouragement of upcoming FDI into Vietnam in general and into the regions in specific. The reasons stated above urge this author to carry out a thesis for PhD qualification with the topic “Researching the determinants of the attraction of FDI flows into Vietnam”. 2. Research related literature: The author’s summarized survey indicates there exists a range of both theoretical and empirical researches done on the proposed topic, or similar, worldwide. However, the number of empirical works on the determinants of attraction and distribution of Vietnam’s FDI remains low. As far as these determinants are concern, the majority of surveyed authors utilize different variables and methods in accordance with their statistical database. Consequently the overall results were divided into two groups, based on primary and secondary data. Representing the first are Hafiz and Giroud --2-- (2004), Lei et al. (2011) and Nguyen et al. (2013). On the other hand, the papers by Parker et al. (2005), Hoàng (2006), Ho (2010), Pham (2011), Nguyen (2011), Hoang et al. (2013) used secondary data. With respect to the factors impacting FDI distribution in Vietnam, due to the limitation of data, the author’s survey indicates only three typical relevant works of Meyer and Nguyen (2005), Nguyen et al. (2008) and Dinh (2009). 3. Research objectives: The main objective of the thesis is to identify the determinants of FDI flows and regional distribution in Vietnam. The thesis focuses primarily on answering the two questions: -What determines the attraction of FDI into Vietnam? -What are the factors that influences FDI distribution across Vietnam’s regions? 4. Research subject and scope: 4.1. Subject The main subjects of the thesis are the determinants of FDI flows and regional distribution in Vietnam. 4.2. Scope (1) Data on Vietnam’s FDI attraction from 1988 to 20131. (2) Data on the feedback of 171 FDI firms on the contemporary investment environment (2012 – 2013). (3) To identify FDI attraction determinants, the thesis uses data of 24 Asian developing countries during 2000 – 2012. (4) Lastly, to indentify factors impacting FDI distribution, data on 63 of Vietnam’s provinces/cities during 2005 – 2013 were gathered. 5. Research methodology: - In order to reflect foreign investors’ opinions on Vietnam’s investment environment, qualitative methods were used. 1 2013 data are preliminary. --3-- - For the analyses of FDI attraction’s determinants, to use an estimating method with unbiased, consistent and effective estimators, I chose the differenced GMM specification rather than the OLS. - Similar to the case of Vietnam as a whole, the method to estimate and testing hypotheses about factors impacting regional FDI distribution is also chosen to be differenced GMM. 6. The innovation and contribution of the thesis:  Theoretical: - With respect to Vietnam, there is no effect of first lagged FDI (FDI_1) on the dependent variable, FDI. - With respect to FDI regional distribution, there exists effect of past information of independent variables on FDI. - The use of interaction terms to analyse the difference in slope coefficients of determinants influencing FDI in comparison with countries in ASIA 24.  Empirical: - Provide empirical evidence on the evaluation of foreign investors on Vietnam’s contemporary investment environment. - Identify the distinct model for FDI attraction determinants and factors impacting FDI distribution. Results confirm that the decisions of foreign investors are affected not only by current information but also past values of the determinants. - Based on the testing results from differenced GMM, three out of 6 hypotheses related to the determinants are significant at 10%. Accordingly, foreign investors choose Vietnam for the policy, market seeking purpose and resource acquiring motivation. - About the factors impacting FDI distribution, testing results by differenced GMM with current and past data are unable to provide sufficient evidence to reject 4 out of 5 given hypotheses. These involve the --4-- quality of regional economic management, market seeking, efficiency seeking and FDI’s cumulative effect. - Based on the results, some policy suggestions aiming for improving investment environment as well as attracting FDI at the country and region levels are proposed. - Lastly, thanks to the analyses based on panel data of ASIA 24, the results not only show the motivation of Vietnam’s FDI firms but also relate to similarities in other countries. As such, the results are hoped to provide a “broader” perspective to policy makers. 