The Relationship Between Foreign Direct Investment Inflows and Trade Openness: Evidence from ASEAN and Related Countries

Abstract

ASEAN countries have entered into many free trade agreements, and foreign direct investment (FDI) inflows have become the engine of economic growth of these countries. Intrigued by the situation, we conducted this research to examine the relationship between trade openness and FDI inflows to ASEAN countries with the moderation effect of free trade agreements (FTAs). Our research used panel data of 21 countries (including Vietnam) that are members of FTAs related to ASEAN countries from the World Bank dataset and reported on a country level from 1995 to 2019. First, the three econometric models, pooled OLS, fixed effects (FEM), and random effects (REM) were applied to test three research hypotheses. Next, the assumption of heteroskedasticity and autocorrelation was tested through White's test and Wooldridge's test, and resulting in that generalised least squares (GLS) was selected because it is an appropriate econometric model with the dataset. The research found the positive impact of trade openness and FTAs on FDI inflows significantly, and the moderating role of FTAs in the relationship between trade openness and FDI inflows significantly at 0.01 level. Our research findings contribute to validate the FDI theory. These results imply that to attract FDI inflows, FDI oriented governments should increase their economy's trade openness and join many free trade agreements.

1. Introduction

Foreign direct investment (FDI) is an investment made by firms outside their home country. Over the last few decades, FDI has been one of the most important aspects of globalization. Also, FDI has been considered as one of the dominant sources of benefits for economic development. The statistics of the World Bank show that FDI net inflows have contributed to the growth rate of GDP from around 0.5% in 1970 to 5.36% in 2017. For the last four years, the trend of FDI inflows has shifted its concentration to developing countries, especially developing Asian nations. According to the UNCTAD database, they were the largest recipient of FDI inflows in 2018. The past research on FDI highlighted that economic factors, like interest rate, inflation rate, exchange rate, GDP growth rate, etc. are the determinants of FDI. Globalization has encouraged the introduction of FTAs among countries in different regions. Therefore, recent studies on FDI have focused on the impact of trade openness and FTAs on FDI flows, and conflicting results were found. For example, the research result of (Ngo et al., 2020) shows that trade openness has a negative impact in short the term and has no significant impact in the long run on attracting FDI inflows to Vietnam. On the other hand (Zisi & Anamali, 2015) examined the determinants of FDI inflows in ASEAN countries and concluded that the level of trade openness has indeed a positive effect on attracting FDI inflows to countries belonging to these three ASEAN countries, Vietnam, Laos, and Cambodia. The situation encouraged us to study the relationship between trade openness and FDI inflows to ASEAN countries with the moderation effect of free trade agreements. The rest of the paper will present the literature review, data sources, methodology, and findings with discussion.

2. Literature Review

2.1. Foreign Direct Investment, Trade Openness, and Free Trade Agreements

2.1.1. Foreign Direct Investment (FDI) Inflows

FDI is basically an investment made to acquire operations outside the economy of the investor. Research on FDI found that there were motives for firms to invest outside of their home countries. According to Dunning (1993), these motives are rent-seeking, market-seeking, efficiency-seeking, and strategic-asset seeking from most industrialized countries. Specifically, foreign firms always seek cheaper factors and inputs of production (rent-seeking) in foreign countries; they exploit new markets in host countries to boost up their sales (market-seeking) and minimize their different trade restrictions like high transport costs and rules of origin. Efficiency-seeking firms try to enter countries having advantages of location, resource endowment, and government regulations to serve a larger market. Strategic asset-seeking firms with primary concerns keep their international position and competitive advantages. Over the last three decades, FDI inflows have provided strong impetus for economic development across countries. FDI serves as an important source of funds for domestic investment thus, promoting capital formation in the host country, a crowding-in effect (Omisakin et al., 2009; Le & Pham, 2020). It also gives opportunities for improving the level of service sectors, wholesale and retail trade, business, and legal services. In addition to private investment, FDI might strongly and positively contribute to the development of the public service sectors in the recipient nations (Phan & Nguyen, 2020).

2.1.2. Trade Openness and FDI Inflows

Theories on international trade explain that countries specializing in the production of goods or services have the most competitive advantages to stimulate economic growth. Accordingly, trade openness can be described as the approach aiming to facilitate international free trade by the removal of government control on the trade of goods and services. Theoretically, trade openness or restrictions could affect FDI inflows positively or negatively, which has been widely discussed for many decades. The relationship between FDI and trade is not a straightforward one - but one where differences exist and sometimes conflicts seen in the literature. There are numerous studies that have investigated the relationship between trade and FDI. These usually assess the relationship from two perspectives: whether trade and FDI are substitutes for each other or complement each other, and the direction of their causality. Many studies on FDI have indicated that trade openness is an important factor considered by foreign countries to invest in other countries.