7. Research structure: Not counting the introduction, conclusion, and policy implications, the thesis is composed of five chapters. CHAPTER 1: FACTORS DECIDE LOCATION OF FOREIGN DIRECT INVESTMENT: THEORY AND EMPIRICAL RESEARCH 1.1. Introduction 1.2. Generalization about Foreign Direct Investment (FDI) Theory has shown that FDI usually is created from the interaction of forces between the host countries and the attractive ones (e.g. Dunning, 1981, 1988, UNCTAD, 2006). Capital flows of FDI will run from countries to countries and FDI could happen thanks to the influences of push factors from host nations and pull factors from attractive ones. Additionally, some investor tends to create motivation push FDI for seeking potential markets or increasing the effectiveness with lower manufacturing cost in attractive ones. --5-- 1.3. Influences of FDI to economic growth FDI could motivate economic growth into two ways which are direct and indirect. Indirect impact is considered as another term which is pervasive impact. The studies which analyze FDI’s impacts to economic growth used different methods and approaches. Some of research mentions to the measurement of FDI’s impacts to economic growth in general according to direct influences, while the others concentrate find out FDI’s impacts to manufacture, international trade, local investment in pervasivedriven. 1.4. Theory about crucial factors of FDI’s position 1.4.1. In national level 1.4.1.1. International Trade Theory The model of first theory explaining foreign investment bases on international trade theory is Heckscher-Ohlin model which was established by Heckscher (1919) and Bertil Ohlin (1933). It concerns that country has better good input factors should export products which is related to those agents and import ones which related to other inputs. 1.4.1.2. The Theory of Firm-Specific Ownership Advantages This theory was taken the initiative by Hymer (1960). This is the first effort to give independent theory to solve foreign investment trends. Hymer gave his view departing from industry economy and maintained that an enterprise which wants to overcome international barriers and participate in production process should have firm-specific ownership advantages. 1.4.1.3. Product Life Cycle Theories This theory was presented firstly in 1965 and developed systematically then by Vernon from 1966. It explains for foreign investment and international trade and considers foreign investment is a stage of product life cycle. --6-- 1.4.1.4. Internalization Theory Internalization Theory was published by Buckley and Casson in 1976, and it bases on Company Theory of Coase (1937). According to this theory, internal transaction (IT) is better than market transaction (MT) when the market isn’t perfect. 1.4.1.5. Eclectic Paradigm Theory (OLI) This model is established meticulously of Dunning (1977, 1979, 1981, 1988, 1996, 1998, 2000, 2001). According to Dunning, a company decides to invest abroad when it has OLI advantages including Ownership Advantages, Location Advantage and Internalization Incentives. In the general base, OLI Theory about determination of FDI’s position will be selected as foundation to build theoretical research model about influent factors of FDI’s attraction and factors affecting FDI’s distribution between regions in Vietnam in Chapter 3 and Chapter 4 of the dissertation. 1.4.2. Distribution of FDI between regions in the nation The first basic theory relating to the distribution of FDI between regions in the nations is Popular Accumulative Impact of Krugman (1991). The second one relates to traditional economic advantages. The final theory is institutional factor. 