Liargovas and Skandalis (2012) and Kumari and Sharma (2017) found that trade openness has a positive and significant relationship with FDI inflows in developing countries. Shah and Khan (2016) proved the importance of trade openness in the attraction of FDI inflows in emerging economies. Wahid et al. (2009) assumed that a country with larger openness to trade is likely to get more FDI inflows. In addition, Peace et al. (2018) showed that international trade has a positive effect on FDI. Sajilan et al. (2019); Ho et al. (2019) found the same results i.e., trade openness has a positive and statistically significant impact on FDI inflows. However, Seyoum et al. (2014) has examined this relationship in 25 sub-Saharan African countries during the years 1977–2009. The study’s result indicated a bidirectional causality relationship between FDI and trade openness. In short, the relationship between FDI inflows and trade openness is very complex, needs greater careful explanation, and may depend on the characteristics of each case. Theoretically, the effect of trade openness on the inflow of FDI varies, followed by the motivation for engaging in FDI activities. Based on the discussion above we propose the first hypothesis:

H1: Trade openness has a significantly positive impact on FDI inflows among ASEAN and related countries.

2.1.3. Free Trade Agreements (FTAs) and FDI Inflows

The effects of FTAs on FDI have been explained in many ways in the literature. Motta and Norman (1996) found that market access improvement and economic integration encourage outside firms to invest in the integrated regional economic block, leading to increased trade volumes between the integrating countries. Neary (2002) demonstrated that the reduction in internal tariffs and the presence of high external tariffs make tariff jumping FDI, from non-member countries, more favorable than exporting. Heinrich and Eby Konan (2000) found that, at a higher level of initial trade distortion, pre-existing investments may be rationalized because firms concentrate their production in a single plant in the Preferential Trade Agreement. At a lower level of initial trade distortion, the market expansion effect will bring in FDI. It is likely that the establishment of FTAs, the tariff-jumping activity would increase FDI due to the high level of most favored nation tariffs. Initially, Puga and Venables (1997) stated that FDI could be directed into FTAs. Raff (2004) confirmed that FTAs may lead to FDI creation or consolidation. Yeyati et al. (2002) and Lederman et al. (2005) presumed a strong positive effect of FTAs on FDI inflows. In a more detailed study, Moon (2009) separates FDI into vertical and horizontal FDI. He finds that, as expected, vertical FDI is increased by FTAs while horizontal FDI is substituted by exports.

For the ASEAN region, the study of Yoo (2016) specified that ASEAN FTA was positively effective to attract vertical FDI to this region, whilst horizontal FDI was dominant before ASEAN FTA. However, the results were not consistent among countries. For instance, in the case of Singapore, ASEAN FTA was not effective to attract FDI. Meanwhile, in the ongoing industrialization of markets of Thailand, Malaysia, and the Philippines, FTA and FDI had a negative relationship. On the other hand, in the emerging industrialization countries like Indonesia, Vietnam, and Cambodia, ASEAN FTA had a positive effect to attract vertical FDI.

Based on the discussion above we propose the second and the third hypotheses:

H2: FTAs have a positive and significant impact on FDI inflows.

H3: FTAs moderate the impact of trade openness on FDI inflows among ASEAN and related countries.

3. Data Sources

The World Bank database that reported on a country level from 1995 to 2019 was the source for this study. We collected the aggregate data of the 21 countries that are members of some FTAs related to ASEAN countries, showed in Table 1 below, excluding Armenia, Belarus, Kazakhstan, Kyrgyzstan. The dataset contains 229 missing values due to the unavailability of the data, which represents only 6% of the total dataset. In addition to country-level data, this study also uses the world uncertainty index, extracted from https://worlduncertaintyindex.com/data/, to capture the effects of world trade uncertainty events on FDI inflows.

4. Methodology

4.1. Model Specification

Our dependent variable is the aggregate net FDI inflow in current US dollars, denoted by inward foreign direct investment (IFDI), in line with several papers (Bobenic et al., 2018; Gupta & Singh, 2016; Nguyen et al., 2020). This value is log-normalized to enable comparability with other variables.

The potential precedents of FDI inflows are selected in accordance with the previous empirical findings and the variables in the models are constructed in the same way for all studied countries, in order to provide comparable results. As noted by UNCTAD (2002), determinants of FDI may be market-related, trade-related, resource-related, efficiency-related, or related to sound economic and/ or political policies. In summation the main independent variables of this study are trade-related, comprising of trade openness and FTAs. Moreover, we also added some control variables, which are broadly grouped into two categories: market-related, that cover the population, GDP growth rate, inflation, exchange rate, and interest rate, and trade-related variables, specifically, world trade uncertainty index.