1.5. Empirical research about attractive factor of FDI The general empirical research reveals that attractive factors of FDI to group of countries, regions or a country aren’t different much. In general, they are divided into three groups: showing government policy (including: national risk/ inflation, exchange rate, cost of capital in the nation, tax policy, economic reformation, bilateral relationship, financial development, ODA support, etc.); deciding economy (including: market’s size, economic growth, opening trade, explorative resources, cost of labour, employee’s capacity/human resource, technological level, facilities, geographic distance, cultural distance,etc.) and institutional factors (including: stable politics, quality of laws, corruption/bureaucratic, etc.). Three groups of --7-- factors show the concepts representing for the base building the empirical research model about influent factors of FDI and FDI’s distribution between regions in Vietnam in chapter 3 and 4 of the dissertation. Additionally, general results of empirical research about influent factors of FDI in Vietnam in national and local levels allow defining research gap in Vietnam. 1.6. Research gap in Vietnam Base on the general of empirical research about influent factor of FDI and FDI’s distribution between regions in Vietnam, it reveals that most of studies haven’t conducted testing violations relating to the facts of changing variance, autocorrelation and endogenous variables. They only limited at initial results of the regression method according to Pooled Ordinary Least Square, therefore the results haven’t been confident. Especially, there hasn’t been any research model using delay variables FDI to study impacts to FDI flows. While, the research results of Campos and Kinoshita (2003); Carstensen and Toubal (2004); Bellak et al. (2008); and Anyanwu (2012) proved that it’s a crucial variable and it should be included in model studying attractive factors of FDI. For the influent factors of FDI’s distribution between regions in Vietnam, there are three relating research including Meyer and Nguyen (2005); Nguyen and Nguyen (2007) and Dinh (2009). If these authors used data from many different sources and inhomogeneous time, the dissertation uses secondary data from the only source which is from General Statistic Office in 2005-2013. Thus, the dissertation ensures the consistency and reliability and updated date (especially Provincial Competitiveness Index – PCI). That help regression results will have high reliability. In addition, general results show that there hasn’t been research paying attention on impacts of past information to FDI’s distribution between regions in Vietnam. The testing results in chapter 4 prove that it is the essential character in making decision of foreign investors in Vietnam. --8-- 1.7. Conclusion of Chapter 1 CHAPTER 2 THE CURRENT SITUATION ON FDI ATTRACTION IN VIETNAM 2.1. Introduction 2.2. The current situation on FDI attraction in Vietnam 2.2.1. Current trend of Vietnam’s FDI According to data from General statistics department (2014), Vietnam’s FDI trend (both registered and implemented) increased from 1988 to 2013. However the total implemented capital compared to registered capital is very low (until end of 2013 the ratio is estimated to be 51.54%). 2.2.2. Vietnam’s FDI composition With respect to industrial composition, the manufacture and processing sectors are estimated to be dominant both in the number of projects and capital attracted, having more than 54.76% the total of registered FDI in Vietnam (8.725/15.932 projects). With respect to investment partners, cumulative speaking, until 31/12/2013, Japan, Singapore, Korean, Taiwan and Virgin Islands (Britain) are five countries that have the largest registered FDI capital in Vietnam. With respect to regional composition, statistics show that among 63 provinces/cities, Ho Chi Minh city is attracting the most FDI ($32,403.2 mil), followed by Ba Ria – Vung Tau ($26,298 mil) and taken the third place is Hanoi ($21,205.6 mil). 2.3. Comparing FDI attracted by Vietnam to that by other countries in the region In the area of ASEAN, until end of 2013 the ranking of countries based on FDI attraction is as follows: the first place is Singapore ($63,722.3 --9-- mil), second is Indonesia ($18,444 mil), followed by Thailand ($12,649.7 mil), Malaysia ($11,582.7 mil) and Vietnam at the fifth place (with $8,900 mil). 2.4. Survey results on Vietnam’s investment environment: The results from surveying 171 FDI firms in Vietnam shows 66.1% of the firms agree that they chose Vietnam in comparison with other countries; 90.1% agree that Vietnam is a destination of ASEAN’s FDI; 62% of the firms investing in Vietnam because of market seeking purpose; 86% suggested that Vietnam investment environment has changed positively ever since joining the WTO (this also shows the satisfactory with current environment in Vietnam); 83.6% suggested that they will continue investing in Vietnam and the next choices would be China and Cambodia. 2.5. Conclusions CHAPTER 3 ANALYSING THE FACTORS AFFECTING FDI ATTRACTION IN VIETNAM 3.1. Introduction 3.2. Hypotheses and model Based on the 6 hypotheses given, theoretical research model of determinants of FDI attraction in Vietnam is established as follows: --10-- Figure 3.1: theoretical research model of determinants of FDI attraction in Vietnam Based on the summary of empirical researches in Chapter 1 and Vietnam’s unique economic – social characteristics, the empirical research model on determinants of FDI attraction in Vietnam is established as follows: ∑ ∑ ( ∑ ( ∑ ) ) And ∑ ∑ ∑ ( ∑ ( ) Where: + i = 1,…, 24: represents 24 Asian nations; + t = 2000,…, 2012 : is an indicator of time; + FDI: Foreign direct investment attracted (USD); ) --11-- + Xi: a set of variables related to policy frameworks, including inflation rate (Infla), exchange rate (ExchRa), domestic credit in private sector (FinDev), official development assistance (ODA). + Yi: a set of variables related to economy, including local gross domestic product (GDP), urban population (UrPop), direct exporting metal (Resour), number of working labour (Labour), number of students in vocational schools (HuCa) and number of landlines (Infras); + Zi: a set of variables related to institutional quality, including corruption control index (Corrupt), regulation quality (Regul) and Law evaluation indicator (Law). + VNt: a set of interactive terms between Vietnam and countries in ASIA 24. These terms are the product of each independent variables and the dummy variable (which receives 1 for Vietnam and 0 otherwise). + FDIit-1 or FDI_1: FDI’s first lag. + FDIVNt-1: the first lag of the interaction between FDI and the dummy variable. 3.3. Data and research methodology 3.3.1. Variables 3.3.2. Data The data on the dependent variable, FDI, are gathered from summary statistics of UNCTAD during 2000 – 2012. With respect to independent variables related to policy frameworks, economics and institutional quality, the author collected from the source of World Bank (during 2000 – 2012). Of those, Corrupt, Regul and Law are ranked from 0 to 100 (with 0 as the lowest rank and 100 as the highest rank). 3.3.3. Research method In the empirical models established, in addition to other independent variables, the first lag of FDI is considered an independent variable, and the data used are panel data from 24 of Asia’s developing countries. Therefore --12-- this is a dynamic panel data model. Thus, to estimate and test proposed hypotheses related to determinants of Vietnam’s FDI attraction, differenced GMM estimator (with fixed effects) is utilized. 3.4. Research results 3.4.1. Variables’ summary statistics On average, during the research period, attracted FDI in ASIA 24 is $8.820 mil, among which the highest value is $280,000 mil and the lowest is -$4.55 mil2. For the independent variables, the ones related to policy frameworks, averaged inflation rate is 7.36%, averaged exchange rate is 1,696.38 LCU/USD, averaged domestic credit rate for private sector compared with GDP is 40.7% and the rate of ODA to GNI is 3.26%. For the economic factors, averaged GDP is $294,000 mil annually with a growth rate of 6.14%, averaged rate of urban population to total population is 46.2% and the regional rate of foreign trade is 88.28% of GDP. In addition, the averaged rate of exporting ore and metal value to total exporting value is more than 7.17% GDP. The averaged number of mobile phone registries on 100 people is more than 45 and the averaged number of vocational students is 8,284,530. For the factors related to institutional policy, regulation quality indicator gets the highest averaged point (40.2), followed by law indicator (39.17) and lastly, corruption control (36.26). 