Firstly, the trade openness of a country is one of the traditional variables for explaining the FDI movements, measured by the sum of export and import, divided by GDP. The same variable was used in most of the previous studies related to FDI inflow determinants (Bobenic et al., 2018; Sanchez-Matrtin et al., 2014; Wei & Zhu, 2007). Exports and imports of goods and services comprise all transactions between residents of a country and the rest of the world, involving a change of ownership from residents to non-residents of general merchandise, non-monetary gold, and services (Bobenic et al., 2018). Both items, as well as GDP, are measured in the same currency i.e., US dollars. The expected effects may differ by the type of investment regarding local market or export orientation, the host country’s foreign exchange control laws, and applied capital taxation. However, for our study, it is expected to have a positive influence on FDI inflows because countries that are more open to trade tend to attract market-seeking FDI (Asongu et al., 2018; Botric & Skuflic, 2006).

Secondly, as a new precedent of FDI inflows, we add FTA as the number of FTA’s entered into and in effect, as shown in Table 1, of each country during the period 1995 to 2019. This variable is expected to have a positive relationship with FDI inflows, meaning that the more FTAs a host country has, the more it attracts FDI from other countries that are in the same FTAs. The value of FTA is also log-normalized.

Table 1: FTAs and Participants

In terms of the global trade uncertainty index (WTUI), we recruit the World Uncertainty Index of Ahir et al. (2019) from https://worlduncertaintyindex.com/data/, which includes the trading uncertainty measures from 143 countries to proxy the global trade uncertainty level. An increase in the global trade uncertainty generates a relative decrease in domestic uncertainty leading some actors to invest and therefore to increase the FDI inflows in the host country (Nguyen et al., 2020).

Regarding the market-related variables, GDP is proxied by the annual GDP growth rate (%). Investment in capital-scarce countries is expected to yield a higher return indicating an inverse relationship between the levels of GDP and the FDI. At the same time, in the case of market-seeking FDI, investors’ predominant intention is to substitute for exports. Hence, we expect a positive association between the GDP growth and the FDI inflows (Brisan & Buiga, 2009; Demirhan & Masca, 2008). Another traditional variable measuring the market size is the number of inhabitants or population (POP), for which we also expect a positive sign as the study of Botric and Skuflic (2006). The value of this variable is also log-normalized.

Inflation (INF) is measured by the harmonized index of consumer prices and reflects the average change over the time in the prices paid by households for a specific, regularly updated basket of consumer goods and services. The coefficient of this variable, on the other hand, is predicted to be negative because a low and stable inflation rate reduces the macroeconomic risks associated with investment and makes the host country more attractive to FDI (Demirhan & Masca, 2008; Wei & Zhu, 2007; Zheng, 2009).

The exchange rate (EXCH), measured by the local currency per USD, is a crucial factor of FDI inflows and some studies on FDI determinants have integrated the exchange rate. Some studies suggest that a depreciation of the host country’s currency attracts FDI (Barrell & Pain, 1996). In the meantime, other research argues that the appreciation of the host currency attracts FDI (Schmidt & Broll, 2009; Waldkirch, 2003). Thus, there is no clear statement as to how exchange rates affect FDI inflows. To compare with other variables, this is also log-normalized.

Finally, interest rate (INTR), proxied by the real interest rate, is expected to have a positive effect on FDI inflows. Higher interest rates increase the value of a country’s currency. Higher interest rates, hence, tend to attract foreign investment, increasing the demand for and value of the home country’s currency (Emmanuel et al., 2019).

Based on these discussions, the specific model used in this research, therefore, takes the following functional form:

\(\begin{aligned} \text { IFDI }_{i t}=& \alpha+\beta_{1} \mathrm{OPEN}_{i t}+\beta_{2} \mathrm{FTA}_{i t}+\beta_{3} \mathrm{POP}_{i t} \\ &+\beta_{4} \mathrm{GDP}_{i t}+\beta_{5} \mathrm{INF}_{i t}+\beta_{6} \mathrm{EXCH}_{i t} \\ &+\beta_{7} \mathrm{INTR}_{i t}+\beta_{8} \mathrm{WTUI}_{i t}+\varepsilon \end{aligned}\)          (1)

\(\begin{aligned} \mathrm{IFDI}_{i t}=& \alpha+\beta_{1} \mathrm{OPEN}_{i t}+\beta_{2} \mathrm{OPEN}_{i t} \times \mathrm{FTA}_{i t} \\ &+\beta_{3} \mathrm{POP}_{i t}+\beta_{4} \mathrm{GDP}_{i t}+\beta_{5} \mathrm{INF}_{i t} \\ &+\beta_{6} \mathrm{EXCH}_{i t}+\beta_{7} \mathrm{INTR}_{i t}+\beta_{8} \mathrm{WTUI}_{i t}+\varepsilon \end{aligned}\)       (2)

where the subscript i denotes country I while t denotes year t; α, β, ε are the intercept, the regression coefficient, and the error term, respectively. The descriptions of variables are presented in Table 2 as below.