3.4.2. Identify correlation coefficient matrix among variables The majority of independent variables used (except for Infla, ExchRa, Open and Corrupt) have correlation coefficients with FDI at significance level less than or equals to 10%. Of those, the ones that have correlation sign contradicting theoretical reasoning are FinDev, ODA, Open and Resour. Furthermore, the correlation coefficients between FDI and GDP is estimated to be highest (0.8478). About the correlation among independent 2 According to UNCTAD, whenever a nation has 1 of the 3 FDI flows (equity, returns for reinvestment and crossfirm debts) that has a negative value and there is not enough positive inflow, the FDI has a negative sign (divestment). --13-- variables, the majority of correlation coefficients among the policy framework group and economic group variables are all small (less than 0.6), with the exception of the correlation between GDP and ODA and HuCa (close to 0.8). These coefficients reflect the close relationship between GDP and ODA and number of vocational students. In contrast, for the factors related to institutional quality the coefficients are high, with the highest being the coefficients between Corrupt and Law (0.88). These results confirm the existent of a multi-collinearity in the model. However, to reach research goal, this violation is acceptable and the variables are still chosen. 3.4.3. Regression results Table 3.7: Results from one step differenced GMM for current data Variable Infla Model 1 0,0179 Model 2 -0,0071 Model 3 Model 4 -0,0155 -0,0155 -0,7287 -3,7990*** -3,8207*** ExchRa -0,6071 FinDev 0,0142 -0,0163* -0,0098 -0,0128 -0,1527* 0,1107* 0,1457** 0,1444** 1,5840** 1,5038** 1,5281** 0,0163 -0,0193 -0,0158 0,0438*** 0,0459*** 0,0479*** 0,0405** 0,0474*** 0,0483*** Infras 0,0128 -0,0002 -0,0196 HuCa 0,5114 0,8735 10,233 -0,0601 -0,1221 Regul 0,2545 0,2634 Law 0,2641 0,3598 -0,0278 -0,0194 ODA GDP UrPop Open Resour Corrupt FDI_1 InflaVN 0,3497*** 0,0336 -10,375 FinDevVN 0,0855 ODAVN 0,9653 --14-- UrPopVN -6,4973 OpenVN 0,9933 ResourVN -32,9104 InfrasVN -10,5608 CorruptVN -27.3501 LawVN -16,9958 FDI_1VN 24,4690 227 185 169 169 Wald Test 14,930 12,530 13,380 7,810 P-value 0,0000 0,0000 0,0000 0,0000 Sagan test: 31,530 37,230 33,730 38,860 P-value 0,012 0,241 0,384 0,188 Arellano-Bond test 0,460 -0,780 -0,890 -0,830 P-value 0,647 0,436 0,376 0,404 No. of observations * : Pvalue <10% ; ** : Pvalue <5%; *** : Pvalue <1% Table 3.8: Results from one step differenced GMM for past data Variable Model 1 Model 2 Model 3 Model 4 Infla_1 -0,0329** -0,0303** -0,0251* -0,0391** ExchRa_1 0,1314 -0,9026* -1,5651 -2,6318 FinDev_1 0,0285*** 0,0082 0,0101 0,0340* -0,1159 -0,1255* -0,0897 -0,0149 -0,8610* -0,6133 -0,5192 UrPop_1 0,1457* 0,0202 0,0866 Open_1 0,0188* 0,0161 0,0332** Resour_1 -0,0122 -0,0104 -0,0065 Infras_1 0,6068*** 0,6192*** 0,5374* HuCa_1 0,0883 0,7062 0,6349 0,0503 0,0112 ODA_1 GDP_1 Corrupt_1 --15-- Regul_1 Law_1 FDI_1 0,5103*** -0,0427 -0,3936 0,0542 0,4246 -0,4159 -0,0579 -0,1978 InflaVN_1 -0,1756 FinDevVN_1 -0,1576 ODAVN_1 -0,2083 UrPopVN_1 2,1666 OpenVN_1 -0,0393 ResourVN_1 -0,2489 InfrasVN_1 -15,124 CorruptVN_1 -3,0109 FDIVN_1 35,952 230 188 159 146 18,59 11,71 7,77 4,15 P-value 0,0000 0,0000 0,0000 0,0000 Sagan test: 19,430 58,610 64,400 88,600 0,195 0,001 0,000 0,000 -0,430 -1,740 -1,720 -2,050 0,664 0,082 0,085 0,040 No. of observations Wald Test P-value Arellano-Bond test P-value * : Pvalue <10% ; ** : Pvalue <5%; *** : Pvalue <1% In table 3.7, the results for one-step differenced GMM for current data of model 3 and 4 indicate that most of the variables related to policy framework group and economic group have significant coefficients. Specifically, at 1% level the significant variables are ExchRa, Open and HuCa, at the 5% level GDP