Table 2: Variable’s Descriptions

4.2. Econometric Approach

In order to evaluate our baseline specification, panel data analysis (Balestra, 1992) has been employed. Panel data has the advantage that it uses all the information available which is not detectable in pure cross-sections or pure time series. It is therefore used extensively in economics and finance research to study cross-country economic issues. Since the number of observations are typically much larger in panel data. Panel data will produce more reliable parameter estimates, and thus enable us to test the robustness of our linear regression results. Panel data also alleviates the problem of multicollinearity because when the explanatory variables vary in two dimensions (cross-section and time series), they are less likely to be highly correlated (Webb & Hall, 2009).

We applied various econometric specifications, to test the sensitivity of our results to change in the underlying empirical models. Firstly, based on the collected data, this paper will report the descriptive statistics of variables to have a deep understanding of the dataset. Next, the authors conducted a wide range of diagnostics to test correlations among variables. Then, we started with standard panel models, pooled OLS, fixed effects, and random effects, depending on the relevant statistical test in line with Benassy-Quere et al. (2007), and the generalized least squares regression (GLS) could be used to fix statistical problems, if any.

5. Empirical Results

5.1. Descriptive Statistics

Descriptive statistics (Table 3) indicate that not every selected variable in the study has an equal number of observations. This means that the panel data is unbalanced with missing observations. The average value of FDI, after logged, are in the range from 15.309 to 26.396 with a standard deviation of 1.895, which reveals that FDI inflows vary across countries. Figure 1 depicts the average net FDI inflows for all countries in the sample. Interestingly, the net FDI inflows into China, which has the most FTAs, are the highest among the Asian countries. Moreover, Table 3 also states the level of openness related to trading has a huge gap between countries. This value ranges from 0.002, Myanmar, to 4.426, Hong Kong.

Table 3: Descriptive Statistics

Figure 1: FDI Inflows from 1995 to 2019 of Selected Countries

5.2. Multicollinearity Test

As two or more independent variables in multiple regression models are highly correlated, it would cause a multicollinearity problem that generates ineffective regressors. The matrix of correlation analysis between individual variables is the easiest way to understand the multicollinearity problem.

The matrix of the correlation coefficient (Table 4) shows that the magnitude correlation between these variables is less than 0.7; therefore, it is unlikely to occur as multicollinearity in the model. Conducted with a variance inflation factor (VIF) test, also resulted in the same conclusion (Table 5). The coefficients VIF of all variables are less than 10 and the average of VIF is equal to 1.38 or there is no multicollinearity phenomenon existing in the regression model.

Table 4: Correlation Matrix

Table 5: Summary of Regression Models and Testing Results

Note: t-statistics in parentheses, *p < 0.1; **p < 0.05; ***p < 0.01

5.3. Regression Results

To test the research, hypotheses H1 and H2 were run regressions with the three models, Pooled OLS, fixed effects (FEM), and random effects (REM). To test assumptions of Pooled OLS model, we performed heteroskedasticity testing through White’s test and autocorrelation by Wooldridge test. White’s test shows a result that Prob > χ² = 0.000, we reject H0 or there is the existence of the heteroskedasticity phenomenon in the model. The autocorrelation testing results in Prob > F = 0.001, or H0 is rejected, i.e., there is autocorrelation problem in the research model.

However, Pooled OLS method is suspect because it does not consider unobserved heterogeneity or characteristics of each enterprise; therefore, the FEM and REWM are used. Finally, choosing the model was based on the Hausman, Time fixed effects, and Breusch-Pagan tests, and the results are shown in Table 5 below.

As seen from the results in Table 5, the best appropriate model is REM, but it still does not fix the autocorrelation and heteroskedasticity problems in the model. According to Wooldridge (2002), GLS could be used to overcome these problems. These results are presented in Table 5, column 4.

The regression test with the model 1 results in the acceptance of hypotheses H1 and H2. To test hypothesis H3, we applied the same procedure with the model 2, and the test result is shown in Table 5, column 5 confirming the moderating role of FTAs in the relationship between trade openness and FDI inflows, in other words, H3 is accepted.

6. Conclusion

Research on FDI has demonstrated that FDI inflows are one of the most important factors to promote the economic development of many countries in the world. Our research conducted with 21 ASEAN and related countries, from 1995 to 2019, provides empirical evidence confirming the positive impact of trade openness, and FTAs on FDI inflows. Our study also proves the moderating role of FTAs in the relationship between trade openness and FDI inflows. Moreover, our study’s findings contribute to confirm the validity of the FDI theory. The findings imply that to attract FDI inflows countries should increase the economy’s trade openness and enter into many free trade agreements.