is tested to be significant. Different from OLS, the first lag of FDI in this case is not significant even at 10%. In addition, the slope differences between Vietnam and the 23 countries in the sample are not significant at the 10% level (the interaction terms of ExchRaVN, --16-- GDPVN, HuCaVN and RegulVN is excluded for causing multicollinearity). With respect to model’s goodness-of-fit when using differenced GMM, for model 3 and 4, the Sargan and Arellano-Bond tests are both insignificant at the 10% level (P-value of those are all larger than 0.1) so that we cannot reject the null hypotheses of exogenous instruments and serial correlation. This shows that GMM method is suitable for fitting the data. Additionally, Wald test results for both models have P-value smaller than 0.01, confirming that the models are suitable for explaining the impact on FDI. Similar reasoning are used to justify the use of past data. Unlike current data, there are only have two variables, Infla_1 and Infras_1 that have significant impact on FDI to ASIA 24 in general and Vietnam in specific (at 10% level). FDI_1 is tested to not have significance in impact on FDI at 10%. Aside from the interaction terms, ExchRaVN_1, GDPVN_1, HuCaVN_1, RegulVN_1 and LawVN_1 is excluded for causing multicollinearity. Testing results show that similar to the case of current data, the interaction terms given are all insignificant at the 10% level. For goodnessof-fit, Wald test results for both models have P-value smaller than 0.01, confirming that the models are suitable for explaining the impact on FDI. However this is not reliable, since the results from Sargan and AreallanoBond of model 3 and 4 are all insignificant at 10% (P-value of these tests are both smaller than 0.1). Accordingly, testing results are sufficient to reject null hypotheses of exogenous instrument variables and second order serial correlation. This means differenced GMM estimator is not suitable for past data of independent variables. 3.4.4. Hypotheses testings Results of differenced GMM estimating method indicate that 3 out of 6 hypotheses cannot be rejected at the 10% level, including the policy frameworks, market seeking and resource seeking purpose. 3.5. Conclusion --17-- CHAPTER 4 ANALYSES OF FACTORS INFLUENCING FDI DISTRIBUTION IN VIETNAM 4.1. Introduction 4.2. Hypotheses and research model From the five given hypotheses, theoretical research model of factors influencing FDI distribution across Vietnam’s regions is established as follows: Figure 4.2: Theoretical research model of factors influencing FDI distribution across Vietnam’s regions Based on summary from empirical researches done by Chenga and Kwan (2000), Sun et al. (2002) and Chen (2009) for China, empirical research model of factors influencing FDI distribution across Vietnam’s regions is consolidated as follows: ∑ ∑ ( And ) --18-- ∑ ∑ ( ) Where: + i = 1,…, 63: represents Vietnam’s 63 provinces/cities; + t = 2005,…, 2013 : is an indicator of time; + FDI: Foreign direct investment (registered) ($mil) ; + Xi: a set of variables related to local economic management capability, including inflation rate (Infla) and provincial competitiveness index (PCI); + Yi: a set of variables related to economy, including local gross domestic product (GDP), urban population (UrPop), direct exporting metal (Resour), number of working labour (Labour), number of students in vocational schools (HuCa) and number of landlines (Infras); + Dummy_tt_HN: Dummy variable which is 1 for provinces/cities within 100 km2 from Hanoi and 0 otherwise; + Dummy_tt_HCM: Dummy variable which is 1 for provinces/cities within 100 km2 from Ho Chi Minh city and 0 otherwise; + FDIit-1 or FDI_1: FDI’s first lag. 4.3. Data and research methodology 4.3.1. Data Data on 63 provinces of Viet Nam during 2005 – 2013 were gathered from the yearbooks of General statistics department and the yearbooks of provinces/cities. From original 64 provinces/cities, Ha Tay and Hanoi were merged in 2007, thereby these regions’ data is incorporated under the name of Hanoi in the period. Other regions are coded from 1 to 63, from Hanoi to Ca Mau. 4.3.2. Research methodology With similar reasoning as in Chapter 3, regression models for factors impacting FDI distribution in Vietnam will also be done for both current and past data of independent variables. --19-- Additionally, the first lag of FDI is used as an independent variable influencing foreign investors’ decisions, and data used are panel data from 63 provinces during 2005 – 2013, therefore in theory this is a dynamic panel data regression. Thus, to estimate and test proposed hypotheses related to factors impacting FDI distribution in Vietnam, differenced GMM estimator (with fixed effects) is utilized, similar to that in estimating and testing for FDI determinants in Chapter 3. 4.4. Research results 4.4.1. Variables’ summary statistics On average, during the research period, registered FDI at provinces/cities reached $mil 345.15 per annum, with a standard deviation of 1,039.68. With respect to variables related to local economic management, the averaged growth rate of consumer index compared to last period is about 107%, provincial competitiveness index is higher than 56.3 points and the leading were regions in the South Eastern and the Cuu Long River delta. As of economic variables, averaged local GDP is 34,192.8 billion VND, with annual growth rate of 11.19%. Urban population is more than 409,000 on average, concentrating mostly at the South Eastern and the Cuu Long River delta. Averaged direct export value is $ 1,263,244,000. Labour force in local businesses has the maximum of 4,000.9 thousands and on average is 768.2 thousands. The number of students in vocational schools is more than 8.479. The number of landlines is reported to be close to 177,000. Overall the South Eastern and the Cuu Long River delta are the most prominent areas in all surveyed variables. 4.4.2. Computing the correlation coefficient matrix among variables In comparison with the dependent variable FDI, the correlation coefficients of independent variables are all quite small (less than 0.6). In contrast to other variables, the sign of the coefficient of Infla is not as expected. Additionally, the coefficients of the rest of the variables are relatively small (the largest being that between GDP and Open, more than --20-- 0.85). These signs may indicate a multi-collinearity in the model. However, to reach research goal, this violation is acceptable and the variables are still chosen. 4.4.3. Regression results According to the two-step GMM estimating with current data (table 4.6), the variables tested to be significant at the 1% level are Infla and Dummy_tt_HCM, at 5% are UrPop, HaCa and Dummy_tt_HN, at 10% we have only Labour. Therefore, the distribution of FDI across Vietnam’s regions is affected by inflation rate, urban population, labour force, number of vocational students. Additionally, the trend of FDI convergence at two biggest cities, Hanoi and Ho Chi Minh, is confirmed. In contrast to current data, testing results for past data only indicate the significance of first lag of Infla_1 (at 1% level). This shows that FDI distribution is solely impacted by past information of inflation. Furthermore, unlike OLS, differenced GMM estimator shows insignificant FDI_1 coefficient even at the 10% level. With respect to model’s goodness-of-fit when using differenced GMM, in both cases of current and past data, the p-values of Wald tests are smaller than 1%, so that we can reject the null hypothesis of jointly insignificant coefficients. This shows that GMM method is suitable for explaining the impact on FDI. On the other hand, the results of Sagan and Arellano-Bond GMM method for three models with both types of data are significant at the 5% level. There is not enough evidence to reject the null hypotheses of exogenous instrument variables and no serial correlation, which mean the GMM estimators are suitable for analyzing the data. In specific, the testing results of independent variables’ impact on FDI are unchanged in all three models, showing the stability and reliability of the estimators. Moreover, the fact that GMM estimating method is appropriate for the two data types confirms that FDI distribution is affected by both current and past information.